d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
a4acc8c3df
freebsd-dev/sys/dev/dc/if_dc.c

4142 lines
109 KiB
C
Raw Normal View History

/* -> /*- for license, minor formatting changes
2005-01-07 02:29:27 +00:00
/*-
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
* Copyright (c) 1997, 1998, 1999
* Bill Paul <wpaul@ee.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.
*/
Remove duplicate FBSDID's, move others to their right place.
2003-11-14 17:16:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* DEC "tulip" clone ethernet driver. Supports the DEC/Intel 21143
* series chips and several workalikes including the following:
*
Add the PCI IDs for the Macronix 98727 and 98732 parts. These are 3.3volt PCI/cardbus chipsets similar to the 98715 (and they have 512-bit hash tables). Also update the man page to mention the 98727/98732 and the SOHOware SFA110A Rev B4 card with the 98715AEC-C chip.
2000-07-17 19:27:41 +00:00
* Macronix 98713/98715/98725/98727/98732 PMAC (www.macronix.com)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
* Macronix/Lite-On 82c115 PNIC II (www.macronix.com)
* Lite-On 82c168/82c169 PNIC (www.litecom.com)
* ASIX Electronics AX88140A (www.asix.com.tw)
* ASIX Electronics AX88141 (www.asix.com.tw)
* ADMtek AL981 (www.admtek.com.tw)
Fix card/device names, no functional change The ADMtek AN985 is the cardbus variant of ADMtek AN983 The Netgear FA511 is just a relabled ADMtek AN985 PR: kern/50574 MFC after: 1 month
2010-01-03 13:59:59 +00:00
* ADMtek AN983 (www.admtek.com.tw)
Fix CardBus spelling. MFC after: 1 month
2010-01-03 23:15:14 +00:00
* ADMtek CardBus AN985 (www.admtek.com.tw)
* Netgear FA511 (www.netgear.com) Appears to be rebadged ADMTek CardBus AN985
Add support for the Davicom DM9102A 10/100 ethernet controller chip. This is just to make sure we initialize the chip correctly: we need to make the sure the port select bit in CSR6 is set properly so that we use the internal PHY for 10/100 support. (The eval boards I have also include an external HomePNA PHY, but I need to play with that more before I can support it.)
2000-01-19 19:03:08 +00:00
* Davicom DM9100, DM9102, DM9102A (www.davicom8.com)
Add support for the Accton EN1217. PR: 18735 Submitted by: Adoal Xu <adoal@iname.com>
2000-06-11 11:54:52 +00:00
* Accton EN1217 (www.accton.com)
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
* Xircom X3201 (www.xircom.com)
Add PCI IDs for some additional cardbus cards. Yes, there really is a RealTek 8139 cardbus device. Unfortunately it doesn't quite work yet because the CIS parser barfs on it. Submitted by msmith, with some small tweaks by me.
2000-10-28 09:00:20 +00:00
* Abocom FE2500
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
* Conexant LANfinity (www.conexant.com)
Add support for 3Com OfficeConnect 10/100B. PR: 49059, 50747 Submitted by: Dax Eckenberg <daxbert@dweebsoft.com> Reviewed by: imp, jhb Approved by: jhb MFC after: 2 weeks
2003-05-12 19:50:21 +00:00
* 3Com OfficeConnect 10/100B 3CSOHO100B (www.3com.com)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
*
* Datasheets for the 21143 are available at developer.intel.com.
* Datasheets for the clone parts can be found at their respective sites.
* (Except for the PNIC; see www.freebsd.org/~wpaul/PNIC/pnic.ps.gz.)
* The PNIC II is essentially a Macronix 98715A chip; the only difference
* worth noting is that its multicast hash table is only 128 bits wide
* instead of 512.
*
* Written by Bill Paul <wpaul@ee.columbia.edu>
* Electrical Engineering Department
* Columbia University, New York City
*/
/*
* The Intel 21143 is the successor to the DEC 21140. It is basically
* the same as the 21140 but with a few new features. The 21143 supports
* three kinds of media attachments:
*
* o MII port, for 10Mbps and 100Mbps support and NWAY
* autonegotiation provided by an external PHY.
* o SYM port, for symbol mode 100Mbps support.
* o 10baseT port.
* o AUI/BNC port.
*
* The 100Mbps SYM port and 10baseT port can be used together in
* combination with the internal NWAY support to create a 10/100
* autosensing configuration.
*
* Note that not all tulip workalikes are handled in this driver: we only
* deal with those which are relatively well behaved. The Winbond is
* handled separately due to its different register offsets and the
* special handling needed for its various bugs. The PNIC is handled
* here, but I'm not thrilled about it.
*
* All of the workalike chips use some form of MII transceiver support
* with the exception of the Macronix chips, which also have a SYM port.
* The ASIX AX88140A is also documented to have a SYM port, but all
* the cards I've seen use an MII transceiver, probably because the
* AX88140A doesn't support internal NWAY.
*/
- Don't pollute opt_global.h with DEVICE_POLLING and introduce opt_device_polling.h - Include opt_device_polling.h into appropriate files. - Embrace with HAVE_KERNEL_OPTION_HEADERS the include in the files that can be compiled as loadable modules. Reviewed by: bde
2005-10-05 10:09:17 +00:00
#ifdef HAVE_KERNEL_OPTION_HEADERS
#include "opt_device_polling.h"
#endif
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include <sys/param.h>
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
#include <sys/endian.h>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include <sys/systm.h>
#include <sys/sockio.h>
#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>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include <sys/socket.h>
#include <net/if.h>
#include <net/if_arp.h>
#include <net/ethernet.h>
#include <net/if_dl.h>
#include <net/if_media.h>
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
#include <net/if_types.h>
#include <net/if_vlan_var.h>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include <net/bpf.h>
#include <machine/bus.h>
#include <machine/resource.h>
#include <sys/bus.h>
#include <sys/rman.h>
#include <dev/mii/mii.h>
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
#include <dev/mii/mii_bitbang.h>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include <dev/mii/miivar.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>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_USEIOSPACE
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Make dc compile after repo-copy.
2005-10-18 06:09:42 +00:00
#include <dev/dc/if_dcreg.h>
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Some integrated Davicom cards in sparc64 boxes have an all zeros MAC address in the EEPROM, and we need to get it from OpenFirmware. This isn't very pretty but time is lacking to do this in a better way this near 5.2-RELEASE. This is a RELENG_5_2 candidate. Original version by: Marius Strobl <marius@alchemy.franken.de> Tested by: Pete Bentley <pete@sorted.org> Reviewed by: jake
2004-01-08 19:08:27 +00:00
#ifdef __sparc64__
#include <dev/ofw/openfirm.h>
#include <machine/ofw_machdep.h>
#endif
- 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(dc, pci, 1, 1, 1);
MODULE_DEPEND(dc, ether, 1, 1, 1);
Depend on miibus. Note that if_aue doesn't strictly depend on usb because it uses the method interface for calls rather than using internal symbols, and because it's a child driver of usb and therefore will not try and do anything unless the parent usb code is loaded at some point. if_aue does strictly depend on miibus as it will fail to link if it is missing.
2000-04-29 13:41:57 +00:00
MODULE_DEPEND(dc, miibus, 1, 1, 1);
Better comment
2005-08-29 18:45:21 +00:00
/*
* "device miibus" is required in kernel config. See GENERIC if you get
* errors here.
*/
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#include "miibus_if.h"
/*
* Various supported device vendors/types and their names.
*/
Remove duplicate const specifiers in many drivers (I hope I got all of them, please let me know if not). Most of these are of the form: static const struct bzzt_type { [...list of members...] } const bzzt_devs[] = { [...list of initializers...] }; The second const is unnecessary, as arrays cannot be modified anyway, and if the elements are const, the whole thing is const automatically (e.g. it is placed in .rodata). I have verified this does not change the binary output of a full kernel build (except for build timestamps embedded in the object files). Reviewed by: yongari, marius MFC after: 1 week
2012-11-05 19:16:27 +00:00
static const struct dc_type dc_devs[] = {
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DEC, DC_DEVICEID_21143), 0,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Intel 21143 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9009), 0,
Add support for the Davicom DM9009 chipset. PR: 46859 Submitted by: Boaz Haberman <boaz@ool-182f8b09.dyn.optonline.net> Approved by: rwatson
2003-01-09 20:51:39 +00:00
"Davicom DM9009 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9100), 0,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Davicom DM9100 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9102), DC_REVISION_DM9102A,
Add support for the Davicom DM9102A 10/100 ethernet controller chip. This is just to make sure we initialize the chip correctly: we need to make the sure the port select bit in CSR6 is set properly so that we use the internal PHY for 10/100 support. (The eval boards I have also include an external HomePNA PHY, but I need to play with that more before I can support it.)
2000-01-19 19:03:08 +00:00
"Davicom DM9102A 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9102), 0,
"Davicom DM9102 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AL981), 0,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"ADMtek AL981 10/100BaseTX" },
Fix card/device names, no functional change The ADMtek AN985 is the cardbus variant of ADMtek AN983 The Netgear FA511 is just a relabled ADMtek AN985 PR: kern/50574 MFC after: 1 month
2010-01-03 13:59:59 +00:00
{ DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AN983), 0,
"ADMtek AN983 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AN985), 0,
Fix CardBus spelling. MFC after: 1 month
2010-01-03 23:15:14 +00:00
"ADMtek AN985 CardBus 10/100BaseTX or clone" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_ADM9511), 0,
Add ADMtek ADM9511 and ADM9513 device ID's. PR: PR51823 Submitted by: Kaho Toshikazu <kaho@elam.kais.kyoto-u.ac.jp> Reviewed by: phk MFC after: 2 days
2003-06-08 10:11:55 +00:00
"ADMtek ADM9511 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_ADM9513), 0,
Add ADMtek ADM9511 and ADM9513 device ID's. PR: PR51823 Submitted by: Kaho Toshikazu <kaho@elam.kais.kyoto-u.ac.jp> Reviewed by: phk MFC after: 2 days
2003-06-08 10:11:55 +00:00
"ADMtek ADM9513 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ASIX, DC_DEVICEID_AX88140A), DC_REVISION_88141,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"ASIX AX88141 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ASIX, DC_DEVICEID_AX88140A), 0,
"ASIX AX88140A 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_98713), DC_REVISION_98713A,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Macronix 98713A 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_98713), 0,
"Macronix 98713 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_CP, DC_DEVICEID_98713_CP), DC_REVISION_98713A,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Compex RL100-TX 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_CP, DC_DEVICEID_98713_CP), 0,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Compex RL100-TX 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_987x5), DC_REVISION_98725,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"Macronix 98725 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_987x5), DC_REVISION_98715AEC_C,
"Macronix 98715AEC-C 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_987x5), 0,
"Macronix 98715/98715A 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_98727), 0,
Add the PCI IDs for the Macronix 98727 and 98732 parts. These are 3.3volt PCI/cardbus chipsets similar to the 98715 (and they have 512-bit hash tables). Also update the man page to mention the 98727/98732 and the SOHOware SFA110A Rev B4 card with the 98715AEC-C chip.
2000-07-17 19:27:41 +00:00
"Macronix 98727/98732 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C115), 0,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"LC82C115 PNIC II 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C168), DC_REVISION_82C169,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
"82c169 PNIC 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C168), 0,
"82c168 PNIC 10/100BaseTX" },
{ DC_DEVID(DC_VENDORID_ACCTON, DC_DEVICEID_EN1217), 0,
Add support for the Accton EN1217. PR: 18735 Submitted by: Adoal Xu <adoal@iname.com>
2000-06-11 11:54:52 +00:00
"Accton EN1217 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ACCTON, DC_DEVICEID_EN2242), 0,
Add support for the Accton EN2242 MiniPCI adapter. This is just an ADMtek Centaur chip, so all we need is the PCI ID. Submitted by: Scott Lang <scottl@FreeBSD.org>
2000-11-14 19:35:22 +00:00
"Accton EN2242 MiniPCI 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_XIRCOM, DC_DEVICEID_X3201), 0,
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
"Xircom X3201 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_DLINK, DC_DEVICEID_DRP32TXD), 0,
Add device-id for the Neteasy DRP-32TXD cardbus 10/100 card. It's another ADMTek AN985 clone. MFC after: 3 days Tested by: Lila liladude at aruba dot it
2006-03-16 20:00:39 +00:00
"Neteasy DRP-32TXD Cardbus 10/100" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ABOCOM, DC_DEVICEID_FE2500), 0,
Add PCI IDs for some additional cardbus cards. Yes, there really is a RealTek 8139 cardbus device. Unfortunately it doesn't quite work yet because the CIS parser barfs on it. Submitted by msmith, with some small tweaks by me.
2000-10-28 09:00:20 +00:00
"Abocom FE2500 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_ABOCOM, DC_DEVICEID_FE2500MX), 0,
Add support for ELECOM/Laneed CardBus FastEtherner Card(LD-CBL/TXA) Submitted by: Masahiro Yamagishi <night@pluto.dti.ne.jp> Approved by: re (scottl)
2003-12-06 02:29:31 +00:00
"Abocom FE2500MX 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_CONEXANT, DC_DEVICEID_RS7112), 0,
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
"Conexant LANfinity MiniPCI 10/100BaseTX" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_HAWKING, DC_DEVICEID_HAWKING_PN672TX), 0,
Add preliminary support for the Hawking PN672TX CardBus cards. # Preliminary because there are some subtle things the NetBSD driver does # that we don't do yet. My card works for me w/o them.
2003-01-10 05:37:15 +00:00
"Hawking CB102 CardBus 10/100" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_PLANEX, DC_DEVICEID_FNW3602T), 0,
Add support for Planex FNW-3602-T(CardBus 100M/10M). Submitted by: kazz <kazz@v001.vaio.ne.jp> Obtained from: [bsd-nomads:16637]
2003-04-18 15:42:25 +00:00
"PlaneX FNW-3602-T CardBus 10/100" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_3COM, DC_DEVICEID_3CSOHOB), 0,
Add support for 3Com OfficeConnect 10/100B. PR: 49059, 50747 Submitted by: Dax Eckenberg <daxbert@dweebsoft.com> Reviewed by: imp, jhb Approved by: jhb MFC after: 2 weeks
2003-05-12 19:50:21 +00:00
"3Com OfficeConnect 10/100B" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_MICROSOFT, DC_DEVICEID_MSMN120), 0,
Add support for a bunch of Microsoft networking products: - MN-110 10/100 USB ethernet (ADMtek Pegasus II, if_aue) - MN-120 10/100 cardbus (ADMtek Centaur-C, if_dc) - MN-130 10/100 PCI (ADMtek Centaur-P, if_dc) Also update dc(4) man page to mention support for MN-120 and MN-130.
2003-07-10 18:43:17 +00:00
"Microsoft MN-120 CardBus 10/100" },
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
{ DC_DEVID(DC_VENDORID_MICROSOFT, DC_DEVICEID_MSMN130), 0,
Add support for a bunch of Microsoft networking products: - MN-110 10/100 USB ethernet (ADMtek Pegasus II, if_aue) - MN-120 10/100 cardbus (ADMtek Centaur-C, if_dc) - MN-130 10/100 PCI (ADMtek Centaur-P, if_dc) Also update dc(4) man page to mention support for MN-120 and MN-130.
2003-07-10 18:43:17 +00:00
"Microsoft MN-130 10/100" },
Fix the last commit. Submitted by: jhb Pointy hat to: glebius
2006-06-08 08:15:21 +00:00
{ DC_DEVID(DC_VENDORID_LINKSYS, DC_DEVICEID_PCMPC200_AB08), 0,
"Linksys PCMPC200 CardBus 10/100" },
{ DC_DEVID(DC_VENDORID_LINKSYS, DC_DEVICEID_PCMPC200_AB09), 0,
"Linksys PCMPC200 CardBus 10/100" },
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
{ DC_DEVID(DC_VENDORID_ULI, DC_DEVICEID_M5261), 0,
"ULi M5261 FastEthernet" },
{ DC_DEVID(DC_VENDORID_ULI, DC_DEVICEID_M5263), 0,
"ULi M5263 FastEthernet" },
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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 dc_probe(device_t);
static int dc_attach(device_t);
static int dc_detach(device_t);
static int dc_suspend(device_t);
static int dc_resume(device_t);
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
static const struct dc_type *dc_devtype(device_t);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
static void dc_discard_rxbuf(struct dc_softc *, int);
static int dc_newbuf(struct dc_softc *, int);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int dc_encap(struct dc_softc *, struct mbuf **);
static void dc_pnic_rx_bug_war(struct dc_softc *, int);
static int dc_rx_resync(struct dc_softc *);
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
static int dc_rxeof(struct dc_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_txeof(struct dc_softc *);
static void dc_tick(void *);
static void dc_tx_underrun(struct dc_softc *);
static void dc_intr(void *);
static void dc_start(struct ifnet *);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
static void dc_start_locked(struct ifnet *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int dc_ioctl(struct ifnet *, u_long, caddr_t);
static void dc_init(void *);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
static void dc_init_locked(struct dc_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_stop(struct dc_softc *);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
static void dc_watchdog(void *);
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
static int dc_shutdown(device_t);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int dc_ifmedia_upd(struct ifnet *);
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
static int dc_ifmedia_upd_locked(struct dc_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_ifmedia_sts(struct ifnet *, struct ifmediareq *);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
static int dc_dma_alloc(struct dc_softc *);
static void dc_dma_free(struct dc_softc *);
static void dc_dma_map_addr(void *, bus_dma_segment_t *, int, int);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_delay(struct dc_softc *);
static void dc_eeprom_idle(struct dc_softc *);
static void dc_eeprom_putbyte(struct dc_softc *, int);
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
static void dc_eeprom_getword(struct dc_softc *, int, uint16_t *);
static void dc_eeprom_getword_pnic(struct dc_softc *, int, uint16_t *);
static void dc_eeprom_getword_xircom(struct dc_softc *, int, uint16_t *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_eeprom_width(struct dc_softc *);
static void dc_read_eeprom(struct dc_softc *, caddr_t, int, int, int);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int dc_miibus_readreg(device_t, int, int);
static int dc_miibus_writereg(device_t, int, int, int);
static void dc_miibus_statchg(device_t);
static void dc_miibus_mediainit(device_t);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_setcfg(struct dc_softc *, int);
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
static void dc_netcfg_wait(struct dc_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static uint32_t dc_mchash_le(struct dc_softc *, const uint8_t *);
static uint32_t dc_mchash_be(const uint8_t *);
static void dc_setfilt_21143(struct dc_softc *);
static void dc_setfilt_asix(struct dc_softc *);
static void dc_setfilt_admtek(struct dc_softc *);
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
static void dc_setfilt_uli(struct dc_softc *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_setfilt_xircom(struct dc_softc *);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_setfilt(struct dc_softc *);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_reset(struct dc_softc *);
static int dc_list_rx_init(struct dc_softc *);
static int dc_list_tx_init(struct dc_softc *);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int dc_read_srom(struct dc_softc *, int);
static int dc_parse_21143_srom(struct dc_softc *);
static int dc_decode_leaf_sia(struct dc_softc *, struct dc_eblock_sia *);
static int dc_decode_leaf_mii(struct dc_softc *, struct dc_eblock_mii *);
static int dc_decode_leaf_sym(struct dc_softc *, struct dc_eblock_sym *);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void dc_apply_fixup(struct dc_softc *, int);
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
static int dc_check_multiport(struct dc_softc *);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
/*
* MII bit-bang glue
*/
static uint32_t dc_mii_bitbang_read(device_t);
static void dc_mii_bitbang_write(device_t, uint32_t);
static const struct mii_bitbang_ops dc_mii_bitbang_ops = {
dc_mii_bitbang_read,
dc_mii_bitbang_write,
{
DC_SIO_MII_DATAOUT, /* MII_BIT_MDO */
DC_SIO_MII_DATAIN, /* MII_BIT_MDI */
DC_SIO_MII_CLK, /* MII_BIT_MDC */
0, /* MII_BIT_DIR_HOST_PHY */
DC_SIO_MII_DIR, /* MII_BIT_DIR_PHY_HOST */
}
};
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#ifdef DC_USEIOSPACE
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_RES SYS_RES_IOPORT
#define DC_RID DC_PCI_CFBIO
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#else
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_RES SYS_RES_MEMORY
#define DC_RID DC_PCI_CFBMA
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
#endif
static device_method_t dc_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, dc_probe),
DEVMETHOD(device_attach, dc_attach),
DEVMETHOD(device_detach, dc_detach),
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
DEVMETHOD(device_suspend, dc_suspend),
DEVMETHOD(device_resume, dc_resume),
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DEVMETHOD(device_shutdown, dc_shutdown),
/* MII interface */
DEVMETHOD(miibus_readreg, dc_miibus_readreg),
DEVMETHOD(miibus_writereg, dc_miibus_writereg),
DEVMETHOD(miibus_statchg, dc_miibus_statchg),
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
DEVMETHOD(miibus_mediainit, dc_miibus_mediainit),
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- 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
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
};
static driver_t dc_driver = {
"dc",
dc_methods,
sizeof(struct dc_softc)
};
static devclass_t dc_devclass;
- Change the module order of these MAC drivers to be last so they are deterministically handled after the corresponding PHY drivers when loaded as modules. Otherwise, when these MAC/PHY driver pairs are compiled into a single module probing the PHY driver may fail. This makes r151438 and r226154 actually work. [1] Reported and tested by: yongari (fxp(4)) - Use DEVMETHOD_END. - Use NULL instead of 0 for pointers. Submitted by: jhb [1] MFC after: 3 days
2012-05-11 02:40:40 +00:00
DRIVER_MODULE_ORDERED(dc, pci, dc_driver, dc_devclass, NULL, NULL,
SI_ORDER_ANY);
DRIVER_MODULE(miibus, dc, miibus_driver, miibus_devclass, NULL, NULL);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_SETBIT(sc, reg, x) \
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, reg, CSR_READ_4(sc, reg) | (x))
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_CLRBIT(sc, reg, x) \
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, reg, CSR_READ_4(sc, reg) & ~(x))
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define SIO_SET(x) DC_SETBIT(sc, DC_SIO, (x))
#define SIO_CLR(x) DC_CLRBIT(sc, DC_SIO, (x))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_delay(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int idx;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
for (idx = (300 / 33) + 1; idx > 0; idx--)
CSR_READ_4(sc, DC_BUSCTL);
}
Be consistent about functions being static. Spotted by: FlexeLint.
2002-10-16 09:04:52 +00:00
static void
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_eeprom_width(struct dc_softc *sc)
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
{
int i;
/* Force EEPROM to idle state. */
dc_eeprom_idle(sc);
/* Enter EEPROM access mode. */
CSR_WRITE_4(sc, DC_SIO, DC_SIO_EESEL);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_ROMCTL_READ);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CS);
dc_delay(sc);
for (i = 3; i--;) {
if (6 & (1 << i))
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_DATAIN);
else
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_DATAIN);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
}
for (i = 1; i <= 12; i++) {
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
if (!(CSR_READ_4(sc, DC_SIO) & DC_SIO_EE_DATAOUT)) {
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
break;
}
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
}
/* Turn off EEPROM access mode. */
dc_eeprom_idle(sc);
if (i < 4 || i > 12)
sc->dc_romwidth = 6;
else
sc->dc_romwidth = i;
/* Enter EEPROM access mode. */
CSR_WRITE_4(sc, DC_SIO, DC_SIO_EESEL);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_ROMCTL_READ);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CS);
dc_delay(sc);
/* Turn off EEPROM access mode. */
dc_eeprom_idle(sc);
}
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_eeprom_idle(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_SIO, DC_SIO_EESEL);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_ROMCTL_READ);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CS);
dc_delay(sc);
for (i = 0; i < 25; i++) {
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
}
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CS);
dc_delay(sc);
CSR_WRITE_4(sc, DC_SIO, 0x00000000);
}
/*
* Send a read command and address to the EEPROM, check for ACK.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_eeprom_putbyte(struct dc_softc *sc, int addr)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int d, i;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
d = DC_EECMD_READ >> 6;
for (i = 3; i--; ) {
if (d & (1 << i))
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_DATAIN);
else
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_DATAIN);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Feed in each bit and strobe the clock.
*/
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
for (i = sc->dc_romwidth; i--;) {
if (addr & (1 << i)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
SIO_SET(DC_SIO_EE_DATAIN);
} else {
SIO_CLR(DC_SIO_EE_DATAIN);
}
dc_delay(sc);
SIO_SET(DC_SIO_EE_CLK);
dc_delay(sc);
SIO_CLR(DC_SIO_EE_CLK);
dc_delay(sc);
}
}
/*
* Read a word of data stored in the EEPROM at address 'addr.'
* The PNIC 82c168/82c169 has its own non-standard way to read
* the EEPROM.
*/
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
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
dc_eeprom_getword_pnic(struct dc_softc *sc, int addr, uint16_t *dest)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t r;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
CSR_WRITE_4(sc, DC_PN_SIOCTL, DC_PN_EEOPCODE_READ | addr);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
for (i = 0; i < DC_TIMEOUT; i++) {
DELAY(1);
r = CSR_READ_4(sc, DC_SIO);
if (!(r & DC_PN_SIOCTL_BUSY)) {
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
*dest = (uint16_t)(r & 0xFFFF);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
return;
}
}
}
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
/*
* Read a word of data stored in the EEPROM at address 'addr.'
* The Xircom X3201 has its own non-standard way to read
* the EEPROM, too.
*/
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
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
dc_eeprom_getword_xircom(struct dc_softc *sc, int addr, uint16_t *dest)
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
SIO_SET(DC_SIO_ROMSEL | DC_SIO_ROMCTL_READ);
addr *= 2;
CSR_WRITE_4(sc, DC_ROM, addr | 0x160);
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
*dest = (uint16_t)CSR_READ_4(sc, DC_SIO) & 0xff;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
addr += 1;
CSR_WRITE_4(sc, DC_ROM, addr | 0x160);
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
*dest |= ((uint16_t)CSR_READ_4(sc, DC_SIO) & 0xff) << 8;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
SIO_CLR(DC_SIO_ROMSEL | DC_SIO_ROMCTL_READ);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Read a word of data stored in the EEPROM at address 'addr.'
*/
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
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
dc_eeprom_getword(struct dc_softc *sc, int addr, uint16_t *dest)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint16_t word = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Force EEPROM to idle state. */
dc_eeprom_idle(sc);
/* Enter EEPROM access mode. */
CSR_WRITE_4(sc, DC_SIO, DC_SIO_EESEL);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_ROMCTL_READ);
dc_delay(sc);
DC_CLRBIT(sc, DC_SIO, DC_SIO_EE_CLK);
dc_delay(sc);
DC_SETBIT(sc, DC_SIO, DC_SIO_EE_CS);
dc_delay(sc);
/*
* Send address of word we want to read.
*/
dc_eeprom_putbyte(sc, addr);
/*
* Start reading bits from EEPROM.
