On Mon, Jun 01, 2009 at 10:26:38AM +0000, Frederik Himpe wrote:
On Mon, 01 Jun 2009 10:36:36 +0100, Daniel P. Berrange wrote:
> On Sat, May 30, 2009 at 12:45:21PM +0000, Frederik Himpe wrote:
>> On Fri, 29 May 2009 19:18:05 +0200, Daniel Veillard wrote:
>>
>> > - test case for exercising the event loop (Daniel Berrange)
>>
>> I think this test is failing for me on Mandriva Cooker (both i586 and
>> x86_64):
>> What could be wrong?
>
> Strange, this test case works fine for me on both archs with Fedora
> 9/10/11 and RHEL 5.
>
> Could you run the test manually with debugging turned on, eg
>
> LIBVIRT_DEBUG=1 ./eventtest
>
> And also try and capture an strace log with timings
>
> strace -ttt ./eventtest
Actually I found out the test failure is not 100% reproducible. I have
seen it failing in steps 6, 9 and IIRC also step 10 of eventtest.
What I think is happening is that poll() is returning much sooner than
we anticipate. When you give it a timeout, it'll return upto 1 timeslice
before the timeout is due. I suspect that this is what is tripping up
the test case on Mandriva. Could you apply the following patch, and
re-build & run LIBVIRT_DEBUG=1 ./eventtest again, and also re-run with
the 'strace -ttt -f ./eventtest' (i forgot to ask for -f last time).
Regards,
Daniel
--
|: Red Hat, Engineering, London -o-
http://people.redhat.com/berrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org -o-
http://ovirt.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|