[FFmpeg-devel] [PATCH 7/8] tests: Add aac_fixed decoder test

[Nedeljko Babic]

>>On Sat, Jul 25, 2015 at 05:37:56AM +0200, Michael Niedermayer wrote:
>>> On Mon, Jul 20, 2015 at 01:36:21PM +0200, Nedeljko Babic wrote:
>>> > Signed-off-by: Nedeljko Babic <nedeljko.babic at imgtec.com>
>>> > ---
>>> >  tests/fate/aac.mak | 58 +++++++++++++++++++++++++++++++++++++++++++++++++++++-
>>> >  1 file changed, 57 insertions(+), 1 deletion(-)
>>> 
>>> these tests fail on clang with -ftrapv
>>> see
>>> http://fate.ffmpeg.org/report.cgi?time=20150723144015&slot=x86_64-freebsd10-clang33-ftrapv
>>
>>i fixed one ftrapv issue but a 2nd remains, it seems the code
>>filters uninitialized and then not used data. Please fix this
>>unless someone else does before
>>
>
>It looks like all the problems (valgrind and clang) have the same cause.
>Furthermore if I enable memory poisoning, there are no problems under valgrind.
>
>I will try to fix this in next few hours, but if I don't finish, I will not be able to work on this in next few weeks.
>

Apparently, these are two different problems after all.

I am sending fix for a bug that pops out with the valgrind and –disable-memory-poisoning.

For problem with clang some more debugging is needed.

>>also SBR seems to fails on ARM
>>http://fate.ffmpeg.org/report.cgi?time=20150725003242&slot=armv7l-panda-gcc4.6-armv5
>>it does pass under qemu ARM though it seems
>>i did test your code on qemu arm not actual arm before i pushed, didnt
>>thought that would make a difference
>
>I don't have ARM hardware at the moment to work on this.
>Maybe it is connected with above problem...

- Nedeljko