*/
for (i = 0x8000; i; i >>= 1) {
SIO_SET(DC_SIO_EE_CLK);
dc_delay(sc);
if (CSR_READ_4(sc, DC_SIO) & DC_SIO_EE_DATAOUT)
word |= i;
dc_delay(sc);
SIO_CLR(DC_SIO_EE_CLK);
dc_delay(sc);
}
/* Turn off EEPROM access mode. */
dc_eeprom_idle(sc);
*dest = word;
}
/*
* Read a sequence of words from the EEPROM.
*/
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
Read the MAC address in the EEPROM in the correct byte order. This is a no-op on little endian architectures, but fixes getting the MAC address for some dc(4) cards on big endian architectures. This is a RELENG_5 candidate. Tested by: gallatin (powerpc), marius (sparc64) First version of the patch written by: gallatin
2004-10-01 15:23:31 +00:00
dc_read_eeprom(struct dc_softc *sc, caddr_t dest, int off, int cnt, int be)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint16_t word = 0, *ptr;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
for (i = 0; i < cnt; i++) {
if (DC_IS_PNIC(sc))
dc_eeprom_getword_pnic(sc, off + i, &word);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
else if (DC_IS_XIRCOM(sc))
dc_eeprom_getword_xircom(sc, off + i, &word);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
else
dc_eeprom_getword(sc, off + i, &word);
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
ptr = (uint16_t *)(dest + (i * 2));
Read the MAC address in the EEPROM in the correct byte order. This is a no-op on little endian architectures, but fixes getting the MAC address for some dc(4) cards on big endian architectures. This is a RELENG_5 candidate. Tested by: gallatin (powerpc), marius (sparc64) First version of the patch written by: gallatin
2004-10-01 15:23:31 +00:00
if (be)
*ptr = be16toh(word);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
else
Read the MAC address in the EEPROM in the correct byte order. This is a no-op on little endian architectures, but fixes getting the MAC address for some dc(4) cards on big endian architectures. This is a RELENG_5 candidate. Tested by: gallatin (powerpc), marius (sparc64) First version of the patch written by: gallatin
2004-10-01 15:23:31 +00:00
*ptr = le16toh(word);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
/*
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
* Write the MII serial port for the MII bit-bang module.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
dc_mii_bitbang_write(device_t dev, uint32_t val)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
struct dc_softc *sc;
- According to the corresponding Linux, NetBSD and OpenSolaris drivers, there should be a 1us delay after every write when bit-banging the MII. Also insert barriers in order to ensure the intended ordering. These changes hopefully will solve the bus wedging occasionally experienced with DM9102A since r182461. - Deobfuscate dc_mii_readreg() a bit.
2008-12-07 23:02:37 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
sc = device_get_softc(dev);
- According to the corresponding Linux, NetBSD and OpenSolaris drivers, there should be a 1us delay after every write when bit-banging the MII. Also insert barriers in order to ensure the intended ordering. These changes hopefully will solve the bus wedging occasionally experienced with DM9102A since r182461. - Deobfuscate dc_mii_readreg() a bit.
2008-12-07 23:02:37 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
CSR_WRITE_4(sc, DC_SIO, val);
- According to the corresponding Linux, NetBSD and OpenSolaris drivers, there should be a 1us delay after every write when bit-banging the MII. Also insert barriers in order to ensure the intended ordering. These changes hopefully will solve the bus wedging occasionally experienced with DM9102A since r182461. - Deobfuscate dc_mii_readreg() a bit.
2008-12-07 23:02:37 +00:00
CSR_BARRIER_4(sc, DC_SIO,
BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
* Read the MII serial port for the MII bit-bang module.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
*/
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
static uint32_t
dc_mii_bitbang_read(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
struct dc_softc *sc;
uint32_t val;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
sc = device_get_softc(dev);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
val = CSR_READ_4(sc, DC_SIO);
- According to the corresponding Linux, NetBSD and OpenSolaris drivers, there should be a 1us delay after every write when bit-banging the MII. Also insert barriers in order to ensure the intended ordering. These changes hopefully will solve the bus wedging occasionally experienced with DM9102A since r182461. - Deobfuscate dc_mii_readreg() a bit.
2008-12-07 23:02:37 +00:00
CSR_BARRIER_4(sc, DC_SIO,
BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
return (val);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_miibus_readreg(device_t dev, int phy, int reg)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
struct dc_softc *sc;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i, rval, phy_reg = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
if (sc->dc_pmode != DC_PMODE_MII) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (phy == (MII_NPHY - 1)) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
switch (reg) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case MII_BMSR:
/*
* Fake something to make the probe
* code think there's a PHY here.
*/
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (BMSR_MEDIAMASK);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
case MII_PHYIDR1:
if (DC_IS_PNIC(sc))
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (DC_VENDORID_LO);
return (DC_VENDORID_DEC);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
case MII_PHYIDR2:
if (DC_IS_PNIC(sc))
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (DC_DEVICEID_82C168);
return (DC_DEVICEID_21143);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
default:
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
} else
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
if (DC_IS_PNIC(sc)) {
CSR_WRITE_4(sc, DC_PN_MII, DC_PN_MIIOPCODE_READ |
(phy << 23) | (reg << 18));
for (i = 0; i < DC_TIMEOUT; i++) {
DELAY(1);
rval = CSR_READ_4(sc, DC_PN_MII);
if (!(rval & DC_PN_MII_BUSY)) {
rval &= 0xFFFF;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (rval == 0xFFFF ? 0 : rval);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
if (sc->dc_type == DC_TYPE_ULI_M5263) {
CSR_WRITE_4(sc, DC_ROM,
((phy << DC_ULI_PHY_ADDR_SHIFT) & DC_ULI_PHY_ADDR_MASK) |
((reg << DC_ULI_PHY_REG_SHIFT) & DC_ULI_PHY_REG_MASK) |
DC_ULI_PHY_OP_READ);
for (i = 0; i < DC_TIMEOUT; i++) {
DELAY(1);
rval = CSR_READ_4(sc, DC_ROM);
if ((rval & DC_ULI_PHY_OP_DONE) != 0) {
return (rval & DC_ULI_PHY_DATA_MASK);
}
}
if (i == DC_TIMEOUT)
device_printf(dev, "phy read timed out\n");
return (0);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (DC_IS_COMET(sc)) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
switch (reg) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case MII_BMCR:
phy_reg = DC_AL_BMCR;
break;
case MII_BMSR:
phy_reg = DC_AL_BMSR;
break;
case MII_PHYIDR1:
phy_reg = DC_AL_VENID;
break;
case MII_PHYIDR2:
phy_reg = DC_AL_DEVID;
break;
case MII_ANAR:
phy_reg = DC_AL_ANAR;
break;
case MII_ANLPAR:
phy_reg = DC_AL_LPAR;
break;
case MII_ANER:
phy_reg = DC_AL_ANER;
break;
default:
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(dev, "phy_read: bad phy register %x\n",
reg);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
rval = CSR_READ_4(sc, phy_reg) & 0x0000FFFF;
if (rval == 0xFFFF)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
return (rval);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Bug fixes that I've put together while working on a project in the office: if_vr: handle the case where vr_encap() returns failure: bust out of the packet sending loop instead of panicking. Also add some missing newlines to some printf()s. if_dc: The miibus_read and miibus_write methods keep swapping in and out of MII mode by fiddling with CSR6 for cards with MII PHYs. This is a hack to support the original Macronix 98713 card which has built-in NWAY that uses an MII-like management interface even though it uses serial transceivers. Conditionalize this so that we only do this on 98713 chips, since it does bad things to genuine tulip chips (and maybe other clones).
2001-01-19 23:55:07 +00:00
if (sc->dc_type == DC_TYPE_98713) {
phy_reg = CSR_READ_4(sc, DC_NETCFG);
CSR_WRITE_4(sc, DC_NETCFG, phy_reg & ~DC_NETCFG_PORTSEL);
}
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
rval = mii_bitbang_readreg(dev, &dc_mii_bitbang_ops, phy, reg);
Bug fixes that I've put together while working on a project in the office: if_vr: handle the case where vr_encap() returns failure: bust out of the packet sending loop instead of panicking. Also add some missing newlines to some printf()s. if_dc: The miibus_read and miibus_write methods keep swapping in and out of MII mode by fiddling with CSR6 for cards with MII PHYs. This is a hack to support the original Macronix 98713 card which has built-in NWAY that uses an MII-like management interface even though it uses serial transceivers. Conditionalize this so that we only do this on 98713 chips, since it does bad things to genuine tulip chips (and maybe other clones).
2001-01-19 23:55:07 +00:00
if (sc->dc_type == DC_TYPE_98713)
CSR_WRITE_4(sc, DC_NETCFG, phy_reg);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
return (rval);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_miibus_writereg(device_t dev, int phy, int reg, int data)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
int i, phy_reg = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
if (DC_IS_PNIC(sc)) {
CSR_WRITE_4(sc, DC_PN_MII, DC_PN_MIIOPCODE_WRITE |
(phy << 23) | (reg << 10) | data);
for (i = 0; i < DC_TIMEOUT; i++) {
if (!(CSR_READ_4(sc, DC_PN_MII) & DC_PN_MII_BUSY))
break;
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
if (sc->dc_type == DC_TYPE_ULI_M5263) {
CSR_WRITE_4(sc, DC_ROM,
((phy << DC_ULI_PHY_ADDR_SHIFT) & DC_ULI_PHY_ADDR_MASK) |
((reg << DC_ULI_PHY_REG_SHIFT) & DC_ULI_PHY_REG_MASK) |
((data << DC_ULI_PHY_DATA_SHIFT) & DC_ULI_PHY_DATA_MASK) |
DC_ULI_PHY_OP_WRITE);
DELAY(1);
return (0);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (DC_IS_COMET(sc)) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
switch (reg) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case MII_BMCR:
phy_reg = DC_AL_BMCR;
break;
case MII_BMSR:
phy_reg = DC_AL_BMSR;
break;
case MII_PHYIDR1:
phy_reg = DC_AL_VENID;
break;
case MII_PHYIDR2:
phy_reg = DC_AL_DEVID;
break;
case MII_ANAR:
phy_reg = DC_AL_ANAR;
break;
case MII_ANLPAR:
phy_reg = DC_AL_LPAR;
break;
case MII_ANER:
phy_reg = DC_AL_ANER;
break;
default:
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(dev, "phy_write: bad phy register %x\n",
reg);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
CSR_WRITE_4(sc, phy_reg, data);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Bug fixes that I've put together while working on a project in the office: if_vr: handle the case where vr_encap() returns failure: bust out of the packet sending loop instead of panicking. Also add some missing newlines to some printf()s. if_dc: The miibus_read and miibus_write methods keep swapping in and out of MII mode by fiddling with CSR6 for cards with MII PHYs. This is a hack to support the original Macronix 98713 card which has built-in NWAY that uses an MII-like management interface even though it uses serial transceivers. Conditionalize this so that we only do this on 98713 chips, since it does bad things to genuine tulip chips (and maybe other clones).
2001-01-19 23:55:07 +00:00
if (sc->dc_type == DC_TYPE_98713) {
phy_reg = CSR_READ_4(sc, DC_NETCFG);
CSR_WRITE_4(sc, DC_NETCFG, phy_reg & ~DC_NETCFG_PORTSEL);
}
- Import the common MII bitbang'ing code from NetBSD and convert drivers to take advantage of it instead of duplicating it. This reduces the size of the i386 GENERIC kernel by about 4k. The only potential in-tree user left unconverted is xe(4), which generally should be changed to use miibus(4) instead of implementing PHY handling on its own, as otherwise it makes not much sense to add a dependency on miibus(4)/mii_bitbang(4) to xe(4) just for the MII bitbang'ing code. The common MII bitbang'ing code also is useful in the embedded space for using GPIO pins to implement MII access. - Based on lessons learnt with dc(4) (see r185750), add bus barriers to the MII bitbang read and write functions of the other drivers converted in order to ensure the intended ordering. Given that register access via an index register as well as register bank/window switching is subject to the same problem, also add bus barriers to the respective functions of smc(4), tl(4) and xl(4). - Sprinkle some const. Thanks to the following testers: Andrew Bliznak (nge(4)), nwhitehorn@ (bm(4)), yongari@ (sis(4) and ste(4)) Thanks to Hans-Joerg Sirtl for supplying hardware to test stge(4). Reviewed by: yongari (subset of drivers) Obtained from: NetBSD (partially)
2011-11-01 16:13:59 +00:00
mii_bitbang_writereg(dev, &dc_mii_bitbang_ops, phy, reg, data);
Bug fixes that I've put together while working on a project in the office: if_vr: handle the case where vr_encap() returns failure: bust out of the packet sending loop instead of panicking. Also add some missing newlines to some printf()s. if_dc: The miibus_read and miibus_write methods keep swapping in and out of MII mode by fiddling with CSR6 for cards with MII PHYs. This is a hack to support the original Macronix 98713 card which has built-in NWAY that uses an MII-like management interface even though it uses serial transceivers. Conditionalize this so that we only do this on 98713 chips, since it does bad things to genuine tulip chips (and maybe other clones).
2001-01-19 23:55:07 +00:00
if (sc->dc_type == DC_TYPE_98713)
CSR_WRITE_4(sc, DC_NETCFG, phy_reg);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_miibus_statchg(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
struct ifnet *ifp;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct mii_data *mii;
struct ifmedia *ifm;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii = device_get_softc(sc->dc_miibus);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
ifp = sc->dc_ifp;
if (mii == NULL || ifp == NULL ||
(ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
return;
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
ifm = &mii->mii_media;
Partially revert r218788. r218788 removed calling dc_setcfg() for !DC_IS_ADMTEK in dc_miibus_statchg(). This change broke link establishment of Intel 21143 with dcphy(4) where it stuck in "ability detect" state without completing auto-negotiation. Also nuke dc_if_media as it's not actually used. Submitted by: marius
2011-11-18 19:38:19 +00:00
if (DC_IS_DAVICOM(sc) && IFM_SUBTYPE(ifm->ifm_media) == IFM_HPNA_1) {
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
dc_setcfg(sc, ifm->ifm_media);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
return;
Partially revert r218788. r218788 removed calling dc_setcfg() for !DC_IS_ADMTEK in dc_miibus_statchg(). This change broke link establishment of Intel 21143 with dcphy(4) where it stuck in "ability detect" state without completing auto-negotiation. Also nuke dc_if_media as it's not actually used. Submitted by: marius
2011-11-18 19:38:19 +00:00
} else if (!DC_IS_ADMTEK(sc))
dc_setcfg(sc, mii->mii_media_active);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
sc->dc_link = 0;
if ((mii->mii_media_status & (IFM_ACTIVE | IFM_AVALID)) ==
(IFM_ACTIVE | IFM_AVALID)) {
switch (IFM_SUBTYPE(mii->mii_media_active)) {
case IFM_10_T:
case IFM_100_TX:
sc->dc_link = 1;
break;
}
}
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
}
/*
* Special support for DM9102A cards with HomePNA PHYs. Note:
* with the Davicom DM9102A/DM9801 eval board that I have, it seems
* to be impossible to talk to the management interface of the DM9801
* PHY (its MDIO pin is not connected to anything). Consequently,
* the driver has to just 'know' about the additional mode and deal
* with it itself. *sigh*
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_miibus_mediainit(device_t dev)
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct mii_data *mii;
struct ifmedia *ifm;
int rev;
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
rev = pci_get_revid(dev);
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
sc = device_get_softc(dev);
mii = device_get_softc(sc->dc_miibus);
ifm = &mii->mii_media;
if (DC_IS_DAVICOM(sc) && rev >= DC_REVISION_DM9102A)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
ifmedia_add(ifm, IFM_ETHER | IFM_HPNA_1, 0, NULL);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_BITS_512 9
#define DC_BITS_128 7
#define DC_BITS_64 6
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
reconst poison re-de u_intXX_t
2003-11-14 05:45:39 +00:00
static uint32_t
dc_mchash_le(struct dc_softc *sc, const uint8_t *addr)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
reconst poison re-de u_intXX_t
2003-11-14 05:45:39 +00:00
uint32_t crc;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Apply patch to the dc driver to handle Macronix MX98715AEC-C/D/E chips, which differ slightly from the Macronix MX98715AEC chip on the sample adapter that I have in that the multicast hash table is only 128 bits wide instead of 512. New adapters are popping up with this chip, and due to improper handling of the smaller hash table, broadcast packets were not being received correctly.
2000-07-15 17:54:30 +00:00
/*
* The hash table on the PNIC II and the MX98715AEC-C/D/E
* chips is only 128 bits wide.
*/
if (sc->dc_flags & DC_128BIT_HASH)
return (crc & ((1 << DC_BITS_128) - 1));
/* The hash table on the MX98715BEC is only 64 bits wide. */
if (sc->dc_flags & DC_64BIT_HASH)
return (crc & ((1 << DC_BITS_64) - 1));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
/* Xircom's hash filtering table is different (read: weird) */
/* Xircom uses the LEAST significant bits */
if (DC_IS_XIRCOM(sc)) {
if ((crc & 0x180) == 0x180)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return ((crc & 0x0F) + (crc & 0x70) * 3 + (14 << 4));
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
else
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return ((crc & 0x1F) + ((crc >> 1) & 0xF0) * 3 +
(12 << 4));
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
}
Apply patch to the dc driver to handle Macronix MX98715AEC-C/D/E chips, which differ slightly from the Macronix MX98715AEC chip on the sample adapter that I have in that the multicast hash table is only 128 bits wide instead of 512. New adapters are popping up with this chip, and due to improper handling of the smaller hash table, broadcast packets were not being received correctly.
2000-07-15 17:54:30 +00:00
return (crc & ((1 << DC_BITS_512) - 1));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Calculate CRC of a multicast group address, return the lower 6 bits.
*/
reconst poison re-de u_intXX_t
2003-11-14 05:45:39 +00:00
static uint32_t
dc_mchash_be(const uint8_t *addr)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Replace handrolled CRC calculation with ether_crc32_[lb]e().
2004-06-09 14:34:04 +00:00
uint32_t crc;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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_be(addr, ETHER_ADDR_LEN);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* Return the filter bit position. */
return ((crc >> 26) & 0x0000003F);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* 21143-style RX filter setup routine. Filter programming is done by
* downloading a special setup frame into the TX engine. 21143, Macronix,
* PNIC, PNIC II and Davicom chips are programmed this way.
*
* We always program the chip using 'hash perfect' mode, i.e. one perfect
* address (our node address) and a 512-bit hash filter for multicast
* frames. We also sneak the broadcast address into the hash filter since
* we need that too.
*/
Be consistent about functions being static. Spotted by: FlexeLint.
2002-10-16 09:04:52 +00:00
static void
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setfilt_21143(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
uint16_t eaddr[(ETHER_ADDR_LEN+1)/2];
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_desc *sframe;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t h, *sp;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifmultiaddr *ifma;
struct ifnet *ifp;
int i;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
i = sc->dc_cdata.dc_tx_prod;
DC_INC(sc->dc_cdata.dc_tx_prod, DC_TX_LIST_CNT);
sc->dc_cdata.dc_tx_cnt++;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sframe = &sc->dc_ldata.dc_tx_list[i];
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
sp = sc->dc_cdata.dc_sbuf;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
bzero(sp, DC_SFRAME_LEN);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sframe->dc_data = htole32(DC_ADDR_LO(sc->dc_saddr));
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sframe->dc_ctl = htole32(DC_SFRAME_LEN | DC_TXCTL_SETUP |
DC_TXCTL_TLINK | DC_FILTER_HASHPERF | DC_TXCTL_FINT);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
sc->dc_cdata.dc_tx_chain[i] = (struct mbuf *)sc->dc_cdata.dc_sbuf;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* If we want promiscuous mode, set the allframes bit. */
if (ifp->if_flags & IFF_PROMISC)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
if (ifp->if_flags & IFF_ALLMULTI)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
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);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
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
h = dc_mchash_le(sc,
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sp[h >> 4] |= htole32(1 << (h & 0xF));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (ifp->if_flags & IFF_BROADCAST) {
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
h = dc_mchash_le(sc, ifp->if_broadcastaddr);
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sp[h >> 4] |= htole32(1 << (h & 0xF));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
/* Set our MAC address. */
bcopy(IF_LLADDR(sc->dc_ifp), eaddr, ETHER_ADDR_LEN);
sp[39] = DC_SP_MAC(eaddr[0]);
sp[40] = DC_SP_MAC(eaddr[1]);
sp[41] = DC_SP_MAC(eaddr[2]);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sframe->dc_status = htole32(DC_TXSTAT_OWN);
Add missing bus_dmamap_sync() in setup frame transmit. MFC after: 3 days
2011-10-24 17:09:22 +00:00
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap, BUS_DMASYNC_PREREAD |
BUS_DMASYNC_PREWRITE);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_stag, sc->dc_smap, BUS_DMASYNC_PREWRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
/*
* The PNIC takes an exceedingly long time to process its
* setup frame; wait 10ms after posting the setup frame
* before proceeding, just so it has time to swallow its
* medicine.
*/
DELAY(10000);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
sc->dc_wdog_timer = 5;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Be consistent about functions being static. Spotted by: FlexeLint.
2002-10-16 09:04:52 +00:00
static void
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setfilt_admtek(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Ethernet hardware address stored in DC_AL_PAR0/DC_AL_PAR1 register is in little endian form. Likewise setting DC_AL_PAR0/DC_AL_PAR1 register expect the address to be in little endian form. For big endian architectures the address should be swapped to get correct one. Change setting/getting ethernet hardware address to big endian architecture frendly. Reported by: Robert Murillo ( billypilgrim782001 at yahoo dot com ) Tested by: Robert Murillo ( billypilgrim782001 at yahoo dot com )
2008-06-08 02:52:26 +00:00
uint8_t eaddr[ETHER_ADDR_LEN];
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
struct ifmultiaddr *ifma;
int h = 0;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t hashes[2] = { 0, 0 };
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* Init our MAC address. */
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
bcopy(IF_LLADDR(sc->dc_ifp), eaddr, ETHER_ADDR_LEN);
Ethernet hardware address stored in DC_AL_PAR0/DC_AL_PAR1 register is in little endian form. Likewise setting DC_AL_PAR0/DC_AL_PAR1 register expect the address to be in little endian form. For big endian architectures the address should be swapped to get correct one. Change setting/getting ethernet hardware address to big endian architecture frendly. Reported by: Robert Murillo ( billypilgrim782001 at yahoo dot com ) Tested by: Robert Murillo ( billypilgrim782001 at yahoo dot com )
2008-06-08 02:52:26 +00:00
CSR_WRITE_4(sc, DC_AL_PAR0, eaddr[3] << 24 | eaddr[2] << 16 |
eaddr[1] << 8 | eaddr[0]);
CSR_WRITE_4(sc, DC_AL_PAR1, eaddr[5] << 8 | eaddr[4]);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* If we want promiscuous mode, set the allframes bit. */
if (ifp->if_flags & IFF_PROMISC)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
if (ifp->if_flags & IFF_ALLMULTI)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* First, zot all the existing hash bits. */
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_AL_MAR0, 0);
CSR_WRITE_4(sc, DC_AL_MAR1, 0);
/*
* If we're already in promisc or allmulti mode, we
* don't have to bother programming the multicast filter.
*/
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (ifp->if_flags & (IFF_PROMISC | IFF_ALLMULTI))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
return;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* Now program new ones. */
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);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
Use only a 64bit hash filter table for ADM-Centaur cards like the Accton EN2242 and the ADMtek AN985 cards. PR: 32699 Submitted by: Jean-Luc Richier <Jean-Luc.Richier@imag.fr> Reviewed by: phk MFC after: 2 weeks
2003-05-01 09:31:01 +00:00
if (DC_IS_CENTAUR(sc))
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
h = dc_mchash_le(sc,
LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
Use only a 64bit hash filter table for ADM-Centaur cards like the Accton EN2242 and the ADMtek AN985 cards. PR: 32699 Submitted by: Jean-Luc Richier <Jean-Luc.Richier@imag.fr> Reviewed by: phk MFC after: 2 weeks
2003-05-01 09:31:01 +00:00
else
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
h = dc_mchash_be(
LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (h < 32)
hashes[0] |= (1 << h);
else
hashes[1] |= (1 << (h - 32));
}
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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_AL_MAR0, hashes[0]);
CSR_WRITE_4(sc, DC_AL_MAR1, hashes[1]);
}
Be consistent about functions being static. Spotted by: FlexeLint.
2002-10-16 09:04:52 +00:00
static void
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setfilt_asix(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
uint32_t eaddr[(ETHER_ADDR_LEN+3)/4];
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
struct ifmultiaddr *ifma;
int h = 0;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t hashes[2] = { 0, 0 };
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
/* Init our MAC address. */
bcopy(IF_LLADDR(sc->dc_ifp), eaddr, ETHER_ADDR_LEN);
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_PAR0);
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
CSR_WRITE_4(sc, DC_AX_FILTDATA, eaddr[0]);
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_PAR1);
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
CSR_WRITE_4(sc, DC_AX_FILTDATA, eaddr[1]);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* If we want promiscuous mode, set the allframes bit. */
if (ifp->if_flags & IFF_PROMISC)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
if (ifp->if_flags & IFF_ALLMULTI)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
/*
* The ASIX chip has a special bit to enable reception
* of broadcast frames.
*/
if (ifp->if_flags & IFF_BROADCAST)
DC_SETBIT(sc, DC_NETCFG, DC_AX_NETCFG_RX_BROAD);
else
DC_CLRBIT(sc, DC_NETCFG, DC_AX_NETCFG_RX_BROAD);
/* first, zot all the existing hash bits */
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_MAR0);
CSR_WRITE_4(sc, DC_AX_FILTDATA, 0);
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_MAR1);
CSR_WRITE_4(sc, DC_AX_FILTDATA, 0);
/*
* If we're already in promisc or allmulti mode, we
* don't have to bother programming the multicast filter.
