d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
cafee33353
freebsd-dev/sys/pci/if_xl.c

3083 lines
78 KiB
C
Raw Normal View History

Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
* Copyright (c) 1997, 1998, 1999
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* Bill Paul <wpaul@ctr.columbia.edu>. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Bill Paul.
* 4. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*
$Id$ -> $FreeBSD$
1999-08-28 01:08:13 +00:00
* $FreeBSD$
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
*/
/*
* 3Com 3c90x Etherlink XL PCI NIC driver
*
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
* Supports the 3Com "boomerang", "cyclone" and "hurricane" PCI
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* bus-master chips (3c90x cards and embedded controllers) including
* the following:
*
* 3Com 3c900-TPO 10Mbps/RJ-45
* 3Com 3c900-COMBO 10Mbps/RJ-45,AUI,BNC
* 3Com 3c905-TX 10/100Mbps/RJ-45
* 3Com 3c905-T4 10/100Mbps/RJ-45
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
* 3Com 3c900B-TPO 10Mbps/RJ-45
* 3Com 3c900B-COMBO 10Mbps/RJ-45,AUI,BNC
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
* 3Com 3c900B-TPC 10Mbps/RJ-45,BNC
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
* 3Com 3c900B-FL 10Mbps/Fiber-optic
Some small updates: - Wait longer for the reset to complete in xl_attach() to try and avoid 'command never completed' warnings. - Clean up a few odds and ends in xl_attach(). - Add PCI ID for the 3c905B-COMBO (a new card). Right now this is treated as a 3c905B; I need to dig up one of these cards for testing before I can make the AUI and BNC ports work. - Add a hack to force reading the I/O address directly from the PCI registers if pci_map_port() fails. I SHOULD NOT HAVE TO DO THIS: SOMEBODY WITH MORE PCI CLUES THAN I SHOULD INVESTIGATE WHY THIS HAPPENS.
1999-04-11 17:44:44 +00:00
* 3Com 3c905B-COMBO 10/100Mbps/RJ-45,AUI,BNC
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* 3Com 3c905B-TX 10/100Mbps/RJ-45
* 3Com 3c905B-FL/FX 10/100Mbps/Fiber-optic
3Com has produced their own Linux driver for the 3c90x/3c90xB series cards. It's GPL'ed of course, but looking over it tonight I learned of Yet Another Fast EtherLink XL Adapter: the 3c980C server adapter. This is basically an updated version of the 3c980 that uses the Tornado ASIC instead of the earlier Hurricane ASIC. The only change here is to add the new PCI device ID (0x9805) and corresponding table entries.
1999-09-15 07:19:34 +00:00
* 3Com 3c905C-TX 10/100Mbps/RJ-45 (Tornado ASIC)
* 3Com 3c980-TX 10/100Mbps server adapter (Hurricane ASIC)
* 3Com 3c980C-TX 10/100Mbps server adapter (Tornado ASIC)
* 3Com 3cSOHO100-TX 10/100Mbps/RJ-45 (Hurricane ASIC)
Update the xl driver to recognize yet another 3c905B/3c905C class NIC: the 3c450-TX HomeConnect. Like the 3cSOHO100-TX OfficeConnect, this NIC uses the same ASIC as the 3c905B/3c905C but is targeted for a particular market segment (home users). It is somewhat less expensive than the 3c905B/3c905C ($49, according to the 3Com web site), comes with its own custom driver kit and is bundled with various goofy Windows software packages designed to demonstrate the niftyness of home networking (networked game demos, etc...). Changes are: - Add PCI ID to list in if_xlreg.h. - Update xl_devs table in if_xl.c. - Update xl_choose_xcvr() to consider the HomeConnect the same as all the other 10baseT/100baseTX cards.
1999-12-16 18:33:57 +00:00
* 3Com 3c450-TX 10/100Mbps/RJ-45 (Tornado ASIC)
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
* 3Com 3c556 10/100Mbps/RJ-45 (MiniPCI, Hurricane ASIC)
* 3Com 3c556B 10/100Mbps/RJ-45 (MiniPCI, Hurricane ASIC)
add support for 3Com 3c575TX Fast Etherlink XL. Device information for 3C575-TX is from NetBSD, sys/dev/cardbus/if_ex_cardbus.c file. Reviewed by: wpaul, imp
2000-11-02 06:49:38 +00:00
* 3Com 3c575TX 10/100Mbps/RJ-45 (Cardbus, Hurricane ASIC)
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
* 3Com 3c575B 10/100Mbps/RJ-45 (Cardbus, Hurricane ASIC)
* 3Com 3c575C 10/100Mbps/RJ-45 (Cardbus, Hurricane ASIC)
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
* 3Com 3cxfem656 10/100Mbps/RJ-45 (Cardbus, Hurricane ASIC)
* 3Com 3cxfem656b 10/100Mbps/RJ-45 (Cardbus, Hurricane ASIC)
* 3Com 3cxfem656c 10/100Mbps/RJ-45 (Cardbus, Tornado ASIC)
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
* Dell Optiplex GX1 on-board 3c918 10/100Mbps/RJ-45
3Com has produced their own Linux driver for the 3c90x/3c90xB series cards. It's GPL'ed of course, but looking over it tonight I learned of Yet Another Fast EtherLink XL Adapter: the 3c980C server adapter. This is basically an updated version of the 3c980 that uses the Tornado ASIC instead of the earlier Hurricane ASIC. The only change here is to add the new PCI device ID (0x9805) and corresponding table entries.
1999-09-15 07:19:34 +00:00
* Dell on-board 3c920 10/100Mbps/RJ-45
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* Dell Precision on-board 3c905B 10/100Mbps/RJ-45
* Dell Latitude laptop docking station embedded 3c905-TX
*
* Written by Bill Paul <wpaul@ctr.columbia.edu>
* Electrical Engineering Department
* Columbia University, New York City
*/
/*
* The 3c90x series chips use a bus-master DMA interface for transfering
* packets to and from the controller chip. Some of the "vortex" cards
* (3c59x) also supported a bus master mode, however for those chips
* you could only DMA packets to/from a contiguous memory buffer. For
* transmission this would mean copying the contents of the queued mbuf
* chain into a an mbuf cluster and then DMAing the cluster. This extra
* copy would sort of defeat the purpose of the bus master support for
* any packet that doesn't fit into a single mbuf.
*
* By contrast, the 3c90x cards support a fragment-based bus master
* mode where mbuf chains can be encapsulated using TX descriptors.
* This is similar to other PCI chips such as the Texas Instruments
* ThunderLAN and the Intel 82557/82558.
*
* The "vortex" driver (if_vx.c) happens to work for the "boomerang"
* bus master chips because they maintain the old PIO interface for
* backwards compatibility, but starting with the 3c905B and the
* "cyclone" chips, the compatibility interface has been dropped.
* Since using bus master DMA is a big win, we use this driver to
* support the PCI "boomerang" chips even though they work with the
* "vortex" driver in order to obtain better performance.
*
* This driver is in the /sys/pci directory because it only supports
* PCI-based NICs.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sockio.h>
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#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>
#include <net/bpf.h>
#include <vm/vm.h> /* for vtophys */
#include <vm/pmap.h> /* for vtophys */
Make the xl and pn drivers work on FreeBSD/alpha and add them to sys/alpha/conf/GENERIC. Note: the PNIC ignores the lower few bits of the RX buffer DMA address, which means we have to add yet another kludge to make it happy. Since we can't offset the packet data, we copy the first few bytes of the received data into a separate mbuf with proper alignment. This puts the IP header where it needs to be to prevent unaligned accesses. Also modified the PNIC driver to use a non-interrupt driven TX strategy. This improves performance somewhat on x86/SMP systems where interrupt delivery doesn't seem to be as fast with an SMP kernel as with a UP kernel.
1999-03-27 20:41:25 +00:00
#include <machine/bus_memio.h>
#include <machine/bus_pio.h>
#include <machine/bus.h>
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
#include <machine/resource.h>
#include <sys/bus.h>
#include <sys/rman.h>
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
#include <dev/mii/mii.h>
#include <dev/mii/miivar.h>
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
#include <pci/pcireg.h>
#include <pci/pcivar.h>
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(xl, miibus, 1, 1, 1);
Add a pointer to "controller miibus0" for people who will not read the commit messages or GENERIC and insist on running -CURRENT. It probably won't work, but it's worth a try.
1999-09-08 15:01:58 +00:00
/* "controller miibus0" required. See GENERIC if you get errors here. */
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
#include "miibus_if.h"
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* The following #define causes the code to use PIO to access the
* chip's registers instead of memory mapped mode. The reason PIO mode
* is on by default is that the Etherlink XL manual seems to indicate
* that only the newer revision chips (3c905B) support both PIO and
* memory mapped access. Since we want to be compatible with the older
* bus master chips, we use PIO here. If you comment this out, the
* driver will use memory mapped I/O, which may be faster but which
* might not work on some devices.
*/
#define XL_USEIOSPACE
#include <pci/if_xlreg.h>
The "easy" fixes for compiling the kernel -Wunused: remove unreferenced static and local variables, goto labels, and functions declared but not defined.
1998-12-07 21:58:50 +00:00
#if !defined(lint)
static const char rcsid[] =
$Id$ -> $FreeBSD$
1999-08-28 01:08:13 +00:00
"$FreeBSD$";
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
#endif
Implement TCP/IP checksum off-loading on send for the 3c905B and later generation cards.
2001-10-22 06:45:42 +00:00
#define XL905B_CSUM_FEATURES (CSUM_IP | CSUM_TCP | CSUM_UDP)
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Various supported device vendors/types and their names.
*/
static struct xl_type xl_devs[] = {
{ TC_VENDORID, TC_DEVICEID_BOOMERANG_10BT,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c900-TPO Etherlink XL" },
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{ TC_VENDORID, TC_DEVICEID_BOOMERANG_10BT_COMBO,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c900-COMBO Etherlink XL" },
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{ TC_VENDORID, TC_DEVICEID_BOOMERANG_10_100BT,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c905-TX Fast Etherlink XL" },
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{ TC_VENDORID, TC_DEVICEID_BOOMERANG_100BT4,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c905-T4 Fast Etherlink XL" },
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
{ TC_VENDORID, TC_DEVICEID_KRAKATOA_10BT,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c900B-TPO Etherlink XL" },
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
{ TC_VENDORID, TC_DEVICEID_KRAKATOA_10BT_COMBO,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c900B-COMBO Etherlink XL" },
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
{ TC_VENDORID, TC_DEVICEID_KRAKATOA_10BT_TPC,
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
"3Com 3c900B-TPC Etherlink XL" },
{ TC_VENDORID, TC_DEVICEID_CYCLONE_10FL,
"3Com 3c900B-FL Etherlink XL" },
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_10_100BT,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c905B-TX Fast Etherlink XL" },
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{ TC_VENDORID, TC_DEVICEID_CYCLONE_10_100BT4,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c905B-T4 Fast Etherlink XL" },
Add support for yet another "cyclone" board, with PCI device id 0x905A. This is a 100BaseFX board with SC fiber media connectors. I don't actually have one of these but I've been told it works with the xl driver. Submitted by: Jason Wright from the openbsd group
1998-10-09 03:59:24 +00:00
{ TC_VENDORID, TC_DEVICEID_CYCLONE_10_100FX,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c905B-FX/SC Fast Etherlink XL" },
Some small updates: - Wait longer for the reset to complete in xl_attach() to try and avoid 'command never completed' warnings. - Clean up a few odds and ends in xl_attach(). - Add PCI ID for the 3c905B-COMBO (a new card). Right now this is treated as a 3c905B; I need to dig up one of these cards for testing before I can make the AUI and BNC ports work. - Add a hack to force reading the I/O address directly from the PCI registers if pci_map_port() fails. I SHOULD NOT HAVE TO DO THIS: SOMEBODY WITH MORE PCI CLUES THAN I SHOULD INVESTIGATE WHY THIS HAPPENS.
1999-04-11 17:44:44 +00:00
{ TC_VENDORID, TC_DEVICEID_CYCLONE_10_100_COMBO,
"3Com 3c905B-COMBO Fast Etherlink XL" },
Head for the hills friends and neighbors, 3Com has yet another 3c90X chipset. First you thrilled to the 3c905, then you trembled at the 3c905B, now gaze in wonder at: the 3c905C! This appears to be another 3c90X series chip called the Tornado (PCI ID 0x10B7/0x9200) and should be equivalent (from the driver API perspective) to the 3c905B, so all we have to do is add the PCI ID to the list.
1999-05-30 18:09:17 +00:00
{ TC_VENDORID, TC_DEVICEID_TORNADO_10_100BT,
"3Com 3c905C-TX Fast Etherlink XL" },
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_10_100BT_SERV,
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
"3Com 3c980 Fast Etherlink XL" },
Dangit: mispelled TORNADO in one place.
1999-09-15 07:20:59 +00:00
{ TC_VENDORID, TC_DEVICEID_TORNADO_10_100BT_SERV,
3Com has produced their own Linux driver for the 3c90x/3c90xB series cards. It's GPL'ed of course, but looking over it tonight I learned of Yet Another Fast EtherLink XL Adapter: the 3c980C server adapter. This is basically an updated version of the 3c980 that uses the Tornado ASIC instead of the earlier Hurricane ASIC. The only change here is to add the new PCI device ID (0x9805) and corresponding table entries.
1999-09-15 07:19:34 +00:00
"3Com 3c980C Fast Etherlink XL" },
Add support for the 3cSOHO100-TX, which is a "hurricane" chipset (cut-down version of the "cyclone" for the small office/home office "cheap bastard" market). Basically the same as a 3c905B but without Wake-on-LAN, ROM socket, etc...
1999-04-12 20:38:45 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_SOHO100TX,
"3Com 3cSOHO100-TX OfficeConnect" },
Update the xl driver to recognize yet another 3c905B/3c905C class NIC: the 3c450-TX HomeConnect. Like the 3cSOHO100-TX OfficeConnect, this NIC uses the same ASIC as the 3c905B/3c905C but is targeted for a particular market segment (home users). It is somewhat less expensive than the 3c905B/3c905C ($49, according to the 3Com web site), comes with its own custom driver kit and is bundled with various goofy Windows software packages designed to demonstrate the niftyness of home networking (networked game demos, etc...). Changes are: - Add PCI ID to list in if_xlreg.h. - Update xl_devs table in if_xl.c. - Update xl_choose_xcvr() to consider the HomeConnect the same as all the other 10baseT/100baseTX cards.
1999-12-16 18:33:57 +00:00
{ TC_VENDORID, TC_DEVICEID_TORNADO_HOMECONNECT,
"3Com 3c450-TX HomeConnect" },
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_556,
"3Com 3c556 Fast Etherlink XL" },
{ TC_VENDORID, TC_DEVICEID_HURRICANE_556B,
"3Com 3c556B Fast Etherlink XL" },
add support for 3Com 3c575TX Fast Etherlink XL. Device information for 3C575-TX is from NetBSD, sys/dev/cardbus/if_ex_cardbus.c file. Reviewed by: wpaul, imp
2000-11-02 06:49:38 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_575A,
"3Com 3c575TX Fast Etherlink XL" },
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_575B,
"3Com 3c575B Fast Etherlink XL" },
{ TC_VENDORID, TC_DEVICEID_HURRICANE_575C,
"3Com 3c575C Fast Etherlink XL" },
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_656,
"3Com 3c656 Fast Etherlink XL" },
{ TC_VENDORID, TC_DEVICEID_HURRICANE_656B,
"3Com 3c656B Fast Etherlink XL" },
{ TC_VENDORID, TC_DEVICEID_TORNADO_656C,
Add device ID for the 3c565C card. I followed exactly the 3c575c, but further tweaks may be necessary down the road. This does nothing with the serial side of the card.
2000-12-01 19:41:14 +00:00
"3Com 3c656C Fast Etherlink XL" },
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{ 0, 0, NULL }
};
Remove __P.
2002-03-20 02:08:01 +00:00
static int xl_probe (device_t);
static int xl_attach (device_t);
static int xl_detach (device_t);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Remove __P.
2002-03-20 02:08:01 +00:00
static int xl_newbuf (struct xl_softc *, struct xl_chain_onefrag *);
static void xl_stats_update (void *);
static int xl_encap (struct xl_softc *, struct xl_chain *,
struct mbuf *);
static int xl_encap_90xB (struct xl_softc *, struct xl_chain *,
struct mbuf *);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Remove __P.
2002-03-20 02:08:01 +00:00
static void xl_rxeof (struct xl_softc *);
static int xl_rx_resync (struct xl_softc *);
static void xl_txeof (struct xl_softc *);
static void xl_txeof_90xB (struct xl_softc *);
static void xl_txeoc (struct xl_softc *);
static void xl_intr (void *);
static void xl_start (struct ifnet *);
static void xl_start_90xB (struct ifnet *);
static int xl_ioctl (struct ifnet *, u_long, caddr_t);
static void xl_init (void *);
static void xl_stop (struct xl_softc *);
static void xl_watchdog (struct ifnet *);
static void xl_shutdown (device_t);
static int xl_suspend (device_t);
static int xl_resume (device_t);
Add suspend/resume hooks to this driver; necessary to overcome problems on HP Omnibook 500. MFC after: 1 week
2001-12-05 10:34:07 +00:00
Remove __P.
2002-03-20 02:08:01 +00:00
static int xl_ifmedia_upd (struct ifnet *);
static void xl_ifmedia_sts (struct ifnet *, struct ifmediareq *);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Remove __P.
2002-03-20 02:08:01 +00:00
static int xl_eeprom_wait (struct xl_softc *);
static int xl_read_eeprom (struct xl_softc *, caddr_t, int, int, int);
static void xl_mii_sync (struct xl_softc *);
static void xl_mii_send (struct xl_softc *, u_int32_t, int);
static int xl_mii_readreg (struct xl_softc *, struct xl_mii_frame *);
static int xl_mii_writereg (struct xl_softc *, struct xl_mii_frame *);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Remove __P.
2002-03-20 02:08:01 +00:00
static void xl_setcfg (struct xl_softc *);
static void xl_setmode (struct xl_softc *, int);
static u_int8_t xl_calchash (caddr_t);
static void xl_setmulti (struct xl_softc *);
static void xl_setmulti_hash (struct xl_softc *);
static void xl_reset (struct xl_softc *);
static int xl_list_rx_init (struct xl_softc *);
static int xl_list_tx_init (struct xl_softc *);
static int xl_list_tx_init_90xB (struct xl_softc *);
static void xl_wait (struct xl_softc *);
static void xl_mediacheck (struct xl_softc *);
static void xl_choose_xcvr (struct xl_softc *, int);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
#ifdef notdef
Remove __P.
