d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
4092996774
freebsd-nq/sys/pci/if_xl.c

3388 lines
88 KiB
C
Raw Normal View History

/* -> /*- for license, minor formatting changes
2005-01-07 02:29:27 +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.
*/
Remove duplicate FBSDID's, move others to their right place.
2003-11-14 17:16:58 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
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 3c555 miniPCI chipset. Submitted by: johannes <johannes@paradise.net.nz> PR: 46603 MFC after: 3 days
2003-01-06 00:46:03 +00:00
* 3Com 3c555 10/100Mbps/RJ-45 (MiniPCI, Laptop Hurricane)
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
Correct typos, mostly s/ a / an / where appropriate. Some whitespace cleanup, especially in troff files.
2003-01-01 18:49:04 +00:00
* chain into an mbuf cluster and then DMAing the cluster. This extra
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
* 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>
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
#include <sys/endian.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 <sys/mbuf.h>
#include <sys/kernel.h>
Add missing <sys/module.h> includes
2004-05-30 20:00:41 +00:00
#include <sys/module.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 <sys/socket.h>
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
#include <sys/taskqueue.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 <net/if.h>
#include <net/if_arp.h>
#include <net/ethernet.h>
#include <net/if_dl.h>
#include <net/if_media.h>
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
#include <net/if_types.h>
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 <net/bpf.h>
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.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>
Prefer new location of pci include files (which have only been in the tree for two or more years now), except in a few places where there's code to be compatible with older versions of FreeBSD.
2003-08-22 07:20:27 +00:00
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
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
- Express hard dependencies on bus (pci, isa, pccard) and network layer (ether). - Don't abuse module names to facilitate ifconfig module loading; such abuse isn't really needed. (And if we do need type information associated with a module then we should make it explicit and not use hacks.)
2003-04-15 06:37:30 +00:00
MODULE_DEPEND(xl, pci, 1, 1, 1);
MODULE_DEPEND(xl, ether, 1, 1, 1);
Depend on miibus. Note that if_aue doesn't strictly depend on usb because it uses the method interface for calls rather than using internal symbols, and because it's a child driver of usb and therefore will not try and do anything unless the parent usb code is loaded at some point. if_aue does strictly depend on miibus as it will fail to link if it is missing.
2000-04-29 13:41:57 +00:00
MODULE_DEPEND(xl, miibus, 1, 1, 1);
s/driver/device/ for config file line
2003-11-06 03:59:03 +00:00
/* "device miibus" 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
#include <pci/if_xlreg.h>
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
/*
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +00:00
* TX Checksumming is disabled by default for two reasons:
* - TX Checksumming will occasionally produce corrupt packets
* - TX Checksumming seems to reduce performance
*
* Only 905B/C cards were reported to have this problem, it is possible
* that later chips _may_ be immune.
*/
#define XL905B_TXCSUM_BROKEN 1
#ifdef XL905B_TXCSUM_BROKEN
#define XL905B_CSUM_FEATURES 0
#else
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)
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +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
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" },
Correctly identify the 3C920B chipset; although it may ship with some Nforce2 boards, it's definitely not an Nvidia chip. Info from: Nvidia's Linux Network driver & pciids.sourceforge.net
2003-01-22 17:14:33 +00:00
{ TC_VENDORID, TC_DEVICEID_TORNADO_10_100BT_920B,
"3Com 3c920B-EMB Integrated Fast Etherlink XL" },
Add PCI and device ID's to if_xl to support: 3C920B-EMB-WNM Integrated Fast Ethernet Controller Submitter reports that the card appears to autonegotiate properly, and operate well with high levels of NFS traffic. PR: 75253 Submitted by: "Oleg V. Nauman" <oleg at reis dot zp dot ua> MFC after: 2 weeks
2005-01-03 19:22:37 +00:00
{ TC_VENDORID, TC_DEVICEID_TORNADO_10_100BT_920B_WNM,
"3Com 3c920B-EMB-WNM Integrated 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 3c555 miniPCI chipset. Submitted by: johannes <johannes@paradise.net.nz> PR: 46603 MFC after: 3 days
2003-01-06 00:46:03 +00:00
{ TC_VENDORID, TC_DEVICEID_HURRICANE_555,
"3Com 3c555 Fast Etherlink XL" },
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 }
};
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +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
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static int xl_newbuf(struct xl_softc *, struct xl_chain_onefrag *);
static void xl_stats_update(void *);
static void xl_stats_update_locked(struct xl_softc *);
static int xl_encap(struct xl_softc *, struct xl_chain *, struct mbuf *);
static void xl_rxeof(struct xl_softc *);
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
static void xl_rxeof_task(void *, int);
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
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_locked(struct ifnet *);
static void xl_start_90xB_locked(struct ifnet *);
static int xl_ioctl(struct ifnet *, u_long, caddr_t);
static void xl_init(void *);
static void xl_init_locked(struct xl_softc *);
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
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
static void xl_poll(struct ifnet *ifp, enum poll_cmd cmd, int count);
static void xl_poll_locked(struct ifnet *ifp, enum poll_cmd cmd, int count);
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#endif
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +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
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +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
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +00:00
static void xl_setcfg(struct xl_softc *);
static void xl_setmode(struct xl_softc *, int);
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_media(struct xl_softc *sc, int *media);
static void xl_choose_xcvr(struct xl_softc *, int);
static void xl_dma_map_addr(void *, bus_dma_segment_t *, int, int);
static void xl_dma_map_rxbuf(void *, bus_dma_segment_t *, int, bus_size_t, int);
static void xl_dma_map_txbuf(void *, bus_dma_segment_t *, int, bus_size_t, 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
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +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
Fix style(9) issues with __P removal. Noticed by: bde
2005-02-24 22:33:05 +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
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;
Sometimes cardbus attachments don't attach, so while we track down this problem put these lines back in. While they should be unnecessary, they appear to be sometimes necessary. Reviewed in concept: dfr Approved by: re (scottl@)
2003-11-28 05:28:29 +00:00
DRIVER_MODULE(xl, cardbus, xl_driver, xl_devclass, 0, 0);
- Express hard dependencies on bus (pci, isa, pccard) and network layer (ether). - Don't abuse module names to facilitate ifconfig module loading; such abuse isn't really needed. (And if we do need type information associated with a module then we should make it explicit and not use hacks.)
2003-04-15 06:37:30 +00:00
DRIVER_MODULE(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
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_dma_map_addr(void *arg, bus_dma_segment_t *segs, int nseg, int error)
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
{
u_int32_t *paddr;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
paddr = arg;
*paddr = segs->ds_addr;
}
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the RX part of this driver too. It's better since the code wasn't dealing with bus_dmamap_load() returning EINPROGRESS, and this can't happen with bus_dmamap_load_mbuf(). Submitted by: jake
2003-04-05 23:24:23 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_dma_map_rxbuf(void *arg, bus_dma_segment_t *segs, int nseg,
bus_size_t mapsize, int error)
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the RX part of this driver too. It's better since the code wasn't dealing with bus_dmamap_load() returning EINPROGRESS, and this can't happen with bus_dmamap_load_mbuf(). Submitted by: jake
2003-04-05 23:24:23 +00:00
{
u_int32_t *paddr;
if (error)
return;
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the RX part of this driver too. It's better since the code wasn't dealing with bus_dmamap_load() returning EINPROGRESS, and this can't happen with bus_dmamap_load_mbuf(). Submitted by: jake
2003-04-05 23:24:23 +00:00
KASSERT(nseg == 1, ("xl_dma_map_rxbuf: too many DMA segments"));
paddr = arg;
*paddr = segs->ds_addr;
}
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_dma_map_txbuf(void *arg, bus_dma_segment_t *segs, int nseg,
bus_size_t mapsize, int error)
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
{
struct xl_list *l;
int i, total_len;
if (error)
return;
KASSERT(nseg <= XL_MAXFRAGS, ("too many DMA segments"));
total_len = 0;
l = arg;
for (i = 0; i < nseg; i++) {
KASSERT(segs[i].ds_len <= MCLBYTES, ("segment size too large"));
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
l->xl_frag[i].xl_addr = htole32(segs[i].ds_addr);
l->xl_frag[i].xl_len = htole32(segs[i].ds_len);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
total_len += segs[i].ds_len;
}
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
l->xl_frag[nseg - 1].xl_len = htole32(segs[nseg - 1].ds_len |
XL_LAST_FRAG);
l->xl_status = htole32(total_len);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
l->xl_next = 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
/*
* 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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_wait(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
{
register int i;
for (i = 0; i < XL_TIMEOUT; i++) {
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
if ((CSR_READ_2(sc, XL_STATUS) & XL_STAT_CMDBUSY) == 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
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "command never completed!\n");
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
}
/*
* 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, \
Be consistent about functions being static. Properly put macro args in (). Spotted by: FlexeLint.
2002-10-16 09:14:59 +00:00
CSR_READ_2(sc, XL_W4_PHY_MGMT) | (x))
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
#define MII_CLR(x) \
CSR_WRITE_2(sc, XL_W4_PHY_MGMT, \
Be consistent about functions being static. Properly put macro args in (). Spotted by: FlexeLint.
2002-10-16 09:14:59 +00:00
CSR_READ_2(sc, XL_W4_PHY_MGMT) & ~(x))
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
/*
* Sync the PHYs by setting data bit and strobing the clock 32 times.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_mii_sync(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
{
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);
Add two MII_SETs to provide a bit more time between operations in xl_sync; this appears to be necessary with certain systems. Requested by: Michael Class <michael_class@gmx.net>
2003-01-06 00:36:13 +00:00
MII_SET(XL_MII_DATA);
Add a tiny bit more delay in the xl_mii_sync function; this is necessary for proper intialization in certain 905B + old system combinations. Tested by: Jakub Miziolek <jxm@obta.uw.edu.pl>
2003-09-29 02:14:04 +00:00
MII_SET(XL_MII_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
MII_CLR(XL_MII_CLK);
Add two MII_SETs to provide a bit more time between operations in xl_sync; this appears to be necessary with certain systems. Requested by: Michael Class <michael_class@gmx.net>
2003-01-06 00:36:13 +00:00
MII_SET(XL_MII_DATA);
Add a tiny bit more delay in the xl_mii_sync function; this is necessary for proper intialization in certain 905B + old system combinations. Tested by: Jakub Miziolek <jxm@obta.uw.edu.pl>
2003-09-29 02:14:04 +00:00
MII_SET(XL_MII_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
}
}
/*
* Clock a series of bits through the MII.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_mii_send(struct xl_softc *sc, u_int32_t bits, int cnt)
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
{
int i;
XL_SEL_WIN(4);
MII_CLR(XL_MII_CLK);
for (i = (0x1 << (cnt - 1)); i; i >>= 1) {
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
if (bits & 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
MII_SET(XL_MII_DATA);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
} else {
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
MII_CLR(XL_MII_DATA);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
MII_CLR(XL_MII_CLK);
MII_SET(XL_MII_CLK);
}
}
/*
* Read an PHY register through the MII.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_mii_readreg(struct xl_softc *sc, struct xl_mii_frame *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
{
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
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Set up frame for RX. */
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_stdelim = XL_MII_STARTDELIM;
frame->mii_opcode = XL_MII_READOP;
frame->mii_turnaround = 0;
frame->mii_data = 0;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Select register window 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
XL_SEL_WIN(4);
CSR_WRITE_2(sc, XL_W4_PHY_MGMT, 0);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Turn on data xmit. */
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
MII_SET(XL_MII_DIR);
xl_mii_sync(sc);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Send command/address info. */
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_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));
MII_SET(XL_MII_CLK);
/* Turn off xmit. */
MII_CLR(XL_MII_DIR);
/* Check for ack */
MII_CLR(XL_MII_CLK);
ack = CSR_READ_2(sc, XL_W4_PHY_MGMT) & XL_MII_DATA;
When reading PHY regs over the i2c bus, the turnaround ACK bit is read one clock edge too late. This bit is driven low by slave (as any other input data bits from slave) when the clock is LOW. The current code did read the bit after the clock was driven high again. Reviewed by: luoqi MFC after: 2 weeks
2003-01-10 08:09:58 +00:00
MII_SET(XL_MII_CLK);
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 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) {
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
for (i = 0; i < 16; 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
MII_CLR(XL_MII_CLK);
MII_SET(XL_MII_CLK);
}
goto fail;
}
for (i = 0x8000; i; i >>= 1) {
MII_CLR(XL_MII_CLK);
if (!ack) {
if (CSR_READ_2(sc, XL_W4_PHY_MGMT) & XL_MII_DATA)
frame->mii_data |= i;
}
MII_SET(XL_MII_CLK);
}
fail:
MII_CLR(XL_MII_CLK);
MII_SET(XL_MII_CLK);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
return (ack ? 1 : 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
}
/*
* Write to a PHY register through the MII.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_mii_writereg(struct xl_softc *sc, struct xl_mii_frame *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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Set up frame for 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
frame->mii_stdelim = XL_MII_STARTDELIM;
frame->mii_opcode = XL_MII_WRITEOP;
frame->mii_turnaround = XL_MII_TURNAROUND;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Select the window 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
XL_SEL_WIN(4);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Turn on data output. */
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
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);
MII_CLR(XL_MII_CLK);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
/* Turn off xmit. */
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
MII_CLR(XL_MII_DIR);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_miibus_readreg(device_t dev, int phy, int 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.