*/
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (ifp->if_flags & (IFF_PROMISC | IFF_ALLMULTI))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
return;
/* now program new ones */
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);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
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
h = dc_mchash_be(LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (h < 32)
hashes[0] |= (1 << h);
else
hashes[1] |= (1 << (h - 32));
}
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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_MAR0);
CSR_WRITE_4(sc, DC_AX_FILTDATA, hashes[0]);
CSR_WRITE_4(sc, DC_AX_FILTIDX, DC_AX_FILTIDX_MAR1);
CSR_WRITE_4(sc, DC_AX_FILTDATA, hashes[1]);
}
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
static void
dc_setfilt_uli(struct dc_softc *sc)
{
uint8_t eaddr[ETHER_ADDR_LEN];
struct ifnet *ifp;
struct ifmultiaddr *ifma;
struct dc_desc *sframe;
uint32_t filter, *sp;
uint8_t *ma;
int i, mcnt;
ifp = sc->dc_ifp;
i = sc->dc_cdata.dc_tx_prod;
DC_INC(sc->dc_cdata.dc_tx_prod, DC_TX_LIST_CNT);
sc->dc_cdata.dc_tx_cnt++;
sframe = &sc->dc_ldata.dc_tx_list[i];
sp = sc->dc_cdata.dc_sbuf;
bzero(sp, DC_SFRAME_LEN);
sframe->dc_data = htole32(DC_ADDR_LO(sc->dc_saddr));
sframe->dc_ctl = htole32(DC_SFRAME_LEN | DC_TXCTL_SETUP |
DC_TXCTL_TLINK | DC_FILTER_PERFECT | DC_TXCTL_FINT);
sc->dc_cdata.dc_tx_chain[i] = (struct mbuf *)sc->dc_cdata.dc_sbuf;
/* Set station address. */
bcopy(IF_LLADDR(sc->dc_ifp), eaddr, ETHER_ADDR_LEN);
*sp++ = DC_SP_MAC(eaddr[1] << 8 | eaddr[0]);
*sp++ = DC_SP_MAC(eaddr[3] << 8 | eaddr[2]);
*sp++ = DC_SP_MAC(eaddr[5] << 8 | eaddr[4]);
/* Set broadcast address. */
*sp++ = DC_SP_MAC(0xFFFF);
*sp++ = DC_SP_MAC(0xFFFF);
*sp++ = DC_SP_MAC(0xFFFF);
/* Extract current filter configuration. */
filter = CSR_READ_4(sc, DC_NETCFG);
filter &= ~(DC_NETCFG_RX_PROMISC | DC_NETCFG_RX_ALLMULTI);
/* Now build perfect filters. */
mcnt = 0;
if_maddr_rlock(ifp);
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
if (mcnt >= DC_ULI_FILTER_NPERF) {
filter |= DC_NETCFG_RX_ALLMULTI;
break;
}
ma = LLADDR((struct sockaddr_dl *)ifma->ifma_addr);
*sp++ = DC_SP_MAC(ma[1] << 8 | ma[0]);
*sp++ = DC_SP_MAC(ma[3] << 8 | ma[2]);
*sp++ = DC_SP_MAC(ma[5] << 8 | ma[4]);
mcnt++;
}
if_maddr_runlock(ifp);
for (; mcnt < DC_ULI_FILTER_NPERF; mcnt++) {
*sp++ = DC_SP_MAC(0xFFFF);
*sp++ = DC_SP_MAC(0xFFFF);
*sp++ = DC_SP_MAC(0xFFFF);
}
if (filter & (DC_NETCFG_TX_ON | DC_NETCFG_RX_ON))
CSR_WRITE_4(sc, DC_NETCFG,
filter & ~(DC_NETCFG_TX_ON | DC_NETCFG_RX_ON));
if (ifp->if_flags & IFF_PROMISC)
filter |= DC_NETCFG_RX_PROMISC | DC_NETCFG_RX_ALLMULTI;
if (ifp->if_flags & IFF_ALLMULTI)
filter |= DC_NETCFG_RX_ALLMULTI;
CSR_WRITE_4(sc, DC_NETCFG,
filter & ~(DC_NETCFG_TX_ON | DC_NETCFG_RX_ON));
if (filter & (DC_NETCFG_TX_ON | DC_NETCFG_RX_ON))
CSR_WRITE_4(sc, DC_NETCFG, filter);
sframe->dc_status = htole32(DC_TXSTAT_OWN);
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap, BUS_DMASYNC_PREREAD |
BUS_DMASYNC_PREWRITE);
bus_dmamap_sync(sc->dc_stag, sc->dc_smap, BUS_DMASYNC_PREWRITE);
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
/*
* Wait some time...
*/
DELAY(1000);
sc->dc_wdog_timer = 5;
}
Be consistent about functions being static. Spotted by: FlexeLint.
2002-10-16 09:04:52 +00:00
static void
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setfilt_xircom(struct dc_softc *sc)
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
{
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
uint16_t eaddr[(ETHER_ADDR_LEN+1)/2];
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
struct ifmultiaddr *ifma;
struct dc_desc *sframe;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t h, *sp;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +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->dc_ifp;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_CLRBIT(sc, DC_NETCFG, (DC_NETCFG_TX_ON | DC_NETCFG_RX_ON));
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
i = sc->dc_cdata.dc_tx_prod;
DC_INC(sc->dc_cdata.dc_tx_prod, DC_TX_LIST_CNT);
sc->dc_cdata.dc_tx_cnt++;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sframe = &sc->dc_ldata.dc_tx_list[i];
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
sp = sc->dc_cdata.dc_sbuf;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
bzero(sp, DC_SFRAME_LEN);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sframe->dc_data = htole32(DC_ADDR_LO(sc->dc_saddr));
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sframe->dc_ctl = htole32(DC_SFRAME_LEN | DC_TXCTL_SETUP |
DC_TXCTL_TLINK | DC_FILTER_HASHPERF | DC_TXCTL_FINT);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
sc->dc_cdata.dc_tx_chain[i] = (struct mbuf *)sc->dc_cdata.dc_sbuf;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
/* If we want promiscuous mode, set the allframes bit. */
if (ifp->if_flags & IFF_PROMISC)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_PROMISC);
if (ifp->if_flags & IFF_ALLMULTI)
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
else
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI);
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);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
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
h = dc_mchash_le(sc,
Add PCI IDs for some additional cardbus cards. Yes, there really is a RealTek 8139 cardbus device. Unfortunately it doesn't quite work yet because the CIS parser barfs on it. Submitted by msmith, with some small tweaks by me.
2000-10-28 09:00:20 +00:00
LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sp[h >> 4] |= htole32(1 << (h & 0xF));
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +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);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (ifp->if_flags & IFF_BROADCAST) {
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
h = dc_mchash_le(sc, ifp->if_broadcastaddr);
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sp[h >> 4] |= htole32(1 << (h & 0xF));
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
}
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
/* Set our MAC address. */
bcopy(IF_LLADDR(sc->dc_ifp), eaddr, ETHER_ADDR_LEN);
sp[0] = DC_SP_MAC(eaddr[0]);
sp[1] = DC_SP_MAC(eaddr[1]);
sp[2] = DC_SP_MAC(eaddr[2]);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_TX_ON);
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ON);
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
sframe->dc_status = htole32(DC_TXSTAT_OWN);
Add missing bus_dmamap_sync() in setup frame transmit. MFC after: 3 days
2011-10-24 17:09:22 +00:00
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap, BUS_DMASYNC_PREREAD |
BUS_DMASYNC_PREWRITE);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_stag, sc->dc_smap, BUS_DMASYNC_PREWRITE);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
/*
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
* Wait some time...
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
*/
DELAY(1000);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
sc->dc_wdog_timer = 5;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setfilt(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (DC_IS_INTEL(sc) || DC_IS_MACRONIX(sc) || DC_IS_PNIC(sc) ||
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
DC_IS_PNICII(sc) || DC_IS_DAVICOM(sc) || DC_IS_CONEXANT(sc))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
dc_setfilt_21143(sc);
if (DC_IS_ASIX(sc))
dc_setfilt_asix(sc);
if (DC_IS_ADMTEK(sc))
dc_setfilt_admtek(sc);
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
if (DC_IS_ULI(sc))
dc_setfilt_uli(sc);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (DC_IS_XIRCOM(sc))
dc_setfilt_xircom(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
static void
dc_netcfg_wait(struct dc_softc *sc)
{
uint32_t isr;
int i;
for (i = 0; i < DC_TIMEOUT; i++) {
isr = CSR_READ_4(sc, DC_ISR);
if (isr & DC_ISR_TX_IDLE &&
((isr & DC_ISR_RX_STATE) == DC_RXSTATE_STOPPED ||
(isr & DC_ISR_RX_STATE) == DC_RXSTATE_WAIT))
break;
DELAY(10);
}
Supress command completion failure warning when the card isn't present. Only call the bus to check if we actually do timeout so we don't affect the normal case (since this case needn't be optimized and this guards against all races).
2011-06-21 20:52:55 +00:00
if (i == DC_TIMEOUT && bus_child_present(sc->dc_dev)) {
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
if (!(isr & DC_ISR_TX_IDLE) && !DC_IS_ASIX(sc))
device_printf(sc->dc_dev,
"%s: failed to force tx to idle state\n", __func__);
if (!((isr & DC_ISR_RX_STATE) == DC_RXSTATE_STOPPED ||
(isr & DC_ISR_RX_STATE) == DC_RXSTATE_WAIT) &&
!DC_HAS_BROKEN_RXSTATE(sc))
device_printf(sc->dc_dev,
"%s: failed to force rx to idle state\n", __func__);
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
* In order to fiddle with the 'full-duplex' and '100Mbps' bits in
* the netconfig register, we first have to put the transmit and/or
* receive logic in the idle state.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_setcfg(struct dc_softc *sc, int media)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
int restart = 0, watchdogreg;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (IFM_SUBTYPE(media) == IFM_NONE)
return;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (CSR_READ_4(sc, DC_NETCFG) & (DC_NETCFG_TX_ON | DC_NETCFG_RX_ON)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
restart = 1;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_CLRBIT(sc, DC_NETCFG, (DC_NETCFG_TX_ON | DC_NETCFG_RX_ON));
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
dc_netcfg_wait(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
if (IFM_SUBTYPE(media) == IFM_100_TX) {
When I tweaked if_dc.c to alter the polling interval for non-MII 21143 chips, I accidentally removed the DC_MII_REDUCED_POLL flag for all 21143 cards. This caused problems with timer-instigated TCP retransmits, which happened to occur at the same time as an MII poll tick on MII-based cards (e.g. D-Link DFE-570TX). Fixed this, plus made some other cleanups. The autoneg fixes for the non-MII cards still work. Also tested the PNIC II now that I have one again.
2000-06-07 17:07:44 +00:00
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_SPEEDSEL);
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_HEARTBEAT);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_pmode == DC_PMODE_MII) {
Close PR 20438. Make fix for preserving LED settings conditional on presence Intel 21143 chip.
2000-08-07 17:03:20 +00:00
if (DC_IS_INTEL(sc)) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* There's a write enable bit here that reads as 1. */
Close PR 20438. Make fix for preserving LED settings conditional on presence Intel 21143 chip.
2000-08-07 17:03:20 +00:00
watchdogreg = CSR_READ_4(sc, DC_WATCHDOG);
watchdogreg &= ~DC_WDOG_CTLWREN;
watchdogreg |= DC_WDOG_JABBERDIS;
Clean up a few things in dc_setcfg() pointed out to be me by aaron@openbsd.com on IRC earlier today.
2000-10-14 00:40:14 +00:00
CSR_WRITE_4(sc, DC_WATCHDOG, watchdogreg);
Close PR 20438. Make fix for preserving LED settings conditional on presence Intel 21143 chip.
2000-08-07 17:03:20 +00:00
} else {
DC_SETBIT(sc, DC_WATCHDOG, DC_WDOG_JABBERDIS);
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_CLRBIT(sc, DC_NETCFG, (DC_NETCFG_PCS |
DC_NETCFG_PORTSEL | DC_NETCFG_SCRAMBLER));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_type == DC_TYPE_98713)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_SETBIT(sc, DC_NETCFG, (DC_NETCFG_PCS |
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_NETCFG_SCRAMBLER));
Add support for the Davicom DM9102A 10/100 ethernet controller chip. This is just to make sure we initialize the chip correctly: we need to make the sure the port select bit in CSR6 is set properly so that we use the internal PHY for 10/100 support. (The eval boards I have also include an external HomePNA PHY, but I need to play with that more before I can support it.)
2000-01-19 19:03:08 +00:00
if (!DC_IS_DAVICOM(sc))
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_CLRBIT(sc, DC_10BTCTRL, 0xFFFF);
} else {
if (DC_IS_PNIC(sc)) {
DC_PN_GPIO_SETBIT(sc, DC_PN_GPIO_SPEEDSEL);
DC_PN_GPIO_SETBIT(sc, DC_PN_GPIO_100TX_LOOP);
DC_SETBIT(sc, DC_PN_NWAY, DC_PN_NWAY_SPEEDSEL);
}
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PCS);
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_SCRAMBLER);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
if (IFM_SUBTYPE(media) == IFM_10_T) {
When I tweaked if_dc.c to alter the polling interval for non-MII 21143 chips, I accidentally removed the DC_MII_REDUCED_POLL flag for all 21143 cards. This caused problems with timer-instigated TCP retransmits, which happened to occur at the same time as an MII poll tick on MII-based cards (e.g. D-Link DFE-570TX). Fixed this, plus made some other cleanups. The autoneg fixes for the non-MII cards still work. Also tested the PNIC II now that I have one again.
2000-06-07 17:07:44 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_SPEEDSEL);
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_HEARTBEAT);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_pmode == DC_PMODE_MII) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* There's a write enable bit here that reads as 1. */
Clean up a few things in dc_setcfg() pointed out to be me by aaron@openbsd.com on IRC earlier today.
2000-10-14 00:40:14 +00:00
if (DC_IS_INTEL(sc)) {
watchdogreg = CSR_READ_4(sc, DC_WATCHDOG);
watchdogreg &= ~DC_WDOG_CTLWREN;
watchdogreg |= DC_WDOG_JABBERDIS;
CSR_WRITE_4(sc, DC_WATCHDOG, watchdogreg);
} else {
DC_SETBIT(sc, DC_WATCHDOG, DC_WDOG_JABBERDIS);
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_CLRBIT(sc, DC_NETCFG, (DC_NETCFG_PCS |
DC_NETCFG_PORTSEL | DC_NETCFG_SCRAMBLER));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_type == DC_TYPE_98713)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PCS);
Add support for the Davicom DM9102A 10/100 ethernet controller chip. This is just to make sure we initialize the chip correctly: we need to make the sure the port select bit in CSR6 is set properly so that we use the internal PHY for 10/100 support. (The eval boards I have also include an external HomePNA PHY, but I need to play with that more before I can support it.)
2000-01-19 19:03:08 +00:00
if (!DC_IS_DAVICOM(sc))
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_CLRBIT(sc, DC_10BTCTRL, 0xFFFF);
} else {
if (DC_IS_PNIC(sc)) {
DC_PN_GPIO_CLRBIT(sc, DC_PN_GPIO_SPEEDSEL);
DC_PN_GPIO_SETBIT(sc, DC_PN_GPIO_100TX_LOOP);
DC_CLRBIT(sc, DC_PN_NWAY, DC_PN_NWAY_SPEEDSEL);
}
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_PCS);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_SCRAMBLER);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
if (DC_IS_INTEL(sc)) {
DC_CLRBIT(sc, DC_SIARESET, DC_SIA_RESET);
DC_CLRBIT(sc, DC_10BTCTRL, 0xFFFF);
if ((media & IFM_GMASK) == IFM_FDX)
DC_SETBIT(sc, DC_10BTCTRL, 0x7F3D);
else
DC_SETBIT(sc, DC_10BTCTRL, 0x7F3F);
DC_SETBIT(sc, DC_SIARESET, DC_SIA_RESET);
DC_CLRBIT(sc, DC_10BTCTRL,
DC_TCTL_AUTONEGENBL);
DELAY(20000);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
/*
* If this is a Davicom DM9102A card with a DM9801 HomePNA
* PHY and we want HomePNA mode, set the portsel bit to turn
* on the external MII port.
*/
if (DC_IS_DAVICOM(sc)) {
Move us yet closer to IFM_* definitions in NetBSD.
2002-04-29 05:32:44 +00:00
if (IFM_SUBTYPE(media) == IFM_HPNA_1) {
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
sc->dc_link = 1;
} else {
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_PORTSEL);
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if ((media & IFM_GMASK) == IFM_FDX) {
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_FULLDUPLEX);
if (sc->dc_pmode == DC_PMODE_SYM && DC_IS_PNIC(sc))
DC_SETBIT(sc, DC_PN_NWAY, DC_PN_NWAY_DUPLEX);
} else {
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_FULLDUPLEX);
if (sc->dc_pmode == DC_PMODE_SYM && DC_IS_PNIC(sc))
DC_CLRBIT(sc, DC_PN_NWAY, DC_PN_NWAY_DUPLEX);
}
if (restart)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_TX_ON | DC_NETCFG_RX_ON);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_reset(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_SETBIT(sc, DC_BUSCTL, DC_BUSCTL_RESET);
for (i = 0; i < DC_TIMEOUT; i++) {
DELAY(10);
if (!(CSR_READ_4(sc, DC_BUSCTL) & DC_BUSCTL_RESET))
break;
}
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
if (DC_IS_ASIX(sc) || DC_IS_ADMTEK(sc) || DC_IS_CONEXANT(sc) ||
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
DC_IS_XIRCOM(sc) || DC_IS_INTEL(sc) || DC_IS_ULI(sc)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DELAY(10000);
DC_CLRBIT(sc, DC_BUSCTL, DC_BUSCTL_RESET);
i = 0;
}
if (i == DC_TIMEOUT)
- Consistently use if_printf() only in interface methods: if_start(), if_watchdog, etc., or in functions used only in these methods. In all other functions in the driver use device_printf(). - Use __func__ instead of typing function name. Submitted by: Alex Lyashkov <umka sevcity.net>
2006-09-15 15:16:12 +00:00
device_printf(sc->dc_dev, "reset never completed!\n");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Wait a little while for the chip to get its brains in order. */
DELAY(1000);
CSR_WRITE_4(sc, DC_IMR, 0x00000000);
CSR_WRITE_4(sc, DC_BUSCTL, 0x00000000);
CSR_WRITE_4(sc, DC_NETCFG, 0x00000000);
Fix some problems reported by Mike Pritchard: - Add a flag DC_TX_INTR_ALWAYS which causes the transmit code to request a TX done interrupt for every packet. The PNIC seems to need this to insure that the sent TX buffers get reaped in a timely fashion. - Try to unreset the SIA as soon as possible after resetting the whole chip. - Change dcphy to support either 10/100 or 10Mbps only NICs. The built-in 21143 ethernet in Compaq Presario machines is 10Mbps only and it doesn't work right if we try to advertise 100Mbps modes during autoneg. When restricted to only 10mbps modes, it works fine. Note that for now, I detect this condition by checking the PCI subsystem ID on this NIC (which has a Compaq vendor/device ID). Yes, I know that's what the SROM is supposed to be for. I'm deliberately ignoring the SROM wherever possible. Sue me. The latter two fixes allow if_dc to work correctly with the built-in ethernet on certain Compaq Presario boxes. There are liable to be quite a few people using these as their home systems who might want to try FreeBSD; may as well be nice to them. Now if anybody out there has an Alpha miata with 10Mbps ethernet and can show me the output from pciconf -l on their system, I'd be grateful.
1999-12-13 21:45:13 +00:00
/*
* Bring the SIA out of reset. In some cases, it looks
* like failing to unreset the SIA soon enough gets it
* into a state where it will never come out of reset
* until we reset the whole chip again.
*/
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
if (DC_IS_INTEL(sc)) {
Fix some problems reported by Mike Pritchard: - Add a flag DC_TX_INTR_ALWAYS which causes the transmit code to request a TX done interrupt for every packet. The PNIC seems to need this to insure that the sent TX buffers get reaped in a timely fashion. - Try to unreset the SIA as soon as possible after resetting the whole chip. - Change dcphy to support either 10/100 or 10Mbps only NICs. The built-in 21143 ethernet in Compaq Presario machines is 10Mbps only and it doesn't work right if we try to advertise 100Mbps modes during autoneg. When restricted to only 10mbps modes, it works fine. Note that for now, I detect this condition by checking the PCI subsystem ID on this NIC (which has a Compaq vendor/device ID). Yes, I know that's what the SROM is supposed to be for. I'm deliberately ignoring the SROM wherever possible. Sue me. The latter two fixes allow if_dc to work correctly with the built-in ethernet on certain Compaq Presario boxes. There are liable to be quite a few people using these as their home systems who might want to try FreeBSD; may as well be nice to them. Now if anybody out there has an Alpha miata with 10Mbps ethernet and can show me the output from pciconf -l on their system, I'd be grateful.
1999-12-13 21:45:13 +00:00
DC_SETBIT(sc, DC_SIARESET, DC_SIA_RESET);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
CSR_WRITE_4(sc, DC_10BTCTRL, 0xFFFFFFFF);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
CSR_WRITE_4(sc, DC_WATCHDOG, 0);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
static const struct dc_type *
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_devtype(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
const struct dc_type *t;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t devid;
uint8_t rev;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
t = dc_devs;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
devid = pci_get_devid(dev);
rev = pci_get_revid(dev);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
while (t->dc_name != NULL) {
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
if (devid == t->dc_devid && rev >= t->dc_minrev)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (t);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
t++;
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (NULL);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Probe for a 21143 or clone chip. Check the PCI vendor and device
* IDs against our list and return a device name if we find a match.
* We do a little bit of extra work to identify the exact type of
* chip. The MX98713 and MX98713A have the same PCI vendor/device ID,
* but different revision IDs. The same is true for 98715/98715A
* chips and the 98725, as well as the ASIX and ADMtek chips. In some
* cases, the exact chip revision affects driver behavior.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_probe(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
const struct dc_type *t;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
t = dc_devtype(dev);
if (t != NULL) {
device_set_desc(dev, t->dc_name);
Return BUS_PROBE_DEFAULT instead of 0.
2005-02-24 21:32:56 +00:00
return (BUS_PROBE_DEFAULT);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (ENXIO);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_apply_fixup(struct dc_softc *sc, int media)
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_mediainfo *m;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint8_t *p;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
int i;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t reg;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m = sc->dc_mi;
while (m != NULL) {
if (m->dc_media == media)
break;
m = m->dc_next;
}
if (m == NULL)
return;
for (i = 0, p = m->dc_reset_ptr; i < m->dc_reset_len; i++, p += 2) {
reg = (p[0] | (p[1] << 8)) << 16;
CSR_WRITE_4(sc, DC_WATCHDOG, reg);
}
for (i = 0, p = m->dc_gp_ptr; i < m->dc_gp_len; i++, p += 2) {
reg = (p[0] | (p[1] << 8)) << 16;
CSR_WRITE_4(sc, DC_WATCHDOG, reg);
}
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_decode_leaf_sia(struct dc_softc *sc, struct dc_eblock_sia *l)
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_mediainfo *m;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
m = malloc(sizeof(struct dc_mediainfo), M_DEVBUF, M_NOWAIT | M_ZERO);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (m == NULL) {
device_printf(sc->dc_dev, "Could not allocate mediainfo\n");
return (ENOMEM);
}
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
switch (l->dc_sia_code & ~DC_SIA_CODE_EXT) {
case DC_SIA_CODE_10BT:
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_media = IFM_10_T;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
break;
case DC_SIA_CODE_10BT_FDX:
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
m->dc_media = IFM_10_T | IFM_FDX;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
break;
case DC_SIA_CODE_10B2:
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_media = IFM_10_2;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
break;
case DC_SIA_CODE_10B5:
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_media = IFM_10_5;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
break;
default:
break;
}
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
/*
* We need to ignore CSR13, CSR14, CSR15 for SIA mode.
* Things apparently already work for cards that do
* supply Media Specific Data.
*/
if (l->dc_sia_code & DC_SIA_CODE_EXT) {
m->dc_gp_len = 2;
m->dc_gp_ptr =
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
(uint8_t *)&l->dc_un.dc_sia_ext.dc_sia_gpio_ctl;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
} else {
m->dc_gp_len = 2;
m->dc_gp_ptr =
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
(uint8_t *)&l->dc_un.dc_sia_noext.dc_sia_gpio_ctl;
Ignore CSR13, CSR14, CSR15 'Media Specific Data' registers for 21143 based cards which use SIA mode. This fixes 10mbit mode for ZNYX ZX346Q cards and other 21143 based cards. PR: 32118 Submitted by: Rene de Vries <rene@tunix.nl> Geert Jan de Groot <GeertJan.deGroot@tunix.nl> Obtained from: BSDI MFC after: 2 weeks
2003-10-05 19:57:10 +00:00
}
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_next = sc->dc_mi;
sc->dc_mi = m;
sc->dc_pmode = DC_PMODE_SIA;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
return (0);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_decode_leaf_sym(struct dc_softc *sc, struct dc_eblock_sym *l)
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_mediainfo *m;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
m = malloc(sizeof(struct dc_mediainfo), M_DEVBUF, M_NOWAIT | M_ZERO);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (m == NULL) {
device_printf(sc->dc_dev, "Could not allocate mediainfo\n");
return (ENOMEM);
}
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
if (l->dc_sym_code == DC_SYM_CODE_100BT)
m->dc_media = IFM_100_TX;
if (l->dc_sym_code == DC_SYM_CODE_100BT_FDX)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
m->dc_media = IFM_100_TX | IFM_FDX;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_gp_len = 2;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
m->dc_gp_ptr = (uint8_t *)&l->dc_sym_gpio_ctl;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m->dc_next = sc->dc_mi;
sc->dc_mi = m;
sc->dc_pmode = DC_PMODE_SYM;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
return (0);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_decode_leaf_mii(struct dc_softc *sc, struct dc_eblock_mii *l)
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_mediainfo *m;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint8_t *p;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
m = malloc(sizeof(struct dc_mediainfo), M_DEVBUF, M_NOWAIT | M_ZERO);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (m == NULL) {
device_printf(sc->dc_dev, "Could not allocate mediainfo\n");
return (ENOMEM);
}
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
/* We abuse IFM_AUTO to represent MII. */
m->dc_media = IFM_AUTO;
m->dc_gp_len = l->dc_gpr_len;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
p = (uint8_t *)l;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
p += sizeof(struct dc_eblock_mii);
m->dc_gp_ptr = p;
p += 2 * l->dc_gpr_len;
m->dc_reset_len = *p;
p++;
m->dc_reset_ptr = p;
m->dc_next = sc->dc_mi;
sc->dc_mi = m;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
return (0);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_read_srom(struct dc_softc *sc, int bits)
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
{
int size;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
size = DC_ROM_SIZE(bits);
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
sc->dc_srom = malloc(size, M_DEVBUF, M_NOWAIT | M_ZERO);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (sc->dc_srom == NULL) {
device_printf(sc->dc_dev, "Could not allocate SROM buffer\n");
return (ENOMEM);
}
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
dc_read_eeprom(sc, (caddr_t)sc->dc_srom, 0, (size / 2), 0);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
return (0);
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_parse_21143_srom(struct dc_softc *sc)
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_leaf_hdr *lhdr;
struct dc_eblock_hdr *hdr;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
int error, have_mii, i, loff;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
char *ptr;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
have_mii = 0;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
loff = sc->dc_srom[27];
lhdr = (struct dc_leaf_hdr *)&(sc->dc_srom[loff]);
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
ptr = (char *)lhdr;
ptr += sizeof(struct dc_leaf_hdr) - 1;
/*
* Look if we got a MII media block.
*/
for (i = 0; i < lhdr->dc_mcnt; i++) {
hdr = (struct dc_eblock_hdr *)ptr;
if (hdr->dc_type == DC_EBLOCK_MII)
have_mii++;
ptr += (hdr->dc_len & 0x7F);
ptr++;
}
/*
* Do the same thing again. Only use SIA and SYM media
* blocks if no MII media block is available.