2002-03-20 02:08:01 +00:00
static void xl_testpacket (struct xl_softc *);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
#endif
Remove __P.
2002-03-20 02:08:01 +00:00
static int xl_miibus_readreg (device_t, int, int);
static int xl_miibus_writereg (device_t, int, int, int);
static void xl_miibus_statchg (device_t);
static void xl_miibus_mediainit (device_t);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
#ifdef XL_USEIOSPACE
#define XL_RES SYS_RES_IOPORT
#define XL_RID XL_PCI_LOIO
#else
#define XL_RES SYS_RES_MEMORY
#define XL_RID XL_PCI_LOMEM
#endif
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
static device_method_t xl_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, xl_probe),
DEVMETHOD(device_attach, xl_attach),
DEVMETHOD(device_detach, xl_detach),
DEVMETHOD(device_shutdown, xl_shutdown),
Add suspend/resume hooks to this driver; necessary to overcome problems on HP Omnibook 500. MFC after: 1 week
2001-12-05 10:34:07 +00:00
DEVMETHOD(device_suspend, xl_suspend),
DEVMETHOD(device_resume, xl_resume),
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
/* bus interface */
DEVMETHOD(bus_print_child, bus_generic_print_child),
DEVMETHOD(bus_driver_added, bus_generic_driver_added),
/* MII interface */
DEVMETHOD(miibus_readreg, xl_miibus_readreg),
DEVMETHOD(miibus_writereg, xl_miibus_writereg),
DEVMETHOD(miibus_statchg, xl_miibus_statchg),
DEVMETHOD(miibus_mediainit, xl_miibus_mediainit),
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
{ 0, 0 }
};
static driver_t xl_driver = {
Grrr. Okay, changing the devnames was a bad idea. Put them back the way they were.
1999-09-20 08:47:11 +00:00
"xl",
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
xl_methods,
sizeof(struct xl_softc)
};
static devclass_t xl_devclass;
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
DRIVER_MODULE(if_xl, cardbus, xl_driver, xl_devclass, 0, 0);
Tweak these for what I hope is the last time: change the DRIVER_MODULE() declaration for the interface driver from "foo" to "if_foo" but leave the declaration for the miibus attached to the interface driver alone. This lets the internal module name be "if_foo" while still allowing the miibus instances to attach to "foo." This should allow ifconfig to autoload driver modules again without breaking the miibus attach.
1999-09-22 06:08:11 +00:00
DRIVER_MODULE(if_xl, pci, xl_driver, xl_devclass, 0, 0);
Un-do the changes to the DRIVER_MODULE() declarations in these drivers. This whole idea isn't going to work until somebody makes the bus/kld code smarter. The idea here is to change the module's internal name from "foo" to "if_foo" so that ifconfig can tell a network driver from a non-network one. However doing this doesn't work correctly no matter how you slice it. For everything to work, you have to change the name in both the driver_t struct and the DRIVER_MODULE() declaration. The problems are: - If you change the name in both places, then the kernel thinks that the device's name is now "if_foo", so you get things like: if_foo0: <FOO ethernet> irq foo at device foo on pcifoo if_foo0: Ethernet address: foo:foo:foo:foo:foo:foo This is bogus. Now the device name doesn't agree with the logical interface name. There's no reason for this, and it violates the principle of least astonishment. - If you leave the name in the driver_t struct as "foo" and only change the names in the DRIVER_MODULE() declaration to "if_foo" then attaching drivers to child devices doesn't work because the names don't agree. This breaks miibus: drivers that need to have miibuses and PHY drivers attached never get them. In other words: damned if you do, damned if you don't. This needs to be thought through some more. Since the drivers that use miibus are broken, I have to change these all back in order to make them work again. Yes this will stop ifconfig from being able to demand load driver modules. On the whole, I'd rather have that than having the drivers not work at all.
1999-09-20 19:06:45 +00:00
DRIVER_MODULE(miibus, xl, miibus_driver, miibus_devclass, 0, 0);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Murphy's law says that it's possible the chip can wedge and
* the 'command in progress' bit may never clear. Hence, we wait
* only a finite amount of time to avoid getting caught in an
* infinite loop. Normally this delay routine would be a macro,
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
* but it isn't called during normal operation so we can afford
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* to make it a function.
*/
static void xl_wait(sc)
struct xl_softc *sc;
{
register int i;
for (i = 0; i < XL_TIMEOUT; i++) {
if (!(CSR_READ_2(sc, XL_STATUS) & XL_STAT_CMDBUSY))
break;
}
Shorten device names so that the pci probe lines don't exceed 80 chars and wrap to a second line. Put 'command never completed' message inside #ifdef DIAGNOSTIC/#endif to stop people worrying about it (it's harmless).
1999-03-31 15:45:15 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (i == XL_TIMEOUT)
printf("xl%d: command never completed!\n", sc->xl_unit);
return;
}
/*
* MII access routines are provided for adapters with external
* PHYs (3c905-TX, 3c905-T4, 3c905B-T4) and those with built-in
* autoneg logic that's faked up to look like a PHY (3c905B-TX).
* Note: if you don't perform the MDIO operations just right,
* it's possible to end up with code that works correctly with
* some chips/CPUs/processor speeds/bus speeds/etc but not
* with others.
*/
#define MII_SET(x) \
CSR_WRITE_2(sc, XL_W4_PHY_MGMT, \
CSR_READ_2(sc, XL_W4_PHY_MGMT) | x)
#define MII_CLR(x) \
CSR_WRITE_2(sc, XL_W4_PHY_MGMT, \
CSR_READ_2(sc, XL_W4_PHY_MGMT) & ~x)
/*
* Sync the PHYs by setting data bit and strobing the clock 32 times.
*/
static void xl_mii_sync(sc)
struct xl_softc *sc;
{
register int i;
XL_SEL_WIN(4);
MII_SET(XL_MII_DIR|XL_MII_DATA);
for (i = 0; i < 32; i++) {
MII_SET(XL_MII_CLK);
DELAY(1);
MII_CLR(XL_MII_CLK);
DELAY(1);
}
return;
}
/*
* Clock a series of bits through the MII.
*/
static void xl_mii_send(sc, bits, cnt)
struct xl_softc *sc;
u_int32_t bits;
int cnt;
{
int i;
XL_SEL_WIN(4);
MII_CLR(XL_MII_CLK);
for (i = (0x1 << (cnt - 1)); i; i >>= 1) {
if (bits & i) {
MII_SET(XL_MII_DATA);
} else {
MII_CLR(XL_MII_DATA);
}
DELAY(1);
MII_CLR(XL_MII_CLK);
DELAY(1);
MII_SET(XL_MII_CLK);
}
}
/*
* Read an PHY register through the MII.
*/
static int xl_mii_readreg(sc, frame)
struct xl_softc *sc;
struct xl_mii_frame *frame;
{
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
int i, ack;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Set up frame for RX.
*/
frame->mii_stdelim = XL_MII_STARTDELIM;
frame->mii_opcode = XL_MII_READOP;
frame->mii_turnaround = 0;
frame->mii_data = 0;
/*
* Select register window 4.
*/
XL_SEL_WIN(4);
CSR_WRITE_2(sc, XL_W4_PHY_MGMT, 0);
/*
* Turn on data xmit.
*/
MII_SET(XL_MII_DIR);
xl_mii_sync(sc);
/*
* Send command/address info.
*/
xl_mii_send(sc, frame->mii_stdelim, 2);
xl_mii_send(sc, frame->mii_opcode, 2);
xl_mii_send(sc, frame->mii_phyaddr, 5);
xl_mii_send(sc, frame->mii_regaddr, 5);
/* Idle bit */
MII_CLR((XL_MII_CLK|XL_MII_DATA));
DELAY(1);
MII_SET(XL_MII_CLK);
DELAY(1);
/* Turn off xmit. */
MII_CLR(XL_MII_DIR);
/* Check for ack */
MII_CLR(XL_MII_CLK);
DELAY(1);
MII_SET(XL_MII_CLK);
DELAY(1);
ack = CSR_READ_2(sc, XL_W4_PHY_MGMT) & XL_MII_DATA;
/*
* Now try reading data bits. If the ack failed, we still
* need to clock through 16 cycles to keep the PHY(s) in sync.
*/
if (ack) {
for(i = 0; i < 16; i++) {
MII_CLR(XL_MII_CLK);
DELAY(1);
MII_SET(XL_MII_CLK);
DELAY(1);
}
goto fail;
}
for (i = 0x8000; i; i >>= 1) {
MII_CLR(XL_MII_CLK);
DELAY(1);
if (!ack) {
if (CSR_READ_2(sc, XL_W4_PHY_MGMT) & XL_MII_DATA)
frame->mii_data |= i;
DELAY(1);
}
MII_SET(XL_MII_CLK);
DELAY(1);
}
fail:
MII_CLR(XL_MII_CLK);
DELAY(1);
MII_SET(XL_MII_CLK);
DELAY(1);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (ack)
return(1);
return(0);
}
/*
* Write to a PHY register through the MII.
*/
static int xl_mii_writereg(sc, frame)
struct xl_softc *sc;
struct xl_mii_frame *frame;
{
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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Set up frame for TX.
*/
frame->mii_stdelim = XL_MII_STARTDELIM;
frame->mii_opcode = XL_MII_WRITEOP;
frame->mii_turnaround = XL_MII_TURNAROUND;
/*
* Select the window 4.
*/
XL_SEL_WIN(4);
/*
* Turn on data output.
*/
MII_SET(XL_MII_DIR);
xl_mii_sync(sc);
xl_mii_send(sc, frame->mii_stdelim, 2);
xl_mii_send(sc, frame->mii_opcode, 2);
xl_mii_send(sc, frame->mii_phyaddr, 5);
xl_mii_send(sc, frame->mii_regaddr, 5);
xl_mii_send(sc, frame->mii_turnaround, 2);
xl_mii_send(sc, frame->mii_data, 16);
/* Idle bit. */
MII_SET(XL_MII_CLK);
DELAY(1);
MII_CLR(XL_MII_CLK);
DELAY(1);
/*
* Turn off xmit.
*/
MII_CLR(XL_MII_DIR);
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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return(0);
}
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
static int xl_miibus_readreg(dev, phy, reg)
device_t dev;
int phy, reg;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct xl_softc *sc;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
struct xl_mii_frame frame;
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
sc = device_get_softc(dev);
Just when I thought it was safe. In the original 3c905-TX NICs, the external NatSemi PHY chip was programmed to respond to MII address 24. In the 3c905B ASICs, the transceiver is internal but it's still mapped to MII address 24. But *some* 3Com 3c905B ASIC revisions map the transceiver control registers to *all* MII addresses (0 through 31). The miibus code probes for PHYs at all MII addresses and because of this unusual behavior, it will attempt to map the same PHY registers several times over, which doesn't work. Naturally, the 3c905B NIC that I tested happened not to exhibit this behavior. The fix is to tweak xl_miibus_readreg() and xl_miibus_writereg() to only respond when attempting to read from MII address 24. This is safe to do since the 3Com documentation indicates that the PHY and/or internal transceiver will always be mapped to address 24, and there are no 3Com XL NICs with more than one PHY.
1999-09-01 03:16:21 +00:00
/*
* Pretend that PHYs are only available at MII address 24.
* This is to guard against problems with certain 3Com ASIC
* revisions that incorrectly map the internal transceiver
* control registers at all MII addresses. This can cause
* the miibus code to attach the same PHY several times over.
*/
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if ((!(sc->xl_flags & XL_FLAG_PHYOK)) && phy != 24)
Just when I thought it was safe. In the original 3c905-TX NICs, the external NatSemi PHY chip was programmed to respond to MII address 24. In the 3c905B ASICs, the transceiver is internal but it's still mapped to MII address 24. But *some* 3Com 3c905B ASIC revisions map the transceiver control registers to *all* MII addresses (0 through 31). The miibus code probes for PHYs at all MII addresses and because of this unusual behavior, it will attempt to map the same PHY registers several times over, which doesn't work. Naturally, the 3c905B NIC that I tested happened not to exhibit this behavior. The fix is to tweak xl_miibus_readreg() and xl_miibus_writereg() to only respond when attempting to read from MII address 24. This is safe to do since the 3Com documentation indicates that the PHY and/or internal transceiver will always be mapped to address 24, and there are no 3Com XL NICs with more than one PHY.
1999-09-01 03:16:21 +00:00
return(0);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
bzero((char *)&frame, sizeof(frame));
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
frame.mii_phyaddr = phy;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
frame.mii_regaddr = reg;
xl_mii_readreg(sc, &frame);
return(frame.mii_data);
}
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
static int xl_miibus_writereg(dev, phy, reg, data)
device_t dev;
int phy, reg, data;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct xl_softc *sc;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
struct xl_mii_frame frame;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
sc = device_get_softc(dev);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if ((!(sc->xl_flags & XL_FLAG_PHYOK)) && phy != 24)
return(0);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
bzero((char *)&frame, sizeof(frame));
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
frame.mii_phyaddr = phy;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
frame.mii_regaddr = reg;
frame.mii_data = data;
xl_mii_writereg(sc, &frame);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
return(0);
}
static void xl_miibus_statchg(dev)
device_t dev;
{
struct xl_softc *sc;
struct mii_data *mii;
sc = device_get_softc(dev);
mii = device_get_softc(sc->xl_miibus);
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
XL_LOCK(sc);
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
xl_setcfg(sc);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
/* Set ASIC's duplex mode to match the PHY. */
XL_SEL_WIN(3);
if ((mii->mii_media_active & IFM_GMASK) == IFM_FDX)
Grrr. Fix PR 27742 correctly this time. (At least I got -stable right.)
2001-06-01 00:34:50 +00:00
CSR_WRITE_1(sc, XL_W3_MAC_CTRL, XL_MACCTRL_DUPLEX);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
else
CSR_WRITE_1(sc, XL_W3_MAC_CTRL,
(CSR_READ_1(sc, XL_W3_MAC_CTRL) & ~XL_MACCTRL_DUPLEX));
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
XL_UNLOCK(sc);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
return;
}
/*
* Special support for the 3c905B-COMBO. This card has 10/100 support
* plus BNC and AUI ports. This means we will have both an miibus attached
* plus some non-MII media settings. In order to allow this, we have to
* add the extra media to the miibus's ifmedia struct, but we can't do
* that during xl_attach() because the miibus hasn't been attached yet.
* So instead, we wait until the miibus probe/attach is done, at which
* point we will get a callback telling is that it's safe to add our
* extra media.
*/
static void xl_miibus_mediainit(dev)
device_t dev;
{
struct xl_softc *sc;
struct mii_data *mii;
struct ifmedia *ifm;
sc = device_get_softc(dev);
mii = device_get_softc(sc->xl_miibus);
ifm = &mii->mii_media;
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
XL_LOCK(sc);
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_media & (XL_MEDIAOPT_AUI|XL_MEDIAOPT_10FL)) {
/*
* Check for a 10baseFL board in disguise.
*/
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
if (bootverbose)
printf("xl%d: found 10baseFL\n", sc->xl_unit);
ifmedia_add(ifm, IFM_ETHER|IFM_10_FL, 0, NULL);
ifmedia_add(ifm, IFM_ETHER|IFM_10_FL|IFM_HDX, 0, NULL);
if (sc->xl_caps & XL_CAPS_FULL_DUPLEX)
ifmedia_add(ifm,
IFM_ETHER|IFM_10_FL|IFM_FDX, 0, NULL);
} else {
if (bootverbose)
printf("xl%d: found AUI\n", sc->xl_unit);
ifmedia_add(ifm, IFM_ETHER|IFM_10_5, 0, NULL);
}
}
if (sc->xl_media & XL_MEDIAOPT_BNC) {
if (bootverbose)
printf("xl%d: found BNC\n", sc->xl_unit);
ifmedia_add(ifm, IFM_ETHER|IFM_10_2, 0, NULL);
}
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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/*
* The EEPROM is slow: give it time to come ready after issuing
* it a command.
*/
static int xl_eeprom_wait(sc)
struct xl_softc *sc;
{
int i;
for (i = 0; i < 100; i++) {
if (CSR_READ_2(sc, XL_W0_EE_CMD) & XL_EE_BUSY)
DELAY(162);
else
break;
}
if (i == 100) {
printf("xl%d: eeprom failed to come ready\n", sc->xl_unit);
return(1);
}
return(0);
}
/*
* Read a sequence of words from the EEPROM. Note that ethernet address
* data is stored in the EEPROM in network byte order.
*/
static int xl_read_eeprom(sc, dest, off, cnt, swap)
struct xl_softc *sc;
caddr_t dest;
int off;
int cnt;
int swap;
{
int err = 0, i;
u_int16_t word = 0, *ptr;
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
#define EEPROM_5BIT_OFFSET(A) ((((A) << 2) & 0x7F00) | ((A) & 0x003F))
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
#define EEPROM_8BIT_OFFSET(A) ((A) & 0x003F)
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
/* WARNING! DANGER!
* It's easy to accidentally overwrite the rom content!
* Note: the 3c575 uses 8bit EEPROM offsets.
*/
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
XL_SEL_WIN(0);
if (xl_eeprom_wait(sc))
return(1);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_flags & XL_FLAG_EEPROM_OFFSET_30)
off += 0x30;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
for (i = 0; i < cnt; i++) {
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_flags & XL_FLAG_8BITROM)
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
CSR_WRITE_2(sc, XL_W0_EE_CMD,
XL_EE_8BIT_READ | EEPROM_8BIT_OFFSET(off + i));
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
else
CSR_WRITE_2(sc, XL_W0_EE_CMD,
XL_EE_READ | EEPROM_5BIT_OFFSET(off + i));
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
err = xl_eeprom_wait(sc);
if (err)
break;
word = CSR_READ_2(sc, XL_W0_EE_DATA);
ptr = (u_int16_t *)(dest + (i * 2));
if (swap)
*ptr = ntohs(word);
else
*ptr = word;
}
return(err ? 1 : 0);
}
/*
* This routine is taken from the 3Com Etherlink XL manual,
* page 10-7. It calculates a CRC of the supplied multicast
* group address and returns the lower 8 bits, which are used
* as the multicast filter position.
* Note: the 3c905B currently only supports a 64-bit hash table,
* which means we really only need 6 bits, but the manual indicates
* that future chip revisions will have a 256-bit hash table,
* hence the routine is set up to calculate 8 bits of position
* info in case we need it some day.
* Note II, The Sequel: _CURRENT_ versions of the 3c905B have a
* 256 bit hash table. This means we have to use all 8 bits regardless.
