Bug#538410: atl1c: ethernet broken on Eee PC model 1005HA-H
On Sat, 2009-07-25 at 13:39 -0300, Ben Armstrong wrote: > Package: linux-image-2.6.30-1-686 > Version: 2.6.30-3 > Severity: normal > > When using a debian-installer daily image on an Eee PC model 1005HA-H, to > which > Otavio just added atl1c, our tester, Yoda-BZH on irc, could not get ethernet > to > work at all. Here is his report: > > > debian-installer/i386/linux vga=771 > initrd=debian-installer/i386/initrd.gz -- quiet > > Choosing language French > Country: France > KB Layout: Françs (fr-latin9) > > At this point I have few lines about atl1c in dmesg : > atl1c :01:00.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19 > atl1c :01:00.0: setting latency timer to 64 > atl1c :01:00.0: PME# disabled > atl1c :01:00.0: PME# disabled > atl1c :01:00.0: version 1.0.0.1-NAPI > > > ** detecting hardware step > In dmesg I now have : > atl1c :01:00.0: irq 27 for MSI/MSI-X > atl1c :01:00.0: alt1c: eth0 NIC Link is Up<100 Mbps Full Duplex> > atl1c :01:00.0: Unable to allocate MSI interrupt Error: -22 > ADDRCONF(NETDEV_UP): eth0: link is not ready > > At this point, dhcp fails to acquire an address. Statically configuring the > network doesn't help, either. It looks like the driver is somehow reporting link state correctly - the PHY shows link up but the kernel sees the link down. I seem to remember that there was a recent change that broke some other drivers that were not doing this quite right; I will check that. > Searching Linus's git tree reveals that there have been some patches to atl1c > since 2.6.30 was released, but we don't know if any of these are relevant to > our problem or not: [...] These don't look like they would address the problem, but perhaps you could try a kernel with these changes anyway? There are experimental packages linked from http://wiki.debian.org/DebianKernel (see the first entry under "snapshots"). Ben. -- Ben Hutchings Power corrupts. Absolute power is kind of neat. - John Lehman, Secretary of the US Navy 1981-1987 signature.asc Description: This is a digitally signed message part
Bug#538410: atl1c: ethernet broken on Eee PC model 1005HA-H
Hi Greg, On Sun, Jul 26, 2009 at 10:58:36PM +0200, Moritz Muehlenhoff wrote: > On Sat, Jul 25, 2009 at 01:39:55PM -0300, Ben Armstrong wrote: > > Package: linux-image-2.6.30-1-686 > > Version: 2.6.30-3 > > Severity: normal > > > > When using a debian-installer daily image on an Eee PC model 1005HA-H, to > > which > > Otavio just added atl1c, our tester, Yoda-BZH on irc, could not get > > ethernet to > > work at all. Here is his report: > > > > > debian-installer/i386/linux vga=771 > > initrd=debian-installer/i386/initrd.gz -- quiet > > > > Choosing language French > > Country: France > > KB Layout: Françs (fr-latin9) > > > > At this point I have few lines about atl1c in dmesg : > > atl1c :01:00.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19 > > atl1c :01:00.0: setting latency timer to 64 > > atl1c :01:00.0: PME# disabled > > atl1c :01:00.0: PME# disabled > > atl1c :01:00.0: version 1.0.0.1-NAPI > > > > ** detecting hardware step > > In dmesg I now have : > > atl1c :01:00.0: irq 27 for MSI/MSI-X > > atl1c :01:00.0: alt1c: eth0 NIC Link is Up<100 Mbps Full Duplex> > > atl1c :01:00.0: Unable to allocate MSI interrupt Error: -22 > > ADDRCONF(NETDEV_UP): eth0: link is not ready > > > > At this point, dhcp fails to acquire an address. Statically configuring the > > network doesn't help, either. > > > > Searching Linus's git tree reveals that there have been some patches to > > atl1c > > since 2.6.30 was released, but we don't know if any of these are relevant to > > our problem or not: > > > > http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Ftorvalds%2Flinux-2.6.git&a=search&h=HEAD&st=commit&s=atl1c > > > > I find it annoying that at least this bug appears to be a duplicate of an > > atl1e > > bug fixed a year ago: > > > > In atl1c this year, after 2.6.30 was released: > > > > atl1c: WAKE_MCAST tested twice, not WAKE_UCAST > > > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=0ed586d075ef65c0268982e5b7f36d0ffaa95547 > > > > In atl1e last year, submitted by the same upstream author! > > > > atl1e: WAKE_MCAST 2x. 1st WAKE_UCAST? > > > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=1a3c4bc61547e5a75fd3b85b425624756da4cffb > > > > Do you need any more info to patch 2.6.30's atl1c for us? We'd like to get > > ethernet working on this model as soon as possible. > > > > Thanks, > > Ben Armstrong, debian-eeepc project leader > > Hi Ben, > could you (or "Yoda BSH") test that applying > 0ed586d075ef65c0268982e5b7f36d0ffaa95547 > on top of 2.6.30 fixes support for atl1c? If so, the best way to proceed > would be to > mail sta...@kernel.org so that the fix is picked up for the kernel stable > update > (2.6.30.4 or .5). In that case it will land in sid rather quickly and users > of the > stock upstream kernel benefit as well. Greg, this concerns the Debian bug report http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=538410 on the atl1c driver: Please merge the following commits into the stable tree for 2.6.30: 0ed586d075ef65c0268982e5b7f36d0ffaa95547 c5ad4f592e27d782faea0a787d9181f192a69ef0 37b76c697f4ac082e9923dfa8e8aecc8bc54a8e1 Tristan Charbonneau confirmed that these fixes - as present in 2.6.31-rc4 - fixes the problem on his hardware. Cheers, Moritz -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#538410: atl1c: ethernet broken on Eee PC model 1005HA-H
