kernel panic when plugging in USB device and closing lid

kernel panic when plugging in USB device and closing lid

[Martin Michlmayr]

My HP Compaq 2510p laptop regularly locks up after I move away from my
desk and then return, plug in my USB keyboard and close the lid.  I've
now found an easy way to reproduce this:

 - Boot in single user mode
 - Plug in a USB device (e.g. keyboard or usb stick)
 - Close the lid
 - Open the lid and you see a kernel panic like this (written down by hand):

BUG: unable to handle kernel <1>BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
IP: [<ffffffff8024949f>] ktime_get_ts:0x17/0x48
PGD 7cceb067 PUD 7cd19067 PMD 0
Oops: 0000 [1] SMP
CPU 1
Module linked in: ...
... lots of register info ...
Process swapper (pid: 0, threadinfo fffff81007f3bc000, task ffff81007f3b40c0)

RIP [<ffffffff8024949f>] ktime_get_ts+0x17/x048
 RSP <ffff81007e88fb90>
CR2: 0000000000000000
---[ end trace dd6b99ae04431d83 ]--
Kernel panic -- not syncing: Attempted to kill the idle task!

This is with 2.6.25.  2.6.26-rc5 looks slightly different.  Screen shots:
http://www.cyrius.com/tmp/2.6.25.jpg
http://www.cyrius.com/tmp/2.6.26-rc5.jpg

I briefly talked to Matthew Garrett on IRC who said this "looks like
state corruption".  He asked me to set /proc/acpi/video/blah/DOS to 7
and see whether I could still reproduce the problem.  After setting
this value, I no longer see the kernel panic.  Matthew: "Right.  The
lid-close SMM seems to blow things up some of the time."

Does anyone have any idea what's going on and how to fix it?

-- 
Martin Michlmayr
http://www.cyrius.com/

Re: kernel panic when plugging in USB device and closing lid

[Matthew Garrett]

On Mon, Jun 16, 2008 at 02:32:38PM +0200, Martin Michlmayr wrote:

> I briefly talked to Matthew Garrett on IRC who said this "looks like
> state corruption".  He asked me to set /proc/acpi/video/blah/DOS to 7
> and see whether I could still reproduce the problem.  After setting
> this value, I no longer see the kernel panic.  Matthew: "Right.  The
> lid-close SMM seems to blow things up some of the time."

If DOS is set to 1, the BIOS seems to do something that makes the kernel 
massively unhappy. Disabling the automatic display switch control avoids 
this, which makes me think it's a BIOS bug of some description.

-- 
Matthew Garrett | mjg59@srcf.ucam.org