Bug 3518

Summary:	disable event LID blocks SLPB and vice versa on wakeup from S3
Product:	ACPI	Reporter:	Pawel Maziarz (drg)
Component:	Power-Sleep-Wake	Assignee:	Shaohua (shaohua.li)
Status:	CLOSED CODE_FIX
Severity:	blocking	CC:	lenb
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	2.6.9-rc3, 2.6.9-rc3-bk6, 2.6.9-rc3-mm2	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	acpidmp proposed patch patch to warnning user patch to warnning user

Description Pawel Maziarz 2004-10-07 06:39:34 UTC

Distribution: Slackware 10
Hardware Environment: Notebook Aristo 2500, CPU Celeron 2.40Ghz
Problem Description:

When i go into S3 mode (echo mem >/sys/power/state), i can wakeup
by pressing SLPB (Fn+ESC on my laptop) or opening LID.
When i do that on a working system, kernel sends events
"button/sleep SLPB 00000080 00000000x" when i'm pressing Fn+ESC
and "button/lid LID 000000080 00000000x" when i'm openning or
closing LID.
But, when i disable LID (echo LID >/proc/acpi/wakeup) *or*
SLPB (echo SLPB >/proc/acpi/wakeup), i do cat /proc/acpi/wakeup
and i see that LID or SLPB is disabled, i can't go back
from S3 state.

Steps to reproduce:
echo LID >/proc/acpi/wakeup or echo SLPB >/proc/acpi/wakeup
and echo mem >/sys/power/state

Comment 1 Shaohua 2004-10-07 19:45:07 UTC

What is the problem? does disabling LID also make SLPB disabled and vice 
versa? If yes, please attach your acpidmp.

Comment 2 Pawel Maziarz 2004-10-07 23:10:50 UTC

Created attachment 3781 [details]
acpidmp

yes, disabling LID also make SLPB disabled and vice versa.

Comment 3 Shaohua 2004-10-08 00:02:53 UTC

Created attachment 3782 [details]
proposed patch

The LID and SLPB have the same wake GPE. Disabling one might also disable
another. The attached patch tries to resolve the issue. Please try.

Comment 4 Pawel Maziarz 2004-10-08 00:52:39 UTC

unfortunately with this patch, when LID or SLPB are disabled, both can wake up system.

Comment 5 Shaohua 2004-10-08 01:11:01 UTC

It's expected behavior. As I said, the LID and SLPB have the same wake GPE, 
you can't enable one but disable the other (it's the hardware design, software 
can't change it). I think both work is better than both failed.

Comment 6 Len Brown 2004-11-10 18:57:03 UTC

patch needs a warning so that user knows why
disabling sleep button didn't work
and that he also needs to disable LID button
to disable sleep button...

Comment 7 Shaohua 2004-11-10 21:37:40 UTC

Created attachment 4008 [details]
patch to warnning user

Ok, a new patch. It warnning user if the problem occurs, and disable/enable all
devices with same GPE, and let user to select.
Such as if BAT0 and BAT1 has the same GPE 1,
echo 'BAT0' >/proc/acpi/wakeup, an warnning reported and both BAT0 and BAT1 are
enabled, so user can choose how to do next step.

Comment 8 Shaohua 2004-11-10 21:50:51 UTC

Created attachment 4009 [details]
patch to warnning user

Oh, possibly more than two devices have the same GPE, the patch considers the
situation.

Comment 9 Len Brown 2004-12-05 20:38:39 UTC

patch in comment #8 applied (w/ printk syntax update) to acpi-test tree

Comment 10 Len Brown 2005-01-26 23:08:20 UTC

shipped in linux-2.6.11-rc1 -- closing.