Menu
▾
▴
devil-linux-develop
|
From: Tim T. <t....@co...> - 2004-05-11 04:47:12
|
Found this patch set for the RTL8169 for 2.4.26, no idea if it fixes the problem but I'll try it if I can figure it out... A couple of questions: - Does DL support the "NMI" watchdog? Is that what that make menuconfig pick is supposed to do? I had it on but it didn't seem to do anyhing. - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being turned on if support in the init.d/boot script, so it must no be comiled in cause the /proc... file does not exist. Thanks- Tim from message thread: <http://oss.sgi.com/archives/netdev/2004-04/msg00341.html> ----------- from <http://www.fr.zoreil.com/linux/kernel/2.4.x/2.4.26/README.txt> : This is an updated version of the r8169 patchset. The fix are directly included in the relevant patches. Apply/compile/test to vanilla 2.4.26 in following order: file | md5 -------------------------------------+--------------------------------- r8169-upgrade-to-2.6.4-rc2.patch b3bc23560b2c642669c4aa4fb25f3572 r8169-dma-api-tx.patch a028c8710ed54d4447d9236f482b6c84 r8169-dma-api-rx-buffers.patch 9d242d745b5054127d7a81587fbbb18c r8169-dma-api-tx-buffers.patch 7370aa7904228bbb19588b1ac18e283c r8169-rx_copybreak.patch dd0a99f3d5cbba194af9ca2fe297b801 r8169-mac-phy-version.patch 6430bfbbad8cb107975d61e4ffcf93a0 r8169-init_one.patch d71baaee42398f173894b08a6e130dbf r8169-timer.patch de72e4825867f22f0b302edb985c9a02 r8169-hw_start.patch 8427a73de3f1c2c72f69da81dde0c9fc r8169-intr_mask.patch 08d386707a2a67a0cddaf3d748df8ff2 r8169-suspend.patch e7832871558326cf9e1a4d8a15cacfa5 r8169-endianness.patch e99ad19f62cca91e48a29f8fcd5f1ceb r8169-getstats.patch b36929bae4b35f6bd8fb6de356eb6475 r8169-ethtool-introduction.patch 2323a525428cd8247e531116d7abbad0 r8169-missing-static.patch ab65bb8ffcab7695663e12b5349c5fe7 r8169-barrier.patch 81a6e52ff808abe0abd9678ea3ed465a r8169-kill-potential-busy-loop.patch 69cdc5de2f9c45d84b2921968d035363 r8169-dac-10.patch b99a95566e141df52de6b63fc707905b r8169-irq-handler-status.patch 0a5b62497984acd4185e81a14226e37b Each patch must be applied on top of the preceeding one: there should be zero warning from the patch command. If the patches do not apply cleanly, please provide an (sh-)history of the issued commands as well as md5sum of the relevant files. Please compile and report after each patch if the driver is stable without/under load. A fix related to a bug under load has been merged in recent 2.4.26-xx. It should help. Comparisons with/without SMP support enabled in the kernel will be welcome. If something goes wrong: - verify that the problem disappears when the latest patch is not applied; - send: - the list of applied patches ; - 'dmesg', 'lspci', 'lsmod', 'cat /proc/interrupts', 'ifconfig' output ; - a few words of explanation ; - disable acpi if it was enabled and try again ; - please Cc: ne...@os.... Depending on your hardware configuration (Intel + apic), you may use the nmi watchdog if things go really scary. Enable it in the configuration menu and include an 'nmi_watchdog=1' option on the kernel boot command-line. The magic sysrq (alt+sysrq+t) may help too. Please check it is not disabled in your /etc/sysctl.conf (distribution dependant ?). Thank you for your contribution. |
|
From: Heiko Z. <he...@zu...> - 2004-05-12 01:50:33
|
Tim Tait wrote: > > Found this patch set for the RTL8169 for 2.4.26, no idea if it fixes the > problem but I'll try it if I can figure it out... So the question is: do we have to include one or more of those patches? > A couple of questions: > - Does DL support the "NMI" watchdog? Is that what that make menuconfig > pick is supposed to do? I had it on but it didn't seem to do anyhing. NMI watchdog? I never saw this anywhere... (but I could jus tbe blind or ignorant) > - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being > turned on if support in the init.d/boot script, so it must no be > comiled in cause the /proc... file does not exist. This is for Kernel debugging, I don't really want to turn this on. Heiko |
