Bug 120737

Summary:	ACPI power button does not work
Product:	[openSUSE] SUSE Linux 10.1	Reporter:	Michael Stather <kontakt>
Component:	Kernel	Assignee:	Thomas Renninger <trenn>
Status:	RESOLVED WONTFIX	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None
Version:	Alpha 1
Target Milestone:	---
Hardware:	Other
OS:	All
Whiteboard:
Found By:	Other	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	acpidmp with ec_burst=0 acpidmp with ec_burst=1 dmesg with ec_burst=0 dmesg with ec_burst=1

Description Michael Stather 2005-10-06 19:17:52 UTC

When I press the power button on my system it doesn´t shut down even if it´s
configured to do so. It´s an Athlon 64 with Ali chipset and ACPI enabled. Which
logs are relevant for that?

Comment 1 Thomas Renninger 2005-10-16 14:06:43 UTC

Stop, acpid, hald, dbus and powersaved services.
Then do:
modprobe button
cat /proc/acpi/event
Then press the power button -> there should a line appear telling you that the power button has been pressed.
If not, it is a kernel bug.
If there appears something it is a userspace bug.

Comment 2 Michael Stather 2005-10-16 22:49:44 UTC

I did what you suggested and there´s no output, so it must be a kernel bug.

Comment 3 Thomas Renninger 2005-10-17 08:44:28 UTC

Hmm, this could be difficult.
Could you try with the boot parameter ec_burst=0. Ohh, this is 10.1 not sure whether ec_burst is on in mainline, best you also try ec_burst=1. Does it work now?
Best is you always try as suggested in comment #1, so we know that no userspace program (which normally fetches the events from /proc/acpi/event) swallows them.
If you do: cat /proc/acpi/event  the program must hang until an event occurs, print it and still hang. If you cannot access the file, try fuser /proc/acpi/event to find out what program does access it.

Please also attach dmesg output (with and without this parameter) and acpidmp output.

Could you describe your hardware a bit. x86_64? laptop? ...

Comment 4 Michael Stather 2005-10-17 22:08:54 UTC

I checked with both ec_burst 0 and 1 but there´s no acpi event.
The requested files are attached.
I´ve an Athlon64 3200+ (with 32bit linux) with an AsRock Socket 939 Dual SATAII board

Comment 5 Michael Stather 2005-10-17 22:09:21 UTC

Created attachment 54445 [details]
acpidmp with ec_burst=0

Comment 6 Michael Stather 2005-10-17 22:09:39 UTC

Created attachment 54446 [details]
acpidmp with ec_burst=1

Comment 7 Michael Stather 2005-10-17 22:09:53 UTC

Created attachment 54447 [details]
dmesg with ec_burst=0

Comment 8 Michael Stather 2005-10-17 22:10:05 UTC

Created attachment 54448 [details]
dmesg with ec_burst=1

Comment 9 Thomas Renninger 2005-12-21 10:53:56 UTC

Sorry, for answering that late... there were a lot other bugs with higher priority...
Does this still happen with recent 10.1 version?
What does /var/log/messages say if the button module is loaed?
Do you see something like:
ACPI: Power Button (FF) [PWRF]
ACPI: Power Button (CM) [PWRB]
or just nothing?

Comment 10 Michael Stather 2005-12-21 12:12:10 UTC

I don´t have this board any more (changed it due to various other problems) but since its overall quality is really crappy I´m sure it´s a BIOS bug (there are much more BIOS bugs in this board...)

Comment 11 Thomas Renninger 2005-12-21 12:38:12 UTC

It's the first report I got for ACPI power/sleep buttons for long, so it is most likely a BIOS bug...
Thanks for reporting, anyways! 
Especially recent 10.1 reports are very welcome.