* On older cards, the upper 2 bits will be ignored. Grrrr....
*/
static u_int8_t xl_calchash(addr)
Silence gcc -Wall -ansi -pedantic and fix minor bug in xl_txeof(). Pointed out by: Eivind
1998-12-10 16:18:43 +00:00
caddr_t addr;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
u_int32_t crc, carry;
int i, j;
u_int8_t c;
/* Compute CRC for the address value. */
crc = 0xFFFFFFFF; /* initial value */
for (i = 0; i < 6; i++) {
c = *(addr + i);
for (j = 0; j < 8; j++) {
carry = ((crc & 0x80000000) ? 1 : 0) ^ (c & 0x01);
crc <<= 1;
c >>= 1;
if (carry)
crc = (crc ^ 0x04c11db6) | carry;
}
}
/* return the filter bit position */
return(crc & 0x000000FF);
}
/*
* NICs older than the 3c905B have only one multicast option, which
* is to enable reception of all multicast frames.
*/
static void xl_setmulti(sc)
struct xl_softc *sc;
{
struct ifnet *ifp;
struct ifmultiaddr *ifma;
u_int8_t rxfilt;
int mcnt = 0;
ifp = &sc->arpcom.ac_if;
XL_SEL_WIN(5);
rxfilt = CSR_READ_1(sc, XL_W5_RX_FILTER);
if (ifp->if_flags & IFF_ALLMULTI) {
rxfilt |= XL_RXFILTER_ALLMULTI;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
return;
}
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)
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
mcnt++;
if (mcnt)
rxfilt |= XL_RXFILTER_ALLMULTI;
else
rxfilt &= ~XL_RXFILTER_ALLMULTI;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
return;
}
/*
* 3c905B adapters have a hash filter that we can program.
*/
static void xl_setmulti_hash(sc)
struct xl_softc *sc;
{
struct ifnet *ifp;
int h = 0, i;
struct ifmultiaddr *ifma;
u_int8_t rxfilt;
int mcnt = 0;
ifp = &sc->arpcom.ac_if;
XL_SEL_WIN(5);
rxfilt = CSR_READ_1(sc, XL_W5_RX_FILTER);
if (ifp->if_flags & IFF_ALLMULTI) {
rxfilt |= XL_RXFILTER_ALLMULTI;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
return;
} else
rxfilt &= ~XL_RXFILTER_ALLMULTI;
/* first, zot all the existing hash bits */
for (i = 0; i < XL_HASHFILT_SIZE; i++)
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_HASH|i);
/* now program new ones */
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) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (ifma->ifma_addr->sa_family != AF_LINK)
continue;
h = xl_calchash(LLADDR((struct sockaddr_dl *)ifma->ifma_addr));
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_HASH|XL_HASH_SET|h);
mcnt++;
}
if (mcnt)
rxfilt |= XL_RXFILTER_MULTIHASH;
else
rxfilt &= ~XL_RXFILTER_MULTIHASH;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
return;
}
#ifdef notdef
static void xl_testpacket(sc)
struct xl_softc *sc;
{
struct mbuf *m;
struct ifnet *ifp;
ifp = &sc->arpcom.ac_if;
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
return;
bcopy(&sc->arpcom.ac_enaddr,
mtod(m, struct ether_header *)->ether_dhost, ETHER_ADDR_LEN);
bcopy(&sc->arpcom.ac_enaddr,
mtod(m, struct ether_header *)->ether_shost, ETHER_ADDR_LEN);
mtod(m, struct ether_header *)->ether_type = htons(3);
mtod(m, unsigned char *)[14] = 0;
mtod(m, unsigned char *)[15] = 0;
mtod(m, unsigned char *)[16] = 0xE3;
m->m_len = m->m_pkthdr.len = sizeof(struct ether_header) + 3;
IF_ENQUEUE(&ifp->if_snd, m);
xl_start(ifp);
return;
}
#endif
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
static void xl_setcfg(sc)
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
struct xl_softc *sc;
{
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
u_int32_t icfg;
XL_SEL_WIN(3);
icfg = CSR_READ_4(sc, XL_W3_INTERNAL_CFG);
icfg &= ~XL_ICFG_CONNECTOR_MASK;
if (sc->xl_media & XL_MEDIAOPT_MII ||
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
sc->xl_media & XL_MEDIAOPT_BT4)
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
icfg |= (XL_XCVR_MII << XL_ICFG_CONNECTOR_BITS);
if (sc->xl_media & XL_MEDIAOPT_BTX)
icfg |= (XL_XCVR_AUTO << XL_ICFG_CONNECTOR_BITS);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
CSR_WRITE_4(sc, XL_W3_INTERNAL_CFG, icfg);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_STOP);
return;
}
static void xl_setmode(sc, media)
struct xl_softc *sc;
int media;
{
u_int32_t icfg;
u_int16_t mediastat;
printf("xl%d: selecting ", sc->xl_unit);
XL_SEL_WIN(4);
mediastat = CSR_READ_2(sc, XL_W4_MEDIA_STATUS);
XL_SEL_WIN(3);
icfg = CSR_READ_4(sc, XL_W3_INTERNAL_CFG);
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_media & XL_MEDIAOPT_BT) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (IFM_SUBTYPE(media) == IFM_10_T) {
printf("10baseT transceiver, ");
sc->xl_xcvr = XL_XCVR_10BT;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_10BT << XL_ICFG_CONNECTOR_BITS);
mediastat |= XL_MEDIASTAT_LINKBEAT|
XL_MEDIASTAT_JABGUARD;
mediastat &= ~XL_MEDIASTAT_SQEENB;
}
}
if (sc->xl_media & XL_MEDIAOPT_BFX) {
if (IFM_SUBTYPE(media) == IFM_100_FX) {
printf("100baseFX port, ");
sc->xl_xcvr = XL_XCVR_100BFX;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_100BFX << XL_ICFG_CONNECTOR_BITS);
mediastat |= XL_MEDIASTAT_LINKBEAT;
mediastat &= ~XL_MEDIASTAT_SQEENB;
}
}
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_media & (XL_MEDIAOPT_AUI|XL_MEDIAOPT_10FL)) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (IFM_SUBTYPE(media) == IFM_10_5) {
printf("AUI port, ");
sc->xl_xcvr = XL_XCVR_AUI;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_AUI << XL_ICFG_CONNECTOR_BITS);
mediastat &= ~(XL_MEDIASTAT_LINKBEAT|
XL_MEDIASTAT_JABGUARD);
mediastat |= ~XL_MEDIASTAT_SQEENB;
}
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (IFM_SUBTYPE(media) == IFM_10_FL) {
printf("10baseFL transceiver, ");
sc->xl_xcvr = XL_XCVR_AUI;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_AUI << XL_ICFG_CONNECTOR_BITS);
mediastat &= ~(XL_MEDIASTAT_LINKBEAT|
XL_MEDIASTAT_JABGUARD);
mediastat |= ~XL_MEDIASTAT_SQEENB;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
if (sc->xl_media & XL_MEDIAOPT_BNC) {
if (IFM_SUBTYPE(media) == IFM_10_2) {
printf("BNC port, ");
sc->xl_xcvr = XL_XCVR_COAX;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_COAX << XL_ICFG_CONNECTOR_BITS);
mediastat &= ~(XL_MEDIASTAT_LINKBEAT|
XL_MEDIASTAT_JABGUARD|
XL_MEDIASTAT_SQEENB);
}
}
if ((media & IFM_GMASK) == IFM_FDX ||
IFM_SUBTYPE(media) == IFM_100_FX) {
printf("full duplex\n");
XL_SEL_WIN(3);
Grrr. Fix PR 27742 correctly this time. (At least I got -stable right.)
2001-06-01 00:34:50 +00:00
CSR_WRITE_1(sc, XL_W3_MAC_CTRL, XL_MACCTRL_DUPLEX);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
} else {
printf("half duplex\n");
XL_SEL_WIN(3);
CSR_WRITE_1(sc, XL_W3_MAC_CTRL,
(CSR_READ_1(sc, XL_W3_MAC_CTRL) & ~XL_MACCTRL_DUPLEX));
}
if (IFM_SUBTYPE(media) == IFM_10_2)
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_START);
else
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_STOP);
CSR_WRITE_4(sc, XL_W3_INTERNAL_CFG, icfg);
XL_SEL_WIN(4);
CSR_WRITE_2(sc, XL_W4_MEDIA_STATUS, mediastat);
DELAY(800);
XL_SEL_WIN(7);
return;
}
static void xl_reset(sc)
struct xl_softc *sc;
{
Some small updates: - Wait longer for the reset to complete in xl_attach() to try and avoid 'command never completed' warnings. - Clean up a few odds and ends in xl_attach(). - Add PCI ID for the 3c905B-COMBO (a new card). Right now this is treated as a 3c905B; I need to dig up one of these cards for testing before I can make the AUI and BNC ports work. - Add a hack to force reading the I/O address directly from the PCI registers if pci_map_port() fails. I SHOULD NOT HAVE TO DO THIS: SOMEBODY WITH MORE PCI CLUES THAN I SHOULD INVESTIGATE WHY THIS HAPPENS.
1999-04-11 17:44:44 +00:00
register int i;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
XL_SEL_WIN(0);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RESET |
When wierdreset flag is set, turn on the DISADVFD flag when we reset rather than all the flags. This prevents setting being read from ROM, which is a problem. If this breaks anything, it will only break the 3C556B cards minipci cards, which mainly exist at rpi as far as rpi has been able to tell. Submitted by: Louis Gerbarg <gerbal@rpi.edu>
2000-10-16 23:02:21 +00:00
((sc->xl_flags & XL_FLAG_WEIRDRESET) ?
XL_RESETOPT_DISADVFD:0));
Some small updates: - Wait longer for the reset to complete in xl_attach() to try and avoid 'command never completed' warnings. - Clean up a few odds and ends in xl_attach(). - Add PCI ID for the 3c905B-COMBO (a new card). Right now this is treated as a 3c905B; I need to dig up one of these cards for testing before I can make the AUI and BNC ports work. - Add a hack to force reading the I/O address directly from the PCI registers if pci_map_port() fails. I SHOULD NOT HAVE TO DO THIS: SOMEBODY WITH MORE PCI CLUES THAN I SHOULD INVESTIGATE WHY THIS HAPPENS.
1999-04-11 17:44:44 +00:00
for (i = 0; i < XL_TIMEOUT; i++) {
DELAY(10);
if (!(CSR_READ_2(sc, XL_STATUS) & XL_STAT_CMDBUSY))
break;
}
if (i == XL_TIMEOUT)
printf("xl%d: reset didn't complete\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Perform an RX reset and TX reset in xl_reset() along with the master reset command. I observed some anomalous behavior while testing a 3c905C with a Dell PowerEdge 4300/500 dual PIII 500Mhz system. The NIC would seem to work correctly most of the time but would sometimes fail to receive certain packets, in particular NFS create requests. I could mount an NFS filesystem from the PowerEdge and do an ls on it, but trying to do a "touch foo" would hang. Monitoring traffic from another host revealed that the client was properly sending an NFS create request but the server was not receiving it. It *did* receive it when I ran the same test with an Intel fxp card. I don't understand the exact mechanics of this strange behavior, but resetting the receiver and transmitter seems to get rid of it. I used to perform an RX and TX reset in xl_init(), but stopped doing it there because on 3c905B and later cards this causes the autoneg session to restart, which would lead to the NIC waiting a long time before exchanging traffic after being brought up the first time. Apparently the receiver and transmitter resets should be performed at least once when initializing the card. Hopefully this will cure problems that people have been having with the 3c905C -- this was the only strange behavior that I have observed with the 3c905C so far which does not appear with the 3c905B or 3c905.
1999-08-02 21:06:16 +00:00
/* Reset TX and RX. */
Pacify users who get all bent out of shape when they see the "xl%d: command never completed" message. The RX reset takes longer complete than it used to, a lot longer in fact than xl_wait() is prepared to wait. When we do the RX reset in xl_reset(), this cases xl_wait() to time out and whine. We wait a little extra time now after the RX reset, which should silence the warning. Thanks to obrien for finally getting me a box with a NIC that causes this problem for me to tinker with.
2001-07-27 20:55:45 +00:00
/* Note: the RX reset takes an absurd amount of time
* on newer versions of the Tornado chips such as those
* on the 3c905CX and newer 3c908C cards. We wait an
* extra amount of time so that xl_wait() doesn't complain
* and annoy the users.
*/
Perform an RX reset and TX reset in xl_reset() along with the master reset command. I observed some anomalous behavior while testing a 3c905C with a Dell PowerEdge 4300/500 dual PIII 500Mhz system. The NIC would seem to work correctly most of the time but would sometimes fail to receive certain packets, in particular NFS create requests. I could mount an NFS filesystem from the PowerEdge and do an ls on it, but trying to do a "touch foo" would hang. Monitoring traffic from another host revealed that the client was properly sending an NFS create request but the server was not receiving it. It *did* receive it when I ran the same test with an Intel fxp card. I don't understand the exact mechanics of this strange behavior, but resetting the receiver and transmitter seems to get rid of it. I used to perform an RX and TX reset in xl_init(), but stopped doing it there because on 3c905B and later cards this causes the autoneg session to restart, which would lead to the NIC waiting a long time before exchanging traffic after being brought up the first time. Apparently the receiver and transmitter resets should be performed at least once when initializing the card. Hopefully this will cure problems that people have been having with the 3c905C -- this was the only strange behavior that I have observed with the 3c905C so far which does not appear with the 3c905B or 3c905.
1999-08-02 21:06:16 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_RESET);
Pacify users who get all bent out of shape when they see the "xl%d: command never completed" message. The RX reset takes longer complete than it used to, a lot longer in fact than xl_wait() is prepared to wait. When we do the RX reset in xl_reset(), this cases xl_wait() to time out and whine. We wait a little extra time now after the RX reset, which should silence the warning. Thanks to obrien for finally getting me a box with a NIC that causes this problem for me to tinker with.
2001-07-27 20:55:45 +00:00
DELAY(100000);
Perform an RX reset and TX reset in xl_reset() along with the master reset command. I observed some anomalous behavior while testing a 3c905C with a Dell PowerEdge 4300/500 dual PIII 500Mhz system. The NIC would seem to work correctly most of the time but would sometimes fail to receive certain packets, in particular NFS create requests. I could mount an NFS filesystem from the PowerEdge and do an ls on it, but trying to do a "touch foo" would hang. Monitoring traffic from another host revealed that the client was properly sending an NFS create request but the server was not receiving it. It *did* receive it when I ran the same test with an Intel fxp card. I don't understand the exact mechanics of this strange behavior, but resetting the receiver and transmitter seems to get rid of it. I used to perform an RX and TX reset in xl_init(), but stopped doing it there because on 3c905B and later cards this causes the autoneg session to restart, which would lead to the NIC waiting a long time before exchanging traffic after being brought up the first time. Apparently the receiver and transmitter resets should be performed at least once when initializing the card. Hopefully this will cure problems that people have been having with the 3c905C -- this was the only strange behavior that I have observed with the 3c905C so far which does not appear with the 3c905B or 3c905.
1999-08-02 21:06:16 +00:00
xl_wait(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_RESET);
xl_wait(sc);
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
if (sc->xl_flags & XL_FLAG_INVERT_LED_PWR ||
sc->xl_flags & XL_FLAG_INVERT_MII_PWR) {
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
XL_SEL_WIN(2);
CSR_WRITE_2(sc, XL_W2_RESET_OPTIONS, CSR_READ_2(sc,
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
XL_W2_RESET_OPTIONS)
| ((sc->xl_flags & XL_FLAG_INVERT_LED_PWR)?XL_RESETOPT_INVERT_LED:0)
| ((sc->xl_flags & XL_FLAG_INVERT_MII_PWR)?XL_RESETOPT_INVERT_MII:0)
);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Wait a little while for the chip to get its brains in order. */
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
DELAY(100000);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/*
* Probe for a 3Com Etherlink XL chip. Check the PCI vendor and device
* IDs against our list and return a device name if we find a match.
*/
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
static int xl_probe(dev)
device_t dev;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
struct xl_type *t;
t = xl_devs;
while(t->xl_name != NULL) {
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
if ((pci_get_vendor(dev) == t->xl_vid) &&
(pci_get_device(dev) == t->xl_did)) {
device_set_desc(dev, t->xl_name);
return(0);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
t++;
}
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
return(ENXIO);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
/*
* This routine is a kludge to work around possible hardware faults
* or manufacturing defects that can cause the media options register
* (or reset options register, as it's called for the first generation
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
* 3c90x adapters) to return an incorrect result. I have encountered
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* one Dell Latitude laptop docking station with an integrated 3c905-TX
* which doesn't have any of the 'mediaopt' bits set. This screws up
* the attach routine pretty badly because it doesn't know what media
* to look for. If we find ourselves in this predicament, this routine
* will try to guess the media options values and warn the user of a
* possible manufacturing defect with his adapter/system/whatever.
*/
static void xl_mediacheck(sc)
struct xl_softc *sc;
{
/*
* If some of the media options bits are set, assume they are
* correct. If not, try to figure it out down below.
* XXX I should check for 10baseFL, but I don't have an adapter
* to test with.
*/
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_media & (XL_MEDIAOPT_MASK & ~XL_MEDIAOPT_VCO)) {
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
/*
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
* Check the XCVR value. If it's not in the normal range
* of values, we need to fake it up here.
*/
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
if (sc->xl_xcvr <= XL_XCVR_AUTO)
return;
else {
printf("xl%d: bogus xcvr value "
"in EEPROM (%x)\n", sc->xl_unit, sc->xl_xcvr);
printf("xl%d: choosing new default based "
"on card type\n", sc->xl_unit);
}
} else {
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media & XL_MEDIAOPT_10FL)
return;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
printf("xl%d: WARNING: no media options bits set in "
"the media options register!!\n", sc->xl_unit);
printf("xl%d: this could be a manufacturing defect in "
"your adapter or system\n", sc->xl_unit);
printf("xl%d: attempting to guess media type; you "
"should probably consult your vendor\n", sc->xl_unit);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
xl_choose_xcvr(sc, 1);
return;
}
static void xl_choose_xcvr(sc, verbose)
struct xl_softc *sc;
int verbose;
{
u_int16_t devid;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Read the device ID from the EEPROM.
* This is what's loaded into the PCI device ID register, so it has
* to be correct otherwise we wouldn't have gotten this far.
