d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
e4a5ee711a
freebsd-nq/sys/dev/sk/if_sk.c

3840 lines
102 KiB
C
Raw Normal View History

Try to create some sort of consistency in how the routings to find the multicast hash are written. There are still two distinct algorithms used, and there actually isn't any reason each driver should have its own copy of this function as they could all share one copy of it (if it grew an additional argument).
2003-11-13 20:55:53 +00:00
/* $OpenBSD: if_sk.c,v 2.33 2003/08/12 05:23:06 nate Exp $ */
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* -> /*- for license, minor formatting changes
2005-01-07 02:29:27 +00:00
/*-
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
* Copyright (c) 1997, 1998, 1999, 2000
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Bill Paul <wpaul@ctr.columbia.edu>. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Bill Paul.
* 4. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*/
/* -> /*- for license, minor formatting changes
2005-01-07 02:29:27 +00:00
/*-
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
* Copyright (c) 2003 Nathan L. Binkert <binkertn@umich.edu>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
Remove duplicate FBSDID's, move others to their right place.
2003-11-14 17:16:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* SysKonnect SK-NET gigabit ethernet driver for FreeBSD. Supports
* the SK-984x series adapters, both single port and dual port.
* References:
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
* The XaQti XMAC II datasheet,
* http://www.freebsd.org/~wpaul/SysKonnect/xmacii_datasheet_rev_c_9-29.pdf
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* The SysKonnect GEnesis manual, http://www.syskonnect.com
*
Universally adopt most conventional spelling of acquire.
2007-05-27 20:50:23 +00:00
* Note: XaQti has been acquired by Vitesse, and Vitesse does not have the
Add actual URL for XMAC II datasheet in comments.
2000-10-20 16:18:16 +00:00
* XMAC II datasheet online. I have put my copy at people.freebsd.org as a
* convenience to others until Vitesse corrects this problem:
*
* http://people.freebsd.org/~wpaul/SysKonnect/xmacii_datasheet_rev_c_9-29.pdf
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
*
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Written by Bill Paul <wpaul@ee.columbia.edu>
* Department of Electrical Engineering
* Columbia University, New York City
*/
/*
* The SysKonnect gigabit ethernet adapters consist of two main
* components: the SysKonnect GEnesis controller chip and the XaQti Corp.
* XMAC II gigabit ethernet MAC. The XMAC provides all of the MAC
* components and a PHY while the GEnesis controller provides a PCI
* interface with DMA support. Each card may have between 512K and
* 2MB of SRAM on board depending on the configuration.
*
* The SysKonnect GEnesis controller can have either one or two XMAC
* chips connected to it, allowing single or dual port NIC configurations.
* SysKonnect has the distinction of being the only vendor on the market
* with a dual port gigabit ethernet NIC. The GEnesis provides dual FIFOs,
* dual DMA queues, packet/MAC/transmit arbiters and direct access to the
* XMAC registers. This driver takes advantage of these features to allow
* both XMACs to operate as independent interfaces.
*/
remove trailing spaces
2006-01-17 05:41:20 +00:00
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <sys/bus.h>
#include <sys/endian.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
Add missing <sys/module.h> includes
2004-05-30 20:00:41 +00:00
#include <sys/module.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <sys/socket.h>
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <sys/sockio.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <sys/queue.h>
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
#include <sys/sysctl.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <net/bpf.h>
#include <net/ethernet.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <net/if.h>
The r48589 promised to remove implicit inclusion of if_var.h soon. Prepare to this event, adding if_var.h to files that do need it. Also, include all includes that now are included due to implicit pollution via if_var.h Sponsored by: Netflix Sponsored by: Nginx, Inc.
2013-10-26 17:58:36 +00:00
#include <net/if_var.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <net/if_arp.h>
#include <net/if_dl.h>
#include <net/if_media.h>
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
#include <net/if_types.h>
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <net/if_vlan_var.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#include <machine/bus.h>
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#include <machine/in_cksum.h>
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
#include <machine/resource.h>
#include <sys/rman.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
#include <dev/mii/mii.h>
#include <dev/mii/miivar.h>
#include <dev/mii/brgphyreg.h>
Prefer new location of pci include files (which have only been in the tree for two or more years now), except in a few places where there's code to be compatible with older versions of FreeBSD.
2003-08-22 07:20:27 +00:00
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#if 0
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#define SK_USEIOSPACE
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
The sk(4) driver has moved to /sys/dev/sk
2006-04-27 00:14:02 +00:00
#include <dev/sk/if_skreg.h>
#include <dev/sk/xmaciireg.h>
#include <dev/sk/yukonreg.h>
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
- Express hard dependencies on bus (pci, isa, pccard) and network layer (ether). - Don't abuse module names to facilitate ifconfig module loading; such abuse isn't really needed. (And if we do need type information associated with a module then we should make it explicit and not use hacks.)
2003-04-15 06:37:30 +00:00
MODULE_DEPEND(sk, pci, 1, 1, 1);
MODULE_DEPEND(sk, ether, 1, 1, 1);
Add a missing MODULE_DEPEND() on miibus.. I was working from KMODDEPS which this driver didn't have.
2000-04-29 15:25:56 +00:00
MODULE_DEPEND(sk, miibus, 1, 1, 1);
Replace FreeBSD 3.x syntax (controller miibus0) with 4.x syntax (device miibus) in time for 7.0 :-)
2005-10-22 05:06:55 +00:00
/* "device miibus" required. See GENERIC if you get errors here. */
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
#include "miibus_if.h"
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
static const struct sk_type sk_devs[] = {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
{
VENDORID_SK,
DEVICEID_SK_V1,
"SysKonnect Gigabit Ethernet (V1.0)"
},
{
VENDORID_SK,
DEVICEID_SK_V2,
"SysKonnect Gigabit Ethernet (V2.0)"
},
Add vendor ID to make Marvell chipset work. E.g. to be found on SMC9452TX it seems Submitted by: Jung-uk Kim <jkim@niksun.com> Tested by: <Radu Bogdan 'veedee' Rusu> veedee@c7.campus.utcluj.ro MFC after: 2 weeks
2003-11-12 23:01:15 +00:00
{
VENDORID_MARVELL,
DEVICEID_SK_V2,
"Marvell Gigabit Ethernet"
},
add pci id for Belkin F5D5005 Gigabit ethernet card.
2004-08-20 05:06:10 +00:00
{
VENDORID_MARVELL,
DEVICEID_BELKIN_5005,
"Belkin F5D5005 Gigabit Ethernet"
},
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
{
VENDORID_3COM,
DEVICEID_3COM_3C940,
"3Com 3C940 Gigabit Ethernet"
},
Ahh, the joys of badge engineering. Tell the sk driver that the Linksys EG1032 is yet another variation. It looks just like the 3c940 except it only has a Marvell logo and no 3com logo.
2003-11-07 22:04:26 +00:00
{
VENDORID_LINKSYS,
DEVICEID_LINKSYS_EG1032,
"Linksys EG1032 Gigabit Ethernet"
},
Support the D-Link DGE-530T. Mine appears to have a blank eeprom, so assume they all do and handle that without alarming the user. Also pull in a bit of defensive code from OpenBSD that triggers when a card is recognised but not properly classified as either Genesis or Yukon. Not that I could ever have needed this. :-) Obtained from: OpenBSD/NetBSD (partially) MFC after: 2 weeks
2004-03-31 12:35:51 +00:00
{
VENDORID_DLINK,
Add device ID for second generation D-Link DGE-530T. PR: kern/99903
2006-07-27 05:06:43 +00:00
DEVICEID_DLINK_DGE530T_A1,
"D-Link DGE-530T Gigabit Ethernet"
},
{
VENDORID_DLINK,
DEVICEID_DLINK_DGE530T_B1,
Support the D-Link DGE-530T. Mine appears to have a blank eeprom, so assume they all do and handle that without alarming the user. Also pull in a bit of defensive code from OpenBSD that triggers when a card is recognised but not properly classified as either Genesis or Yukon. Not that I could ever have needed this. :-) Obtained from: OpenBSD/NetBSD (partially) MFC after: 2 weeks
2004-03-31 12:35:51 +00:00
"D-Link DGE-530T Gigabit Ethernet"
},
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{ 0, 0, NULL }
};
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int skc_probe(device_t);
static int skc_attach(device_t);
static int skc_detach(device_t);
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
static int skc_shutdown(device_t);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int skc_suspend(device_t);
static int skc_resume(device_t);
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
static bus_dma_tag_t skc_get_dma_tag(device_t, device_t);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_detach(device_t);
static int sk_probe(device_t);
static int sk_attach(device_t);
static void sk_tick(void *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void sk_yukon_tick(void *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_intr(void *);
static void sk_intr_xmac(struct sk_if_softc *);
static void sk_intr_bcom(struct sk_if_softc *);
static void sk_intr_yukon(struct sk_if_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static __inline void sk_rxcksum(struct ifnet *, struct mbuf *, u_int32_t);
static __inline int sk_rxvalid(struct sk_softc *, u_int32_t, u_int32_t);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_rxeof(struct sk_if_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void sk_jumbo_rxeof(struct sk_if_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_txeof(struct sk_if_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void sk_txcksum(struct ifnet *, struct mbuf *, struct sk_tx_desc *);
static int sk_encap(struct sk_if_softc *, struct mbuf **);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_start(struct ifnet *);
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
static void sk_start_locked(struct ifnet *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_ioctl(struct ifnet *, u_long, caddr_t);
static void sk_init(void *);
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
static void sk_init_locked(struct sk_if_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_init_xmac(struct sk_if_softc *);
static void sk_init_yukon(struct sk_if_softc *);
static void sk_stop(struct sk_if_softc *);
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
static void sk_watchdog(void *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_ifmedia_upd(struct ifnet *);
static void sk_ifmedia_sts(struct ifnet *, struct ifmediareq *);
static void sk_reset(struct sk_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static __inline void sk_discard_rxbuf(struct sk_if_softc *, int);
static __inline void sk_discard_jumbo_rxbuf(struct sk_if_softc *, int);
static int sk_newbuf(struct sk_if_softc *, int);
static int sk_jumbo_newbuf(struct sk_if_softc *, int);
static void sk_dmamap_cb(void *, bus_dma_segment_t *, int, int);
static int sk_dma_alloc(struct sk_if_softc *);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
static int sk_dma_jumbo_alloc(struct sk_if_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void sk_dma_free(struct sk_if_softc *);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
static void sk_dma_jumbo_free(struct sk_if_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_init_rx_ring(struct sk_if_softc *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int sk_init_jumbo_rx_ring(struct sk_if_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_init_tx_ring(struct sk_if_softc *);
static u_int32_t sk_win_read_4(struct sk_softc *, int);
static u_int16_t sk_win_read_2(struct sk_softc *, int);
static u_int8_t sk_win_read_1(struct sk_softc *, int);
static void sk_win_write_4(struct sk_softc *, int, u_int32_t);
static void sk_win_write_2(struct sk_softc *, int, u_int32_t);
static void sk_win_write_1(struct sk_softc *, int, u_int32_t);
Remove __P.
2002-03-20 02:08:01 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_miibus_readreg(device_t, int, int);
static int sk_miibus_writereg(device_t, int, int, int);
static void sk_miibus_statchg(device_t);
Remove __P.
2002-03-20 02:08:01 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_xmac_miibus_readreg(struct sk_if_softc *, int, int);
static int sk_xmac_miibus_writereg(struct sk_if_softc *, int, int,
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
int);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_xmac_miibus_statchg(struct sk_if_softc *);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int sk_marv_miibus_readreg(struct sk_if_softc *, int, int);
static int sk_marv_miibus_writereg(struct sk_if_softc *, int, int,
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
int);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void sk_marv_miibus_statchg(struct sk_if_softc *);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static uint32_t sk_xmchash(const uint8_t *);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void sk_setfilt(struct sk_if_softc *, u_int16_t *, int);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
static void sk_rxfilter(struct sk_if_softc *);
static void sk_rxfilter_genesis(struct sk_if_softc *);
static void sk_rxfilter_yukon(struct sk_if_softc *);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
static int sysctl_int_range(SYSCTL_HANDLER_ARGS, int low, int high);
static int sysctl_hw_sk_int_mod(SYSCTL_HANDLER_ARGS);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
/* Tunables. */
static int jumbo_disable = 0;
TUNABLE_INT("hw.skc.jumbo_disable", &jumbo_disable);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/*
* It seems that SK-NET GENESIS supports very simple checksum offload
* capability for Tx and I believe it can generate 0 checksum value for
* UDP packets in Tx as the hardware can't differenciate UDP packets from
* TCP packets. 0 chcecksum value for UDP packet is an invalid one as it
* means sender didn't perforam checksum computation. For the safety I
* disabled UDP checksum offload capability at the moment. Alternatively
* we can intrduce a LINK0/LINK1 flag as hme(4) did in its Tx checksum
* offload routine.
*/
#define SK_CSUM_FEATURES (CSUM_TCP)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/*
* Note that we have newbus methods for both the GEnesis controller
* itself and the XMAC(s). The XMACs are children of the GEnesis, and
* the miibus code is a child of the XMACs. We need to do it this way
* so that the miibus drivers can access the PHY registers on the
* right PHY. It's not quite what I had in mind, but it's the only
* design that achieves the desired effect.
*/
static device_method_t skc_methods[] = {
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
/* Device interface */
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
DEVMETHOD(device_probe, skc_probe),
DEVMETHOD(device_attach, skc_attach),
DEVMETHOD(device_detach, skc_detach),
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
DEVMETHOD(device_suspend, skc_suspend),
DEVMETHOD(device_resume, skc_resume),
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
DEVMETHOD(device_shutdown, skc_shutdown),
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
DEVMETHOD(bus_get_dma_tag, skc_get_dma_tag),
- There's no need to overwrite the default device method with the default one. Interestingly, these are actually the default for quite some time (bus_generic_driver_added(9) since r52045 and bus_generic_print_child(9) since r52045) but even recently added device drivers do this unnecessarily. Discussed with: jhb, marcel - While at it, use DEVMETHOD_END. Discussed with: jhb - Also while at it, use __FBSDID.
2011-11-22 21:28:20 +00:00
DEVMETHOD_END
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
};
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
static driver_t skc_driver = {
Grrr. Okay, changing the devnames was a bad idea. Put them back the way they were.
1999-09-20 08:47:11 +00:00
"skc",
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
skc_methods,
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
sizeof(struct sk_softc)
};
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
static devclass_t skc_devclass;
static device_method_t sk_methods[] = {
/* Device interface */
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
DEVMETHOD(device_probe, sk_probe),
DEVMETHOD(device_attach, sk_attach),
DEVMETHOD(device_detach, sk_detach),
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
DEVMETHOD(device_shutdown, bus_generic_shutdown),
/* MII interface */
DEVMETHOD(miibus_readreg, sk_miibus_readreg),
DEVMETHOD(miibus_writereg, sk_miibus_writereg),
DEVMETHOD(miibus_statchg, sk_miibus_statchg),
- There's no need to overwrite the default device method with the default one. Interestingly, these are actually the default for quite some time (bus_generic_driver_added(9) since r52045 and bus_generic_print_child(9) since r52045) but even recently added device drivers do this unnecessarily. Discussed with: jhb, marcel - While at it, use DEVMETHOD_END. Discussed with: jhb - Also while at it, use __FBSDID.
2011-11-22 21:28:20 +00:00
DEVMETHOD_END
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
};
static driver_t sk_driver = {
"sk",
sk_methods,
sizeof(struct sk_if_softc)
};
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
static devclass_t sk_devclass;
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
DRIVER_MODULE(skc, pci, skc_driver, skc_devclass, NULL, NULL);
DRIVER_MODULE(sk, skc, sk_driver, sk_devclass, NULL, NULL);
DRIVER_MODULE(miibus, sk, miibus_driver, miibus_devclass, NULL, NULL);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Convert sk(4) to use the new bus_alloc_resources() API and bus_{read,write}_* macros. Submitted by: Antoine Brodin <antoine.brodin AT laposte DOT net> Reviewed by: imp (initial version)
2006-07-19 04:12:59 +00:00
static struct resource_spec sk_res_spec_io[] = {
{ SYS_RES_IOPORT, PCIR_BAR(1), RF_ACTIVE },
{ SYS_RES_IRQ, 0, RF_ACTIVE | RF_SHAREABLE },
{ -1, 0, 0 }
};
static struct resource_spec sk_res_spec_mem[] = {
{ SYS_RES_MEMORY, PCIR_BAR(0), RF_ACTIVE },
{ SYS_RES_IRQ, 0, RF_ACTIVE | RF_SHAREABLE },
{ -1, 0, 0 }
};
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
#define SK_SETBIT(sc, reg, x) \
CSR_WRITE_4(sc, reg, CSR_READ_4(sc, reg) | x)
#define SK_CLRBIT(sc, reg, x) \
CSR_WRITE_4(sc, reg, CSR_READ_4(sc, reg) & ~x)
#define SK_WIN_SETBIT_4(sc, reg, x) \
sk_win_write_4(sc, reg, sk_win_read_4(sc, reg) | x)
#define SK_WIN_CLRBIT_4(sc, reg, x) \
sk_win_write_4(sc, reg, sk_win_read_4(sc, reg) & ~x)
#define SK_WIN_SETBIT_2(sc, reg, x) \
sk_win_write_2(sc, reg, sk_win_read_2(sc, reg) | x)
#define SK_WIN_CLRBIT_2(sc, reg, x) \
sk_win_write_2(sc, reg, sk_win_read_2(sc, reg) & ~x)
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static u_int32_t
sk_win_read_4(sc, reg)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
return(CSR_READ_4(sc, SK_WIN_BASE + SK_REG(reg)));
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#else
return(CSR_READ_4(sc, reg));
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static u_int16_t
sk_win_read_2(sc, reg)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
return(CSR_READ_2(sc, SK_WIN_BASE + SK_REG(reg)));
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#else
return(CSR_READ_2(sc, reg));
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static u_int8_t
sk_win_read_1(sc, reg)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
return(CSR_READ_1(sc, SK_WIN_BASE + SK_REG(reg)));
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#else
return(CSR_READ_1(sc, reg));
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_win_write_4(sc, reg, val)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
u_int32_t val;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
CSR_WRITE_4(sc, SK_WIN_BASE + SK_REG(reg), val);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#else
CSR_WRITE_4(sc, reg, val);
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_win_write_2(sc, reg, val)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
u_int32_t val;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
CSR_WRITE_2(sc, SK_WIN_BASE + SK_REG(reg), val);
#else
CSR_WRITE_2(sc, reg, val);
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_win_write_1(sc, reg, val)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
int reg;
u_int32_t val;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#ifdef SK_USEIOSPACE
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_RAP, SK_WIN(reg));
CSR_WRITE_1(sc, SK_WIN_BASE + SK_REG(reg), val);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
#else
CSR_WRITE_1(sc, reg, val);
#endif
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
sk_miibus_readreg(dev, phy, reg)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
device_t dev;
int phy, reg;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int v;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc_if = device_get_softc(dev);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_LOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch(sc_if->sk_softc->sk_type) {
case SK_GENESIS:
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
v = sk_xmac_miibus_readreg(sc_if, phy, reg);
break;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
v = sk_marv_miibus_readreg(sc_if, phy, reg);
break;
default:
v = 0;
break;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_UNLOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (v);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
static int
sk_miibus_writereg(dev, phy, reg, val)
device_t dev;
int phy, reg, val;
{
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int v;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sc_if = device_get_softc(dev);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_LOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch(sc_if->sk_softc->sk_type) {
case SK_GENESIS:
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
v = sk_xmac_miibus_writereg(sc_if, phy, reg, val);
break;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
v = sk_marv_miibus_writereg(sc_if, phy, reg, val);
break;
default:
v = 0;
break;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_UNLOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (v);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
static void
sk_miibus_statchg(dev)
device_t dev;
{
struct sk_if_softc *sc_if;
sc_if = device_get_softc(dev);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_LOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch(sc_if->sk_softc->sk_type) {
case SK_GENESIS:
sk_xmac_miibus_statchg(sc_if);
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_marv_miibus_statchg(sc_if);
break;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_MII_UNLOCK(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
return;
}
static int
sk_xmac_miibus_readreg(sc_if, phy, reg)
struct sk_if_softc *sc_if;
int phy, reg;
{
int i;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
SK_XM_WRITE_2(sc_if, XM_PHY_ADDR, reg|(phy << 8));
SK_XM_READ_2(sc_if, XM_PHY_DATA);
if (sc_if->sk_phytype != SK_PHYTYPE_XMAC) {
for (i = 0; i < SK_TIMEOUT; i++) {
DELAY(1);
if (SK_XM_READ_2(sc_if, XM_MMUCMD) &
XM_MMUCMD_PHYDATARDY)
break;
}
if (i == SK_TIMEOUT) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if_printf(sc_if->sk_ifp, "phy failed to come ready\n");
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
return(0);
}
}
DELAY(1);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
i = SK_XM_READ_2(sc_if, XM_PHY_DATA);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
return(i);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_writereg(sc_if, phy, reg, val)
struct sk_if_softc *sc_if;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
int phy, reg, val;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
int i;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
SK_XM_WRITE_2(sc_if, XM_PHY_ADDR, reg|(phy << 8));
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
for (i = 0; i < SK_TIMEOUT; i++) {
if (!(SK_XM_READ_2(sc_if, XM_MMUCMD) & XM_MMUCMD_PHYBUSY))
break;
}
if (i == SK_TIMEOUT) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if_printf(sc_if->sk_ifp, "phy failed to come ready\n");
return (ETIMEDOUT);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
SK_XM_WRITE_2(sc_if, XM_PHY_DATA, val);
for (i = 0; i < SK_TIMEOUT; i++) {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
DELAY(1);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (!(SK_XM_READ_2(sc_if, XM_MMUCMD) & XM_MMUCMD_PHYBUSY))
break;
}
if (i == SK_TIMEOUT)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if_printf(sc_if->sk_ifp, "phy write timed out\n");
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
return(0);
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_statchg(sc_if)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct sk_if_softc *sc_if;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
{
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct mii_data *mii;
mii = device_get_softc(sc_if->sk_miibus);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/*
* If this is a GMII PHY, manually set the XMAC's
* duplex mode accordingly.
