Bug 45921

Summary:	resume from hibernate fails when "Suspend to RAM and standby" is disabled
Product:	Power Management	Reporter:	Andreas Steinmetz (ast)
Component:	Hibernation/Suspend	Assignee:	Lan Tianyu (tianyu.lan)
Status:	CLOSED CODE_FIX
Severity:	normal	CC:	florian, rjw, rui.zhang, tianyu.lan
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	3.5.[01]	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	Kernel config Modules loaded Broken config do acpisleep dmi check if CONFIG_ACPI_SLEEP is set refreshed patch based on Rafael' comments

Description Andreas Steinmetz 2012-08-13 23:13:07 UTC

The system in question can't suspend to ram, resume results in a reset and the system hanging in a broken bios (power off required).

Hibernate, however, works, but only, if CONFIG_SUSPEND=y and CONFIG_SUSPEND_FREEZER=y, otherwise resume from hibernate fails with two displays full of vertical bars and an otherwise fully locked up system.

The difference, as far as I can remember, is that if both options are set these vertical bars appear twice shortly on the main display during hibernation, when the options are not set they don't seem to appear.

The system is an Ivy Bridge system using the integrated HD4000 graphics, so i915 is in use.

The behavior is the same with 3.5.0 and 3.5.1. Kernel config attached.

Comment 1 Andreas Steinmetz 2012-08-13 23:14:04 UTC

Created attachment 77641 [details]
Kernel config

Comment 2 Andreas Steinmetz 2012-08-13 23:14:30 UTC

Created attachment 77651 [details]
Modules loaded

Comment 3 Rafael J. Wysocki 2012-08-14 23:16:31 UTC

May I see the .config that doesn't work too, please?

Comment 4 Andreas Steinmetz 2012-08-15 00:26:05 UTC

Created attachment 77731 [details]
Broken config

As per diff to the working config:

@@ -438,8 +438,7 @@
 # Power management and ACPI options
 #
 CONFIG_ARCH_HIBERNATION_HEADER=y
-CONFIG_SUSPEND=y
-CONFIG_SUSPEND_FREEZER=y
+# CONFIG_SUSPEND is not set
 CONFIG_HIBERNATE_CALLBACKS=y
 CONFIG_HIBERNATION=y
 CONFIG_PM_STD_PARTITION=""

Shouldn't affect hibernate, but unfortunately does.

Comment 5 Zhang Rui 2012-11-28 14:20:59 UTC

please attach the dmi output of this laptop.

Comment 6 Zhang Rui 2012-11-28 14:55:44 UTC

Created attachment 87551 [details]
do acpisleep dmi check if CONFIG_ACPI_SLEEP is set

please try this patch and see if it helps.

Comment 7 Zhang Rui 2012-11-29 04:44:50 UTC

Created attachment 87681 [details]
refreshed patch based on Rafael' comments

please try this patch instead.

Comment 8 Florian Mickler 2012-12-22 09:20:17 UTC

A patch referencing this bug report has been merged in Linux v3.8-rc1:

commit 0ac1b1d7b7424cd6f129b5454b504b3cae746f0e
Author: Zhang Rui <rui.zhang@intel.com>
Date:   Fri Nov 30 12:57:03 2012 +0100

    ACPI: do acpisleep dmi check when CONFIG_ACPI_SLEEP is set

Comment 9 Lan Tianyu 2013-03-31 14:22:45 UTC

ping ... Does Comment #8 patch fix the issue?

Comment 10 Lan Tianyu 2013-04-09 06:09:42 UTC

Since there is already a fix bug patch and no response. Close the bug as code fix and if still exit in the last upstream kernel. Please feel free to reopen.