|
From: Friedrich L. <fl...@fl...> - 2004-05-14 00:00:24
|
Heiko Zuerker wrote on 12.05.2004 03:46 MET: > Tim Tait wrote: > >> - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being >> turned on if support in the init.d/boot script, so it must no be >> comiled in cause the /proc... file does not exist. > > This is for Kernel debugging, I don't really want to turn this on. No not really for kernel debugging only. Magic-SysRequest would be really nice. Didn't we include the patch to achive that via the network? You can even turn it off or on during runtime (if it is compiled in). From /etc/init.d/boot.proc on my SuSE 8.2: 25 # check if sysrq should be enabled 26 # 27 if test -e /proc/sys/kernel/sysrq ; then 28 if test "$ENABLE_SYSRQ" = yes ; then 29 echo "1" > /proc/sys/kernel/sysrq 30 else 31 echo "0" > /proc/sys/kernel/sysrq 32 fi 33 fi BTW take a look at /usr/src/linux/Documentation/sysrq.txt -- MfG / Regards Friedrich Lobenstock ____________________________________________________________________ Friedrich Lobenstock Linux Services Lobenstock URL: http://www.lsl.at/ Email: fl...@fl... ____________________________________________________________________ |
|
From: Heiko Z. <he...@zu...> - 2004-05-14 00:16:33
|
Friedrich Lobenstock wrote: > Heiko Zuerker wrote on 12.05.2004 03:46 MET: > >> Tim Tait wrote: >> >>> - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being >>> turned on if support in the init.d/boot script, so it must no be >>> comiled in cause the /proc... file does not exist. >> >> >> This is for Kernel debugging, I don't really want to turn this on. > > > No not really for kernel debugging only. Magic-SysRequest would be > really nice. Didn't we include the patch to achive that via the network? Nope to yet. > You can even turn it off or on during runtime (if it is compiled in). > From /etc/init.d/boot.proc on my SuSE 8.2: > > 25 # check if sysrq should be enabled > 26 # > 27 if test -e /proc/sys/kernel/sysrq ; then > 28 if test "$ENABLE_SYSRQ" = yes ; then > 29 echo "1" > /proc/sys/kernel/sysrq > 30 else > 31 echo "0" > /proc/sys/kernel/sysrq > 32 fi > 33 fi We currently don;'t have it enabled in the Kernel config. > BTW take a look at /usr/src/linux/Documentation/sysrq.txt I rather keep watching the DVD Heiko |
|
From: Tim T. <t....@co...> - 2004-05-12 02:12:12
|
Tim Tait wrote: > > Found this patch set for the RTL8169 for 2.4.26, no idea if it fixes > the problem but I'll try it if I can figure it out... > > A couple of questions: > - Does DL support the "NMI" watchdog? Is that what that make > menuconfig pick is supposed to do? I had it on but it didn't seem to > do anyhing. > > - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being > turned on if support in the init.d/boot script, so it must no be > comiled in cause the /proc... file does not exist. > > Thanks- > > Tim > > from message thread: > <http://oss.sgi.com/archives/netdev/2004-04/msg00341.html> > > ----------- from > <http://www.fr.zoreil.com/linux/kernel/2.4.x/2.4.26/README.txt> : > > This is an updated version of the r8169 patchset. The fix are directly > included in the relevant patches. > > Apply/compile/test to vanilla 2.4.26 in following order: > > file | md5 > -------------------------------------+--------------------------------- > r8169-upgrade-to-2.6.4-rc2.patch b3bc23560b2c642669c4aa4fb25f3572 > r8169-dma-api-tx.patch a028c8710ed54d4447d9236f482b6c84 > r8169-dma-api-rx-buffers.patch 9d242d745b5054127d7a81587fbbb18c > r8169-dma-api-tx-buffers.patch 7370aa7904228bbb19588b1ac18e283c > r8169-rx_copybreak.patch dd0a99f3d5cbba194af9ca2fe297b801 > r8169-mac-phy-version.patch 6430bfbbad8cb107975d61e4ffcf93a0 > r8169-init_one.patch d71baaee42398f173894b08a6e130dbf > r8169-timer.patch de72e4825867f22f0b302edb985c9a02 > r8169-hw_start.patch 