*/
xl_read_eeprom(sc, (caddr_t)&devid, XL_EE_PRODID, 1, 0);
switch(devid) {
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
case TC_DEVICEID_BOOMERANG_10BT: /* 3c900-TPO */
case TC_DEVICEID_KRAKATOA_10BT: /* 3c900B-TPO */
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_media = XL_MEDIAOPT_BT;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
sc->xl_xcvr = XL_XCVR_10BT;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing 10BaseT "
"transceiver\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
case TC_DEVICEID_BOOMERANG_10BT_COMBO: /* 3c900-COMBO */
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
case TC_DEVICEID_KRAKATOA_10BT_COMBO: /* 3c900B-COMBO */
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_media = XL_MEDIAOPT_BT|XL_MEDIAOPT_BNC|XL_MEDIAOPT_AUI;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
sc->xl_xcvr = XL_XCVR_10BT;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing COMBO "
"(AUI/BNC/TP)\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
case TC_DEVICEID_KRAKATOA_10BT_TPC: /* 3c900B-TPC */
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
sc->xl_media = XL_MEDIAOPT_BT|XL_MEDIAOPT_BNC;
sc->xl_xcvr = XL_XCVR_10BT;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing TPC (BNC/TP)\n", sc->xl_unit);
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
break;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
case TC_DEVICEID_CYCLONE_10FL: /* 3c900B-FL */
sc->xl_media = XL_MEDIAOPT_10FL;
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
sc->xl_xcvr = XL_XCVR_AUI;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing 10baseFL\n", sc->xl_unit);
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
break;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
case TC_DEVICEID_BOOMERANG_10_100BT: /* 3c905-TX */
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
case TC_DEVICEID_HURRICANE_556: /* 3c556 */
case TC_DEVICEID_HURRICANE_556B: /* 3c556B */
add support for 3Com 3c575TX Fast Etherlink XL. Device information for 3C575-TX is from NetBSD, sys/dev/cardbus/if_ex_cardbus.c file. Reviewed by: wpaul, imp
2000-11-02 06:49:38 +00:00
case TC_DEVICEID_HURRICANE_575A: /* 3c575TX */
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
case TC_DEVICEID_HURRICANE_575B: /* 3c575B */
case TC_DEVICEID_HURRICANE_575C: /* 3c575C */
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
case TC_DEVICEID_HURRICANE_656: /* 3c656 */
case TC_DEVICEID_HURRICANE_656B: /* 3c656B */
case TC_DEVICEID_TORNADO_656C: /* 3c656C */
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_media = XL_MEDIAOPT_MII;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
sc->xl_xcvr = XL_XCVR_MII;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing MII\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
case TC_DEVICEID_BOOMERANG_100BT4: /* 3c905-T4 */
case TC_DEVICEID_CYCLONE_10_100BT4: /* 3c905B-T4 */
sc->xl_media = XL_MEDIAOPT_BT4;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
sc->xl_xcvr = XL_XCVR_MII;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing 100BaseT4/MII\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
case TC_DEVICEID_HURRICANE_10_100BT: /* 3c905B-TX */
case TC_DEVICEID_HURRICANE_10_100BT_SERV:/*3c980-TX */
3Com has produced their own Linux driver for the 3c90x/3c90xB series cards. It's GPL'ed of course, but looking over it tonight I learned of Yet Another Fast EtherLink XL Adapter: the 3c980C server adapter. This is basically an updated version of the 3c980 that uses the Tornado ASIC instead of the earlier Hurricane ASIC. The only change here is to add the new PCI device ID (0x9805) and corresponding table entries.
1999-09-15 07:19:34 +00:00
case TC_DEVICEID_TORNADO_10_100BT_SERV: /* 3c980C-TX */
Add support for the 3cSOHO100-TX, which is a "hurricane" chipset (cut-down version of the "cyclone" for the small office/home office "cheap bastard" market). Basically the same as a 3c905B but without Wake-on-LAN, ROM socket, etc...
1999-04-12 20:38:45 +00:00
case TC_DEVICEID_HURRICANE_SOHO100TX: /* 3cSOHO100-TX */
Whoops, forgot to update xl_mediacheck() to account for the 3c905C.
1999-05-30 18:11:47 +00:00
case TC_DEVICEID_TORNADO_10_100BT: /* 3c905C-TX */
Update the xl driver to recognize yet another 3c905B/3c905C class NIC: the 3c450-TX HomeConnect. Like the 3cSOHO100-TX OfficeConnect, this NIC uses the same ASIC as the 3c905B/3c905C but is targeted for a particular market segment (home users). It is somewhat less expensive than the 3c905B/3c905C ($49, according to the 3Com web site), comes with its own custom driver kit and is bundled with various goofy Windows software packages designed to demonstrate the niftyness of home networking (networked game demos, etc...). Changes are: - Add PCI ID to list in if_xlreg.h. - Update xl_devs table in if_xl.c. - Update xl_choose_xcvr() to consider the HomeConnect the same as all the other 10baseT/100baseTX cards.
1999-12-16 18:33:57 +00:00
case TC_DEVICEID_TORNADO_HOMECONNECT: /* 3c450-TX */
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_media = XL_MEDIAOPT_BTX;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
sc->xl_xcvr = XL_XCVR_AUTO;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing 10/100 internal\n", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
case TC_DEVICEID_CYCLONE_10_100_COMBO: /* 3c905B-COMBO */
sc->xl_media = XL_MEDIAOPT_BTX|XL_MEDIAOPT_BNC|XL_MEDIAOPT_AUI;
sc->xl_xcvr = XL_XCVR_AUTO;
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
if (verbose)
printf("xl%d: guessing 10/100 "
"plus BNC/AUI\n", sc->xl_unit);
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
break;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
default:
printf("xl%d: unknown device ID: %x -- "
"defaulting to 10baseT\n", sc->xl_unit, devid);
sc->xl_media = XL_MEDIAOPT_BT;
break;
}
return;
}
/*
* Attach the interface. Allocate softc structures, do ifmedia
* setup and ethernet/BPF attach.
*/
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
static int xl_attach(dev)
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
device_t dev;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
Fixed printf format errors (none detected in GENERIC again).
1998-08-17 19:06:47 +00:00
u_char eaddr[ETHER_ADDR_LEN];
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
u_int32_t command;
struct xl_softc *sc;
struct ifnet *ifp;
int media = IFM_ETHER|IFM_100_TX|IFM_FDX;
Small tweak to newbus changes: return error status on failure correctly in xl_attach() (not a problem if the attach never fails, but if it does the function would still return 0, which is wrong).
1999-07-22 03:59:22 +00:00
int unit, error = 0, rid;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
sc = device_get_softc(dev);
unit = device_get_unit(dev);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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->xl_mtx, device_get_nameunit(dev), MTX_NETWORK_LOCK,
MTX_DEF | MTX_RECURSE);
Initialize/grab the mutex earlier in the attach phase, so that bailing out to the fail: label where we release/destroy the mutex will work without exploding.
2000-12-04 22:46:50 +00:00
XL_LOCK(sc);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
sc->xl_flags = 0;
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_556 ||
pci_get_device(dev) == TC_DEVICEID_HURRICANE_556B)
sc->xl_flags |= XL_FLAG_FUNCREG | XL_FLAG_PHYOK |
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
XL_FLAG_EEPROM_OFFSET_30 | XL_FLAG_WEIRDRESET |
XL_FLAG_INVERT_LED_PWR | XL_FLAG_INVERT_MII_PWR;
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_556)
sc->xl_flags |= XL_FLAG_8BITROM;
add support for 3Com 3c575TX Fast Etherlink XL. Device information for 3C575-TX is from NetBSD, sys/dev/cardbus/if_ex_cardbus.c file. Reviewed by: wpaul, imp
2000-11-02 06:49:38 +00:00
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_575A ||
pci_get_device(dev) == TC_DEVICEID_HURRICANE_575B ||
Add device ID for the 3c565C card. I followed exactly the 3c575c, but further tweaks may be necessary down the road. This does nothing with the serial side of the card.
2000-12-01 19:41:14 +00:00
pci_get_device(dev) == TC_DEVICEID_HURRICANE_575C ||
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
pci_get_device(dev) == TC_DEVICEID_HURRICANE_656B ||
pci_get_device(dev) == TC_DEVICEID_TORNADO_656C)
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
sc->xl_flags |= XL_FLAG_FUNCREG | XL_FLAG_PHYOK |
XL_FLAG_EEPROM_OFFSET_30 | XL_FLAG_8BITROM;
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_656)
sc->xl_flags |= XL_FLAG_FUNCREG | XL_FLAG_PHYOK;
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_575B)
sc->xl_flags |= XL_FLAG_INVERT_LED_PWR;
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_575C)
sc->xl_flags |= XL_FLAG_INVERT_MII_PWR;
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
if (pci_get_device(dev) == TC_DEVICEID_TORNADO_656C)
Add device ID for the 3c565C card. I followed exactly the 3c575c, but further tweaks may be necessary down the road. This does nothing with the serial side of the card.
2000-12-01 19:41:14 +00:00
sc->xl_flags |= XL_FLAG_INVERT_MII_PWR;
Add support for the 3c656B cardbus adapter. This is one half of a dual function card. It needs pretty much the same flags as the 656C, except that it seems to need both the INVERT_MII_PWR and INVERT_LED_PWR flags set. Tested with cardbus in -current as of today. Also added support for the 3c656, which looks to be the same as the 656B, except it doesn't need the EEPROM_8BIT flag. I think. This one is untested, but the added support should not break any of the other cards.
2001-08-28 00:40:18 +00:00
if (pci_get_device(dev) == TC_DEVICEID_HURRICANE_656 ||
pci_get_device(dev) == TC_DEVICEID_HURRICANE_656B)
sc->xl_flags |= XL_FLAG_INVERT_MII_PWR |
XL_FLAG_INVERT_LED_PWR;
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
Make two changes: If I'm reading the manual correctly, the 3c905B actually loses its PCI configuration during the transition from D3(hot) back to D0, not during the transition from D0 to D3(hot). This means it should be possible to save the existing PCI settings, restet the power state, then restore the PCI settings afterwards. Changed xl_attach() to attempt this first thing before the normal PCI setup. I'm not certain this will work correctly, but it shouldn't hurt. If xl_init() is called while an autoneg session is in progress, the autoneg timeout and chip state will get clobbered. Try to avoid this by checking sc->xl_autoneg at the start of xl_init() and defer the initialization until later if it's set. (xl_init() is always called at the end of an autoneg session by xl_autoneg_mii().) Problem pointed out by: Larry Baird <lab@gta.com>
1998-08-19 15:07:46 +00:00
/*
* If this is a 3c905B, we have to check one extra thing.
* The 905B supports power management and may be placed in
* a low-power mode (D3 mode), typically by certain operating
* systems which shall not be named. The PCI BIOS is supposed
* to reset the NIC and bring it out of low-power mode, but
* some do not. Consequently, we have to see if this chip
* supports power management, and if so, make sure it's not
* in low-power mode. If power management is available, the
* capid byte will be 0x01.
*
* I _think_ that what actually happens is that the chip
* loses its PCI configuration during the transition from
* D3 back to D0; this means that it should be possible for
* us to save the PCI iobase, membase and IRQ, put the chip
* back in the D0 state, then restore the PCI config ourselves.
*/
Use pci_get_powerstate()/pci_set_powerstate() which now exists in the PCI code. This saves each driver from having to grovel around looking for the right registers to twiddle. I should eventually convert the other PCI drivers to do this; for now, these three are ones which I know need power state handling.
2000-12-18 21:53:05 +00:00
if (pci_get_powerstate(dev) != PCI_POWERSTATE_D0) {
u_int32_t iobase, membase, irq;
/* Save important PCI config data. */
iobase = pci_read_config(dev, XL_PCI_LOIO, 4);
membase = pci_read_config(dev, XL_PCI_LOMEM, 4);
irq = pci_read_config(dev, XL_PCI_INTLINE, 4);
/* Reset the power state. */
printf("xl%d: chip is in D%d power mode "
"-- setting to D0\n", unit,
pci_get_powerstate(dev));
pci_set_powerstate(dev, PCI_POWERSTATE_D0);
/* Restore PCI config data. */
pci_write_config(dev, XL_PCI_LOIO, iobase, 4);
pci_write_config(dev, XL_PCI_LOMEM, membase, 4);
pci_write_config(dev, XL_PCI_INTLINE, irq, 4);
Make two changes: If I'm reading the manual correctly, the 3c905B actually loses its PCI configuration during the transition from D3(hot) back to D0, not during the transition from D0 to D3(hot). This means it should be possible to save the existing PCI settings, restet the power state, then restore the PCI settings afterwards. Changed xl_attach() to attempt this first thing before the normal PCI setup. I'm not certain this will work correctly, but it shouldn't hurt. If xl_init() is called while an autoneg session is in progress, the autoneg timeout and chip state will get clobbered. Try to avoid this by checking sc->xl_autoneg at the start of xl_init() and defer the initialization until later if it's set. (xl_init() is always called at the end of an autoneg session by xl_autoneg_mii().) Problem pointed out by: Larry Baird <lab@gta.com>
1998-08-19 15:07:46 +00:00
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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);
Apply patch supplied by Jonathan Chen: use the correct arguments to pci_enable_io(). We need to use SYS_RES_IOPORT/SYS_RES_MEMORY instead of PCIM_CMD_PORTEN/PCIM_CMD_MEMEN.
2001-07-09 17:58:42 +00:00
pci_enable_io(dev, SYS_RES_IOPORT);
pci_enable_io(dev, SYS_RES_MEMORY);
Use the correct register name. s/PCI_COMMAND_STATUS_REG/PCIR_COMMAND/
2000-05-28 16:13:43 +00:00
command = pci_read_config(dev, PCIR_COMMAND, 4);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
#ifdef XL_USEIOSPACE
if (!(command & PCIM_CMD_PORTEN)) {
printf("xl%d: failed to enable I/O ports!\n", unit);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
goto fail;
}
#else
if (!(command & PCIM_CMD_MEMEN)) {
printf("xl%d: failed to enable memory mapping!\n", unit);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
goto fail;
}
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
#endif
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
rid = XL_RID;
sc->xl_res = bus_alloc_resource(dev, XL_RES, &rid,
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
0, ~0, 1, RF_ACTIVE);
if (sc->xl_res == NULL) {
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
printf ("xl%d: couldn't map ports/memory\n", unit);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
goto fail;
}
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
sc->xl_btag = rman_get_bustag(sc->xl_res);
sc->xl_bhandle = rman_get_bushandle(sc->xl_res);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_flags & XL_FLAG_FUNCREG) {
rid = XL_PCI_FUNCMEM;
sc->xl_fres = bus_alloc_resource(dev, SYS_RES_MEMORY, &rid,
0, ~0, 1, RF_ACTIVE);
if (sc->xl_fres == NULL) {
printf ("xl%d: couldn't map ports/memory\n", unit);
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
error = ENXIO;
goto fail;
}
sc->xl_ftag = rman_get_bustag(sc->xl_fres);
sc->xl_fhandle = rman_get_bushandle(sc->xl_fres);
}
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
rid = 0;
sc->xl_irq = bus_alloc_resource(dev, SYS_RES_IRQ, &rid, 0, ~0, 1,
RF_SHAREABLE | RF_ACTIVE);
if (sc->xl_irq == NULL) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
printf("xl%d: couldn't map interrupt\n", unit);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_fres != NULL)
bus_release_resource(dev, SYS_RES_MEMORY,
XL_PCI_FUNCMEM, sc->xl_fres);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
goto fail;
}
error = bus_setup_intr(dev, sc->xl_irq, INTR_TYPE_NET,
xl_intr, sc, &sc->xl_intrhand);
if (error) {
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->xl_irq);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_fres != NULL)
bus_release_resource(dev, SYS_RES_MEMORY,
XL_PCI_FUNCMEM, sc->xl_fres);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
printf("xl%d: couldn't set up irq\n", unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
goto fail;
}
/* Reset the adapter. */
xl_reset(sc);
/*
* Get station address from the EEPROM.
*/
if (xl_read_eeprom(sc, (caddr_t)&eaddr, XL_EE_OEM_ADR0, 3, 1)) {
printf("xl%d: failed to read station address\n", sc->xl_unit);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_teardown_intr(dev, sc->xl_irq, sc->xl_intrhand);
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->xl_irq);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_fres != NULL)
bus_release_resource(dev, SYS_RES_MEMORY,
XL_PCI_FUNCMEM, sc->xl_fres);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
goto fail;
}
/*
* A 3Com chip was detected. Inform the world.
*/
printf("xl%d: Ethernet address: %6D\n", unit, eaddr, ":");
sc->xl_unit = unit;
callout_handle_init(&sc->xl_stat_ch);
bcopy(eaddr, (char *)&sc->arpcom.ac_enaddr, ETHER_ADDR_LEN);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
sc->xl_ldata = contigmalloc(sizeof(struct xl_list_data), M_DEVBUF,
Change contigmalloc() lower memory bound from 1MB to 0 to improve chances of allocations succeeding on systems with small amounts of RAM. Pointed out by: bde
1999-09-25 17:29:02 +00:00
M_NOWAIT, 0, 0xffffffff, PAGE_SIZE, 0);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
if (sc->xl_ldata == NULL) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
printf("xl%d: no memory for list buffers!\n", unit);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_teardown_intr(dev, sc->xl_irq, sc->xl_intrhand);
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->xl_irq);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_fres != NULL)
bus_release_resource(dev, SYS_RES_MEMORY,
XL_PCI_FUNCMEM, sc->xl_fres);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
error = ENXIO;
Some small updates: - Wait longer for the reset to complete in xl_attach() to try and avoid 'command never completed' warnings. - Clean up a few odds and ends in xl_attach(). - Add PCI ID for the 3c905B-COMBO (a new card). Right now this is treated as a 3c905B; I need to dig up one of these cards for testing before I can make the AUI and BNC ports work. - Add a hack to force reading the I/O address directly from the PCI registers if pci_map_port() fails. I SHOULD NOT HAVE TO DO THIS: SOMEBODY WITH MORE PCI CLUES THAN I SHOULD INVESTIGATE WHY THIS HAPPENS.
1999-04-11 17:44:44 +00:00
goto fail;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
bzero(sc->xl_ldata, sizeof(struct xl_list_data));
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
/*
* Figure out the card type. 3c905B adapters have the
* 'supportsNoTxLength' bit set in the capabilities
* word in the EEPROM.