*/
if (sc_if->sk_phytype != SK_PHYTYPE_XMAC) {
if ((mii->mii_media_active & IFM_GMASK) == IFM_FDX) {
SK_XM_SETBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_GMIIFDX);
} else {
SK_XM_CLRBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_GMIIFDX);
}
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
static int
sk_marv_miibus_readreg(sc_if, phy, reg)
struct sk_if_softc *sc_if;
int phy, reg;
{
u_int16_t val;
int i;
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
if (sc_if->sk_phytype != SK_PHYTYPE_MARV_COPPER &&
sc_if->sk_phytype != SK_PHYTYPE_MARV_FIBER) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
return(0);
}
SK_YU_WRITE_2(sc_if, YUKON_SMICR, YU_SMICR_PHYAD(phy) |
YU_SMICR_REGAD(reg) | YU_SMICR_OP_READ);
remove trailing spaces
2006-01-17 05:41:20 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
for (i = 0; i < SK_TIMEOUT; i++) {
DELAY(1);
val = SK_YU_READ_2(sc_if, YUKON_SMICR);
if (val & YU_SMICR_READ_VALID)
break;
}
if (i == SK_TIMEOUT) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if_printf(sc_if->sk_ifp, "phy failed to come ready\n");
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
return(0);
}
remove trailing spaces
2006-01-17 05:41:20 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
val = SK_YU_READ_2(sc_if, YUKON_SMIDR);
return(val);
}
static int
sk_marv_miibus_writereg(sc_if, phy, reg, val)
struct sk_if_softc *sc_if;
int phy, reg, val;
{
int i;
SK_YU_WRITE_2(sc_if, YUKON_SMIDR, val);
SK_YU_WRITE_2(sc_if, YUKON_SMICR, YU_SMICR_PHYAD(phy) |
YU_SMICR_REGAD(reg) | YU_SMICR_OP_WRITE);
for (i = 0; i < SK_TIMEOUT; i++) {
DELAY(1);
Fix incorrect busy check for PHY write operation. While I'm here remove unnecessary return statement.
2006-08-02 02:36:59 +00:00
if ((SK_YU_READ_2(sc_if, YUKON_SMICR) & YU_SMICR_BUSY) == 0)
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
break;
}
Fix incorrect busy check for PHY write operation. While I'm here remove unnecessary return statement.
2006-08-02 02:36:59 +00:00
if (i == SK_TIMEOUT)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if_printf(sc_if->sk_ifp, "phy write timeout\n");
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
return(0);
}
static void
sk_marv_miibus_statchg(sc_if)
struct sk_if_softc *sc_if;
{
return;
}
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
#define HASH_BITS 6
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static u_int32_t
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
sk_xmchash(addr)
Don't use caddr_t in mchash(). Also use C99 spellings over BSD ones. Requested by: bde,imp
2003-12-08 07:54:15 +00:00
const uint8_t *addr;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Don't use caddr_t in mchash(). Also use C99 spellings over BSD ones. Requested by: bde,imp
2003-12-08 07:54:15 +00:00
uint32_t crc;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Compute CRC for the address value. */
Replace handrolled CRC calculation with ether_crc32_[lb]e().
2004-06-09 14:34:04 +00:00
crc = ether_crc32_le(addr, ETHER_ADDR_LEN);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
return (~crc & ((1 << HASH_BITS) - 1));
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_setfilt(sc_if, addr, slot)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int16_t *addr;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
int slot;
{
int base;
base = XM_RXFILT_ENTRY(slot);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_XM_WRITE_2(sc_if, base, addr[0]);
SK_XM_WRITE_2(sc_if, base + 2, addr[1]);
SK_XM_WRITE_2(sc_if, base + 4, addr[2]);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
sk_rxfilter(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_softc *sc;
SK_IF_LOCK_ASSERT(sc_if);
sc = sc_if->sk_softc;
if (sc->sk_type == SK_GENESIS)
sk_rxfilter_genesis(sc_if);
else
sk_rxfilter_yukon(sc_if);
}
static void
sk_rxfilter_genesis(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc_if->sk_ifp;
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
u_int32_t hashes[2] = { 0, 0 }, mode;
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
int h = 0, i;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifmultiaddr *ifma;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int16_t dummy[] = { 0, 0, 0 };
u_int16_t maddr[(ETHER_ADDR_LEN+1)/2];
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK_ASSERT(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
mode = SK_XM_READ_4(sc_if, XM_MODE);
mode &= ~(XM_MODE_RX_PROMISC | XM_MODE_RX_USE_HASH |
XM_MODE_RX_USE_PERFECT);
/* First, zot all the existing perfect filters. */
for (i = 1; i < XM_RXFILT_MAX; i++)
sk_setfilt(sc_if, dummy, i);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Now program new ones. */
if (ifp->if_flags & IFF_ALLMULTI || ifp->if_flags & IFF_PROMISC) {
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
if (ifp->if_flags & IFF_ALLMULTI)
mode |= XM_MODE_RX_USE_HASH;
if (ifp->if_flags & IFF_PROMISC)
mode |= XM_MODE_RX_PROMISC;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
hashes[0] = 0xFFFFFFFF;
hashes[1] = 0xFFFFFFFF;
} else {
i = 1;
Use if_maddr_rlock()/if_maddr_runlock() rather than IF_ADDR_LOCK()/ IF_ADDR_UNLOCK() across network device drivers when accessing the per-interface multicast address list, if_multiaddrs. This will allow us to change the locking strategy without affecting our driver programming interface or binary interface. For two wireless drivers, remove unnecessary locking, since they don't actually access the multicast address list. Approved by: re (kib) MFC after: 6 weeks
2009-06-26 11:45:06 +00:00
if_maddr_rlock(ifp);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
TAILQ_FOREACH_REVERSE(ifma, &ifp->if_multiaddrs, ifmultihead,
ifma_link) {
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
/*
* Program the first XM_RXFILT_MAX multicast groups
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
* into the perfect filter.
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
*/
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
bcopy(LLADDR((struct sockaddr_dl *)ifma->ifma_addr),
maddr, ETHER_ADDR_LEN);
if (i < XM_RXFILT_MAX) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sk_setfilt(sc_if, maddr, i);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
mode |= XM_MODE_RX_USE_PERFECT;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
i++;
continue;
}
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
h = sk_xmchash((const uint8_t *)maddr);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (h < 32)
hashes[0] |= (1 << h);
else
hashes[1] |= (1 << (h - 32));
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
mode |= XM_MODE_RX_USE_HASH;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Use if_maddr_rlock()/if_maddr_runlock() rather than IF_ADDR_LOCK()/ IF_ADDR_UNLOCK() across network device drivers when accessing the per-interface multicast address list, if_multiaddrs. This will allow us to change the locking strategy without affecting our driver programming interface or binary interface. For two wireless drivers, remove unnecessary locking, since they don't actually access the multicast address list. Approved by: re (kib) MFC after: 6 weeks
2009-06-26 11:45:06 +00:00
if_maddr_runlock(ifp);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
SK_XM_WRITE_4(sc_if, XM_MODE, mode);
SK_XM_WRITE_4(sc_if, XM_MAR0, hashes[0]);
SK_XM_WRITE_4(sc_if, XM_MAR2, hashes[1]);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
static void
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
sk_rxfilter_yukon(sc_if)
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
struct sk_if_softc *sc_if;
{
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
struct ifnet *ifp;
u_int32_t crc, hashes[2] = { 0, 0 }, mode;
struct ifmultiaddr *ifma;
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK_ASSERT(sc_if);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
ifp = sc_if->sk_ifp;
mode = SK_YU_READ_2(sc_if, YUKON_RCR);
if (ifp->if_flags & IFF_PROMISC)
mode &= ~(YU_RCR_UFLEN | YU_RCR_MUFLEN);
else if (ifp->if_flags & IFF_ALLMULTI) {
mode |= YU_RCR_UFLEN | YU_RCR_MUFLEN;
hashes[0] = 0xFFFFFFFF;
hashes[1] = 0xFFFFFFFF;
} else {
mode |= YU_RCR_UFLEN;
Use if_maddr_rlock()/if_maddr_runlock() rather than IF_ADDR_LOCK()/ IF_ADDR_UNLOCK() across network device drivers when accessing the per-interface multicast address list, if_multiaddrs. This will allow us to change the locking strategy without affecting our driver programming interface or binary interface. For two wireless drivers, remove unnecessary locking, since they don't actually access the multicast address list. Approved by: re (kib) MFC after: 6 weeks
2009-06-26 11:45:06 +00:00
if_maddr_rlock(ifp);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
crc = ether_crc32_be(LLADDR((struct sockaddr_dl *)
ifma->ifma_addr), ETHER_ADDR_LEN);
/* Just want the 6 least significant bits. */
crc &= 0x3f;
/* Set the corresponding bit in the hash table. */
hashes[crc >> 5] |= 1 << (crc & 0x1f);
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
}
Use if_maddr_rlock()/if_maddr_runlock() rather than IF_ADDR_LOCK()/ IF_ADDR_UNLOCK() across network device drivers when accessing the per-interface multicast address list, if_multiaddrs. This will allow us to change the locking strategy without affecting our driver programming interface or binary interface. For two wireless drivers, remove unnecessary locking, since they don't actually access the multicast address list. Approved by: re (kib) MFC after: 6 weeks
2009-06-26 11:45:06 +00:00
if_maddr_runlock(ifp);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
if (hashes[0] != 0 || hashes[1] != 0)
mode |= YU_RCR_MUFLEN;
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
}
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
SK_YU_WRITE_2(sc_if, YUKON_MCAH1, hashes[0] & 0xffff);
SK_YU_WRITE_2(sc_if, YUKON_MCAH2, (hashes[0] >> 16) & 0xffff);
SK_YU_WRITE_2(sc_if, YUKON_MCAH3, hashes[1] & 0xffff);
SK_YU_WRITE_2(sc_if, YUKON_MCAH4, (hashes[1] >> 16) & 0xffff);
SK_YU_WRITE_2(sc_if, YUKON_RCR, mode);
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
sk_init_rx_ring(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_ring_data *rd;
bus_addr_t addr;
u_int32_t csum_start;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
int i;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sc_if->sk_cdata.sk_rx_cons = 0;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
csum_start = (ETHER_HDR_LEN + sizeof(struct ip)) << 16 |
ETHER_HDR_LEN;
rd = &sc_if->sk_rdata;
bzero(rd->sk_rx_ring, sizeof(struct sk_rx_desc) * SK_RX_RING_CNT);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
for (i = 0; i < SK_RX_RING_CNT; i++) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sk_newbuf(sc_if, i) != 0)
return (ENOBUFS);
if (i == (SK_RX_RING_CNT - 1))
addr = SK_RX_RING_ADDR(sc_if, 0);
else
addr = SK_RX_RING_ADDR(sc_if, i + 1);
rd->sk_rx_ring[i].sk_next = htole32(SK_ADDR_LO(addr));
rd->sk_rx_ring[i].sk_csum_start = htole32(csum_start);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_cdata.sk_rx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return(0);
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int
sk_init_jumbo_rx_ring(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_ring_data *rd;
bus_addr_t addr;
u_int32_t csum_start;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
int i;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sc_if->sk_cdata.sk_jumbo_rx_cons = 0;
csum_start = ((ETHER_HDR_LEN + sizeof(struct ip)) << 16) |
ETHER_HDR_LEN;
rd = &sc_if->sk_rdata;
bzero(rd->sk_jumbo_rx_ring,
sizeof(struct sk_rx_desc) * SK_JUMBO_RX_RING_CNT);
for (i = 0; i < SK_JUMBO_RX_RING_CNT; i++) {
if (sk_jumbo_newbuf(sc_if, i) != 0)
return (ENOBUFS);
if (i == (SK_JUMBO_RX_RING_CNT - 1))
addr = SK_JUMBO_RX_RING_ADDR(sc_if, 0);
else
addr = SK_JUMBO_RX_RING_ADDR(sc_if, i + 1);
rd->sk_jumbo_rx_ring[i].sk_next = htole32(SK_ADDR_LO(addr));
rd->sk_jumbo_rx_ring[i].sk_csum_start = htole32(csum_start);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_cdata.sk_jumbo_rx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void
sk_init_tx_ring(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_ring_data *rd;
struct sk_txdesc *txd;
bus_addr_t addr;
int i;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
STAILQ_INIT(&sc_if->sk_cdata.sk_txfreeq);
STAILQ_INIT(&sc_if->sk_cdata.sk_txbusyq);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sc_if->sk_cdata.sk_tx_prod = 0;
sc_if->sk_cdata.sk_tx_cons = 0;
sc_if->sk_cdata.sk_tx_cnt = 0;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
rd = &sc_if->sk_rdata;
bzero(rd->sk_tx_ring, sizeof(struct sk_tx_desc) * SK_TX_RING_CNT);
for (i = 0; i < SK_TX_RING_CNT; i++) {
if (i == (SK_TX_RING_CNT - 1))
addr = SK_TX_RING_ADDR(sc_if, 0);
else
addr = SK_TX_RING_ADDR(sc_if, i + 1);
rd->sk_tx_ring[i].sk_next = htole32(SK_ADDR_LO(addr));
txd = &sc_if->sk_cdata.sk_txdesc[i];
STAILQ_INSERT_TAIL(&sc_if->sk_cdata.sk_txfreeq, txd, tx_q);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static __inline void
sk_discard_rxbuf(sc_if, idx)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int idx;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_rx_desc *r;
struct sk_rxdesc *rxd;
struct mbuf *m;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
r = &sc_if->sk_rdata.sk_rx_ring[idx];
rxd = &sc_if->sk_cdata.sk_rxdesc[idx];
m = rxd->rx_m;
r->sk_ctl = htole32(m->m_len | SK_RXSTAT | SK_OPCODE_CSUM);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static __inline void
sk_discard_jumbo_rxbuf(sc_if, idx)
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int idx;
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_rx_desc *r;
struct sk_rxdesc *rxd;
struct mbuf *m;
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
r = &sc_if->sk_rdata.sk_jumbo_rx_ring[idx];
rxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[idx];
m = rxd->rx_m;
r->sk_ctl = htole32(m->m_len | SK_RXSTAT | SK_OPCODE_CSUM);
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int
sk_newbuf(sc_if, idx)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int idx;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_rx_desc *r;
struct sk_rxdesc *rxd;
struct mbuf *m;
bus_dma_segment_t segs[1];
bus_dmamap_t map;
int nsegs;
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (m == NULL)
return (ENOBUFS);
m->m_len = m->m_pkthdr.len = MCLBYTES;
m_adj(m, ETHER_ALIGN);
if (bus_dmamap_load_mbuf_sg(sc_if->sk_cdata.sk_rx_tag,
sc_if->sk_cdata.sk_rx_sparemap, m, segs, &nsegs, 0) != 0) {
m_freem(m);
return (ENOBUFS);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
KASSERT(nsegs == 1, ("%s: %d segments returned!", __func__, nsegs));
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
rxd = &sc_if->sk_cdata.sk_rxdesc[idx];
if (rxd->rx_m != NULL) {
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_tag, rxd->rx_dmamap,
BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc_if->sk_cdata.sk_rx_tag, rxd->rx_dmamap);
}
map = rxd->rx_dmamap;
rxd->rx_dmamap = sc_if->sk_cdata.sk_rx_sparemap;
sc_if->sk_cdata.sk_rx_sparemap = map;
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_tag, rxd->rx_dmamap,
BUS_DMASYNC_PREREAD);
rxd->rx_m = m;
r = &sc_if->sk_rdata.sk_rx_ring[idx];
r->sk_data_lo = htole32(SK_ADDR_LO(segs[0].ds_addr));
r->sk_data_hi = htole32(SK_ADDR_HI(segs[0].ds_addr));
r->sk_ctl = htole32(segs[0].ds_len | SK_RXSTAT | SK_OPCODE_CSUM);
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int
sk_jumbo_newbuf(sc_if, idx)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int idx;
{
struct sk_rx_desc *r;
struct sk_rxdesc *rxd;
struct mbuf *m;
bus_dma_segment_t segs[1];
bus_dmamap_t map;
int nsegs;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m = m_getjcl(M_NOWAIT, MT_DATA, M_PKTHDR, MJUM9BYTES);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (m == NULL)
return (ENOBUFS);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
m->m_pkthdr.len = m->m_len = MJUM9BYTES;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/*
* Adjust alignment so packet payload begins on a
* longword boundary. Mandatory for Alpha, useful on
* x86 too.
*/
m_adj(m, ETHER_ALIGN);
Replace the mbuf external reference counting code with something that should be better. The old code counted references to mbuf clusters by using the offset of the cluster from the start of memory allocated for mbufs and clusters as an index into an array of chars, which did the reference counting. If the external storage was not a cluster then reference counting had to be done by the code using that external storage. NetBSD's system of linked lists of mbufs was cosidered, but Alfred felt it would have locking issues when the kernel was made more SMP friendly. The system implimented uses a pool of unions to track external storage. The union contains an int for counting the references and a pointer for forming a free list. The reference counts are incremented and decremented atomically and so should be SMP friendly. This system can track reference counts for any sort of external storage. Access to the reference counting stuff is now through macros defined in mbuf.h, so it should be easier to make changes to the system in the future. The possibility of storing the reference count in one of the referencing mbufs was considered, but was rejected 'cos it would often leave extra mbufs allocated. Storing the reference count in the cluster was also considered, but because the external storage may not be a cluster this isn't an option. The size of the pool of reference counters is available in the stats provided by "netstat -m". PR: 19866 Submitted by: Bosko Milekic <bmilekic@dsuper.net> Reviewed by: alfred (glanced at by others on -net)
2000-08-19 08:32:59 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (bus_dmamap_load_mbuf_sg(sc_if->sk_cdata.sk_jumbo_rx_tag,
sc_if->sk_cdata.sk_jumbo_rx_sparemap, m, segs, &nsegs, 0) != 0) {
m_freem(m);
return (ENOBUFS);
}
KASSERT(nsegs == 1, ("%s: %d segments returned!", __func__, nsegs));
rxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[idx];
if (rxd->rx_m != NULL) {
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_tag, rxd->rx_dmamap,
BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc_if->sk_cdata.sk_jumbo_rx_tag,
rxd->rx_dmamap);
}
map = rxd->rx_dmamap;
rxd->rx_dmamap = sc_if->sk_cdata.sk_jumbo_rx_sparemap;
sc_if->sk_cdata.sk_jumbo_rx_sparemap = map;
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_tag, rxd->rx_dmamap,
BUS_DMASYNC_PREREAD);
rxd->rx_m = m;
r = &sc_if->sk_rdata.sk_jumbo_rx_ring[idx];
r->sk_data_lo = htole32(SK_ADDR_LO(segs[0].ds_addr));
r->sk_data_hi = htole32(SK_ADDR_HI(segs[0].ds_addr));
r->sk_ctl = htole32(segs[0].ds_len | SK_RXSTAT | SK_OPCODE_CSUM);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
/*
* Set media options.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
sk_ifmedia_upd(ifp)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
struct sk_if_softc *sc_if = ifp->if_softc;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct mii_data *mii;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii = device_get_softc(sc_if->sk_miibus);
sk_init(sc_if);
mii_mediachg(mii);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return(0);
}
/*
* Report current media status.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_ifmedia_sts(ifp, ifmr)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp;
struct ifmediareq *ifmr;
{
struct sk_if_softc *sc_if;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct mii_data *mii;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc_if = ifp->if_softc;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii = device_get_softc(sc_if->sk_miibus);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii_pollstat(mii);
ifmr->ifm_active = mii->mii_media_active;
ifmr->ifm_status = mii->mii_media_status;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
sk_ioctl(ifp, command, data)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp;
u_long command;
caddr_t data;
{
struct sk_if_softc *sc_if = ifp->if_softc;
struct ifreq *ifr = (struct ifreq *) data;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int error, mask;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct mii_data *mii;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = 0;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
switch(command) {
case SIOCSIFMTU:
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
if (ifr->ifr_mtu < ETHERMIN || ifr->ifr_mtu > SK_JUMBO_MTU)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
error = EINVAL;
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
else if (ifp->if_mtu != ifr->ifr_mtu) {
if (sc_if->sk_jumbo_disable != 0 &&
ifr->ifr_mtu > SK_MAX_FRAMELEN)
error = EINVAL;
else {
SK_IF_LOCK(sc_if);
ifp->if_mtu = ifr->ifr_mtu;
if (ifp->if_drv_flags & IFF_DRV_RUNNING) {
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
sk_init_locked(sc_if);
}
SK_IF_UNLOCK(sc_if);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
break;
case SIOCSIFFLAGS:
* When adding/deleting multicast addresses, only whack the address filter if the interface is marked RUNNING. Obtained from: NetBSD if_sk.c rev. 1.12 * Don't initialize the card (and start an autonegotiation) every time the IP address changes. Makes 'dhclient sk0' invocations way faster and more consistant. i.e. one DHCPREQUEST elicits the DHCPACK. Obtained from: OpenBSD if_sk.c rev. 1.56 * Additional locking changes in sk_ioctl. PR: kern/61296 should see improvements by the last two. Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:18:58 +00:00
SK_IF_LOCK(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (ifp->if_flags & IFF_UP) {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING) {
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
if ((ifp->if_flags ^ sc_if->sk_if_flags)
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
& (IFF_PROMISC | IFF_ALLMULTI))
sk_rxfilter(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
} else
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
sk_init_locked(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
} else {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sk_stop(sc_if);
}
sc_if->sk_if_flags = ifp->if_flags;
* When adding/deleting multicast addresses, only whack the address filter if the interface is marked RUNNING. Obtained from: NetBSD if_sk.c rev. 1.12 * Don't initialize the card (and start an autonegotiation) every time the IP address changes. Makes 'dhclient sk0' invocations way faster and more consistant. i.e. one DHCPREQUEST elicits the DHCPACK. Obtained from: OpenBSD if_sk.c rev. 1.56 * Additional locking changes in sk_ioctl. PR: kern/61296 should see improvements by the last two. Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:18:58 +00:00
SK_IF_UNLOCK(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK(sc_if);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
sk_rxfilter(sc_if);
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_UNLOCK(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
case SIOCGIFMEDIA:
case SIOCSIFMEDIA:
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii = device_get_softc(sc_if->sk_miibus);
error = ifmedia_ioctl(ifp, ifr, &mii->mii_media, command);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
case SIOCSIFCAP:
SK_IF_LOCK(sc_if);
if (sc_if->sk_softc->sk_type == SK_GENESIS) {
SK_IF_UNLOCK(sc_if);
break;
}
mask = ifr->ifr_reqcap ^ ifp->if_capenable;
Some revision of Yukon controller generates corrupted frame when TX checksum offloading is enabled. The frame has a valid checksum value so payload might be modified during TX checksum calculation. Disable TX checksum offloading but give users chance to enable it when they know their controller works without problems with TX checksum offloading. Reported by: Andrzej Tobola <ato <> iem dot pw dot edu dot pl>
2010-07-09 21:21:08 +00:00
if ((mask & IFCAP_TXCSUM) != 0 &&
(IFCAP_TXCSUM & ifp->if_capabilities) != 0) {
ifp->if_capenable ^= IFCAP_TXCSUM;
if ((ifp->if_capenable & IFCAP_TXCSUM) != 0)
ifp->if_hwassist |= SK_CSUM_FEATURES;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
else
Some revision of Yukon controller generates corrupted frame when TX checksum offloading is enabled. The frame has a valid checksum value so payload might be modified during TX checksum calculation. Disable TX checksum offloading but give users chance to enable it when they know their controller works without problems with TX checksum offloading. Reported by: Andrzej Tobola <ato <> iem dot pw dot edu dot pl>
2010-07-09 21:21:08 +00:00
ifp->if_hwassist &= ~SK_CSUM_FEATURES;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
Some revision of Yukon controller generates corrupted frame when TX checksum offloading is enabled. The frame has a valid checksum value so payload might be modified during TX checksum calculation. Disable TX checksum offloading but give users chance to enable it when they know their controller works without problems with TX checksum offloading. Reported by: Andrzej Tobola <ato <> iem dot pw dot edu dot pl>
2010-07-09 21:21:08 +00:00
if ((mask & IFCAP_RXCSUM) != 0 &&
(IFCAP_RXCSUM & ifp->if_capabilities) != 0)
ifp->if_capenable ^= IFCAP_RXCSUM;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_UNLOCK(sc_if);
break;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
default:
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
error = ether_ioctl(ifp, command, data);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (error);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
/*
* Probe for a SysKonnect GEnesis chip. Check the PCI vendor and device
* IDs against our list and return a device name if we find a match.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
skc_probe(dev)
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
device_t dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
const struct sk_type *t = sk_devs;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
while(t->sk_name != NULL) {
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
if ((pci_get_vendor(dev) == t->sk_vid) &&
(pci_get_device(dev) == t->sk_did)) {
Only allow the sk(4) driver to attach to revision 2 of the LinkSys EG1032 cards and teach the re(4) driver to attach to revision 3 cards. Submitted by: Fredrik Lindberg fli+freebsd-current at shapeshifter dot se MFC after: 2 weeks Reviewed by: imp, mdodd
2005-10-14 18:51:30 +00:00
/*
* Only attach to rev. 2 of the Linksys EG1032 adapter.
