[FFmpeg-devel] [PATCH] Fix uninitialized reads for fate-vsynth1-asv2 test.

[Clément Bœsch]

On Sun, Jan 15, 2012 at 01:57:07PM +0100, Reimar Döffinger wrote:
> On Sun, Jan 15, 2012 at 01:10:00PM +0100, Michael Niedermayer wrote:
> > On Sun, Jan 15, 2012 at 11:04:23AM +0100, Reimar Döffinger wrote:
> > > This is not a real error and memsetting always even when the
> > > size did not change is overkill, but I'd still like to propose
> > > this.
> > > 
> > > Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
> > 
> > should do no harm, thus ok
> 
> Pushed.
> Let's see how far a full valgrind run (with uninitialized reads not
> disabled).
> Zlib init function seems broken, so I needed a suppression file for that
> already.
> I see the valgrind fate also has a custom suppression file.
> Maybe we should collect these and add a valgrind suppression file into
> git?

The current suppression file is the following:

    % cat fate/eval-strtod.supp
    {
       eval-strtod
       Memcheck:Addr8
       fun:__GI___strncasecmp_l
       fun:____strtod_l_internal
       fun:av_strtod
    }

This helps masking this issue: http://sourceware.org/bugzilla/show_bug.cgi?id=12424

I don't know if lavf-mxf is the blocking the rest of the tests, but we
should be fixed "soon" if you're last patch goes upstream.

About the current uninit read the current patch is fixing, it should be
ignored with --undef-value-errors=no (it's slow enough so I disabled
it...)

BTW, the box get down pretty often and I have to restart the run instance
regulary; also it's more like 5-6 hours a run instead of what I thought
was 1 or 2 ;)

Anyway, I looked into IOC (integer overflow checker) but when I finally
got it working, it leaked so much on just one single test that it wasn't
really usable (OOM killed after a few minutes, the time to fill the swap).
Maybe I did something wrong, but I don't have much time to investigate.

-- 
Clément B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20120115/921cf97a/attachment.asc>