8427a73de3f1c2c72f69da81dde0c9fc > r8169-intr_mask.patch 08d386707a2a67a0cddaf3d748df8ff2 > r8169-suspend.patch e7832871558326cf9e1a4d8a15cacfa5 > r8169-endianness.patch e99ad19f62cca91e48a29f8fcd5f1ceb > r8169-getstats.patch b36929bae4b35f6bd8fb6de356eb6475 > r8169-ethtool-introduction.patch 2323a525428cd8247e531116d7abbad0 > r8169-missing-static.patch ab65bb8ffcab7695663e12b5349c5fe7 > r8169-barrier.patch 81a6e52ff808abe0abd9678ea3ed465a > r8169-kill-potential-busy-loop.patch 69cdc5de2f9c45d84b2921968d035363 > r8169-dac-10.patch b99a95566e141df52de6b63fc707905b > r8169-irq-handler-status.patch 0a5b62497984acd4185e81a14226e37b > > Each patch must be applied on top of the preceeding one: there > should be zero warning from the patch command. <http://www.fr.zoreil.com/linux/kernel/2.4.x/2.4.26/r8169-blob.tar.bz2> These patches did the trick, works like a charm at gigabit speeds now. About 25% faster thoughput too than in 2.4.25 or stock 2.4.26! I've never done a source tree patch submission before, so how does this patch get rolled into DL? Should the diffs be combined into one? Tim |
|
From: Heiko Z. <he...@zu...> - 2004-05-12 02:25:28
|
Tim Tait wrote: > Tim Tait wrote: > >> >> Found this patch set for the RTL8169 for 2.4.26, no idea if it fixes >> the problem but I'll try it if I can figure it out... >> >> A couple of questions: >> - Does DL support the "NMI" watchdog? Is that what that make >> menuconfig pick is supposed to do? I had it on but it didn't seem to >> do anyhing. >> >> - Where do I turn on the "magic" Alt-sysrq-T thingy? I see it being >> turned on if support in the init.d/boot script, so it must no be >> comiled in cause the /proc... file does not exist. >> >> Thanks- >> >> Tim >> >> from message thread: >> <http://oss.sgi.com/archives/netdev/2004-04/msg00341.html> >> >> ----------- from >> <http://www.fr.zoreil.com/linux/kernel/2.4.x/2.4.26/README.txt> : >> >> This is an updated version of the r8169 patchset. The fix are directly >> included in the relevant patches. >> >> Apply/compile/test to vanilla 2.4.26 in following order: >> >> file | md5 >> -------------------------------------+--------------------------------- >> r8169-upgrade-to-2.6.4-rc2.patch b3bc23560b2c642669c4aa4fb25f3572 >> r8169-dma-api-tx.patch a028c8710ed54d4447d9236f482b6c84 >> r8169-dma-api-rx-buffers.patch 9d242d745b5054127d7a81587fbbb18c >> r8169-dma-api-tx-buffers.patch 7370aa7904228bbb19588b1ac18e283c >> r8169-rx_copybreak.patch dd0a99f3d5cbba194af9ca2fe297b801 >> r8169-mac-phy-version.patch 6430bfbbad8cb107975d61e4ffcf93a0 >> r8169-init_one.patch d71baaee42398f173894b08a6e130dbf >> r8169-timer.patch de72e4825867f22f0b302edb985c9a02 >> r8169-hw_start.patch 8427a73de3f1c2c72f69da81dde0c9fc >> r8169-intr_mask.patch 08d386707a2a67a0cddaf3d748df8ff2 >> r8169-suspend.patch e7832871558326cf9e1a4d8a15cacfa5 >> r8169-endianness.patch e99ad19f62cca91e48a29f8fcd5f1ceb >> r8169-getstats.patch b36929bae4b35f6bd8fb6de356eb6475 >> r8169-ethtool-introduction.patch 2323a525428cd8247e531116d7abbad0 >> r8169-missing-static.patch ab65bb8ffcab7695663e12b5349c5fe7 >> r8169-barrier.patch 81a6e52ff808abe0abd9678ea3ed465a >> r8169-kill-potential-busy-loop.patch 69cdc5de2f9c45d84b2921968d035363 >> r8169-dac-10.patch b99a95566e141df52de6b63fc707905b >> r8169-irq-handler-status.patch 0a5b62497984acd4185e81a14226e37b >> >> Each patch must be applied on top of the preceeding one: there >> should be zero warning from the patch command. > > > <http://www.fr.zoreil.com/linux/kernel/2.4.x/2.4.26/r8169-blob.tar.bz2> > > These patches did the trick, works like a charm at gigabit speeds now. > About 25% faster thoughput too than in 2.4.25 or stock 2.4.26! > > I've never done a source tree patch submission before, so how does this > patch get rolled into DL? Should the diffs be combined into one? We can use the patches as-is I think. I'll look into it when I get some more time. Any comments anybody? Heiko |