* Rev. 3 is supported by re(4).
*/
if ((t->sk_vid == VENDORID_LINKSYS) &&
(t->sk_did == DEVICEID_LINKSYS_EG1032) &&
(pci_get_subdevice(dev) !=
SUBDEVICEID_LINKSYS_EG1032_REV2)) {
t++;
continue;
}
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
device_set_desc(dev, t->sk_name);
Return BUS_PROBE_DEFAULT instead of 0.
2005-02-24 21:32:56 +00:00
return (BUS_PROBE_DEFAULT);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
t++;
}
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
return(ENXIO);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
/*
* Force the GEnesis into reset, then bring it out of reset.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_reset(sc)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
CSR_WRITE_2(sc, SK_CSR, SK_CSR_SW_RESET);
CSR_WRITE_2(sc, SK_CSR, SK_CSR_MASTER_RESET);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
if (SK_YUKON_FAMILY(sc->sk_type))
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
CSR_WRITE_2(sc, SK_LINK_CTRL, SK_LINK_RESET_SET);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
DELAY(1000);
CSR_WRITE_2(sc, SK_CSR, SK_CSR_SW_UNRESET);
DELAY(2);
CSR_WRITE_2(sc, SK_CSR, SK_CSR_MASTER_UNRESET);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
if (SK_YUKON_FAMILY(sc->sk_type))
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
CSR_WRITE_2(sc, SK_LINK_CTRL, SK_LINK_RESET_CLEAR);
if (sc->sk_type == SK_GENESIS) {
/* Configure packet arbiter */
sk_win_write_2(sc, SK_PKTARB_CTL, SK_PKTARBCTL_UNRESET);
sk_win_write_2(sc, SK_RXPA1_TINIT, SK_PKTARB_TIMEOUT);
sk_win_write_2(sc, SK_TXPA1_TINIT, SK_PKTARB_TIMEOUT);
sk_win_write_2(sc, SK_RXPA2_TINIT, SK_PKTARB_TIMEOUT);
sk_win_write_2(sc, SK_TXPA2_TINIT, SK_PKTARB_TIMEOUT);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Enable RAM interface */
sk_win_write_4(sc, SK_RAMCTL, SK_RAMCTL_UNRESET);
/*
* Configure interrupt moderation. The moderation timer
* defers interrupts specified in the interrupt moderation
* timer mask based on the timeout specified in the interrupt
* moderation timer init register. Each bit in the timer
fix interrupt moderation timer frequencies for Yukon Obtained from: OpenBSD
2006-01-17 05:57:44 +00:00
* register represents one tick, so to specify a timeout in
* microseconds, we have to multiply by the correct number of
* ticks-per-microsecond.
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
*/
fix interrupt moderation timer frequencies for Yukon Obtained from: OpenBSD
2006-01-17 05:57:44 +00:00
switch (sc->sk_type) {
case SK_GENESIS:
sc->sk_int_ticks = SK_IMTIMER_TICKS_GENESIS;
break;
default:
sc->sk_int_ticks = SK_IMTIMER_TICKS_YUKON;
break;
}
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
if (bootverbose)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(sc->sk_dev, "interrupt moderation is %d us\n",
sc->sk_int_mod);
fix interrupt moderation timer frequencies for Yukon Obtained from: OpenBSD
2006-01-17 05:57:44 +00:00
sk_win_write_4(sc, SK_IMTIMERINIT, SK_IM_USECS(sc->sk_int_mod,
sc->sk_int_ticks));
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
sk_win_write_4(sc, SK_IMMR, SK_ISR_TX1_S_EOF|SK_ISR_TX2_S_EOF|
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_ISR_RX1_EOF|SK_ISR_RX2_EOF);
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
sk_win_write_1(sc, SK_IMTIMERCTL, SK_IMCTL_START);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_probe(dev)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
device_t dev;
{
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
struct sk_softc *sc;
sc = device_get_softc(device_get_parent(dev));
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/*
* Not much to do here. We always know there will be
* at least one XMAC present, and if there are two,
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
* skc_attach() will create a second device instance
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
* for us.
*/
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch (sc->sk_type) {
case SK_GENESIS:
device_set_desc(dev, "XaQti Corp. XMAC II");
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
device_set_desc(dev, "Marvell Semiconductor, Inc. Yukon");
break;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Return BUS_PROBE_DEFAULT instead of 0.
2005-02-24 21:32:56 +00:00
return (BUS_PROBE_DEFAULT);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Each XMAC chip is attached as a separate logical IP interface.
* Single port cards will have only one logical interface of course.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_attach(dev)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
device_t dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct sk_softc *sc;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
struct ifnet *ifp;
If we got an invalid station address, generate random address. This might be caused by broken BIOS. Reported by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 21:54:32 +00:00
u_int32_t r;
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
int error, i, phy, port;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
u_char eaddr[6];
If we got an invalid station address, generate random address. This might be caused by broken BIOS. Reported by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 21:54:32 +00:00
u_char inv_mac[] = {0, 0, 0, 0, 0, 0};
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
if (dev == NULL)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return(EINVAL);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = 0;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc_if = device_get_softc(dev);
sc = device_get_softc(device_get_parent(dev));
port = *(int *)device_get_ivars(dev);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sc_if->sk_if_dev = dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc_if->sk_port = port;
sc_if->sk_softc = sc;
sc->sk_if[port] = sc_if;
if (port == SK_PORT_A)
sc_if->sk_tx_bmu = SK_BMU_TXS_CSR0;
if (port == SK_PORT_B)
sc_if->sk_tx_bmu = SK_BMU_TXS_CSR1;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_init_mtx(&sc_if->sk_tick_ch, &sc_if->sk_softc->sk_mtx, 0);
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
callout_init_mtx(&sc_if->sk_watchdog_ch, &sc_if->sk_softc->sk_mtx, 0);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sk_dma_alloc(sc_if) != 0) {
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
error = ENOMEM;
goto fail;
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
sk_dma_jumbo_alloc(sc_if);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp = if_alloc(IFT_ETHER);
if (ifp == NULL) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(sc_if->sk_if_dev, "can not if_alloc()\n");
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
error = ENOSPC;
goto fail;
}
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
ifp->if_softc = sc_if;
if_initname(ifp, device_get_name(dev), device_get_unit(dev));
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
Mark sk(4) as capable of handling extended VLAN frames. NICs based on XMAC II chip should be ready for this in their initial mode of operation, and Yukon-based NICs are configured so by the driver. PR: kern/79998 MFC after: 1 month
2005-10-11 22:55:16 +00:00
/*
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
* SK_GENESIS has a bug in checksum offload - From linux.
Mark sk(4) as capable of handling extended VLAN frames. NICs based on XMAC II chip should be ready for this in their initial mode of operation, and Yukon-based NICs are configured so by the driver. PR: kern/79998 MFC after: 1 month
2005-10-11 22:55:16 +00:00
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sc_if->sk_softc->sk_type != SK_GENESIS) {
Some revision of Yukon controller generates corrupted frame when TX checksum offloading is enabled. The frame has a valid checksum value so payload might be modified during TX checksum calculation. Disable TX checksum offloading but give users chance to enable it when they know their controller works without problems with TX checksum offloading. Reported by: Andrzej Tobola <ato <> iem dot pw dot edu dot pl>
2010-07-09 21:21:08 +00:00
ifp->if_capabilities = IFCAP_TXCSUM | IFCAP_RXCSUM;
ifp->if_hwassist = 0;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
} else {
ifp->if_capabilities = 0;
ifp->if_hwassist = 0;
}
ifp->if_capenable = ifp->if_capabilities;
Some revision of Yukon controller generates corrupted frame when TX checksum offloading is enabled. The frame has a valid checksum value so payload might be modified during TX checksum calculation. Disable TX checksum offloading but give users chance to enable it when they know their controller works without problems with TX checksum offloading. Reported by: Andrzej Tobola <ato <> iem dot pw dot edu dot pl>
2010-07-09 21:21:08 +00:00
/*
* Some revision of Yukon controller generates corrupted
* frame when TX checksum offloading is enabled. The
* frame has a valid checksum value so payload might be
* modified during TX checksum calculation. Disable TX
* checksum offloading but give users chance to enable it
* when they know their controller works without problems
* with TX checksum offloading.
*/
ifp->if_capenable &= ~IFCAP_TXCSUM;
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
ifp->if_ioctl = sk_ioctl;
ifp->if_start = sk_start;
ifp->if_init = sk_init;
Enable sk(4) for ATLQ. Thoroughly tested by: Ender <ender NO tog SPAM net> MFC after: 4 weeks
2004-12-24 14:13:38 +00:00
IFQ_SET_MAXLEN(&ifp->if_snd, SK_TX_RING_CNT - 1);
ifp->if_snd.ifq_drv_maxlen = SK_TX_RING_CNT - 1;
IFQ_SET_READY(&ifp->if_snd);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Get station address for this interface. Note that
* dual port cards actually come with three station
* addresses: one for each port, plus an extra. The
* extra one is used by the SysKonnect driver software
* as a 'virtual' station address for when both ports
* are operating in failover mode. Currently we don't
* use this extra address.
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
for (i = 0; i < ETHER_ADDR_LEN; i++)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
eaddr[i] =
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sk_win_read_1(sc, SK_MAC0_0 + (port * 8) + i);
If we got an invalid station address, generate random address. This might be caused by broken BIOS. Reported by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 21:54:32 +00:00
/* Verify whether the station address is invalid or not. */
if (bcmp(eaddr, inv_mac, sizeof(inv_mac)) == 0) {
device_printf(sc_if->sk_if_dev,
"Generating random ethernet address\n");
r = arc4random();
/*
* Set OUI to convenient locally assigned address. 'b'
* is 0x62, which has the locally assigned bit set, and
* the broadcast/multicast bit clear.
*/
eaddr[0] = 'b';
eaddr[1] = 's';
eaddr[2] = 'd';
eaddr[3] = (r >> 16) & 0xff;
eaddr[4] = (r >> 8) & 0xff;
eaddr[5] = (r >> 0) & 0xff;
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Set up RAM buffer addresses. The NIC will have a certain
* amount of SRAM on it, somewhere between 512K and 2MB. We
* need to divide this up a) between the transmitter and
* receiver and b) between the two XMACs, if this is a
* dual port NIC. Our algotithm is to divide up the memory
* evenly so that everyone gets a fair share.
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
*
* Just to be contrary, Yukon2 appears to have separate memory
* for each MAC.
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
*/
Remove incomplete Yukon II support code which was added in if_sk.c, rev 1.123. Yukon II users should use msk(4).
2006-12-14 04:10:26 +00:00
if (sk_win_read_1(sc, SK_CONFIG) & SK_CONFIG_SINGLEMAC) {
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
u_int32_t chunk, val;
chunk = sc->sk_ramsize / 2;
val = sc->sk_rboff / sizeof(u_int64_t);
sc_if->sk_rx_ramstart = val;
val += (chunk / sizeof(u_int64_t));
sc_if->sk_rx_ramend = val - 1;
sc_if->sk_tx_ramstart = val;
val += (chunk / sizeof(u_int64_t));
sc_if->sk_tx_ramend = val - 1;
} else {
u_int32_t chunk, val;
chunk = sc->sk_ramsize / 4;
val = (sc->sk_rboff + (chunk * 2 * sc_if->sk_port)) /
sizeof(u_int64_t);
sc_if->sk_rx_ramstart = val;
val += (chunk / sizeof(u_int64_t));
sc_if->sk_rx_ramend = val - 1;
sc_if->sk_tx_ramstart = val;
val += (chunk / sizeof(u_int64_t));
sc_if->sk_tx_ramend = val - 1;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/* Read and save PHY type and set PHY address */
sc_if->sk_phytype = sk_win_read_1(sc, SK_EPROM1) & 0xF;
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
if (!SK_YUKON_FAMILY(sc->sk_type)) {
switch(sc_if->sk_phytype) {
case SK_PHYTYPE_XMAC:
sc_if->sk_phyaddr = SK_PHYADDR_XMAC;
break;
case SK_PHYTYPE_BCOM:
sc_if->sk_phyaddr = SK_PHYADDR_BCOM;
break;
default:
device_printf(sc->sk_dev, "unsupported PHY type: %d\n",
sc_if->sk_phytype);
error = ENODEV;
SK_IF_UNLOCK(sc_if);
goto fail;
}
} else {
if (sc_if->sk_phytype < SK_PHYTYPE_MARV_COPPER &&
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
sc->sk_pmd != 'S') {
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
/* not initialized, punt */
sc_if->sk_phytype = SK_PHYTYPE_MARV_COPPER;
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
sc->sk_coppertype = 1;
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sc_if->sk_phyaddr = SK_PHYADDR_MARV;
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
if (!(sc->sk_coppertype))
Add some incomplete support for Marvell Yukon EC controllers based on OpenBSD changes. With these changes, PHY part of the driver becomes functional (it senses media changes and negotiates speed just fine), previously it just hang with no PHY message, but no data goes through interface (error message is "can not stop transfer of Tx/Rx descriptor). Hopefully somebody with more clue/free time will be able to pick up after me.
2006-04-28 03:17:37 +00:00
sc_if->sk_phytype = SK_PHYTYPE_MARV_FIBER;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
}
The sk driver developed a bug when the multicast code was changed to use TAILQ macros. The sk_attach_xmac() routine calls sk_init_xmac() before doing the transceiver probe, but *before* ether_ifattach() is called. This causes sk_init_xmac() to call sk_setmulti(), which tries to do a TAILQ_FOREACH(), which it can't do because ether_ifattach() hasn't done a TAILQ_INIT() yet. This causes a NULL pointer dereference and panic in sk_setmulti() at driver load/initialization time. Fixed by calling ether_ifattach() before the MII probe. The code in RELENG_4 still uses the old way of enumerating the multicast list and doesn't have this problem. Yet.
2001-05-09 18:22:42 +00:00
/*
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
* Call MI attach routine. Can't hold locks when calling into ether_*.
The sk driver developed a bug when the multicast code was changed to use TAILQ macros. The sk_attach_xmac() routine calls sk_init_xmac() before doing the transceiver probe, but *before* ether_ifattach() is called. This causes sk_init_xmac() to call sk_setmulti(), which tries to do a TAILQ_FOREACH(), which it can't do because ether_ifattach() hasn't done a TAILQ_INIT() yet. This causes a NULL pointer dereference and panic in sk_setmulti() at driver load/initialization time. Fixed by calling ether_ifattach() before the MII probe. The code in RELENG_4 still uses the old way of enumerating the multicast list and doesn't have this problem. Yet.
2001-05-09 18:22:42 +00:00
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_UNLOCK(sc_if);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ether_ifattach(ifp, eaddr);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK(sc_if);
/*
* The hardware should be ready for VLAN_MTU by default:
* XMAC II has 0x8100 in VLAN Tag Level 1 register initially;
* YU_SMR_MFL_VLAN is set by this driver in Yukon.
*
*/
ifp->if_capabilities |= IFCAP_VLAN_MTU;
ifp->if_capenable |= IFCAP_VLAN_MTU;
/*
* Tell the upper layer(s) we support long frames.
* Must appear after the call to ether_ifattach() because
* ether_ifattach() sets ifi_hdrlen to the default value.
*/
Use define from if_var.h to access a field inside struct if_data, that resides in struct ifnet. Sponsored by: Nginx, Inc.
2014-08-30 19:55:54 +00:00
ifp->if_hdrlen = sizeof(struct ether_vlan_header);
The sk driver developed a bug when the multicast code was changed to use TAILQ macros. The sk_attach_xmac() routine calls sk_init_xmac() before doing the transceiver probe, but *before* ether_ifattach() is called. This causes sk_init_xmac() to call sk_setmulti(), which tries to do a TAILQ_FOREACH(), which it can't do because ether_ifattach() hasn't done a TAILQ_INIT() yet. This causes a NULL pointer dereference and panic in sk_setmulti() at driver load/initialization time. Fixed by calling ether_ifattach() before the MII probe. The code in RELENG_4 still uses the old way of enumerating the multicast list and doesn't have this problem. Yet.
2001-05-09 18:22:42 +00:00
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
* Do miibus setup.
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
*/
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
phy = MII_PHY_ANY;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch (sc->sk_type) {
case SK_GENESIS:
sk_init_xmac(sc_if);
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
if (sc_if->sk_phytype == SK_PHYTYPE_XMAC)
phy = 0;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_init_yukon(sc_if);
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
phy = 0;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
break;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_UNLOCK(sc_if);
Convert the PHY drivers to honor the mii_flags passed down and convert the NIC drivers as well as the PHY drivers to take advantage of the mii_attach() introduced in r213878 to get rid of certain hacks. For the most part these were: - Artificially limiting miibus_{read,write}reg methods to certain PHY addresses; we now let mii_attach() only probe the PHY at the desired address(es) instead. - PHY drivers setting MIIF_* flags based on the NIC driver they hang off from, partly even based on grabbing and using the softc of the parent; we now pass these flags down from the NIC to the PHY drivers via mii_attach(). This got us rid of all such hacks except those of brgphy() in combination with bce(4) and bge(4), which is way beyond what can be expressed with simple flags. While at it, I took the opportunity to change the NIC drivers to pass up the error returned by mii_attach() (previously by mii_phy_probe()) and unify the error message used in this case where and as appropriate as mii_attach() actually can fail for a number of reasons, not just because of no PHY(s) being present at the expected address(es). Reviewed by: jhb, yongari
2010-10-15 14:52:11 +00:00
error = mii_attach(dev, &sc_if->sk_miibus, ifp, sk_ifmedia_upd,
sk_ifmedia_sts, BMSR_DEFCAPMASK, phy, MII_OFFSET_ANY, 0);
if (error != 0) {
device_printf(sc_if->sk_if_dev, "attaching PHYs failed\n");
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
ether_ifdetach(ifp);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
goto fail;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
fail:
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (error) {
/* Access should be ok even though lock has been dropped */
sc->sk_if[port] = NULL;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_detach(dev);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
}
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
return(error);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
/*
* Attach the interface. Allocate softc structures, do ifmedia
* setup and ethernet/BPF attach.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
skc_attach(dev)
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
device_t dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
struct sk_softc *sc;
Remove incomplete Yukon II support code which was added in if_sk.c, rev 1.123. Yukon II users should use msk(4).
2006-12-14 04:10:26 +00:00
int error = 0, *port;
Set ramsize depending on a value from eprom instead of using hardcoded 128k for Yukon devices. 88E8001 only has 64k of on-chip RAM[1]. [1] http://www.marvell.com/products/pcconn/yukon/Yukon_88E8001_10_073103_final.pdf Tested by: amd64, current Approved by: rwatson (mentor) MFC after: 1 week
2004-11-13 20:21:57 +00:00
uint8_t skrs;
Make a missing or empty VPD non-fatal, as it was prior to rev 1.131. With this change, my D-Link DGE-530T rev A1 is operational again.
2007-05-10 13:10:34 +00:00
const char *pname = NULL;
provide routines to access VPD data at the PCI layer... remove sk's own implementation, and use the new calls to get the data... Reviewed by: -arch
2006-10-09 16:15:56 +00:00
char *revstr;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
sc = device_get_softc(dev);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sc->sk_dev = dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Change callers of mtx_init() to pass in an appropriate lock type name. In most cases NULL is passed, but in some cases such as network driver locks (which use the MTX_NETWORK_LOCK macro) and UMA zone locks, a name is used. Tested on: i386, alpha, sparc64
2002-04-04 21:03:38 +00:00
mtx_init(&sc->sk_mtx, device_get_nameunit(dev), MTX_NETWORK_LOCK,
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
MTX_DEF);
mtx_init(&sc->sk_mii_mtx, "sk_mii_mutex", NULL, MTX_DEF);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Map control/status registers.