*/
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
ptr = (char *)lhdr;
ptr += sizeof(struct dc_leaf_hdr) - 1;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = 0;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
for (i = 0; i < lhdr->dc_mcnt; i++) {
hdr = (struct dc_eblock_hdr *)ptr;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
switch (hdr->dc_type) {
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
case DC_EBLOCK_MII:
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_decode_leaf_mii(sc, (struct dc_eblock_mii *)hdr);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
break;
case DC_EBLOCK_SIA:
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
if (! have_mii)
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_decode_leaf_sia(sc,
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
(struct dc_eblock_sia *)hdr);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
break;
case DC_EBLOCK_SYM:
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
if (! have_mii)
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_decode_leaf_sym(sc,
Only use a SIA/SYM media info block if no MII block is detected. The submitter of PR 32118 told me that this patch also fixes autoselecting for znyx 4 port cards (10baseT, 100baseTX did work already). PR: 32118 Reviewed by: imp Approved by: rwatson (re)
2003-05-15 16:53:29 +00:00
(struct dc_eblock_sym *)hdr);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
break;
default:
/* Don't care. Yet. */
break;
}
ptr += (hdr->dc_len & 0x7F);
ptr++;
}
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
return (error);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
}
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
static void
dc_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nseg, int error)
{
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_addr_t *paddr;
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
KASSERT(nseg == 1,
("%s: wrong number of segments (%d)", __func__, nseg));
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
paddr = arg;
*paddr = segs->ds_addr;
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
static int
dc_dma_alloc(struct dc_softc *sc)
{
int error, i;
error = bus_dma_tag_create(bus_get_dma_tag(sc->dc_dev), 1, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
BUS_SPACE_MAXSIZE_32BIT, 0, BUS_SPACE_MAXSIZE_32BIT, 0,
NULL, NULL, &sc->dc_ptag);
if (error) {
device_printf(sc->dc_dev,
"failed to allocate parent DMA tag\n");
goto fail;
}
/* Allocate a busdma tag and DMA safe memory for TX/RX descriptors. */
error = bus_dma_tag_create(sc->dc_ptag, DC_LIST_ALIGN, 0,
BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, DC_RX_LIST_SZ, 1,
DC_RX_LIST_SZ, 0, NULL, NULL, &sc->dc_rx_ltag);
if (error) {
device_printf(sc->dc_dev, "failed to create RX list DMA tag\n");
goto fail;
}
error = bus_dma_tag_create(sc->dc_ptag, DC_LIST_ALIGN, 0,
BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL, DC_TX_LIST_SZ, 1,
DC_TX_LIST_SZ, 0, NULL, NULL, &sc->dc_tx_ltag);
if (error) {
device_printf(sc->dc_dev, "failed to create TX list DMA tag\n");
goto fail;
}
/* RX descriptor list. */
error = bus_dmamem_alloc(sc->dc_rx_ltag,
(void **)&sc->dc_ldata.dc_rx_list, BUS_DMA_NOWAIT |
BUS_DMA_ZERO | BUS_DMA_COHERENT, &sc->dc_rx_lmap);
if (error) {
device_printf(sc->dc_dev,
"failed to allocate DMA'able memory for RX list\n");
goto fail;
}
error = bus_dmamap_load(sc->dc_rx_ltag, sc->dc_rx_lmap,
sc->dc_ldata.dc_rx_list, DC_RX_LIST_SZ, dc_dma_map_addr,
&sc->dc_ldata.dc_rx_list_paddr, BUS_DMA_NOWAIT);
if (error) {
device_printf(sc->dc_dev,
"failed to load DMA'able memory for RX list\n");
goto fail;
}
/* TX descriptor list. */
error = bus_dmamem_alloc(sc->dc_tx_ltag,
(void **)&sc->dc_ldata.dc_tx_list, BUS_DMA_NOWAIT |
BUS_DMA_ZERO | BUS_DMA_COHERENT, &sc->dc_tx_lmap);
if (error) {
device_printf(sc->dc_dev,
"failed to allocate DMA'able memory for TX list\n");
goto fail;
}
error = bus_dmamap_load(sc->dc_tx_ltag, sc->dc_tx_lmap,
sc->dc_ldata.dc_tx_list, DC_TX_LIST_SZ, dc_dma_map_addr,
&sc->dc_ldata.dc_tx_list_paddr, BUS_DMA_NOWAIT);
if (error) {
device_printf(sc->dc_dev,
"cannot load DMA'able memory for TX list\n");
goto fail;
}
/*
* Allocate a busdma tag and DMA safe memory for the multicast
* setup frame.
*/
error = bus_dma_tag_create(sc->dc_ptag, DC_LIST_ALIGN, 0,
BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL,
DC_SFRAME_LEN + DC_MIN_FRAMELEN, 1, DC_SFRAME_LEN + DC_MIN_FRAMELEN,
0, NULL, NULL, &sc->dc_stag);
if (error) {
device_printf(sc->dc_dev,
"failed to create DMA tag for setup frame\n");
goto fail;
}
error = bus_dmamem_alloc(sc->dc_stag, (void **)&sc->dc_cdata.dc_sbuf,
BUS_DMA_NOWAIT, &sc->dc_smap);
if (error) {
device_printf(sc->dc_dev,
"failed to allocate DMA'able memory for setup frame\n");
goto fail;
}
error = bus_dmamap_load(sc->dc_stag, sc->dc_smap, sc->dc_cdata.dc_sbuf,
DC_SFRAME_LEN, dc_dma_map_addr, &sc->dc_saddr, BUS_DMA_NOWAIT);
if (error) {
device_printf(sc->dc_dev,
"cannot load DMA'able memory for setup frame\n");
goto fail;
}
/* Allocate a busdma tag for RX mbufs. */
error = bus_dma_tag_create(sc->dc_ptag, DC_RXBUF_ALIGN, 0,
BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL,
MCLBYTES, 1, MCLBYTES, 0, NULL, NULL, &sc->dc_rx_mtag);
if (error) {
device_printf(sc->dc_dev, "failed to create RX mbuf tag\n");
goto fail;
}
/* Allocate a busdma tag for TX mbufs. */
error = bus_dma_tag_create(sc->dc_ptag, 1, 0,
BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR, NULL, NULL,
MCLBYTES * DC_MAXFRAGS, DC_MAXFRAGS, MCLBYTES,
0, NULL, NULL, &sc->dc_tx_mtag);
if (error) {
device_printf(sc->dc_dev, "failed to create TX mbuf tag\n");
goto fail;
}
/* Create the TX/RX busdma maps. */
for (i = 0; i < DC_TX_LIST_CNT; i++) {
error = bus_dmamap_create(sc->dc_tx_mtag, 0,
&sc->dc_cdata.dc_tx_map[i]);
if (error) {
device_printf(sc->dc_dev,
"failed to create TX mbuf dmamap\n");
goto fail;
}
}
for (i = 0; i < DC_RX_LIST_CNT; i++) {
error = bus_dmamap_create(sc->dc_rx_mtag, 0,
&sc->dc_cdata.dc_rx_map[i]);
if (error) {
device_printf(sc->dc_dev,
"failed to create RX mbuf dmamap\n");
goto fail;
}
}
error = bus_dmamap_create(sc->dc_rx_mtag, 0, &sc->dc_sparemap);
if (error) {
device_printf(sc->dc_dev,
"failed to create spare RX mbuf dmamap\n");
goto fail;
}
fail:
return (error);
}
static void
dc_dma_free(struct dc_softc *sc)
{
int i;
/* RX buffers. */
if (sc->dc_rx_mtag != NULL) {
for (i = 0; i < DC_RX_LIST_CNT; i++) {
if (sc->dc_cdata.dc_rx_map[i] != NULL)
bus_dmamap_destroy(sc->dc_rx_mtag,
sc->dc_cdata.dc_rx_map[i]);
}
if (sc->dc_sparemap != NULL)
bus_dmamap_destroy(sc->dc_rx_mtag, sc->dc_sparemap);
bus_dma_tag_destroy(sc->dc_rx_mtag);
}
/* TX buffers. */
if (sc->dc_rx_mtag != NULL) {
for (i = 0; i < DC_TX_LIST_CNT; i++) {
if (sc->dc_cdata.dc_tx_map[i] != NULL)
bus_dmamap_destroy(sc->dc_tx_mtag,
sc->dc_cdata.dc_tx_map[i]);
}
bus_dma_tag_destroy(sc->dc_tx_mtag);
}
/* RX descriptor list. */
if (sc->dc_rx_ltag) {
if (sc->dc_rx_lmap != NULL)
bus_dmamap_unload(sc->dc_rx_ltag, sc->dc_rx_lmap);
if (sc->dc_rx_lmap != NULL && sc->dc_ldata.dc_rx_list != NULL)
bus_dmamem_free(sc->dc_rx_ltag, sc->dc_ldata.dc_rx_list,
sc->dc_rx_lmap);
bus_dma_tag_destroy(sc->dc_rx_ltag);
}
/* TX descriptor list. */
if (sc->dc_tx_ltag) {
if (sc->dc_tx_lmap != NULL)
bus_dmamap_unload(sc->dc_tx_ltag, sc->dc_tx_lmap);
if (sc->dc_tx_lmap != NULL && sc->dc_ldata.dc_tx_list != NULL)
bus_dmamem_free(sc->dc_tx_ltag, sc->dc_ldata.dc_tx_list,
sc->dc_tx_lmap);
bus_dma_tag_destroy(sc->dc_tx_ltag);
}
/* multicast setup frame. */
if (sc->dc_stag) {
if (sc->dc_smap != NULL)
bus_dmamap_unload(sc->dc_stag, sc->dc_smap);
if (sc->dc_smap != NULL && sc->dc_cdata.dc_sbuf != NULL)
bus_dmamem_free(sc->dc_stag, sc->dc_cdata.dc_sbuf,
sc->dc_smap);
bus_dma_tag_destroy(sc->dc_stag);
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_attach(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
uint32_t eaddr[(ETHER_ADDR_LEN+3)/4];
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t command;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct ifnet *ifp;
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
struct dc_mediainfo *m;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t reg, revision;
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
uint16_t *srom;
int error, mac_offset, n, phy, rid, tmp;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint8_t *mac;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
- Consistently use if_printf() only in interface methods: if_start(), if_watchdog, etc., or in functions used only in these methods. In all other functions in the driver use device_printf(). - Use __func__ instead of typing function name. Submitted by: Alex Lyashkov <umka sevcity.net>
2006-09-15 15:16:12 +00:00
sc->dc_dev = dev;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_mtx, device_get_nameunit(dev), MTX_NETWORK_LOCK,
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
MTX_DEF);
Remove burn bridges code that saved/restored the pci config registers that are now handled in the pci bus layer. They are no longer necessary.
2004-06-28 20:07:03 +00:00
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
rid = DC_RID;
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->dc_res = bus_alloc_resource_any(dev, DC_RES, &rid, RF_ACTIVE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_res == NULL) {
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(dev, "couldn't map ports/memory\n");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
error = ENXIO;
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
goto fail;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
sc->dc_btag = rman_get_bustag(sc->dc_res);
sc->dc_bhandle = rman_get_bushandle(sc->dc_res);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* Allocate interrupt. */
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
rid = 0;
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->dc_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid,
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
RF_SHAREABLE | RF_ACTIVE);
if (sc->dc_irq == NULL) {
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(dev, "couldn't map interrupt\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
error = ENXIO;
goto fail;
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Need this info to decide on a chip type. */
sc->dc_info = dc_devtype(dev);
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
revision = pci_get_revid(dev);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = 0;
Minor correction to comment: PNIC and XIRCOM have eeprom, its just non-standard.
2003-02-12 07:16:15 +00:00
/* Get the eeprom width, but PNIC and XIRCOM have diff eeprom */
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
if (sc->dc_info->dc_devid !=
DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C168) &&
sc->dc_info->dc_devid !=
DC_DEVID(DC_VENDORID_XIRCOM, DC_DEVICEID_X3201))
Fix the breakage resulting from Rev. 1.80. Get the eeprom width for all but two cards. This should fix broken cards like these: DM9102 (Davicom, DEVICE_ID: 0x9002) DM9009 (Davicom, DEVICE_ID: 0x9009) DM9100 (Davicom, DEVICE_ID: 0x9100) 98713/98713_CP (Macronix PMAC, DEVICE_ID: 0x0512) 98713_CP (Macronix PMAC, DEVICE_ID: 0x0512) 987x5 (Macronix PMAC, DEVICE_ID: 0x0531) 98727 (Macronix PMAC, DEVICE_ID: 0x0532) 82C115 (Lite-On PNIC II, DEVICE_ID: 0xc115) AX88140A (ASIX Dev_ID: DEVICE_ID: 0x1400) EN1217 (Accton EN1217, DEVICE_ID: 0x1217) Note that these cards sould still work in STABLE. Reviewed by: imp
2003-02-07 23:12:51 +00:00
dc_eeprom_width(sc);
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
switch (sc->dc_info->dc_devid) {
case DC_DEVID(DC_VENDORID_DEC, DC_DEVICEID_21143):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_21143;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_POLL | DC_TX_USE_TX_INTR;
When I tweaked if_dc.c to alter the polling interval for non-MII 21143 chips, I accidentally removed the DC_MII_REDUCED_POLL flag for all 21143 cards. This caused problems with timer-instigated TCP retransmits, which happened to occur at the same time as an MII poll tick on MII-based cards (e.g. D-Link DFE-570TX). Fixed this, plus made some other cleanups. The autoneg fixes for the non-MII cards still work. Also tested the PNIC II now that I have one again.
2000-06-07 17:07:44 +00:00
sc->dc_flags |= DC_REDUCED_MII_POLL;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
/* Save EEPROM contents so we can parse them later. */
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_read_srom(sc, sc->dc_romwidth);
if (error != 0)
goto fail;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9009):
case DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9100):
case DC_DEVID(DC_VENDORID_DAVICOM, DC_DEVICEID_DM9102):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_DM9102;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_COALESCE | DC_TX_INTR_ALWAYS;
sc->dc_flags |= DC_REDUCED_MII_POLL | DC_TX_STORENFWD;
All davicom cards seem to need DC_TX_ALIGN.
2003-08-27 08:13:34 +00:00
sc->dc_flags |= DC_TX_ALIGN;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_pmode = DC_PMODE_MII;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
Yet another bug fix/optimization for the Davicom DM9100/9102: increase the PCI latency timer value to 0x80. Davicom's Linux driver does this, and it drastically reduces the number of TX underruns in my tests. (Note: this is done only for the Davicom chips. I'm not sure it's a good idea to do it for all of them.) Again, still waiting on confirmation before merging to stable.
2000-10-27 00:15:04 +00:00
/* Increase the latency timer value. */
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
pci_write_config(dev, PCIR_LATTIMER, 0x80, 1);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AL981):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_AL981;
sc->dc_flags |= DC_TX_USE_TX_INTR;
sc->dc_flags |= DC_TX_ADMTEK_WAR;
sc->dc_pmode = DC_PMODE_MII;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_read_srom(sc, sc->dc_romwidth);
if (error != 0)
goto fail;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
Fix card/device names, no functional change The ADMtek AN985 is the cardbus variant of ADMtek AN983 The Netgear FA511 is just a relabled ADMtek AN985 PR: kern/50574 MFC after: 1 month
2010-01-03 13:59:59 +00:00
case DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AN983):
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_AN985):
case DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_ADM9511):
case DC_DEVID(DC_VENDORID_ADMTEK, DC_DEVICEID_ADM9513):
case DC_DEVID(DC_VENDORID_DLINK, DC_DEVICEID_DRP32TXD):
case DC_DEVID(DC_VENDORID_ABOCOM, DC_DEVICEID_FE2500):
case DC_DEVID(DC_VENDORID_ABOCOM, DC_DEVICEID_FE2500MX):
case DC_DEVID(DC_VENDORID_ACCTON, DC_DEVICEID_EN2242):
case DC_DEVID(DC_VENDORID_HAWKING, DC_DEVICEID_HAWKING_PN672TX):
case DC_DEVID(DC_VENDORID_PLANEX, DC_DEVICEID_FNW3602T):
case DC_DEVID(DC_VENDORID_3COM, DC_DEVICEID_3CSOHOB):
case DC_DEVID(DC_VENDORID_MICROSOFT, DC_DEVICEID_MSMN120):
case DC_DEVID(DC_VENDORID_MICROSOFT, DC_DEVICEID_MSMN130):
Fix the last commit. Submitted by: jhb Pointy hat to: glebius
2006-06-08 08:15:21 +00:00
case DC_DEVID(DC_VENDORID_LINKSYS, DC_DEVICEID_PCMPC200_AB08):
case DC_DEVID(DC_VENDORID_LINKSYS, DC_DEVICEID_PCMPC200_AB09):
Fix card/device names, no functional change The ADMtek AN985 is the cardbus variant of ADMtek AN983 The Netgear FA511 is just a relabled ADMtek AN985 PR: kern/50574 MFC after: 1 month
2010-01-03 13:59:59 +00:00
sc->dc_type = DC_TYPE_AN983;
Use only a 64bit hash filter table for ADM-Centaur cards like the Accton EN2242 and the ADMtek AN985 cards. PR: 32699 Submitted by: Jean-Luc Richier <Jean-Luc.Richier@imag.fr> Reviewed by: phk MFC after: 2 weeks
2003-05-01 09:31:01 +00:00
sc->dc_flags |= DC_64BIT_HASH;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_flags |= DC_TX_USE_TX_INTR;
sc->dc_flags |= DC_TX_ADMTEK_WAR;
sc->dc_pmode = DC_PMODE_MII;
Don't read the MAC address from a copy of the EEPROM in the softc that has been recorded earlier and overwrite it again later by reading it directly from the EEPROM again. Read the MAC address from the PAR0/PAR1 registers instead, which are autoloaded on reboot. Tested on AN985, AN983B. According to the datasheets, it should also work for the AL981 (I don't have such a chip on a card at home) PR: 52988 Submitted by: Andrew Gordon <arg-bsd@arg.me.uk> MFC after: 2 weeks
2003-09-16 05:01:27 +00:00
/* Don't read SROM for - auto-loaded on reset */
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_98713):
case DC_DEVID(DC_VENDORID_CP, DC_DEVICEID_98713_CP):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (revision < DC_REVISION_98713A) {
sc->dc_type = DC_TYPE_98713;
}
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
if (revision >= DC_REVISION_98713A) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_98713A;
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
sc->dc_flags |= DC_21143_NWAY;
}
sc->dc_flags |= DC_REDUCED_MII_POLL;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_POLL | DC_TX_USE_TX_INTR;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_987x5):
case DC_DEVID(DC_VENDORID_ACCTON, DC_DEVICEID_EN1217):
Apply patch to the dc driver to handle Macronix MX98715AEC-C/D/E chips, which differ slightly from the Macronix MX98715AEC chip on the sample adapter that I have in that the multicast hash table is only 128 bits wide instead of 512. New adapters are popping up with this chip, and due to improper handling of the smaller hash table, broadcast packets were not being received correctly.
2000-07-15 17:54:30 +00:00
/*
* Macronix MX98715AEC-C/D/E parts have only a
* 128-bit hash table. We need to deal with these
* in the same manner as the PNIC II so that we
* get the right number of bits out of the
* CRC routine.
*/
if (revision >= DC_REVISION_98715AEC_C &&
revision < DC_REVISION_98725)
sc->dc_flags |= DC_128BIT_HASH;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_987x5;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_POLL | DC_TX_USE_TX_INTR;
sc->dc_flags |= DC_REDUCED_MII_POLL | DC_21143_NWAY;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_MX, DC_DEVICEID_98727):
Add the PCI IDs for the Macronix 98727 and 98732 parts. These are 3.3volt PCI/cardbus chipsets similar to the 98715 (and they have 512-bit hash tables). Also update the man page to mention the 98727/98732 and the SOHOware SFA110A Rev B4 card with the 98715AEC-C chip.
2000-07-17 19:27:41 +00:00
sc->dc_type = DC_TYPE_987x5;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_POLL | DC_TX_USE_TX_INTR;
sc->dc_flags |= DC_REDUCED_MII_POLL | DC_21143_NWAY;
Add the PCI IDs for the Macronix 98727 and 98732 parts. These are 3.3volt PCI/cardbus chipsets similar to the 98715 (and they have 512-bit hash tables). Also update the man page to mention the 98727/98732 and the SOHOware SFA110A Rev B4 card with the 98715AEC-C chip.
2000-07-17 19:27:41 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C115):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_PNICII;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_POLL | DC_TX_USE_TX_INTR | DC_128BIT_HASH;
sc->dc_flags |= DC_REDUCED_MII_POLL | DC_21143_NWAY;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_LO, DC_DEVICEID_82C168):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_PNIC;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_STORENFWD | DC_TX_INTR_ALWAYS;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_flags |= DC_PNIC_RX_BUG_WAR;
sc->dc_pnic_rx_buf = malloc(DC_RXLEN * 5, M_DEVBUF, M_NOWAIT);
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (sc->dc_pnic_rx_buf == NULL) {
device_printf(sc->dc_dev,
"Could not allocate PNIC RX buffer\n");
error = ENOMEM;
goto fail;
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (revision < DC_REVISION_82C169)
sc->dc_pmode = DC_PMODE_SYM;
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_ASIX, DC_DEVICEID_AX88140A):
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_type = DC_TYPE_ASIX;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
sc->dc_flags |= DC_TX_USE_TX_INTR | DC_TX_INTR_FIRSTFRAG;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_flags |= DC_REDUCED_MII_POLL;
sc->dc_pmode = DC_PMODE_MII;
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_XIRCOM, DC_DEVICEID_X3201):
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
sc->dc_type = DC_TYPE_XIRCOM;
Avoid an unnecessary copy of a packet if it is already in a single mbuf. Introduce an additional device flag for those NICs which require the transmit buffers to be aligned to 32-bit boundaries. (the equivalen fix for STABLE is slightly simpler because there are no supported chips which require this alignment there.)
2001-12-11 02:47:53 +00:00
sc->dc_flags |= DC_TX_INTR_ALWAYS | DC_TX_COALESCE |
Patch up some existing style bugs and some that crept in with the DEVICE_POLLING stuff.
2001-12-15 02:51:21 +00:00
DC_TX_ALIGN;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
/*
* We don't actually need to coalesce, but we're doing
* it to obtain a double word aligned buffer.
Avoid an unnecessary copy of a packet if it is already in a single mbuf. Introduce an additional device flag for those NICs which require the transmit buffers to be aligned to 32-bit boundaries. (the equivalen fix for STABLE is slightly simpler because there are no supported chips which require this alignment there.)
2001-12-11 02:47:53 +00:00
* The DC_TX_COALESCE flag is required.
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
*/
Dynamically configure the width of the srom. This code comes from OpenBSD who got the code (or the idea) from the NetBSD tlp driver. This gets some cardbus dc cards working (either completely or nearly so). It also appears to get additional pci cards working, without breaking working ones. # Maybe some additional work is needed here. Also, the cardbus attachment # might need to match on the CIS rather than on the vendor/device so we have # a finer level of detail as to what the card is. Technically, the # vendor/device fields are undefined for CardBus (even though most cards are # using common silicon with pci models).
2002-10-07 05:26:35 +00:00
sc->dc_pmode = DC_PMODE_MII;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
break;
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
case DC_DEVID(DC_VENDORID_CONEXANT, DC_DEVICEID_RS7112):
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
sc->dc_type = DC_TYPE_CONEXANT;
sc->dc_flags |= DC_TX_INTR_ALWAYS;
sc->dc_flags |= DC_REDUCED_MII_POLL;
sc->dc_pmode = DC_PMODE_MII;
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
error = dc_read_srom(sc, sc->dc_romwidth);
if (error != 0)
goto fail;
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
break;
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
case DC_DEVID(DC_VENDORID_ULI, DC_DEVICEID_M5261):
case DC_DEVID(DC_VENDORID_ULI, DC_DEVICEID_M5263):
if (sc->dc_info->dc_devid ==
DC_DEVID(DC_VENDORID_ULI, DC_DEVICEID_M5261))
sc->dc_type = DC_TYPE_ULI_M5261;
else
sc->dc_type = DC_TYPE_ULI_M5263;
/* TX buffers should be aligned on 4 byte boundary. */
sc->dc_flags |= DC_TX_INTR_ALWAYS | DC_TX_COALESCE |
DC_TX_ALIGN;
sc->dc_pmode = DC_PMODE_MII;
error = dc_read_srom(sc, sc->dc_romwidth);
if (error != 0)
goto fail;
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
default:
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
device_printf(dev, "unknown device: %x\n",
sc->dc_info->dc_devid);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
/* Save the cache line size. */
Add support for the Davicom DM9102A 10/100 ethernet controller chip. This is just to make sure we initialize the chip correctly: we need to make the sure the port select bit in CSR6 is set properly so that we use the internal PHY for 10/100 support. (The eval boards I have also include an external HomePNA PHY, but I need to play with that more before I can support it.)
2000-01-19 19:03:08 +00:00
if (DC_IS_DAVICOM(sc))
sc->dc_cachesize = 0;
else
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
sc->dc_cachesize = pci_get_cachelnsz(dev);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Reset the adapter. */
dc_reset(sc);
/* Take 21143 out of snooze mode */
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (DC_IS_INTEL(sc) || DC_IS_XIRCOM(sc)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
command = pci_read_config(dev, DC_PCI_CFDD, 4);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
command &= ~(DC_CFDD_SNOOZE_MODE | DC_CFDD_SLEEP_MODE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
pci_write_config(dev, DC_PCI_CFDD, command, 4);
}
/*
* Try to learn something about the supported media.
* We know that ASIX and ADMtek and Davicom devices
* will *always* be using MII media, so that's a no-brainer.
* The tricky ones are the Macronix/PNIC II and the
* Intel 21143.
*/
Always check memory allocation failure. If driver encounter memory allocation error, do not attach driver. Reviewed by: marius
2011-02-18 01:56:25 +00:00
if (DC_IS_INTEL(sc)) {
error = dc_parse_21143_srom(sc);
if (error != 0)
goto fail;
} else if (DC_IS_MACRONIX(sc) || DC_IS_PNICII(sc)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_type == DC_TYPE_98713)
sc->dc_pmode = DC_PMODE_MII;
else
sc->dc_pmode = DC_PMODE_SYM;
} else if (!sc->dc_pmode)
sc->dc_pmode = DC_PMODE_MII;
/*
* Get station address from the EEPROM.
*/
switch(sc->dc_type) {
case DC_TYPE_98713:
case DC_TYPE_98713A:
case DC_TYPE_987x5:
case DC_TYPE_PNICII:
dc_read_eeprom(sc, (caddr_t)&mac_offset,
(DC_EE_NODEADDR_OFFSET / 2), 1, 0);
dc_read_eeprom(sc, (caddr_t)&eaddr, (mac_offset / 2), 3, 0);
break;
case DC_TYPE_PNIC:
dc_read_eeprom(sc, (caddr_t)&eaddr, 0, 3, 1);
break;
case DC_TYPE_DM9102:
Some integrated Davicom cards in sparc64 boxes have an all zeros MAC address in the EEPROM, and we need to get it from OpenFirmware. This isn't very pretty but time is lacking to do this in a better way this near 5.2-RELEASE. This is a RELENG_5_2 candidate. Original version by: Marius Strobl <marius@alchemy.franken.de> Tested by: Pete Bentley <pete@sorted.org> Reviewed by: jake
2004-01-08 19:08:27 +00:00
dc_read_eeprom(sc, (caddr_t)&eaddr, DC_EE_NODEADDR, 3, 0);
#ifdef __sparc64__
/*
* If this is an onboard dc(4) the station address read from
- Rev. 1.175 fixed compilation on sparc64 but also backed out zeroing of the eaddr array (introduced in rev. 1.174) prior to writing to it. As dc_read_eeprom() is told to write only 3 16-bit words to eaddr but eaddr in fact is somewhat larger removal of the zeroing defeated the check whether the MAC address is all zero as there can be some random garbage in eaddr past the 3 words written to it and the check verifys all bits in eaddr. Solve this by changing the check to verify only the 3 words (happenning to be ETHER_ADDR_LEN bytes) written to eaddr. - While here change the notation of "FCode" in a nearby comment to the official way. Ok'ed by: marcel, ru
2005-12-05 12:32:20 +00:00
* the EEPROM is all zero and we have to get it from the FCode.
