Bug 12019

Summary:	Major increase in the number of wakeups from C3
Product:	ACPI	Reporter:	Mircea Gherzan (mgherzan)
Component:	Power-Processor	Assignee:	power-management_other
Status:	CLOSED CODE_FIX
Severity:	normal	CC:	rjw, venki
Priority:	P1
Hardware:	All
OS:	Linux
Kernel Version:	2.6.28-rc3	Subsystem:
Regression:	Yes	Bisected commit-id:
Bug Depends on:
Bug Blocks:	11808
Attachments:	.27 boot .28 boot .28 loaded modules .27 powertop output, 20 seconds .28 powertop output, 20 seconds -rc5 powertop output

Description Mircea Gherzan 2008-11-13 07:02:38 UTC

Latest working kernel version: 2.6.27.5
Earliest failing kernel version: 2.6.28-rc3
Distribution: Debian Sid
Hardware Environment: Fujitsu-Siemens Lifebook E8110, Intel T2400, 1.5G RAM, GMA950, iwl3945
Software Environment: Gnome 2.22.3
Problem Description:

The processor is woken up from C3 by "extra timer interrupt" around 100 times per second. In .27, under the same conditions, I am able to achieve approx 10 wakeups per second, therefore better C3 residency times.

The testing was done with the RF kill switch ON.

Steps to reproduce: run powertop, apply all hints.

Comment 1 Mircea Gherzan 2008-11-13 07:03:18 UTC

Created attachment 18845 [details]
.27 boot

Comment 2 Mircea Gherzan 2008-11-13 07:03:35 UTC

Created attachment 18846 [details]
.28 boot

Comment 3 Mircea Gherzan 2008-11-13 07:04:00 UTC

Created attachment 18847 [details]
.28 loaded modules

Comment 4 Mircea Gherzan 2008-11-13 07:05:00 UTC

Created attachment 18848 [details]
.27 powertop output, 20 seconds

Comment 5 Mircea Gherzan 2008-11-13 07:05:23 UTC

Created attachment 18849 [details]
.28 powertop output, 20 seconds

Comment 6 Rafael J. Wysocki 2008-11-16 09:47:23 UTC

Notify-Also : Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>

Comment 7 Rafael J. Wysocki 2008-11-16 09:48:55 UTC

Notify-Also : Thomas Gleixner <tglx@linutronix.de>

Comment 8 ykzhao 2008-11-16 18:03:37 UTC

Hi, Mircea
    From the log in comment #28 it seems that the system is waken up from C3 by "extra timer interrupt" about 117 times. It seems that the above test is done under graphics mode. (In the 2.6.28-rc3 kernel there exists the following info. "hrtimer_wakeup").
    Will you please do the above test under console mode and see whether the problem still exists? When doing test under console mode, please don't load i915 drm driver.
   
   Thanks.

Comment 9 Andreas Mohr 2008-11-17 07:35:24 UTC

Duplicate of #11905 ? If so, then it should work on -rc5.

Comment 10 Mircea Gherzan 2008-11-17 15:36:39 UTC

Created attachment 18899 [details]
-rc5 powertop output

Comment 11 Mircea Gherzan 2008-11-17 15:39:35 UTC

Fixed in -rc5, as seen in the attachment from comment #10.