*/
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
if ((sc->xl_flags & XL_FLAG_PHYOK) == 0 && phy != 24)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (0);
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
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;
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +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
xl_mii_readreg(sc, &frame);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (frame.mii_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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_miibus_writereg(device_t dev, int phy, int reg, int 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);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
if ((sc->xl_flags & XL_FLAG_PHYOK) == 0 && phy != 24)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (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
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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_miibus_statchg(device_t dev)
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
{
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
struct xl_softc *sc;
struct mii_data *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 = device_get_softc(dev);
mii = device_get_softc(sc->xl_miibus);
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,
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
(CSR_READ_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
}
/*
* 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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_miibus_mediainit(device_t dev)
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
{
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
struct xl_softc *sc;
struct mii_data *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
struct ifmedia *ifm;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
sc = device_get_softc(dev);
mii = device_get_softc(sc->xl_miibus);
ifm = &mii->mii_media;
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
if (sc->xl_media & (XL_MEDIAOPT_AUI | XL_MEDIAOPT_10FL)) {
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
/*
* Check for a 10baseFL board in disguise.
*/
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
if (bootverbose)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"found 10baseFL\n");
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
ifmedia_add(ifm, IFM_ETHER | IFM_10_FL, 0, NULL);
ifmedia_add(ifm, IFM_ETHER | IFM_10_FL|IFM_HDX, 0,
NULL);
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_caps & XL_CAPS_FULL_DUPLEX)
ifmedia_add(ifm,
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
IFM_ETHER | IFM_10_FL | IFM_FDX, 0, NULL);
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 {
if (bootverbose)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "found AUI\n");
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
ifmedia_add(ifm, IFM_ETHER | IFM_10_5, 0, NULL);
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_BNC) {
if (bootverbose)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "found BNC\n");
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
ifmedia_add(ifm, IFM_ETHER | IFM_10_2, 0, NULL);
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
}
/*
* The EEPROM is slow: give it time to come ready after issuing
* it a command.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_eeprom_wait(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
{
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) {
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "eeprom failed to come ready\n");
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (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
}
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
/*
* Read a sequence of words from the EEPROM. Note that ethernet address
* data is stored in the EEPROM in network byte order.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_read_eeprom(struct xl_softc *sc, caddr_t dest, int off, int cnt, int swap)
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
{
int err = 0, i;
u_int16_t word = 0, *ptr;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
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)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
/*
* XXX: WARNING! DANGER!
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
* 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))
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (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
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)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
CSR_WRITE_2(sc, XL_W0_EE_CMD,
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_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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
*ptr = word;
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 whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (err ? 1 : 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
}
/*
* NICs older than the 3c905B have only one multicast option, which
* is to enable reception of all multicast frames.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_setmulti(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
{
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
struct ifmultiaddr *ifma;
u_int8_t rxfilt;
int mcnt = 0;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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_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;
}
Modify device drivers supporting multicast addresses to lock if_addr_mtx over iteration of their multicast address lists when synchronizing the hardware address filter with the network stack-maintained list. Problem reported by: Ed Maste (emaste at phaedrus dot sandvine dot ca> MFC after: 1 week
2005-08-03 00:18:35 +00:00
IF_ADDR_LOCK(ifp);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link)
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++;
Modify device drivers supporting multicast addresses to lock if_addr_mtx over iteration of their multicast address lists when synchronizing the hardware address filter with the network stack-maintained list. Problem reported by: Ed Maste (emaste at phaedrus dot sandvine dot ca> MFC after: 1 week
2005-08-03 00:18:35 +00:00
IF_ADDR_UNLOCK(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
if (mcnt)
rxfilt |= XL_RXFILTER_ALLMULTI;
else
rxfilt &= ~XL_RXFILTER_ALLMULTI;
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RX_SET_FILT|rxfilt);
}
/*
* 3c905B adapters have a hash filter that we can program.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_setmulti_hash(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
{
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
int h = 0, i;
struct ifmultiaddr *ifma;
u_int8_t rxfilt;
int mcnt = 0;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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_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 */
Modify device drivers supporting multicast addresses to lock if_addr_mtx over iteration of their multicast address lists when synchronizing the hardware address filter with the network stack-maintained list. Problem reported by: Ed Maste (emaste at phaedrus dot sandvine dot ca> MFC after: 1 week
2005-08-03 00:18:35 +00:00
IF_ADDR_LOCK(ifp);
Convert if_multiaddrs from LIST to TAILQ so that it can be traversed backwards in the three drivers which want to do that. Reviewed by: mikeh
2001-02-06 10:12:15 +00:00
TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
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;
Replace handrolled CRC calculation with ether_crc32_[lb]e().
2004-06-09 14:34:04 +00:00
/*
* 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....
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
*/
Replace handrolled CRC calculation with ether_crc32_[lb]e().
2004-06-09 14:34:04 +00:00
h = ether_crc32_be(LLADDR((struct sockaddr_dl *)
ifma->ifma_addr), ETHER_ADDR_LEN) & 0xFF;
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
CSR_WRITE_2(sc, XL_COMMAND,
h | XL_CMD_RX_SET_HASH | XL_HASH_SET);
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++;
}
Modify device drivers supporting multicast addresses to lock if_addr_mtx over iteration of their multicast address lists when synchronizing the hardware address filter with the network stack-maintained list. Problem reported by: Ed Maste (emaste at phaedrus dot sandvine dot ca> MFC after: 1 week
2005-08-03 00:18:35 +00:00
IF_ADDR_UNLOCK(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
if (mcnt)
rxfilt |= XL_RXFILTER_MULTIHASH;
else
rxfilt &= ~XL_RXFILTER_MULTIHASH;
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
CSR_WRITE_2(sc, XL_COMMAND, rxfilt | XL_CMD_RX_SET_FILT);
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
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_testpacket(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 mbuf *m;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
MGETHDR(m, M_DONTWAIT, MT_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
if (m == NULL)
return;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
bcopy(&IFP2ENADDR(sc->xl_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
mtod(m, struct ether_header *)->ether_dhost, ETHER_ADDR_LEN);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
bcopy(&IFP2ENADDR(sc->xl_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
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;
Bring in the first chunk of altq driver modifications. This covers the following drivers: bfe(4), em(4), fxp(4), lnc(4), tun(4), de(4) rl(4), sis(4) and xl(4) More patches are pending on: http://peoples.freebsd.org/~mlaier/ Please take a look and tell me if "your" driver is missing, so I can fix this. Tested-by: many No-objection: -current, -net
2004-07-02 12:16:02 +00:00
IFQ_ENQUEUE(&ifp->if_snd, m);
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_start(ifp);
}
#endif
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_setcfg(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
{
- 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;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/*XL_LOCK_ASSERT(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
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);
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_setmode(struct xl_softc *sc, int 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
{
u_int32_t icfg;
u_int16_t mediastat;
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
char *pmsg = "", *dmsg = "";
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 fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_LOCK_ASSERT(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
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) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
pmsg = "10baseT transceiver";
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_xcvr = XL_XCVR_10BT;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_10BT << XL_ICFG_CONNECTOR_BITS);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
mediastat |= XL_MEDIASTAT_LINKBEAT |
XL_MEDIASTAT_JABGUARD;
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
mediastat &= ~XL_MEDIASTAT_SQEENB;
}
}
if (sc->xl_media & XL_MEDIAOPT_BFX) {
if (IFM_SUBTYPE(media) == IFM_100_FX) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
pmsg = "100baseFX port";
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_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) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
pmsg = "AUI port";
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_xcvr = XL_XCVR_AUI;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_AUI << XL_ICFG_CONNECTOR_BITS);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
mediastat &= ~(XL_MEDIASTAT_LINKBEAT |
XL_MEDIASTAT_JABGUARD);
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
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) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
pmsg = "10baseFL transceiver";
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;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_AUI << XL_ICFG_CONNECTOR_BITS);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
mediastat &= ~(XL_MEDIASTAT_LINKBEAT |
XL_MEDIASTAT_JABGUARD);
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
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) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
pmsg = "AUI port";
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_xcvr = XL_XCVR_COAX;
icfg &= ~XL_ICFG_CONNECTOR_MASK;
icfg |= (XL_XCVR_COAX << XL_ICFG_CONNECTOR_BITS);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
mediastat &= ~(XL_MEDIASTAT_LINKBEAT |
XL_MEDIASTAT_JABGUARD | 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 ((media & IFM_GMASK) == IFM_FDX ||
IFM_SUBTYPE(media) == IFM_100_FX) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
dmsg = "full";
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);
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 {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
dmsg = "half";
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);
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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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_4(sc, XL_W3_INTERNAL_CFG, icfg);
XL_SEL_WIN(4);
CSR_WRITE_2(sc, XL_W4_MEDIA_STATUS, mediastat);
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +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
DELAY(800);
XL_SEL_WIN(7);
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "selecting %s, %s duplex\n", pmsg, dmsg);
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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_reset(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
{
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;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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_SEL_WIN(0);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_RESET |
((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
Modify the xl_reset() routine slightly so that, if we're using memory mapped I/O mode, we pause for .1 seconds after issuing the reset command before trying to poll the 'command busy' bit in the status register. With my 3c575C cardbus NIC, my Sony Picturebook locks up when it tries to read the status register immediately after the reset. This appears to be a problem only with certain NICs on certain hardware, but the added delay should not hurt cards that already work. This bug seems to have been brought to light by the fact that the xl driver now defaults to memory mapped I/O mode instead of programmed I/O mode like it used to. With PIO mode, the delay isn't needed and everything works (which is why this NIC worked with 5.0-RELEASE but not 5.1). I suspect that what's happening is that when the chip is reset, it takes a little while for the memory-mapped decoding logic to recover. Trying to access the chip's registers during this period causes an error condition of some kind that wedges the system.
2003-06-29 01:38:57 +00:00
/*
* If we're using memory mapped register mode, pause briefly
* after issuing the reset command before trying to access any
* other registers. With my 3c575C cardbus card, failing to do
* this results in the system locking up while trying to poll
* the command busy bit in the status register.
*/
if (sc->xl_flags & XL_FLAG_USE_MMIO)
DELAY(100000);
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "reset didn't complete\n");
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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
if (sc->xl_flags & 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
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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
CSR_WRITE_2(sc, XL_W2_RESET_OPTIONS,
CSR_READ_2(sc, 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
}
/*
* 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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_probe(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;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
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 BUS_PROBE_DEFAULT instead of 0.