Some integrated Davicom cards in sparc64 boxes have an all zeros MAC address in the EEPROM, and we need to get it from OpenFirmware. This isn't very pretty but time is lacking to do this in a better way this near 5.2-RELEASE. This is a RELENG_5_2 candidate. Original version by: Marius Strobl <marius@alchemy.franken.de> Tested by: Pete Bentley <pete@sorted.org> Reviewed by: jake
2004-01-08 19:08:27 +00:00
*/
- Rev. 1.175 fixed compilation on sparc64 but also backed out zeroing of the eaddr array (introduced in rev. 1.174) prior to writing to it. As dc_read_eeprom() is told to write only 3 16-bit words to eaddr but eaddr in fact is somewhat larger removal of the zeroing defeated the check whether the MAC address is all zero as there can be some random garbage in eaddr past the 3 words written to it and the check verifys all bits in eaddr. Solve this by changing the check to verify only the 3 words (happenning to be ETHER_ADDR_LEN bytes) written to eaddr. - While here change the notation of "FCode" in a nearby comment to the official way. Ok'ed by: marcel, ru
2005-12-05 12:32:20 +00:00
if (eaddr[0] == 0 && (eaddr[1] & ~0xffff) == 0)
Fix on sparc64.
2005-12-04 07:45:12 +00:00
OF_getetheraddr(dev, (caddr_t)&eaddr);
Some integrated Davicom cards in sparc64 boxes have an all zeros MAC address in the EEPROM, and we need to get it from OpenFirmware. This isn't very pretty but time is lacking to do this in a better way this near 5.2-RELEASE. This is a RELENG_5_2 candidate. Original version by: Marius Strobl <marius@alchemy.franken.de> Tested by: Pete Bentley <pete@sorted.org> Reviewed by: jake
2004-01-08 19:08:27 +00:00
#endif
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case DC_TYPE_21143:
case DC_TYPE_ASIX:
dc_read_eeprom(sc, (caddr_t)&eaddr, DC_EE_NODEADDR, 3, 0);
break;
case DC_TYPE_AL981:
Fix card/device names, no functional change The ADMtek AN985 is the cardbus variant of ADMtek AN983 The Netgear FA511 is just a relabled ADMtek AN985 PR: kern/50574 MFC after: 1 month
2010-01-03 13:59:59 +00:00
case DC_TYPE_AN983:
Ethernet hardware address stored in DC_AL_PAR0/DC_AL_PAR1 register is in little endian form. Likewise setting DC_AL_PAR0/DC_AL_PAR1 register expect the address to be in little endian form. For big endian architectures the address should be swapped to get correct one. Change setting/getting ethernet hardware address to big endian architecture frendly. Reported by: Robert Murillo ( billypilgrim782001 at yahoo dot com ) Tested by: Robert Murillo ( billypilgrim782001 at yahoo dot com )
2008-06-08 02:52:26 +00:00
reg = CSR_READ_4(sc, DC_AL_PAR0);
mac = (uint8_t *)&eaddr[0];
mac[0] = (reg >> 0) & 0xff;
mac[1] = (reg >> 8) & 0xff;
mac[2] = (reg >> 16) & 0xff;
mac[3] = (reg >> 24) & 0xff;
reg = CSR_READ_4(sc, DC_AL_PAR1);
mac[4] = (reg >> 0) & 0xff;
mac[5] = (reg >> 8) & 0xff;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
case DC_TYPE_CONEXANT:
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
bcopy(sc->dc_srom + DC_CONEXANT_EE_NODEADDR, &eaddr,
ETHER_ADDR_LEN);
Add support for Conexant LANfinity miniPCI controllers. People who have laptops with this chip should test this and report back as I don't have access to this hardware myself. People with -stable systems should try the patch at: http://www.freebsd.org/~wpaul/conexant.patch.gz Submitted by: Phil Kernick <Phil@Kernick.org>
2001-09-04 17:10:11 +00:00
break;
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
case DC_TYPE_XIRCOM:
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* The MAC comes from the CIS. */
Xircom cards store the MAC address in the CIS, so get it from the new pci_get_ether accesor, which gets it from the CIS for cardbus cards (and from other pci-like buses via whatever mechanism is used there). Submitted by: sam Approved by: re (blanket)
2002-11-27 07:04:10 +00:00
mac = pci_get_ether(dev);
if (!mac) {
device_printf(dev, "No station address in CIS!\n");
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
error = ENXIO;
Xircom cards store the MAC address in the CIS, so get it from the new pci_get_ether accesor, which gets it from the CIS for cardbus cards (and from other pci-like buses via whatever mechanism is used there). Submitted by: sam Approved by: re (blanket)
2002-11-27 07:04:10 +00:00
goto fail;
}
bcopy(mac, eaddr, ETHER_ADDR_LEN);
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
break;
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
case DC_TYPE_ULI_M5261:
case DC_TYPE_ULI_M5263:
srom = (uint16_t *)sc->dc_srom;
if (srom == NULL || *srom == 0xFFFF || *srom == 0) {
/*
* No valid SROM present, read station address
* from ID Table.
*/
device_printf(dev,
"Reading station address from ID Table.\n");
CSR_WRITE_4(sc, DC_BUSCTL, 0x10000);
CSR_WRITE_4(sc, DC_SIARESET, 0x01C0);
CSR_WRITE_4(sc, DC_10BTCTRL, 0x0000);
CSR_WRITE_4(sc, DC_10BTCTRL, 0x0010);
CSR_WRITE_4(sc, DC_10BTCTRL, 0x0000);
CSR_WRITE_4(sc, DC_SIARESET, 0x0000);
CSR_WRITE_4(sc, DC_SIARESET, 0x01B0);
mac = (uint8_t *)eaddr;
for (n = 0; n < ETHER_ADDR_LEN; n++)
mac[n] = (uint8_t)CSR_READ_4(sc, DC_10BTCTRL);
CSR_WRITE_4(sc, DC_SIARESET, 0x0000);
CSR_WRITE_4(sc, DC_BUSCTL, 0x0000);
DELAY(10);
} else
dc_read_eeprom(sc, (caddr_t)&eaddr, DC_EE_NODEADDR, 3,
0);
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
default:
dc_read_eeprom(sc, (caddr_t)&eaddr, DC_EE_NODEADDR, 3, 0);
break;
}
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
bcopy(eaddr, sc->dc_eaddr, sizeof(eaddr));
/*
* If we still have invalid station address, see whether we can
* find station address for chip 0. Some multi-port controllers
* just store station address for chip 0 if they have a shared
* SROM.
*/
if ((sc->dc_eaddr[0] == 0 && (sc->dc_eaddr[1] & ~0xffff) == 0) ||
(sc->dc_eaddr[0] == 0xffffffff &&
(sc->dc_eaddr[1] & 0xffff) == 0xffff)) {
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
error = dc_check_multiport(sc);
if (error == 0) {
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
bcopy(sc->dc_eaddr, eaddr, sizeof(eaddr));
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
/* Extract media information. */
if (DC_IS_INTEL(sc) && sc->dc_srom != NULL) {
while (sc->dc_mi != NULL) {
m = sc->dc_mi->dc_next;
free(sc->dc_mi, M_DEVBUF);
sc->dc_mi = m;
}
error = dc_parse_21143_srom(sc);
if (error != 0)
goto fail;
}
} else if (error == ENOMEM)
goto fail;
else
error = 0;
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if ((error = dc_dma_alloc(sc)) != 0)
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
goto fail;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_ifp = if_alloc(IFT_ETHER);
if (ifp == NULL) {
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(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;
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifp->if_softc = sc;
Replace the if_name and if_unit members of struct ifnet with new members if_xname, if_dname, and if_dunit. if_xname is the name of the interface and if_dname/unit are the driver name and instance. This change paves the way for interface renaming and enhanced pseudo device creation and configuration symantics. Approved By: re (in principle) Reviewed By: njl, imp Tested On: i386, amd64, sparc64 Obtained From: NetBSD (if_xname)
2003-10-31 18:32:15 +00:00
if_initname(ifp, device_get_name(dev), device_get_unit(dev));
Conditionalize IFF_NEEDSGIANT, like everything else here, on IS_MPSAFE. The driver doesn't look any less safe without Giant than with, and works with IS_MPSAFE set to 1 here, so others should probably test it as such.
2004-10-01 07:01:38 +00:00
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifp->if_ioctl = dc_ioctl;
ifp->if_start = dc_start;
ifp->if_init = dc_init;
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
IFQ_SET_MAXLEN(&ifp->if_snd, DC_TX_LIST_CNT - 1);
ifp->if_snd.ifq_drv_maxlen = DC_TX_LIST_CNT - 1;
IFQ_SET_READY(&ifp->if_snd);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
* Do MII setup. If this is a 21143, check for a PHY on the
* MII bus after applying any necessary fixups to twiddle the
* GPIO bits. If we don't end up finding a PHY, restore the
* old selection (SIA only or SIA/SYM) and attach the dcphy
* driver instead.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
tmp = 0;
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
if (DC_IS_INTEL(sc)) {
dc_apply_fixup(sc, IFM_AUTO);
tmp = sc->dc_pmode;
sc->dc_pmode = DC_PMODE_MII;
}
Fix newer Xircom CBE2-100 cards that were reporting dc0: MII without any PHY! We have to enable the connection to the MII first. Doing so fixes the problem cards without breaking the older, working cards. Bad card provided by: deischen
2005-04-18 03:31:29 +00:00
/*
* Setup General Purpose port mode and data so the tulip can talk
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
* to the MII. This needs to be done before mii_attach so that
Fix newer Xircom CBE2-100 cards that were reporting dc0: MII without any PHY! We have to enable the connection to the MII first. Doing so fixes the problem cards without breaking the older, working cards. Bad card provided by: deischen
2005-04-18 03:31:29 +00:00
* we can actually see them.
*/
if (DC_IS_XIRCOM(sc)) {
CSR_WRITE_4(sc, DC_SIAGP, DC_SIAGP_WRITE_EN | DC_SIAGP_INT1_EN |
DC_SIAGP_MD_GP2_OUTPUT | DC_SIAGP_MD_GP0_OUTPUT);
DELAY(10);
CSR_WRITE_4(sc, DC_SIAGP, DC_SIAGP_INT1_EN |
DC_SIAGP_MD_GP2_OUTPUT | DC_SIAGP_MD_GP0_OUTPUT);
DELAY(10);
}
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;
/*
* Note: both the AL981 and AN983 have internal PHYs, however the
* AL981 provides direct access to the PHY registers while the AN983
* uses a serial MII interface. The AN983's MII interface is also
* buggy in that you can read from any MII address (0 to 31), but
* only address 1 behaves normally. To deal with both cases, we
* pretend that the PHY is at MII address 1.
*/
if (DC_IS_ADMTEK(sc))
phy = DC_ADMTEK_PHYADDR;
/*
* Note: the ukphy probes of the RS7112 report a PHY at MII address
* 0 (possibly HomePNA?) and 1 (ethernet) so we only respond to the
* correct one.
*/
if (DC_IS_CONEXANT(sc))
phy = DC_CONEXANT_PHYADDR;
error = mii_attach(dev, &sc->dc_miibus, ifp, dc_ifmedia_upd,
dc_ifmedia_sts, BMSR_DEFCAPMASK, phy, MII_OFFSET_ANY, 0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (error && DC_IS_INTEL(sc)) {
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
sc->dc_pmode = tmp;
if (sc->dc_pmode != DC_PMODE_SIA)
sc->dc_pmode = DC_PMODE_SYM;
When I tweaked if_dc.c to alter the polling interval for non-MII 21143 chips, I accidentally removed the DC_MII_REDUCED_POLL flag for all 21143 cards. This caused problems with timer-instigated TCP retransmits, which happened to occur at the same time as an MII poll tick on MII-based cards (e.g. D-Link DFE-570TX). Fixed this, plus made some other cleanups. The autoneg fixes for the non-MII cards still work. Also tested the PNIC II now that I have one again.
2000-06-07 17:07:44 +00:00
sc->dc_flags |= DC_21143_NWAY;
Special-case the LED twiddling code so that it doesn't do anything on the NEC VersaPro NoteBook PC. This 21143 implementation has no LEDs, and flipping the LED control bits somehow stops it from establishing a link. We check the subsystem ID and don't flip the LED control bits for the NEC NIC.
2000-09-07 18:51:04 +00:00
/*
* For non-MII cards, we need to have the 21143
* drive the LEDs. Except there are some systems
* like the NEC VersaPro NoteBook PC which have no
* LEDs, and twiddling these bits has adverse effects
* on them. (I.e. you suddenly can't get a link.)
*/
- Switch on the full 32-bit device ID to avoid collisions between the vendor-specific device ids across vendors. - Include the revision in the dc_devs[] array instead of special casing the revid handling in dc_devtype(). - Use PCI bus accessors to read registers instead of pci_read_config() where possible. - Use an 8-bit write to update the latency timer. - Use PCIR_xxx constants and remove unused DC_xxx related to standard PCI config registers. MFC after: 1 week
2006-06-03 20:41:55 +00:00
if (!(pci_get_subvendor(dev) == 0x1033 &&
pci_get_subdevice(dev) == 0x8028))
Special-case the LED twiddling code so that it doesn't do anything on the NEC VersaPro NoteBook PC. This 21143 implementation has no LEDs, and flipping the LED control bits somehow stops it from establishing a link. We check the subsystem ID and don't flip the LED control bits for the NEC NIC.
2000-09-07 18:51:04 +00:00
sc->dc_flags |= DC_TULIP_LEDS;
- There's no need to ignore the return value of mii_attach(9) when attaching dcphy(4) (CID 9283). - In dc_detach(), check whether ifp is NULL as dc_attach() may call the former without ifp being allocated (CID 4288). Found with: Coverity Prevent(tm)
2011-11-18 21:22:43 +00:00
error = mii_attach(dev, &sc->dc_miibus, ifp, dc_ifmedia_upd,
dc_ifmedia_sts, BMSR_DEFCAPMASK, MII_PHY_ANY,
MII_OFFSET_ANY, 0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
if (error) {
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
device_printf(dev, "attaching PHYs failed\n");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
goto fail;
}
Move the DC_SETBIT to the attach function. Reviewed by: phk
2003-02-05 21:21:26 +00:00
if (DC_IS_ADMTEK(sc)) {
/*
* Set automatic TX underrun recovery for the ADMtek chips
*/
DC_SETBIT(sc, DC_AL_CR, DC_AL_CR_ATUR);
}
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
/*
* Tell the upper layer(s) we support long frames.
*/
ifp->if_data.ifi_hdrlen = sizeof(struct ether_vlan_header);
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_capabilities |= IFCAP_VLAN_MTU;
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
ifp->if_capenable = ifp->if_capabilities;
First driver with user-configurable polling(4).
2004-04-11 13:47:15 +00:00
#ifdef DEVICE_POLLING
ifp->if_capabilities |= IFCAP_POLLING;
#endif
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
callout_init_mtx(&sc->dc_stat_ch, &sc->dc_mtx, 0);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
callout_init_mtx(&sc->dc_wdog_ch, &sc->dc_mtx, 0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
/*
* Call MI attach routine.
*/
Fix the misalignment bugs differently than was done in the previous commit. Copy the ethernet address into a local buffer, which we know is sufficiently aligned for the width of the memory accesses that we do. This also eliminates all suspicious and potentionally harmful casts. In collaboration with: ru
2005-12-02 01:23:20 +00:00
ether_ifattach(ifp, (caddr_t)eaddr);
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +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
/* Hook interrupt last to avoid having to lock softc */
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
error = bus_setup_intr(dev, sc->dc_irq, 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, dc_intr, sc, &sc->dc_intrhand);
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
if (error) {
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
device_printf(dev, "couldn't set up irq\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);
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;
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
}
Teach the 'dc' driver how to pick up settings left over by the SRM on alpha systems. This is an expedient if not entirely elegant solution to the problem. Submitted by: gallatin Approved by: jkh
2000-03-11 05:20:56 +00:00
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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)
dc_detach(dev);
Remove bogus locking from dc(4). Instead, move interrupt allocation and ether_ifattach() to end. This fixes a "could sleep" case and simplifies error exit cases as well. Also be sure to set errno and clean up resources in !mac error case. Tested by: Ryan Beasley
2003-01-14 19:31:27 +00:00
return (error);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
/*
* 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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_detach(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct ifnet *ifp;
struct dc_mediainfo *m;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
Add missing ()'s so that these drivers all compile again. Noticed by: jake Tested on: i386 (compile)
2003-03-31 20:22:00 +00:00
KASSERT(mtx_initialized(&sc->dc_mtx), ("dc mutex not initialized"));
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
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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#ifdef DEVICE_POLLING
- There's no need to ignore the return value of mii_attach(9) when attaching dcphy(4) (CID 9283). - In dc_detach(), check whether ifp is NULL as dc_attach() may call the former without ifp being allocated (CID 4288). Found with: Coverity Prevent(tm)
2011-11-18 21:22:43 +00:00
if (ifp != NULL && ifp->if_capenable & IFCAP_POLLING)
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
ether_poll_deregister(ifp);
#endif
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 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)) {
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Revert part of revision 1.97 by calling dc_stop() unconditionally in dc_detach() instead of only calling it if the hardware is preset. This is a workaround for page faults in softclock() after a `dc' device was detached, caused by not disabling a timer before freeing its memory. The bus_child_present() checks should probably be re-added later, but only to avoid the hardware accesses and not the other resource cleanups in dc_stop(). Approved by: njl
2003-04-17 08:36:52 +00:00
dc_stop(sc);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
callout_drain(&sc->dc_stat_ch);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
callout_drain(&sc->dc_wdog_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);
}
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
if (sc->dc_miibus)
device_delete_child(dev, sc->dc_miibus);
bus_generic_detach(dev);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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->dc_intrhand)
bus_teardown_intr(dev, sc->dc_irq, sc->dc_intrhand);
if (sc->dc_irq)
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->dc_irq);
if (sc->dc_res)
bus_release_resource(dev, DC_RES, DC_RID, sc->dc_res);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- There's no need to ignore the return value of mii_attach(9) when attaching dcphy(4) (CID 9283). - In dc_detach(), check whether ifp is NULL as dc_attach() may call the former without ifp being allocated (CID 4288). Found with: Coverity Prevent(tm)
2011-11-18 21:22:43 +00:00
if (ifp != NULL)
Make dc compile after repo-copy.
2005-10-18 06:09:42 +00:00
if_free(ifp);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
dc_dma_free(sc);
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +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
free(sc->dc_pnic_rx_buf, M_DEVBUF);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
while (sc->dc_mi != NULL) {
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +00:00
m = sc->dc_mi->dc_next;
free(sc->dc_mi, M_DEVBUF);
sc->dc_mi = m;
}
Ooops. Need to free dc_srom on detach to not leak memory. Pointy Hat to: The Mad Redhead of Niwot
2002-10-07 05:29:13 +00:00
free(sc->dc_srom, M_DEVBUF);
Add support for parsing the media blocks from the SROM on 21143 adapters. This is necessary in order to make this driver work with the built-in ethernet on the alpha Miata machines. These systems have a 21143-PC chip on-board and optional daughtercards with either a 10/100 MII transceiver or a 10baseT/10base2 transceiver. In both cases, you need to twiddle the GPIO bits on the controller in order to turn the transceivers on, and you have to read the media info from the SROM in order to find out what bits to twiddle.
2000-10-05 17:36:14 +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
mtx_destroy(&sc->dc_mtx);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Initialize the transmit descriptors.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_list_tx_init(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_chain_data *cd;
struct dc_list_data *ld;
int i, nexti;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
cd = &sc->dc_cdata;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
ld = &sc->dc_ldata;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
for (i = 0; i < DC_TX_LIST_CNT; i++) {
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
if (i == DC_TX_LIST_CNT - 1)
nexti = 0;
else
nexti = i + 1;
Make sure to clear status word of TX descriptor in dc_list_tx_init(). Do not update if_opackets if the transmission had failed.
2011-02-18 22:58:46 +00:00
ld->dc_tx_list[i].dc_status = 0;
ld->dc_tx_list[i].dc_ctl = 0;
ld->dc_tx_list[i].dc_data = 0;
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
ld->dc_tx_list[i].dc_next = htole32(DC_TXDESC(sc, nexti));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
cd->dc_tx_chain[i] = NULL;
}
cd->dc_tx_prod = cd->dc_tx_cons = cd->dc_tx_cnt = 0;
For controllers that have TX interrupt moderation capability, request TX completion interrupt for every 8-th frames. Previously dc(4) requested TX completion interrupt if number of queued TX descriptors is greater than 64. This caused a lot of TX completion interrupt under high TX load once driver queued more than 64 TX descriptors. It's quite normal to see more than 64 queued TX descriptors under high TX load. This change reduces the number of TX completion interrupts to be less than 17k under high TX load. Because this change does not generate TX completion interrupt for each frame, add reclaiming transmitted buffers in dc_tick not to generate false watchdog timeouts. While I'm here add check for queued descriptors in dc_txeof() since there is no more work to do when there is no pending descriptors.
2011-02-18 20:38:05 +00:00
cd->dc_tx_pkts = 0;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap,
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Initialize the RX descriptors and allocate mbufs for them. Note that
* we arrange the descriptors in a closed ring, so that the last descriptor
* points back to the first.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_list_rx_init(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_chain_data *cd;
struct dc_list_data *ld;
int i, nexti;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
cd = &sc->dc_cdata;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
ld = &sc->dc_ldata;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
for (i = 0; i < DC_RX_LIST_CNT; i++) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (dc_newbuf(sc, i) != 0)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (ENOBUFS);
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
if (i == DC_RX_LIST_CNT - 1)
nexti = 0;
else
nexti = i + 1;
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
ld->dc_rx_list[i].dc_next = htole32(DC_RXDESC(sc, nexti));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
cd->dc_rx_prod = 0;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_rx_ltag, sc->dc_rx_lmap,
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Initialize an RX descriptor and attach an MBUF cluster.
*/
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
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
dc_newbuf(struct dc_softc *sc, int i)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
struct mbuf *m;
bus_dmamap_t map;
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
bus_dma_segment_t segs[1];
int error, nseg;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (m == NULL)
return (ENOBUFS);
m->m_len = m->m_pkthdr.len = MCLBYTES;
m_adj(m, sizeof(u_int64_t));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* If this is a PNIC chip, zero the buffer. This is part
* of the workaround for the receive bug in the 82c168 and
* 82c169 chips.
*/
if (sc->dc_flags & DC_PNIC_RX_BUG_WAR)
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bzero(mtod(m, char *), m->m_len);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
error = bus_dmamap_load_mbuf_sg(sc->dc_rx_mtag, sc->dc_sparemap,
m, segs, &nseg, 0);
if (error) {
m_freem(m);
return (error);
}
KASSERT(nseg == 1, ("%s: wrong number of segments (%d)", __func__,
nseg));
if (sc->dc_cdata.dc_rx_chain[i] != NULL)
bus_dmamap_unload(sc->dc_rx_mtag, sc->dc_cdata.dc_rx_map[i]);
map = sc->dc_cdata.dc_rx_map[i];
sc->dc_cdata.dc_rx_map[i] = sc->dc_sparemap;
sc->dc_sparemap = map;
sc->dc_cdata.dc_rx_chain[i] = m;
bus_dmamap_sync(sc->dc_rx_mtag, sc->dc_cdata.dc_rx_map[i],
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
BUS_DMASYNC_PREREAD);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_rx_list[i].dc_ctl = htole32(DC_RXCTL_RLINK | DC_RXLEN);
sc->dc_ldata.dc_rx_list[i].dc_data =
htole32(DC_ADDR_LO(segs[0].ds_addr));
sc->dc_ldata.dc_rx_list[i].dc_status = htole32(DC_RXSTAT_OWN);
bus_dmamap_sync(sc->dc_rx_ltag, sc->dc_rx_lmap,
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Grrrrr.
* The PNIC chip has a terrible bug in it that manifests itself during
* periods of heavy activity. The exact mode of failure if difficult to
* pinpoint: sometimes it only happens in promiscuous mode, sometimes it
* will happen on slow machines. The bug is that sometimes instead of
* uploading one complete frame during reception, it uploads what looks
* like the entire contents of its FIFO memory. The frame we want is at
* the end of the whole mess, but we never know exactly how much data has
* been uploaded, so salvaging the frame is hard.
*
* There is only one way to do it reliably, and it's disgusting.
* Here's what we know:
*
* - We know there will always be somewhere between one and three extra
* descriptors uploaded.
*
* - We know the desired received frame will always be at the end of the
* total data upload.
*
* - We know the size of the desired received frame because it will be
* provided in the length field of the status word in the last descriptor.
*
* Here's what we do:
*
* - When we allocate buffers for the receive ring, we bzero() them.
* This means that we know that the buffer contents should be all
* zeros, except for data uploaded by the chip.
*
* - We also force the PNIC chip to upload frames that include the
* ethernet CRC at the end.
*
* - We gather all of the bogus frame data into a single buffer.
*
* - We then position a pointer at the end of this buffer and scan
* backwards until we encounter the first non-zero byte of data.
* This is the end of the received frame. We know we will encounter
* some data at the end of the frame because the CRC will always be
* there, so even if the sender transmits a packet of all zeros,
* we won't be fooled.
*
* - We know the size of the actual received frame, so we subtract
* that value from the current pointer location. This brings us
* to the start of the actual received packet.
*
* - We copy this into an mbuf and pass it on, along with the actual
* frame length.
*
* The performance hit is tremendous, but it beats dropping frames all
* the time.
*/
Consistently use a tab character instead of space after #define. No functional changes.