*/
xl_read_eeprom(sc, (caddr_t)&sc->xl_caps, XL_EE_CAPS, 1, 0);
if (sc->xl_caps & XL_CAPS_NO_TXLENGTH)
sc->xl_type = XL_TYPE_905B;
else
sc->xl_type = XL_TYPE_90X;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp = &sc->arpcom.ac_if;
ifp->if_softc = sc;
ifp->if_unit = unit;
ifp->if_name = "xl";
ifp->if_mtu = ETHERMTU;
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
ifp->if_ioctl = xl_ioctl;
ifp->if_output = ether_output;
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
if (sc->xl_type == XL_TYPE_905B) {
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
ifp->if_start = xl_start_90xB;
Implement TCP/IP checksum off-loading on send for the 3c905B and later generation cards.
2001-10-22 06:45:42 +00:00
ifp->if_hwassist = XL905B_CSUM_FEATURES;
ifp->if_capabilities = IFCAP_HWCSUM;
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
} else
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
ifp->if_start = xl_start;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp->if_watchdog = xl_watchdog;
ifp->if_init = xl_init;
ifp->if_baudrate = 10000000;
Remember to initialize ifp->if_snd.ifq_maxlen.
1999-02-01 21:25:52 +00:00
ifp->if_snd.ifq_maxlen = XL_TX_LIST_CNT - 1;
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
ifp->if_capenable = ifp->if_capabilities;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Now we have to see what sort of media we have.
* This includes probing for an MII interace and a
* possible PHY.
*/
XL_SEL_WIN(3);
sc->xl_media = CSR_READ_2(sc, XL_W3_MEDIA_OPT);
if (bootverbose)
printf("xl%d: media options word: %x\n", sc->xl_unit,
sc->xl_media);
xl_read_eeprom(sc, (char *)&sc->xl_xcvr, XL_EE_ICFG_0, 2, 0);
sc->xl_xcvr &= XL_ICFG_CONNECTOR_MASK;
sc->xl_xcvr >>= XL_ICFG_CONNECTOR_BITS;
Modify the xl_mediacheck() routine to also test for the case where the XCVR value read from the EEPROM is completely wrong. I've had one report of a 3c900 card that returns an xcvr value of 14, which is impossible (the manual states that all vales above 8 are reserved). If the value is out of the expe Add PCI vendor ID for the 3c980-TX server adapter card, which apparently also uses the cyclone chip. Graciously supplied Mats O Jansson <maja@cntw.com>. Also noted by Mats, the 10mpbs cyclone adapters should be named 3c900B, not 3c905B. I haven't actually encountered a 10mbps only cyclone adapter yet, nor anybody who has one, but this makes sense given the naming scheme used for the older boomerang adapters.
1998-09-04 16:22:15 +00:00
xl_mediacheck(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (sc->xl_media & XL_MEDIAOPT_MII || sc->xl_media & XL_MEDIAOPT_BTX
|| sc->xl_media & XL_MEDIAOPT_BT4) {
if (bootverbose)
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
printf("xl%d: found MII/AUTO\n", sc->xl_unit);
xl_setcfg(sc);
if (mii_phy_probe(dev, &sc->xl_miibus,
xl_ifmedia_upd, xl_ifmedia_sts)) {
printf("xl%d: no PHY found!\n", sc->xl_unit);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_teardown_intr(dev, sc->xl_irq, sc->xl_intrhand);
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->xl_irq);
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
contigfree(sc->xl_ldata,
sizeof(struct xl_list_data), M_DEVBUF);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
error = ENXIO;
goto fail;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
goto done;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
Close PR #13665. I managed to figure out the problem, no thanks to the submitter, who *still* hasn't bothered to answer me back. The thing which the submitter completely failed to mention is that his 3c900B-TPO card has the transceiver selection in the EEPROM set to "auto." You can tweak the setting using the 3C90XCFG.EXE utility that 3Com provides with the card. I'm not sure if it's supposed to default to auto or if the user fiddled with it. Currently, the xl driver only does autoselection for 10/100 NICs (i.e. those with NWAY autonegotiation capabilities). For the 10baseT, 10base5, 10base2, 10baseFL and 100baseFX cards, the driver sets the default media to whatever the EEPROM transceiver selector says. The problem is that the "auto" selection is mistakenly identified as "10/100 NWAY autoselection mode" and this is not handled correctly: the default media ends up being chosen as 100baseTX, which doesn't work because we've only added 10baseT media types to the ifmedia word. This leads to a panic in ifmedia_set() (something else which the submitter never bothered to mention). A workaround for this is to re-run the 3C90XCFG.EXE utility and change the transceiver selection to something besides "auto." I have also patched the driver to watch for the "auto" setting in the non-miibus case and select a reasonable default based on the card type instead of falling through to 100baseTX and exploding. PR: misc/13665
1999-09-20 20:26:14 +00:00
/*
* Sanity check. If the user has selected "auto" and this isn't
* a 10/100 card of some kind, we need to force the transceiver
* type to something sane.
*/
if (sc->xl_xcvr == XL_XCVR_AUTO)
xl_choose_xcvr(sc, bootverbose);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Do ifmedia setup.
*/
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifmedia_init(&sc->ifmedia, 0, xl_ifmedia_upd, xl_ifmedia_sts);
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_media & XL_MEDIAOPT_BT) {
if (bootverbose)
printf("xl%d: found 10baseT\n", sc->xl_unit);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_T, 0, NULL);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_T|IFM_HDX, 0, NULL);
if (sc->xl_caps & XL_CAPS_FULL_DUPLEX)
ifmedia_add(&sc->ifmedia,
IFM_ETHER|IFM_10_T|IFM_FDX, 0, NULL);
}
if (sc->xl_media & (XL_MEDIAOPT_AUI|XL_MEDIAOPT_10FL)) {
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
/*
* Check for a 10baseFL board in disguise.
*/
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
if (bootverbose)
printf("xl%d: found 10baseFL\n", sc->xl_unit);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_FL, 0, NULL);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_FL|IFM_HDX,
0, NULL);
if (sc->xl_caps & XL_CAPS_FULL_DUPLEX)
ifmedia_add(&sc->ifmedia,
IFM_ETHER|IFM_10_FL|IFM_FDX, 0, NULL);
} else {
if (bootverbose)
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
printf("xl%d: found AUI\n", sc->xl_unit);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_5, 0, NULL);
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
if (sc->xl_media & XL_MEDIAOPT_BNC) {
if (bootverbose)
printf("xl%d: found BNC\n", sc->xl_unit);
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_10_2, 0, NULL);
}
if (sc->xl_media & XL_MEDIAOPT_BFX) {
if (bootverbose)
printf("xl%d: found 100baseFX\n", sc->xl_unit);
ifp->if_baudrate = 100000000;
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_100_FX, 0, NULL);
}
/* Choose a default media. */
switch(sc->xl_xcvr) {
case XL_XCVR_10BT:
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
media = IFM_ETHER|IFM_10_T;
xl_setmode(sc, media);
break;
case XL_XCVR_AUI:
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
media = IFM_ETHER|IFM_10_FL;
xl_setmode(sc, media);
} else {
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
media = IFM_ETHER|IFM_10_5;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
xl_setmode(sc, media);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
case XL_XCVR_COAX:
media = IFM_ETHER|IFM_10_2;
xl_setmode(sc, media);
break;
case XL_XCVR_AUTO:
case XL_XCVR_100BTX:
case XL_XCVR_MII:
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
/* Chosen by miibus */
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
case XL_XCVR_100BFX:
media = IFM_ETHER|IFM_100_FX;
break;
default:
printf("xl%d: unknown XCVR type: %d\n", sc->xl_unit,
sc->xl_xcvr);
- If the OACTIVE flag is set on entry to xl_start(), check to see if the transmitter is wedged. If so, try to unwedge it, process any descriptors that might need to be free()d, then proceed. - Disable the 'background' autonegotiation performed during bootstrap. What happens currently is that the driver starts an autoneg session, the sets a timeout in the ifnet structure and returns. Later, when the timer expires, the watchdog routine calls the autoneg handler to check the results of the session. The problem with this is that the session may not complete until some point after we have started to mount NFS filesystems, which can cause the mounts to fail. This is especially troublesome if booting with an NFS rootfs: we need the interface up and running before reaching the mountroot() code. The default behavior now is to do the autoneg synchronously, i.e. wait 5 seconds for the autoneg to complete before exiting the driver attach routine. People who want the old behavior can compile the driver with XL_BACKGROUND_AUTONEG #defined. This has no effect on autoneg sessions initiated by 'ifconfig xl0 media autoselect.' This slows the probe down a little, but it's either that or botching NFS mounts at bootup. - If xl_setmode_mii() is called and there's an autoneg session in progress, cancel it, _then_ set the modes.
1998-09-08 23:42:10 +00:00
/*
* This will probably be wrong, but it prevents
* the ifmedia code from panicking.
*/
media = IFM_ETHER|IFM_10_T;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
}
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus == NULL)
ifmedia_set(&sc->ifmedia, media);
done:
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
Make all Ethernet drivers attach using ether_ifattach() and detach using ether_ifdetach(). The former consolidates the operations of if_attach(), ng_ether_attach(), and bpfattach(). The latter consolidates the corresponding detach operations. Reviewed by: julian, freebsd-net
2000-07-13 22:54:34 +00:00
* Call MI attach routine.
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
*/
Make all Ethernet drivers attach using ether_ifattach() and detach using ether_ifdetach(). The former consolidates the operations of if_attach(), ng_ether_attach(), and bpfattach(). The latter consolidates the corresponding detach operations. Reviewed by: julian, freebsd-net
2000-07-13 22:54:34 +00:00
ether_ifattach(ifp, ETHER_BPF_SUPPORTED);
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
XL_UNLOCK(sc);
return(0);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
fail:
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
XL_UNLOCK(sc);
mtx_destroy(&sc->xl_mtx);
Small tweak to newbus changes: return error status on failure correctly in xl_attach() (not a problem if the attach never fails, but if it does the function would still return 0, which is wrong).
1999-07-22 03:59:22 +00:00
return(error);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
}
static int xl_detach(dev)
device_t dev;
{
struct xl_softc *sc;
struct ifnet *ifp;
sc = device_get_softc(dev);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
XL_LOCK(sc);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
ifp = &sc->arpcom.ac_if;
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
xl_reset(sc);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
xl_stop(sc);
Make all Ethernet drivers attach using ether_ifattach() and detach using ether_ifdetach(). The former consolidates the operations of if_attach(), ng_ether_attach(), and bpfattach(). The latter consolidates the corresponding detach operations. Reviewed by: julian, freebsd-net
2000-07-13 22:54:34 +00:00
ether_ifdetach(ifp, ETHER_BPF_SUPPORTED);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
/* Delete any miibus and phy devices attached to this interface */
if (sc->xl_miibus != NULL) {
bus_generic_detach(dev);
device_delete_child(dev, sc->xl_miibus);
}
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
bus_teardown_intr(dev, sc->xl_irq, sc->xl_intrhand);
bus_release_resource(dev, SYS_RES_IRQ, 0, sc->xl_irq);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_fres != NULL)
bus_release_resource(dev, SYS_RES_MEMORY,
XL_PCI_FUNCMEM, sc->xl_fres);
Some more small newbus cleanups. Remember to free all resources in case of failures in foo_attach(), simplify iospace/memspace things a little.
1999-07-23 02:06:57 +00:00
bus_release_resource(dev, XL_RES, XL_RID, sc->xl_res);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
ifmedia_removeall(&sc->ifmedia);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
contigfree(sc->xl_ldata, sizeof(struct xl_list_data), M_DEVBUF);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +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
XL_UNLOCK(sc);
mtx_destroy(&sc->xl_mtx);
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
return(0);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
/*
* Initialize the transmit descriptors.
*/
static int xl_list_tx_init(sc)
struct xl_softc *sc;
{
struct xl_chain_data *cd;
struct xl_list_data *ld;
int i;
cd = &sc->xl_cdata;
ld = sc->xl_ldata;
for (i = 0; i < XL_TX_LIST_CNT; i++) {
cd->xl_tx_chain[i].xl_ptr = &ld->xl_tx_list[i];
if (i == (XL_TX_LIST_CNT - 1))
cd->xl_tx_chain[i].xl_next = NULL;
else
cd->xl_tx_chain[i].xl_next = &cd->xl_tx_chain[i + 1];
}
cd->xl_tx_free = &cd->xl_tx_chain[0];
cd->xl_tx_tail = cd->xl_tx_head = NULL;
return(0);
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
/*
* Initialize the transmit descriptors.
*/
static int xl_list_tx_init_90xB(sc)
struct xl_softc *sc;
{
struct xl_chain_data *cd;
struct xl_list_data *ld;
int i;
cd = &sc->xl_cdata;
ld = sc->xl_ldata;
for (i = 0; i < XL_TX_LIST_CNT; i++) {
cd->xl_tx_chain[i].xl_ptr = &ld->xl_tx_list[i];
cd->xl_tx_chain[i].xl_phys = vtophys(&ld->xl_tx_list[i]);
if (i == (XL_TX_LIST_CNT - 1))
cd->xl_tx_chain[i].xl_next = &cd->xl_tx_chain[0];
else
cd->xl_tx_chain[i].xl_next = &cd->xl_tx_chain[i + 1];
if (i == 0)
cd->xl_tx_chain[i].xl_prev =
&cd->xl_tx_chain[XL_TX_LIST_CNT - 1];
else
cd->xl_tx_chain[i].xl_prev =
&cd->xl_tx_chain[i - 1];
}
bzero((char *)ld->xl_tx_list,
sizeof(struct xl_list) * XL_TX_LIST_CNT);
ld->xl_tx_list[0].xl_status = XL_TXSTAT_EMPTY;
cd->xl_tx_prod = 1;
cd->xl_tx_cons = 1;
cd->xl_tx_cnt = 0;
return(0);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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.
*/
static int xl_list_rx_init(sc)
struct xl_softc *sc;
{
struct xl_chain_data *cd;
struct xl_list_data *ld;
int i;
cd = &sc->xl_cdata;
ld = sc->xl_ldata;
for (i = 0; i < XL_RX_LIST_CNT; i++) {
cd->xl_rx_chain[i].xl_ptr =
(struct xl_list_onefrag *)&ld->xl_rx_list[i];
Fix handling of ENOBUFS condition. During reception, the filled mbuf cluster from the RX descriptor is passed up to the higher layers and replaced with an empty buffer for the next time the descriptor comes up in the RX ring. The xl_newbuf() routine returns ENOBUFS if it can't obtain a new mbuf cluster, but this return value was being ignored. Now, if buffer allocation fails, we leave the old one in place and drop the packet. This is rude, but there's not much else that can be done in this situation. Without this, the driver can cause a panic if the system runs out of MBUF clusters. Now it will complain loudly, but it shouldn't cause a panic. Also added another pair of missing newlines to some printf()s.
1998-08-31 15:26:09 +00:00
if (xl_newbuf(sc, &cd->xl_rx_chain[i]) == ENOBUFS)
return(ENOBUFS);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (i == (XL_RX_LIST_CNT - 1)) {
cd->xl_rx_chain[i].xl_next = &cd->xl_rx_chain[0];
ld->xl_rx_list[i].xl_next =
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
vtophys(&ld->xl_rx_list[0]);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
} else {
cd->xl_rx_chain[i].xl_next = &cd->xl_rx_chain[i + 1];
ld->xl_rx_list[i].xl_next =
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
vtophys(&ld->xl_rx_list[i + 1]);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
}
cd->xl_rx_head = &cd->xl_rx_chain[0];
return(0);
}
/*
* Initialize an RX descriptor and attach an MBUF cluster.
*/
static int xl_newbuf(sc, c)
struct xl_softc *sc;
struct xl_chain_onefrag *c;
{
struct mbuf *m_new = NULL;
MGETHDR(m_new, M_DONTWAIT, MT_DATA);
Remove printf's on mbuf/cluster allocation failures. There are now equivalent and less dangerous (rate limited) messages in the mbuf allocation code. MFC after: 3 days
2001-12-14 05:56:35 +00:00
if (m_new == NULL)
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return(ENOBUFS);
MCLGET(m_new, M_DONTWAIT);
if (!(m_new->m_flags & M_EXT)) {
m_freem(m_new);
return(ENOBUFS);
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
m_new->m_len = m_new->m_pkthdr.len = MCLBYTES;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
/* Force longword alignment for packet payload. */
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
m_adj(m_new, ETHER_ALIGN);
Make the xl and pn drivers work on FreeBSD/alpha and add them to sys/alpha/conf/GENERIC. Note: the PNIC ignores the lower few bits of the RX buffer DMA address, which means we have to add yet another kludge to make it happy. Since we can't offset the packet data, we copy the first few bytes of the received data into a separate mbuf with proper alignment. This puts the IP header where it needs to be to prevent unaligned accesses. Also modified the PNIC driver to use a non-interrupt driven TX strategy. This improves performance somewhat on x86/SMP systems where interrupt delivery doesn't seem to be as fast with an SMP kernel as with a UP kernel.
1999-03-27 20:41:25 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
c->xl_mbuf = m_new;
c->xl_ptr->xl_frag.xl_addr = vtophys(mtod(m_new, caddr_t));
c->xl_ptr->xl_frag.xl_len = MCLBYTES | XL_LAST_FRAG;
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
c->xl_ptr->xl_status = 0;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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
static int xl_rx_resync(sc)
struct xl_softc *sc;
{
struct xl_chain_onefrag *pos;
int i;
pos = sc->xl_cdata.xl_rx_head;
for (i = 0; i < XL_RX_LIST_CNT; i++) {
if (pos->xl_ptr->xl_status)
break;
pos = pos->xl_next;
}
if (i == XL_RX_LIST_CNT)
return(0);
sc->xl_cdata.xl_rx_head = pos;
return(EAGAIN);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* A frame has been uploaded: pass the resulting mbuf chain up to
* the higher level protocols.
*/
static void xl_rxeof(sc)
struct xl_softc *sc;
{
struct ether_header *eh;
struct mbuf *m;
struct ifnet *ifp;
struct xl_chain_onefrag *cur_rx;
int total_len = 0;
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
u_int32_t rxstat;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp = &sc->arpcom.ac_if;
again:
while((rxstat = sc->xl_cdata.xl_rx_head->xl_ptr->xl_status)) {
cur_rx = sc->xl_cdata.xl_rx_head;
Fix handling of ENOBUFS condition. During reception, the filled mbuf cluster from the RX descriptor is passed up to the higher layers and replaced with an empty buffer for the next time the descriptor comes up in the RX ring. The xl_newbuf() routine returns ENOBUFS if it can't obtain a new mbuf cluster, but this return value was being ignored. Now, if buffer allocation fails, we leave the old one in place and drop the packet. This is rude, but there's not much else that can be done in this situation. Without this, the driver can cause a panic if the system runs out of MBUF clusters. Now it will complain loudly, but it shouldn't cause a panic. Also added another pair of missing newlines to some printf()s.