2005-02-24 21:32:56 +00:00
return (BUS_PROBE_DEFAULT);
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++;
}
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_mediacheck(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
{
/*
* 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
/*
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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 {
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"bogus xcvr value in EEPROM (%x)\n", sc->xl_xcvr);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"choosing new default based on card type\n");
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
}
} 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;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"WARNING: no media options bits set in the media options register!!\n");
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"this could be a manufacturing defect in your adapter or system\n");
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"attempting to guess media type; you should probably consult your vendor\n");
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
}
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);
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_choose_xcvr(struct xl_softc *sc, int verbose)
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
{
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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"guessing 10BaseT transceiver\n");
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"guessing COMBO (AUI/BNC/TP)\n");
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "guessing TPC (BNC/TP)\n");
- 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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "guessing 10baseFL\n");
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 3c555 miniPCI chipset. Submitted by: johannes <johannes@paradise.net.nz> PR: 46603 MFC after: 3 days
2003-01-06 00:46:03 +00:00
case TC_DEVICEID_HURRICANE_555: /* 3c555 */
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 */
Correctly identify the 3C920B chipset; although it may ship with some Nforce2 boards, it's definitely not an Nvidia chip. Info from: Nvidia's Linux Network driver & pciids.sourceforge.net
2003-01-22 17:14:33 +00:00
case TC_DEVICEID_TORNADO_10_100BT_920B: /* 3c920B-EMB */
Add PCI and device ID's to if_xl to support: 3C920B-EMB-WNM Integrated Fast Ethernet Controller Submitter reports that the card appears to autonegotiate properly, and operate well with high levels of NFS traffic. PR: 75253 Submitted by: "Oleg V. Nauman" <oleg at reis dot zp dot ua> MFC after: 2 weeks
2005-01-03 19:22:37 +00:00
case TC_DEVICEID_TORNADO_10_100BT_920B_WNM: /* 3c920B-EMB-WNM */
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "guessing MII\n");
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"guessing 100baseT4/MII\n");
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"guessing 10/100 internal\n");
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)
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"guessing 10/100 plus BNC/AUI\n");
- 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:
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"unknown device ID: %x -- defaulting to 10baseT\n", 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
sc->xl_media = XL_MEDIAOPT_BT;
break;
}
}
/*
* Attach the interface. Allocate softc structures, do ifmedia
* setup and ethernet/BPF attach.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_attach(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];
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
u_int16_t xcvr[2];
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;
struct ifnet *ifp;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
int media;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
int unit, error = 0, rid, res;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
uint16_t did;
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,
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
MTX_DEF);
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
ifmedia_init(&sc->ifmedia, 0, xl_ifmedia_upd, xl_ifmedia_sts);
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
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
did = pci_get_device(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
sc->xl_flags = 0;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_555)
Ooops, use the full name of XL_FLAG_EEPROM_OFFSET_30 Noticed by: Niels Chr. Bank-Pedersen <ncbp@bank-pedersen.dk>
2003-01-06 01:55:12 +00:00
sc->xl_flags |= XL_FLAG_EEPROM_OFFSET_30 | XL_FLAG_PHYOK;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_556 ||
did == TC_DEVICEID_HURRICANE_556B)
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 |= 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;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_555 ||
did == TC_DEVICEID_HURRICANE_556)
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 |= XL_FLAG_8BITROM;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_556B)
Add some magic bits necessary to turn the transmitter on for some (newer) 556B chips. Requested & tested by: Dinesh Nambisan <dinesh@nambisan.net> Magic bits found by: Dave Dribin & Donald Becker MFC After: 3 days
2002-10-22 02:33:50 +00:00
sc->xl_flags |= XL_FLAG_NO_XCVR_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
The 575A doesn't have funcregs in memio. So don't claim that it does. This gets my 575A card probing. Card provided by: James Flemer MFC After: 3 days
2005-07-27 06:38:55 +00:00
if (did == TC_DEVICEID_HURRICANE_575B ||
did == TC_DEVICEID_HURRICANE_575C ||
did == TC_DEVICEID_HURRICANE_656B ||
did == TC_DEVICEID_TORNADO_656C)
sc->xl_flags |= XL_FLAG_FUNCREG;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_575A ||
did == TC_DEVICEID_HURRICANE_575B ||
did == TC_DEVICEID_HURRICANE_575C ||
did == TC_DEVICEID_HURRICANE_656B ||
did == TC_DEVICEID_TORNADO_656C)
The 575A doesn't have funcregs in memio. So don't claim that it does. This gets my 575A card probing. Card provided by: James Flemer MFC After: 3 days
2005-07-27 06:38:55 +00:00
sc->xl_flags |= XL_FLAG_PHYOK | XL_FLAG_EEPROM_OFFSET_30 |
XL_FLAG_8BITROM;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_656)
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
sc->xl_flags |= XL_FLAG_FUNCREG | XL_FLAG_PHYOK;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_575B)
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_INVERT_LED_PWR;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_575C)
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_INVERT_MII_PWR;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == 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;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if (did == TC_DEVICEID_HURRICANE_656 ||
did == TC_DEVICEID_HURRICANE_656B)
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
sc->xl_flags |= XL_FLAG_INVERT_MII_PWR |
XL_FLAG_INVERT_LED_PWR;
Add PCI and device ID's to if_xl to support: 3C920B-EMB-WNM Integrated Fast Ethernet Controller Submitter reports that the card appears to autonegotiate properly, and operate well with high levels of NFS traffic. PR: 75253 Submitted by: "Oleg V. Nauman" <oleg at reis dot zp dot ua> MFC after: 2 weeks
2005-01-03 19:22:37 +00:00
if (did == TC_DEVICEID_TORNADO_10_100BT_920B ||
did == TC_DEVICEID_TORNADO_10_100BT_920B_WNM)
Complete the support of the on-board xl(4) on nVidia nForce2 mobo's. Submitted by: Mikko S. Hyvarinen <morphy@morphy.iki.fi>
2003-01-19 04:47:54 +00:00
sc->xl_flags |= XL_FLAG_PHYOK;
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
switch (did) {
Boomerang 10/100BT (found in 2c905-TX) chips apparently suffer the same problems as their Hurricane 575* bretheren in that one could set the memory mapped port, but that has no effect. Add a quirk for this. # I'll have to see if I can dig up documentation on these parts to see # if there's someway software can know this other than a table...
2004-04-13 19:34:20 +00:00
case TC_DEVICEID_BOOMERANG_10_100BT: /* 3c905-TX */
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
case TC_DEVICEID_HURRICANE_575A:
case TC_DEVICEID_HURRICANE_575B:
case TC_DEVICEID_HURRICANE_575C:
sc->xl_flags |= XL_FLAG_NO_MMIO;
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
/*
* 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);
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
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +00:00
if ((sc->xl_flags & XL_FLAG_NO_MMIO) == 0) {
rid = XL_PCI_LOMEM;
res = SYS_RES_MEMORY;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->xl_res = bus_alloc_resource_any(dev, res, &rid, RF_ACTIVE);
- Avoid calling pci_get_device() more than once in a single function. - Provide a mechanism to prevent the use of MMIO. - Prevent the use of MMIO for all 3c575 cardbus cards.
2003-09-14 16:33:48 +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
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
if (sc->xl_res != NULL) {
sc->xl_flags |= XL_FLAG_USE_MMIO;
if (bootverbose)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "using memory mapped I/O\n");
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
} else {
rid = XL_PCI_LOIO;
res = SYS_RES_IOPORT;
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->xl_res = bus_alloc_resource_any(dev, res, &rid, RF_ACTIVE);
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
if (sc->xl_res == NULL) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "couldn't map ports/memory\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = ENXIO;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
goto fail;
}
if (bootverbose)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "using port I/O\n");
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->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;
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->xl_fres = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &rid,
RF_ACTIVE);
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) {
The 575A doesn't have funcregs in memio. So don't claim that it does. This gets my 575A card probing. Card provided by: James Flemer MFC After: 3 days
2005-07-27 06:38:55 +00:00
device_printf(dev, "couldn't map funcreg memory\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = ENXIO;
goto fail;
Add 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_ftag = rman_get_bustag(sc->xl_fres);
sc->xl_fhandle = rman_get_bushandle(sc->xl_fres);
}
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
/* Allocate interrupt */
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;
Convert callers to the new bus_alloc_resource_any(9) API. Submitted by: Mark Santcroos <marks@ripe.net> Reviewed by: imp, dfr, bde
2004-03-17 17:50:55 +00:00
sc->xl_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &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
RF_SHAREABLE | RF_ACTIVE);
if (sc->xl_irq == NULL) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "couldn't map interrupt\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = ENXIO;
goto fail;
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
}
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
/* Initialize interface name. */
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc->xl_ifp = if_alloc(IFT_ETHER);
if (ifp == NULL) {
device_printf(dev, "can not if_alloc()\n");
error = ENOSPC;
goto fail;
}
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
ifp->if_softc = sc;
if_initname(ifp, device_get_name(dev), 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
/* Reset the adapter. */
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +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
xl_reset(sc);
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +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
/*
* Get station address from the EEPROM.
*/
if (xl_read_eeprom(sc, (caddr_t)&eaddr, XL_EE_OEM_ADR0, 3, 1)) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "failed to read station address\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = ENXIO;
goto fail;
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_unit = unit;
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
callout_init_mtx(&sc->xl_stat_callout, &sc->xl_mtx, 0);
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
TASK_INIT(&sc->xl_task, 0, xl_rxeof_task, 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 xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
/*
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
* Now allocate a tag for the DMA descriptor lists and a chunk
* of DMA-able memory based on the tag. Also obtain the DMA
* addresses of the RX and TX ring, which we'll need later.
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
* All of our lists are allocated as a contiguous block
* of memory.
*/
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
error = bus_dma_tag_create(NULL, 8, 0,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
XL_RX_LIST_SZ, 1, XL_RX_LIST_SZ, 0, NULL, NULL,
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
&sc->xl_ldata.xl_rx_tag);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "failed to allocate rx dma tag\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
}
error = bus_dmamem_alloc(sc->xl_ldata.xl_rx_tag,
Use the BUS_DMA_ZERO flag instead of bzero()'ing DMA memory.
2003-07-27 13:56:03 +00:00
(void **)&sc->xl_ldata.xl_rx_list, BUS_DMA_NOWAIT | BUS_DMA_ZERO,
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
&sc->xl_ldata.xl_rx_dmamap);
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "no memory for rx list buffers!\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
bus_dma_tag_destroy(sc->xl_ldata.xl_rx_tag);
sc->xl_ldata.xl_rx_tag = NULL;
goto fail;
}
error = bus_dmamap_load(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap, sc->xl_ldata.xl_rx_list,
XL_RX_LIST_SZ, xl_dma_map_addr,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
&sc->xl_ldata.xl_rx_dmaaddr, BUS_DMA_NOWAIT);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "cannot get dma address of the rx ring!\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
bus_dmamem_free(sc->xl_ldata.xl_rx_tag, sc->xl_ldata.xl_rx_list,
sc->xl_ldata.xl_rx_dmamap);
bus_dma_tag_destroy(sc->xl_ldata.xl_rx_tag);
sc->xl_ldata.xl_rx_tag = NULL;
goto fail;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
}
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
error = bus_dma_tag_create(NULL, 8, 0,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
XL_TX_LIST_SZ, 1, XL_TX_LIST_SZ, 0, NULL, NULL,
Mega busdma API commit. Add two new arguments to bus_dma_tag_create(): lockfunc and lockfuncarg. Lockfunc allows a driver to provide a function for managing its locking semantics while using busdma. At the moment, this is used for the asynchronous busdma_swi and callback mechanism. Two lockfunc implementations are provided: busdma_lock_mutex() performs standard mutex operations on the mutex that is specified from lockfuncarg. dftl_lock() is a panic implementation and is defaulted to when NULL, NULL are passed to bus_dma_tag_create(). The only time that NULL, NULL should ever be used is when the driver ensures that bus_dmamap_load() will not be deferred. Drivers that do not provide their own locking can pass busdma_lock_mutex,&Giant args in order to preserve the former behaviour. sparc64 and powerpc do not provide real busdma_swi functions, so this is largely a noop on those platforms. The busdma_swi on is64 is not properly locked yet, so warnings will be emitted on this platform when busdma callback deferrals happen. If anyone gets panics or warnings from dflt_lock() being called, please let me know right away. Reviewed by: tmm, gibbs
2003-07-01 15:52:06 +00:00
&sc->xl_ldata.xl_tx_tag);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "failed to allocate tx dma tag\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
}
error = bus_dmamem_alloc(sc->xl_ldata.xl_tx_tag,
Use the BUS_DMA_ZERO flag instead of bzero()'ing DMA memory.