*/
Big round of minor updates: - Use pci_get_powerstate()/pci_set_powerstate() in all the other drivers that need them so we don't have to fiddle with the PCI power management registers directly. - Use pci_enable_busmaster()/pci_enable_io() to turn on busmastering and PIO/memory mapped accesses. - Add support to the RealTek driver for the D-Link DFE-530TX+ which has a RealTek 8139 with its own PCI ID. (Submitted by Jason Wright) - Have the SiS 900/National DP83815 driver be sure to disable PME mode in sis_reset(). This apparently fixes a problem on some motherboards where the DP83815 chip fails to receive packets. (Submitted by Chuck McCrobie <mccrobie@cablespeed.com>)
2001-02-21 20:54:22 +00:00
pci_enable_busmaster(dev);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Convert sk(4) to use the new bus_alloc_resources() API and bus_{read,write}_* macros. Submitted by: Antoine Brodin <antoine.brodin AT laposte DOT net> Reviewed by: imp (initial version)
2006-07-19 04:12:59 +00:00
/* Allocate resources */
#ifdef SK_USEIOSPACE
sc->sk_res_spec = sk_res_spec_io;
#else
sc->sk_res_spec = sk_res_spec_mem;
#endif
error = bus_alloc_resources(dev, sc->sk_res_spec, sc->sk_res);
if (error) {
if (sc->sk_res_spec == sk_res_spec_mem)
sc->sk_res_spec = sk_res_spec_io;
else
sc->sk_res_spec = sk_res_spec_mem;
error = bus_alloc_resources(dev, sc->sk_res_spec, sc->sk_res);
if (error) {
device_printf(dev, "couldn't allocate %s resources\n",
sc->sk_res_spec == sk_res_spec_mem ? "memory" :
"I/O");
goto fail;
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
sc->sk_type = sk_win_read_1(sc, SK_CHIPVER);
sc->sk_rev = (sk_win_read_1(sc, SK_CONFIG) >> 4) & 0xf;
/* Bail out if chip is not recognized. */
if (sc->sk_type != SK_GENESIS && !SK_YUKON_FAMILY(sc->sk_type)) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(dev, "unknown device: chipver=%02x, rev=%x\n",
sc->sk_type, sc->sk_rev);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
error = ENXIO;
goto fail;
}
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev),
SYSCTL_CHILDREN(device_get_sysctl_tree(dev)),
OID_AUTO, "int_mod", CTLTYPE_INT|CTLFLAG_RW,
&sc->sk_int_mod, 0, sysctl_hw_sk_int_mod, "I",
"SK interrupt moderation");
/* Pull in device tunables. */
sc->sk_int_mod = SK_IM_DEFAULT;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = resource_int_value(device_get_name(dev), device_get_unit(dev),
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
"int_mod", &sc->sk_int_mod);
if (error == 0) {
if (sc->sk_int_mod < SK_IM_MIN ||
sc->sk_int_mod > SK_IM_MAX) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(dev, "int_mod value out of range; "
"using default: %d\n", SK_IM_DEFAULT);
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
sc->sk_int_mod = SK_IM_DEFAULT;
}
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Reset the adapter. */
sk_reset(sc);
Set ramsize depending on a value from eprom instead of using hardcoded 128k for Yukon devices. 88E8001 only has 64k of on-chip RAM[1]. [1] http://www.marvell.com/products/pcconn/yukon/Yukon_88E8001_10_073103_final.pdf Tested by: amd64, current Approved by: rwatson (mentor) MFC after: 1 week
2004-11-13 20:21:57 +00:00
skrs = sk_win_read_1(sc, SK_EPROM0);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sc->sk_type == SK_GENESIS) {
/* Read and save RAM size and RAMbuffer offset */
Set ramsize depending on a value from eprom instead of using hardcoded 128k for Yukon devices. 88E8001 only has 64k of on-chip RAM[1]. [1] http://www.marvell.com/products/pcconn/yukon/Yukon_88E8001_10_073103_final.pdf Tested by: amd64, current Approved by: rwatson (mentor) MFC after: 1 week
2004-11-13 20:21:57 +00:00
switch(skrs) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
case SK_RAMSIZE_512K_64:
sc->sk_ramsize = 0x80000;
sc->sk_rboff = SK_RBOFF_0;
break;
case SK_RAMSIZE_1024K_64:
sc->sk_ramsize = 0x100000;
sc->sk_rboff = SK_RBOFF_80000;
break;
case SK_RAMSIZE_1024K_128:
sc->sk_ramsize = 0x100000;
sc->sk_rboff = SK_RBOFF_0;
break;
case SK_RAMSIZE_2048K_128:
sc->sk_ramsize = 0x200000;
sc->sk_rboff = SK_RBOFF_0;
break;
default:
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(dev, "unknown ram size: %d\n", skrs);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
error = ENXIO;
goto fail;
}
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
} else { /* SK_YUKON_FAMILY */
Cleanup debugging code and put it under bootverbose (includes minor style polishing). Approved by: rwatson (mentor)
2005-01-29 19:26:53 +00:00
if (skrs == 0x00)
Set ramsize depending on a value from eprom instead of using hardcoded 128k for Yukon devices. 88E8001 only has 64k of on-chip RAM[1]. [1] http://www.marvell.com/products/pcconn/yukon/Yukon_88E8001_10_073103_final.pdf Tested by: amd64, current Approved by: rwatson (mentor) MFC after: 1 week
2004-11-13 20:21:57 +00:00
sc->sk_ramsize = 0x20000;
Cleanup debugging code and put it under bootverbose (includes minor style polishing). Approved by: rwatson (mentor)
2005-01-29 19:26:53 +00:00
else
Set ramsize depending on a value from eprom instead of using hardcoded 128k for Yukon devices. 88E8001 only has 64k of on-chip RAM[1]. [1] http://www.marvell.com/products/pcconn/yukon/Yukon_88E8001_10_073103_final.pdf Tested by: amd64, current Approved by: rwatson (mentor) MFC after: 1 week
2004-11-13 20:21:57 +00:00
sc->sk_ramsize = skrs * (1<<12);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc->sk_rboff = SK_RBOFF_0;
}
/* Read and save physical media type */
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
sc->sk_pmd = sk_win_read_1(sc, SK_PMDTYPE);
if (sc->sk_pmd == 'T' || sc->sk_pmd == '1')
sc->sk_coppertype = 1;
else
sc->sk_coppertype = 0;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
/* Determine whether to name it with VPD PN or just make it up.
* Marvell Yukon VPD PN seems to freqently be bogus. */
switch (pci_get_device(dev)) {
case DEVICEID_SK_V1:
case DEVICEID_BELKIN_5005:
case DEVICEID_3COM_3C940:
case DEVICEID_LINKSYS_EG1032:
Add device ID for second generation D-Link DGE-530T. PR: kern/99903
2006-07-27 05:06:43 +00:00
case DEVICEID_DLINK_DGE530T_A1:
case DEVICEID_DLINK_DGE530T_B1:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
/* Stay with VPD PN. */
Make a missing or empty VPD non-fatal, as it was prior to rev 1.131. With this change, my D-Link DGE-530T rev A1 is operational again.
2007-05-10 13:10:34 +00:00
(void) pci_get_vpd_ident(dev, &pname);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
break;
case DEVICEID_SK_V2:
/* YUKON VPD PN might bear no resemblance to reality. */
switch (sc->sk_type) {
case SK_GENESIS:
/* Stay with VPD PN. */
Make a missing or empty VPD non-fatal, as it was prior to rev 1.131. With this change, my D-Link DGE-530T rev A1 is operational again.
2007-05-10 13:10:34 +00:00
(void) pci_get_vpd_ident(dev, &pname);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
break;
case SK_YUKON:
pname = "Marvell Yukon Gigabit Ethernet";
break;
case SK_YUKON_LITE:
pname = "Marvell Yukon Lite Gigabit Ethernet";
break;
case SK_YUKON_LP:
pname = "Marvell Yukon LP Gigabit Ethernet";
break;
default:
pname = "Marvell Yukon (Unknown) Gigabit Ethernet";
break;
}
/* Yukon Lite Rev. A0 needs special test. */
if (sc->sk_type == SK_YUKON || sc->sk_type == SK_YUKON_LP) {
u_int32_t far;
u_int8_t testbyte;
/* Save flash address register before testing. */
far = sk_win_read_4(sc, SK_EP_ADDR);
sk_win_write_1(sc, SK_EP_ADDR+0x03, 0xff);
testbyte = sk_win_read_1(sc, SK_EP_ADDR+0x03);
if (testbyte != 0x00) {
/* Yukon Lite Rev. A0 detected. */
sc->sk_type = SK_YUKON_LITE;
sc->sk_rev = SK_YUKON_LITE_REV_A0;
/* Restore flash address register. */
sk_win_write_4(sc, SK_EP_ADDR, far);
}
}
break;
default:
device_printf(dev, "unknown device: vendor=%04x, device=%04x, "
"chipver=%02x, rev=%x\n",
pci_get_vendor(dev), pci_get_device(dev),
sc->sk_type, sc->sk_rev);
error = ENXIO;
goto fail;
}
if (sc->sk_type == SK_YUKON_LITE) {
switch (sc->sk_rev) {
case SK_YUKON_LITE_REV_A0:
revstr = "A0";
break;
case SK_YUKON_LITE_REV_A1:
revstr = "A1";
break;
case SK_YUKON_LITE_REV_A3:
revstr = "A3";
break;
default:
revstr = "";
break;
}
} else {
revstr = "";
}
* Correct an off-by-one reading vpd ro data. * Announce some more fields from ro area for better debugging of broken sk(4)s on various boards. Submitted by: Bjoern A. Zeeb <bzeeb-lists@lists.zabbadoz.net>
2004-11-01 17:21:04 +00:00
/* Announce the product name and more VPD data if there. */
Make a missing or empty VPD non-fatal, as it was prior to rev 1.131. With this change, my D-Link DGE-530T rev A1 is operational again.
2007-05-10 13:10:34 +00:00
if (pname != NULL)
device_printf(dev, "%s rev. %s(0x%x)\n",
pname, revstr, sc->sk_rev);
Cleanup debugging code and put it under bootverbose (includes minor style polishing). Approved by: rwatson (mentor)
2005-01-29 19:26:53 +00:00
if (bootverbose) {
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
device_printf(dev, "chip ver = 0x%02x\n", sc->sk_type);
device_printf(dev, "chip rev = 0x%02x\n", sc->sk_rev);
Cleanup debugging code and put it under bootverbose (includes minor style polishing). Approved by: rwatson (mentor)
2005-01-29 19:26:53 +00:00
device_printf(dev, "SK_EPROM0 = 0x%02x\n", skrs);
device_printf(dev, "SRAM size = 0x%06x\n", sc->sk_ramsize);
* Correct an off-by-one reading vpd ro data. * Announce some more fields from ro area for better debugging of broken sk(4)s on various boards. Submitted by: Bjoern A. Zeeb <bzeeb-lists@lists.zabbadoz.net>
2004-11-01 17:21:04 +00:00
}
Commit more debugging output. This is a little bit of using a large hammer, but sk(4) is so prevalent on AMD64 motherboards we need to reduce the number of round trips in the mailing lists trying to get sufficient information to make sure we've got a handle on all the problems and are working towards making sk(4) solid. Submitted by: bz
2004-11-14 20:17:01 +00:00
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc->sk_devs[SK_PORT_A] = device_add_child(dev, "sk", -1);
Deal with failed malloc calls[1]. While there also check for failed device_add_child calls. Found by: Coventry Analysis tool[1]. Submitted by: sam[1] Approved by: pjd (mentor) MFC after: 1 week
2005-04-25 10:18:24 +00:00
if (sc->sk_devs[SK_PORT_A] == NULL) {
device_printf(dev, "failed to add child for PORT_A\n");
error = ENXIO;
goto fail;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
port = malloc(sizeof(int), M_DEVBUF, M_NOWAIT);
Deal with failed malloc calls[1]. While there also check for failed device_add_child calls. Found by: Coventry Analysis tool[1]. Submitted by: sam[1] Approved by: pjd (mentor) MFC after: 1 week
2005-04-25 10:18:24 +00:00
if (port == NULL) {
device_printf(dev, "failed to allocate memory for "
"ivars of PORT_A\n");
error = ENXIO;
goto fail;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
*port = SK_PORT_A;
device_set_ivars(sc->sk_devs[SK_PORT_A], port);
Remove incomplete Yukon II support code which was added in if_sk.c, rev 1.123. Yukon II users should use msk(4).
2006-12-14 04:10:26 +00:00
if (!(sk_win_read_1(sc, SK_CONFIG) & SK_CONFIG_SINGLEMAC)) {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc->sk_devs[SK_PORT_B] = device_add_child(dev, "sk", -1);
Deal with failed malloc calls[1]. While there also check for failed device_add_child calls. Found by: Coventry Analysis tool[1]. Submitted by: sam[1] Approved by: pjd (mentor) MFC after: 1 week
2005-04-25 10:18:24 +00:00
if (sc->sk_devs[SK_PORT_B] == NULL) {
device_printf(dev, "failed to add child for PORT_B\n");
error = ENXIO;
goto fail;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
port = malloc(sizeof(int), M_DEVBUF, M_NOWAIT);
Deal with failed malloc calls[1]. While there also check for failed device_add_child calls. Found by: Coventry Analysis tool[1]. Submitted by: sam[1] Approved by: pjd (mentor) MFC after: 1 week
2005-04-25 10:18:24 +00:00
if (port == NULL) {
device_printf(dev, "failed to allocate memory for "
"ivars of PORT_B\n");
error = ENXIO;
goto fail;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
*port = SK_PORT_B;
device_set_ivars(sc->sk_devs[SK_PORT_B], port);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Turn on the 'driver is loaded' LED. */
CSR_WRITE_2(sc, SK_LED, SK_LED_GREEN_ON);
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
error = bus_generic_attach(dev);
if (error) {
device_printf(dev, "failed to attach port(s)\n");
goto fail;
}
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
/* Hook interrupt last to avoid having to lock softc */
Convert sk(4) to use the new bus_alloc_resources() API and bus_{read,write}_* macros. Submitted by: Antoine Brodin <antoine.brodin AT laposte DOT net> Reviewed by: imp (initial version)
2006-07-19 04:12:59 +00:00
error = bus_setup_intr(dev, sc->sk_res[1], INTR_TYPE_NET|INTR_MPSAFE,
o break newbus api: add a new argument of type driver_filter_t to bus_setup_intr() o add an int return code to all fast handlers o retire INTR_FAST/IH_FAST For more info: http://docs.freebsd.org/cgi/getmsg.cgi?fetch=465712+0+current/freebsd-current Reviewed by: many Approved by: re@
2007-02-23 12:19:07 +00:00
NULL, sk_intr, sc, &sc->sk_intrhand);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (error) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(dev, "couldn't set up irq\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
}
Add call to bus_generic_attach() at the end of sk_attach(). It turns out that if you kldload this driver, all the subordinate devices are probed/attached as expected. But this is not the case when the driver is statically compiled into the kernel. Since I do most of my testing with modules, I failed to notice this. I'm not sure if it's intended behavior or not. I think it may be, but it seems a little counter-intuitive.
2000-08-02 18:19:00 +00:00
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
fail:
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (error)
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
skc_detach(dev);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
return(error);
}
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
/*
* Shutdown hardware and free up resources. This can be called any
* time after the mutex has been initialized. It is called in both
* the error case in attach and the normal detach case so it needs
* to be careful about only freeing resources that have actually been
* allocated.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_detach(dev)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
device_t dev;
{
struct sk_if_softc *sc_if;
struct ifnet *ifp;
sc_if = device_get_softc(dev);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
KASSERT(mtx_initialized(&sc_if->sk_softc->sk_mtx),
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
("sk mutex not initialized in sk_detach"));
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_IF_LOCK(sc_if);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
/* These should only be active if attach_xmac succeeded */
Use newly minted device_is_attached rather than device_is_alive to see if attach succeeded. device_is_alive just tells us that probe succeeded. Since we were using it to do things like detach net interfaces, this caused problems when there were errors in the attach routine. Symptoms of problem reported by: martin blapp
2003-04-21 18:34:04 +00:00
if (device_is_attached(dev)) {
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
sk_stop(sc_if);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
/* Can't hold locks while calling detach */
SK_IF_UNLOCK(sc_if);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_drain(&sc_if->sk_tick_ch);
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
callout_drain(&sc_if->sk_watchdog_ch);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
ether_ifdetach(ifp);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
SK_IF_LOCK(sc_if);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
}
Fix "struct ifnet" leaks when attach() fails in the middle, e.g. when mii_phy_probe() or bus_setup_intr() fails. For drivers that call their detach() in this case, call if_free() there to cover this case too.
2005-09-16 11:11:51 +00:00
if (ifp)
if_free(ifp);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
/*
* We're generally called from skc_detach() which is using
* device_delete_child() to get to here. It's already trashed
* miibus for us, so don't do it here or we'll panic.
*/
/*
if (sc_if->sk_miibus != NULL)
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
device_delete_child(dev, sc_if->sk_miibus);
fix LOR's in sk. Original patch from dwhite. This moves the memory allocation earlier on in sk_attach so we don't have to lock until a bit later. PR: 69752
2004-08-20 06:22:04 +00:00
*/
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
bus_generic_detach(dev);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
sk_dma_jumbo_free(sc_if);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sk_dma_free(sc_if);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_IF_UNLOCK(sc_if);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
return(0);
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
skc_detach(dev)
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
device_t dev;
{
struct sk_softc *sc;
sc = device_get_softc(dev);
Fix KASSERT syntax errors. Please compile LINT before commiting.
2003-04-01 08:10:21 +00:00
KASSERT(mtx_initialized(&sc->sk_mtx), ("sk mutex not initialized"));
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (device_is_alive(dev)) {
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
if (sc->sk_devs[SK_PORT_A] != NULL) {
free(device_get_ivars(sc->sk_devs[SK_PORT_A]), M_DEVBUF);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
device_delete_child(dev, sc->sk_devs[SK_PORT_A]);
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
}
if (sc->sk_devs[SK_PORT_B] != NULL) {
free(device_get_ivars(sc->sk_devs[SK_PORT_B]), M_DEVBUF);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
device_delete_child(dev, sc->sk_devs[SK_PORT_B]);
Do not try to free non allocated memory in error case. Do our best to plug some memory leaks (VPD data, jumbo memory buffer,...). Log if we cannot free because memory still in use[1]. Change locking to avoid ''acquiring duplicate lock of same type: "network driver"'' and potential deadlock. Also seems to fix LOR #063. [1] This change does not solve problems if buffers are still in use when unloading if_sk.ko. There is ongoing work which will address jumbogram allocations in a more general way. PR: kern/75677 (with changes, no mii fixes in here) Tested by: net, Antoine Brodin (slightly different version) Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:27:22 +00:00
}
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
bus_generic_detach(dev);
}
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (sc->sk_intrhand)
Convert sk(4) to use the new bus_alloc_resources() API and bus_{read,write}_* macros. Submitted by: Antoine Brodin <antoine.brodin AT laposte DOT net> Reviewed by: imp (initial version)
2006-07-19 04:12:59 +00:00
bus_teardown_intr(dev, sc->sk_res[1], sc->sk_intrhand);
bus_release_resources(dev, sc->sk_res_spec, sc->sk_res);
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
mtx_destroy(&sc->sk_mii_mtx);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
mtx_destroy(&sc->sk_mtx);
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
return(0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
static bus_dma_tag_t
skc_get_dma_tag(device_t bus, device_t child __unused)
{
return (bus_get_dma_tag(bus));
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_dmamap_arg {
bus_addr_t sk_busaddr;
};
static void
sk_dmamap_cb(arg, segs, nseg, error)
void *arg;
bus_dma_segment_t *segs;
int nseg;
int error;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_dmamap_arg *ctx;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (error != 0)
return;
fix the missing lock in sk_jfree (verified w/ an assert) also fix up handling and proding of the tx, _OACTIVE is now handled better... Submitted by: Peter Edwards (sk_jfree) Obtained from: OpenBSD and/or NetBSD (tx prod)
2004-11-15 19:37:21 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
ctx = arg;
ctx->sk_busaddr = segs[0].ds_addr;
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/*
* Allocate jumbo buffer storage. The SysKonnect adapters support
* "jumbograms" (9K frames), although SysKonnect doesn't currently
* use them in their drivers. In order for us to use them, we need
* large 9K receive buffers, however standard mbuf clusters are only
* 2048 bytes in size. Consequently, we need to allocate and manage
* our own jumbo buffer pool. Fortunately, this does not require an
* excessive amount of additional code.
*/
static int
sk_dma_alloc(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_dmamap_arg ctx;
struct sk_txdesc *txd;
struct sk_rxdesc *rxd;
int error, i;
/* create parent tag */
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
* XXX
* This driver should use BUS_SPACE_MAXADDR for lowaddr argument
* in bus_dma_tag_create(9) as the NIC would support DAC mode.
* However bz@ reported that it does not work on amd64 with > 4GB
* RAM. Until we have more clues of the breakage, disable DAC mode
* by limiting DMA address to be in 32bit address space.
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
*/
Change the remainder of the drivers for DMA'ing devices enabled in the sparc64 GENERIC and the sound device drivers known working on sparc64 to use bus_get_dma_tag() to obtain the parent DMA tag so we can get rid of the sparc64_root_dma_tag kludge eventually. Except for ath(4), sk(4), stge(4) and ti(4) these changes are runtime tested (unless I booted up the wrong kernels again...).