Hi Tested with a 2.6.31-rc4 which contains both patches (missing/misplaced parenthesis fix and WAKE_UCAST fix) and it works fine. Thanks a lot -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#538410: atl1c: ethernet broken on Eee PC model 1005HA-H
On Sat, Jul 25, 2009 at 01:39:55PM -0300, Ben Armstrong wrote: > Package: linux-image-2.6.30-1-686 > Version: 2.6.30-3 > Severity: normal > > When using a debian-installer daily image on an Eee PC model 1005HA-H, to > which > Otavio just added atl1c, our tester, Yoda-BZH on irc, could not get ethernet > to > work at all. Here is his report: > > > debian-installer/i386/linux vga=771 > initrd=debian-installer/i386/initrd.gz -- quiet > > Choosing language French > Country: France > KB Layout: Françs (fr-latin9) > > At this point I have few lines about atl1c in dmesg : > atl1c :01:00.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19 > atl1c :01:00.0: setting latency timer to 64 > atl1c :01:00.0: PME# disabled > atl1c :01:00.0: PME# disabled > atl1c :01:00.0: version 1.0.0.1-NAPI > > ** detecting hardware step > In dmesg I now have : > atl1c :01:00.0: irq 27 for MSI/MSI-X > atl1c :01:00.0: alt1c: eth0 NIC Link is Up<100 Mbps Full Duplex> > atl1c :01:00.0: Unable to allocate MSI interrupt Error: -22 > ADDRCONF(NETDEV_UP): eth0: link is not ready > > At this point, dhcp fails to acquire an address. Statically configuring the > network doesn't help, either. > > Searching Linus's git tree reveals that there have been some patches to atl1c > since 2.6.30 was released, but we don't know if any of these are relevant to > our problem or not: > > http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Ftorvalds%2Flinux-2.6.git&a=search&h=HEAD&st=commit&s=atl1c > > I find it annoying that at least this bug appears to be a duplicate of an > atl1e > bug fixed a year ago: > > In atl1c this year, after 2.6.30 was released: > > atl1c: WAKE_MCAST tested twice, not WAKE_UCAST > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=0ed586d075ef65c0268982e5b7f36d0ffaa95547 > > In atl1e last year, submitted by the same upstream author! > > atl1e: WAKE_MCAST 2x. 1st WAKE_UCAST? > > http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=1a3c4bc61547e5a75fd3b85b425624756da4cffb > > Do you need any more info to patch 2.6.30's atl1c for us? We'd like to get > ethernet working on this model as soon as possible. > > Thanks, > Ben Armstrong, debian-eeepc project leader Hi Ben, could you (or "Yoda BSH") test that applying 0ed586d075ef65c0268982e5b7f36d0ffaa95547 on top of 2.6.30 fixes support for atl1c? If so, the best way to proceed would be to mail sta...@kernel.org so that the fix is picked up for the kernel stable update (2.6.30.4 or .5). In that case it will land in sid rather quickly and users of the stock upstream kernel benefit as well. Cheers, Moritz -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#538410: atl1c: ethernet broken on Eee PC model 1005HA-H
Package: linux-image-2.6.30-1-686 Version: 2.6.30-3 Severity: normal When using a debian-installer daily image on an Eee PC model 1005HA-H, to which Otavio just added atl1c, our tester, Yoda-BZH on irc, could not get ethernet to work at all. Here is his report: > debian-installer/i386/linux vga=771 initrd=debian-installer/i386/initrd.gz -- quiet Choosing language French Country: France KB Layout: Françs (fr-latin9) At this point I have few lines about atl1c in dmesg : atl1c :01:00.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19 atl1c :01:00.0: setting latency timer to 64 atl1c :01:00.0: PME# disabled atl1c :01:00.0: PME# disabled atl1c :01:00.0: version 1.0.0.1-NAPI ** detecting hardware step In dmesg I now have : atl1c :01:00.0: irq 27 for MSI/MSI-X atl1c :01:00.0: alt1c: eth0 NIC Link is Up<100 Mbps Full Duplex> atl1c :01:00.0: Unable to allocate MSI interrupt Error: -22 ADDRCONF(NETDEV_UP): eth0: link is not ready At this point, dhcp fails to acquire an address. Statically configuring the network doesn't help, either. Searching Linus's git tree reveals that there have been some patches to atl1c since 2.6.30 was released, but we don't know if any of these are relevant to our problem or not: http://git.kernel.org/?p=linux%2Fkernel%2Fgit%2Ftorvalds%2Flinux-2.6.git&a=search&h=HEAD&st=commit&s=atl1c I find it annoying that at least this bug appears to be a duplicate of an atl1e bug fixed a year ago: In atl1c this year, after 2.6.30 was released: atl1c: WAKE_MCAST tested twice, not WAKE_UCAST http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=0ed586d075ef65c0268982e5b7f36d0ffaa95547 In atl1e last year, submitted by the same upstream author! atl1e: WAKE_MCAST 2x. 1st WAKE_UCAST? http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=1a3c4bc61547e5a75fd3b85b425624756da4cffb Do you need any more info to patch 2.6.30's atl1c for us? We'd like to get ethernet working on this model as soon as possible. Thanks, Ben Armstrong, debian-eeepc project leader -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