2003-07-27 13:56:03 +00:00
(void **)&sc->xl_ldata.xl_tx_list, BUS_DMA_NOWAIT | BUS_DMA_ZERO,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
&sc->xl_ldata.xl_tx_dmamap);
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "no memory for list buffers!\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
bus_dma_tag_destroy(sc->xl_ldata.xl_tx_tag);
sc->xl_ldata.xl_tx_tag = NULL;
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
}
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
error = bus_dmamap_load(sc->xl_ldata.xl_tx_tag,
sc->xl_ldata.xl_tx_dmamap, sc->xl_ldata.xl_tx_list,
XL_TX_LIST_SZ, xl_dma_map_addr,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
&sc->xl_ldata.xl_tx_dmaaddr, BUS_DMA_NOWAIT);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "cannot get dma address of the tx ring!\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
bus_dmamem_free(sc->xl_ldata.xl_tx_tag, sc->xl_ldata.xl_tx_list,
sc->xl_ldata.xl_tx_dmamap);
bus_dma_tag_destroy(sc->xl_ldata.xl_tx_tag);
sc->xl_ldata.xl_tx_tag = NULL;
goto fail;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
}
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
/*
* Allocate a DMA tag for the mapping of mbufs.
*/
error = bus_dma_tag_create(NULL, 1, 0,
Correct maxsize/maxsegsz parameters to bus_dma_tag_create().
2003-04-09 15:05:19 +00:00
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR, NULL, NULL,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
MCLBYTES * XL_MAXFRAGS, XL_MAXFRAGS, MCLBYTES, 0, NULL,
NULL, &sc->xl_mtag);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "failed to allocate mbuf dma tag\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +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
Fix panic on sparc64 introduced in my last commit. I really wish the busdma APIs were more consistent accross architectures. We should probably move all the other DMA map creations in xl_attach() where we can really handle them failing, since xl_init() is void and shouldn't fail. Pointy hat to: mux Tested by: Anders Andersson <anders@hack.org>
2003-02-19 18:33:29 +00:00
/* We need a spare DMA map for the RX ring. */
error = bus_dmamap_create(sc->xl_mtag, 0, &sc->xl_tmpmap);
if (error)
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
Fix panic on sparc64 introduced in my last commit. I really wish the busdma APIs were more consistent accross architectures. We should probably move all the other DMA map creations in xl_attach() where we can really handle them failing, since xl_init() is void and shouldn't fail. Pointy hat to: mux Tested by: Anders Andersson <anders@hack.org>
2003-02-19 18:33:29 +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
/*
* Figure out the card type. 3c905B adapters have the
* 'supportsNoTxLength' bit set in the capabilities
* word in the EEPROM.
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
* Note: my 3c575C cardbus card lies. It returns a value
* of 0x1578 for its capabilities word, which is somewhat
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
* nonsensical. Another way to distinguish a 3c90x chip
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
* from a 3c90xB/C chip is to check for the 'supportsLargePackets'
* bit. This will only be set for 3c90x boomerage chips.
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_read_eeprom(sc, (caddr_t)&sc->xl_caps, XL_EE_CAPS, 1, 0);
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
if (sc->xl_caps & XL_CAPS_NO_TXLENGTH ||
!(sc->xl_caps & XL_CAPS_LARGE_PKTS))
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_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->if_mtu = ETHERMTU;
ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST;
ifp->if_ioctl = xl_ioctl;
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
ifp->if_capabilities = IFCAP_VLAN_MTU;
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) {
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;
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +00:00
#ifdef XL905B_TXCSUM_BROKEN
ifp->if_capabilities |= IFCAP_RXCSUM;
#else
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
ifp->if_capabilities |= IFCAP_HWCSUM;
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +00:00
#endif
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
}
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
ifp->if_capenable = ifp->if_capabilities;
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
ifp->if_capabilities |= IFCAP_POLLING;
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#endif
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +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;
Bring in the first chunk of altq driver modifications. This covers the following drivers: bfe(4), em(4), fxp(4), lnc(4), tun(4), de(4) rl(4), sis(4) and xl(4) More patches are pending on: http://peoples.freebsd.org/~mlaier/ Please take a look and tell me if "your" driver is missing, so I can fix this. Tested-by: many No-objection: -current, -net
2004-07-02 12:16:02 +00:00
IFQ_SET_MAXLEN(&ifp->if_snd, XL_TX_LIST_CNT - 1);
ifp->if_snd.ifq_drv_maxlen = XL_TX_LIST_CNT - 1;
IFQ_SET_READY(&ifp->if_snd);
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)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "media options word: %x\n", sc->xl_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
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
xl_read_eeprom(sc, (char *)&xcvr, XL_EE_ICFG_0, 2, 0);
sc->xl_xcvr = xcvr[0] | xcvr[1] << 16;
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_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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
if (sc->xl_media & XL_MEDIAOPT_MII ||
sc->xl_media & XL_MEDIAOPT_BTX ||
sc->xl_media & XL_MEDIAOPT_BT4) {
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 (bootverbose)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found MII/AUTO\n");
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
xl_setcfg(sc);
if (mii_phy_probe(dev, &sc->xl_miibus,
xl_ifmedia_upd, xl_ifmedia_sts)) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "no PHY found!\n");
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
error = ENXIO;
goto fail;
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.
*/
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
if (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
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.
*/
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)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found 10baseT\n");
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
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)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found 10baseFL\n");
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
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)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found AUI\n");
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
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)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found BNC\n");
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_add(&sc->ifmedia, IFM_ETHER|IFM_10_2, 0, NULL);
}
if (sc->xl_media & XL_MEDIAOPT_BFX) {
if (bootverbose)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "found 100baseFX\n");
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_baudrate = 100000000;
ifmedia_add(&sc->ifmedia, IFM_ETHER|IFM_100_FX, 0, NULL);
}
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
media = IFM_ETHER|IFM_100_TX|IFM_FDX;
xl_choose_media(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
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:
Add some magic bits necessary to turn the transmitter on for some (newer) 556B chips. Requested & tested by: Dinesh Nambisan <dinesh@nambisan.net> Magic bits found by: Dave Dribin & Donald Becker MFC After: 3 days
2002-10-22 02:33:50 +00:00
if (sc->xl_flags & XL_FLAG_NO_XCVR_PWR) {
XL_SEL_WIN(0);
CSR_WRITE_2(sc, XL_W0_MFG_ID, XL_NO_XCVR_PWR_MAGICBITS);
}
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
*/
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
ether_ifattach(ifp, eaddr);
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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
error = bus_setup_intr(dev, sc->xl_irq, INTR_TYPE_NET | INTR_MPSAFE,
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
xl_intr, sc, &sc->xl_intrhand);
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
device_printf(dev, "couldn't set up irq\n");
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
ether_ifdetach(ifp);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
goto fail;
Make sure to free the correct resources when the card fails to attach properly. (Broken in the previous commit.) Noticed by: "Niels Chr. Bank-Pedersen" <ncbp@bank-pedersen.dk>
2003-03-19 01:48:14 +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
fail:
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (error)
xl_detach(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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/*
* Choose a default media.
* XXX This is a leaf function only called by xl_attach() and
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
* acquires/releases the non-recursible driver mutex to
* satisfy lock assertions.
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
*/
static void
xl_choose_media(struct xl_softc *sc, int *media)
{
XL_LOCK(sc);
switch (sc->xl_xcvr) {
case XL_XCVR_10BT:
*media = IFM_ETHER|IFM_10_T;
xl_setmode(sc, *media);
break;
case XL_XCVR_AUI:
if (sc->xl_type == XL_TYPE_905B &&
sc->xl_media == XL_MEDIAOPT_10FL) {
*media = IFM_ETHER|IFM_10_FL;
xl_setmode(sc, *media);
} else {
*media = IFM_ETHER|IFM_10_5;
xl_setmode(sc, *media);
}
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:
/* Chosen by miibus */
break;
case XL_XCVR_100BFX:
*media = IFM_ETHER|IFM_100_FX;
break;
default:
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "unknown XCVR type: %d\n",
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
sc->xl_xcvr);
/*
* This will probably be wrong, but it prevents
* the ifmedia code from panicking.
*/
*media = IFM_ETHER|IFM_10_T;
break;
}
XL_UNLOCK(sc);
}
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
/*
* Shutdown hardware and free up resources. This can be called any
* time after the mutex has been initialized. It is called in both
* the error case in attach and the normal detach case so it needs
* to be careful about only freeing resources that have actually been
* allocated.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_detach(device_t 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
{
struct xl_softc *sc;
struct ifnet *ifp;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
int rid, 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
sc = device_get_softc(dev);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc->xl_ifp;
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
Add missing ()'s so that these drivers all compile again. Noticed by: jake Tested on: i386 (compile)
2003-03-31 20:22:00 +00:00
KASSERT(mtx_initialized(&sc->xl_mtx), ("xl mutex not initialized"));
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
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#ifdef DEVICE_POLLING
if (ifp->if_capenable & IFCAP_POLLING)
ether_poll_deregister(ifp);
#endif
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
if (sc->xl_flags & XL_FLAG_USE_MMIO) {
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
rid = XL_PCI_LOMEM;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
res = SYS_RES_MEMORY;
} else {
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
rid = XL_PCI_LOIO;
Instead of relying on a compile time define to determine whether the xl driver should use port or memory based IO, determine it dynamically at runtime, preferring MMIO where possible. This helps us support newer arches which dislike port based access better. Tested on i386 & sparc64, with 3c900, 905, 905b, and 905C cards. (in varying combinations by both jake and myself)
2003-03-18 06:29:51 +00:00
res = SYS_RES_IOPORT;
}
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
/* These should only be active if attach succeeded */
Use newly minted device_is_attached rather than device_is_alive to see if attach succeeded. device_is_alive just tells us that probe succeeded. Since we were using it to do things like detach net interfaces, this caused problems when there were errors in the attach routine. Symptoms of problem reported by: martin blapp
2003-04-21 18:34:04 +00:00
if (device_is_attached(dev)) {
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_LOCK(sc);
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
xl_reset(sc);
xl_stop(sc);
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_UNLOCK(sc);
taskqueue_drain(taskqueue_swi, &sc->xl_task);
callout_drain(&sc->xl_stat_callout);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
ether_ifdetach(ifp);
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
}
Fix "struct ifnet" leaks when attach() fails in the middle, e.g. when mii_phy_probe() or bus_setup_intr() fails. For drivers that call their detach() in this case, call if_free() there to cover this case too.
