Bug 525147

Summary: Lockups when using performance counters
Product: [Fedora] Fedora Reporter: Søren Sandmann Pedersen <sandmann>
Component: kernelAssignee: Kernel Maintainer List <kernel-maint>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: medium Docs Contact:
Priority: low    
Version: rawhideCC: dougsland, gansalmon, itamar, kem, kernel-maint, sandmann
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2009-10-16 10:04:14 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
/proc/cpuinfo
none
The attached program
none
A version that compiles without warnings. none

Description Søren Sandmann Pedersen 2009-09-23 14:40:26 UTC 
Description of problem:

The attached programs sometimes causes lockups. Generally I have to run immediately after logging in and rebooting to provoke the lockup.

The system is a quad core Core 2; I'll attach cpuinfo.
Comment 1 Søren Sandmann Pedersen 2009-09-23 14:42:43 UTC 
Created attachment 362307 [details]
/proc/cpuinfo
Comment 2 Chuck Ebbert 2009-09-24 03:24:52 UTC 
(In reply to comment #2)
> Description of problem:
> 
> The attached programs sometimes causes lockups. Generally I have to run
> immediately after logging in and rebooting to provoke the lockup.

What attached programs?
Comment 3 Søren Sandmann Pedersen 2009-09-24 08:59:18 UTC 
Created attachment 362449 [details]
The attached program

This one, sorry. I'm not sure what went wrong.
Comment 4 Chuck Ebbert 2009-09-24 15:07:43 UTC 
    int cpu = (int)data;

should be

    int cpu = (int)*data;

You should probably build with -Wall and fix every warning before looking for bugs.
Comment 5 Søren Sandmann Pedersen 2009-09-24 15:23:31 UTC 
Created attachment 362515 [details]
A version that compiles without warnings.

Here is a version that compiles without warnings with -Wall.

Note that I am using this:

    pthread_create (&t, NULL, thread, (void *)i);

to cast a small integer to a pointer, then casting back later. There is nothing wrong with that, I don't think.
Comment 6 Chuck Ebbert 2009-09-27 10:30:50 UTC 
Can you try with a 2.6.31.1 kernel, which has several perfcounter fixes? Is it completely wedged to the point that the sysrq keys don't work?
Comment 7 Søren Sandmann Pedersen 2009-10-16 10:04:14 UTC 
It is completely wedged, yes. I haven't seen it since upgrading to 2.6.31.1, so maybe it's gone.