#7397 closed defect (needs_more_info)
[ffmpeg-qsv][h264_ffmpeg]decode dump file has garbage on iHD driver.
Reported by: | a | Owned by: | |
---|---|---|---|
Priority: | normal | Component: | undetermined |
Version: | unspecified | Keywords: | qsv |
Cc: | Blocked By: | ||
Blocking: | Reproduced by developer: | no | |
Analyzed by developer: | no |
Description
Summary of the bug:
How to reproduce:
1.ffmpeg -hwaccel qsv -hwaccel_device /dev/dri/renderD128 -v verbose -y -c:v h264_qsv -i /media//ffmpeg_h264/crop-to-container-dims-canon.mov -vf hwdownload,format=nv12 -pix_fmt yuv420p -f rawvideo -y dump.yuv -vsync 0 2. 7yuv to open dump.yuv ffmpeg version : N-91726-g56f68a0
bug descript
The dump file md5 is not equal to reference, and the dump file has garbage.
Attachments (1)
Change History (10)
by , 6 years ago
Attachment: | crop-to-container-dims-canon.mov added |
---|
comment:1 by , 6 years ago
Summary: | ffmpeg qsv h264 decode dump file has garbage on iHD driver. → [ffmpeg-qsv][h264_ffmpeg]decode dump file has garbage on iHD driver. |
---|
comment:2 by , 6 years ago
comment:3 by , 6 years ago
Keywords: | qsv added; ffmpeg-qsv h264 decode removed |
---|
comment:4 by , 6 years ago
Keywords: | ffmpeg-qsv h264 decode added; qsv removed |
---|
SKL OS: ubuntu 16.04 kernel: 4.15.0
libva b6c50dad4d7d14c507108d9f468662e2d7ae1c4d https://github.com/01org/libva.git
VPG_Driver ed04556a6676fd77ff845508bc54bfc60915450c https://github.com/intel/media-driver
libva-utils 9a10ad663349732decd668a426cfc349b0010d1d https://github.com/01org/libva-utils.git
Mediasdk 7ea683585f0f84f1bea035a405855312c92c46c0 https://github.com/Intel-Media-SDK/MediaSDK
FFMPEG 56f68a099cc607658118e00cad30569103ae3751 https://git.ffmpeg.org/ffmpeg.git
comment:5 by , 6 years ago
Keywords: | qsv added; ffmpeg-qsv h264 decode removed |
---|
comment:6 by , 6 years ago
This seems to work for me? Or at least, I don't get garbage in the output.
It isn't bit-exact because the input is full-range and libmfx doesn't seem to have any way of communicating that, but the data is correct. (Decoding with iHD directly via VAAPI supports the range signalling and therefore matches software behaviour exactly, so this is only a libmfx problem.)
comment:7 by , 6 years ago
Tested with the latest version of MSDK: c083a2724b99851af6d69bbb35e3130b39a1f702
$ ffmpeg -i crop-to-container-dims-canon.mov -c:v copy src.h264 $ ./sample_decode h264 -i src.h264 -i420 -o msdk.yuv
ffmpeg -hwaccel qsv -hwaccel_device /dev/dri/renderD128 -v verbose -y -c:v h264_qsv -i crop-to-container-dims-canon.mov -vf hwdownload,format=nv12 -pix_fmt yuv420p -f rawvideo -y dump.yuv -vsync 0
Both msdk.yuv and dump.yuv have garbage in my test.
Report an issue in MediaSDK:
https://github.com/Intel-Media-SDK/MediaSDK/issues/715
comment:8 by , 6 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
dump.yuv don't have garbage on latest code base.
comment:9 by , 6 years ago
Resolution: | fixed → needs_more_info |
---|
Please add the commit that fixed the issue.
failed case