Bug 804370

Summary: High energy comsumption after last update
Product: [Fedora] Fedora Reporter: Martin Angermeier <martinangermeier>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED NOTABUG QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 16CC: gansalmon, itamar, jonathan, kernel-maint, madhu.chinakonda, martinangermeier
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2012-03-20 15:19:06 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
Yum.log 03-18-2012 none

Description Martin Angermeier 2012-03-18 11:09:07 UTC
Created attachment 570877 [details]
Yum.log 03-18-2012

Description of problem:
After updating to kernel version 3.2.10-3 I got around 25W energy consumption on my notebook, up from 8,5W (both Idle). I got a Sony Vaio-S with an Intel i3 (Sandy Bridge) and hybrid graphics (AMD HD 6450). I enabled RC6 and switched off the discrete gpu (AMD) using vgaswitcheroo. 


Version-Release number of selected component (if applicable):
kernel 3.2.10-2

How reproducible:
Every boot


Steps to Reproduce:
1. boot up
2. idle system
3. switch on energy monitor or powertop
  
Actual results:
~ 25W power consumption 

Expected results:
~ 8,5W

Additional info:
latest yum update attached

Comment 1 Martin Angermeier 2012-03-18 14:23:42 UTC
PS: 

Version-Release number of selected component (if applicable):
kernel 3.2.10-2 (mistype, should be 3.2.10-3)

Comment 2 Dave Jones 2012-03-19 15:13:35 UTC
does powertop show anything interesting ?

Comment 3 Martin Angermeier 2012-03-19 16:59:33 UTC
Well, that's weird. Ich booted up today, updated the following packages:

Mar 18 19:44:07 Updated: rpmfusion-nonfree-release-16-3.noarch
Mar 18 21:13:29 Installed: lame-libs-3.98.4-1.fc14.x86_64
Mar 18 21:13:30 Installed: x264-libs-0.0.0-0.31.20110811.fc16.x86_64
Mar 18 21:13:31 Installed: a52dec-0.7.4-15.fc11.x86_64
Mar 18 21:13:32 Installed: libsidplay-1.36.60-2.fc15.x86_64
Mar 18 21:13:32 Installed: librtmp-2.4-0.1.20110811gitc58cfb3e.fc16.x86_64
Mar 18 21:13:33 Installed: libmad-0.15.1b-13.fc12.x86_64
Mar 18 21:13:34 Installed: xvidcore-1.3.2-2.fc15.x86_64
Mar 18 21:13:35 Installed: libva-1.0.15-1.fc16.x86_64
Mar 18 21:13:36 Installed: ffmpeg-libs-0.8.9-1.fc16.x86_64
Mar 18 21:13:37 Installed: libid3tag-0.15.1b-11.fc15.x86_64
Mar 18 21:13:38 Installed: opencore-amr-0.1.2-1.fc12.x86_64
Mar 18 21:13:38 Installed: libmpeg2-0.5.1-8.fc12.x86_64
Mar 18 21:13:39 Installed: twolame-libs-0.3.13-1.fc16.x86_64
Mar 18 21:13:40 Installed: gstreamer-plugins-ugly-0.10.18-3.fc16.x86_64
Mar 18 21:13:41 Installed: gstreamer-ffmpeg-0.10.13-1.fc16.x86_64
Mar 19 08:19:08 Updated: rpmfusion-free-release-16-3.noarch
Mar 19 08:19:11 Updated: ffmpeg-libs-0.8.10-1.fc16.x86_64
Mar 19 17:43:09 Updated: liblouis-2.4.1-1.fc16.x86_64
Mar 19 17:43:09 Updated: liblouis-python-2.4.1-1.fc16.noarch

and now I got an energy consumption around 9W, quite normal. 
I don't know why this happened. As I wrote the bug, my GPU was running on full-speed, CPU 800MHz (power save and deepest idle state). PowerTOP shows normal values, just a slight increase from 3.2.9

Comment 4 Martin Angermeier 2012-03-20 10:47:25 UTC
Daily update:

Today 12W in idle (Firefox with Facebook, pdf-Reader, display brightness 33%)

Powertop shows something interesting, 3% C0 power state for cpu0, 18% for cpu1 and ~900ms CPU time for
/usr/libeexec/tracker-miner-fs and ~ 1000 pkts/s for network-interface wlan0

Under 3.2.9 today 8W

Comment 5 Martin Angermeier 2012-03-20 10:55:54 UTC
It seems that cpupower is changing CPU-states more quickly.
Governor is set to powersave but it's behaviour is more like ondemand it seems

Comment 6 Dave Jones 2012-03-20 15:19:06 UTC
that profile is anything but idle.  even if you're not touching anything, firefox and facebook will be using cpu (they use ridiculous amounts of javascript).

also tracker running is obviously going to draw power as it runs around indexing things.

I don't see anything kernel related here that needs fixing.
If it does spike up again, do some profiling with powertop/top/iotop and find out what it's doing. If the machine really is completely idle and pulling lots of power, perhaps we can do something.