Bug 448151

Summary: Throttling seems weird, doesn't appear to enter C-states, runs hot
Product: [Fedora] Fedora Reporter: James <james>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED CURRENTRELEASE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: low    
Version: 9CC: nhorman
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: 2.6.27.5-32.fc9.x86_64 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-11-12 08:56:52 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:

Description James 2008-05-23 19:30:15 UTC 
Description of problem:
The information under /proc/acpi/processor/CPU?/throttling seems odd. First, the
CPU never seemed to enter even C1. [Normally, I'd expect to see it in C1 some of
the time, I have to boot nohz=off because of another bug: see Bug 443155] This
is probably why the system is idling a good few degrees higher under this kernel. 

Throttling seems different as well. Under kernel-2.6.24.7-92.fc8,
CPU0/throttling would be:

Could not determine current throttling state.

but CPU1/throttling had:

state count:             8
active state:            T0
state available: T0 to T7
states:
   *T0:                  100%
    T1:                  88%
    T2:                  75%
    T3:                  63%
    T4:                  50%
    T5:                  38%
    T6:                  25%
    T7:                  13%

Under 2.6.25.4-10.fc8, CPU0/throttling is now like CPU1 EXCEPT the asterisk is
stuck on the T3 state.

Version-Release number of selected component (if applicable):
kernel-2.6.25.4-10.fc8

How reproducible:
Always.

Steps to Reproduce:
1. Boot.
  
Additional info:
For acpidump, see bug 448150.
Comment 1 James 2008-05-31 00:04:41 UTC 
Same with kernel-2.6.25.4-14.fc8, eats battery.
Comment 2 James 2008-06-13 09:39:27 UTC 
Enters C-states when I remove "nohz=off" with kernel-2.6.25.6-24.fc8, but
throttling oddness remains and still seems to run hotter than
kernel-2.6.24.7-92.fc8.
Comment 3 James 2008-06-17 09:43:05 UTC 
Things still seem odd with kernel-2.6.25.6-27.fc8. powertop reports a few
hundred wakeups per second here and the processor spending a good amount of time
in C3 (but perhaps this is not what the hardware is *actually* doing), so
nothing unusual there... but sometimes powertop reports garbage (e.g., tens of
thousands --- or even a negative number --- of wups) and the sysfs counts for
C-state usage seem to reset every so often.
Comment 4 James 2008-06-25 18:16:43 UTC 
Even odder in kernel-2.6.25.9-37.fc8, throttling for both cores seems suck in T8.
Comment 5 James 2008-07-27 11:42:14 UTC 
Back to same throttling strangeness as 2.6.25.4-10.fc8 with
kernel-2.6.25.12-62.fc8 (see opening comment). kernel-2.6.24.7-92.fc8 runs
cooler on this machine with "nohz=off" than 2.6.25.4-10.fc8 in tickless mode.
Comment 6 James 2008-09-09 07:53:32 UTC 
Still present in kernel-2.6.26.3-14.fc8.
Comment 7 James 2008-11-12 08:56:52 UTC 
Appears to be fixed in kernel-2.6.27.5-32.fc9.x86_64, until a suspend/resume cycle. Will report a new bug.