2011-02-19 03:23:19 +00:00
#define DC_WHOLEFRAME (DC_RXSTAT_FIRSTFRAG | DC_RXSTAT_LASTFRAG)
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_pnic_rx_bug_war(struct dc_softc *sc, int idx)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_desc *cur_rx;
struct dc_desc *c = NULL;
struct mbuf *m = NULL;
unsigned char *ptr;
int i, total_len;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t rxstat = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
i = sc->dc_pnic_rx_bug_save;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
cur_rx = &sc->dc_ldata.dc_rx_list[idx];
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ptr = sc->dc_pnic_rx_buf;
Change a : bzero(ptr, sizeof(DC_RXLEN * 5)); which should obviously be: bzero(ptr, DC_RXLEN * 5); Looks like this bug may have reduced the effectiveness of the workaround for the hardware bug in the PNIC chips. MFC after: 1 week
2003-06-01 22:09:12 +00:00
bzero(ptr, DC_RXLEN * 5);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Copy all the bytes from the bogus buffers. */
while (1) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
c = &sc->dc_ldata.dc_rx_list[i];
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
rxstat = le32toh(c->dc_status);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
m = sc->dc_cdata.dc_rx_chain[i];
bcopy(mtod(m, char *), ptr, DC_RXLEN);
ptr += DC_RXLEN;
/* If this is the last buffer, break out. */
if (i == idx || rxstat & DC_RXSTAT_LASTFRAG)
break;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
dc_discard_rxbuf(sc, i);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_INC(i, DC_RX_LIST_CNT);
}
/* Find the length of the actual receive frame. */
total_len = DC_RXBYTES(rxstat);
/* Scan backwards until we hit a non-zero byte. */
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
while (*ptr == 0x00)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ptr--;
/* Round off. */
if ((uintptr_t)(ptr) & 0x3)
ptr -= 1;
/* Now find the start of the frame. */
ptr -= total_len;
if (ptr < sc->dc_pnic_rx_buf)
ptr = sc->dc_pnic_rx_buf;
/*
* Now copy the salvaged frame to the last mbuf and fake up
* the status word to make it look like a successful
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
* frame reception.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
*/
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
bcopy(ptr, mtod(m, char *), total_len);
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
cur_rx->dc_status = htole32(rxstat | DC_RXSTAT_FIRSTFRAG);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
/*
* This routine searches the RX ring for dirty descriptors in the
* event that the rxeof routine falls out of sync with the chip's
* current descriptor pointer. This may happen sometimes as a result
* of a "no RX buffer available" condition that happens when the chip
* consumes all of the RX buffers before the driver has a chance to
* process the RX ring. This routine may need to be called more than
* once to bring the driver back in sync with the chip, however we
* should still be getting RX DONE interrupts to drive the search
* for new packets in the RX ring, so we should catch up eventually.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_rx_resync(struct dc_softc *sc)
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_desc *cur_rx;
int i, pos;
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
pos = sc->dc_cdata.dc_rx_prod;
for (i = 0; i < DC_RX_LIST_CNT; i++) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
cur_rx = &sc->dc_ldata.dc_rx_list[pos];
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
if (!(le32toh(cur_rx->dc_status) & DC_RXSTAT_OWN))
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
break;
DC_INC(pos, DC_RX_LIST_CNT);
}
/* If the ring really is empty, then just return. */
if (i == DC_RX_LIST_CNT)
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
/* We've fallen behing the chip: catch it. */
sc->dc_cdata.dc_rx_prod = pos;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (EAGAIN);
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
static void
dc_discard_rxbuf(struct dc_softc *sc, int i)
{
struct mbuf *m;
if (sc->dc_flags & DC_PNIC_RX_BUG_WAR) {
m = sc->dc_cdata.dc_rx_chain[i];
bzero(mtod(m, char *), m->m_len);
}
sc->dc_ldata.dc_rx_list[i].dc_ctl = htole32(DC_RXCTL_RLINK | DC_RXLEN);
sc->dc_ldata.dc_rx_list[i].dc_status = htole32(DC_RXSTAT_OWN);
bus_dmamap_sync(sc->dc_rx_ltag, sc->dc_rx_lmap, BUS_DMASYNC_PREREAD |
BUS_DMASYNC_PREWRITE);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* A frame has been uploaded: pass the resulting mbuf chain up to
* the higher level protocols.
*/
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_rxeof(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
struct mbuf *m;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
struct dc_desc *cur_rx;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
int i, total_len, rx_npkts;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t rxstat;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
DC_LOCK_ASSERT(sc);
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->dc_ifp;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
rx_npkts = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_rx_ltag, sc->dc_rx_lmap, BUS_DMASYNC_POSTREAD |
BUS_DMASYNC_POSTWRITE);
for (i = sc->dc_cdata.dc_rx_prod;
(ifp->if_drv_flags & IFF_DRV_RUNNING) != 0;
DC_INC(i, DC_RX_LIST_CNT)) {
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
#ifdef DEVICE_POLLING
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING) {
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
if (sc->rxcycles <= 0)
break;
sc->rxcycles--;
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
#endif
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
cur_rx = &sc->dc_ldata.dc_rx_list[i];
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
rxstat = le32toh(cur_rx->dc_status);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if ((rxstat & DC_RXSTAT_OWN) != 0)
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
m = sc->dc_cdata.dc_rx_chain[i];
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_rx_mtag, sc->dc_cdata.dc_rx_map[i],
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
BUS_DMASYNC_POSTREAD);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
total_len = DC_RXBYTES(rxstat);
Remove too expensive bus_dmamap_sync(9) call in dc_rx_resync(). With this change, driver may not notice updated descriptor status change when bounce buffers are active. However, rxeof() in next run will handle the synchronization. Change dc_rxeof() a bit to return the number of processed frames in RX descriptor ring. Previously it returned the number of frames that were successfully passed to upper stack which in turn means it ignored frames that were discarded due to errors. The number of processed frames in RX descriptor ring is used to detect whether driver is out of sync with controller's current descriptor pointer. Returning number of processed frames reduces unnecessary (probably wrong) re-synchronization. Reviewed by: marius
2011-03-16 17:09:51 +00:00
rx_npkts++;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_flags & DC_PNIC_RX_BUG_WAR) {
if ((rxstat & DC_WHOLEFRAME) != DC_WHOLEFRAME) {
if (rxstat & DC_RXSTAT_FIRSTFRAG)
sc->dc_pnic_rx_bug_save = i;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if ((rxstat & DC_RXSTAT_LASTFRAG) == 0)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
continue;
dc_pnic_rx_bug_war(sc, i);
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
rxstat = le32toh(cur_rx->dc_status);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
total_len = DC_RXBYTES(rxstat);
}
}
/*
* If an error occurs, update stats, clear the
* status word and leave the mbuf cluster in place:
* it should simply get re-used next time this descriptor
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
* comes up in the ring. However, don't report long
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
* frames as errors since they could be vlans.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
*/
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if ((rxstat & DC_RXSTAT_RXERR)) {
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
if (!(rxstat & DC_RXSTAT_GIANT) ||
(rxstat & (DC_RXSTAT_CRCERR | DC_RXSTAT_DRIBBLE |
DC_RXSTAT_MIIERE | DC_RXSTAT_COLLSEEN |
DC_RXSTAT_RUNT | DC_RXSTAT_DE))) {
ifp->if_ierrors++;
if (rxstat & DC_RXSTAT_COLLSEEN)
ifp->if_collisions++;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
dc_discard_rxbuf(sc, i);
if (rxstat & DC_RXSTAT_CRCERR)
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
continue;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
else {
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
return (rx_npkts);
Add VLAN for the dc(4) driver (ie long frame). The patch is 2 parts. One to notify the system that the MTU for VLAN can be 1500 so the vlan will automatically be configured with a 1500 MTU the other is to ignore the error case if the received frame is to long. The frame size notification came from code in the SIS driver, and the support for long frames derived from the NetBSD Tulip driver. Tested on: 4 port D-Link adapter DFE-570TX 4 Intel 21143 Netgear card with 82c169 PNIC 10/100BaseTX Reviewed by: ru (manpage), wpaul (not objected to), archie Approved by: imp Obtained from: NetBSD
2002-01-16 16:33:58 +00:00
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
/* No errors; receive the packet. */
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
total_len -= ETHER_CRC_LEN;
- Wrap code optimized for architectures without alignment constraints in #ifdef __NO_STRICT_ALIGNMENT rather than #ifdef __i386__. This means that amd64 now also uses the optimized code. [1] While at it, fix a nearby style(9) bug. - Remove the hw.dc_quick SYSCTL, which allowed to turn off the above mentioned optimization, as like the equivalent and already removed - In dc_setcfg() suppress printing a warning when forcing the receiver and transceiver to idle state times out for chips where the status bits in question just never change (observed in detail with DM9102A) and therefore the warning would be highly likely false positive. [2] - In dc_ifmedia_sts() add a missing DC_UNLOCK(). Tested by: Hans-Joerg Sirtl on amd64 [1] PR: 82681 [2] Obtained from: NetBSD tlp(4) [2] MFC after: 1 week
2006-10-29 20:24:27 +00:00
#ifdef __NO_STRICT_ALIGNMENT
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
/*
- Wrap code optimized for architectures without alignment constraints in #ifdef __NO_STRICT_ALIGNMENT rather than #ifdef __i386__. This means that amd64 now also uses the optimized code. [1] While at it, fix a nearby style(9) bug. - Remove the hw.dc_quick SYSCTL, which allowed to turn off the above mentioned optimization, as like the equivalent and already removed - In dc_setcfg() suppress printing a warning when forcing the receiver and transceiver to idle state times out for chips where the status bits in question just never change (observed in detail with DM9102A) and therefore the warning would be highly likely false positive. [2] - In dc_ifmedia_sts() add a missing DC_UNLOCK(). Tested by: Hans-Joerg Sirtl on amd64 [1] PR: 82681 [2] Obtained from: NetBSD tlp(4) [2] MFC after: 1 week
2006-10-29 20:24:27 +00:00
* On architectures without alignment problems we try to
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
* allocate a new buffer for the receive ring, and pass up
* the one where the packet is already, saving the expensive
* copy done in m_devget().
* If we are on an architecture with alignment problems, or
* if the allocation fails, then use m_devget and leave the
* existing buffer in the receive ring.
*/
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (dc_newbuf(sc, i) != 0) {
dc_discard_rxbuf(sc, i);
ifp->if_iqdrops++;
continue;
}
m->m_pkthdr.rcvif = ifp;
m->m_pkthdr.len = m->m_len = total_len;
#else
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
{
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
struct mbuf *m0;
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
m0 = m_devget(mtod(m, char *), total_len,
ETHER_ALIGN, ifp, NULL);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
dc_discard_rxbuf(sc, i);
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
if (m0 == NULL) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
ifp->if_iqdrops++;
For i386 architecture, remove an expensive m_devget() (and the underlying unaligned bcopy) on incoming packets that are already available (albeit unaligned) in a buffer. The performance improvement varies, depending on CPU and memory speed, but can be quite large especially on slow CPUs. I have seen over 50% increase on forwarding speed on the sis driver for the 486/133 (embedded systems), which does exactly the same thing. The behaviour is controlled by a sysctl variable, hw.dc_quick which defaults to 1. Set it to 0 to restore the old behaviour. After running a few experiments (in userland, though) I am convinced that doing the m_devget() is detrimental to performance in almost all cases. Even if your CPU has degraded performance with misaligned data, the bcopy() in the driver has the same overhead due to misaligment as the one that you save in the uiomove(), plus you do one extra copy and pollute the cache. But more often than not, you do not even have to touch the payload, e.g. when you are forwarding packets, and even in the often-cited case of NFS, you often end up passing a pointer to the payload to the disk controller. In any case, you can play with the sysctl variable to toggle between the two behaviours, and see if it makes a difference. MFC-after: 3 days
2001-11-29 22:46:48 +00:00
continue;
}
m = m0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
#endif
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifp->if_ipackets++;
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
DC_UNLOCK(sc);
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);
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
DC_LOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
sc->dc_cdata.dc_rx_prod = i;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
return (rx_npkts);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* A frame was downloaded to the chip. It's safe for us to clean up
* the list buffers.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_txeof(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
struct dc_desc *cur_tx;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
int idx, setup;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t ctl, txstat;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
For controllers that have TX interrupt moderation capability, request TX completion interrupt for every 8-th frames. Previously dc(4) requested TX completion interrupt if number of queued TX descriptors is greater than 64. This caused a lot of TX completion interrupt under high TX load once driver queued more than 64 TX descriptors. It's quite normal to see more than 64 queued TX descriptors under high TX load. This change reduces the number of TX completion interrupts to be less than 17k under high TX load. Because this change does not generate TX completion interrupt for each frame, add reclaiming transmitted buffers in dc_tick not to generate false watchdog timeouts. While I'm here add check for queued descriptors in dc_txeof() since there is no more work to do when there is no pending descriptors.
2011-02-18 20:38:05 +00:00
if (sc->dc_cdata.dc_tx_cnt == 0)
return;
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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Go through our tx list and free mbufs for those
* frames that have been transmitted.
*/
Fix a regression introduced in r218832. For TX status check, driver should use a TX list DMA tag. MFC after: 3 days
2011-10-24 17:05:59 +00:00
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap, BUS_DMASYNC_POSTREAD |
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
BUS_DMASYNC_POSTWRITE);
setup = 0;
for (idx = sc->dc_cdata.dc_tx_cons; idx != sc->dc_cdata.dc_tx_prod;
DC_INC(idx, DC_TX_LIST_CNT), sc->dc_cdata.dc_tx_cnt--) {
cur_tx = &sc->dc_ldata.dc_tx_list[idx];
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
txstat = le32toh(cur_tx->dc_status);
ctl = le32toh(cur_tx->dc_ctl);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (txstat & DC_TXSTAT_OWN)
break;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (sc->dc_cdata.dc_tx_chain[idx] == NULL)
continue;
if (ctl & DC_TXCTL_SETUP) {
cur_tx->dc_ctl = htole32(ctl & ~DC_TXCTL_SETUP);
setup++;
bus_dmamap_sync(sc->dc_stag, sc->dc_smap,
BUS_DMASYNC_POSTWRITE);
/*
* Yes, the PNIC is so brain damaged
* that it will sometimes generate a TX
* underrun error while DMAing the RX
* filter setup frame. If we detect this,
* we have to send the setup frame again,
* or else the filter won't be programmed
* correctly.
*/
if (DC_IS_PNIC(sc)) {
if (txstat & DC_TXSTAT_ERRSUM)
dc_setfilt(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_cdata.dc_tx_chain[idx] = NULL;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
continue;
}
Fix the "conexant chips don't work in full duplexmode" problem. According to Phil Kernick: "The problem is that in full duplex mode, the Conexant chip always reports a carrier lost error, even when the frame is successfully sent. So, if we have a Conexant chip, then ignore carrier lost when in full duplex mode." Since the Xircom chips seem to have the same issue and since we already have a workaround for this, just expand the workaround test to also check for DC_IS_CONEXANT().
2001-12-19 18:23:45 +00:00
if (DC_IS_XIRCOM(sc) || DC_IS_CONEXANT(sc)) {
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
/*
* XXX: Why does my Xircom taunt me so?
* For some reason it likes setting the CARRLOST flag
Fix the "conexant chips don't work in full duplexmode" problem. According to Phil Kernick: "The problem is that in full duplex mode, the Conexant chip always reports a carrier lost error, even when the frame is successfully sent. So, if we have a Conexant chip, then ignore carrier lost when in full duplex mode." Since the Xircom chips seem to have the same issue and since we already have a workaround for this, just expand the workaround test to also check for DC_IS_CONEXANT().
2001-12-19 18:23:45 +00:00
* even when the carrier is there. wtf?!?
* Who knows, but Conexant chips have the
* same problem. Maybe they took lessons
* from Xircom.
*/
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (/*sc->dc_type == DC_TYPE_21143 &&*/
sc->dc_pmode == DC_PMODE_MII &&
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
((txstat & 0xFFFF) & ~(DC_TXSTAT_ERRSUM |
Fix the "conexant chips don't work in full duplexmode" problem. According to Phil Kernick: "The problem is that in full duplex mode, the Conexant chip always reports a carrier lost error, even when the frame is successfully sent. So, if we have a Conexant chip, then ignore carrier lost when in full duplex mode." Since the Xircom chips seem to have the same issue and since we already have a workaround for this, just expand the workaround test to also check for DC_IS_CONEXANT().
2001-12-19 18:23:45 +00:00
DC_TXSTAT_NOCARRIER)))
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
txstat &= ~DC_TXSTAT_ERRSUM;
} else {
if (/*sc->dc_type == DC_TYPE_21143 &&*/
sc->dc_pmode == DC_PMODE_MII &&
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
((txstat & 0xFFFF) & ~(DC_TXSTAT_ERRSUM |
DC_TXSTAT_NOCARRIER | DC_TXSTAT_CARRLOST)))
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
txstat &= ~DC_TXSTAT_ERRSUM;
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (txstat & DC_TXSTAT_ERRSUM) {
ifp->if_oerrors++;
if (txstat & DC_TXSTAT_EXCESSCOLL)
ifp->if_collisions++;
if (txstat & DC_TXSTAT_LATECOLL)
ifp->if_collisions++;
if (!(txstat & DC_TXSTAT_UNDERRUN)) {
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
return;
}
Make sure to clear status word of TX descriptor in dc_list_tx_init(). Do not update if_opackets if the transmission had failed.
2011-02-18 22:58:46 +00:00
} else
ifp->if_opackets++;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifp->if_collisions += (txstat & DC_TXSTAT_COLLCNT) >> 3;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_tx_mtag, sc->dc_cdata.dc_tx_map[idx],
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->dc_tx_mtag, sc->dc_cdata.dc_tx_map[idx]);
m_freem(sc->dc_cdata.dc_tx_chain[idx]);
sc->dc_cdata.dc_tx_chain[idx] = NULL;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
sc->dc_cdata.dc_tx_cons = idx;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (sc->dc_cdata.dc_tx_cnt <= DC_TX_LIST_CNT - DC_TX_LIST_RSVD) {
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;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (sc->dc_cdata.dc_tx_cnt == 0)
sc->dc_wdog_timer = 0;
}
if (setup > 0)
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_tick(void *xsc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct mii_data *mii;
struct ifnet *ifp;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t r;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = xsc;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK_ASSERT(sc);
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->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii = device_get_softc(sc->dc_miibus);
For controllers that have TX interrupt moderation capability, request TX completion interrupt for every 8-th frames. Previously dc(4) requested TX completion interrupt if number of queued TX descriptors is greater than 64. This caused a lot of TX completion interrupt under high TX load once driver queued more than 64 TX descriptors. It's quite normal to see more than 64 queued TX descriptors under high TX load. This change reduces the number of TX completion interrupts to be less than 17k under high TX load. Because this change does not generate TX completion interrupt for each frame, add reclaiming transmitted buffers in dc_tick not to generate false watchdog timeouts. While I'm here add check for queued descriptors in dc_txeof() since there is no more work to do when there is no pending descriptors.
2011-02-18 20:38:05 +00:00
/*
* Reclaim transmitted frames for controllers that do
* not generate TX completion interrupt for every frame.
*/
if (sc->dc_flags & DC_TX_USE_TX_INTR)
dc_txeof(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (sc->dc_flags & DC_REDUCED_MII_POLL) {
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
if (sc->dc_flags & DC_21143_NWAY) {
r = CSR_READ_4(sc, DC_10BTSTAT);
if (IFM_SUBTYPE(mii->mii_media_active) ==
IFM_100_TX && (r & DC_TSTAT_LS100)) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc->dc_link = 0;
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
mii_mediachg(mii);
}
if (IFM_SUBTYPE(mii->mii_media_active) ==
IFM_10_T && (r & DC_TSTAT_LS10)) {
sc->dc_link = 0;
mii_mediachg(mii);
}
Tweak the DC_REDUCED_MII_POLL code in dc_mii_tick() for the DC_IS_INTEL() case. The idea is to reduce how often we call mii_tick(), however currently it may not be called often enough, which prevents autonegotiation from being driven correctly. This should improve the chances of successfully autonegotiating media settings on non-MII 21143 NICs. (Still waiting for confirmation from some testers, but the code is clearly wrong in any case.)
1999-12-07 19:18:41 +00:00
if (sc->dc_link == 0)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii_tick(mii);
} else {
For chips with a broken DC_ISR_RX_STATE which f.e. never signals stopped nor the waiting state and also no other means to check whether the receiver is idle (see also r163774), we have no choice than to call mii_tick(9) unconditionally even in the case of the DC_REDUCED_MII_POLL handling as far as the RX side is concerned. This isn't necessarily worse than checking whether RX is idle though because unlike as with TX we're racing with the hardware, which might receive packets any time while we poll the MII, anyway. Reported and tested by: Jacob Owens Reviewed by: yongari MFC after: 3 days
2008-08-29 20:31:41 +00:00
/*
* For NICs which never report DC_RXSTATE_WAIT, we
* have to bite the bullet...
*/
if ((DC_HAS_BROKEN_RXSTATE(sc) || (CSR_READ_4(sc,
DC_ISR) & DC_ISR_RX_STATE) == DC_RXSTATE_WAIT) &&
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
sc->dc_cdata.dc_tx_cnt == 0)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii_tick(mii);
}
} else
mii_tick(mii);
/*
* When the init routine completes, we expect to be able to send
* packets right away, and in fact the network code will send a
* gratuitous ARP the moment the init routine marks the interface
* as running. However, even though the MAC may have been initialized,
* there may be a delay of a few seconds before the PHY completes
* autonegotiation and the link is brought up. Any transmissions
* made during that delay will be lost. Dealing with this is tricky:
* we can't just pause in the init routine while waiting for the
* PHY to come ready since that would bring the whole system to
* a screeching halt for several seconds.
*
* What we do here is prevent the TX start routine from sending
* any packets until a link has been established. After the
* interface has been initialized, the tick routine will poll
* the state of the PHY until the IFM_ACTIVE flag is set. Until
* that time, packets will stay in the send queue, and once the
* link comes up, they will be flushed out to the wire.
*/
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
if (sc->dc_link != 0 && !IFQ_DRV_IS_EMPTY(&ifp->if_snd))
dc_start_locked(ifp);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
if (sc->dc_flags & DC_21143_NWAY && !sc->dc_link)
Change the driver to allocate its own callout structure, and modify the interface to use callout_* instead of timeout(). Also add an IS_MPSAFE #define (currently off) which will mark the driver as mpsafe to the upper layers.
2000-11-25 08:00:10 +00:00
callout_reset(&sc->dc_stat_ch, hz/10, dc_tick, sc);
Rework the support for the internal autonegotiation on the 21143 and workalike chips (Macronix 98713A/98715 and PNIC II). Timing is somewhat critical: you need to bring the link as soon as possible after NWAY is done, and the old one second polling interval was too long. Now we poll every 10th of a second until NWAY completes (at which point we return to the 1 second interval again to keep an eye on the link state). I tested all the other cards I had on hand to make sure I didn't bust any of them and they seem to work (including the MII-based 21143 card). This should fix some autoneg problems with DE500-BA cards and the built-in 10/100 ethernet on some alpha systems. (Now before anyone asks why I never noticed this before, the old code worked just find with the Intel swich I used for testing back in NY. Apparently not all switches are as picky about the timing.)
2000-05-31 05:40:53 +00:00
else
Change the driver to allocate its own callout structure, and modify the interface to use callout_* instead of timeout(). Also add an IS_MPSAFE #define (currently off) which will mark the driver as mpsafe to the upper layers.
2000-11-25 08:00:10 +00:00
callout_reset(&sc->dc_stat_ch, hz, dc_tick, sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
/*
* A transmit underrun has occurred. Back off the transmit threshold,
* or switch to store and forward mode if we have to.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_tx_underrun(struct dc_softc *sc)
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
{
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
uint32_t netcfg, isr;
int i, reinit;
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
reinit = 0;
netcfg = CSR_READ_4(sc, DC_NETCFG);
device_printf(sc->dc_dev, "TX underrun -- ");
if ((sc->dc_flags & DC_TX_STORENFWD) == 0) {
if (sc->dc_txthresh + DC_TXTHRESH_INC > DC_TXTHRESH_MAX) {
printf("using store and forward mode\n");
netcfg |= DC_NETCFG_STORENFWD;
} else {
printf("increasing TX threshold\n");
sc->dc_txthresh += DC_TXTHRESH_INC;
netcfg &= ~DC_NETCFG_TX_THRESH;
netcfg |= sc->dc_txthresh;
}
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
if (DC_IS_INTEL(sc)) {
/*
* The real 21143 requires that the transmitter be idle
* in order to change the transmit threshold or store
* and forward state.
*/
CSR_WRITE_4(sc, DC_NETCFG, netcfg & ~DC_NETCFG_TX_ON);
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
for (i = 0; i < DC_TIMEOUT; i++) {
isr = CSR_READ_4(sc, DC_ISR);
if (isr & DC_ISR_TX_IDLE)
break;
DELAY(10);
}
if (i == DC_TIMEOUT) {
device_printf(sc->dc_dev,
"%s: failed to force tx to idle state\n",
__func__);
reinit++;
}
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
}
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
} else {
printf("resetting\n");
reinit++;
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
}
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
if (reinit == 0) {
CSR_WRITE_4(sc, DC_NETCFG, netcfg);
if (DC_IS_INTEL(sc))
CSR_WRITE_4(sc, DC_NETCFG, netcfg | DC_NETCFG_TX_ON);
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
} else {
Rearrange dc_tx_underrun() a bit to correctly set TX FIFO threshold value. Controllers that always require "store and forward" mode( Davicom and PNIC 82C168) have no way to recover from TX underrun except completely reinitializing hardware. Previously only Davicom was reinitialized and the TX FIFO threshold was changed not to use "store and forward" mode after reinitialization since the default FIFO threshold value was 0. This effectively disabled Davicom controller's "store and forward" mode once it encountered TX underruns. In theory, this can cause watchodg timeouts. Intel 21143 controller requires TX MAC should be idle before changing TX FIFO threshold. So driver tried to disable TX MAC and checked whether it saw the idle state of TX MAC. Driver should perform full hardware reinitialization on failing to enter to idle state and it should not touch TX MAC again once it performed full reinitialization. While I'm here remove resetting TX FIFO threshold to 0 when interface is put into down state. If driver ever encountered TX underrun, it's likely to trigger TX underrun again whenever interface is brought to up again. Keeping old/learned TX FIFO threshold value shall reduce the chance of seeing TX underrns in next run.