2005-09-16 11:11:51 +00:00
if (ifp)
if_free(ifp);
Revise attach/detach resource cleanup - Unconditionally call *_stop() if device is in the tree. This is to prevent callouts from happening after the device is gone. Checks for bus_child_present() should be added in the future to keep from touching potentially non-existent hardware in *_detach(). Found by iedowse@. - Always check for and free miibus children, even if the device is not in the tree since some failure cases could have gotten here. - Call ether_ifdetach() in the irq setup failure case - ti(4), xl(4): move ifmedia_init() calls to the beginning of attach so that ifmedia_removeall() can be unconditionally called on detach. There is no way to detect whether ifmedia has been initialized without using a separate variable (as tl(4) does). - Add comments to indicate assumptions of code path
2003-04-17 20:32:06 +00:00
if (sc->xl_miibus)
device_delete_child(dev, sc->xl_miibus);
bus_generic_detach(dev);
ifmedia_removeall(&sc->ifmedia);
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
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (sc->xl_intrhand)
bus_teardown_intr(dev, sc->xl_irq, sc->xl_intrhand);
if (sc->xl_irq)
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);
Clean up locking and resource management for pci/if_* - Remove locking of the softc in the attach method, instead depending on bus_setup_intr being at the end of attach (delaying interrupt enable until after ether_ifattach is called) - Call *_detach directly in the error case of attach, depending on checking in detach to only free resources that were allocated. This puts all resource freeing in one place, avoiding thinkos that lead to memory leaks. - Add bus_child_present check to calls to *_stop in the detach method to be sure hw is present before touching its registers. - Remove bzero softc calls since device_t should do this for us. - dc: move interrupt allocation back where it was before. It was unnecessary to move it. This reverts part of 1.88 - rl: move irq allocation before ether_ifattach. Problems might have been caused by allocating the irq after enabling interrupts on the card. - rl: call rl_stop before ether_ifdetach - sf: call sf_stop before ether_ifdetach - sis: add missed free of sis_tag - sis: check errors from tag creation - sis: move dmamem_alloc and dmamap_load to happen at same time as tag creation - sk: remove duplicate initialization of sk_dev - ste: add missed bus_generic_detach - ti: call ti_stop before ether_ifdetach - ti: add missed error setting in ti_rdata alloc failure - vr: add missed error setting in I/O, memory mapping cases - xl: add missed error setting in I/O, memory mapping cases - xl: remove multi-level goto on attach failure - xl: move dmamem_alloc and dmamap_load to happen at same time as tag creation - Calls to free(9) are unconditional because it is valid to call free with a null pointer. Reviewed by: imp, mdodd
2003-03-31 17:29:43 +00:00
if (sc->xl_res)
bus_release_resource(dev, res, rid, sc->xl_res);
if (sc->xl_mtag) {
bus_dmamap_destroy(sc->xl_mtag, sc->xl_tmpmap);
bus_dma_tag_destroy(sc->xl_mtag);
}
if (sc->xl_ldata.xl_rx_tag) {
bus_dmamap_unload(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap);
bus_dmamem_free(sc->xl_ldata.xl_rx_tag, sc->xl_ldata.xl_rx_list,
sc->xl_ldata.xl_rx_dmamap);
bus_dma_tag_destroy(sc->xl_ldata.xl_rx_tag);
}
if (sc->xl_ldata.xl_tx_tag) {
bus_dmamap_unload(sc->xl_ldata.xl_tx_tag,
sc->xl_ldata.xl_tx_dmamap);
bus_dmamem_free(sc->xl_ldata.xl_tx_tag, sc->xl_ldata.xl_tx_list,
sc->xl_ldata.xl_tx_dmamap);
bus_dma_tag_destroy(sc->xl_ldata.xl_tx_tag);
}
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
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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_list_tx_init(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_chain_data *cd;
struct xl_list_data *ld;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
int error, 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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
cd = &sc->xl_cdata;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
ld = &sc->xl_ldata;
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 < XL_TX_LIST_CNT; i++) {
cd->xl_tx_chain[i].xl_ptr = &ld->xl_tx_list[i];
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = bus_dmamap_create(sc->xl_mtag, 0,
&cd->xl_tx_chain[i].xl_map);
if (error)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
cd->xl_tx_chain[i].xl_phys = ld->xl_tx_dmaaddr +
i * sizeof(struct xl_list);
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_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;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(ld->xl_tx_tag, ld->xl_tx_dmamap, BUS_DMASYNC_PREWRITE);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
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.
*/
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_list_tx_init_90xB(struct xl_softc *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
{
struct xl_chain_data *cd;
struct xl_list_data *ld;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
int error, i;
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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
cd = &sc->xl_cdata;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
ld = &sc->xl_ldata;
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
for (i = 0; i < XL_TX_LIST_CNT; i++) {
cd->xl_tx_chain[i].xl_ptr = &ld->xl_tx_list[i];
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = bus_dmamap_create(sc->xl_mtag, 0,
&cd->xl_tx_chain[i].xl_map);
if (error)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
cd->xl_tx_chain[i].xl_phys = ld->xl_tx_dmaaddr +
i * sizeof(struct xl_list);
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 (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];
}
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bzero(ld->xl_tx_list, XL_TX_LIST_SZ);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
ld->xl_tx_list[0].xl_status = htole32(XL_TXSTAT_EMPTY);
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
cd->xl_tx_prod = 1;
cd->xl_tx_cons = 1;
cd->xl_tx_cnt = 0;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(ld->xl_tx_tag, ld->xl_tx_dmamap, BUS_DMASYNC_PREWRITE);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
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
}
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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_list_rx_init(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_chain_data *cd;
struct xl_list_data *ld;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
int error, i, next;
u_int32_t nextptr;
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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
cd = &sc->xl_cdata;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
ld = &sc->xl_ldata;
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 < XL_RX_LIST_CNT; i++) {
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
cd->xl_rx_chain[i].xl_ptr = &ld->xl_rx_list[i];
error = bus_dmamap_create(sc->xl_mtag, 0,
&cd->xl_rx_chain[i].xl_map);
if (error)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = xl_newbuf(sc, &cd->xl_rx_chain[i]);
if (error)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
if (i == (XL_RX_LIST_CNT - 1))
next = 0;
else
next = i + 1;
nextptr = ld->xl_rx_dmaaddr +
next * sizeof(struct xl_list_onefrag);
cd->xl_rx_chain[i].xl_next = &cd->xl_rx_chain[next];
ld->xl_rx_list[i].xl_next = htole32(nextptr);
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(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(ld->xl_rx_tag, ld->xl_rx_dmamap, BUS_DMASYNC_PREWRITE);
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];
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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 an RX descriptor and attach an MBUF cluster.
- Fix mbuf leak when we successfully allocate a new mbuf but fail to bus_dmamap_load() it. - Make it so reusing mbufs when we can't allocate (or map) new ones actually works. We were previously trying to reuse a mbuf which was already bus_dmamap_unload()'ed. Reviewed by: silby
2003-02-18 18:50:54 +00:00
* If we fail to do so, we need to leave the old mbuf and
* the old DMA map untouched so that it can be reused.
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
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_newbuf(struct xl_softc *sc, struct xl_chain_onefrag *c)
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 mbuf *m_new = NULL;
- Fix mbuf leak when we successfully allocate a new mbuf but fail to bus_dmamap_load() it. - Make it so reusing mbufs when we can't allocate (or map) new ones actually works. We were previously trying to reuse a mbuf which was already bus_dmamap_unload()'ed. Reviewed by: silby
2003-02-18 18:50:54 +00:00
bus_dmamap_t map;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
int error;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
u_int32_t baddr;
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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(sc);
Make xl use m_getcl() to allocate an mbuf and a cluster in one shot, as opposed to one after the other. This is faster in both -CURRENT and -STABLE. Additionally, there is less code duplication for error-checking. One thing to note is that this code seems to return(1) when no buffers are available; perhaps ENOBUFS should be the correct return value? Partially submitted & tested by: Hiten Pandya <hiten@unixdaemons.com> MFC after: 1 week
2003-02-22 14:46:31 +00:00
m_new = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
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)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
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
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
Use bus_dmamap_load_mbuf() instead of bus_dmamap_load() for the RX part of this driver too. It's better since the code wasn't dealing with bus_dmamap_load() returning EINPROGRESS, and this can't happen with bus_dmamap_load_mbuf(). Submitted by: jake
2003-04-05 23:24:23 +00:00
error = bus_dmamap_load_mbuf(sc->xl_mtag, sc->xl_tmpmap, m_new,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
xl_dma_map_rxbuf, &baddr, BUS_DMA_NOWAIT);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
m_freem(m_new);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp, "can't map mbuf (error %d)\n",
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
error);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
}
- Fix mbuf leak when we successfully allocate a new mbuf but fail to bus_dmamap_load() it. - Make it so reusing mbufs when we can't allocate (or map) new ones actually works. We were previously trying to reuse a mbuf which was already bus_dmamap_unload()'ed. Reviewed by: silby
2003-02-18 18:50:54 +00:00
bus_dmamap_unload(sc->xl_mtag, c->xl_map);
map = c->xl_map;
c->xl_map = sc->xl_tmpmap;
sc->xl_tmpmap = map;
c->xl_mbuf = m_new;
c->xl_ptr->xl_frag.xl_len = htole32(m_new->m_len | XL_LAST_FRAG);
c->xl_ptr->xl_status = 0;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
c->xl_ptr->xl_frag.xl_addr = htole32(baddr);
bus_dmamap_sync(sc->xl_mtag, c->xl_map, BUS_DMASYNC_PREREAD);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_rx_resync(struct xl_softc *sc)
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
{
struct xl_chain_onefrag *pos;
int i;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
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)
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
sc->xl_cdata.xl_rx_head = pos;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (EAGAIN);
It appears that under certain circumstances that I still can't quite pin down, the dc driver and receiver can fall out of sync with one another, resulting in a condition where the chip continues to receive packets but the driver never notices. Normally, the receive handler checks each descriptor starting from the current producer index to see if the chip has relinquished ownership, indicating that a packet has been received. The driver hands the packet off to ether_input() and then prepares the descriptor to receive another frame before moving on to the next descriptor in the ring. But sometimes, the chip appears to skip a descriptor. This leaves the driver testing the status word in a descriptor that never gets updated. The driver still gets "RX done" interrupts but never advances further into the RX ring, until the ring fills up and the chip interrupts again to signal an error condition. Sometimes, the driver will remain in this desynchronized state, resulting in spotty performance until the interface is reset. Fortunately, it's fairly simple to detect this condition: if we call the rxeof routine but the number of received packets doesn't increase, we suspect that there could be a problem. In this case, we call a new routine called dc_rx_resync(), which scans ahead in the RX ring to see if there's a frame waiting for us somewhere beyond that the driver thinks is the current producer index. If it finds one, it bumps up the index and calls the rxeof handler again to snarf up the packet and bring the driver back in sync with the chip. (It may actually do this several times in the event that there's more than one "hole" in the ring.) So far the only card supported by if_dc which has exhibited this problem is a LinkSys LNE100TX v2.0 (82c115 PNIC II), and it only seems to happen on one particular system, however the fix is general enough and has low enough overhead that we may as well apply it for all supported chipsets. I also implemented the same fix for the 3Com xl driver, which is apparently vulnerable to the same problem. Problem originally noted and patch tested by: Matt Dillon
2000-01-03 15:28:47 +00:00
}
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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_rxeof(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
{
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
struct mbuf *m;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
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
Drop the driver lock around calls to if_input to avoid a LOR when the packets are immediately returned for sending (e.g. when bridging or packet forwarding). There are more efficient ways to do this but for now use the least intrusive approach. Reviewed by: imp, rwatson
2003-11-14 19:00:32 +00:00
XL_LOCK_ASSERT(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
again:
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_rx_tag, sc->xl_ldata.xl_rx_dmamap,
BUS_DMASYNC_POSTREAD);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while ((rxstat = le32toh(sc->xl_cdata.xl_rx_head->xl_ptr->xl_status))) {
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING) {
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
if (sc->rxcycles <= 0)
break;
sc->rxcycles--;
}
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#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
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;
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
total_len = rxstat & XL_RXSTAT_LENMASK;
/*
* Since we have told the chip to allow large frames,
* we need to trap giant frame errors in software. We allow
* a little more than the normal frame size to account for
* frames with VLAN tags.
*/
if (total_len > XL_MAX_FRAMELEN)
rxstat |= (XL_RXSTAT_UP_ERROR|XL_RXSTAT_OVERSIZE);
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
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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
* comes up in the ring.
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 (rxstat & XL_RXSTAT_UP_ERROR) {
ifp->if_ierrors++;
cur_rx->xl_ptr->xl_status = 0;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap, BUS_DMASYNC_PREWRITE);
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
continue;
}
/*
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
* If the error bit was not set, the upload complete
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
* bit should be set which means we have a valid packet.
* If not, something truly strange has happened.
*/
if (!(rxstat & XL_RXSTAT_UP_CMPLT)) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp,
"bad receive status -- packet dropped\n");
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;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap, BUS_DMASYNC_PREWRITE);
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
continue;
}
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
/* No errors; receive the packet. */
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(sc->xl_mtag, cur_rx->xl_map,
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
BUS_DMASYNC_POSTREAD);
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
m = cur_rx->xl_mbuf;
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.