1998-08-31 15:26:09 +00:00
sc->xl_cdata.xl_rx_head = cur_rx->xl_next;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* 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
* comes up in the ring.
*/
if (rxstat & XL_RXSTAT_UP_ERROR) {
ifp->if_ierrors++;
cur_rx->xl_ptr->xl_status = 0;
continue;
}
/*
* If there error bit was not set, the upload complete
* bit should be set which means we have a valid packet.
* If not, something truly strange has happened.
*/
if (!(rxstat & XL_RXSTAT_UP_CMPLT)) {
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
printf("xl%d: bad receive status -- "
"packet dropped", sc->xl_unit);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp->if_ierrors++;
cur_rx->xl_ptr->xl_status = 0;
continue;
}
/* No errors; receive the packet. */
m = cur_rx->xl_mbuf;
total_len = cur_rx->xl_ptr->xl_status & XL_RXSTAT_LENMASK;
Fix handling of ENOBUFS condition. During reception, the filled mbuf cluster from the RX descriptor is passed up to the higher layers and replaced with an empty buffer for the next time the descriptor comes up in the RX ring. The xl_newbuf() routine returns ENOBUFS if it can't obtain a new mbuf cluster, but this return value was being ignored. Now, if buffer allocation fails, we leave the old one in place and drop the packet. This is rude, but there's not much else that can be done in this situation. Without this, the driver can cause a panic if the system runs out of MBUF clusters. Now it will complain loudly, but it shouldn't cause a panic. Also added another pair of missing newlines to some printf()s.
1998-08-31 15:26:09 +00:00
/*
* Try to conjure up a new mbuf cluster. If that
* fails, it means we have an out of memory condition and
* should leave the buffer in place and continue. This will
* result in a lost packet, but there's little else we
* can do in this situation.
*/
if (xl_newbuf(sc, cur_rx) == ENOBUFS) {
ifp->if_ierrors++;
cur_rx->xl_ptr->xl_status = 0;
continue;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
ifp->if_ipackets++;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
eh = mtod(m, struct ether_header *);
m->m_pkthdr.rcvif = ifp;
Add bridging support (tested in 3.1-RELEASE by Steven Vetzal <svetzal@icom.ca>).
1999-04-16 01:56:06 +00:00
m->m_pkthdr.len = m->m_len = total_len;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Remove header from mbuf and pass it on. */
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
m_adj(m, sizeof(struct ether_header));
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
if (sc->xl_type == XL_TYPE_905B) {
/* Do IP checksum checking. */
if (rxstat & XL_RXSTAT_IPCKOK)
m->m_pkthdr.csum_flags |= CSUM_IP_CHECKED;
if (!(rxstat & XL_RXSTAT_IPCKERR))
m->m_pkthdr.csum_flags |= CSUM_IP_VALID;
if ((rxstat & XL_RXSTAT_TCPCOK &&
!(rxstat & XL_RXSTAT_TCPCKERR)) ||
(rxstat & XL_RXSTAT_UDPCKOK &&
!(rxstat & XL_RXSTAT_UDPCKERR))) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID|CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ether_input(ifp, eh, m);
}
/*
* Handle the 'end of channel' condition. When the upload
* engine hits the end of the RX ring, it will stall. This
* is our cue to flush the RX ring, reload the uplist pointer
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
* register and unstall the engine.
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
* XXX This is actually a little goofy. With the ThunderLAN
* chip, you get an interrupt when the receiver hits the end
* of the receive ring, which tells you exactly when you
* you need to reload the ring pointer. Here we have to
* fake it. I'm mad at myself for not being clever enough
* to avoid the use of a goto here.
*/
if (CSR_READ_4(sc, XL_UPLIST_PTR) == 0 ||
CSR_READ_4(sc, XL_UPLIST_STATUS) & XL_PKTSTAT_UP_STALLED) {
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_UP_STALL);
xl_wait(sc);
CSR_WRITE_4(sc, XL_UPLIST_PTR,
vtophys(&sc->xl_ldata->xl_rx_list[0]));
sc->xl_cdata.xl_rx_head = &sc->xl_cdata.xl_rx_chain[0];
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_UP_UNSTALL);
goto again;
}
return;
}
/*
* A frame was downloaded to the chip. It's safe for us to clean up
* the list buffers.
*/
static void xl_txeof(sc)
struct xl_softc *sc;
{
struct xl_chain *cur_tx;
struct ifnet *ifp;
ifp = &sc->arpcom.ac_if;
/* Clear the timeout timer. */
ifp->if_timer = 0;
/*
* Go through our tx list and free mbufs for those
* frames that have been uploaded. Note: the 3c905B
* sets a special bit in the status word to let us
* know that a frame has been downloaded, but the
* original 3c900/3c905 adapters don't do that.
* Consequently, we have to use a different test if
* xl_type != XL_TYPE_905B.
*/
while(sc->xl_cdata.xl_tx_head != NULL) {
cur_tx = sc->xl_cdata.xl_tx_head;
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
if (CSR_READ_4(sc, XL_DOWNLIST_PTR))
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_cdata.xl_tx_head = cur_tx->xl_next;
m_freem(cur_tx->xl_mbuf);
cur_tx->xl_mbuf = NULL;
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
ifp->if_opackets++;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
cur_tx->xl_next = sc->xl_cdata.xl_tx_free;
sc->xl_cdata.xl_tx_free = cur_tx;
}
if (sc->xl_cdata.xl_tx_head == NULL) {
ifp->if_flags &= ~IFF_OACTIVE;
sc->xl_cdata.xl_tx_tail = NULL;
} else {
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
if (CSR_READ_4(sc, XL_DMACTL) & XL_DMACTL_DOWN_STALLED ||
!CSR_READ_4(sc, XL_DOWNLIST_PTR)) {
Close PR #8384: Revert the transmission packet queueing strategy changes. Clearly I missed something while debugging this, although I never encountered any problems on my test machines. Also make one other minor change: jack up the TX reclaim threshold for 3c90xB adapters in order to stave off 'transmission error: 82' errors. Document the existence of the tx reclaim register (for inspecting the current reclaim threshold) in register window 5 (if_xlreg.h).
1998-10-22 15:52:25 +00:00
CSR_WRITE_4(sc, XL_DOWNLIST_PTR,
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
vtophys(sc->xl_cdata.xl_tx_head->xl_ptr));
Close PR #8384: Revert the transmission packet queueing strategy changes. Clearly I missed something while debugging this, although I never encountered any problems on my test machines. Also make one other minor change: jack up the TX reclaim threshold for 3c90xB adapters in order to stave off 'transmission error: 82' errors. Document the existence of the tx reclaim register (for inspecting the current reclaim threshold) in register window 5 (if_xlreg.h).
1998-10-22 15:52:25 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_UNSTALL);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
return;
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
static void xl_txeof_90xB(sc)
struct xl_softc *sc;
{
struct xl_chain *cur_tx = NULL;
struct ifnet *ifp;
int idx;
ifp = &sc->arpcom.ac_if;
idx = sc->xl_cdata.xl_tx_cons;
while(idx != sc->xl_cdata.xl_tx_prod) {
cur_tx = &sc->xl_cdata.xl_tx_chain[idx];
if (!(cur_tx->xl_ptr->xl_status & XL_TXSTAT_DL_COMPLETE))
break;
if (cur_tx->xl_mbuf != NULL) {
m_freem(cur_tx->xl_mbuf);
cur_tx->xl_mbuf = NULL;
}
ifp->if_opackets++;
sc->xl_cdata.xl_tx_cnt--;
XL_INC(idx, XL_TX_LIST_CNT);
ifp->if_timer = 0;
}
sc->xl_cdata.xl_tx_cons = idx;
if (cur_tx != NULL)
ifp->if_flags &= ~IFF_OACTIVE;
return;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* TX 'end of channel' interrupt handler. Actually, we should
* only get a 'TX complete' interrupt if there's a transmit error,
* so this is really TX error handler.
*/
static void xl_txeoc(sc)
struct xl_softc *sc;
{
u_int8_t txstat;
while((txstat = CSR_READ_1(sc, XL_TX_STATUS))) {
if (txstat & XL_TXSTATUS_UNDERRUN ||
txstat & XL_TXSTATUS_JABBER ||
txstat & XL_TXSTATUS_RECLAIM) {
printf("xl%d: transmission error: %x\n",
sc->xl_unit, txstat);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_RESET);
xl_wait(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
if (sc->xl_type == XL_TYPE_905B) {
if (sc->xl_cdata.xl_tx_cnt) {
int i;
struct xl_chain *c;
i = sc->xl_cdata.xl_tx_cons;
c = &sc->xl_cdata.xl_tx_chain[i];
CSR_WRITE_4(sc, XL_DOWNLIST_PTR,
c->xl_phys);
CSR_WRITE_1(sc, XL_DOWN_POLL, 64);
}
} else {
if (sc->xl_cdata.xl_tx_head != NULL)
CSR_WRITE_4(sc, XL_DOWNLIST_PTR,
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
vtophys(sc->xl_cdata.xl_tx_head->xl_ptr));
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Remember to set this for the
* first generation 3c90X chips.
*/
CSR_WRITE_1(sc, XL_TX_FREETHRESH, XL_PACKET_SIZE >> 8);
Dynamically increase TX start threshold if TX underruns are detected.
1999-05-05 17:05:07 +00:00
if (txstat & XL_TXSTATUS_UNDERRUN &&
sc->xl_tx_thresh < XL_PACKET_SIZE) {
sc->xl_tx_thresh += XL_MIN_FRAMELEN;
printf("xl%d: tx underrun, increasing tx start"
" threshold to %d bytes\n", sc->xl_unit,
sc->xl_tx_thresh);
}
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
CSR_WRITE_2(sc, XL_COMMAND,
Dynamically increase TX start threshold if TX underruns are detected.
1999-05-05 17:05:07 +00:00
XL_CMD_TX_SET_START|sc->xl_tx_thresh);
Close PR #8384: Revert the transmission packet queueing strategy changes. Clearly I missed something while debugging this, although I never encountered any problems on my test machines. Also make one other minor change: jack up the TX reclaim threshold for 3c90xB adapters in order to stave off 'transmission error: 82' errors. Document the existence of the tx reclaim register (for inspecting the current reclaim threshold) in register window 5 (if_xlreg.h).
1998-10-22 15:52:25 +00:00
if (sc->xl_type == XL_TYPE_905B) {
CSR_WRITE_2(sc, XL_COMMAND,
XL_CMD_SET_TX_RECLAIM|(XL_PACKET_SIZE >> 4));
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_ENABLE);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_UNSTALL);
} else {
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_ENABLE);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_UNSTALL);
}
/*
* Write an arbitrary byte to the TX_STATUS register
* to clear this interrupt/error and advance to the next.
*/
CSR_WRITE_1(sc, XL_TX_STATUS, 0x01);
}
return;
}
static void xl_intr(arg)
void *arg;
{
struct xl_softc *sc;
struct ifnet *ifp;
u_int16_t status;
sc = arg;
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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp = &sc->arpcom.ac_if;
Add support for cardbus card's chips. This will make the 3c575 cards work once the rest of the cardbus infrastructure has been committed. Submitted by: Jonathan Chen <jon@spook.org>
2000-10-16 23:16:02 +00:00
while((status = CSR_READ_2(sc, XL_STATUS)) & XL_INTRS && status != 0xFFFF) {
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
CSR_WRITE_2(sc, XL_COMMAND,
XL_CMD_INTR_ACK|(status & XL_INTRS));
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 & XL_STAT_UP_COMPLETE) {
int curpkts;
curpkts = ifp->if_ipackets;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
xl_rxeof(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
if (curpkts == ifp->if_ipackets) {
while (xl_rx_resync(sc))
xl_rxeof(sc);
}
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
if (status & XL_STAT_DOWN_COMPLETE) {
if (sc->xl_type == XL_TYPE_905B)
xl_txeof_90xB(sc);
else
xl_txeof(sc);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (status & XL_STAT_TX_COMPLETE) {
ifp->if_oerrors++;
xl_txeoc(sc);
}
if (status & XL_STAT_ADFAIL) {
xl_reset(sc);
xl_init(sc);
}
if (status & XL_STAT_STATSOFLOW) {
sc->xl_stats_no_timeout = 1;
xl_stats_update(sc);
sc->xl_stats_no_timeout = 0;
}
}
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
if (ifp->if_snd.ifq_head != NULL)
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
(*ifp->if_start)(ifp);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Fix one instance of XL_LOCK() that should have been XL_UNLOCK(). After doing this so many times, I guess I was entitled to at least one typo. Thanks to all who spotted this.
2000-10-15 21:15:38 +00:00
XL_UNLOCK(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
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
static void xl_stats_update(xsc)
void *xsc;
{
struct xl_softc *sc;
struct ifnet *ifp;
struct xl_stats xl_stats;
u_int8_t *p;
int i;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct mii_data *mii = NULL;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
bzero((char *)&xl_stats, sizeof(struct xl_stats));
sc = xsc;
ifp = &sc->arpcom.ac_if;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus != NULL)
mii = device_get_softc(sc->xl_miibus);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
p = (u_int8_t *)&xl_stats;
/* Read all the stats registers. */
XL_SEL_WIN(6);
for (i = 0; i < 16; i++)
*p++ = CSR_READ_1(sc, XL_W6_CARRIER_LOST + i);
ifp->if_ierrors += xl_stats.xl_rx_overrun;
ifp->if_collisions += xl_stats.xl_tx_multi_collision +
xl_stats.xl_tx_single_collision +
xl_stats.xl_tx_late_collision;
/*
* Boomerang and cyclone chips have an extra stats counter
* in window 4 (BadSSD). We have to read this too in order
* to clear out all the stats registers and avoid a statsoflow
* interrupt.
*/
XL_SEL_WIN(4);
CSR_READ_1(sc, XL_W4_BADSSD);
Fix a problem where stats overflow interrupts would cause a major slowdown, and re-enable stats overflow interrupts. For future reference, the bug was in our code, and not some bug in the 3com chips. Reviewed by: wpaul MFC after: 2 days
2001-12-17 22:24:19 +00:00
if ((mii != NULL) && (!sc->xl_stats_no_timeout))
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
mii_tick(mii);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
XL_SEL_WIN(7);
if (!sc->xl_stats_no_timeout)
sc->xl_stat_ch = timeout(xl_stats_update, sc, hz);
return;
}
/*
* Encapsulate an mbuf chain in a descriptor by coupling the mbuf data
* pointers to the fragment pointers.
*/
static int xl_encap(sc, c, m_head)
struct xl_softc *sc;
struct xl_chain *c;
struct mbuf *m_head;
{
int frag = 0;
struct xl_frag *f = NULL;
int total_len;
struct mbuf *m;
/*
* Start packing the mbufs in this chain into
* the fragment pointers. Stop when we run out
* of fragments or hit the end of the mbuf chain.
*/
m = m_head;
total_len = 0;
for (m = m_head, frag = 0; m != NULL; m = m->m_next) {
if (m->m_len != 0) {
if (frag == XL_MAXFRAGS)
break;
total_len+= m->m_len;
c->xl_ptr->xl_frag[frag].xl_addr =
vtophys(mtod(m, vm_offset_t));
c->xl_ptr->xl_frag[frag].xl_len = m->m_len;
frag++;
}
}
/*
* Handle special case: we used up all 63 fragments,
* but we have more mbufs left in the chain. Copy the
* data into an mbuf cluster. Note that we don't
* bother clearing the values in the other fragment
* pointers/counters; it wouldn't gain us anything,
* and would waste cycles.
*/
if (m != NULL) {
struct mbuf *m_new = NULL;
MGETHDR(m_new, M_DONTWAIT, MT_DATA);
if (m_new == NULL) {
printf("xl%d: no memory for tx list", sc->xl_unit);
return(1);
}
if (m_head->m_pkthdr.len > MHLEN) {
MCLGET(m_new, M_DONTWAIT);
if (!(m_new->m_flags & M_EXT)) {
m_freem(m_new);
printf("xl%d: no memory for tx list",
sc->xl_unit);
return(1);
}
}
m_copydata(m_head, 0, m_head->m_pkthdr.len,
mtod(m_new, caddr_t));
m_new->m_pkthdr.len = m_new->m_len = m_head->m_pkthdr.len;
m_freem(m_head);
m_head = m_new;
f = &c->xl_ptr->xl_frag[0];
f->xl_addr = vtophys(mtod(m_new, caddr_t));
f->xl_len = total_len = m_new->m_len;
frag = 1;
}
c->xl_mbuf = m_head;
c->xl_ptr->xl_frag[frag - 1].xl_len |= XL_LAST_FRAG;
c->xl_ptr->xl_status = total_len;
c->xl_ptr->xl_next = 0;
return(0);
}
/*
* 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.
*/
static void xl_start(ifp)
struct ifnet *ifp;
{
struct xl_softc *sc;
struct mbuf *m_head = NULL;
struct xl_chain *prev = NULL, *cur_tx = NULL, *start_tx;
sc = ifp->if_softc;
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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Check for an available queue slot. If there are none,
* punt.
*/
if (sc->xl_cdata.xl_tx_free == NULL) {
Apply patch graciously provided by Jason Wright <jason@thought.net> from the OpenBSD group to fix a problem with the default ifmedia not being set properly in some cases with a 3c905B, leading to a panic in ifmedia_set(). Also apply a patch to force the transmit start routine to check the transmitter to make sure it isn't wedged if the outbound tx queue appears full. This seems to cure some problems with 'watchdog timeout' errors cropping up in some cases. I tried to do this before by checking for the IFF_OACTIVE flag on entry to xl_start(), but if the IFF_OACTIVE flag is set, ether_output() won't even call xl_start(). It should work now. Lastly, increase the size of the TX queue from 10 descriptors to 16 to hopefully make it less likely that the TX queue will fill up.
1998-09-25 17:34:19 +00:00
xl_txeoc(sc);
xl_txeof(sc);
if (sc->xl_cdata.xl_tx_free == NULL) {
ifp->if_flags |= IFF_OACTIVE;
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
XL_UNLOCK(sc);
Apply patch graciously provided by Jason Wright <jason@thought.net> from the OpenBSD group to fix a problem with the default ifmedia not being set properly in some cases with a 3c905B, leading to a panic in ifmedia_set(). Also apply a patch to force the transmit start routine to check the transmitter to make sure it isn't wedged if the outbound tx queue appears full. This seems to cure some problems with 'watchdog timeout' errors cropping up in some cases. I tried to do this before by checking for the IFF_OACTIVE flag on entry to xl_start(), but if the IFF_OACTIVE flag is set, ether_output() won't even call xl_start(). It should work now. Lastly, increase the size of the TX queue from 10 descriptors to 16 to hopefully make it less likely that the TX queue will fill up.