2007-01-21 19:32:51 +00:00
error = bus_dma_tag_create(
bus_get_dma_tag(sc_if->sk_if_dev),/* parent */
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
BUS_SPACE_MAXSIZE_32BIT, /* maxsize */
0, /* nsegments */
BUS_SPACE_MAXSIZE_32BIT, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_parent_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create parent DMA tag\n");
goto fail;
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* create tag for Tx ring */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
SK_RING_ALIGN, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
SK_TX_RING_SZ, /* maxsize */
1, /* nsegments */
SK_TX_RING_SZ, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_tx_ring_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate Tx ring DMA tag\n");
goto fail;
}
/* create tag for Rx ring */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
SK_RING_ALIGN, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
SK_RX_RING_SZ, /* maxsize */
1, /* nsegments */
SK_RX_RING_SZ, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_rx_ring_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate Rx ring DMA tag\n");
goto fail;
}
/* create tag for Tx buffers */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
MCLBYTES * SK_MAXTXSEGS, /* maxsize */
SK_MAXTXSEGS, /* nsegments */
MCLBYTES, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_tx_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate Tx DMA tag\n");
goto fail;
}
/* create tag for Rx buffers */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
MCLBYTES, /* maxsize */
1, /* nsegments */
MCLBYTES, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_rx_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate Rx DMA tag\n");
goto fail;
}
/* allocate DMA'able memory and load the DMA map for Tx ring */
error = bus_dmamem_alloc(sc_if->sk_cdata.sk_tx_ring_tag,
Allocate the DMA memory shared between the host and the controller as coherent. MFC after: 2 weeks
2011-03-11 22:21:12 +00:00
(void **)&sc_if->sk_rdata.sk_tx_ring, BUS_DMA_NOWAIT |
BUS_DMA_COHERENT | BUS_DMA_ZERO, &sc_if->sk_cdata.sk_tx_ring_map);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate DMA'able memory for Tx ring\n");
goto fail;
}
ctx.sk_busaddr = 0;
error = bus_dmamap_load(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map, sc_if->sk_rdata.sk_tx_ring,
SK_TX_RING_SZ, sk_dmamap_cb, &ctx, BUS_DMA_NOWAIT);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to load DMA'able memory for Tx ring\n");
goto fail;
}
sc_if->sk_rdata.sk_tx_ring_paddr = ctx.sk_busaddr;
/* allocate DMA'able memory and load the DMA map for Rx ring */
error = bus_dmamem_alloc(sc_if->sk_cdata.sk_rx_ring_tag,
Allocate the DMA memory shared between the host and the controller as coherent. MFC after: 2 weeks
2011-03-11 22:21:12 +00:00
(void **)&sc_if->sk_rdata.sk_rx_ring, BUS_DMA_NOWAIT |
BUS_DMA_COHERENT | BUS_DMA_ZERO, &sc_if->sk_cdata.sk_rx_ring_map);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate DMA'able memory for Rx ring\n");
goto fail;
}
ctx.sk_busaddr = 0;
error = bus_dmamap_load(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_cdata.sk_rx_ring_map, sc_if->sk_rdata.sk_rx_ring,
SK_RX_RING_SZ, sk_dmamap_cb, &ctx, BUS_DMA_NOWAIT);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to load DMA'able memory for Rx ring\n");
goto fail;
}
sc_if->sk_rdata.sk_rx_ring_paddr = ctx.sk_busaddr;
/* create DMA maps for Tx buffers */
for (i = 0; i < SK_TX_RING_CNT; i++) {
txd = &sc_if->sk_cdata.sk_txdesc[i];
txd->tx_m = NULL;
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
txd->tx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = bus_dmamap_create(sc_if->sk_cdata.sk_tx_tag, 0,
&txd->tx_dmamap);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create Tx dmamap\n");
goto fail;
}
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* create DMA maps for Rx buffers */
if ((error = bus_dmamap_create(sc_if->sk_cdata.sk_rx_tag, 0,
&sc_if->sk_cdata.sk_rx_sparemap)) != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create spare Rx dmamap\n");
goto fail;
}
for (i = 0; i < SK_RX_RING_CNT; i++) {
rxd = &sc_if->sk_cdata.sk_rxdesc[i];
rxd->rx_m = NULL;
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
rxd->rx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = bus_dmamap_create(sc_if->sk_cdata.sk_rx_tag, 0,
&rxd->rx_dmamap);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create Rx dmamap\n");
goto fail;
}
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
fail:
return (error);
}
static int
sk_dma_jumbo_alloc(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_dmamap_arg ctx;
struct sk_rxdesc *jrxd;
int error, i;
if (jumbo_disable != 0) {
device_printf(sc_if->sk_if_dev, "disabling jumbo frame support\n");
sc_if->sk_jumbo_disable = 1;
return (0);
}
/* create tag for jumbo Rx ring */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
SK_RING_ALIGN, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
SK_JUMBO_RX_RING_SZ, /* maxsize */
1, /* nsegments */
SK_JUMBO_RX_RING_SZ, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_jumbo_rx_ring_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate jumbo Rx ring DMA tag\n");
goto jumbo_fail;
}
/* create tag for jumbo Rx buffers */
error = bus_dma_tag_create(sc_if->sk_cdata.sk_parent_tag,/* parent */
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
MJUM9BYTES, /* maxsize */
1, /* nsegments */
MJUM9BYTES, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc_if->sk_cdata.sk_jumbo_rx_tag);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate jumbo Rx DMA tag\n");
goto jumbo_fail;
}
/* allocate DMA'able memory and load the DMA map for jumbo Rx ring */
error = bus_dmamem_alloc(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
Allocate the DMA memory shared between the host and the controller as coherent. MFC after: 2 weeks
2011-03-11 22:21:12 +00:00
(void **)&sc_if->sk_rdata.sk_jumbo_rx_ring, BUS_DMA_NOWAIT |
BUS_DMA_COHERENT | BUS_DMA_ZERO,
&sc_if->sk_cdata.sk_jumbo_rx_ring_map);
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to allocate DMA'able memory for jumbo Rx ring\n");
goto jumbo_fail;
}
ctx.sk_busaddr = 0;
error = bus_dmamap_load(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_cdata.sk_jumbo_rx_ring_map,
sc_if->sk_rdata.sk_jumbo_rx_ring, SK_JUMBO_RX_RING_SZ, sk_dmamap_cb,
&ctx, BUS_DMA_NOWAIT);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to load DMA'able memory for jumbo Rx ring\n");
goto jumbo_fail;
}
sc_if->sk_rdata.sk_jumbo_rx_ring_paddr = ctx.sk_busaddr;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* create DMA maps for jumbo Rx buffers */
if ((error = bus_dmamap_create(sc_if->sk_cdata.sk_jumbo_rx_tag, 0,
&sc_if->sk_cdata.sk_jumbo_rx_sparemap)) != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create spare jumbo Rx dmamap\n");
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
goto jumbo_fail;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
for (i = 0; i < SK_JUMBO_RX_RING_CNT; i++) {
jrxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[i];
jrxd->rx_m = NULL;
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
jrxd->rx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = bus_dmamap_create(sc_if->sk_cdata.sk_jumbo_rx_tag, 0,
&jrxd->rx_dmamap);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"failed to create jumbo Rx dmamap\n");
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
goto jumbo_fail;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
jumbo_fail:
sk_dma_jumbo_free(sc_if);
device_printf(sc_if->sk_if_dev, "disabling jumbo frame support due to "
"resource shortage\n");
sc_if->sk_jumbo_disable = 1;
return (0);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
static void
sk_dma_free(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_txdesc *txd;
struct sk_rxdesc *rxd;
int i;
/* Tx ring */
if (sc_if->sk_cdata.sk_tx_ring_tag) {
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_tx_ring_paddr)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_unload(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map);
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_tx_ring)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamem_free(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_rdata.sk_tx_ring,
sc_if->sk_cdata.sk_tx_ring_map);
sc_if->sk_rdata.sk_tx_ring = NULL;
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
sc_if->sk_rdata.sk_tx_ring_paddr = 0;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dma_tag_destroy(sc_if->sk_cdata.sk_tx_ring_tag);
sc_if->sk_cdata.sk_tx_ring_tag = NULL;
}
/* Rx ring */
if (sc_if->sk_cdata.sk_rx_ring_tag) {
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_rx_ring_paddr)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_unload(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_cdata.sk_rx_ring_map);
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_rx_ring)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamem_free(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_rdata.sk_rx_ring,
sc_if->sk_cdata.sk_rx_ring_map);
sc_if->sk_rdata.sk_rx_ring = NULL;
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
sc_if->sk_rdata.sk_rx_ring_paddr = 0;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dma_tag_destroy(sc_if->sk_cdata.sk_rx_ring_tag);
sc_if->sk_cdata.sk_rx_ring_tag = NULL;
}
/* Tx buffers */
if (sc_if->sk_cdata.sk_tx_tag) {
for (i = 0; i < SK_TX_RING_CNT; i++) {
txd = &sc_if->sk_cdata.sk_txdesc[i];
if (txd->tx_dmamap) {
bus_dmamap_destroy(sc_if->sk_cdata.sk_tx_tag,
txd->tx_dmamap);
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
txd->tx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
}
bus_dma_tag_destroy(sc_if->sk_cdata.sk_tx_tag);
sc_if->sk_cdata.sk_tx_tag = NULL;
}
/* Rx buffers */
if (sc_if->sk_cdata.sk_rx_tag) {
for (i = 0; i < SK_RX_RING_CNT; i++) {
rxd = &sc_if->sk_cdata.sk_rxdesc[i];
if (rxd->rx_dmamap) {
bus_dmamap_destroy(sc_if->sk_cdata.sk_rx_tag,
rxd->rx_dmamap);
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
rxd->rx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
}
if (sc_if->sk_cdata.sk_rx_sparemap) {
bus_dmamap_destroy(sc_if->sk_cdata.sk_rx_tag,
sc_if->sk_cdata.sk_rx_sparemap);
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
sc_if->sk_cdata.sk_rx_sparemap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
bus_dma_tag_destroy(sc_if->sk_cdata.sk_rx_tag);
sc_if->sk_cdata.sk_rx_tag = NULL;
}
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
if (sc_if->sk_cdata.sk_parent_tag) {
bus_dma_tag_destroy(sc_if->sk_cdata.sk_parent_tag);
sc_if->sk_cdata.sk_parent_tag = NULL;
}
}
static void
sk_dma_jumbo_free(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_rxdesc *jrxd;
int i;
/* jumbo Rx ring */
if (sc_if->sk_cdata.sk_jumbo_rx_ring_tag) {
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_jumbo_rx_ring_paddr)
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
bus_dmamap_unload(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_cdata.sk_jumbo_rx_ring_map);
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
if (sc_if->sk_rdata.sk_jumbo_rx_ring)
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
bus_dmamem_free(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_rdata.sk_jumbo_rx_ring,
sc_if->sk_cdata.sk_jumbo_rx_ring_map);
sc_if->sk_rdata.sk_jumbo_rx_ring = NULL;
Fix various NIC drivers to properly cleanup static DMA resources. In particular, don't check the value of the bus_dma map against NULL to determine if either bus_dmamem_alloc() or bus_dmamap_load() succeeded. Instead, assume that bus_dmamap_load() succeeeded (and thus that bus_dmamap_unload() should be called) if the bus address for a resource is non-zero, and assume that bus_dmamem_alloc() succeeded (and thus that bus_dmamem_free() should be called) if the virtual address for a resource is not NULL. In many cases these bugs could result in leaks when a driver was detached. Reviewed by: yongari MFC after: 2 weeks
2014-06-11 14:53:58 +00:00
sc_if->sk_rdata.sk_jumbo_rx_ring_paddr = 0;
Nuke local jumbo allocator and switch to use of UMA backed page allocator for jumbo frame. o Removed unneeded jlist lock which was used to manage jumbo buffers. o Don't reinitialize hardware if MTU was not changed. o Added additional check for minimal MTU size. o Added a new tunable hw.skc.jumbo_disable to disable jumbo frame support for the driver. The tunable could be set for systems that do not need to use jumbo frames and it would save (9K * number of Rx descriptors) bytes kernel memory. o Jumbo buffer allocation failure is no longer critical error for the operation of sk(4). If sk(4) encounter the allocation failure it just disables jumbo frame support and continues to work without user intervention. With these changes jumbo frame performance of sk(4) was slightly increased and users should not encounter jumbo buffer allocation failure. Previously sk(4) tried to allocate physically contiguous memory, 3388KB for 256 Rx descriptors. Sometimes that amount of contiguous memory region could not be available for running systems which in turn resulted in failure of loading the driver. Tested by: Cy Schubert < Cy.Schubert () komquats dot com >
2008-02-14 01:10:48 +00:00
bus_dma_tag_destroy(sc_if->sk_cdata.sk_jumbo_rx_ring_tag);
sc_if->sk_cdata.sk_jumbo_rx_ring_tag = NULL;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* jumbo Rx buffers */
if (sc_if->sk_cdata.sk_jumbo_rx_tag) {
for (i = 0; i < SK_JUMBO_RX_RING_CNT; i++) {
jrxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[i];
if (jrxd->rx_dmamap) {
bus_dmamap_destroy(
sc_if->sk_cdata.sk_jumbo_rx_tag,
jrxd->rx_dmamap);
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
jrxd->rx_dmamap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
}
if (sc_if->sk_cdata.sk_jumbo_rx_sparemap) {
bus_dmamap_destroy(sc_if->sk_cdata.sk_jumbo_rx_tag,
sc_if->sk_cdata.sk_jumbo_rx_sparemap);
Prefer NULL over integer 0 for pointer type.
2008-02-14 01:25:01 +00:00
sc_if->sk_cdata.sk_jumbo_rx_sparemap = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
}
bus_dma_tag_destroy(sc_if->sk_cdata.sk_jumbo_rx_tag);
sc_if->sk_cdata.sk_jumbo_rx_tag = NULL;
}
}
static void
sk_txcksum(ifp, m, f)
struct ifnet *ifp;
struct mbuf *m;
struct sk_tx_desc *f;
{
struct ip *ip;
u_int16_t offset;
u_int8_t *p;
offset = sizeof(struct ip) + ETHER_HDR_LEN;
for(; m && m->m_len == 0; m = m->m_next)
;
if (m == NULL || m->m_len < ETHER_HDR_LEN) {
if_printf(ifp, "%s: m_len < ETHER_HDR_LEN\n", __func__);
/* checksum may be corrupted */
goto sendit;
}
if (m->m_len < ETHER_HDR_LEN + sizeof(u_int32_t)) {
if (m->m_len != ETHER_HDR_LEN) {
if_printf(ifp, "%s: m_len != ETHER_HDR_LEN\n",
__func__);
/* checksum may be corrupted */
goto sendit;
}
for(m = m->m_next; m && m->m_len == 0; m = m->m_next)
;
if (m == NULL) {
offset = sizeof(struct ip) + ETHER_HDR_LEN;
/* checksum may be corrupted */
goto sendit;
}
ip = mtod(m, struct ip *);
} else {
p = mtod(m, u_int8_t *);
p += ETHER_HDR_LEN;
ip = (struct ip *)p;
}
offset = (ip->ip_hl << 2) + ETHER_HDR_LEN;
sendit:
f->sk_csum_startval = 0;
f->sk_csum_start = htole32(((offset + m->m_pkthdr.csum_data) & 0xffff) |
(offset << 16));
}
static int
sk_encap(sc_if, m_head)
struct sk_if_softc *sc_if;
struct mbuf **m_head;
{
struct sk_txdesc *txd;
struct sk_tx_desc *f = NULL;
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
struct mbuf *m;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dma_segment_t txsegs[SK_MAXTXSEGS];
u_int32_t cflags, frag, si, sk_ctl;
int error, i, nseg;
SK_IF_LOCK_ASSERT(sc_if);
if ((txd = STAILQ_FIRST(&sc_if->sk_cdata.sk_txfreeq)) == NULL)
return (ENOBUFS);
error = bus_dmamap_load_mbuf_sg(sc_if->sk_cdata.sk_tx_tag,
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
txd->tx_dmamap, *m_head, txsegs, &nseg, 0);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (error == EFBIG) {
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m = m_defrag(*m_head, M_NOWAIT);
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
if (m == NULL) {
m_freem(*m_head);
*m_head = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (ENOMEM);
}
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
*m_head = m;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
error = bus_dmamap_load_mbuf_sg(sc_if->sk_cdata.sk_tx_tag,
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
txd->tx_dmamap, *m_head, txsegs, &nseg, 0);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (error != 0) {
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
m_freem(*m_head);
*m_head = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (error);
}
} else if (error != 0)
return (error);
if (nseg == 0) {
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
m_freem(*m_head);
*m_head = NULL;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
return (EIO);
}
if (sc_if->sk_cdata.sk_tx_cnt + nseg >= SK_TX_RING_CNT) {
bus_dmamap_unload(sc_if->sk_cdata.sk_tx_tag, txd->tx_dmamap);
return (ENOBUFS);
}
Fix invalid reference of mbuf chains. Use proper pointer dereference to inform modified mbuf chains to caller. In collaboration with: glebius
2006-08-12 01:24:37 +00:00
m = *m_head;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if ((m->m_pkthdr.csum_flags & sc_if->sk_ifp->if_hwassist) != 0)
cflags = SK_OPCODE_CSUM;
else
cflags = SK_OPCODE_DEFAULT;
si = frag = sc_if->sk_cdata.sk_tx_prod;
for (i = 0; i < nseg; i++) {
f = &sc_if->sk_rdata.sk_tx_ring[frag];
f->sk_data_lo = htole32(SK_ADDR_LO(txsegs[i].ds_addr));
f->sk_data_hi = htole32(SK_ADDR_HI(txsegs[i].ds_addr));
sk_ctl = txsegs[i].ds_len | cflags;
if (i == 0) {
if (cflags == SK_OPCODE_CSUM)
sk_txcksum(sc_if->sk_ifp, m, f);
sk_ctl |= SK_TXCTL_FIRSTFRAG;
} else
sk_ctl |= SK_TXCTL_OWN;
f->sk_ctl = htole32(sk_ctl);
sc_if->sk_cdata.sk_tx_cnt++;
SK_INC(frag, SK_TX_RING_CNT);
}
sc_if->sk_cdata.sk_tx_prod = frag;
/* set EOF on the last desciptor */
frag = (frag + SK_TX_RING_CNT - 1) % SK_TX_RING_CNT;
f = &sc_if->sk_rdata.sk_tx_ring[frag];
f->sk_ctl |= htole32(SK_TXCTL_LASTFRAG | SK_TXCTL_EOF_INTR);
/* turn the first descriptor ownership to NIC */
f = &sc_if->sk_rdata.sk_tx_ring[si];
f->sk_ctl |= htole32(SK_TXCTL_OWN);
STAILQ_REMOVE_HEAD(&sc_if->sk_cdata.sk_txfreeq, tx_q);
STAILQ_INSERT_TAIL(&sc_if->sk_cdata.sk_txbusyq, txd, tx_q);
txd->tx_m = m;
/* sync descriptors */
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_tag, txd->tx_dmamap,
BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_start(ifp)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp;
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
{
struct sk_if_softc *sc_if;
sc_if = ifp->if_softc;
SK_IF_LOCK(sc_if);
sk_start_locked(ifp);
SK_IF_UNLOCK(sc_if);
return;
}
static void
sk_start_locked(ifp)
struct ifnet *ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
struct sk_softc *sc;
struct sk_if_softc *sc_if;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct mbuf *m_head;
int enq;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc_if = ifp->if_softc;
sc = sc_if->sk_softc;
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK_ASSERT(sc_if);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
for (enq = 0; !IFQ_DRV_IS_EMPTY(&ifp->if_snd) &&
sc_if->sk_cdata.sk_tx_cnt < SK_TX_RING_CNT - 1; ) {
Enable sk(4) for ATLQ. Thoroughly tested by: Ender <ender NO tog SPAM net> MFC after: 4 weeks
2004-12-24 14:13:38 +00:00
IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (m_head == NULL)
break;
/*
* Pack the data into the transmit ring. If we
* don't have room, set the OACTIVE flag and wait
* for the NIC to drain the ring.
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sk_encap(sc_if, &m_head)) {
if (m_head == NULL)
break;
Enable sk(4) for ATLQ. Thoroughly tested by: Ender <ender NO tog SPAM net> MFC after: 4 weeks
2004-12-24 14:13:38 +00:00
IFQ_DRV_PREPEND(&ifp->if_snd, m_head);
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags |= IFF_DRV_OACTIVE;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
enq++;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
BPF_MTAP(ifp, m_head);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (enq > 0) {
/* Transmit */
fix the missing lock in sk_jfree (verified w/ an assert) also fix up handling and proding of the tx, _OACTIVE is now handled better... Submitted by: Peter Edwards (sk_jfree) Obtained from: OpenBSD and/or NetBSD (tx prod)
2004-11-15 19:37:21 +00:00
CSR_WRITE_4(sc, sc_if->sk_tx_bmu, SK_TXBMU_TX_START);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
fix the missing lock in sk_jfree (verified w/ an assert) also fix up handling and proding of the tx, _OACTIVE is now handled better... Submitted by: Peter Edwards (sk_jfree) Obtained from: OpenBSD and/or NetBSD (tx prod)
2004-11-15 19:37:21 +00:00
/* Set a timeout in case the chip goes out to lunch. */
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
sc_if->sk_watchdog_timer = 5;
fix the missing lock in sk_jfree (verified w/ an assert) also fix up handling and proding of the tx, _OACTIVE is now handled better... Submitted by: Peter Edwards (sk_jfree) Obtained from: OpenBSD and/or NetBSD (tx prod)
2004-11-15 19:37:21 +00:00
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
sk_watchdog(arg)
void *arg;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
struct sk_if_softc *sc_if;
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
struct ifnet *ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
ifp = arg;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc_if = ifp->if_softc;
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
SK_IF_LOCK_ASSERT(sc_if);
if (sc_if->sk_watchdog_timer == 0 || --sc_if->sk_watchdog_timer)
goto done;
Fix watchdog timeout errors seen on a few systems. SK-NET GENESIS document says reading SK_ISSR should stop generating further interrupts(Since we drop a driver lock before invoking ifp->if_input handler we should disable interrupts in ISR in order to protect integrity of softc from subsequent interrupts). But it seems that there is possibility of loosing interrupts between reading SK_ISSR and determining which interrupts are reported. To cope with the situation we continuously read SK_ISSR register until there are no interrupts. However, it seems that the above work around doesn't fix all cases. To protect watchdog handler from triggering false alarm add a work around code which try to reclaim pending Tx descriptors before resetting hardware. This should fix occasional watchdog timeout errors seen on this driver. Reported by: Frank Behrens <frank AT pinky dot sax dot de > Tested by: Frank Behrens <frank AT pinky dot sax dot de >
2006-06-07 09:05:20 +00:00
/*
Fix typo in comment. Submitted by: brad AT OpenBSD DOT org
2006-07-05 04:56:50 +00:00
* Reclaim first as there is a possibility of losing Tx completion
* interrupts.
