[Bug 48061] PCI issues affecting 3ware Inc 7xxx/8xxx-series and RTL-8169 Gigabit Ethernet

[lusis]

Public bug reported:

Binary package hint: linux-image-2.6.15-23-amd64-k8

Since upgrading to dapper and the new kernel, I've gotten several errors
and system freezes (which eventually clear up after a few minutes)
relating to different PCI devices.

The first symptom after rebooting into 2.6.15-23 and dapper was
attempting to view my themes in Thunderbird. I rebooted and kept having
the same problem after doign various tasks. I finally disabled beagle
when I saw in dmesg that the problem was in the 3Ware card. I thought
maybe something was up with the driver in 2.6.15-23 and beagle was
pushing too much I/O.

Reboot again and disable beagle. Restart X and problem still happens. I
can reproduce it with Azureus. I realized after the last lockup and
waiting about 10 minutes everything comes back. The lock ups eventually
happen again.

Finally get back into dmesg and notice the following:

Jun  2 06:48:36 localhost kernel: [19852.454073] 3w-xxxx: scsi0: PCI Abort: clearing.
Jun  2 06:48:57 localhost kernel: [19873.193049] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:49:08 localhost kernel: [19883.752039] sd 0:0:0:0: WARNING: Command (0x2a) timed out, resetting card.
Jun  2 06:49:09 localhost kernel: [19885.190279] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:49:21 localhost kernel: [19897.187507] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:49:33 localhost kernel: [19909.184736] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:49:38 localhost kernel: [19913.785099] 3w-xxxx: scsi0: AEN drain failed, retrying.
Jun  2 06:49:45 localhost kernel: [19921.181964] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:49:57 localhost kernel: [19933.179194] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 06:50:09 localhost kernel: [19945.176423] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:19:26 localhost -- MARK --
Jun  2 07:23:36 localhost kernel: [21952.012110] 3w-xxxx: scsi0: PCI Abort: clearing.
Jun  2 07:23:51 localhost kernel: [21966.707481] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:24:06 localhost kernel: [21981.992177] sd 0:0:0:0: WARNING: Command (0x28) timed out, resetting card.
Jun  2 07:24:09 localhost kernel: [21984.703325] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:24:36 localhost kernel: [22012.026232] 3w-xxxx: scsi0: AEN drain failed, retrying.
Jun  2 07:24:51 localhost kernel: [22026.693627] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:25:03 localhost kernel: [22038.690856] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:25:15 localhost kernel: [22050.688084] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:25:27 localhost kernel: [22062.685314] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:25:39 localhost kernel: [22074.682542] NETDEV WATCHDOG: eth0: transmit timed out
Jun  2 07:25:57 localhost kernel: [22092.678385] NETDEV WATCHDOG: eth0: transmit timed out


So it seems that the problem was happening without on the network as well which leads me to look for a pci or (upon further googling) a possible apic issue in 2.6.15?

I rebooted into 2.6.12-10 and it's been fine thus far with no errors. I
don't have X running right now because of the nvidia/xorg change in
dapper but I had been running the previous 2.6.12-10 configuration for
the past several months with no problem.

I can attach whatever files needed (dmi dump, dmesg, messages, lspci)
just let me know what would provide the most information. I think I can
easily trigger the problem with the new kernel by starting Azureus again
and multiple ISO downloads.

** Affects: linux-source-2.6.15 (Ubuntu)
       Severity: Normal
       Priority: (none set)
         Status: Unconfirmed

-- 
PCI issues affecting 3ware Inc 7xxx/8xxx-series and RTL-8169 Gigabit Ethernet
https://launchpad.net/bugs/48061