[FFmpeg-devel] [PATCH] tests/fate/filter-video: increase fuzz for fate-filter-refcmp-psnr-rgb

[Michael Niedermayer]

On Thu, Dec 06, 2018 at 10:42:24PM +0100, Carl Eugen Hoyos wrote:
> 2018-12-06 22:41 GMT+01:00, Carl Eugen Hoyos <ceffmpeg at gmail.com>:
> > 2018-12-06 22:19 GMT+01:00, Michael Niedermayer <michael at niedermayer.cc>:
> >> Fixes: test failure on powerpc
> >>
> >> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> >> ---
> >>  tests/fate/filter-video.mak | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/tests/fate/filter-video.mak b/tests/fate/filter-video.mak
> >> index 8bbdc04896..1042e96e54 100644
> >> --- a/tests/fate/filter-video.mak
> >> +++ b/tests/fate/filter-video.mak
> >> @@ -782,7 +782,7 @@ fate-filter-meta-4560-rotate0: CMD = framecrc -flags
> >> +bitexact -c:a aac_fixed -i
> >>  REFCMP_DEPS = FFMPEG LAVFI_INDEV TESTSRC2_FILTER AVGBLUR_FILTER
> >> METADATA_FILTER
> >>
> >>  FATE_FILTER_SAMPLES-$(call ALLYES, $(REFCMP_DEPS) PSNR_FILTER) +=
> >> fate-filter-refcmp-psnr-rgb
> >> -fate-filter-refcmp-psnr-rgb: CMD = refcmp_metadata psnr rgb24 0.001
> >> +fate-filter-refcmp-psnr-rgb: CMD = refcmp_metadata psnr rgb24 0.002
> >
> > This is not needed on any ppc system where I tested recently including
> > Power7 Linux BE and Power8 Linux LE and aix BE.
> 
> This is of course not meant to stop this patch, I am just wondering how
> to reproduce.

Yes, let me start from mostly the begin

configure --assert-level=2 --target-exec='qemu-201812/build/ppc64abi32-linux-user/qemu-ppc64abi32 -L /usr/powerpc-linux-gnu/' --cpu=power8 --samples=fate-suite/ --enable-gpl --cross-prefix=/usr/powerpc-linux-gnu/bin/ --cc='ccache powerpc-linux-gnu-gcc-4.8' --arch=ppc --target-os=linux --enable-cross-compile --disable-iconv --disable-pthreads

make -k  -j12  fate-filter-pp2 fate-filter-pp3 fate-wavpack-cuesheet  fate-lossless-monkeysaudio-399 fate-eval fate-parseutils 
make: *** [fate-parseutils] Error 1
make: *** [fate-eval] Error 1
make: *** [fate-wavpack-cuesheet] Error 1
make: *** [fate-filter-pp3] Error 1
make: *** [fate-filter-pp2] Error 1

(fate-lossless-monkeysaudio-399 seems to have stoped failing, maybe because 
 --cpu is different than when i tested this previously)

Oddly adding --extra-cflags=-funsafe-math-optimizations
fixes 3 of these, only these remain
make: *** [fate-parseutils] Error 1
make: *** [fate-eval] Error 1
And not just that but the failures in fate-eval are significantly fewer

but it added these failures:
make: *** [fate-filter-tremolo] Error 1
make: *** [fate-filter-refcmp-psnr-rgb] Error 1

and slightly increasing the fuzz factor eliminated the second which is
what the patch does
I think the patch makes sense independant of this long story as it allows
one fate test to pass with some compiler flags

thx

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If a bugfix only changes things apparently unrelated to the bug with no
further explanation, that is a good sign that the bugfix is wrong.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20181207/7d7a3da2/attachment.sig>