Resume regression: CPU1 not restarting on Core Duo with 2.6.22-9
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
acpi (Ubuntu) |
New
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: acpi
I'm use Ubuntu Gutsy on a Fujitsu-Siemens Amilo Si-1520.
It has a Core Duo CPU (it has 2 cores).
Suspend to RAM works most of the time with 2.6.22-8 and earlier kernels.
(Sometimes on suspend it hangs with a corrupt text mode screen, but I suspect that is the Intel video driver. Sometimes on resume it simply boots instead of resuming, but this bug report isn't about that).
With 2.6.22-9 kernels, I have found it consistently fails to resume one of the CPU cores, and the caps lock light flashes continuously after resume. /proc/cpuinfo shows only one core after resume, and the CPU frequency monitoring panel applet (which shows both cores normally) crashes.
During resume, a screen full of text errors mentioning ACPI appears quickly. (This doesn't happen with the older kernels), and then X reappears. Most of the error messages are lost, but a capture of /dev/vcs7 is attached.
In /var/log/messages, there's a suggestive message "Error taking CPU1 up: -5". All messages around the time of suspend and resume are attached. Even if there's a failure to restart CPU1, it shouldn't cause a "kernel panic" as appears on vcs7, I guess, but maybe that's the cause of the restart failure.
Ubuntu Gutsy kernels 2.6.22-8 and 2.6.22-7 don't have this problem; I've never seen it with them. 2.6.22-9 has it consistently on every resume.
Thanks
Thanks
The kernel panic message isn't lost. Here's the output of "dmesg" from the start of that suspend to the end of the resume. It's a kernel Oops, unable to handle paging request, and therefore a fairly clear bug.