*/
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (xl_newbuf(sc, cur_rx)) {
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
ifp->if_ierrors++;
cur_rx->xl_ptr->xl_status = 0;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap, BUS_DMASYNC_PREWRITE);
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
continue;
}
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_rx_tag,
sc->xl_ldata.xl_rx_dmamap, BUS_DMASYNC_PREWRITE);
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
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
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
if (ifp->if_capenable & IFCAP_RXCSUM) {
Implement TCP/IP checksum off-loading on receive. Announce rxcsum capabilities. Reviewed by: wpaul
2001-09-23 05:13:12 +00:00
/* 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;
}
}
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
Drop the driver lock around calls to if_input to avoid a LOR when the packets are immediately returned for sending (e.g. when bridging or packet forwarding). There are more efficient ways to do this but for now use the least intrusive approach. Reviewed by: imp, rwatson
2003-11-14 19:00:32 +00:00
XL_UNLOCK(sc);
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
(*ifp->if_input)(ifp, m);
Drop the driver lock around calls to if_input to avoid a LOR when the packets are immediately returned for sending (e.g. when bridging or packet forwarding). There are more efficient ways to do this but for now use the least intrusive approach. Reviewed by: imp, rwatson
2003-11-14 19:00:32 +00:00
XL_LOCK(sc);
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
/*
* If we are running from the taskqueue, the interface
* might have been stopped while we were passing the last
* packet up the network stack.
*/
if (!(ifp->if_drv_flags & IFF_DRV_RUNNING))
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
}
/*
* 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);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
CSR_WRITE_4(sc, XL_UPLIST_PTR, sc->xl_ldata.xl_rx_dmaaddr);
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_rx_head = &sc->xl_cdata.xl_rx_chain[0];
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_UP_UNSTALL);
goto again;
}
}
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
/*
* Taskqueue wrapper for xl_rxeof().
*/
static void
xl_rxeof_task(void *arg, int pending)
{
struct xl_softc *sc = (struct xl_softc *)arg;
NET_LOCK_GIANT() when entering network code. Pointy hat to: glebius Reported by: rodrigc
2005-07-13 14:06:13 +00:00
NET_LOCK_GIANT();
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
XL_LOCK(sc);
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
if (sc->xl_ifp->if_drv_flags & IFF_DRV_RUNNING)
xl_rxeof(sc);
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
XL_UNLOCK(sc);
NET_LOCK_GIANT() when entering network code. Pointy hat to: glebius Reported by: rodrigc
2005-07-13 14:06:13 +00:00
NET_UNLOCK_GIANT();
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +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
/*
* A frame was downloaded to the chip. It's safe for us to clean up
* the list buffers.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_txeof(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_chain *cur_tx;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
/* 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.
*/
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while (sc->xl_cdata.xl_tx_head != 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
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;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(sc->xl_mtag, cur_tx->xl_map,
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
BUS_DMASYNC_POSTWRITE);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_unload(sc->xl_mtag, cur_tx->xl_map);
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
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) {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
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 = 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,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
sc->xl_cdata.xl_tx_head->xl_phys);
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
}
}
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_txeof_90xB(struct xl_softc *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
{
struct xl_chain *cur_tx = NULL;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_ifp;
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
int idx;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(sc);
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_ldata.xl_tx_tag, sc->xl_ldata.xl_tx_dmamap,
BUS_DMASYNC_POSTREAD);
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_cons;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while (idx != sc->xl_cdata.xl_tx_prod) {
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
cur_tx = &sc->xl_cdata.xl_tx_chain[idx];
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
if (!(le32toh(cur_tx->xl_ptr->xl_status) &
XL_TXSTAT_DL_COMPLETE))
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
break;
if (cur_tx->xl_mbuf != NULL) {
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_sync(sc->xl_mtag, cur_tx->xl_map,
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
BUS_DMASYNC_POSTWRITE);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_unload(sc->xl_mtag, cur_tx->xl_map);
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_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)
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
Add 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
/*
* 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.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_txeoc(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
{
u_int8_t txstat;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(sc);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while ((txstat = CSR_READ_1(sc, XL_TX_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
if (txstat & XL_TXSTATUS_UNDERRUN ||
txstat & XL_TXSTATUS_JABBER ||
txstat & XL_TXSTATUS_RECLAIM) {
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"transmission error: %x\n", txstat);
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_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;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
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,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
sc->xl_cdata.xl_tx_head->xl_phys);
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;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
if_printf(sc->xl_ifp,
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
"tx underrun, increasing tx start threshold to %d bytes\n", sc->xl_tx_thresh);
Dynamically increase TX start threshold if TX underruns are detected.
1999-05-05 17:05:07 +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,
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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
* to clear this interrupt/error and advance to the 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
*/
CSR_WRITE_1(sc, XL_TX_STATUS, 0x01);
}
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_intr(void *arg)
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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *sc = arg;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
u_int16_t status;
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
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
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING) {
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
XL_UNLOCK(sc);
return;
}
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#endif
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while ((status = CSR_READ_2(sc, XL_STATUS)) & XL_INTRS &&
status != 0xFFFF) {
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) {
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
int curpkts;
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
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);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_init_locked(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_STATSOFLOW) {
sc->xl_stats_no_timeout = 1;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_stats_update_locked(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
sc->xl_stats_no_timeout = 0;
}
}
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) {
if (sc->xl_type == XL_TYPE_905B)
xl_start_90xB_locked(ifp);
else
xl_start_locked(ifp);
}
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
XL_UNLOCK(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
}
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
static void
xl_poll(struct ifnet *ifp, enum poll_cmd cmd, int count)
{
struct xl_softc *sc = ifp->if_softc;
XL_LOCK(sc);
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
xl_poll_locked(ifp, cmd, count);
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
XL_UNLOCK(sc);
}
static void
xl_poll_locked(struct ifnet *ifp, enum poll_cmd cmd, int count)
{
struct xl_softc *sc = ifp->if_softc;
XL_LOCK_ASSERT(sc);
sc->rxcycles = count;
xl_rxeof(sc);
if (sc->xl_type == XL_TYPE_905B)
xl_txeof_90xB(sc);
else
xl_txeof(sc);
if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) {
if (sc->xl_type == XL_TYPE_905B)
xl_start_90xB_locked(ifp);
else
xl_start_locked(ifp);
}
if (cmd == POLL_AND_CHECK_STATUS) {
u_int16_t status;
status = CSR_READ_2(sc, XL_STATUS);
if (status & XL_INTRS && status != 0xFFFF) {
CSR_WRITE_2(sc, XL_COMMAND,
XL_CMD_INTR_ACK|(status & XL_INTRS));
if (status & XL_STAT_TX_COMPLETE) {
ifp->if_oerrors++;
xl_txeoc(sc);
}
if (status & XL_STAT_ADFAIL) {
xl_reset(sc);
xl_init_locked(sc);
}
if (status & XL_STAT_STATSOFLOW) {
sc->xl_stats_no_timeout = 1;
xl_stats_update_locked(sc);
sc->xl_stats_no_timeout = 0;
}
}
}
}
#endif /* DEVICE_POLLING */
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/*
* XXX: This is an entry point for callout which needs to take the lock.
*/
static void
xl_stats_update(void *xsc)
{
struct xl_softc *sc = xsc;
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 fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_LOCK_ASSERT(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_stats_update_locked(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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_stats_update_locked(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
{
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
bzero((char *)&xl_stats, sizeof(struct xl_stats));
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 +
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
xl_stats.xl_tx_single_collision + xl_stats.xl_tx_late_collision;
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
/*
* 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)
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
callout_reset(&sc->xl_stat_callout, hz, xl_stats_update, 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
}
/*
* Encapsulate an mbuf chain in a descriptor by coupling the mbuf data
* pointers to the fragment pointers.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_encap(struct xl_softc *sc, struct xl_chain *c, struct mbuf *m_head)
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(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
int error;
Fix up the long mbuf chain recovery code to use m_defrag; the old code messed up on B & C chipsets because it lost the packet header and therefore the flag indicating the need for hardware checksums. MFC after: 2 weeks
2003-03-29 06:34:25 +00:00
u_int32_t status;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_ifp;
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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
/*
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
* Start packing the mbufs in this chain into
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 fragment pointers. Stop when we run out
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
* of fragments or hit the end of the mbuf chain.
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(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = bus_dmamap_load_mbuf(sc->xl_mtag, c->xl_map, m_head,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
xl_dma_map_txbuf, c->xl_ptr, BUS_DMA_NOWAIT);
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
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
if (error && error != EFBIG) {
m_freem(m_head);
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "can't map mbuf (error %d)\n", error);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (1);
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +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
/*
* 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.
*/
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
Make xl use m_getcl() to allocate an mbuf and a cluster in one shot, as opposed to one after the other. This is faster in both -CURRENT and -STABLE. Additionally, there is less code duplication for error-checking. One thing to note is that this code seems to return(1) when no buffers are available; perhaps ENOBUFS should be the correct return value? Partially submitted & tested by: Hiten Pandya <hiten@unixdaemons.com> MFC after: 1 week
2003-02-22 14:46:31 +00:00
struct mbuf *m_new;
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 up the long mbuf chain recovery code to use m_defrag; the old code messed up on B & C chipsets because it lost the packet header and therefore the flag indicating the need for hardware checksums. MFC after: 2 weeks
2003-03-29 06:34:25 +00:00
m_new = m_defrag(m_head, M_DONTWAIT);
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 (m_new == NULL) {
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
m_freem(m_head);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (1);
Fix up the long mbuf chain recovery code to use m_defrag; the old code messed up on B & C chipsets because it lost the packet header and therefore the flag indicating the need for hardware checksums. MFC after: 2 weeks
2003-03-29 06:34:25 +00:00
} else {
m_head = m_new;
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 up the long mbuf chain recovery code to use m_defrag; the old code messed up on B & C chipsets because it lost the packet header and therefore the flag indicating the need for hardware checksums. MFC after: 2 weeks
2003-03-29 06:34:25 +00:00
error = bus_dmamap_load_mbuf(sc->xl_mtag, c->xl_map,
- Ensure that the busdma API won't do deferred loads by using the BUS_DMA_NOWAIT flag, since the code can't handle this. - Use NULL, NULL for the lockfunc and lockfuncarg parameters of bus_dma_tag_create() since deferred loads can't happen now.
2003-07-04 11:46:23 +00:00
m_head, xl_dma_map_txbuf, c->xl_ptr, BUS_DMA_NOWAIT);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
if (error) {
Fix up the long mbuf chain recovery code to use m_defrag; the old code messed up on B & C chipsets because it lost the packet header and therefore the flag indicating the need for hardware checksums. MFC after: 2 weeks
2003-03-29 06:34:25 +00:00
m_freem(m_head);
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "can't map mbuf (error %d)\n", error);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (1);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +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
}
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
if (sc->xl_type == XL_TYPE_905B) {
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
status = XL_TXSTAT_RND_DEFEAT;
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +00:00
#ifndef XL905B_TXCSUM_BROKEN
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
if (m_head->m_pkthdr.csum_flags) {
if (m_head->m_pkthdr.csum_flags & CSUM_IP)
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
status |= XL_TXSTAT_IPCKSUM;
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
if (m_head->m_pkthdr.csum_flags & CSUM_TCP)
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
status |= XL_TXSTAT_TCPCKSUM;
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
if (m_head->m_pkthdr.csum_flags & CSUM_UDP)
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
status |= XL_TXSTAT_UDPCKSUM;
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +00:00
}
solid reports that it is buggy *and* that it slows down transmit speed. Buggy report: Matt Dillon & others Slowness report: I can't find the e-mail MFC After: 1 minute
2004-03-19 23:20:23 +00:00
#endif
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
c->xl_ptr->xl_status = htole32(status);
Fix bug with 3c90xB cards and newer. We weren't trying to copy the mbuf chain into an mbuf cluster when there is more than 63 mbufs in the chain. We were trying with older cards though.
2002-12-18 23:16:22 +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_head;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
bus_dmamap_sync(sc->xl_mtag, c->xl_map, BUS_DMASYNC_PREWRITE);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
/*
* 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.
