| Message ID | 20190321144829.74557-1-derek.buitenhuis@gmail.com |
|---|---|
| State | Superseded |
| Headers | show |
On 3/21/2019 11:48 AM, Derek Buitenhuis wrote: > This makes sure we don't regress on 70c8c8a818f39bc262565ec29fae2baffb3e1660. > > Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> > --- > Sample: http://chromashift.org/two_first_slice.mp4 (Trimmed to 500KiB like Carl requested). > MD5: 58b637d5500c2911d8cfe99fee86cb04 > --- > tests/fate/hevc.mak | 3 +++ > tests/ref/fate/hevc-two-first-slice | 12 ++++++++++++ > 2 files changed, 15 insertions(+) > create mode 100644 tests/ref/fate/hevc-two-first-slice > > diff --git a/tests/fate/hevc.mak b/tests/fate/hevc.mak > index db3ea19340..5ec6e279b9 100644 > --- a/tests/fate/hevc.mak > +++ b/tests/fate/hevc.mak > @@ -253,6 +253,9 @@ fate-hevc-extradata-reload: CMD = framemd5 -i $(TARGET_SAMPLES)/hevc/extradata-r > fate-hevc-monochrome-crop: CMD = probeframes -show_entries frame=width,height:stream=width,height $(TARGET_SAMPLES)/hevc/hevc-monochrome.hevc > FATE_HEVC_FFPROBE-$(call DEMDEC, HEVC, HEVC) += fate-hevc-monochrome-crop > > +hevc-two-first-slice: CMD = threads=2 framemd5 -i $(TARGET_SAMPLES)/hevc/two_first_slice.mp4 -sws_flags bitexact -t 00:00.20 -an > +FATE_HEVC += hevc-two-first-slice > + > FATE_SAMPLES_AVCONV += $(FATE_HEVC-yes) > FATE_SAMPLES_FFPROBE += $(FATE_HEVC_FFPROBE-yes) > > diff --git a/tests/ref/fate/hevc-two-first-slice b/tests/ref/fate/hevc-two-first-slice > new file mode 100644 > index 0000000000..8ce7a066c7 > --- /dev/null > +++ b/tests/ref/fate/hevc-two-first-slice > @@ -0,0 +1,12 @@ > +#format: frame checksums > +#version: 2 > +#hash: MD5 > +#tb 0: 1/15 > +#media_type 0: video > +#codec_id 0: rawvideo > +#dimensions 0: 1920x1080 > +#sar 0: 0/1 > +#stream#, dts, pts, duration, size, hash > +0, 0, 0, 1, 3110400, c88154281ada839d3f209b0eab012577 > +0, 1, 1, 1, 3110400, a2623ba4639cf246b803c08e7233112f > +0, 2, 2, 1, 3110400, 9208fcdb16664e91ada77d15cc29ace9 Are you sure the faulty packet is the third one? When i debugged this the one that reproduced the issue was the 31st. Or is this a cut sample starting from the faulty RAP?
On 21/03/2019 15:26, James Almer wrote: > Are you sure the faulty packet is the third one? When i debugged this > the one that reproduced the issue was the 31st. Or is this a cut sample > starting from the faulty RAP? Definitely sure it's covered within this command, and file's number of packets. Easy to verify: Run the fate test with your patch reverted, and it deadlocks. - Derek
On 21/03/2019 15:29, Derek Buitenhuis wrote: > Definitely sure it's covered within this command, and file's number of packets. > > Easy to verify: Run the fate test with your patch reverted, and it deadlocks. Scratch that. I was mistaken, it doesn't work with the trimmed file... Re-testing, and re-sending. Also had two lines staged but not amended... woops. - Derek
2019-03-21 16:33 GMT+01:00, Derek Buitenhuis <derek.buitenhuis@gmail.com>: > On 21/03/2019 15:29, Derek Buitenhuis wrote: >> Definitely sure it's covered within this command, and file's number of >> packets. >> >> Easy to verify: Run the fate test with your patch reverted, and it >> deadlocks. > > Scratch that. I was mistaken, it doesn't work with the trimmed file... I tested the following again: $ dd if=nal_header_deadlock.mp4 of=nal_header_deadlock_cut.mp4 bs=100k count=5 $ md5sum nal_header_deadlock_cut.mp4 58b637d5500c2911d8cfe99fee86cb04 nal_header_deadlock_cut.mp4 This file hangs with 3ac474892, the last commit before James's fix. Carl Eugen
diff --git a/tests/fate/hevc.mak b/tests/fate/hevc.mak index db3ea19340..5ec6e279b9 100644 --- a/tests/fate/hevc.mak +++ b/tests/fate/hevc.mak @@ -253,6 +253,9 @@ fate-hevc-extradata-reload: CMD = framemd5 -i $(TARGET_SAMPLES)/hevc/extradata-r fate-hevc-monochrome-crop: CMD = probeframes -show_entries frame=width,height:stream=width,height $(TARGET_SAMPLES)/hevc/hevc-monochrome.hevc FATE_HEVC_FFPROBE-$(call DEMDEC, HEVC, HEVC) += fate-hevc-monochrome-crop +hevc-two-first-slice: CMD = threads=2 framemd5 -i $(TARGET_SAMPLES)/hevc/two_first_slice.mp4 -sws_flags bitexact -t 00:00.20 -an +FATE_HEVC += hevc-two-first-slice + FATE_SAMPLES_AVCONV += $(FATE_HEVC-yes) FATE_SAMPLES_FFPROBE += $(FATE_HEVC_FFPROBE-yes) diff --git a/tests/ref/fate/hevc-two-first-slice b/tests/ref/fate/hevc-two-first-slice new file mode 100644 index 0000000000..8ce7a066c7 --- /dev/null +++ b/tests/ref/fate/hevc-two-first-slice @@ -0,0 +1,12 @@ +#format: frame checksums +#version: 2 +#hash: MD5 +#tb 0: 1/15 +#media_type 0: video +#codec_id 0: rawvideo +#dimensions 0: 1920x1080 +#sar 0: 0/1 +#stream#, dts, pts, duration, size, hash +0, 0, 0, 1, 3110400, c88154281ada839d3f209b0eab012577 +0, 1, 1, 1, 3110400, a2623ba4639cf246b803c08e7233112f +0, 2, 2, 1, 3110400, 9208fcdb16664e91ada77d15cc29ace9
This makes sure we don't regress on 70c8c8a818f39bc262565ec29fae2baffb3e1660. Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com> --- Sample: http://chromashift.org/two_first_slice.mp4 (Trimmed to 500KiB like Carl requested). MD5: 58b637d5500c2911d8cfe99fee86cb04 --- tests/fate/hevc.mak | 3 +++ tests/ref/fate/hevc-two-first-slice | 12 ++++++++++++ 2 files changed, 15 insertions(+) create mode 100644 tests/ref/fate/hevc-two-first-slice