Fix watchdog timeout errors seen on a few systems. SK-NET GENESIS document says reading SK_ISSR should stop generating further interrupts(Since we drop a driver lock before invoking ifp->if_input handler we should disable interrupts in ISR in order to protect integrity of softc from subsequent interrupts). But it seems that there is possibility of loosing interrupts between reading SK_ISSR and determining which interrupts are reported. To cope with the situation we continuously read SK_ISSR register until there are no interrupts. However, it seems that the above work around doesn't fix all cases. To protect watchdog handler from triggering false alarm add a work around code which try to reclaim pending Tx descriptors before resetting hardware. This should fix occasional watchdog timeout errors seen on this driver. Reported by: Frank Behrens <frank AT pinky dot sax dot de > Tested by: Frank Behrens <frank AT pinky dot sax dot de >
2006-06-07 09:05:20 +00:00
*/
sk_txeof(sc_if);
if (sc_if->sk_cdata.sk_tx_cnt != 0) {
if_printf(sc_if->sk_ifp, "watchdog timeout\n");
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
Fix watchdog timeout errors seen on a few systems. SK-NET GENESIS document says reading SK_ISSR should stop generating further interrupts(Since we drop a driver lock before invoking ifp->if_input handler we should disable interrupts in ISR in order to protect integrity of softc from subsequent interrupts). But it seems that there is possibility of loosing interrupts between reading SK_ISSR and determining which interrupts are reported. To cope with the situation we continuously read SK_ISSR register until there are no interrupts. However, it seems that the above work around doesn't fix all cases. To protect watchdog handler from triggering false alarm add a work around code which try to reclaim pending Tx descriptors before resetting hardware. This should fix occasional watchdog timeout errors seen on this driver. Reported by: Frank Behrens <frank AT pinky dot sax dot de > Tested by: Frank Behrens <frank AT pinky dot sax dot de >
2006-06-07 09:05:20 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
sk_init_locked(sc_if);
}
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
done:
callout_reset(&sc_if->sk_watchdog_ch, hz, sk_watchdog, ifp);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
static int
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
skc_shutdown(dev)
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
device_t dev;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
{
struct sk_softc *sc;
Convert the SysKonnect gigabit ethernet driver to newbus.
1999-07-22 04:04:12 +00:00
sc = device_get_softc(dev);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_LOCK(sc);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Turn off the 'driver is loaded' LED. */
CSR_WRITE_2(sc, SK_LED, SK_LED_GREEN_OFF);
/*
* Reset the GEnesis controller. Doing this should also
* assert the resets on the attached XMAC(s).
*/
sk_reset(sc);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_UNLOCK(sc);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
return (0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static int
skc_suspend(dev)
device_t dev;
{
struct sk_softc *sc;
struct sk_if_softc *sc_if0, *sc_if1;
struct ifnet *ifp0 = NULL, *ifp1 = NULL;
sc = device_get_softc(dev);
SK_LOCK(sc);
sc_if0 = sc->sk_if[SK_PORT_A];
sc_if1 = sc->sk_if[SK_PORT_B];
if (sc_if0 != NULL)
ifp0 = sc_if0->sk_ifp;
if (sc_if1 != NULL)
ifp1 = sc_if1->sk_ifp;
if (ifp0 != NULL)
sk_stop(sc_if0);
if (ifp1 != NULL)
sk_stop(sc_if1);
sc->sk_suspended = 1;
SK_UNLOCK(sc);
return (0);
}
static int
skc_resume(dev)
device_t dev;
{
struct sk_softc *sc;
struct sk_if_softc *sc_if0, *sc_if1;
struct ifnet *ifp0 = NULL, *ifp1 = NULL;
sc = device_get_softc(dev);
SK_LOCK(sc);
sc_if0 = sc->sk_if[SK_PORT_A];
sc_if1 = sc->sk_if[SK_PORT_B];
if (sc_if0 != NULL)
ifp0 = sc_if0->sk_ifp;
if (sc_if1 != NULL)
ifp1 = sc_if1->sk_ifp;
if (ifp0 != NULL && ifp0->if_flags & IFF_UP)
sk_init_locked(sc_if0);
if (ifp1 != NULL && ifp1->if_flags & IFF_UP)
sk_init_locked(sc_if1);
sc->sk_suspended = 0;
SK_UNLOCK(sc);
return (0);
}
/*
* According to the data sheet from SK-NET GENESIS the hardware can compute
* two Rx checksums at the same time(Each checksum start position is
* programmed in Rx descriptors). However it seems that TCP/UDP checksum
* does not work at least on my Yukon hardware. I tried every possible ways
* to get correct checksum value but couldn't get correct one. So TCP/UDP
* checksum offload was disabled at the moment and only IP checksum offload
* was enabled.
* As nomral IP header size is 20 bytes I can't expect it would give an
* increase in throughput. However it seems it doesn't hurt performance in
* my testing. If there is a more detailed information for checksum secret
* of the hardware in question please contact yongari@FreeBSD.org to add
* TCP/UDP checksum offload support.
*/
static __inline void
sk_rxcksum(ifp, m, csum)
struct ifnet *ifp;
struct mbuf *m;
u_int32_t csum;
{
struct ether_header *eh;
struct ip *ip;
int32_t hlen, len, pktlen;
u_int16_t csum1, csum2, ipcsum;
pktlen = m->m_pkthdr.len;
if (pktlen < sizeof(struct ether_header) + sizeof(struct ip))
return;
eh = mtod(m, struct ether_header *);
if (eh->ether_type != htons(ETHERTYPE_IP))
return;
ip = (struct ip *)(eh + 1);
if (ip->ip_v != IPVERSION)
return;
hlen = ip->ip_hl << 2;
pktlen -= sizeof(struct ether_header);
if (hlen < sizeof(struct ip))
return;
if (ntohs(ip->ip_len) < hlen)
return;
if (ntohs(ip->ip_len) != pktlen)
return;
csum1 = htons(csum & 0xffff);
csum2 = htons((csum >> 16) & 0xffff);
ipcsum = in_addword(csum1, ~csum2 & 0xffff);
/* checksum fixup for IP options */
len = hlen - sizeof(struct ip);
if (len > 0) {
/*
* If the second checksum value is correct we can compute IP
* checksum with simple math. Unfortunately the second checksum
* value is wrong so we can't verify the checksum from the
* value(It seems there is some magic here to get correct
* value). If the second checksum value is correct it also
* means we can get TCP/UDP checksum) here. However, it still
* needs pseudo header checksum calculation due to hardware
* limitations.
*/
return;
}
m->m_pkthdr.csum_flags = CSUM_IP_CHECKED;
if (ipcsum == 0xffff)
m->m_pkthdr.csum_flags |= CSUM_IP_VALID;
}
static __inline int
sk_rxvalid(sc, stat, len)
struct sk_softc *sc;
u_int32_t stat, len;
{
if (sc->sk_type == SK_GENESIS) {
if ((stat & XM_RXSTAT_ERRFRAME) == XM_RXSTAT_ERRFRAME ||
XM_RXSTAT_BYTES(stat) != len)
return (0);
} else {
if ((stat & (YU_RXSTAT_CRCERR | YU_RXSTAT_LONGERR |
YU_RXSTAT_MIIERR | YU_RXSTAT_BADFC | YU_RXSTAT_GOODFC |
YU_RXSTAT_JABBER)) != 0 ||
(stat & YU_RXSTAT_RXOK) != YU_RXSTAT_RXOK ||
YU_RXSTAT_BYTES(stat) != len)
return (0);
}
return (1);
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_rxeof(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Drop the driver lock around calls to if_input to avoid a LOR when the packets are immediately returned for sending (e.g. when bridging or packet forwarding). There are more efficient ways to do this but for now use the least intrusive approach. Reviewed by: imp, rwatson
2003-11-14 19:00:32 +00:00
struct sk_softc *sc;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct mbuf *m;
struct ifnet *ifp;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_rx_desc *cur_rx;
struct sk_rxdesc *rxd;
int cons, prog;
u_int32_t csum, rxstat, sk_ctl;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Drop the driver lock around calls to if_input to avoid a LOR when the packets are immediately returned for sending (e.g. when bridging or packet forwarding). There are more efficient ways to do this but for now use the least intrusive approach. Reviewed by: imp, rwatson
2003-11-14 19:00:32 +00:00
sc = sc_if->sk_softc;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK_ASSERT(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_cdata.sk_rx_ring_map, BUS_DMASYNC_POSTREAD);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
prog = 0;
for (cons = sc_if->sk_cdata.sk_rx_cons; prog < SK_RX_RING_CNT;
prog++, SK_INC(cons, SK_RX_RING_CNT)) {
cur_rx = &sc_if->sk_rdata.sk_rx_ring[cons];
sk_ctl = le32toh(cur_rx->sk_ctl);
if ((sk_ctl & SK_RXCTL_OWN) != 0)
break;
rxd = &sc_if->sk_cdata.sk_rxdesc[cons];
rxstat = le32toh(cur_rx->sk_xmac_rxstat);
if ((sk_ctl & (SK_RXCTL_STATUS_VALID | SK_RXCTL_FIRSTFRAG |
SK_RXCTL_LASTFRAG)) != (SK_RXCTL_STATUS_VALID |
SK_RXCTL_FIRSTFRAG | SK_RXCTL_LASTFRAG) ||
SK_RXBYTES(sk_ctl) < SK_MIN_FRAMELEN ||
SK_RXBYTES(sk_ctl) > SK_MAX_FRAMELEN ||
sk_rxvalid(sc, rxstat, SK_RXBYTES(sk_ctl)) == 0) {
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sk_discard_rxbuf(sc_if, cons);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
continue;
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
m = rxd->rx_m;
csum = le32toh(cur_rx->sk_csum);
if (sk_newbuf(sc_if, cons) != 0) {
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* reuse old buffer */
sk_discard_rxbuf(sc_if, cons);
continue;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
m->m_pkthdr.rcvif = ifp;
m->m_pkthdr.len = m->m_len = SK_RXBYTES(sk_ctl);
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if ((ifp->if_capenable & IFCAP_RXCSUM) != 0)
sk_rxcksum(ifp, m, csum);
SK_IF_UNLOCK(sc_if);
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
(*ifp->if_input)(ifp, m);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (prog > 0) {
sc_if->sk_cdata.sk_rx_cons = cons;
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_ring_tag,
sc_if->sk_cdata.sk_rx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
}
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void
sk_jumbo_rxeof(sc_if)
struct sk_if_softc *sc_if;
{
struct sk_softc *sc;
struct mbuf *m;
struct ifnet *ifp;
struct sk_rx_desc *cur_rx;
struct sk_rxdesc *jrxd;
int cons, prog;
u_int32_t csum, rxstat, sk_ctl;
sc = sc_if->sk_softc;
ifp = sc_if->sk_ifp;
SK_IF_LOCK_ASSERT(sc_if);
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_cdata.sk_jumbo_rx_ring_map, BUS_DMASYNC_POSTREAD);
prog = 0;
for (cons = sc_if->sk_cdata.sk_jumbo_rx_cons;
prog < SK_JUMBO_RX_RING_CNT;
prog++, SK_INC(cons, SK_JUMBO_RX_RING_CNT)) {
cur_rx = &sc_if->sk_rdata.sk_jumbo_rx_ring[cons];
sk_ctl = le32toh(cur_rx->sk_ctl);
if ((sk_ctl & SK_RXCTL_OWN) != 0)
break;
jrxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[cons];
rxstat = le32toh(cur_rx->sk_xmac_rxstat);
if ((sk_ctl & (SK_RXCTL_STATUS_VALID | SK_RXCTL_FIRSTFRAG |
SK_RXCTL_LASTFRAG)) != (SK_RXCTL_STATUS_VALID |
SK_RXCTL_FIRSTFRAG | SK_RXCTL_LASTFRAG) ||
SK_RXBYTES(sk_ctl) < SK_MIN_FRAMELEN ||
SK_RXBYTES(sk_ctl) > SK_JUMBO_FRAMELEN ||
sk_rxvalid(sc, rxstat, SK_RXBYTES(sk_ctl)) == 0) {
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
sk_discard_jumbo_rxbuf(sc_if, cons);
continue;
}
m = jrxd->rx_m;
csum = le32toh(cur_rx->sk_csum);
if (sk_jumbo_newbuf(sc_if, cons) != 0) {
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* reuse old buffer */
sk_discard_jumbo_rxbuf(sc_if, cons);
continue;
}
m->m_pkthdr.rcvif = ifp;
m->m_pkthdr.len = m->m_len = SK_RXBYTES(sk_ctl);
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if ((ifp->if_capenable & IFCAP_RXCSUM) != 0)
sk_rxcksum(ifp, m, csum);
SK_IF_UNLOCK(sc_if);
(*ifp->if_input)(ifp, m);
SK_IF_LOCK(sc_if);
}
if (prog > 0) {
sc_if->sk_cdata.sk_jumbo_rx_cons = cons;
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_ring_tag,
sc_if->sk_cdata.sk_jumbo_rx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_txeof(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_txdesc *txd;
only clear the IFF_OACTIVE flag when we have a chance of being able to queue a packet to the hardware... instead of when the hardware queue is empty.. don't initalize cur_tx now that it doesn't need to be... Pointed out by: bde
2004-11-17 21:35:22 +00:00
struct sk_tx_desc *cur_tx;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int32_t idx, sk_ctl;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
txd = STAILQ_FIRST(&sc_if->sk_cdata.sk_txbusyq);
if (txd == NULL)
return;
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map, BUS_DMASYNC_POSTREAD);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Go through our tx ring and free mbufs for those
* frames that have been sent.
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
for (idx = sc_if->sk_cdata.sk_tx_cons;; SK_INC(idx, SK_TX_RING_CNT)) {
if (sc_if->sk_cdata.sk_tx_cnt <= 0)
break;
cur_tx = &sc_if->sk_rdata.sk_tx_ring[idx];
sk_ctl = le32toh(cur_tx->sk_ctl);
if (sk_ctl & SK_TXCTL_OWN)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
break;
sc_if->sk_cdata.sk_tx_cnt--;
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if ((sk_ctl & SK_TXCTL_LASTFRAG) == 0)
continue;
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_tag, txd->tx_dmamap,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc_if->sk_cdata.sk_tx_tag, txd->tx_dmamap);
Mechanically convert to if_inc_counter().
2014-09-19 03:51:26 +00:00
if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
m_freem(txd->tx_m);
txd->tx_m = NULL;
STAILQ_REMOVE_HEAD(&sc_if->sk_cdata.sk_txbusyq, tx_q);
STAILQ_INSERT_TAIL(&sc_if->sk_cdata.sk_txfreeq, txd, tx_q);
txd = STAILQ_FIRST(&sc_if->sk_cdata.sk_txbusyq);
}
only clear the IFF_OACTIVE flag when we have a chance of being able to queue a packet to the hardware... instead of when the hardware queue is empty.. don't initalize cur_tx now that it doesn't need to be... Pointed out by: bde
2004-11-17 21:35:22 +00:00
sc_if->sk_cdata.sk_tx_cons = idx;
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
sc_if->sk_watchdog_timer = sc_if->sk_cdata.sk_tx_cnt > 0 ? 5 : 0;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_ring_tag,
sc_if->sk_cdata.sk_tx_ring_map,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_tick(xsc_if)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
void *xsc_if;
{
struct sk_if_softc *sc_if;
struct mii_data *mii;
struct ifnet *ifp;
int i;
sc_if = xsc_if;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii = device_get_softc(sc_if->sk_miibus);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (!(ifp->if_flags & IFF_UP))
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
return;
if (sc_if->sk_phytype == SK_PHYTYPE_BCOM) {
sk_intr_bcom(sc_if);
return;
}
/*
* According to SysKonnect, the correct way to verify that
* the link has come back up is to poll bit 0 of the GPIO
* register three times. This pin has the signal from the
* link_sync pin connected to it; if we read the same link
* state 3 times in a row, we know the link is up.
*/
for (i = 0; i < 3; i++) {
if (SK_XM_READ_2(sc_if, XM_GPIO) & XM_GPIO_GP0_SET)
break;
}
if (i != 3) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_reset(&sc_if->sk_tick_ch, hz, sk_tick, sc_if);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
return;
}
/* Turn the GP0 interrupt back on. */
SK_XM_CLRBIT_2(sc_if, XM_IMR, XM_IMR_GP0_SET);
SK_XM_READ_2(sc_if, XM_ISR);
mii_tick(mii);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_stop(&sc_if->sk_tick_ch);
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
static void
sk_yukon_tick(xsc_if)
void *xsc_if;
{
struct sk_if_softc *sc_if;
struct mii_data *mii;
sc_if = xsc_if;
mii = device_get_softc(sc_if->sk_miibus);
mii_tick(mii);
callout_reset(&sc_if->sk_tick_ch, hz, sk_yukon_tick, sc_if);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_intr_bcom(sc_if)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct sk_if_softc *sc_if;
{
struct mii_data *mii;
struct ifnet *ifp;
int status;
mii = device_get_softc(sc_if->sk_miibus);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
SK_XM_CLRBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_TX_ENB|XM_MMUCMD_RX_ENB);
/*
* Read the PHY interrupt register to make sure
* we clear any pending interrupts.
*/
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
status = sk_xmac_miibus_readreg(sc_if, SK_PHYADDR_BCOM, BRGPHY_MII_ISR);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sk_init_xmac(sc_if);
return;
}
if (status & (BRGPHY_ISR_LNK_CHG|BRGPHY_ISR_AN_PR)) {
int lstat;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
lstat = sk_xmac_miibus_readreg(sc_if, SK_PHYADDR_BCOM,
BRGPHY_MII_AUXSTS);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
if (!(lstat & BRGPHY_AUXSTS_LINK) && sc_if->sk_link) {
mii_mediachg(mii);
/* Turn off the link LED. */
- Call mii_pollstat() after we bring up the link on a 1000baseTX card after autoneg so we make sure to set the link state and duplex mode correctly. - Make sure to set the 'ignore pause frames' bit on the XMAC. - Small linewrap fix.
2000-06-06 02:56:37 +00:00
SK_IF_WRITE_1(sc_if, 0,
SK_LINKLED1_CTL, SK_LINKLED_OFF);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc_if->sk_link = 0;
} else if (status & BRGPHY_ISR_LNK_CHG) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_writereg(sc_if, SK_PHYADDR_BCOM,
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
BRGPHY_MII_IMR, 0xFF00);
mii_tick(mii);
sc_if->sk_link = 1;
/* Turn on the link LED. */
SK_IF_WRITE_1(sc_if, 0, SK_LINKLED1_CTL,
SK_LINKLED_ON|SK_LINKLED_LINKSYNC_OFF|
SK_LINKLED_BLINK_OFF);
} else {
mii_tick(mii);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_reset(&sc_if->sk_tick_ch, hz, sk_tick, sc_if);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
}
}
SK_XM_SETBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_TX_ENB|XM_MMUCMD_RX_ENB);
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_intr_xmac(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
struct sk_softc *sc;
u_int16_t status;
sc = sc_if->sk_softc;
status = SK_XM_READ_2(sc_if, XM_ISR);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/*
* Link has gone down. Start MII tick timeout to
* watch for link resync.
*/
if (sc_if->sk_phytype == SK_PHYTYPE_XMAC) {
if (status & XM_ISR_GP0_SET) {
SK_XM_SETBIT_2(sc_if, XM_IMR, XM_IMR_GP0_SET);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_reset(&sc_if->sk_tick_ch, hz, sk_tick, sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
if (status & XM_ISR_AUTONEG_DONE) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_reset(&sc_if->sk_tick_ch, hz, sk_tick, sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
}
if (status & XM_IMR_TX_UNDERRUN)
SK_XM_SETBIT_4(sc_if, XM_MODE, XM_MODE_FLUSH_TXFIFO);
if (status & XM_IMR_RX_OVERRUN)
SK_XM_SETBIT_4(sc_if, XM_MODE, XM_MODE_FLUSH_RXFIFO);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
status = SK_XM_READ_2(sc_if, XM_ISR);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
static void
sk_intr_yukon(sc_if)
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int8_t status;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
status = SK_IF_READ_1(sc_if, 0, SK_GMAC_ISR);
/* RX overrun */
if ((status & SK_GMAC_INT_RX_OVER) != 0) {
SK_IF_WRITE_1(sc_if, 0, SK_RXMF1_CTRL_TEST,
SK_RFCTL_RX_FIFO_OVER);
}
/* TX underrun */
if ((status & SK_GMAC_INT_TX_UNDER) != 0) {
SK_IF_WRITE_1(sc_if, 0, SK_RXMF1_CTRL_TEST,
SK_TFCTL_TX_FIFO_UNDER);
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_intr(xsc)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
void *xsc;
{
struct sk_softc *sc = xsc;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_if_softc *sc_if0, *sc_if1;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct ifnet *ifp0 = NULL, *ifp1 = NULL;
u_int32_t status;
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_LOCK(sc);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
status = CSR_READ_4(sc, SK_ISSR);
if (status == 0 || status == 0xffffffff || sc->sk_suspended)
goto done_locked;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc_if0 = sc->sk_if[SK_PORT_A];
sc_if1 = sc->sk_if[SK_PORT_B];
if (sc_if0 != NULL)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp0 = sc_if0->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
if (sc_if1 != NULL)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp1 = sc_if1->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Fix watchdog timeout errors seen on a few systems. SK-NET GENESIS document says reading SK_ISSR should stop generating further interrupts(Since we drop a driver lock before invoking ifp->if_input handler we should disable interrupts in ISR in order to protect integrity of softc from subsequent interrupts). But it seems that there is possibility of loosing interrupts between reading SK_ISSR and determining which interrupts are reported. To cope with the situation we continuously read SK_ISSR register until there are no interrupts. However, it seems that the above work around doesn't fix all cases. To protect watchdog handler from triggering false alarm add a work around code which try to reclaim pending Tx descriptors before resetting hardware. This should fix occasional watchdog timeout errors seen on this driver. Reported by: Frank Behrens <frank AT pinky dot sax dot de > Tested by: Frank Behrens <frank AT pinky dot sax dot de >
2006-06-07 09:05:20 +00:00
for (; (status &= sc->sk_intrmask) != 0;) {
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Handle receive interrupts first. */
if (status & SK_ISR_RX1_EOF) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp0->if_mtu > SK_MAX_FRAMELEN)
sk_jumbo_rxeof(sc_if0);
else
sk_rxeof(sc_if0);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_BMU_RX_CSR0,
SK_RXBMU_CLR_IRQ_EOF|SK_RXBMU_RX_START);
}
if (status & SK_ISR_RX2_EOF) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp1->if_mtu > SK_MAX_FRAMELEN)
sk_jumbo_rxeof(sc_if1);
else
sk_rxeof(sc_if1);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_BMU_RX_CSR1,
SK_RXBMU_CLR_IRQ_EOF|SK_RXBMU_RX_START);
}
/* Then transmit interrupts. */
if (status & SK_ISR_TX1_S_EOF) {
sk_txeof(sc_if0);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
CSR_WRITE_4(sc, SK_BMU_TXS_CSR0, SK_TXBMU_CLR_IRQ_EOF);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
if (status & SK_ISR_TX2_S_EOF) {
sk_txeof(sc_if1);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
CSR_WRITE_4(sc, SK_BMU_TXS_CSR1, SK_TXBMU_CLR_IRQ_EOF);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
/* Then MAC interrupts. */
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (status & SK_ISR_MAC1 &&
ifp0->if_drv_flags & IFF_DRV_RUNNING) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sc->sk_type == SK_GENESIS)
sk_intr_xmac(sc_if0);
else
sk_intr_yukon(sc_if0);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (status & SK_ISR_MAC2 &&
ifp1->if_drv_flags & IFF_DRV_RUNNING) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sc->sk_type == SK_GENESIS)
sk_intr_xmac(sc_if1);
else
correct typo in interrupt handling for the 2nd port of 2-port cards Submitted by: luigi Reviewed by: checking original openbsd code Approved by: re (scottl)
2003-11-29 19:33:01 +00:00
sk_intr_yukon(sc_if1);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
if (status & SK_ISR_EXTERNAL_REG) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (ifp0 != NULL &&
sc_if0->sk_phytype == SK_PHYTYPE_BCOM)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sk_intr_bcom(sc_if0);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (ifp1 != NULL &&
sc_if1->sk_phytype == SK_PHYTYPE_BCOM)
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sk_intr_bcom(sc_if1);
}
Fix watchdog timeout errors seen on a few systems. SK-NET GENESIS document says reading SK_ISSR should stop generating further interrupts(Since we drop a driver lock before invoking ifp->if_input handler we should disable interrupts in ISR in order to protect integrity of softc from subsequent interrupts). But it seems that there is possibility of loosing interrupts between reading SK_ISSR and determining which interrupts are reported. To cope with the situation we continuously read SK_ISSR register until there are no interrupts. However, it seems that the above work around doesn't fix all cases. To protect watchdog handler from triggering false alarm add a work around code which try to reclaim pending Tx descriptors before resetting hardware. This should fix occasional watchdog timeout errors seen on this driver. Reported by: Frank Behrens <frank AT pinky dot sax dot de > Tested by: Frank Behrens <frank AT pinky dot sax dot de >
2006-06-07 09:05:20 +00:00
status = CSR_READ_4(sc, SK_ISSR);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
CSR_WRITE_4(sc, SK_IMR, sc->sk_intrmask);
Enable sk(4) for ATLQ. Thoroughly tested by: Ender <ender NO tog SPAM net> MFC after: 4 weeks
2004-12-24 14:13:38 +00:00
if (ifp0 != NULL && !IFQ_DRV_IS_EMPTY(&ifp0->if_snd))
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
sk_start_locked(ifp0);
Enable sk(4) for ATLQ. Thoroughly tested by: Ender <ender NO tog SPAM net> MFC after: 4 weeks
2004-12-24 14:13:38 +00:00
if (ifp1 != NULL && !IFQ_DRV_IS_EMPTY(&ifp1->if_snd))
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
sk_start_locked(ifp1);
Call sk_start() at the end of sk_intr() if there's packets in the interface send queue that need to be processed.
