[FFmpeg-devel] [PATCH] fate-valgrind.supp: Suppress uninitialized warnings for h264.

Reimar Döffinger Reimar.Doeffinger at gmx.de
Fri Feb 17 18:20:11 CET 2012


On Fri, Feb 17, 2012 at 06:06:42PM +0100, Michael Niedermayer wrote:
> > > > Option 1) Fix the uninitialized data source
> 
> theres not really a uninitialized data source

Which you did not say, neither the message nor the patch itself
contained any hint that you investigated this at all nor
what the conclusions are.
I still don't think it a great idea to reduce the usefulness
of the whole valgrind test for those "few" failing ones,
but after finding the cause, fine.
But only with _proper_ documentation what this is about.
And possibly a bug report to valgrind - I am not sure
they will want to fix it, but there is a chance that they
just forgot about the redzone special-case when porting the
code from x86 to x86_64.

> > > > Option 3) Add those suppressions only for the files with the issue
> > > 
> > > Might be the quickest, smuggle a $TEST_VG_OPTS into the generated
> > > valgrind command-line and have those tests set that variable.
> > > (will still need some time/fiddling to get working I'm sure)
> 
> sounds like hacks that noone will maintain

Well, it would offer a solution to the no-undef FATE test not being
able to use --valgrind (though I actually think that is not true,
it should be able to use --valgrind='valgrind --undef-value-errors=no').


More information about the ffmpeg-devel mailing list