*/
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_start(struct ifnet *ifp)
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
{
struct xl_softc *sc = ifp->if_softc;
XL_LOCK(sc);
if (sc->xl_type == XL_TYPE_905B)
xl_start_90xB_locked(ifp);
else
xl_start_locked(ifp);
XL_UNLOCK(sc);
}
static void
xl_start_locked(struct ifnet *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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *sc = ifp->if_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
struct mbuf *m_head = NULL;
struct xl_chain *prev = NULL, *cur_tx = NULL, *start_tx;
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
struct xl_chain *prev_tx;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
u_int32_t status;
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
int error;
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
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
XL_LOCK_ASSERT(sc);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +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
/*
* 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) {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags |= IFF_DRV_OACTIVE;
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;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
while (sc->xl_cdata.xl_tx_free != NULL) {
Bring in the first chunk of altq driver modifications. This covers the following drivers: bfe(4), em(4), fxp(4), lnc(4), tun(4), de(4) rl(4), sis(4) and xl(4) More patches are pending on: http://peoples.freebsd.org/~mlaier/ Please take a look and tell me if "your" driver is missing, so I can fix this. Tested-by: many No-objection: -current, -net
2004-07-02 12:16:02 +00:00
IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head);
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 (m_head == NULL)
break;
/* Pick a descriptor off the free list. */
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
prev_tx = 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
cur_tx = sc->xl_cdata.xl_tx_free;
/* Pack the data into the descriptor. */
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
error = xl_encap(sc, cur_tx, m_head);
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
if (error) {
cur_tx = prev_tx;
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
continue;
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
}
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
sc->xl_cdata.xl_tx_free = cur_tx->xl_next;
cur_tx->xl_next = 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
/* Chain it together. */
if (prev != NULL) {
prev->xl_next = cur_tx;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
prev->xl_ptr->xl_next = htole32(cur_tx->xl_phys);
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
}
prev = cur_tx;
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
BPF_MTAP(ifp, cur_tx->xl_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
}
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.
*/
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
if (cur_tx == NULL)
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;
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.
*/
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
cur_tx->xl_ptr->xl_status = htole32(le32toh(cur_tx->xl_ptr->xl_status) |
XL_TXSTAT_DL_INTR);
bus_dmamap_sync(sc->xl_ldata.xl_tx_tag, sc->xl_ldata.xl_tx_dmamap,
BUS_DMASYNC_PREWRITE);
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
/*
* 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;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
sc->xl_cdata.xl_tx_tail->xl_ptr->xl_next =
htole32(start_tx->xl_phys);
status = sc->xl_cdata.xl_tx_tail->xl_ptr->xl_status;
sc->xl_cdata.xl_tx_tail->xl_ptr->xl_status =
htole32(le32toh(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))
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
CSR_WRITE_4(sc, XL_DOWNLIST_PTR, start_tx->xl_phys);
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
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.
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
*
- #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
* 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. :)
*/
Calling xl_rxeof() at the end of xl_start_locked() leads to recursion in case of IP fast forwarding. Enqueue a taskqueue(9) task instead of calling xl_rxeof() directly. Reported & tested by: Slava Alpatov Reviewed by: wpaul MFC after: 1 week
2005-05-20 07:16:09 +00:00
taskqueue_enqueue(taskqueue_swi, &sc->xl_task);
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(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
static void
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
xl_start_90xB_locked(struct ifnet *ifp)
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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *sc = ifp->if_softc;
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
struct mbuf *m_head = NULL;
struct xl_chain *prev = NULL, *cur_tx = NULL, *start_tx;
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
struct xl_chain *prev_tx;
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
int error, idx;
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
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +00:00
XL_LOCK_ASSERT(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
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (ifp->if_drv_flags & IFF_DRV_OACTIVE)
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +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
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) {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags |= IFF_DRV_OACTIVE;
Add 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
break;
}
Bring in the first chunk of altq driver modifications. This covers the following drivers: bfe(4), em(4), fxp(4), lnc(4), tun(4), de(4) rl(4), sis(4) and xl(4) More patches are pending on: http://peoples.freebsd.org/~mlaier/ Please take a look and tell me if "your" driver is missing, so I can fix this. Tested-by: many No-objection: -current, -net
2004-07-02 12:16:02 +00:00
IFQ_DRV_DEQUEUE(&ifp->if_snd, m_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 (m_head == NULL)
break;
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
prev_tx = cur_tx;
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
cur_tx = &sc->xl_cdata.xl_tx_chain[idx];
/* Pack the data into the descriptor. */
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
error = xl_encap(sc, cur_tx, m_head);
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +00:00
if (error) {
cur_tx = prev_tx;
o Only try to recopy the mbuf into an mbuf cluster if bus_dmamap_load_mbuf() returned EFBIG. o Fix mbuf leaks in an error (rare) code path. o Reuse the TX descriptor if xl_encap() failed instead of just picking the next one. o Better error messages.
2003-01-07 01:01:20 +00:00
continue;
Fix up callers of xl_encap so that they handle a failure response properly (likely due to mbuf exhaustion.) Previously, the driver got somewhat wedged. Also, remove the annoying messages printed every time xl_encap couldn't allocate a mbuf; they served no useful purpose, and just made an mbuf exhaustion situation more annoying. MFC after: 1 week
2003-04-08 01:05:54 +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
/* Chain it together. */
if (prev != NULL)
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
prev->xl_ptr->xl_next = htole32(cur_tx->xl_phys);
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
prev = cur_tx;
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
BPF_MTAP(ifp, cur_tx->xl_mbuf);
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_INC(idx, XL_TX_LIST_CNT);
sc->xl_cdata.xl_tx_cnt++;
}
/*
* If there are no packets queued, bail.
*/
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
if (cur_tx == NULL)
Further improve locking in xl(4): - Avoid an additional lock acquire/release when leaving xl_intr(), by changing xl_start*() to xl_start*_locked(), and calling the appropriate routine by chip revision (as the DMA descriptors are different). - Simplify the appropriate routines now that they are called with the lock held. This should save a significant amount of CPU cycles spent on servicing each interrupt for both UP and SMP whilst remaining MPSAFE. Tested by: rwatson
2004-07-09 02:28:23 +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
/*
* 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.
*/
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
cur_tx->xl_ptr->xl_status = htole32(le32toh(cur_tx->xl_ptr->xl_status) |
XL_TXSTAT_DL_INTR);
bus_dmamap_sync(sc->xl_ldata.xl_tx_tag, sc->xl_ldata.xl_tx_dmamap,
BUS_DMASYNC_PREWRITE);
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
/* Start transmission */
sc->xl_cdata.xl_tx_prod = idx;
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
start_tx->xl_prev->xl_ptr->xl_next = htole32(start_tx->xl_phys);
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
/*
* Set a timeout in case the chip goes out to lunch.
*/
ifp->if_timer = 5;
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_init(void *xsc)
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 = xsc;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK(sc);
xl_init_locked(sc);
XL_UNLOCK(sc);
}
static void
xl_init_locked(struct xl_softc *sc)
{
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_ifp;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
int error, 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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(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,
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
IFP2ENADDR(sc->xl_ifp)[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
}
/* 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. */
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = xl_list_rx_init(sc);
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "initialization of the rx ring failed (%d)\n",
error);
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
xl_stop(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)
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = xl_list_tx_init_90xB(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
else
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
error = xl_list_tx_init(sc);
if (error) {
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "initialization of the tx ring failed (%d)\n",
error);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
xl_stop(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
return;
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +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 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);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
CSR_WRITE_4(sc, XL_UPLIST_PTR, sc->xl_ldata.xl_rx_dmaaddr);
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_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,
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
sc->xl_cdata.xl_tx_chain[0].xl_phys);
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_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.
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
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);
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
/*
* increase packet size to allow reception of 802.1q or ISL packets.
* For the 3c90x chip, set the 'allow large packets' bit in the MAC
* control register. For 3c90xB/C chips, use the RX packet size
* register.
*/
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
if (sc->xl_type == XL_TYPE_905B)
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
CSR_WRITE_2(sc, XL_W3_MAXPKTSIZE, XL_PACKET_SIZE);
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
else {
u_int8_t macctl;
macctl = CSR_READ_1(sc, XL_W3_MAC_CTRL);
macctl |= XL_MACCTRL_ALLOW_LARGE_PACK;
CSR_WRITE_1(sc, XL_W3_MAC_CTRL, macctl);
}
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;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_stats_update_locked(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
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);
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
#ifdef DEVICE_POLLING
/* Disable interrupts if we are polling. */
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
if (ifp->if_capenable & IFCAP_POLLING)
xl(4) meets polling(4). Hardware for this work kindly provided by Eric Masson. MFC after: 3 weeks
2005-03-26 20:22:58 +00:00
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB|0);
else
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#endif
- #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|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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/* XXX Downcall to miibus. */
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);
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags |= IFF_DRV_RUNNING;
ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
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 fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
callout_reset(&sc->xl_stat_callout, hz, xl_stats_update, 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 media options.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_ifmedia_upd(struct ifnet *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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *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
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
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_LOCK(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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);
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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
switch (IFM_SUBTYPE(ifm->ifm_media)) {
- 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_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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (0);
- 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;
default:
break;
}
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
if (sc->xl_media & XL_MEDIAOPT_MII ||
sc->xl_media & XL_MEDIAOPT_BTX ||
sc->xl_media & XL_MEDIAOPT_BT4) {
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
xl_init_locked(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);
}
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_UNLOCK(sc);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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
}
/*
* Report current media status.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr)
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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *sc = ifp->if_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
u_int32_t icfg;
Report media status for bitrate PHYs.
2003-08-23 18:09:25 +00:00
u_int16_t status = 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;
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 fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
XL_LOCK(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
Report media status for bitrate PHYs.
2003-08-23 18:09:25 +00:00
XL_SEL_WIN(4);
status = CSR_READ_2(sc, XL_W4_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
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;
Report media status for bitrate PHYs.
