Phantom powerbutton event?

[jukka holm]

PC randomly shutdowns itself on Hoary, same pc works fine with warty and other
distributions and operating systems. My debugging has found out that somehow the
powerbutton is beeing pressed till the pc shuts down. I have latest updates, the
bug has been around as long as Ive used hoary (about 2 months). Tried both k7 &
386 kernel versions. Tried noacpi kernel option. Tried disable powernowd. The
shutdown happens after 1 to 24 hours of uptime. No matter if im using pc or not.

I replaced /etc/acpi/powerbutton.sh with a script that logs date & ps -ef
output, it tells that the script is being run 20-30 times in 2-4 seconds and
then pc shuts down.

Dont know if reporting a bug is the right way to deal with this, but I am more
than eager to find a solution and I feel kinda helpless without support.

http://ubuntuforums.org/showthread.php?p=89003#post89003: http://ubuntuforums.org/showthread.php?p=89003#post89003

[Matt Zimmerman (mdz)]

I would expect acpi=off to work around this, but in the forum you said that your
system failed to boot with acpi=off.

Please attach dmesg output, both with and without acpi=off, and the output from
"sudo dmidecode".

Have you checked whether a BIOS upgrade is available from the manufacturer?

[jukka holm (jukka-holm-gmail)]

Created an attachment (id=1589)
Dmesg output

[jukka holm (jukka-holm-gmail)]

Created an attachment (id=1590)
Dmiedecode output

[jukka holm (jukka-holm-gmail)]

(In reply to comment #1)
> I would expect acpi=off to work around this, but in the forum you said that your
> system failed to boot with acpi=off.

System hangs on "Initialising Advanced Power Management Daemon" if using
acpi=off. This is with hoary.

>
> Please attach dmesg output, both with and without acpi=off, and the output from
> "sudo dmidecode".

Cant get those with acpi=off. Attaching without.

> Have you checked whether a BIOS upgrade is available from the manufacturer?

No updates since 2002.

What does nolacpi option actually do? I have had no crash with that option yet.
Over 24h uptime, its not exceptional, but its good.

[jukka holm (jukka-holm-gmail)]

(In reply to comment #4)
> What does nolacpi option actually do? I have had no crash with that option yet.
> Over 24h uptime, its not exceptional, but its good.

Nevermind, just took a bit longer than usually.

This time for some reason the powerbutton.sh had not been run (no log created).

[jukka holm (jukka-holm-gmail)]

There are now two others complaining kind of a similar problem at the
ubuntuforums http://ubuntuforums.org/showthread.php?t=18345

[Fabio Massimo Di Nitto (fabbione)]

Ok let's try to isolate one problem at a time.

Can you guys please try to install mbmon from universe and run it with output
redirected to some log files?

note that it might or might not work, but it is still a test that we can do
to isolate the temperature problem.

If the machine stops again, please add the output from mbmon and let see
if the cpu was overheating or something.

Fabio

PS mbmon does not need any kernel module. It either works or not on your mobo.

[jukka holm (jukka-holm-gmail)]

OK, Im running it right now. Ill post when the next shutdown occurs.

[jukka holm (jukka-holm-gmail)]

So it took a little longer.. One thing less to suspect now - temperature log
didnt show up any variation before shutdown.

[jukka holm (jukka-holm-gmail)]

Seems like others are suffering from this:

http://ubuntuforums.org/showthread.php?p=102295

[jukka holm (jukka-holm-gmail)]

For some reason not untill now I hadnt look for /var/log/acpid . Anyhow, dont
know if this makes any difference, but it is full of this:
[Mon Mar 21 19:39:30 2005] BEGIN HANDLER MESSAGES
[Mon Mar 21 19:39:30 2005] END HANDLER MESSAGES
[Mon Mar 21 19:39:30 2005] action exited with status 0
[Mon Mar 21 19:39:30 2005] completed event "button/power PWRF 00000080
00000046"[Mon Mar 21 19:39:30 2005] received event "button/power PWRF 00000080
00000047"
[Mon Mar 21 19:39:30 2005] executing action "/etc/acpi/powerbtn.sh"
[Mon Mar 21 19:39:30 2005] BEGIN HANDLER MESSAGES
[Mon Mar 21 21:42:02 2005] starting up
[Mon Mar 21 21:42:02 2005] 21 rules loaded

Any way to disable the powerbutton event somehow? Removing the
/etc/acpi/powerbtn.sh doesnt help. acpi=off would propably help, but it hangs on
the part where it tries to load apm-daemon. How could I debug that? Any help
appreciated - Im willing to give any help/info I can.

In my humble opinion this is quite serious problem, as it seems that others are
suffering fom it also.

[Matthew Garrett (mjg59)]

Could you attach the contents of /proc/acpi/dsdt please?

[jukka holm (jukka-holm-gmail)]

Created an attachment (id=1804)
cat /proc/acpi/dsdt > dsdt

[Matthew Garrett (mjg59)]

Ok. As a vague attempt at a workaround, does this happen if you remove the
button module?

[jukka holm (jukka-holm-gmail)]

OK, now I have modprobe -r button and it doesnt show up anymore in lsmod. Lets
wait and see what happens.

[jukka holm (jukka-holm-gmail)]

Now I have over 50h uptime with the button module unloaded. Lets wait couple
days more and hope that this "workaround" really works.

[jukka holm (jukka-holm-gmail)]

No, removing button module did not stop the computer from shutting down.

[jukka holm (jukka-holm-gmail)]

This was hardware related afterall. There were slight overclock on cpu which did
not cause any problems in any other environments, even had 2 months uptime with
warty with the same settings. Have not had the problem anymore.