Re: [FFmpeg-devel] [PATCH 1/3] avcodec/dxtory: Remove code that corrupts dimensions

[Michael Niedermayer]

On Fri, Aug 10, 2018 at 09:05:42AM +0200, Paul B Mahol wrote:
> On 2/20/18, Michael Niedermayer  wrote:
> > On Sat, Feb 17, 2018 at 10:25:10PM +0100, Michael Niedermayer wrote:
> >> Fixes: Timeout
> >> Fixes: 5796/clusterfuzz-testcase-minimized-5206729085157376
> >>
> >> Does someone have a valid sample that triggers this path ?
> >>
> >> Found-by: continuous fuzzing process
> >> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> >> Signed-off-by: Michael Niedermayer 
> >> ---
> >>  libavcodec/dxtory.c | 6 +-
> >>  1 file changed, 1 insertion(+), 5 deletions(-)
> >
> > will apply
> >
> > [...]
> > --
> > Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
> >
> > Complexity theory is the science of finding the exact solution to an
> > approximation. Benchmarking OTOH is finding an approximation of the exact
> >
> 
> This change breaks this sample: http://0x0.st/s4Ot.avi
> 
> Please revert ASAP.

Can you explain how it breaks this sample ?
if i look at the images from:
make -j12 && ./ffmpeg -ss 1 -i s4Ot.avi -vf eq=1:0.5 test-git%d.png

teh first image to differ is test-*7.png
if i load it and the reverted in gimp as layers its easy vissible that
one has been vertically scaled to 1024 while the other is not.
Which is kind of expected.
the image contains nothing but a circle and a dot in the middle
if i now take the paintbrush tool and set its size to 132 then its
easy to see when hovering over the circle that the one from git is
round while the one from the change reverted is not round
it is round in the other images (looked at image5)
So i presume that the correct height is 1024 and not 1004 as the
code in case of a revert would make it.

Do you see some artifacts that are in the git version which are not in the
reverted one ?

Feel free to revert of course, if you think that theres a major issue
that this fixes. Iam not reverting as i failed to reproduce an issue except
the printed messages asking for a sample

Thanks
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Concerning the gods, I have no means of knowing whether they exist or not
or of what sort they may be, because of the obscurity of the subject, and
the brevity of human life -- Protagoras


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 1/3] avcodec/dxtory: Remove code that corrupts dimensions

[Paul B Mahol]

On 2/20/18, Michael Niedermayer  wrote:
> On Sat, Feb 17, 2018 at 10:25:10PM +0100, Michael Niedermayer wrote:
>> Fixes: Timeout
>> Fixes: 5796/clusterfuzz-testcase-minimized-5206729085157376
>>
>> Does someone have a valid sample that triggers this path ?
>>
>> Found-by: continuous fuzzing process
>> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
>> Signed-off-by: Michael Niedermayer 
>> ---
>>  libavcodec/dxtory.c | 6 +-
>>  1 file changed, 1 insertion(+), 5 deletions(-)
>
> will apply
>
> [...]
> --
> Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> Complexity theory is the science of finding the exact solution to an
> approximation. Benchmarking OTOH is finding an approximation of the exact
>

This change breaks this sample: http://0x0.st/s4Ot.avi

Please revert ASAP.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH 1/3] avcodec/dxtory: Remove code that corrupts dimensions

[Michael Niedermayer]

On Sat, Feb 17, 2018 at 10:25:10PM +0100, Michael Niedermayer wrote:
> Fixes: Timeout
> Fixes: 5796/clusterfuzz-testcase-minimized-5206729085157376
> 
> Does someone have a valid sample that triggers this path ?
> 
> Found-by: continuous fuzzing process 
> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> Signed-off-by: Michael Niedermayer 
> ---
>  libavcodec/dxtory.c | 6 +-
>  1 file changed, 1 insertion(+), 5 deletions(-)

will apply

[...]
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Complexity theory is the science of finding the exact solution to an
approximation. Benchmarking OTOH is finding an approximation of the exact


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

[FFmpeg-devel] [PATCH 1/3] avcodec/dxtory: Remove code that corrupts dimensions

[Michael Niedermayer]

Fixes: Timeout
Fixes: 5796/clusterfuzz-testcase-minimized-5206729085157376

Does someone have a valid sample that triggers this path ?

Found-by: continuous fuzzing process 
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer 
---
 libavcodec/dxtory.c | 6 +-
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/libavcodec/dxtory.c b/libavcodec/dxtory.c
index e736cec8db..285ca38efb 100644
--- a/libavcodec/dxtory.c
+++ b/libavcodec/dxtory.c
@@ -305,11 +305,7 @@ static int dxtory_decode_v2(AVCodecContext *avctx, AVFrame 
*pic,
 }
 
 if (avctx->height - line) {
-av_log(avctx, AV_LOG_VERBOSE,
-   "Not enough slice data available, "
-   "cropping the frame by %d pixels\n",
-avctx->height - line);
-avctx->height = line;
+avpriv_request_sample(avctx, "Not enough slice data available");
 }
 
 return 0;
-- 
2.16.1

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel