Bug 113886

interesting (from dmesg after resume): The APIC error. "noapic", remember? ;-)

Restarting tasks... done
Thawing cpus ...
Booting processor 1/1 eip 3000
Initializing CPU#1
Calibrating delay using timer specific routine.. 5586.45 BogoMIPS (lpj=11172906)
CPU: After generic identify, caps: bfebfbff 00000000 00000000 00000000 00004400
00000000 00000000
CPU: After vendor identify, caps: bfebfbff 00000000 00000000 00000000 00004400
00000000 00000000
CPU: Trace cache: 12K uops, L1 D cache: 8K
CPU: L2 cache: 512K
CPU: Physical Processor ID: 0
CPU: After all inits, caps: bfebfbff 00000000 00000000 00000080 00004400
00000000 00000000
Intel machine check architecture supported.
Intel machine check reporting enabled on CPU#1.
CPU1: Intel P4/Xeon Extended MCE MSRs (12) available
CPU1: Thermal monitoring enabled
CPU1: Intel(R) Pentium(R) 4 CPU 2.80GHz stepping 09
APIC error on CPU1: 00(40)
CPU1 is up

Can you try if the same problem happens with suspend-to-RAM? That way we may get
Intel people to fix it... 

APIC error after you used noapic is interesting... There are actually two apics
(OpenAPIC and LocalAPIC), the second one can be turned off with nolapic IIRC.
That may make message go away.

Holger will get the machine, so he will check this ;-)

How so often: Suspend to ram suspends but doesn"t resume. Screen remains black,
no keyboard action possible. Powersave logs say that it resumed correctly.

nolapic (if spelled correctly) does not prevent the apic error to be shown.

suspend to ram might need the "apcpi_sleep=s3_bios",
"acpi_sleep=s3_bios,s3_mode" or even the binary NVidia X driver (and the README
at http://www.susewiki.org/index.php?title=Suspend_NVidia_HOWTO) to get the
display back. If powersave logs say that it resumed correctly, it probably has.
Can you trigger a clean shutdown with the powerbutton after resume? Network access?

in /usr/src/linux/Documentation/power/video.txt the P10-554 is mentioned as

Toshiba Satellite P10-554       s3_bios,s3_mode (4)(****)
(****) Not with SMP kernel, UP only.

but the (****) may be outdated.

So it works even with running X server with "acpi_sleep=s3_bios,s3_mode noapic".
And suprisingly also with smp kernel!

Error message
  APIC error on CPU1: 00(40)
  CPU1 is up
is also shown.

With apic enabled and init=/bin/bash during resume:

Restarting tasks... done
Thawing cpus ...
Booting processor 1/1 eip 3000
Initializing CPU#1
Stuck ??
Inquiring remote APIC #1...
... APIC #1 ID: failed
... APIC #1 VERSION: failed
... APIC #1 SPIV: failed
Error taking cpu 1 up: -22
Kernel panic - not syncing: Not enough cpus

BTW andi is pushing for noapic to be the default, so this issue may go away
automagically :-).

but only on UP, so no, it won't help us :-)

No, as mentioned in comment #7, it works also with smp kernel and noapic. So it
would help. Or did I misunderstood you?

only UP will get noapic, SMP will not IIUC

So, suspend-to-RAM works even in SMP+apic mode, suspend-to-disk does not.  I'll
have to try to reproduce it here. Yes, screenshot of the problem would be nice.

I'll need more info on this one. I tried suspending on SMP machine here, and it works in all modes I tried -- UP/noapic, UP/apic and SMP/apic. It was with pretty recent vanilla kernel, but I do not think this area changed much.

*** Bug 134548 has been marked as a duplicate of this bug. ***

Holger has the machine

I attach an image of the freeze. Picture done with SUSE kernel 2.6.16-2-smp and init=/bin/bash. The kernel panic appears after reloading the image. No 'noapic' given.

BTW: s2ram has exactly the same error. If giving noapic as boot parameter s2disk and s2ram work good if acpi_sleep=s3_bios,s3_mode is given (s2ram -a 3).

Created attachment 74375 [details]
Image of kernel panic

Thanks...

...we try to kick second CPU but fail for some reason. We could actually continue in this case (not panic) -- one CPU is enough :-). Backtrace above is innocent.

Could you try to emulate hotplug with echo > /sys/... ? That should produce failure, too. Then, I'm afraid this needs to go to bugzilla.kernel.org, so that Intel/IBM people take a look. This is more problem with underlying cpu hotplug subsystem than with suspend... (And intel people actually care about suspend to ram).

I already tried CPU hotplugging. It works as expected. I can enable and disable cpu1 as often as I like.

Bug filed upstrea:
http://bugzilla.kernel.org/show_bug.cgi?id=6270

Thanks, lets see if Intel people can debug it. What is the bugzilla state for "being solved in another bugzilla"? ;-)

Not sure either. But WONTFIX seems to be fit most, because you actually won't fix it in regard to this bug ;-)

Ok, agreed. Hopefully intel people can solve it.