1999-09-25 04:50:27 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
done_locked:
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
SK_UNLOCK(sc);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_init_xmac(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
struct sk_softc *sc;
struct ifnet *ifp;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int16_t eaddr[(ETHER_ADDR_LEN+1)/2];
- Merge from r249476: Ensure that PCI bus BUS_GET_DMA_TAG() method sees the actual PCI device which makes the request for DMA tag, instead of some descendant of the PCI device, by creating a pass-through trampoline. - Sprinkle const on tables. - Use NULL instead of 0 for pointers. - Take advantage of nitems(). MFC after: 1 week
2013-05-30 12:16:55 +00:00
static const struct sk_bcom_hack bhack[] = {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
{ 0x18, 0x0c20 }, { 0x17, 0x0012 }, { 0x15, 0x1104 }, { 0x17, 0x0013 },
{ 0x15, 0x0404 }, { 0x17, 0x8006 }, { 0x15, 0x0132 }, { 0x17, 0x8006 },
{ 0x15, 0x0232 }, { 0x17, 0x800D }, { 0x15, 0x000F }, { 0x18, 0x0420 },
{ 0, 0 } };
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK_ASSERT(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc = sc_if->sk_softc;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Unreset the XMAC. */
SK_IF_WRITE_2(sc_if, 0, SK_TXF1_MACCTL, SK_TXMACCTL_XMAC_UNRESET);
DELAY(1000);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/* Reset the XMAC's internal state. */
SK_XM_SETBIT_2(sc_if, XM_GPIO, XM_GPIO_RESETMAC);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Save the XMAC II revision */
sc_if->sk_xmac_rev = XM_XMAC_REV(SK_XM_READ_4(sc_if, XM_DEVID));
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/*
* Perform additional initialization for external PHYs,
* namely for the 1000baseTX cards that use the XMAC's
* GMII mode.
*/
if (sc_if->sk_phytype == SK_PHYTYPE_BCOM) {
int i = 0;
u_int32_t val;
/* Take PHY out of reset. */
val = sk_win_read_4(sc, SK_GPIO);
if (sc_if->sk_port == SK_PORT_A)
val |= SK_GPIO_DIR0|SK_GPIO_DAT0;
else
val |= SK_GPIO_DIR2|SK_GPIO_DAT2;
sk_win_write_4(sc, SK_GPIO, val);
/* Enable GMII mode on the XMAC. */
SK_XM_SETBIT_2(sc_if, XM_HWCFG, XM_HWCFG_GMIIMODE);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_writereg(sc_if, SK_PHYADDR_BCOM,
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
BRGPHY_MII_BMCR, BRGPHY_BMCR_RESET);
DELAY(10000);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_writereg(sc_if, SK_PHYADDR_BCOM,
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
BRGPHY_MII_IMR, 0xFFF0);
/*
* Early versions of the BCM5400 apparently have
* a bug that requires them to have their reserved
* registers initialized to some magic values. I don't
* know what the numbers do, I'm just the messenger.
*/
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sk_xmac_miibus_readreg(sc_if, SK_PHYADDR_BCOM, 0x03)
== 0x6041) {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
while(bhack[i].reg) {
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_xmac_miibus_writereg(sc_if, SK_PHYADDR_BCOM,
bhack[i].reg, bhack[i].val);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
i++;
}
}
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Set station address */
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
bcopy(IF_LLADDR(sc_if->sk_ifp), eaddr, ETHER_ADDR_LEN);
SK_XM_WRITE_2(sc_if, XM_PAR0, eaddr[0]);
SK_XM_WRITE_2(sc_if, XM_PAR1, eaddr[1]);
SK_XM_WRITE_2(sc_if, XM_PAR2, eaddr[2]);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_XM_SETBIT_4(sc_if, XM_MODE, XM_MODE_RX_USE_STATION);
if (ifp->if_flags & IFF_BROADCAST) {
SK_XM_CLRBIT_4(sc_if, XM_MODE, XM_MODE_RX_NOBROAD);
} else {
SK_XM_SETBIT_4(sc_if, XM_MODE, XM_MODE_RX_NOBROAD);
}
/* We don't need the FCS appended to the packet. */
SK_XM_SETBIT_2(sc_if, XM_RXCMD, XM_RXCMD_STRIPFCS);
/* We want short frames padded to 60 bytes. */
SK_XM_SETBIT_2(sc_if, XM_TXCMD, XM_TXCMD_AUTOPAD);
/*
* Enable the reception of all error frames. This is is
* a necessary evil due to the design of the XMAC. The
* XMAC's receive FIFO is only 8K in size, however jumbo
* frames can be up to 9000 bytes in length. When bad
* frame filtering is enabled, the XMAC's RX FIFO operates
* in 'store and forward' mode. For this to work, the
* entire frame has to fit into the FIFO, but that means
* that jumbo frames larger than 8192 bytes will be
* truncated. Disabling all bad frame filtering causes
* the RX FIFO to operate in streaming mode, in which
* case the XMAC will start transfering frames out of the
* RX FIFO as soon as the FIFO threshold is reached.
*/
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp->if_mtu > SK_MAX_FRAMELEN) {
SK_XM_SETBIT_4(sc_if, XM_MODE, XM_MODE_RX_BADFRAMES|
XM_MODE_RX_GIANTS|XM_MODE_RX_RUNTS|XM_MODE_RX_CRCERRS|
XM_MODE_RX_INRANGELEN);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_XM_SETBIT_2(sc_if, XM_RXCMD, XM_RXCMD_BIGPKTOK);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
} else
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_XM_CLRBIT_2(sc_if, XM_RXCMD, XM_RXCMD_BIGPKTOK);
/*
* Bump up the transmit threshold. This helps hold off transmit
* underruns when we're blasting traffic from both ports at once.
*/
SK_XM_WRITE_2(sc_if, XM_TX_REQTHRESH, SK_XM_TX_FIFOTHRESH);
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
/* Set Rx filter */
sk_rxfilter_genesis(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Clear and enable interrupts */
SK_XM_READ_2(sc_if, XM_ISR);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
if (sc_if->sk_phytype == SK_PHYTYPE_XMAC)
SK_XM_WRITE_2(sc_if, XM_IMR, XM_INTRS);
else
SK_XM_WRITE_2(sc_if, XM_IMR, 0xFFFF);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Configure MAC arbiter */
switch(sc_if->sk_xmac_rev) {
case XM_XMAC_REV_B2:
sk_win_write_1(sc, SK_RCINIT_RX1, SK_RCINIT_XMAC_B2);
sk_win_write_1(sc, SK_RCINIT_TX1, SK_RCINIT_XMAC_B2);
sk_win_write_1(sc, SK_RCINIT_RX2, SK_RCINIT_XMAC_B2);
sk_win_write_1(sc, SK_RCINIT_TX2, SK_RCINIT_XMAC_B2);
sk_win_write_1(sc, SK_MINIT_RX1, SK_MINIT_XMAC_B2);
sk_win_write_1(sc, SK_MINIT_TX1, SK_MINIT_XMAC_B2);
sk_win_write_1(sc, SK_MINIT_RX2, SK_MINIT_XMAC_B2);
sk_win_write_1(sc, SK_MINIT_TX2, SK_MINIT_XMAC_B2);
sk_win_write_1(sc, SK_RECOVERY_CTL, SK_RECOVERY_XMAC_B2);
break;
case XM_XMAC_REV_C1:
sk_win_write_1(sc, SK_RCINIT_RX1, SK_RCINIT_XMAC_C1);
sk_win_write_1(sc, SK_RCINIT_TX1, SK_RCINIT_XMAC_C1);
sk_win_write_1(sc, SK_RCINIT_RX2, SK_RCINIT_XMAC_C1);
sk_win_write_1(sc, SK_RCINIT_TX2, SK_RCINIT_XMAC_C1);
sk_win_write_1(sc, SK_MINIT_RX1, SK_MINIT_XMAC_C1);
sk_win_write_1(sc, SK_MINIT_TX1, SK_MINIT_XMAC_C1);
sk_win_write_1(sc, SK_MINIT_RX2, SK_MINIT_XMAC_C1);
sk_win_write_1(sc, SK_MINIT_TX2, SK_MINIT_XMAC_C1);
sk_win_write_1(sc, SK_RECOVERY_CTL, SK_RECOVERY_XMAC_B2);
break;
default:
break;
}
sk_win_write_2(sc, SK_MACARB_CTL,
SK_MACARBCTL_UNRESET|SK_MACARBCTL_FASTOE_OFF);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc_if->sk_link = 1;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
put function's name at begining of column...
2004-08-20 05:09:31 +00:00
static void
sk_init_yukon(sc_if)
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
struct sk_if_softc *sc_if;
{
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int32_t phy, v;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
u_int16_t reg;
Enable jumbo frames on Yukon variants of sk(4). In contrast to OpenBSD we enable jumbo frame support depending on MTU setting (like done for xmac). Approved by: pjd (mentor) Obtained from: OpenBSD if_sk.c r1.52 (YU_SMR_MFL_JUMBO flag) Tested by: Heinz Knocke <knockefreebsd at o2 dot pl> MFC after: 5 days
2004-12-20 10:07:12 +00:00
struct sk_softc *sc;
struct ifnet *ifp;
Fix a long standing bug in programming station address for Yukon controllers. sk(4) never reprogrammed station address for Yukon controllers so overriding station address with ifconfig(8) was not possible. Fix the bug by reprogramming all registers that control station address, flow-control and virtual station address. Virtual station address has no use at this moment since driver does not make use of fail over feature. Tested by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 22:07:22 +00:00
u_int8_t *eaddr;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
int i;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_LOCK_ASSERT(sc_if);
Enable jumbo frames on Yukon variants of sk(4). In contrast to OpenBSD we enable jumbo frame support depending on MTU setting (like done for xmac). Approved by: pjd (mentor) Obtained from: OpenBSD if_sk.c r1.52 (YU_SMR_MFL_JUMBO flag) Tested by: Heinz Knocke <knockefreebsd at o2 dot pl> MFC after: 5 days
2004-12-20 10:07:12 +00:00
sc = sc_if->sk_softc;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
Enable jumbo frames on Yukon variants of sk(4). In contrast to OpenBSD we enable jumbo frame support depending on MTU setting (like done for xmac). Approved by: pjd (mentor) Obtained from: OpenBSD if_sk.c r1.52 (YU_SMR_MFL_JUMBO flag) Tested by: Heinz Knocke <knockefreebsd at o2 dot pl> MFC after: 5 days
2004-12-20 10:07:12 +00:00
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
if (sc->sk_type == SK_YUKON_LITE &&
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
sc->sk_rev >= SK_YUKON_LITE_REV_A3) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/*
* Workaround code for COMA mode, set PHY reset.
* Otherwise it will not correctly take chip out of
* powerdown (coma)
*/
v = sk_win_read_4(sc, SK_GPIO);
v |= SK_GPIO_DIR9 | SK_GPIO_DAT9;
sk_win_write_4(sc, SK_GPIO, v);
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* GMAC and GPHY Reset */
SK_IF_WRITE_4(sc_if, 0, SK_GPHY_CTRL, SK_GPHY_RESET_SET);
SK_IF_WRITE_4(sc_if, 0, SK_GMAC_CTRL, SK_GMAC_RESET_SET);
DELAY(1000);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sc->sk_type == SK_YUKON_LITE &&
sc->sk_rev >= SK_YUKON_LITE_REV_A3) {
/*
* Workaround code for COMA mode, clear PHY reset
*/
v = sk_win_read_4(sc, SK_GPIO);
v |= SK_GPIO_DIR9;
v &= ~SK_GPIO_DAT9;
sk_win_write_4(sc, SK_GPIO, v);
}
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
phy = SK_GPHY_INT_POL_HI | SK_GPHY_DIS_FC | SK_GPHY_DIS_SLEEP |
SK_GPHY_ENA_XC | SK_GPHY_ANEG_ALL | SK_GPHY_ENA_PAUSE;
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
if (sc->sk_coppertype)
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
phy |= SK_GPHY_COPPER;
Re-wrok PHY setup, media handling and dual-port detection. With this change SysKonnect SK-9521 v2.0 and SK-9821 v2.0 adapter now works. Obtained from: OpenBSD Reported by: Ganbold ganbold ! micom ( mng $ net Tested by: Ganbold ganbold ! micom ( mng $ net
2006-05-15 04:50:33 +00:00
else
phy |= SK_GPHY_FIBER;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
SK_IF_WRITE_4(sc_if, 0, SK_GPHY_CTRL, phy | SK_GPHY_RESET_SET);
DELAY(1000);
SK_IF_WRITE_4(sc_if, 0, SK_GPHY_CTRL, phy | SK_GPHY_RESET_CLEAR);
SK_IF_WRITE_4(sc_if, 0, SK_GMAC_CTRL, SK_GMAC_LOOP_OFF |
SK_GMAC_PAUSE_ON | SK_GMAC_RESET_CLEAR);
/* unused read of the interrupt source register */
SK_IF_READ_2(sc_if, 0, SK_GMAC_ISR);
reg = SK_YU_READ_2(sc_if, YUKON_PAR);
/* MIB Counter Clear Mode set */
reg |= YU_PAR_MIB_CLR;
SK_YU_WRITE_2(sc_if, YUKON_PAR, reg);
/* MIB Counter Clear Mode clear */
reg &= ~YU_PAR_MIB_CLR;
SK_YU_WRITE_2(sc_if, YUKON_PAR, reg);
/* receive control reg */
Fix multicast and promiscuous mode handling for Yukon devices. Submitted by: Jung-uk Kim <jkim@niksun.com>
2004-01-21 22:32:15 +00:00
SK_YU_WRITE_2(sc_if, YUKON_RCR, YU_RCR_CRCR);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* transmit parameter register */
SK_YU_WRITE_2(sc_if, YUKON_TPR, YU_TPR_JAM_LEN(0x3) |
YU_TPR_JAM_IPG(0xb) | YU_TPR_JAM2DATA_IPG(0x1a) );
/* serial mode register */
Enable jumbo frames on Yukon variants of sk(4). In contrast to OpenBSD we enable jumbo frame support depending on MTU setting (like done for xmac). Approved by: pjd (mentor) Obtained from: OpenBSD if_sk.c r1.52 (YU_SMR_MFL_JUMBO flag) Tested by: Heinz Knocke <knockefreebsd at o2 dot pl> MFC after: 5 days
2004-12-20 10:07:12 +00:00
reg = YU_SMR_DATA_BLIND(0x1c) | YU_SMR_MFL_VLAN | YU_SMR_IPG_DATA(0x1e);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp->if_mtu > SK_MAX_FRAMELEN)
Enable jumbo frames on Yukon variants of sk(4). In contrast to OpenBSD we enable jumbo frame support depending on MTU setting (like done for xmac). Approved by: pjd (mentor) Obtained from: OpenBSD if_sk.c r1.52 (YU_SMR_MFL_JUMBO flag) Tested by: Heinz Knocke <knockefreebsd at o2 dot pl> MFC after: 5 days
2004-12-20 10:07:12 +00:00
reg |= YU_SMR_MFL_JUMBO;
SK_YU_WRITE_2(sc_if, YUKON_SMR, reg);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Fix a long standing bug in programming station address for Yukon controllers. sk(4) never reprogrammed station address for Yukon controllers so overriding station address with ifconfig(8) was not possible. Fix the bug by reprogramming all registers that control station address, flow-control and virtual station address. Virtual station address has no use at this moment since driver does not make use of fail over feature. Tested by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 22:07:22 +00:00
/* Setup Yukon's station address */
eaddr = IF_LLADDR(sc_if->sk_ifp);
for (i = 0; i < 3; i++)
SK_YU_WRITE_2(sc_if, SK_MAC0_0 + i * 4,
eaddr[i * 2] | eaddr[i * 2 + 1] << 8);
/* Set GMAC source address of flow control. */
for (i = 0; i < 3; i++)
remove trailing spaces
2006-01-17 05:41:20 +00:00
SK_YU_WRITE_2(sc_if, YUKON_SAL1 + i * 4,
Fix a long standing bug in programming station address for Yukon controllers. sk(4) never reprogrammed station address for Yukon controllers so overriding station address with ifconfig(8) was not possible. Fix the bug by reprogramming all registers that control station address, flow-control and virtual station address. Virtual station address has no use at this moment since driver does not make use of fail over feature. Tested by: "Mikhail T." <mi+thun <> aldan.algebra.com> MFC after: 1 week
2010-11-06 22:07:22 +00:00
eaddr[i * 2] | eaddr[i * 2 + 1] << 8);
/* Set GMAC virtual address. */
for (i = 0; i < 3; i++)
SK_YU_WRITE_2(sc_if, YUKON_SAL2 + i * 4,
eaddr[i * 2] | eaddr[i * 2 + 1] << 8);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
Separate multicast filtering of SysKonnect GENESIS and Marvell Yukon from common multicast handling code. Yukon uses hash-based multicast filtering(big endian form) but GENESIS uses perfect multicast filtering as well as hash-based one(little endian form). Due to the differences of multicast filtering there is no much sense to have a common code. o Remove sk_setmulti() and introduce sk_rxfilter_yukon(), sk_rxfilter_yukon() that handles multicast filtering setup. o Have sk_rxfilter_{yukon, genesis} handle promiscuous mode and nuke sk_setpromisc(). This simplifies ioctl handler as well as giving a chance to check validity of Rx control register of Yukon. o Don't reinitialize controller when IFF_ALLMULTI flags is changed. o Nuke sk_gmchash(), it's not needed anymore. o Always reconfigure Rx control register whenever a new multicast filtering condition is changed. This fixes multicast filtering setup on Yukon. PR: kern/134051
2009-05-01 03:24:03 +00:00
/* Set Rx filter */
sk_rxfilter_yukon(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* enable interrupt mask for counter overflows */
SK_YU_WRITE_2(sc_if, YUKON_TIMR, 0);
SK_YU_WRITE_2(sc_if, YUKON_RIMR, 0);
SK_YU_WRITE_2(sc_if, YUKON_TRIMR, 0);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* Configure RX MAC FIFO Flush Mask */
v = YU_RXSTAT_FOFL | YU_RXSTAT_CRCERR | YU_RXSTAT_MIIERR |
YU_RXSTAT_BADFC | YU_RXSTAT_GOODFC | YU_RXSTAT_RUNT |
YU_RXSTAT_JABBER;
SK_IF_WRITE_2(sc_if, 0, SK_RXMF1_FLUSH_MASK, v);
/* Disable RX MAC FIFO Flush for YUKON-Lite Rev. A0 only */
if (sc->sk_type == SK_YUKON_LITE && sc->sk_rev == SK_YUKON_LITE_REV_A0)
v = SK_TFCTL_OPERATION_ON;
else
v = SK_TFCTL_OPERATION_ON | SK_RFCTL_FIFO_FLUSH_ON;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* Configure RX MAC FIFO */
SK_IF_WRITE_1(sc_if, 0, SK_RXMF1_CTRL_TEST, SK_RFCTL_RESET_CLEAR);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_WRITE_2(sc_if, 0, SK_RXMF1_CTRL_TEST, v);
/* Increase flush threshould to 64 bytes */
SK_IF_WRITE_2(sc_if, 0, SK_RXMF1_FLUSH_THRESHOLD,
SK_RFCTL_FIFO_THRESHOLD + 1);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* Configure TX MAC FIFO */
SK_IF_WRITE_1(sc_if, 0, SK_TXMF1_CTRL_TEST, SK_TFCTL_RESET_CLEAR);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_IF_WRITE_2(sc_if, 0, SK_TXMF1_CTRL_TEST, SK_TFCTL_OPERATION_ON);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/*
* Note that to properly initialize any part of the GEnesis chip,
* you first have to take it out of reset mode.