2011-03-08 19:49:16 +00:00
sc->dc_ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
dc_init_locked(sc);
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
}
}
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
#ifdef DEVICE_POLLING
static poll_handler_t dc_poll;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
static int
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
dc_poll(struct ifnet *ifp, enum poll_cmd cmd, int count)
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc = ifp->if_softc;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
int rx_npkts = 0;
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
DC_LOCK(sc);
if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) {
DC_UNLOCK(sc);
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
return (rx_npkts);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
}
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
sc->rxcycles = count;
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
rx_npkts = dc_rxeof(sc);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
dc_txeof(sc);
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 (!IFQ_IS_EMPTY(&ifp->if_snd) &&
!(ifp->if_drv_flags & IFF_DRV_OACTIVE))
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_start_locked(ifp);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
if (cmd == POLL_AND_CHECK_STATUS) { /* also check status register */
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t status;
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
status = CSR_READ_4(sc, DC_ISR);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
status &= (DC_ISR_RX_WATDOGTIMEO | DC_ISR_RX_NOBUF |
DC_ISR_TX_NOBUF | DC_ISR_TX_IDLE | DC_ISR_TX_UNDERRUN |
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
DC_ISR_BUS_ERR);
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
if (!status) {
DC_UNLOCK(sc);
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
return (rx_npkts);
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
}
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
/* ack what we have */
CSR_WRITE_4(sc, DC_ISR, status);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (status & (DC_ISR_RX_WATDOGTIMEO | DC_ISR_RX_NOBUF)) {
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t r = CSR_READ_4(sc, DC_FRAMESDISCARDED);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
ifp->if_ierrors += (r & 0xffff) + ((r >> 17) & 0x7ff);
if (dc_rx_resync(sc))
dc_rxeof(sc);
}
/* restart transmit unit if necessary */
if (status & DC_ISR_TX_IDLE && sc->dc_cdata.dc_tx_cnt)
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
if (status & DC_ISR_TX_UNDERRUN)
dc_tx_underrun(sc);
if (status & DC_ISR_BUS_ERR) {
- Consistently use if_printf() only in interface methods: if_start(), if_watchdog, etc., or in functions used only in these methods. In all other functions in the driver use device_printf(). - Use __func__ instead of typing function name. Submitted by: Alex Lyashkov <umka sevcity.net>
2006-09-15 15:16:12 +00:00
if_printf(ifp, "%s: bus error\n", __func__);
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +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
DC_UNLOCK(sc);
When user_frac in the polling subsystem is low it is going to busy the CPU for too long period than necessary. Additively, interfaces are kept polled (in the tick) even if no more packets are available. In order to avoid such situations a new generic mechanism can be implemented in proactive way, keeping track of the time spent on any packet and fragmenting the time for any tick, stopping the processing as soon as possible. In order to implement such mechanism, the polling handler needs to change, returning the number of packets processed. While the intended logic is not part of this patch, the polling KPI is broken by this commit, adding an int return value and the new flag IFCAP_POLLING_NOCOUNT (which will signal that the return value is meaningless for the installed handler and checking should be skipped). Bump __FreeBSD_version in order to signal such situation. Reviewed by: emaste Sponsored by: Sandvine Incorporated
2009-05-30 15:14:44 +00:00
return (rx_npkts);
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
}
#endif /* DEVICE_POLLING */
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_intr(void *arg)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct ifnet *ifp;
s/u_intXX_t/uintXX_t/g
2011-02-19 03:32:10 +00:00
uint32_t r, status;
Remove too expensive bus_dmamap_sync(9) call in dc_rx_resync(). With this change, driver may not notice updated descriptor status change when bounce buffers are active. However, rxeof() in next run will handle the synchronization. Change dc_rxeof() a bit to return the number of processed frames in RX descriptor ring. Previously it returned the number of frames that were successfully passed to upper stack which in turn means it ignored frames that were discarded due to errors. The number of processed frames in RX descriptor ring is used to detect whether driver is out of sync with controller's current descriptor pointer. Returning number of processed frames reduces unnecessary (probably wrong) re-synchronization. Reviewed by: marius
2011-03-16 17:09:51 +00:00
int n;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = arg;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (sc->suspended)
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
return;
Take luigi's suggestion and move the check for nothing to do to before the lock so we don't have lots of null lock/release pairs.
2001-02-20 04:43:59 +00:00
DC_LOCK(sc);
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
status = CSR_READ_4(sc, DC_ISR);
if (status == 0xFFFFFFFF || (status & DC_INTRS) == 0) {
DC_UNLOCK(sc);
return;
}
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->dc_ifp;
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
#ifdef DEVICE_POLLING
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING) {
DC_UNLOCK(sc);
return;
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
#endif
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Disable interrupts. */
CSR_WRITE_4(sc, DC_IMR, 0x00000000);
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
for (n = 16; n > 0; n--) {
if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
break;
/* Ack interrupts. */
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_ISR, status);
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
if (status & DC_ISR_RX_OK) {
Remove too expensive bus_dmamap_sync(9) call in dc_rx_resync(). With this change, driver may not notice updated descriptor status change when bounce buffers are active. However, rxeof() in next run will handle the synchronization. Change dc_rxeof() a bit to return the number of processed frames in RX descriptor ring. Previously it returned the number of frames that were successfully passed to upper stack which in turn means it ignored frames that were discarded due to errors. The number of processed frames in RX descriptor ring is used to detect whether driver is out of sync with controller's current descriptor pointer. Returning number of processed frames reduces unnecessary (probably wrong) re-synchronization. Reviewed by: marius
2011-03-16 17:09:51 +00:00
if (dc_rxeof(sc) == 0) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
while (dc_rx_resync(sc))
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
dc_rxeof(sc);
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
if (status & (DC_ISR_TX_OK | DC_ISR_TX_NOBUF))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
dc_txeof(sc);
if (status & DC_ISR_TX_IDLE) {
dc_txeof(sc);
if (sc->dc_cdata.dc_tx_cnt) {
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_TX_ON);
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
}
}
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
if (status & DC_ISR_TX_UNDERRUN)
dc_tx_underrun(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if ((status & DC_ISR_RX_WATDOGTIMEO)
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
|| (status & DC_ISR_RX_NOBUF)) {
Count how many frames driver lost in interrupt handler. This register is cleared on read so make sure to clear it in driver initialization phase.
2011-02-19 03:01:24 +00:00
r = CSR_READ_4(sc, DC_FRAMESDISCARDED);
ifp->if_ierrors += (r & 0xffff) + ((r >> 17) & 0x7ff);
Remove too expensive bus_dmamap_sync(9) call in dc_rx_resync(). With this change, driver may not notice updated descriptor status change when bounce buffers are active. However, rxeof() in next run will handle the synchronization. Change dc_rxeof() a bit to return the number of processed frames in RX descriptor ring. Previously it returned the number of frames that were successfully passed to upper stack which in turn means it ignored frames that were discarded due to errors. The number of processed frames in RX descriptor ring is used to detect whether driver is out of sync with controller's current descriptor pointer. Returning number of processed frames reduces unnecessary (probably wrong) re-synchronization. Reviewed by: marius
2011-03-16 17:09:51 +00:00
if (dc_rxeof(sc) == 0) {
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
while (dc_rx_resync(sc))
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
dc_rxeof(sc);
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd))
dc_start_locked(ifp);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (status & DC_ISR_BUS_ERR) {
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
DC_UNLOCK(sc);
return;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
status = CSR_READ_4(sc, DC_ISR);
if (status == 0xFFFFFFFF || (status & DC_INTRS) == 0)
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/* Re-enable interrupts. */
Rearrange interrupt handler a bit and remove forever loop. Previously dc(4) always checked whether there is pending interrupts and this consumed a lot of CPU cycles in interrupt handler. Limit the number of processing for TX/RX frames to 16. Also allow sending frames in the loop not to starve TX under high RX load. Reading DC_ISR register should be protected with driver lock, otherwise interrupt handler could be run(e.g. link state change) before the completion of dc_init_locked(). While I'm here remove unneeded code.
2011-02-19 01:37:28 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
CSR_WRITE_4(sc, DC_IMR, DC_INTRS);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
DC_UNLOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Encapsulate an mbuf chain in a descriptor by coupling the mbuf data
* pointers to the fragment pointers.
*/
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
Fix a problem where m_defrag would allocate a new mbuf to replace the chain passed into dc_encap, which dc_start was unaware of. This caused the old (now invalid) mbuf to be passed to BPF_MTAP. Spotted by: Kenjiro Cho <kjc@csl.sony.co.jp>
2003-10-19 23:05:19 +00:00
dc_encap(struct dc_softc *sc, struct mbuf **m_head)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
bus_dma_segment_t segs[DC_MAXFRAGS];
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_t map;
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
struct dc_desc *f;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct mbuf *m;
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
int cur, defragged, error, first, frag, i, idx, nseg;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
m = NULL;
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
defragged = 0;
if (sc->dc_flags & DC_TX_COALESCE &&
((*m_head)->m_next != NULL || sc->dc_flags & DC_TX_ALIGN)) {
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);
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
defragged = 1;
} else {
/*
* Count the number of frags in this chain to see if we
* need to m_collapse. Since the descriptor list is shared
* by all packets, we'll m_collapse long chains so that they
* do not use up the entire list, even if they would fit.
*/
i = 0;
for (m = *m_head; m != NULL; m = m->m_next)
i++;
if (i > DC_TX_LIST_CNT / 4 ||
DC_TX_LIST_CNT - i + sc->dc_cdata.dc_tx_cnt <=
DC_TX_LIST_RSVD) {
Mechanically substitute flags from historic mbuf allocator with malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
m = m_collapse(*m_head, M_NOWAIT, DC_MAXFRAGS);
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
defragged = 1;
}
}
if (defragged != 0) {
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
if (m == NULL) {
m_freem(*m_head);
*m_head = NULL;
Update if_dc to use m_defrag, removing the semi-duplicate dc_coal function. Also, use m_defrag where appropriate to defrag long mbuf chains in the same fashion as was done in if_sis.c. Before this change, if_dc would blow up and take down the interface if fed a really long mbuf chain. MFC after: 2 weeks
2003-03-29 21:27:07 +00:00
return (ENOBUFS);
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
}
Fix a problem where m_defrag would allocate a new mbuf to replace the chain passed into dc_encap, which dc_start was unaware of. This caused the old (now invalid) mbuf to be passed to BPF_MTAP. Spotted by: Kenjiro Cho <kjc@csl.sony.co.jp>
2003-10-19 23:05:19 +00:00
*m_head = m;
Update if_dc to use m_defrag, removing the semi-duplicate dc_coal function. Also, use m_defrag where appropriate to defrag long mbuf chains in the same fashion as was done in if_sis.c. Before this change, if_dc would blow up and take down the interface if fed a really long mbuf chain. MFC after: 2 weeks
2003-03-29 21:27:07 +00:00
}
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
idx = sc->dc_cdata.dc_tx_prod;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
error = bus_dmamap_load_mbuf_sg(sc->dc_tx_mtag,
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
sc->dc_cdata.dc_tx_map[idx], *m_head, segs, &nseg, 0);
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
if (defragged != 0 || (m = m_collapse(*m_head, M_NOWAIT,
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
DC_MAXFRAGS)) == NULL) {
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
m_freem(*m_head);
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
*m_head = NULL;
- Use m_collapse(9) instead of m_defrag(9) if possible. This results in a noticeable reduction in system time spent. - If bus_dmamap_load_mbuf_sg(9) fails with EFBIG and we already have defragmented the mbuf chain, don't bother to defragment and load it a second time just yet as it's likely to fail again anyway. MFC after: 3 days
2008-08-23 20:57:48 +00:00
return (defragged != 0 ? error : ENOBUFS);
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
}
*m_head = m;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
error = bus_dmamap_load_mbuf_sg(sc->dc_tx_mtag,
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
sc->dc_cdata.dc_tx_map[idx], *m_head, segs, &nseg, 0);
if (error != 0) {
m_freem(*m_head);
*m_head = NULL;
return (error);
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
}
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
} else if (error != 0)
return (error);
KASSERT(nseg <= DC_MAXFRAGS,
("%s: wrong number of segments (%d)", __func__, nseg));
if (nseg == 0) {
m_freem(*m_head);
*m_head = NULL;
return (EIO);
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
}
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
/* Check descriptor overruns. */
if (sc->dc_cdata.dc_tx_cnt + nseg > DC_TX_LIST_CNT - DC_TX_LIST_RSVD) {
bus_dmamap_unload(sc->dc_tx_mtag, sc->dc_cdata.dc_tx_map[idx]);
return (ENOBUFS);
}
bus_dmamap_sync(sc->dc_tx_mtag, sc->dc_cdata.dc_tx_map[idx],
BUS_DMASYNC_PREWRITE);
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
first = cur = frag = sc->dc_cdata.dc_tx_prod;
for (i = 0; i < nseg; i++) {
if ((sc->dc_flags & DC_TX_ADMTEK_WAR) &&
(frag == (DC_TX_LIST_CNT - 1)) &&
(first != sc->dc_cdata.dc_tx_first)) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_unload(sc->dc_tx_mtag,
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
sc->dc_cdata.dc_tx_map[first]);
m_freem(*m_head);
*m_head = NULL;
return (ENOBUFS);
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
f = &sc->dc_ldata.dc_tx_list[frag];
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
f->dc_ctl = htole32(DC_TXCTL_TLINK | segs[i].ds_len);
if (i == 0) {
f->dc_status = 0;
f->dc_ctl |= htole32(DC_TXCTL_FIRSTFRAG);
} else
f->dc_status = htole32(DC_TXSTAT_OWN);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
f->dc_data = htole32(DC_ADDR_LO(segs[i].ds_addr));
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
cur = frag;
DC_INC(frag, DC_TX_LIST_CNT);
}
sc->dc_cdata.dc_tx_prod = frag;
sc->dc_cdata.dc_tx_cnt += nseg;
sc->dc_cdata.dc_tx_chain[cur] = *m_head;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_tx_list[cur].dc_ctl |= htole32(DC_TXCTL_LASTFRAG);
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
if (sc->dc_flags & DC_TX_INTR_FIRSTFRAG)
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_tx_list[first].dc_ctl |=
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
htole32(DC_TXCTL_FINT);
if (sc->dc_flags & DC_TX_INTR_ALWAYS)
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_tx_list[cur].dc_ctl |= htole32(DC_TXCTL_FINT);
For controllers that have TX interrupt moderation capability, request TX completion interrupt for every 8-th frames. Previously dc(4) requested TX completion interrupt if number of queued TX descriptors is greater than 64. This caused a lot of TX completion interrupt under high TX load once driver queued more than 64 TX descriptors. It's quite normal to see more than 64 queued TX descriptors under high TX load. This change reduces the number of TX completion interrupts to be less than 17k under high TX load. Because this change does not generate TX completion interrupt for each frame, add reclaiming transmitted buffers in dc_tick not to generate false watchdog timeouts. While I'm here add check for queued descriptors in dc_txeof() since there is no more work to do when there is no pending descriptors.
2011-02-18 20:38:05 +00:00
if (sc->dc_flags & DC_TX_USE_TX_INTR &&
++sc->dc_cdata.dc_tx_pkts >= 8) {
sc->dc_cdata.dc_tx_pkts = 0;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_tx_list[cur].dc_ctl |= htole32(DC_TXCTL_FINT);
For controllers that have TX interrupt moderation capability, request TX completion interrupt for every 8-th frames. Previously dc(4) requested TX completion interrupt if number of queued TX descriptors is greater than 64. This caused a lot of TX completion interrupt under high TX load once driver queued more than 64 TX descriptors. It's quite normal to see more than 64 queued TX descriptors under high TX load. This change reduces the number of TX completion interrupts to be less than 17k under high TX load. Because this change does not generate TX completion interrupt for each frame, add reclaiming transmitted buffers in dc_tick not to generate false watchdog timeouts. While I'm here add check for queued descriptors in dc_txeof() since there is no more work to do when there is no pending descriptors.
2011-02-18 20:38:05 +00:00
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_ldata.dc_tx_list[first].dc_status = htole32(DC_TXSTAT_OWN);
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
/*
* Swap the last and the first dmamaps to ensure the map for
* this transmission is placed at the last descriptor.
*/
map = sc->dc_cdata.dc_tx_map[cur];
sc->dc_cdata.dc_tx_map[cur] = sc->dc_cdata.dc_tx_map[first];
sc->dc_cdata.dc_tx_map[first] = map;
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_start(struct ifnet *ifp)
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
{
struct dc_softc *sc;
sc = ifp->if_softc;
DC_LOCK(sc);
dc_start_locked(ifp);
DC_UNLOCK(sc);
}
- Const'ify the dc_devs array. - Correct the maxsize parameter when creating the mbufs busdma tag to reflect the actual requirement of dc(4). - Move the KASSERT in dc_newbuf() to the right spot. - Also convert the TX side to take advantage of bus_dmamap_load_mbuf_sg(9). - Move the comment regarding dc_start_locked() to the right spot. MFC after: 2 weeks
2008-03-24 17:38:24 +00:00
/*
* Main transmit routine
* To avoid having to do mbuf copies, we put pointers to the mbuf data
* regions directly in the transmit lists. We also save a copy of the
* pointers since the transmit list fragment pointers are physical
* addresses.
*/
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
static void
dc_start_locked(struct ifnet *ifp)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
struct mbuf *m_head;
int queued;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = ifp->if_softc;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK_ASSERT(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
Send frames only when there is a valid link and driver is running as well as controller has enough free TX descriptors. Remove check for number of queued frames before attempting to transmit. I guess it was added to allow draining queued frames even if there is no link. I'm under the impression this type of check should be done in upper layer. No other drivers in tree do that.
2011-02-19 00:00:32 +00:00
if ((ifp->if_drv_flags & (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)) !=
IFF_DRV_RUNNING || sc->dc_link == 0)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
return;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
sc->dc_cdata.dc_tx_first = sc->dc_cdata.dc_tx_prod;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
for (queued = 0; !IFQ_DRV_IS_EMPTY(&ifp->if_snd); ) {
/*
* If there's no way we can send any packets, return now.
*/
if (sc->dc_cdata.dc_tx_cnt > DC_TX_LIST_CNT - DC_TX_LIST_RSVD) {
ifp->if_drv_flags |= IFF_DRV_OACTIVE;
break;
}
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (m_head == NULL)
break;
Fix a problem where m_defrag would allocate a new mbuf to replace the chain passed into dc_encap, which dc_start was unaware of. This caused the old (now invalid) mbuf to be passed to BPF_MTAP. Spotted by: Kenjiro Cho <kjc@csl.sony.co.jp>
2003-10-19 23:05:19 +00:00
if (dc_encap(sc, &m_head)) {
o In order to reduce bug and code duplication fold handling of NICs requiring DC_TX_ALIGN or DC_TX_COALESCE, which was previously done in dc_start_locked(), into dc_encap(). o In dc_encap(): - If m_defrag() fails just drop the packet like other NIC drivers do. This should only happen when there's a mbuf shortage, in which case it was possible to end up with an IFQ full of packets which couldn't be processed as they couldn't be defragmented as they were taking up all the mbufs themselves. This includes adjusting dc_start_locked() to not trying to prepend the mbuf (chain) if dc_encap() has freed it. - Likewise, if bus_dmamap_load_mbuf() fails as dc_dma_map_txbuf() failed, free the mbuf possibly allocated by the above call to m_defrag() and drop the packet. o In dc_txeof(): - Don't clear IFF_DRV_OACTIVE unless there are at least 6 free TX descriptors. Further down the road dc_encap() will bail if there are only 5 or fewer free TX descriptors, causing dc_start_locked() to abort and prepend the dequeued mbuf again so it makes no sense to pretend we could process mbufs again when in fact we won't. While at it replace this magic 5 with a macro DC_TX_LIST_RSVD. - Just always assign idx to sc->dc_cdata.dc_tx_cons; it doesn't make much sense to exclude the idx == sc->dc_cdata.dc_tx_cons case. o In dc_dma_map_txbuf() there's no need to set sc->dc_cdata.dc_tx_err to error if the latter is != 0, bus_dmamap_load_mbuf() already returns the same error value in that case anyway. o For less overhead, convert to use bus_dmamap_load_mbuf_sg() for loading RX buffers. o Remove some banal and/or outdated comments. Approved by: re (kensmith) MFC after: 1 week
2007-08-05 11:28:19 +00:00
if (m_head == NULL)
break;
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +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;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
queued++;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
if (queued > 0) {
/* Transmit */
if (!(sc->dc_flags & DC_TX_POLL))
CSR_WRITE_4(sc, DC_TXSTART, 0xFFFFFFFF);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
/*
* Set a timeout in case the chip goes out to lunch.
*/
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
sc->dc_wdog_timer = 5;
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_init(void *xsc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc = xsc;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
dc_init_locked(sc);
Remove various bits of conditional Alpha code and fixup a few comments.
2006-05-12 05:04:46 +00:00
DC_UNLOCK(sc);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
}
static void
dc_init_locked(struct dc_softc *sc)
{
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->dc_ifp;
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct mii_data *mii;
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
struct ifmedia *ifm;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK_ASSERT(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
if ((ifp->if_drv_flags & IFF_DRV_RUNNING) != 0)
return;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii = device_get_softc(sc->dc_miibus);
/*
* Cancel pending I/O and free all RX/TX buffers.
*/
dc_stop(sc);
dc_reset(sc);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
if (DC_IS_INTEL(sc)) {
ifm = &mii->mii_media;
dc_apply_fixup(sc, ifm->ifm_media);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Set cache alignment and burst length.
*/
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
if (DC_IS_ASIX(sc) || DC_IS_DAVICOM(sc) || DC_IS_ULI(sc))
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_BUSCTL, 0);
else
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
CSR_WRITE_4(sc, DC_BUSCTL, DC_BUSCTL_MRME | DC_BUSCTL_MRLE);
Enable round-robin arbitration between transmit and receive unit in the 21143, instead of giving priority to the receive unit. This gives a 10-15% performance improvement in the forwarding rate under heavy load. Reviewed-by: Bill Paul
2001-10-27 00:59:17 +00:00
/*
* Evenly share the bus between receive and transmit process.
*/
if (DC_IS_INTEL(sc))
DC_SETBIT(sc, DC_BUSCTL, DC_BUSCTL_ARBITRATION);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (DC_IS_DAVICOM(sc) || DC_IS_INTEL(sc)) {
DC_SETBIT(sc, DC_BUSCTL, DC_BURSTLEN_USECA);
} else {
DC_SETBIT(sc, DC_BUSCTL, DC_BURSTLEN_16LONG);
}
if (sc->dc_flags & DC_TX_POLL)
DC_SETBIT(sc, DC_BUSCTL, DC_TXPOLL_1);
switch(sc->dc_cachesize) {
case 32:
DC_SETBIT(sc, DC_BUSCTL, DC_CACHEALIGN_32LONG);
break;
case 16:
DC_SETBIT(sc, DC_BUSCTL, DC_CACHEALIGN_16LONG);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case 8:
DC_SETBIT(sc, DC_BUSCTL, DC_CACHEALIGN_8LONG);
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case 0:
default:
DC_SETBIT(sc, DC_BUSCTL, DC_CACHEALIGN_NONE);
break;
}
if (sc->dc_flags & DC_TX_STORENFWD)
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_STORENFWD);
else {
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
if (sc->dc_txthresh > DC_TXTHRESH_MAX) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_STORENFWD);
} else {
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_STORENFWD);
DC_SETBIT(sc, DC_NETCFG, sc->dc_txthresh);
}
}
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_NO_RXCRC);
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_TX_BACKOFF);
if (DC_IS_MACRONIX(sc) || DC_IS_PNICII(sc)) {
/*
* The app notes for the 98713 and 98715A say that
* in order to have the chips operate properly, a magic
* number must be written to CSR16. Macronix does not
* document the meaning of these bits so there's no way
* to know exactly what they do. The 98713 has a magic
* number all its own; the rest all use a different one.
*/
DC_CLRBIT(sc, DC_MX_MAGICPACKET, 0xFFFF0000);
if (sc->dc_type == DC_TYPE_98713)
DC_SETBIT(sc, DC_MX_MAGICPACKET, DC_MX_MAGIC_98713);
else
DC_SETBIT(sc, DC_MX_MAGICPACKET, DC_MX_MAGIC_98715);
}
NEWCARD/Cardbus - This commit adds support for Xircom X3201 based cardbus cards. Support for the TDK 78Q2120 MII is also added. IBM Etherjet, Intel and Xircom cards uses these chips. Note that as a result of this commit, some Intel/DEC 21143 based cardbus cards will also attach, but not get link. That is being looked at.
2000-10-19 08:34:32 +00:00
if (DC_IS_XIRCOM(sc)) {
/*
* setup General Purpose Port mode and data so the tulip
* can talk to the MII.
*/
CSR_WRITE_4(sc, DC_SIAGP, DC_SIAGP_WRITE_EN | DC_SIAGP_INT1_EN |
DC_SIAGP_MD_GP2_OUTPUT | DC_SIAGP_MD_GP0_OUTPUT);
DELAY(10);
CSR_WRITE_4(sc, DC_SIAGP, DC_SIAGP_INT1_EN |
DC_SIAGP_MD_GP2_OUTPUT | DC_SIAGP_MD_GP0_OUTPUT);
DELAY(10);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_TX_THRESH);
Apply patch to allow TX underrun handling without issuing a complete chip reset. Just temporarily turn off the transmitter instead. Submitted by: Stephen McKay <mckay@freebsd.org>
2001-02-22 19:26:55 +00:00
DC_SETBIT(sc, DC_NETCFG, DC_TXTHRESH_MIN);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Init circular RX list. */
if (dc_list_rx_init(sc) == ENOBUFS) {
- Consistently use if_printf() only in interface methods: if_start(), if_watchdog, etc., or in functions used only in these methods. In all other functions in the driver use device_printf(). - Use __func__ instead of typing function name. Submitted by: Alex Lyashkov <umka sevcity.net>
2006-09-15 15:16:12 +00:00
device_printf(sc->dc_dev,
Use device_printf() and if_printf() and remove dc_unit from softc.
2005-08-10 20:33:46 +00:00
"initialization failed: no memory for rx buffers\n");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
dc_stop(sc);
return;
}
/*
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
* Init TX descriptors.
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
*/
dc_list_tx_init(sc);
/*
* Load the address of the RX list.
*/
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
CSR_WRITE_4(sc, DC_RXADDR, DC_RXDESC(sc, 0));
CSR_WRITE_4(sc, DC_TXADDR, DC_TXDESC(sc, 0));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Enable interrupts.
*/
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
#ifdef DEVICE_POLLING
/*
* ... but only if we are not polling, and make sure they are off in
* the case of polling. Some cards (e.g. fxp) turn interrupts on
* after a reset.