1998-09-25 17:34:19 +00:00
return;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
start_tx = sc->xl_cdata.xl_tx_free;
while(sc->xl_cdata.xl_tx_free != NULL) {
IF_DEQUEUE(&ifp->if_snd, m_head);
if (m_head == NULL)
break;
/* Pick a descriptor off the free list. */
cur_tx = sc->xl_cdata.xl_tx_free;
sc->xl_cdata.xl_tx_free = cur_tx->xl_next;
cur_tx->xl_next = NULL;
/* Pack the data into the descriptor. */
xl_encap(sc, cur_tx, m_head);
/* Chain it together. */
if (prev != NULL) {
prev->xl_next = cur_tx;
prev->xl_ptr->xl_next = vtophys(cur_tx->xl_ptr);
}
prev = cur_tx;
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
if (ifp->if_bpf)
bpf_mtap(ifp, cur_tx->xl_mbuf);
}
Add sanity check to foo_start() routines: in the unlikely (though apparently possible) event that the transmit start routine is called with and empty if_snd queue, bail out instead of dereferencing unilitialized transmit list pointers and panicking.
1998-12-05 02:21:44 +00:00
/*
* If there are no packets queued, bail.
*/
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
if (cur_tx == NULL) {
XL_UNLOCK(sc);
Add sanity check to foo_start() routines: in the unlikely (though apparently possible) event that the transmit start routine is called with and empty if_snd queue, bail out instead of dereferencing unilitialized transmit list pointers and panicking.
1998-12-05 02:21:44 +00:00
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
}
Add sanity check to foo_start() routines: in the unlikely (though apparently possible) event that the transmit start routine is called with and empty if_snd queue, bail out instead of dereferencing unilitialized transmit list pointers and panicking.
1998-12-05 02:21:44 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Place the request for the upload interrupt
* in the last descriptor in the chain. This way, if
* we're chaining several packets at once, we'll only
* get an interupt once for the whole chain rather than
* once for each packet.
*/
cur_tx->xl_ptr->xl_status |= XL_TXSTAT_DL_INTR;
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
/*
* Queue the packets. If the TX channel is clear, update
* the downlist pointer register.
*/
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_STALL);
xl_wait(sc);
Grrr.... forgot one line from the previous fix.
1999-07-08 00:42:02 +00:00
if (sc->xl_cdata.xl_tx_head != NULL) {
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
sc->xl_cdata.xl_tx_tail->xl_next = start_tx;
sc->xl_cdata.xl_tx_tail->xl_ptr->xl_next =
vtophys(start_tx->xl_ptr);
sc->xl_cdata.xl_tx_tail->xl_ptr->xl_status &=
~XL_TXSTAT_DL_INTR;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc->xl_cdata.xl_tx_tail = cur_tx;
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
} else {
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
sc->xl_cdata.xl_tx_head = start_tx;
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
sc->xl_cdata.xl_tx_tail = cur_tx;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
Fix a potential race condition that can occur in xl_start(). If the NIC clears out the transmit queue and zeroes the downlist pointer register, but xl_txeof() isn't called before xl_start() tries to queue more packets, xl_start() will think that the DMA is still in progress and not update the downlist register again, thus causing packets to sit in the transmit queue forever. Patch provided by: Russell T Hunt <alaric@MIT.EDU>
1999-07-07 21:49:14 +00:00
if (!CSR_READ_4(sc, XL_DOWNLIST_PTR))
CSR_WRITE_4(sc, XL_DOWNLIST_PTR, vtophys(start_tx->xl_ptr));
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_UNSTALL);
XL_SEL_WIN(7);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Set a timeout in case the chip goes out to lunch.
*/
ifp->if_timer = 5;
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
/*
* XXX Under certain conditions, usually on slower machines
* where interrupts may be dropped, it's possible for the
* adapter to chew up all the buffers in the receive ring
* and stall, without us being able to do anything about it.
* To guard against this, we need to make a pass over the
* RX queue to make sure there aren't any packets pending.
* Doing it here means we can flush the receive ring at the
* same time the chip is DMAing the transmit descriptors we
* just gave it.
*
* 3Com goes to some lengths to emphasize the Parallel Tasking (tm)
* nature of their chips in all their marketing literature;
* we may as well take advantage of it. :)
*/
xl_rxeof(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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
static int xl_encap_90xB(sc, c, m_head)
struct xl_softc *sc;
struct xl_chain *c;
struct mbuf *m_head;
{
int frag = 0;
struct xl_frag *f = NULL;
struct mbuf *m;
struct xl_list *d;
/*
* Start packing the mbufs in this chain into
* the fragment pointers. Stop when we run out
* of fragments or hit the end of the mbuf chain.
*/
d = c->xl_ptr;
d->xl_status = 0;
d->xl_next = 0;
for (m = m_head, frag = 0; m != NULL; m = m->m_next) {
if (m->m_len != 0) {
if (frag == XL_MAXFRAGS)
break;
f = &d->xl_frag[frag];
f->xl_addr = vtophys(mtod(m, vm_offset_t));
f->xl_len = m->m_len;
frag++;
}
}
c->xl_mbuf = m_head;
c->xl_ptr->xl_frag[frag - 1].xl_len |= XL_LAST_FRAG;
c->xl_ptr->xl_status = XL_TXSTAT_RND_DEFEAT;
Implement TCP/IP checksum off-loading on send for the 3c905B and later generation cards.
2001-10-22 06:45:42 +00:00
if (m_head->m_pkthdr.csum_flags) {
if (m_head->m_pkthdr.csum_flags & CSUM_IP)
c->xl_ptr->xl_status |= XL_TXSTAT_IPCKSUM;
if (m_head->m_pkthdr.csum_flags & CSUM_TCP)
c->xl_ptr->xl_status |= XL_TXSTAT_TCPCKSUM;
if (m_head->m_pkthdr.csum_flags & CSUM_UDP)
c->xl_ptr->xl_status |= XL_TXSTAT_UDPCKSUM;
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
return(0);
}
static void xl_start_90xB(ifp)
struct ifnet *ifp;
{
struct xl_softc *sc;
struct mbuf *m_head = NULL;
struct xl_chain *prev = NULL, *cur_tx = NULL, *start_tx;
int idx;
sc = ifp->if_softc;
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
XL_LOCK(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +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
if (ifp->if_flags & IFF_OACTIVE) {
XL_UNLOCK(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
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
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
idx = sc->xl_cdata.xl_tx_prod;
start_tx = &sc->xl_cdata.xl_tx_chain[idx];
while (sc->xl_cdata.xl_tx_chain[idx].xl_mbuf == NULL) {
if ((XL_TX_LIST_CNT - sc->xl_cdata.xl_tx_cnt) < 3) {
ifp->if_flags |= IFF_OACTIVE;
break;
}
IF_DEQUEUE(&ifp->if_snd, m_head);
if (m_head == NULL)
break;
cur_tx = &sc->xl_cdata.xl_tx_chain[idx];
/* Pack the data into the descriptor. */
xl_encap_90xB(sc, cur_tx, m_head);
/* Chain it together. */
if (prev != NULL)
prev->xl_ptr->xl_next = cur_tx->xl_phys;
prev = cur_tx;
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
if (ifp->if_bpf)
bpf_mtap(ifp, cur_tx->xl_mbuf);
XL_INC(idx, XL_TX_LIST_CNT);
sc->xl_cdata.xl_tx_cnt++;
}
/*
* If there are no packets queued, bail.
*/
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
if (cur_tx == NULL) {
XL_UNLOCK(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
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
}
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
/*
* Place the request for the upload interrupt
* in the last descriptor in the chain. This way, if
* we're chaining several packets at once, we'll only
* get an interupt once for the whole chain rather than
* once for each packet.
*/
cur_tx->xl_ptr->xl_status |= XL_TXSTAT_DL_INTR;
/* Start transmission */
sc->xl_cdata.xl_tx_prod = idx;
start_tx->xl_prev->xl_ptr->xl_next = start_tx->xl_phys;
/*
* Set a timeout in case the chip goes out to lunch.
*/
ifp->if_timer = 5;
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
XL_UNLOCK(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
return;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
static void xl_init(xsc)
void *xsc;
{
struct xl_softc *sc = xsc;
struct ifnet *ifp = &sc->arpcom.ac_if;
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
int i;
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
u_int16_t rxfilt = 0;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct mii_data *mii = NULL;
Make two changes: If I'm reading the manual correctly, the 3c905B actually loses its PCI configuration during the transition from D3(hot) back to D0, not during the transition from D0 to D3(hot). This means it should be possible to save the existing PCI settings, restet the power state, then restore the PCI settings afterwards. Changed xl_attach() to attempt this first thing before the normal PCI setup. I'm not certain this will work correctly, but it shouldn't hurt. If xl_init() is called while an autoneg session is in progress, the autoneg timeout and chip state will get clobbered. Try to avoid this by checking sc->xl_autoneg at the start of xl_init() and defer the initialization until later if it's set. (xl_init() is always called at the end of an autoneg session by xl_autoneg_mii().) Problem pointed out by: Larry Baird <lab@gta.com>
1998-08-19 15:07:46 +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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Cancel pending I/O and free all RX/TX buffers.
*/
xl_stop(sc);
Close PR# 15986: issue an RX reset command when initializing the interface, but only for those cards that don't use miibus (i.e. all the 10mbps only cards, and the 100baseFX card). PR: kern/15986
2000-01-09 21:12:59 +00:00
if (sc->xl_miibus == NULL) {
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_RESET);
xl_wait(sc);
}
Small tweak: just reset the transmit block instead of doing a global reset in xl_init(). This achieves the effect that I wanted without totally resetting the chip.
1999-10-25 20:52:32 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_RESET);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
xl_wait(sc);
Small tweak: just reset the transmit block instead of doing a global reset in xl_init(). This achieves the effect that I wanted without totally resetting the chip.
1999-10-25 20:52:32 +00:00
DELAY(10000);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus != NULL)
mii = device_get_softc(sc->xl_miibus);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Init our MAC address */
XL_SEL_WIN(2);
for (i = 0; i < ETHER_ADDR_LEN; i++) {
CSR_WRITE_1(sc, XL_W2_STATION_ADDR_LO + i,
sc->arpcom.ac_enaddr[i]);
}
/* Clear the station mask. */
for (i = 0; i < 3; i++)
CSR_WRITE_2(sc, XL_W2_STATION_MASK_LO + (i * 2), 0);
Do not issue RX and TX reset commands in xl_stop() and xl_init(). On the 3c905B, the RX and TX reset commands also reset the cyclone chip's internal PHY, which causes it to restart its autonegotiation session. This takes a second or two to complete, which makes the interface seem to stop responding for a few seconds every time you do something that reinitializes it. Issuing the RX and TX resets on the older 3c905 boomerang adapters doesn't cause any delay because the boomerang chip requires an external PHY. This should fix the problem where people doing network installs via 3c905B cards experience a delay after the interface is first initialized, among other things.
1999-02-11 23:59:29 +00:00
#ifdef notdef
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Reset TX and RX. */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_RESET);
xl_wait(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_RESET);
xl_wait(sc);
Do not issue RX and TX reset commands in xl_stop() and xl_init(). On the 3c905B, the RX and TX reset commands also reset the cyclone chip's internal PHY, which causes it to restart its autonegotiation session. This takes a second or two to complete, which makes the interface seem to stop responding for a few seconds every time you do something that reinitializes it. Issuing the RX and TX resets on the older 3c905 boomerang adapters doesn't cause any delay because the boomerang chip requires an external PHY. This should fix the problem where people doing network installs via 3c905B cards experience a delay after the interface is first initialized, among other things.
1999-02-11 23:59:29 +00:00
#endif
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Init circular RX list. */
Fix handling of ENOBUFS condition. During reception, the filled mbuf cluster from the RX descriptor is passed up to the higher layers and replaced with an empty buffer for the next time the descriptor comes up in the RX ring. The xl_newbuf() routine returns ENOBUFS if it can't obtain a new mbuf cluster, but this return value was being ignored. Now, if buffer allocation fails, we leave the old one in place and drop the packet. This is rude, but there's not much else that can be done in this situation. Without this, the driver can cause a panic if the system runs out of MBUF clusters. Now it will complain loudly, but it shouldn't cause a panic. Also added another pair of missing newlines to some printf()s.
1998-08-31 15:26:09 +00:00
if (xl_list_rx_init(sc) == ENOBUFS) {
printf("xl%d: initialization failed: no "
"memory for rx buffers\n", sc->xl_unit);
xl_stop(sc);
Add suspend/resume hooks to this driver; necessary to overcome problems on HP Omnibook 500. MFC after: 1 week
2001-12-05 10:34:07 +00:00
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/* Init TX descriptors. */
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
if (sc->xl_type == XL_TYPE_905B)
xl_list_tx_init_90xB(sc);
else
xl_list_tx_init(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* Set the TX freethresh value.
* Note that this has no effect on 3c905B "cyclone"
* cards but is required for 3c900/3c905 "boomerang"
* cards in order to enable the download engine.
*/
CSR_WRITE_1(sc, XL_TX_FREETHRESH, XL_PACKET_SIZE >> 8);
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
/* Set the TX start threshold for best performance. */
Dynamically increase TX start threshold if TX underruns are detected.
1999-05-05 17:05:07 +00:00
sc->xl_tx_thresh = XL_MIN_FRAMELEN;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_SET_START|sc->xl_tx_thresh);
Bunch of updates: - Try to unbreak what I broke by screwing with the tx queuing again. I'm waiting for a few more people to test out this code and report back before I move it into current. Hopefully it will be soon. Basically I reverted to the old TX queuing strategy. - Add experimental support for the 3c900B-FL (10mbps ST fiber). The card should be detected properly and the 10baseFL mode supported, but again I'm still waiting for word from a tester to see if this actually works. It shouldn't affect the other cards though; all the differences are in media selection. - Set the TX start threshold register to get better performance. - Increase the size of the RX and TX rings. UDP performance was pretty bad because the TX ring was too small. Should be substantially better now (I can saturate the link with either TCP or UDP now). - Change some of the #defines to reflect proper 3Com ASIC names (boomerang, cyclone, krakatoa, hurricane). - Simplify and reorganize interrupt handler; ack all interrupts right away and then process them. This avoids a potential race condition. (Noted by Matt Dillon.) - Reorganize the bridging code to eliminate using a goto to jump into the middle of an if() {} clause. Sorry, that just made my brain itch. - Use m_adj() in xl_rxeof(). - Make the payload alignment in xl_newbuf() the default (instead of just conditionally defined for the alpha) to improve NFS performance (avoids need for nfs_realign()).
1999-04-30 16:15:43 +00:00
Close PR #8384: Revert the transmission packet queueing strategy changes. Clearly I missed something while debugging this, although I never encountered any problems on my test machines. Also make one other minor change: jack up the TX reclaim threshold for 3c90xB adapters in order to stave off 'transmission error: 82' errors. Document the existence of the tx reclaim register (for inspecting the current reclaim threshold) in register window 5 (if_xlreg.h).
1998-10-22 15:52:25 +00:00
/*
* If this is a 3c905B, also set the tx reclaim threshold.
* This helps cut down on the number of tx reclaim errors
* that could happen on a busy network. The chip multiplies
* the register value by 16 to obtain the actual threshold
* in bytes, so we divide by 16 when setting the value here.
* The existing threshold value can be examined by reading
* the register at offset 9 in window 5.
*/
if (sc->xl_type == XL_TYPE_905B) {
CSR_WRITE_2(sc, XL_COMMAND,
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
XL_CMD_SET_TX_RECLAIM|(XL_PACKET_SIZE >> 4));
Close PR #8384: Revert the transmission packet queueing strategy changes. Clearly I missed something while debugging this, although I never encountered any problems on my test machines. Also make one other minor change: jack up the TX reclaim threshold for 3c90xB adapters in order to stave off 'transmission error: 82' errors. Document the existence of the tx reclaim register (for inspecting the current reclaim threshold) in register window 5 (if_xlreg.h).
1998-10-22 15:52:25 +00:00
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Set RX filter bits. */
XL_SEL_WIN(5);
rxfilt = CSR_READ_1(sc, XL_W5_RX_FILTER);
/* Set the individual bit to receive frames for this host only. */
rxfilt |= XL_RXFILTER_INDIVIDUAL;
/* If we want promiscuous mode, set the allframes bit. */
if (ifp->if_flags & IFF_PROMISC) {
rxfilt |= XL_RXFILTER_ALLFRAMES;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
} else {
rxfilt &= ~XL_RXFILTER_ALLFRAMES;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
}
/*
* Set capture broadcast bit to capture broadcast frames.
*/
if (ifp->if_flags & IFF_BROADCAST) {
rxfilt |= XL_RXFILTER_BROADCAST;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
} else {
rxfilt &= ~XL_RXFILTER_BROADCAST;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
}
/*
* Program the multicast filter, if necessary.
*/
if (sc->xl_type == XL_TYPE_905B)
xl_setmulti_hash(sc);
else
xl_setmulti(sc);
/*
* Load the address of the RX list. We have to
* stall the upload engine before we can manipulate
* the uplist pointer register, then unstall it when
* we're finished. We also have to wait for the
* stall command to complete before proceeding.
* Note that we have to do this after any RX resets
* have completed since the uplist register is cleared
* by a reset.
*/
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_UP_STALL);
xl_wait(sc);
CSR_WRITE_4(sc, XL_UPLIST_PTR, vtophys(&sc->xl_ldata->xl_rx_list[0]));
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_UP_UNSTALL);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
xl_wait(sc);
if (sc->xl_type == XL_TYPE_905B) {
/* Set polling interval */
CSR_WRITE_1(sc, XL_DOWN_POLL, 64);
/* Load the address of the TX list */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_STALL);
xl_wait(sc);
CSR_WRITE_4(sc, XL_DOWNLIST_PTR,
vtophys(&sc->xl_ldata->xl_tx_list[0]));
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_DOWN_UNSTALL);
xl_wait(sc);
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/*
* If the coax transceiver is on, make sure to enable
* the DC-DC converter.