*/
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_init(xsc)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
void *xsc;
{
struct sk_if_softc *sc_if = xsc;
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK(sc_if);
sk_init_locked(sc_if);
SK_IF_UNLOCK(sc_if);
return;
}
static void
sk_init_locked(sc_if)
struct sk_if_softc *sc_if;
{
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_softc *sc;
struct ifnet *ifp;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
struct mii_data *mii;
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
u_int16_t reg;
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
u_int32_t imr;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
int error;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK_ASSERT(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc = sc_if->sk_softc;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii = device_get_softc(sc_if->sk_miibus);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
* When adding/deleting multicast addresses, only whack the address filter if the interface is marked RUNNING. Obtained from: NetBSD if_sk.c rev. 1.12 * Don't initialize the card (and start an autonegotiation) every time the IP address changes. Makes 'dhclient sk0' invocations way faster and more consistant. i.e. one DHCPREQUEST elicits the DHCPACK. Obtained from: OpenBSD if_sk.c rev. 1.56 * Additional locking changes in sk_ioctl. PR: kern/61296 should see improvements by the last two. Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:18:58 +00:00
return;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Cancel pending I/O and free all RX/TX buffers. */
sk_stop(sc_if);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sc->sk_type == SK_GENESIS) {
/* Configure LINK_SYNC LED */
SK_IF_WRITE_1(sc_if, 0, SK_LINKLED1_CTL, SK_LINKLED_ON);
SK_IF_WRITE_1(sc_if, 0, SK_LINKLED1_CTL,
SK_LINKLED_LINKSYNC_ON);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* Configure RX LED */
remove trailing spaces
2006-01-17 05:41:20 +00:00
SK_IF_WRITE_1(sc_if, 0, SK_RXLED1_CTL,
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
SK_RXLEDCTL_COUNTER_START);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
/* Configure TX LED */
SK_IF_WRITE_1(sc_if, 0, SK_TXLED1_CTL,
SK_TXLEDCTL_COUNTER_START);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/*
* Configure descriptor poll timer
*
* SK-NET GENESIS data sheet says that possibility of losing Start
* transmit command due to CPU/cache related interim storage problems
* under certain conditions. The document recommends a polling
* mechanism to send a Start transmit command to initiate transfer
* of ready descriptors regulary. To cope with this issue sk(4) now
* enables descriptor poll timer to initiate descriptor processing
* periodically as defined by SK_DPT_TIMER_MAX. However sk(4) still
* issue SK_TXBMU_TX_START to Tx BMU to get fast execution of Tx
* command instead of waiting for next descriptor polling time.
* The same rule may apply to Rx side too but it seems that is not
* needed at the moment.
* Since sk(4) uses descriptor polling as a last resort there is no
* need to set smaller polling time than maximum allowable one.
*/
SK_IF_WRITE_4(sc_if, 0, SK_DPT_INIT, SK_DPT_TIMER_MAX);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Configure I2C registers */
/* Configure XMAC(s) */
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch (sc->sk_type) {
case SK_GENESIS:
sk_init_xmac(sc_if);
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
sk_init_yukon(sc_if);
break;
}
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
mii_mediachg(mii);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
if (sc->sk_type == SK_GENESIS) {
/* Configure MAC FIFOs */
SK_IF_WRITE_4(sc_if, 0, SK_RXF1_CTL, SK_FIFO_UNRESET);
SK_IF_WRITE_4(sc_if, 0, SK_RXF1_END, SK_FIFO_END);
SK_IF_WRITE_4(sc_if, 0, SK_RXF1_CTL, SK_FIFO_ON);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
SK_IF_WRITE_4(sc_if, 0, SK_TXF1_CTL, SK_FIFO_UNRESET);
SK_IF_WRITE_4(sc_if, 0, SK_TXF1_END, SK_FIFO_END);
SK_IF_WRITE_4(sc_if, 0, SK_TXF1_CTL, SK_FIFO_ON);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Configure transmit arbiter(s) */
SK_IF_WRITE_1(sc_if, 0, SK_TXAR1_COUNTERCTL,
SK_TXARCTL_ON|SK_TXARCTL_FSYNC_ON);
/* Configure RAMbuffers */
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_CTLTST, SK_RBCTL_UNRESET);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_START, sc_if->sk_rx_ramstart);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_WR_PTR, sc_if->sk_rx_ramstart);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_RD_PTR, sc_if->sk_rx_ramstart);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_END, sc_if->sk_rx_ramend);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_CTLTST, SK_RBCTL_ON);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_CTLTST, SK_RBCTL_UNRESET);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_CTLTST, SK_RBCTL_STORENFWD_ON);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_START, sc_if->sk_tx_ramstart);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_WR_PTR, sc_if->sk_tx_ramstart);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_RD_PTR, sc_if->sk_tx_ramstart);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_END, sc_if->sk_tx_ramend);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_CTLTST, SK_RBCTL_ON);
/* Configure BMUs */
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_BMU_CSR, SK_RXBMU_ONLINE);
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp->if_mtu > SK_MAX_FRAMELEN) {
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_CURADDR_LO,
SK_ADDR_LO(SK_JUMBO_RX_RING_ADDR(sc_if, 0)));
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_CURADDR_HI,
SK_ADDR_HI(SK_JUMBO_RX_RING_ADDR(sc_if, 0)));
} else {
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_CURADDR_LO,
SK_ADDR_LO(SK_RX_RING_ADDR(sc_if, 0)));
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_CURADDR_HI,
SK_ADDR_HI(SK_RX_RING_ADDR(sc_if, 0)));
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_IF_WRITE_4(sc_if, 1, SK_TXQS1_BMU_CSR, SK_TXBMU_ONLINE);
SK_IF_WRITE_4(sc_if, 1, SK_TXQS1_CURADDR_LO,
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
SK_ADDR_LO(SK_TX_RING_ADDR(sc_if, 0)));
SK_IF_WRITE_4(sc_if, 1, SK_TXQS1_CURADDR_HI,
SK_ADDR_HI(SK_TX_RING_ADDR(sc_if, 0)));
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Init descriptors */
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (ifp->if_mtu > SK_MAX_FRAMELEN)
error = sk_init_jumbo_rx_ring(sc_if);
else
error = sk_init_rx_ring(sc_if);
if (error != 0) {
device_printf(sc_if->sk_if_dev,
"initialization failed: no memory for rx buffers\n");
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sk_stop(sc_if);
return;
}
sk_init_tx_ring(sc_if);
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
/* Set interrupt moderation if changed via sysctl. */
imr = sk_win_read_4(sc, SK_IMTIMERINIT);
fix interrupt moderation timer frequencies for Yukon Obtained from: OpenBSD
2006-01-17 05:57:44 +00:00
if (imr != SK_IM_USECS(sc->sk_int_mod, sc->sk_int_ticks)) {
sk_win_write_4(sc, SK_IMTIMERINIT, SK_IM_USECS(sc->sk_int_mod,
sc->sk_int_ticks));
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
if (bootverbose)
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
device_printf(sc_if->sk_if_dev,
"interrupt moderation is %d us.\n",
sc->sk_int_mod);
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Configure interrupt handling */
CSR_READ_4(sc, SK_ISSR);
if (sc_if->sk_port == SK_PORT_A)
sc->sk_intrmask |= SK_INTRS1;
else
sc->sk_intrmask |= SK_INTRS2;
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
sc->sk_intrmask |= SK_ISR_EXTERNAL_REG;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
CSR_WRITE_4(sc, SK_IMR, sc->sk_intrmask);
/* Start BMUs. */
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_BMU_CSR, SK_RXBMU_RX_START);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch(sc->sk_type) {
case SK_GENESIS:
/* Enable XMACs TX and RX state machines */
SK_XM_CLRBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_IGNPAUSE);
SK_XM_SETBIT_2(sc_if, XM_MMUCMD, XM_MMUCMD_TX_ENB|XM_MMUCMD_RX_ENB);
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
reg = SK_YU_READ_2(sc_if, YUKON_GPCR);
reg |= YU_GPCR_TXEN | YU_GPCR_RXEN;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
#if 0
/* XXX disable 100Mbps and full duplex mode? */
reg &= ~(YU_GPCR_SPEED | YU_GPCR_DPLX_DIS);
#endif
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
SK_YU_WRITE_2(sc_if, YUKON_GPCR, reg);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* Activate descriptor polling timer */
SK_IF_WRITE_4(sc_if, 0, SK_DPT_TIMER_CTRL, SK_DPT_TCTL_START);
/* start transfer of Tx descriptors */
CSR_WRITE_4(sc, sc_if->sk_tx_bmu, SK_TXBMU_TX_START);
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags |= IFF_DRV_RUNNING;
ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
switch (sc->sk_type) {
case SK_YUKON:
case SK_YUKON_LITE:
case SK_YUKON_LP:
callout_reset(&sc_if->sk_tick_ch, hz, sk_yukon_tick, sc_if);
break;
}
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
callout_reset(&sc_if->sk_watchdog_ch, hz, sk_watchdog, ifp);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
style: put return values on a line by themselves. fix some paste issues where whitespace was used instead of tabs.
2002-08-23 23:49:02 +00:00
static void
sk_stop(sc_if)
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
struct sk_if_softc *sc_if;
{
int i;
struct sk_softc *sc;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
struct sk_txdesc *txd;
struct sk_rxdesc *rxd;
struct sk_rxdesc *jrxd;
Remember to clear the IFF_RUNNING and IFF_OACTIVE flags in sf_stop() and sk_stop().
1999-07-25 05:16:05 +00:00
struct ifnet *ifp;
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
u_int32_t val;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
SK_IF_LOCK_ASSERT(sc_if);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
sc = sc_if->sk_softc;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc_if->sk_ifp;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
callout_stop(&sc_if->sk_tick_ch);
Use our own timer for watchdog instead of if_watchdog/if_timer interface.
2007-04-02 04:43:41 +00:00
callout_stop(&sc_if->sk_watchdog_ch);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
/* stop Tx descriptor polling timer */
SK_IF_WRITE_4(sc_if, 0, SK_DPT_TIMER_CTRL, SK_DPT_TCTL_STOP);
/* stop transfer of Tx descriptors */
CSR_WRITE_4(sc, sc_if->sk_tx_bmu, SK_TXBMU_TX_STOP);
for (i = 0; i < SK_TIMEOUT; i++) {
val = CSR_READ_4(sc, sc_if->sk_tx_bmu);
if ((val & SK_TXBMU_TX_STOP) == 0)
break;
DELAY(1);
}
if (i == SK_TIMEOUT)
device_printf(sc_if->sk_if_dev,
"can not stop transfer of Tx descriptor\n");
/* stop transfer of Rx descriptors */
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_BMU_CSR, SK_RXBMU_RX_STOP);
for (i = 0; i < SK_TIMEOUT; i++) {
val = SK_IF_READ_4(sc_if, 0, SK_RXQ1_BMU_CSR);
if ((val & SK_RXBMU_RX_STOP) == 0)
break;
DELAY(1);
}
if (i == SK_TIMEOUT)
device_printf(sc_if->sk_if_dev,
"can not stop transfer of Rx descriptor\n");
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
if (sc_if->sk_phytype == SK_PHYTYPE_BCOM) {
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
/* Put PHY back into reset. */
val = sk_win_read_4(sc, SK_GPIO);
if (sc_if->sk_port == SK_PORT_A) {
val |= SK_GPIO_DIR0;
val &= ~SK_GPIO_DAT0;
} else {
val |= SK_GPIO_DIR2;
val &= ~SK_GPIO_DAT2;
}
sk_win_write_4(sc, SK_GPIO, val);
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Turn off various components of this interface. */
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
SK_XM_SETBIT_2(sc_if, XM_GPIO, XM_GPIO_RESETMAC);
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
switch (sc->sk_type) {
case SK_GENESIS:
SK_IF_WRITE_2(sc_if, 0, SK_TXF1_MACCTL, SK_TXMACCTL_XMAC_RESET);
SK_IF_WRITE_4(sc_if, 0, SK_RXF1_CTL, SK_FIFO_RESET);
break;
case SK_YUKON:
* Improve chip identification. Obtained from: NetBSD if_sk.c rev. 1.11 * Take PHY out of reset for Yukon Lite Rev. A3. Submitted by: postings on net@ in thread "skc0: no PHY found", 2005-02-22 Tested by: net Approved by: rwatson (mentor) MFC after: 5 days
2005-03-17 14:21:51 +00:00
case SK_YUKON_LITE:
case SK_YUKON_LP:
Add support for SK-9521 V2.0 and 3COM 3C940. Tested at 100Mbit only, using Asus P4P800 onboard 3C940. The -stable version of this patch I have in use for ~2 weeks now, and works just fine for me. Based on: Nathan L. Binkert's patch for OpenBSD Patch submitted by and thanks to: Jung-uk Kim <jkim@niksun.com> MFC after: 2 weeks
2003-09-20 10:53:08 +00:00
SK_IF_WRITE_1(sc_if,0, SK_RXMF1_CTRL_TEST, SK_RFCTL_RESET_SET);
SK_IF_WRITE_1(sc_if,0, SK_TXMF1_CTRL_TEST, SK_TFCTL_RESET_SET);
break;
}
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
SK_IF_WRITE_4(sc_if, 0, SK_RXQ1_BMU_CSR, SK_RXBMU_OFFLINE);
SK_IF_WRITE_4(sc_if, 0, SK_RXRB1_CTLTST, SK_RBCTL_RESET|SK_RBCTL_OFF);
SK_IF_WRITE_4(sc_if, 1, SK_TXQS1_BMU_CSR, SK_TXBMU_OFFLINE);
SK_IF_WRITE_4(sc_if, 1, SK_TXRBS1_CTLTST, SK_RBCTL_RESET|SK_RBCTL_OFF);
SK_IF_WRITE_1(sc_if, 0, SK_TXAR1_COUNTERCTL, SK_TXARCTL_OFF);
SK_IF_WRITE_1(sc_if, 0, SK_RXLED1_CTL, SK_RXLEDCTL_COUNTER_STOP);
SK_IF_WRITE_1(sc_if, 0, SK_TXLED1_CTL, SK_RXLEDCTL_COUNTER_STOP);
SK_IF_WRITE_1(sc_if, 0, SK_LINKLED1_CTL, SK_LINKLED_OFF);
SK_IF_WRITE_1(sc_if, 0, SK_LINKLED1_CTL, SK_LINKLED_LINKSYNC_OFF);
/* Disable interrupts */
if (sc_if->sk_port == SK_PORT_A)
sc->sk_intrmask &= ~SK_INTRS1;
else
sc->sk_intrmask &= ~SK_INTRS2;
CSR_WRITE_4(sc, SK_IMR, sc->sk_intrmask);
Reoganize/update the SysKonnect driver: - Break out the support for the XMAC II's PHY into an miibus driver. - Reorganize the probe/attach stuff using newbus. Each XMAC is now attached to the parent GEnesis controller using newbus. This is necessary since each XMAC must also have an attached miibus, and the miibus read/write register routines need to be able to get at the softc struct for each XMAC, not the one for the parent controller. This allows me to get rid of the grotty code I added for selecting the unit numbers for the ifnet interfaces: the unit numbers are now derived from the newbus-assigned unit numbers, which should track with the ifnet interface numbers. I think. At the very least, there should never be any collisions. - Add support for the SK-9821 and SK-9822 1000baseTX adapters. Special thanks to SysKonnect for loaning me two adapters for testing.
2000-04-22 02:16:41 +00:00
SK_XM_READ_2(sc_if, XM_ISR);
SK_XM_WRITE_2(sc_if, XM_IMR, 0xFFFF);
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
/* Free RX and TX mbufs still in the queues. */
for (i = 0; i < SK_RX_RING_CNT; i++) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
rxd = &sc_if->sk_cdata.sk_rxdesc[i];
if (rxd->rx_m != NULL) {
bus_dmamap_sync(sc_if->sk_cdata.sk_rx_tag,
rxd->rx_dmamap, BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc_if->sk_cdata.sk_rx_tag,
rxd->rx_dmamap);
m_freem(rxd->rx_m);
rxd->rx_m = NULL;
}
}
for (i = 0; i < SK_JUMBO_RX_RING_CNT; i++) {
jrxd = &sc_if->sk_cdata.sk_jumbo_rxdesc[i];
if (jrxd->rx_m != NULL) {
bus_dmamap_sync(sc_if->sk_cdata.sk_jumbo_rx_tag,
jrxd->rx_dmamap, BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc_if->sk_cdata.sk_jumbo_rx_tag,
jrxd->rx_dmamap);
m_freem(jrxd->rx_m);
jrxd->rx_m = NULL;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
}
for (i = 0; i < SK_TX_RING_CNT; i++) {
Bring busdmafied sk(4) to all architectures. - MPSAFE. No more recursive lock required. - bus_dma(9) conversion. I think it should work on all architectures. - optimized Rx handler for each normal and jumbo frames. Previously sk(4) used jumbo frame management code to handle normal sized frames. As the handler needs an additional lock to protect jumbo frame management structure from races, it used two lock operations for each received packet. Now sk(4) uses single lock operation for normal frame.(Jumbo frame still needs two lock operations as before.) The hardware supports DMA scatter operations for Rx descriptors such that it's possible to take advantagee of m_cljget(9) for jumbo frames. However, due to a unknown reasons it resulted in poor performance on sparc64. So I dropped m_cljget(9) approach. This should be revisited since it would reduce one lock operation for jumbo frame handling. - Tx TCP/Rx IP checksum offload support. According to the data sheet of SK-NET GENESIS the hardware supports Rx IP/TCP/UDP offload. But I couldn't make it work on my Yukon hardware. So Rx TCP/UDP was disabled at the moment. It seems that newer Yukon chips can support Tx UDP checksum offload too. But I need more documentation first. - Added more wait time in reading VPD data. It seems that ASUS LOM takes a very long time to respond VPD read signal. - Added an additional lock for MII register access callbacks. - Added more strict received packet validation routine. Previously it passed corrupted packets to upper layers under certain conditions. - A new function sk_yukon_tick() to handle auto-negotiation properly. - Interrupt handler now checks shared interrupt source and protects the interrupt handler from NULL pointer dereference which was caused by odd status word value. The status word can returns 0xffffffff if cable is unplugged while Rx/Tx/auto-negotiation is in progress. - suspend/resume support(not tested). - Added Rx/Tx FIFO flush routine for Yukon - Activate Tx descriptor poll timer in order to protect possible loss of SK_TXBMU_TX_START command. Previously the driver continuously issued SK_TXBMU_TX_START when it notices pending Tx descriptors not processed yet in interrupt handler. That approach would add additional PCI write access overhead under high Tx load situations and it might fail if the first SK_TXBMU_TX_START was lost and no interrupt is generated from the first SK_TXBMU_TX_START command. - s/printf/if_printf/, s/printf/device_printf/, Axe sk_unit in softc. - Setting multicast/station address is now safe on strict-alignment architectures. - Fix long standing bug in VLAN header length setup. - Added/corrected register definitions for Yukon. (Register information from Linux skge driver.) - Added Rx status definition for Marvell Yukon/XaQti XMAC. (Rx status register information from Linux skge driver.) - Update if_oerrors if we encounter watchdog error. - callout(9) conversion Special thanks to jkim who let me know RX status differences between Yukon and XaQti XMAC. It seems that there is still occasional watchdog timeout error but I couldn't reproduce it and need more information to analyze it from users. Tested by: bz(amd64), me(i386, sparc64), current ML Frank Behrens frank ! pinky ( sax $ de
2006-04-27 05:59:09 +00:00
txd = &sc_if->sk_cdata.sk_txdesc[i];
if (txd->tx_m != NULL) {
bus_dmamap_sync(sc_if->sk_cdata.sk_tx_tag,
txd->tx_dmamap, BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc_if->sk_cdata.sk_tx_tag,
txd->tx_dmamap);
m_freem(txd->tx_m);
txd->tx_m = NULL;
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
}
}
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags &= ~(IFF_DRV_RUNNING|IFF_DRV_OACTIVE);
* Solve "No PHY found" problem for more Yukon Lite variants. * Catch a bus attach error. * Improve locking. MFC after: 6 days
2005-08-13 11:13:12 +00:00
This commit adds driver support for the SysKonnect SK-984x series gigabit ethernet adapters. This includes two single port cards (single mode and multimode fiber) and two dual port cards (also single mode and multimode fiber). SysKonnect is currently the only vendor with a dual port gigabit ethernet NIC. The ports on dual port adapters are treated as separate network interfaces. Thus, if you have an SK-9844 dual port SX card, you should have both sk0 and sk1 interfaces attached. Dual port cards are implemented using two XMAC II chips connected to a single SysKonnect GEnesis controller. Hence, dual port cards are really one PCI device, as opposed to two separate PCI devices connected through a PCI to PCI bridge. Note that SysKonnect's drivers use the two ports for failover purposes rather that as two separate interfaces, plus they don't support jumbo frames. This applies to their Linux driver too. :) Support is provided for hardware multicast filtering, BPF and jumbo frames. The SysKonnect cards support TCP checksum offload however this feature is not currently enabled (hopefully it will be once we get checksum offload support). There are still a few things that need to be implemeted, like the ability to communicate with the on-board LM80 voltage/temperature monitor, but I wanted to get the driver under CVS control and into -current so people could bang on it. A big thanks for SysKonnect for making all their programming info for these cards (and for their FDDI and token ring cards) available without NDA (see www.syskonnect.com).
1999-07-09 04:30:09 +00:00
return;
}
* Lower interrupt moderation timer 200->100. Obtained from: NetBSD if_sk.c rev. 1.11 * Make interrupt moderation configurable via sysctl/tuneable. PR: kern/41220 Approved by: rwatson (mentor)
2005-03-17 14:23:13 +00:00
static int
sysctl_int_range(SYSCTL_HANDLER_ARGS, int low, int high)
{
int error, value;
if (!arg1)
return (EINVAL);
value = *(int *)arg1;
error = sysctl_handle_int(oidp, &value, 0, req);
if (error || !req->newptr)
return (error);
if (value < low || value > high)
return (EINVAL);
*(int *)arg1 = value;
return (0);
}
static int
sysctl_hw_sk_int_mod(SYSCTL_HANDLER_ARGS)
{
return (sysctl_int_range(oidp, arg1, arg2, req, SK_IM_MIN, SK_IM_MAX));
}
Reference in New Issue Copy Permalink