Bug 9344

Summary:	2.6.24 regression: hibernation hangs on "Suspending console" in low-battery condition
Product:	Power Management	Reporter:	Rafael J. Wysocki (rjwysocki)
Component:	Hibernation/Suspend	Assignee:	Alexey Starikovskiy (astarikovskiy)
Status:	REJECTED UNREPRODUCIBLE
Severity:	normal	CC:	acpi-bugzilla, arvidjaar, astarikovskiy, mingo
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.24-rc1	Subsystem:
Regression:	Yes	Bisected commit-id:
Bug Depends on:
Bug Blocks:	7216, 9243
Attachments:	Always return valid status

Description Rafael J. Wysocki 2007-11-11 10:15:54 UTC

Subject         : 2.6.24 regression:  hibernation hangs on "Suspending console" in low-battery condition
Submitter       : Andrey Borzenkov <arvidjaar@mail.ru>
References      : http://lkml.org/lkml/2007/11/11/28
Handled-By      : "Rafael J. Wysocki" <rjw@sisk.pl>

Comment 1 Rafael J. Wysocki 2007-11-19 10:27:40 UTC

Suspend-to-RAM in normal battery conditions (or without battery)
works. With 5% threshold there's a visible indication of alarm from *firmware* (battery LED flashing) and the subsequent "emergency" suspend fails.

Comment 2 Alexey Starikovskiy 2007-11-19 10:34:56 UTC

Could you please try to comment out battery_update at battery_notify()?

Comment 3 Alexey Starikovskiy 2007-11-19 10:56:01 UTC

Created attachment 13623 [details]
Always return valid status

There is also this patch, might help.

Comment 4 Rafael J. Wysocki 2007-11-20 16:52:48 UTC

Andrey?

Comment 5 Andrey Borzenkov 2007-11-20 19:54:47 UTC

I already used this patch in rc2 - it made no difference (as I mentioned in original mail thread). In rc3 I am unable to trigger this problem with or without this patch. I tried to set BAT1/alarm to some higher value but either I do not understand what it does or it was actually red herring.

Well if it won't be back in rc4 I'd say it has to be removed from list of regressions.

I had to test and you understand, it takes several hours for battery to discharge so I can do it once a day at the most and I have to do it several times to be sure ... it takes time, you know ...

Comment 6 Ingo Molnar 2007-12-04 03:26:25 UTC

I'm closing this regression as suggested by Andrey - please re-open it if you see this problem agan.