*/
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING)
Device Polling code for -current. Non-SMP, i386-only, no polling in the idle loop at the moment. To use this code you must compile a kernel with options DEVICE_POLLING and at runtime enable polling with sysctl kern.polling.enable=1 The percentage of CPU reserved to userland can be set with sysctl kern.polling.user_frac=NN (default is 50) while the remainder is used by polling device drivers and netisr's. These are the only two variables that you should need to touch. There are a few more parameters in kern.polling but the default values are adequate for all purposes. See the code in kern_poll.c for more details on them. Polling in the idle loop will be implemented shortly by introducing a kernel thread which does the job. Until then, the amount of CPU dedicated to polling will never exceed (100-user_frac). The equivalent (actually, better) code for -stable is at http://info.iet.unipi.it/~luigi/polling/ and also supports polling in the idle loop. NOTE to Alpha developers: There is really nothing in this code that is i386-specific. If you move the 2 lines supporting the new option from sys/conf/{files,options}.i386 to sys/conf/{files,options} I am pretty sure that this should work on the Alpha as well, just that I do not have a suitable test box to try it. If someone feels like trying it, I would appreciate it. NOTE to other developers: sure some things could be done better, and as always I am open to constructive criticism, which a few of you have already given and I greatly appreciated. However, before proposing radical architectural changes, please take some time to possibly try out this code, or at the very least read the comments in kern_poll.c, especially re. the reason why I am using a soft netisr and cannot (I believe) replace it with a simple timeout. Quick description of files touched by this commit: sys/conf/files.i386 new file kern/kern_poll.c sys/conf/options.i386 new option sys/i386/i386/trap.c poll in trap (disabled by default) sys/kern/kern_clock.c initialization and hardclock hooks. sys/kern/kern_intr.c minor swi_net changes sys/kern/kern_poll.c the bulk of the code. sys/net/if.h new flag sys/net/if_var.h declaration for functions used in device drivers. sys/net/netisr.h NETISR_POLL sys/dev/fxp/if_fxp.c sys/dev/fxp/if_fxpvar.h sys/pci/if_dc.c sys/pci/if_dcreg.h sys/pci/if_sis.c sys/pci/if_sisreg.h device driver modifications
2001-12-14 17:56:12 +00:00
CSR_WRITE_4(sc, DC_IMR, 0x00000000);
else
#endif
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_IMR, DC_INTRS);
CSR_WRITE_4(sc, DC_ISR, 0xFFFFFFFF);
Add support for ALi/ULi, now NVIDIA, M5261/M5263 PCI FastEthernet controller which is found on ULi M1563 South Bridge & M1689 Bridge. These controllers look like a tulip clone. M5263 controller does not support MII bitbang so use DC_ROM register to access MII registers. Like other tulip variants, ULi controller uses a setup frame to configure RX filter and uses new setup frame format. It's not clear to me whether the controller supports a hash based multicast filtering so this patch uses 14 perfect multicast filter to filter multicast frames. If number of multicast addresses is greater than 14, controller is put into a mode that receives all multicast frames. Due to lack of access to M5261, this change was not tested with M5261 but it probably works. Many thanks to Marco who provided remote access to M5263. Tested by: Marco Steinbach <coco <> executive-computing dot de>, Martin MATO <martin.mato <> orange dot fr>
2011-10-24 20:48:02 +00:00
/* Initialize TX jabber and RX watchdog timer. */
if (DC_IS_ULI(sc))
CSR_WRITE_4(sc, DC_WATCHDOG, DC_WDOG_JABBERCLK |
DC_WDOG_HOSTUNJAB);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/* Enable transmitter. */
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_TX_ON);
Make the blinkylights on non-MII 21143 cards work. We need to enable the link and activity LED control bits in CSR15 in order for the controller to drive the LEDs correctly. This was largely done for the ZNYX multiport cards, but should also work with the DEC DE500-BA and other non-MII cards.
2000-09-01 23:59:54 +00:00
/*
* If this is an Intel 21143 and we're not using the
* MII port, program the LED control pins so we get
* link and activity indications.
*/
Special-case the LED twiddling code so that it doesn't do anything on the NEC VersaPro NoteBook PC. This 21143 implementation has no LEDs, and flipping the LED control bits somehow stops it from establishing a link. We check the subsystem ID and don't flip the LED control bits for the NEC NIC.
2000-09-07 18:51:04 +00:00
if (sc->dc_flags & DC_TULIP_LEDS) {
Make the blinkylights on non-MII 21143 cards work. We need to enable the link and activity LED control bits in CSR15 in order for the controller to drive the LEDs correctly. This was largely done for the ZNYX multiport cards, but should also work with the DEC DE500-BA and other non-MII cards.
2000-09-01 23:59:54 +00:00
CSR_WRITE_4(sc, DC_WATCHDOG,
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
DC_WDOG_CTLWREN | DC_WDOG_LINK | DC_WDOG_ACTIVITY);
Special-case the LED twiddling code so that it doesn't do anything on the NEC VersaPro NoteBook PC. This 21143 implementation has no LEDs, and flipping the LED control bits somehow stops it from establishing a link. We check the subsystem ID and don't flip the LED control bits for the NEC NIC.
2000-09-07 18:51:04 +00:00
CSR_WRITE_4(sc, DC_WATCHDOG, 0);
Make the blinkylights on non-MII 21143 cards work. We need to enable the link and activity LED control bits in CSR15 in order for the controller to drive the LEDs correctly. This was largely done for the ZNYX multiport cards, but should also work with the DEC DE500-BA and other non-MII cards.
2000-09-01 23:59:54 +00:00
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Load the RX/multicast filter. We do this sort of late
* because the filter programming scheme on the 21143 and
* some clones requires DMAing a setup frame via the TX
* engine, and we need the transmitter enabled for that.
*/
dc_setfilt(sc);
/* Enable receiver. */
DC_SETBIT(sc, DC_NETCFG, DC_NETCFG_RX_ON);
CSR_WRITE_4(sc, DC_RXSTART, 0xFFFFFFFF);
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;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
dc_ifmedia_upd_locked(sc);
Apply special PHY initialization code for 21143 controllers before any other media configuration. Otherwise some 21143 controller cannot establish a link. While I'm here remove the PHY initialization code in dc_setcfg(). Since dc_setcfg() is called whenever link state is changed, having the PHY initialization code in dc_setcfg() resulted in continuous link flips. After driver resets SIA, use default SIA transmit/receive configuration instead of disabling autosense/autonegotiation. Otherwise, controller fails to establish a link as well as losing auto-negotiation capability. For manual media configuration, always configure 21143 controller with specified media to ensure media change. This change makes ANA-6922 establish link with/without auto-negotiation. While I'm here be more strict on link UP/DOWN detection logic. Many thanks to marius who fixed several bugs in initial patch and even tested the patch on a couple of dc(4) controllers. PR: kern/79262 Reviewed by: marius Tested by: marius
2011-02-18 02:58:09 +00:00
Count how many frames driver lost in interrupt handler. This register is cleared on read so make sure to clear it in driver initialization phase.
2011-02-19 03:01:24 +00:00
/* Clear missed frames and overflow counter. */
CSR_READ_4(sc, DC_FRAMESDISCARDED);
If this is a Davicom DM9102A and we're enabling the homePNA link, force dc_link to 1 and don't activate the tick routine. Without this, dc_start() always thinks the link is down and never transmits in homePNA mode.
2000-09-20 00:59:17 +00:00
/* Don't start the ticker if this is a homePNA link. */
Move us yet closer to IFM_* definitions in NetBSD.
2002-04-29 05:32:44 +00:00
if (IFM_SUBTYPE(mii->mii_media.ifm_media) == IFM_HPNA_1)
If this is a Davicom DM9102A and we're enabling the homePNA link, force dc_link to 1 and don't activate the tick routine. Without this, dc_start() always thinks the link is down and never transmits in homePNA mode.
2000-09-20 00:59:17 +00:00
sc->dc_link = 1;
else {
if (sc->dc_flags & DC_21143_NWAY)
Change the driver to allocate its own callout structure, and modify the interface to use callout_* instead of timeout(). Also add an IS_MPSAFE #define (currently off) which will mark the driver as mpsafe to the upper layers.
2000-11-25 08:00:10 +00:00
callout_reset(&sc->dc_stat_ch, hz/10, dc_tick, sc);
If this is a Davicom DM9102A and we're enabling the homePNA link, force dc_link to 1 and don't activate the tick routine. Without this, dc_start() always thinks the link is down and never transmits in homePNA mode.
2000-09-20 00:59:17 +00:00
else
Change the driver to allocate its own callout structure, and modify the interface to use callout_* instead of timeout(). Also add an IS_MPSAFE #define (currently off) which will mark the driver as mpsafe to the upper layers.
2000-11-25 08:00:10 +00:00
callout_reset(&sc->dc_stat_ch, hz, dc_tick, sc);
If this is a Davicom DM9102A and we're enabling the homePNA link, force dc_link to 1 and don't activate the tick routine. Without this, dc_start() always thinks the link is down and never transmits in homePNA mode.
2000-09-20 00:59:17 +00:00
}
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
sc->dc_wdog_timer = 0;
callout_reset(&sc->dc_wdog_ch, hz, dc_watchdog, sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_ifmedia_upd(struct ifnet *ifp)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
int error;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = ifp->if_softc;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
error = dc_ifmedia_upd_locked(sc);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
return (error);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
When driver is run for the first time there would be no established link such that calling dc_setcfg() right after media change would be meaningless unless controller in question is not Davicom DM9102. Ideally dc_setcfg() should be called when speed/duplex is resolved otherwise it would reprogram controller with wrong speed/duplex information. Because MII status change callback already calls dc_setcfg() I think calling dc_setcfg() in dc_init_locked() is wrong. For instance, it would take some time to establish a link after mii_mediachg(), so blindly calling dc_setcfg() right after mii_mediachg() will always yield wrong media configuration. Extend dc_ifmedia_upd() to handle media change and still allow 21143 and Davidcom controllers program speed/duplex regardless of current resolved speed/duplex of link. In theory 21143 may not need to call dc_setcfg() right after media change, but leave it as it is because there are too many variants to test that change. Probably dc(4) shall need a PHY reset in dc_ifmedia_upd() but it's hard to verify correctness of the change. This change reliably makes ULi M5263 establish a link. While I'm here correctly report media change result. Previously it always reported a success.
2011-10-24 20:26:37 +00:00
static int
dc_ifmedia_upd_locked(struct dc_softc *sc)
{
struct mii_data *mii;
struct ifmedia *ifm;
int error;
DC_LOCK_ASSERT(sc);
sc->dc_link = 0;
mii = device_get_softc(sc->dc_miibus);
error = mii_mediachg(mii);
if (error == 0) {
ifm = &mii->mii_media;
if (DC_IS_INTEL(sc))
dc_setcfg(sc, ifm->ifm_media);
else if (DC_IS_DAVICOM(sc) &&
IFM_SUBTYPE(ifm->ifm_media) == IFM_HPNA_1)
dc_setcfg(sc, ifm->ifm_media);
}
return (error);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* 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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct mii_data *mii;
struct ifmedia *ifm;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = ifp->if_softc;
mii = device_get_softc(sc->dc_miibus);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
mii_pollstat(mii);
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
ifm = &mii->mii_media;
if (DC_IS_DAVICOM(sc)) {
Move us yet closer to IFM_* definitions in NetBSD.
2002-04-29 05:32:44 +00:00
if (IFM_SUBTYPE(ifm->ifm_media) == IFM_HPNA_1) {
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
ifmr->ifm_active = ifm->ifm_media;
ifmr->ifm_status = 0;
- Wrap code optimized for architectures without alignment constraints in #ifdef __NO_STRICT_ALIGNMENT rather than #ifdef __i386__. This means that amd64 now also uses the optimized code. [1] While at it, fix a nearby style(9) bug. - Remove the hw.dc_quick SYSCTL, which allowed to turn off the above mentioned optimization, as like the equivalent and already removed - In dc_setcfg() suppress printing a warning when forcing the receiver and transceiver to idle state times out for chips where the status bits in question just never change (observed in detail with DM9102A) and therefore the warning would be highly likely false positive. [2] - In dc_ifmedia_sts() add a missing DC_UNLOCK(). Tested by: Hans-Joerg Sirtl on amd64 [1] PR: 82681 [2] Obtained from: NetBSD tlp(4) [2] MFC after: 1 week
2006-10-29 20:24:27 +00:00
DC_UNLOCK(sc);
Add support for DM9102A boards with Davicom DM9801 HomePNA PHYs.
2000-01-24 17:19:37 +00:00
return;
}
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifmr->ifm_active = mii->mii_media_active;
ifmr->ifm_status = mii->mii_media_status;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc = ifp->if_softc;
struct ifreq *ifr = (struct ifreq *)data;
struct mii_data *mii;
int error = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
switch (command) {
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
case SIOCSIFFLAGS:
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
if (ifp->if_flags & IFF_UP) {
Fix handling of IFF_ALLMULTI. The same bug in various forms affects the following drivers: dc mn sf sk ste ti tl xl an bge em gem gx ie lge sr aue cue kue wi xe Approved by: re
2002-11-25 19:28:01 +00:00
int need_setfilt = (ifp->if_flags ^ sc->dc_if_flags) &
(IFF_PROMISC | IFF_ALLMULTI);
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 handling of IFF_ALLMULTI. The same bug in various forms affects the following drivers: dc mn sf sk ste ti tl xl an bge em gem gx ie lge sr aue cue kue wi xe Approved by: re
2002-11-25 19:28:01 +00:00
if (need_setfilt)
dc_setfilt(sc);
} else {
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
dc_stop(sc);
}
sc->dc_if_flags = ifp->if_flags;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
There is no need to execute filter configuration when driver is not running. Remove wrong driver state change in dc_setfilt_xircom(). While I'm here nuke unnecessary assignments.
2011-02-18 23:08:05 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
dc_setfilt(sc);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
case SIOCGIFMEDIA:
case SIOCSIFMEDIA:
mii = device_get_softc(sc->dc_miibus);
error = ifmedia_ioctl(ifp, ifr, &mii->mii_media, command);
break;
First driver with user-configurable polling(4).
2004-04-11 13:47:15 +00:00
case SIOCSIFCAP:
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#ifdef DEVICE_POLLING
if (ifr->ifr_reqcap & IFCAP_POLLING &&
!(ifp->if_capenable & IFCAP_POLLING)) {
error = ether_poll_register(dc_poll, ifp);
if (error)
return(error);
DC_LOCK(sc);
/* Disable interrupts */
CSR_WRITE_4(sc, DC_IMR, 0x00000000);
ifp->if_capenable |= IFCAP_POLLING;
DC_UNLOCK(sc);
return (error);
}
if (!(ifr->ifr_reqcap & IFCAP_POLLING) &&
ifp->if_capenable & IFCAP_POLLING) {
error = ether_poll_deregister(ifp);
/* Enable interrupts. */
DC_LOCK(sc);
CSR_WRITE_4(sc, DC_IMR, DC_INTRS);
ifp->if_capenable &= ~IFCAP_POLLING;
DC_UNLOCK(sc);
return (error);
}
#endif /* DEVICE_POLLING */
First driver with user-configurable polling(4).
2004-04-11 13:47:15 +00:00
break;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
break;
}
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
return (error);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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 callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
dc_watchdog(void *xsc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
struct dc_softc *sc = xsc;
struct ifnet *ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
DC_LOCK_ASSERT(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
if (sc->dc_wdog_timer == 0 || --sc->dc_wdog_timer != 0) {
callout_reset(&sc->dc_wdog_ch, hz, dc_watchdog, sc);
return;
}
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
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
ifp = sc->dc_ifp;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
ifp->if_oerrors++;
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
device_printf(sc->dc_dev, "watchdog timeout\n");
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Remove unnecessary controller reinitialization which resulted in link flips during alias address insertion or dhclient operation. While I'm here remove dc_reset() in DC_ISR_BUS_ERR case. Device is fully reinitialized again in dc_init_locked().
2011-02-23 02:40:31 +00:00
ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Add ALTQ support for dc(4), based upon a mostly-working patch from mlaier.
2004-10-01 07:04:09 +00:00
if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd))
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_start_locked(ifp);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
/*
* Stop the adapter and free any mbufs allocated to the
* RX and TX lists.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_stop(struct dc_softc *sc)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct ifnet *ifp;
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
struct dc_list_data *ld;
struct dc_chain_data *cd;
int i;
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
uint32_t ctl, netcfg;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK_ASSERT(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
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->dc_ifp;
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
ld = &sc->dc_ldata;
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
cd = &sc->dc_cdata;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
Change the driver to allocate its own callout structure, and modify the interface to use callout_* instead of timeout(). Also add an IS_MPSAFE #define (currently off) which will mark the driver as mpsafe to the upper layers.
2000-11-25 08:00:10 +00:00
callout_stop(&sc->dc_stat_ch);
- Use our own callout (the dc_tick() callout uses varying periods depending on the NIC and isn't used at all with HomePNA links) instead of if_slowtimo() for driving dc_watchdog() in order to avoid races accessing if_timer. - Use bus_get_dma_tag() so dc(4) works on platforms requiring it. - Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that. - Remove an alpha remnant in dc_softc.
2006-12-06 01:56:38 +00:00
callout_stop(&sc->dc_wdog_ch);
sc->dc_wdog_timer = 0;
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
sc->dc_link = 0;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +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);
MFS (merge from stable): rev 1.9.2.28, fix ordering of IFF_RUNNING mods. The reason we are required to commit to -current first is so that later MFC's do not risk the loss of existing bug fixes. Even if this was not strictly required in -current, it should still be fixed there too.
2001-12-07 00:57:57 +00:00
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
netcfg = CSR_READ_4(sc, DC_NETCFG);
if (netcfg & (DC_NETCFG_RX_ON | DC_NETCFG_TX_ON))
CSR_WRITE_4(sc, DC_NETCFG,
netcfg & ~(DC_NETCFG_RX_ON | DC_NETCFG_TX_ON));
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_IMR, 0x00000000);
Revise r220046 by introducing dc_netcfg_wait() which waits the end of active DMA cycle. dc_setcfg() also has to wait until the DMA engine is stopped so using a common function to handle the job is better than duplicating the code. No objection from: marius
2011-03-28 19:08:53 +00:00
/* Wait the completion of TX/RX SM. */
if (netcfg & (DC_NETCFG_RX_ON | DC_NETCFG_TX_ON))
dc_netcfg_wait(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
CSR_WRITE_4(sc, DC_TXADDR, 0x00000000);
CSR_WRITE_4(sc, DC_RXADDR, 0x00000000);
/*
* Free data in the RX lists.
*/
for (i = 0; i < DC_RX_LIST_CNT; i++) {
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
if (cd->dc_rx_chain[i] != NULL) {
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bus_dmamap_sync(sc->dc_rx_mtag,
cd->dc_rx_map[i], BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc->dc_rx_mtag,
cd->dc_rx_map[i]);
Convert the dc(4) driver to the busdma API. This is a necessary step to have this driver working on sparc64. It still needs to be made endian-clean before it can work there. Special thanks to dragonk@evilcode.net for sending me a dc(4) card so that I was able to do this work. Many cheers to all the people that tested this change, thanks to them, this change shouldn't break anything :-). Tested by: marcel (i386 and ia64), ru (i386), wilko (alpha), mbr (i386), wpaul (i386) and Will Saxon <WillS@housing.ufl.edu> (i386)
2003-07-06 21:45:31 +00:00
m_freem(cd->dc_rx_chain[i]);
cd->dc_rx_chain[i] = NULL;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bzero(ld->dc_rx_list, DC_RX_LIST_SZ);
bus_dmamap_sync(sc->dc_rx_ltag, sc->dc_rx_lmap,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Free the TX list buffers.
*/
for (i = 0; i < DC_TX_LIST_CNT; i++) {
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
if (cd->dc_tx_chain[i] != NULL) {
Make the dc(4) driver endian-clean, so to that it works on sparc64. There are such cards in Netra X1 boxes, which should thus be fully supported now. Tested by: jake
2003-07-09 15:03:10 +00:00
ctl = le32toh(ld->dc_tx_list[i].dc_ctl);
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
if (ctl & DC_TXCTL_SETUP) {
bus_dmamap_sync(sc->dc_stag, sc->dc_smap,
BUS_DMASYNC_POSTWRITE);
} else {
bus_dmamap_sync(sc->dc_tx_mtag,
cd->dc_tx_map[i], BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->dc_tx_mtag,
cd->dc_tx_map[i]);
m_freem(cd->dc_tx_chain[i]);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Various style(9) and readability fixes.
2003-07-06 14:36:33 +00:00
cd->dc_tx_chain[i] = NULL;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
}
Split common TX/RX descriptor DMA tag to TX and RX DMA tags respectively and fix all bus_dma(9) issues seen when bounce buffers are used. o Setup frame handling had no bus_dmamap_sync(9) which prevented driver from configuring RX filter. Add missing bus_dmamap_sync(9) in both dc_setfilt_21143()/dc_setfilt_xircom() and dc_txeof(). o Use bus_addr_t for DMA segment instead of using u_int32_t. o Introduce dc_dma_alloc()/dc_dma_free() functions to allocate/free DMA'able memory. o Create two DMA descriptor list for each TX/RX lists. This change will minimize the size of bounce buffers that would be used in each TX/RX path. Previously driver had to copy both TX/RX lists when bounce buffer is active. o 21143 data sheet says descriptor list requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Setup frame requires 4 bytes alignment. Remove PAGE_SIZE alignment restriction and use sizeof(struct dc_dec). o Add missing DMA map unload for both setup frame and TX/RX descriptor list. o Overhaul RX handling logic such that make driver always allocate new RX buffer with dc_newbuf(). Previously driver allowed to copy received frame with m_devget(9) after passing the descriptor ownership to controller. This can lead to passing wrong frame to upper stack. o Introduce dc_discard_rxbuf() which will discard received frame and reuse loaded DMA map and RX mbuf. o Correct several wrong bus_dmamap_sync(9) usage in dc_rxeof and dc_txeof. The TX/RX descriptor lists are updated by both driver and HW so READ/WRITE semantics should be used. o If driver failed to allocate new RX buffer, update if_iqdrops counter instead of if_ierrors since driver received the frame without errors. o Make sure to unload loaded setup frame DMA map in dc_txeof and clear the mark of setup frame of the TX descriptor in dc_txeof(). o Add check for possible TX descriptor overruns in dc_encap() and move check for free buffer to caller, dc_start_locked(). o Swap the loaded DMA map and the last DMA map for multi-segmented frames. Since dc_txeof() assumes the last descriptor of the frame has the DMA map, driver should swap the first and the last DMA map in dc_encap(). Previously driver tried to unload not-yet-loaded DMA map such that the loaded DMA map was not unloaded at all for multi-segmented frames. o Rewrite DC_RXDESC/DC_TXDESC macro to simpler one. o Remove definition of ETHER_ALIGN, it's already defined in ethernet.h. With this changes, dc(4) works with bounce buffers and it shall also fix issues which might have shown in PAE environments. Tested by: marius
2011-02-19 02:47:10 +00:00
bzero(ld->dc_tx_list, DC_TX_LIST_SZ);
bus_dmamap_sync(sc->dc_tx_ltag, sc->dc_tx_lmap,
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
/*
* Device suspend routine. Stop the interface and save some PCI
* settings in case the BIOS doesn't restore them properly on
* resume.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_suspend(device_t dev)
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
sc = device_get_softc(dev);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
dc_stop(sc);
sc->suspended = 1;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
return (0);
}
/*
* Device resume routine. Restore some PCI settings in case the BIOS
* doesn't, re-enable busmastering, and restart the interface if
* appropriate.
*/
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
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_resume(device_t dev)
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
struct ifnet *ifp;
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
sc = device_get_softc(dev);
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->dc_ifp;
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
/* reinitialize interface if necessary */
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Make consistent; turn spaces into tabs where there is a mixture.
2002-09-04 18:14:17 +00:00
if (ifp->if_flags & IFF_UP)
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
dc_init_locked(sc);
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
sc->suspended = 0;
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Add suspend/resume code mostly merged from fxp/rl driver. This is temporary hack, better and generalized solution probably should be implemented at lower layer(MII or PCI?). Tested by: shoko.araki@soliton.co.jp MFC after: 1 week
2002-05-06 10:55:42 +00:00
return (0);
}
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
/*
* Stop all chip I/O so that the kernel's probe routines don't
* get confused by errant DMAs when rebooting.
*/
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
static int
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
dc_shutdown(device_t dev)
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
{
- Style(9) fixes, most notably : o Remove register keyword o ANSIfy prototypes o Remove "return;" at the end of void functions o Remove trailing spaces o Don't align local variables with tabs and reorder them o Don't use /* FOO */ at the end of a #ifdef FOO block if it's a small block - Other non-functional changes : o 6 -> ETHER_ADDR_LEN o Don't initialize if_output; ether_ifattach() does it for us
2003-06-01 19:26:34 +00:00
struct dc_softc *sc;
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
sc = device_get_softc(dev);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_LOCK(sc);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
dc_stop(sc);
Fixup locking and mark MPSAFE: - Add locked versions of start and init. The SRM_MEDIA code in dc_init() stayed in dc_init() instead of moving to dc_init_locked() to make the locking saner. - Use callout_init_mtx(). - Fixup locking in detach and ioctl. - Lock the driver in the ifmedia callouts. - Don't recurse on the driver lock. - De-spl. MFC after: 3 days
2005-08-18 19:10:07 +00:00
DC_UNLOCK(sc);
Fix function prototype for device_shutdown method.
2007-11-22 02:45:00 +00:00
return (0);
Add the if_dc driver and remove all of the al, ax, dm, pn and mx drivers which it replaces. The new driver supports all of the chips supported by the ones it replaces, as well as many DEC/Intel 21143 10/100 cards. This also completes my quest to convert things to miibus and add Alpha support.
1999-12-04 17:41:31 +00:00
}
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
static int
dc_check_multiport(struct dc_softc *sc)
{
struct dc_softc *dsc;
devclass_t dc;
device_t child;
uint8_t *eaddr;
int unit;
dc = devclass_find("dc");
for (unit = 0; unit < devclass_get_maxunit(dc); unit++) {
child = devclass_get_device(dc, unit);
if (child == NULL)
continue;
if (child == sc->dc_dev)
continue;
if (device_get_parent(child) != device_get_parent(sc->dc_dev))
continue;
if (unit > device_get_unit(sc->dc_dev))
continue;
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
if (device_is_attached(child) == 0)
continue;
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
dsc = device_get_softc(child);
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
device_printf(sc->dc_dev,
"Using station address of %s as base\n",
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
device_get_nameunit(child));
bcopy(dsc->dc_eaddr, sc->dc_eaddr, ETHER_ADDR_LEN);
eaddr = (uint8_t *)sc->dc_eaddr;
eaddr[5]++;
When driver have to use base softc due to lack of SROM on second port, copy SROM information from base softc as well and run SROM parser again. This change is necessary for some dual port controllers to make dc(4) correctly detect PHY media based on first port configuration table. While I'm here add a check for validity of the base softc before duplicating SROM contents from base softc. If driver failed to attach to the first port it can access invalid area. PR: kern/79262 Reviewed by: marius
2011-02-18 02:14:53 +00:00
/* Prepare SROM to parse again. */
if (DC_IS_INTEL(sc) && dsc->dc_srom != NULL &&
sc->dc_romwidth != 0) {
free(sc->dc_srom, M_DEVBUF);
sc->dc_romwidth = dsc->dc_romwidth;
sc->dc_srom = malloc(DC_ROM_SIZE(sc->dc_romwidth),
M_DEVBUF, M_NOWAIT);
if (sc->dc_srom == NULL) {
device_printf(sc->dc_dev,
"Could not allocate SROM buffer\n");
return (ENOMEM);
}
bcopy(dsc->dc_srom, sc->dc_srom,
DC_ROM_SIZE(sc->dc_romwidth));
}
It seems some multi-port dc(4) controllers shares SROM of the first port such that reading station address from second port always returned 0xFF:0xFF:0xFF:0xFF:0xFF:0xFF Unfortunately it seems there is no easy way to know whether SROM is shared or not. Workaround the issue by traversing dc(4) device list and see whether we're using second port and use station address of controller 0 as base station address of second port. PR: kern/79262 MFC after: 2 weeks
2010-10-14 17:22:38 +00:00
return (0);
}
return (ENOENT);
}
Reference in New Issue Copy Permalink