*/
XL_SEL_WIN(3);
if (sc->xl_xcvr == XL_XCVR_COAX)
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_START);
else
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_STOP);
Close PR #27742: allow the xl driver to receive VLAN tagged frames by setting the 'max packet size' register in window 3. This only works for cards based on the cyclone or newer chipsets (i.e. it won't work with the original 3c905/boomerang cards). There is a trick which will work with the boomerang, which is to turn on the 'large packets ok' bit in the MAC control register, however this lets the chip accept any frame up to 4K in length, which is larger than the mbuf cluster buffers we use to receive frames. If somebody sends us such a frame and the chip DMAs it to us, it could write past the end of the cluster buffer and clobber something. PR: kern/27742
2001-05-31 22:08:45 +00:00
/* increase packet size to allow reception of 802.1q or ISL packets */
if (sc->xl_type == XL_TYPE_905B)
CSR_WRITE_2(sc, XL_W3_MAXPKTSIZE, XL_PACKET_SIZE);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Clear out the stats counters. */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_STATS_DISABLE);
sc->xl_stats_no_timeout = 1;
xl_stats_update(sc);
sc->xl_stats_no_timeout = 0;
XL_SEL_WIN(4);
CSR_WRITE_2(sc, XL_W4_NET_DIAG, XL_NETDIAG_UPPER_BYTES_ENABLE);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_STATS_ENABLE);
/*
* Enable interrupts.
*/
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ACK|0xFF);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_STAT_ENB|XL_INTRS);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB|XL_INTRS);
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
if (sc->xl_flags & XL_FLAG_FUNCREG)
bus_space_write_4(sc->xl_ftag, sc->xl_fhandle, 4, 0x8000);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Set the RX early threshold */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_THRESH|(XL_PACKET_SIZE >>2));
CSR_WRITE_2(sc, XL_DMACTL, XL_DMACTL_UP_RX_EARLY);
/* Enable receiver and transmitter. */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_ENABLE);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
xl_wait(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_ENABLE);
xl_wait(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (mii != NULL)
mii_mediachg(mii);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Select window 7 for normal operations. */
XL_SEL_WIN(7);
ifp->if_flags |= IFF_RUNNING;
ifp->if_flags &= ~IFF_OACTIVE;
sc->xl_stat_ch = timeout(xl_stats_update, sc, hz);
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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/*
* Set media options.
*/
static int xl_ifmedia_upd(ifp)
struct ifnet *ifp;
{
struct xl_softc *sc;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct ifmedia *ifm = NULL;
struct mii_data *mii = NULL;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc = ifp->if_softc;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus != NULL)
mii = device_get_softc(sc->xl_miibus);
if (mii == NULL)
ifm = &sc->ifmedia;
else
ifm = &mii->mii_media;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
switch(IFM_SUBTYPE(ifm->ifm_media)) {
case IFM_100_FX:
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
case IFM_10_FL:
- Close PR #11136: add PCI ID for another new cyclone device: the 3c900B-TPC (twisted pair and coax). Treated similarly to the 3c900B-COMBO, except no AUI port. - Fix media selection so that it's possible to select the AUI and BNC ports on the 3c905B-COMBO. This board is now fully supported. - Change TX queueing strategy to hopefully be more efficient by avoiding register accesses in xl_start(). Should provide small performance improvement and a little better reliability.
1999-04-15 03:18:33 +00:00
case IFM_10_2:
case IFM_10_5:
xl_setmode(sc, ifm->ifm_media);
return(0);
break;
default:
break;
}
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (sc->xl_media & XL_MEDIAOPT_MII || sc->xl_media & XL_MEDIAOPT_BTX
|| sc->xl_media & XL_MEDIAOPT_BT4) {
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
xl_init(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
} else {
xl_setmode(sc, ifm->ifm_media);
}
return(0);
}
/*
* Report current media status.
*/
static void xl_ifmedia_sts(ifp, ifmr)
struct ifnet *ifp;
struct ifmediareq *ifmr;
{
struct xl_softc *sc;
u_int32_t icfg;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct mii_data *mii = NULL;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
sc = ifp->if_softc;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus != NULL)
mii = device_get_softc(sc->xl_miibus);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
XL_SEL_WIN(3);
icfg = CSR_READ_4(sc, XL_W3_INTERNAL_CFG) & XL_ICFG_CONNECTOR_MASK;
icfg >>= XL_ICFG_CONNECTOR_BITS;
ifmr->ifm_active = IFM_ETHER;
switch(icfg) {
case XL_XCVR_10BT:
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
ifmr->ifm_active = IFM_ETHER|IFM_10_T;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (CSR_READ_1(sc, XL_W3_MAC_CTRL) & XL_MACCTRL_DUPLEX)
ifmr->ifm_active |= IFM_FDX;
else
ifmr->ifm_active |= IFM_HDX;
break;
case XL_XCVR_AUI:
Fix the media selection for the 3c900B-FL 10baseFL adapter. It actually uses the AUI port with an on-board AUI to 10baseFL transceiver, not the 10baseT port like I had earlier suspected. The 3c900B-FL should be properly supported now.
1999-05-04 20:52:30 +00:00
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
ifmr->ifm_active = IFM_ETHER|IFM_10_FL;
if (CSR_READ_1(sc, XL_W3_MAC_CTRL) & XL_MACCTRL_DUPLEX)
ifmr->ifm_active |= IFM_FDX;
else
ifmr->ifm_active |= IFM_HDX;
} else
ifmr->ifm_active = IFM_ETHER|IFM_10_5;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
case XL_XCVR_COAX:
ifmr->ifm_active = IFM_ETHER|IFM_10_2;
break;
/*
* XXX MII and BTX/AUTO should be separate cases.
*/
case XL_XCVR_100BTX:
case XL_XCVR_AUTO:
case XL_XCVR_MII:
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (mii != NULL) {
mii_pollstat(mii);
ifmr->ifm_active = mii->mii_media_active;
ifmr->ifm_status = mii->mii_media_status;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
}
break;
case XL_XCVR_100BFX:
ifmr->ifm_active = IFM_ETHER|IFM_100_FX;
break;
default:
printf("xl%d: unknown XCVR type: %d\n", sc->xl_unit, icfg);
break;
}
return;
}
static int xl_ioctl(ifp, command, data)
struct ifnet *ifp;
u_long command;
caddr_t data;
{
struct xl_softc *sc = ifp->if_softc;
struct ifreq *ifr = (struct ifreq *) data;
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
int error = 0;
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
struct mii_data *mii = NULL;
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
u_int8_t rxfilt;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
switch(command) {
case SIOCSIFADDR:
case SIOCGIFADDR:
case SIOCSIFMTU:
error = ether_ioctl(ifp, command, data);
break;
case SIOCSIFFLAGS:
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
XL_SEL_WIN(5);
rxfilt = CSR_READ_1(sc, XL_W5_RX_FILTER);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
if (ifp->if_flags & IFF_UP) {
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
if (ifp->if_flags & IFF_RUNNING &&
ifp->if_flags & IFF_PROMISC &&
!(sc->xl_if_flags & IFF_PROMISC)) {
rxfilt |= XL_RXFILTER_ALLFRAMES;
CSR_WRITE_2(sc, XL_COMMAND,
XL_CMD_RX_SET_FILT|rxfilt);
XL_SEL_WIN(7);
} else if (ifp->if_flags & IFF_RUNNING &&
!(ifp->if_flags & IFF_PROMISC) &&
sc->xl_if_flags & IFF_PROMISC) {
rxfilt &= ~XL_RXFILTER_ALLFRAMES;
CSR_WRITE_2(sc, XL_COMMAND,
XL_CMD_RX_SET_FILT|rxfilt);
XL_SEL_WIN(7);
} else
xl_init(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
} else {
if (ifp->if_flags & IFF_RUNNING)
xl_stop(sc);
}
Make some small tweaks: - When setting/clearing promisc mode, just update the filter, don't reset the whole interface. - Call xl_init() in xl_ifmedia_upd() when setting miibus media modes. This fixes a problem with the 3c905B-COMBO where switching from 10base5/AUI or 10base2/BNC to a 10/100 mode doesn't always work right. - Attempt to reset the interface in xl_init() so that we know we're getting the receive and transmit rings reset properly.
1999-10-14 21:49:17 +00:00
sc->xl_if_flags = ifp->if_flags;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
error = 0;
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
if (sc->xl_type == XL_TYPE_905B)
xl_setmulti_hash(sc);
else
xl_setmulti(sc);
error = 0;
break;
case SIOCGIFMEDIA:
case SIOCSIFMEDIA:
Convert the 3Com XL driver to miibus. This one is a little tricky due to the fact that there are non-MII cards supported by the same driver and I don't have all of the cards available for testing. There's also the 3c905B-COMBO which has MII, AUI and BNC media ports all in one package. Supporting the COMBO is difficult because we have to add the 10base5 and 10base2 media types to the same ifmedia struct as the MII-attached types, however there is no way to force the miibus and child PHYs into existence before xl_attach() completes, so there is no ifmedia struct available in xl_attach(). What we do inistead is use the mediainit method as a callback: when a child PHY is attached, it calls the miibus mediainit routine which selects a default media. This routing also calls the NIC driver's mediainit method (if it implements one) at which point we can safely add the other media types.
1999-08-29 15:52:19 +00:00
if (sc->xl_miibus != NULL)
mii = device_get_softc(sc->xl_miibus);
if (mii == NULL)
error = ifmedia_ioctl(ifp, ifr,
&sc->ifmedia, command);
else
error = ifmedia_ioctl(ifp, ifr,
&mii->mii_media, command);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
break;
default:
error = EINVAL;
break;
}
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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return(error);
}
static void xl_watchdog(ifp)
struct ifnet *ifp;
{
struct xl_softc *sc;
u_int16_t status = 0;
sc = ifp->if_softc;
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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp->if_oerrors++;
XL_SEL_WIN(4);
status = CSR_READ_2(sc, XL_W4_MEDIA_STATUS);
printf("xl%d: watchdog timeout\n", sc->xl_unit);
if (status & XL_MEDIASTAT_CARRIER)
printf("xl%d: no carrier - transceiver cable problem?\n",
sc->xl_unit);
xl_txeoc(sc);
Apply patch graciously provided by Jason Wright <jason@thought.net> from the OpenBSD group to fix a problem with the default ifmedia not being set properly in some cases with a 3c905B, leading to a panic in ifmedia_set(). Also apply a patch to force the transmit start routine to check the transmitter to make sure it isn't wedged if the outbound tx queue appears full. This seems to cure some problems with 'watchdog timeout' errors cropping up in some cases. I tried to do this before by checking for the IFF_OACTIVE flag on entry to xl_start(), but if the IFF_OACTIVE flag is set, ether_output() won't even call xl_start(). It should work now. Lastly, increase the size of the TX queue from 10 descriptors to 16 to hopefully make it less likely that the TX queue will fill up.
1998-09-25 17:34:19 +00:00
xl_txeof(sc);
xl_rxeof(sc);
Perform an RX reset and TX reset in xl_reset() along with the master reset command. I observed some anomalous behavior while testing a 3c905C with a Dell PowerEdge 4300/500 dual PIII 500Mhz system. The NIC would seem to work correctly most of the time but would sometimes fail to receive certain packets, in particular NFS create requests. I could mount an NFS filesystem from the PowerEdge and do an ls on it, but trying to do a "touch foo" would hang. Monitoring traffic from another host revealed that the client was properly sending an NFS create request but the server was not receiving it. It *did* receive it when I ran the same test with an Intel fxp card. I don't understand the exact mechanics of this strange behavior, but resetting the receiver and transmitter seems to get rid of it. I used to perform an RX and TX reset in xl_init(), but stopped doing it there because on 3c905B and later cards this causes the autoneg session to restart, which would lead to the NIC waiting a long time before exchanging traffic after being brought up the first time. Apparently the receiver and transmitter resets should be performed at least once when initializing the card. Hopefully this will cure problems that people have been having with the 3c905C -- this was the only strange behavior that I have observed with the 3c905C so far which does not appear with the 3c905B or 3c905.
1999-08-02 21:06:16 +00:00
xl_reset(sc);
Apply patch graciously provided by Jason Wright <jason@thought.net> from the OpenBSD group to fix a problem with the default ifmedia not being set properly in some cases with a 3c905B, leading to a panic in ifmedia_set(). Also apply a patch to force the transmit start routine to check the transmitter to make sure it isn't wedged if the outbound tx queue appears full. This seems to cure some problems with 'watchdog timeout' errors cropping up in some cases. I tried to do this before by checking for the IFF_OACTIVE flag on entry to xl_start(), but if the IFF_OACTIVE flag is set, ether_output() won't even call xl_start(). It should work now. Lastly, increase the size of the TX queue from 10 descriptors to 16 to hopefully make it less likely that the TX queue will fill up.
1998-09-25 17:34:19 +00:00
xl_init(sc);
if (ifp->if_snd.ifq_head != NULL)
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
(*ifp->if_start)(ifp);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/*
* Stop the adapter and free any mbufs allocated to the
* RX and TX lists.
*/
static void xl_stop(sc)
struct xl_softc *sc;
{
register int i;
struct ifnet *ifp;
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
XL_LOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
ifp = &sc->arpcom.ac_if;
ifp->if_timer = 0;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_DISABLE);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_STATS_DISABLE);
- #define mask of enabled interrupts/indications in if_xlreg.h instead of constructing local copy in xl_init() - disable interrupts on entry to xl_intr(), re-enable them on exit. - fix a few typos in some comments
1998-08-24 17:51:38 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_DISCARD);
xl_wait(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_DISABLE);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_COAX_STOP);
DELAY(800);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
#ifdef foo
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_RESET);
xl_wait(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_TX_RESET);
xl_wait(sc);
Do not issue RX and TX reset commands in xl_stop() and xl_init(). On the 3c905B, the RX and TX reset commands also reset the cyclone chip's internal PHY, which causes it to restart its autonegotiation session. This takes a second or two to complete, which makes the interface seem to stop responding for a few seconds every time you do something that reinitializes it. Issuing the RX and TX resets on the older 3c905 boomerang adapters doesn't cause any delay because the boomerang chip requires an external PHY. This should fix the problem where people doing network installs via 3c905B cards experience a delay after the interface is first initialized, among other things.
1999-02-11 23:59:29 +00:00
#endif
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ACK|XL_STAT_INTLATCH);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_STAT_ENB|0);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB|0);
Add support for the 3Com 556 and 556B mini-pci adapters used on some laptops. I've checked that this still works with the other cards and it works with the 3c556 that I have access to, but I want to check that it works with the 556B mentioned in PR #20878 before I close out the PR and merge to -stable.
2000-08-28 20:40:03 +00:00
if (sc->xl_flags & XL_FLAG_FUNCREG) bus_space_write_4 (sc->xl_ftag, sc->xl_fhandle, 4, 0x8000);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
/* Stop the stats updater. */
untimeout(xl_stats_update, sc, sc->xl_stat_ch);
/*
* Free data in the RX lists.
*/
for (i = 0; i < XL_RX_LIST_CNT; i++) {
if (sc->xl_cdata.xl_rx_chain[i].xl_mbuf != NULL) {
m_freem(sc->xl_cdata.xl_rx_chain[i].xl_mbuf);
sc->xl_cdata.xl_rx_chain[i].xl_mbuf = NULL;
}
}
bzero((char *)&sc->xl_ldata->xl_rx_list,
sizeof(sc->xl_ldata->xl_rx_list));
/*
* Free the TX list buffers.
*/
for (i = 0; i < XL_TX_LIST_CNT; i++) {
if (sc->xl_cdata.xl_tx_chain[i].xl_mbuf != NULL) {
m_freem(sc->xl_cdata.xl_tx_chain[i].xl_mbuf);
sc->xl_cdata.xl_tx_chain[i].xl_mbuf = NULL;
}
}
bzero((char *)&sc->xl_ldata->xl_tx_list,
sizeof(sc->xl_ldata->xl_tx_list));
ifp->if_flags &= ~(IFF_RUNNING | IFF_OACTIVE);
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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
/*
* Stop all chip I/O so that the kernel's probe routines don't
* get confused by errant DMAs when rebooting.
*/
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
static void xl_shutdown(dev)
device_t dev;
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
{
Convert the xl driver to newbus. It is now possible to make this driver into a loadable module, and all of the platform dependencies are gone (except for the alpha_XXX_dmamap() thing, which is another issue -- I still don't know how to use the busdma stuff with a network driver). Also increase the delay in xl_reset(); testing on a 486/66 with a 3c905C shows that reading the EEPROM fails immediately after a reset. Waiting a little longer after the reset completes seems to fix it.
1999-07-20 21:23:17 +00:00
struct xl_softc *sc;
sc = device_get_softc(dev);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +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
XL_LOCK(sc);
Add an alternate transmit strategy for 3c90xB adapters based on the transmit strategy used in the 3Com Linux driver. The new strategy is to use transmit descriptor polling -- that is, the NIC polls the descriptors to see when new packets are available for transmission. The advantage to the new scheme is that no register accesses are needed in the transmit routine. The old scheme requires several register accesses to stall the TX engine, update the TX DMA list pointer register, then unstall the TX engine. Hopefully the new scheme will provide improved transmit performance with less CPU overhead. This only affects the 3c90xB or 3c90xC cards, not the 3c90x cards. This means the original 3c900 and 3c905 cards are unaffected. Newer cards include the 3c900B series, the 3c905B, 3c980, 3c980B, 3c905C and 3c905C, and the 3cSOHO100-TX OfficeConnect.
1999-09-20 00:24:11 +00:00
xl_reset(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
xl_stop(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
XL_UNLOCK(sc);
Import the (Fast) Etherlink XL driver. I'm reasonally confident in its stability now. ALso modify /sys/conf/files, /sys/i386/conf/GENERIC and /sys/i386/conf/LINT to add entries for the XL driver. Deactivate support for the XL adapters in the vortex driver. LAstly, add a man page. (Also added an MLINKS entry for the ThunderLAN man page which I forgot previously.)
1998-08-16 17:14:59 +00:00
return;
}
Add suspend/resume hooks to this driver; necessary to overcome problems on HP Omnibook 500. MFC after: 1 week
2001-12-05 10:34:07 +00:00
static int xl_suspend(dev)
device_t dev;
{
struct xl_softc *sc;
sc = device_get_softc(dev);
XL_LOCK(sc);
xl_stop(sc);
XL_UNLOCK(sc);
return(0);
}
static int xl_resume(dev)
device_t dev;
{
struct xl_softc *sc;
struct ifnet *ifp;
sc = device_get_softc(dev);
XL_LOCK(sc);
ifp = &sc->arpcom.ac_if;
xl_reset(sc);
if (ifp->if_flags & IFF_UP)
xl_init(sc);
XL_UNLOCK(sc);
return(0);
}
Reference in New Issue Copy Permalink