Created attachment 314639 [details] Output of emerge --info I have an HP EliteBook 8440p on which resume after suspend (using sys-power/pm-utils) is failing with a kernel panic in 2/10 cases. I already started debugging the issue and found the kernel either crashing with "BUG: unable to handle kernel paging request at (address)" or "BUG: unable to handle kernel NULL pointer dereference at (address)". The stack trace belonging always seems to be related to a different process. Latest Ubuntu 12.04 is working fine with suspend/resume on this machine. I already tried to track down the problem disabling suspicious hard-/software: * disabled Broadcom WiFi card via BIOS: no effect * unloaded broadcom drivers from kernel before suspend: no effect * try suspend without X running: no effect * unloaded VirtualBox kernel modules from kernel before suspend: no effect I must note that we are using VirtualBox and x11-drivers/nvidia-drivers on this machine. I tested the same setup on the successor of the EliteBook 8440p, the EliteBook 8460p. It uses an Intel WiFi chip and ATI graphics. Here, suspend/resume works fine. Attached you'll find the output of emerge --info, lsmod, and several dumps I made using the serial console.
Created attachment 314641 [details] Output of lsmod
Created attachment 314643 [details] Kernel dumps after resume from suspend
Can you do a fresh boot straight into text mode (i.e. don't load the nvidia stuff) with some suspend/resume cycles, to see if it ever panics without nvidia or virtualbox drivers? Assuming that still shows the problem, next step would be to build your kernel with memory debugging features on and see if you can catch a memory corruption bug closer to when it happens instead of during resume.
(In reply to comment #3) > Can you do a fresh boot straight into text mode (i.e. don't load the nvidia > stuff) with some suspend/resume cycles, to see if it ever panics without > nvidia or virtualbox drivers? > > Assuming that still shows the problem, next step would be to build your > kernel with memory debugging features on and see if you can catch a memory > corruption bug closer to when it happens instead of during resume. Still panics without VirtualBox and nvidia drivers. Any hint on which memory debugging features to use and how to debug?
Have you tried with a vanilla kernel?
Yes I have and I can confirm that it also fails with a vanilla kernel.
Can you please post a kernel oops with VirtualBox not loaded?
That should have read *without* VirtualBox loaded.