Kernel panics when using 8.10 on Thinkpad T61p in docking station

[Larry Hastings]

Binary package hint: kernel-image-2.6.27-3-generic-di

I have installed 8.10 on my Lenovo Thinkpad T61p. If I use the laptop on its own, it runs for days (and is fabulous, by the way). If I drop it into the docking station, it kernel panics, often within five minutes. The entire machine will be frozen, with just the Caps Lock light flashing at me.

Yesterday I powered up the machine in the dock and walked away for ten minutes. When I came back I discovered the kernel had panicked--and I hadn't even logged in yet.

I figured the most likely culprit was driving the 30" Dell widescreen monitor hooked up to the docking station. The laptop doesn't have its own DVI port, so I can only drive the monitor at 2560x1600 with the DVI port on the docking station. I'm using the NVidia 177 (beta) binary driver (nvidia-glx-177).

However! I've also experienced some kernel panics even when not using the docking station. The minimal configuration that panicked the kernel had a USB hub with a USB mouse and keyboard attached; it didn't happen nearly as soon as when using the docking station, it panicked maybe an hour or two after booting. I disconnected the USB hub, hard reset, and the machine ran all day.

The only unconventional things about this install:
1) The root partition is encrypted (I used the alternate installer).
2) The home partition is encrypted, and formatted with ZFS. It's automatically mounted during startup with an Upstart script I wrote.
I can't imagine those are related to the panic, but I've been wrong before.

I'm *very* interested in solving this problem, but I'm only marginally competent at Linux. To start with, I have no idea how to find any error message associated with the kernel panic. Please--how do I find out why the kernel panicked?

[Leann Ogasawara (leannogasawara)]

Hi Larry,

Would you be able to capture the dmesg output which contains the kernel panic you are seeing and attach it to this bug report? If not, even if you would be able to take a digital photo of the kernel panic and attach it to this bug report that would be fine. Additionally, you may want to take a peek in /var/log/kern.log.0 and see if it is logged there. Thanks.

[Larry Hastings (larry-hastings)]

I never got dmesg output, and never saw the panics in any log files. But today I finally hooked up a serial console and captured the kernel panic output.

I've attached the logs from two different kernel panics.

Since originally posting this message, I believe I've narrowed it down to "wireless activity". If I use WiFi, the more general WiFi activity going on near me the more likely a kernel panic becomes. This week I was in a room with 100 people all trying to use WiFi, and my machine would panic in only a minute or two. If I use WiFi at home where there's little activity it stays up for a long time, and if I use wired network or my CDMA card it is absolutely stable.

I saw some kcrypt stuff in the trace, so I think I should mention: my root partition is encrypted, using the Intrepid "alternate installer" encrypted root partition support. My home encryption is encrypted too; I'm running a ZFS-on-FUSE filesystem on there.

I'd be happy to answer any questions you like, and it's pretty easy for me to generate fresh kernel panics at work, so let me know what more I can do to help kill this bug. In the meantime, I have a workaround that seems to do the trick: turn off WiFi using the laptop's hard WiFi kill switch.

[Larry Hastings (larry-hastings)]

Second kernel panic log. (Could I have attached both to a single comment? Too late now.)

[Larry Hastings (larry-hastings)]

Oh! I should have mentioned. The most salient part of the log file is surely right at the top:

[ 524.700004] kernel BUG at /build/buildd/linux-2.6.27/drivers/net/wireless/iwl
wifi/iwl-tx.c:1198!

Finally, I'm current as of updates from earlier today. These logs were generated with Intrepid's 2.6.27-4.

Thanks!

[ltmon (lukemonahan)]

Just to let you know this isn't a completely isolated problem. I haven't gotten a serial console hooked up yet (any easier way to get crash dumps than this?), but I do get freezes with capslock flashing on a T61P. This happened with 8.04 also.

The freeze happens overwhelmingly more often when I'm in a docking station, at work. Possibly this is related to the wireless issue, as even though I'm not using wireless, there's a lot more wireless activity in the general area than when I'm at home. I use wireless exclusively at home, but get no such issues.

I'll monitor this, and post a log when I can get one. I'm also going to try with the hardware wireless switch set to off for a while and see if it's any more stable.

[Leann Ogasawara (leannogasawara)]

Hi Larry,

Care to try installing the linux-backports-modules package. I've had confirmation from other bug reporters who were having kernel panics with the iwlwifi drivers that it resolved the panics. The linux-backports-modules package most recently pulled in an updated version of the compat-wireless stack:

https://edge.launchpad.net/ubuntu/+source/linux-backports-modules-2.6.27

linux-backports-modules-2.6.27 (2.6.27-7.3) intrepid; urgency=low

  [Tim Gardner]

  * Added iwlwifi firmware
  * Added the upstream compat-wireless-2.6 tree.
  * Updated compat-wireless to wireless-testing tag master-2008-10-14
  * Set CONFIG_WIRELESS_OLD_REGULATORY=y by default.

[Mathias Bjerke (mathbje)]

Hi,

I have the exact same problem. After a few minutes Ubuntu 8.10 get a kernel panic, and the Caps Lock keep flashing until I force a reboot.

Did you have any luck with the new iwlwifi drivers?

[ltmon (lukemonahan)]

I've been running with linux-backports-modules-intrepid installed for a few days now without incident. Usually I would get a panic a long time before this, so I'm happy to consider it fixed for my own particular situation in any case.

[Larry Hastings (larry-hastings)]

I think I had *some* luck with the backport, but then I had to give up on it as I got the "scheduling from the idle thread!" log spew bug (launchpad #289581); there's a fix for that in intrepid-proposed, so I switched to that, but that seemed to preclude the backport or something. (I didn't pin the new kernel or anything, I just added intrepid-proposed to my sources.) Anyway, the machine has been stable as of late, though I don't feel like I've given it a wifi torture test since upgrading to -10.