2003-08-23 18:09:25 +00:00
ifmr->ifm_status = IFM_AVALID;
if ((status & XL_MEDIASTAT_CARRIER) == 0)
ifmr->ifm_status |= IFM_ACTIVE;
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 whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
switch (icfg) {
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 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:
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "unknown XCVR type: %d\n", icfg);
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;
}
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_ioctl(struct ifnet *ifp, u_long command, caddr_t 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
{
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
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
switch (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
case SIOCSIFFLAGS:
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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_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) {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING &&
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
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);
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
} else if (ifp->if_drv_flags & IFF_DRV_RUNNING &&
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
!(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);
style: Move the { back to the else line to match the }. Discussed with: glebius X-MFC after: 5.3-Release
2004-09-26 09:01:02 +00:00
} else {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0)
Do not call xl_init_locked() unconditionally when we are bringed UP. Call it only if we weren't UP before. In some cases xl_init causes long media re-negotiation, and ppp(8) fails to open PPPoE connection because it sets IFF_UP every time before opening PPPoE connection. PR: kern/69133 Patch by: mdodd Approved by: wpaul, julian (mentor) MFC after: 1 week
2004-09-09 12:16:29 +00:00
xl_init_locked(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
} else {
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
if (ifp->if_drv_flags & IFF_DRV_RUNNING)
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);
}
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;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
error = 0;
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/* XXX Downcall from if_addmulti() possibly with locks held. */
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
if (sc->xl_type == XL_TYPE_905B)
xl_setmulti_hash(sc);
else
xl_setmulti(sc);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +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
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;
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
case SIOCSIFCAP:
Big polling(4) cleanup. o Axe poll in trap. o Axe IFF_POLLING flag from if_flags. o Rework revision 1.21 (Giant removal), in such a way that poll_mtx is not dropped during call to polling handler. This fixes problem with idle polling. o Make registration and deregistration from polling in a functional way, insted of next tick/interrupt. o Obsolete kern.polling.enable. Polling is turned on/off with ifconfig. Detailed kern_poll.c changes: - Remove polling handler flags, introduced in 1.21. The are not needed now. - Forget and do not check if_flags, if_capenable and if_drv_flags. - Call all registered polling handlers unconditionally. - Do not drop poll_mtx, when entering polling handlers. - In ether_poll() NET_LOCK_GIANT prior to locking poll_mtx. - In netisr_poll() axe the block, where polling code asks drivers to unregister. - In netisr_poll() and ether_poll() do polling always, if any handlers are present. - In ether_poll_[de]register() remove a lot of error hiding code. Assert that arguments are correct, instead. - In ether_poll_[de]register() use standard return values in case of error or success. - Introduce poll_switch() that is a sysctl handler for kern.polling.enable. poll_switch() goes through interface list and enabled/disables polling. A message that kern.polling.enable is deprecated is printed. Detailed driver changes: - On attach driver announces IFCAP_POLLING in if_capabilities, but not in if_capenable. - On detach driver calls ether_poll_deregister() if polling is enabled. - In polling handler driver obtains its lock and checks IFF_DRV_RUNNING flag. If there is no, then unlocks and returns. - In ioctl handler driver checks for IFCAP_POLLING flag requested to be set or cleared. Driver first calls ether_poll_[de]register(), then obtains driver lock and [dis/en]ables interrupts. - In interrupt handler driver checks IFCAP_POLLING flag in if_capenable. If present, then returns.This is important to protect from spurious interrupts. Reviewed by: ru, sam, jhb
2005-10-01 18:56:19 +00:00
#ifdef DEVICE_POLLING
if (ifr->ifr_reqcap & IFCAP_POLLING &&
!(ifp->if_capenable & IFCAP_POLLING)) {
error = ether_poll_register(xl_poll, ifp);
if (error)
return(error);
XL_LOCK(sc);
/* Disable interrupts */
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB|0);
ifp->if_capenable |= IFCAP_POLLING;
XL_UNLOCK(sc);
return (error);
}
if (!(ifr->ifr_reqcap & IFCAP_POLLING) &&
ifp->if_capenable & IFCAP_POLLING) {
error = ether_poll_deregister(ifp);
/* Enable interrupts. */
XL_LOCK(sc);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ACK|0xFF);
CSR_WRITE_2(sc, XL_COMMAND, XL_CMD_INTR_ENB|XL_INTRS);
if (sc->xl_flags & XL_FLAG_FUNCREG)
bus_space_write_4(sc->xl_ftag, sc->xl_fhandle,
4, 0x8000);
ifp->if_capenable &= ~IFCAP_POLLING;
XL_UNLOCK(sc);
return (error);
}
#endif /* DEVICE_POLLING */
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK(sc);
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +00:00
ifp->if_capenable = ifr->ifr_reqcap;
if (ifp->if_capenable & IFCAP_TXCSUM)
ifp->if_hwassist = XL905B_CSUM_FEATURES;
else
ifp->if_hwassist = 0;
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_UNLOCK(sc);
Support for large frames for VLANs was added by tweaking the packet size register, present only on 3c90xB and later NICs. This meant that you could not use a 1500 byte MTU with VLANs on original 3c905/3c900 cards (boomerang chipset). The boomerang chip does support large frames though, just not in the same way: you can set the 'allow large frames' bit in the MAC control register to receive frames up to 4K in size. Changes: - Set the 'allow large frames' bit for boomerang chips and increase the packet size register for cyclone and later chips. This allows us to use IFCAP_VLAN_MTU on all supported xl(4) NICs. - Actually set the IFCAP_VLAN_MTU flag in the capabilities word in xl_attach(). - Change the method used to detect older boomerang chips. My 3c575C cardbus NIC was being incorrectly identified as 3c90x chip instead of 3c90xB because the capabilities word in its EEPROM reports a bizzare value. In addition to checking for the supportsNoTxLength bit, also check for the absence of the supportsLargePackets bit. Both of these cases denote a 3c90xB chip. - Make RX and TX checksums configurable via the SIOCSIFCAP ioctl. - Avoid an unecessary le32toh() in xl_rxeof(): we already have the received frame size in the lower 16 bits of rxstat, no need to read it again. Tested with 3c905-TX, 3c900-TPO, 3c980C and 3c575C NICs.
2003-07-10 05:24:33 +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:
o track either_ifattach/ether_ifdetach API changes o use if_input for input packet processing o don't strip the Ethernet header for input packets o use BPF_* macros bpf tapping o call ether_ioctl to handle default ioctl case o track vlan changes Reviewed by: many Approved by: re
2002-11-14 23:49:09 +00:00
error = ether_ioctl(ifp, command, data);
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;
}
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (error);
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
}
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
/*
* XXX: Invoked from ifnet slow timer. Lock coverage needed.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_watchdog(struct ifnet *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
{
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
struct xl_softc *sc = ifp->if_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
u_int16_t status = 0;
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);
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "watchdog timeout\n");
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_MEDIASTAT_CARRIER)
Use if_printf() and device_printf() where appropriate, i.e.: - Use device_printf() during device probe/attach. - Move if_xname initialization to before xl_reset() is called. - Use if_printf() at all other times after struct ifnet has been initialized.
2004-07-05 00:15:23 +00:00
if_printf(ifp, "no carrier - transceiver cable problem?\n");
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_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);
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_init_locked(sc);
Another missing ! in front of IFQ_DRV_IS_EMPTY, this time in an even more sensitive, but less excercised location (the watchdog). While here use the *_start_locked function directly to avoid drop, grab, drop lock. I have to be very careful with future ALTQ patches! Found & reviewed by: rwatson MFC after: 3 days
2004-11-08 16:16:04 +00:00
if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) {
if (sc->xl_type == XL_TYPE_905B)
xl_start_90xB_locked(ifp);
else
xl_start_locked(ifp);
}
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
Another missing ! in front of IFQ_DRV_IS_EMPTY, this time in an even more sensitive, but less excercised location (the watchdog). While here use the *_start_locked function directly to avoid drop, grab, drop lock. I have to be very careful with future ALTQ patches! Found & reviewed by: rwatson MFC after: 3 days
2004-11-08 16:16:04 +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
}
/*
* Stop the adapter and free any mbufs allocated to the
* RX and TX lists.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_stop(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
{
register int i;
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
struct ifnet *ifp = sc->xl_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
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
XL_LOCK_ASSERT(sc);
First round of converting network drivers from spls to mutexes. This takes care of all the 10/100 and gigE PCI drivers that I've done. Next will be the wireless drivers, then the USB ones. I may pick up some stragglers along the way. I'm sort of playing this by ear: if anyone spots any places where I've screwed up horribly, please let me know.
2000-10-13 17:54:19 +00:00
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_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);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +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. */
Various fixups to locking: - Remove a lot of superfluous locking during attach. There is no need to lock access to the driver until some other thread has a way of getting to it. For ethernet drivers the other ways include registering an interrupt handler via bus_setup_intr(), calling ether_ifattach() to hook into the network stack, and kicking off a callout-driven timer via callout_reset(). - Use callout_* rather than timeout/untimeout. - Break out of xl_rxeof() if IFF_DRV_RUNNING is clear after ifp->if_input returns to handle the case where the interface was stopped while we were passing a packet up the stack. Don't call xl_rxeof() in xl_rxeof_task() unless IFF_DRV_RUNNING is set. With these fixes in place, any outstanding task will gracefully terminate as soon as it gets a chance to run after the interface has been stopped via xl_stop(). As a result, taskqueue_drain() is no longer required in xl_stop(). The task is still drained in detach() however to make sure that detach() can safely destroy the driver mutex at the end of the function. - Lock the driver lock in the ifmedia callouts and don't lock across ifmedia_ioctl() in xl_ioctl(). Note: glebius came up with most of (3) as well independently. I took a rather roundabout way of arriving at the same conclusion. MFC after: 3 days
2005-08-18 19:24:30 +00:00
callout_stop(&sc->xl_stat_callout);
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
/*
* 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) {
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_unload(sc->xl_mtag,
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
sc->xl_cdata.xl_rx_chain[i].xl_map);
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_destroy(sc->xl_mtag,
Make the xl driver work on sparc64: - Add conversions to/from little endian for fields that the NIC accesses by DMA as required. - Add some bus_dmamap_sync() calls, and correct some existing ones. - Read the receiver information from the EEPROM in an endian-neutral manner. - Load all RX and TX descriptors in a single DMA map up front, and get the bus addresses of individual descriptors by address arithmetic; this fixes multiple use of the descriptor tags, which would have undesired effects. It seems that xl still does not work on e250 boxen, for reasons which are not clear yet. Reviewed by: mux
2003-01-19 00:23:59 +00:00
sc->xl_cdata.xl_rx_chain[i].xl_map);
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
m_freem(sc->xl_cdata.xl_rx_chain[i].xl_mbuf);
sc->xl_cdata.xl_rx_chain[i].xl_mbuf = NULL;
}
}
Do not attempt to clean up data that has not been initialized yet. This fixes two kernel panics on boot when the xl driver fails to allocate bus/port/memory resources. Reviewed by: silence on -net
2004-08-06 09:08:33 +00:00
if (sc->xl_ldata.xl_rx_list != NULL)
bzero(sc->xl_ldata.xl_rx_list, XL_RX_LIST_SZ);
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
/*
* 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) {
Convert the xl(4) driver to the busdma API. This should make it possible to use this driver under ia64, sparc64 (though there may be endianness issues with this one) and other archs. Tested on: i386, alpha (gallatin)
2002-12-17 00:08:49 +00:00
bus_dmamap_unload(sc->xl_mtag,
sc->xl_cdata.xl_tx_chain[i].xl_map);
bus_dmamap_destroy(sc->xl_mtag,
sc->xl_cdata.xl_tx_chain[i].xl_map);
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
m_freem(sc->xl_cdata.xl_tx_chain[i].xl_mbuf);
sc->xl_cdata.xl_tx_chain[i].xl_mbuf = NULL;
}
}
Do not attempt to clean up data that has not been initialized yet. This fixes two kernel panics on boot when the xl driver fails to allocate bus/port/memory resources. Reviewed by: silence on -net
2004-08-06 09:08:33 +00:00
if (sc->xl_ldata.xl_tx_list != NULL)
bzero(sc->xl_ldata.xl_tx_list, XL_TX_LIST_SZ);
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
Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to ifnet.if_drv_flags. Device drivers are now responsible for synchronizing access to these flags, as they are in if_drv_flags. This helps prevent races between the network stack and device driver in maintaining the interface flags field. Many __FreeBSD__ and __FreeBSD_version checks maintained and continued; some less so. Reviewed by: pjd, bz MFC after: 7 days
2005-08-09 10:20:02 +00:00
ifp->if_drv_flags &= ~(IFF_DRV_RUNNING | IFF_DRV_OACTIVE);
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 all chip I/O so that the kernel's probe routines don't
* get confused by errant DMAs when rebooting.
*/
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static void
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_shutdown(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
}
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
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_suspend(device_t dev)
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
{
struct xl_softc *sc;
sc = device_get_softc(dev);
XL_LOCK(sc);
xl_stop(sc);
XL_UNLOCK(sc);
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (0);
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
}
style: put return types on a line by themselves.
2002-08-24 00:02:03 +00:00
static int
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
xl_resume(device_t dev)
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
{
struct xl_softc *sc;
struct ifnet *ifp;
sc = device_get_softc(dev);
Stop embedding struct ifnet at the top of driver softcs. Instead the struct ifnet or the layer 2 common structure it was embedded in have been replaced with a struct ifnet pointer to be filled by a call to the new function, if_alloc(). The layer 2 common structure is also allocated via if_alloc() based on the interface type. It is hung off the new struct ifnet member, if_l2com. This change removes the size of these structures from the kernel ABI and will allow us to better manage them as interfaces come and go. Other changes of note: - Struct arpcom is no longer referenced in normal interface code. Instead the Ethernet address is accessed via the IFP2ENADDR() macro. To enforce this ac_enaddr has been renamed to _ac_enaddr. - The second argument to ether_ifattach is now always the mac address from driver private storage rather than sometimes being ac_enaddr. Reviewed by: sobomax, sam
2005-06-10 16:49:24 +00:00
ifp = sc->xl_ifp;
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
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
XL_LOCK(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_reset(sc);
if (ifp->if_flags & IFF_UP)
Reintroduce and clean up locking in xl(4). - Eliminate the use of a recursive mutex. - Mark the driver as INTR_MPSAFE. - Split the default media choice code out into xl_choose_media() to avoid making poor assumptions about the state of the lock during attach. - The miibus upcall/downcall paths may still be racy. Change to commented-out locking assertions there for now. - Tested with nfsclient, routed, ssh, ntp, dhclient and quagga bgpd. - This needs SMP test coverage. I do not have such resources. Tested on: UP, !debug.mpsafenet && debug.mpsafenet Hardware: 3C905B-TX (0x905510b7)
2004-07-05 02:34:35 +00:00
xl_init_locked(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);
ANSIfy function definitions. Remove unnecessary return keywords. Other minor stylistic changes.
2004-07-04 23:31:28 +00:00
Fix whitespace, indentation, long line wrapping and comments.
2004-07-04 22:20:52 +00:00
return (0);
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
}
Reference in New Issue Copy Permalink