The current code in VA intel-driver is suboptimal, yet incorrect. If the bitstream is in grayscale format, we have to try and allocate YUV 4:0:0 surfaces instead. you can get some useful information from Bug730047(https://bugzilla.gnome.org/show_bug.cgi?id=730047) 1. Testing Env: ======================================================================== Libva: (master)0337703ffedc8562299c2d65acfae76c380043c6 Libva_intel_driver: (master)b48ba793b83096be87cc4a1258be80737f26fb3b Gstreamer10: (1.0)4e880d4d1e151ea64f83c28b5c3e1bbc06c57903 Gst_plugins_base10: (1.0)2dd3f028c1e6dea799d7496639f53220818b20b1 Gst_plugins_good10: (1.0)643d425f51f81b56deec16c01162637546708ee5 Gst_plugins_bad10: (1.0)0587ab41b4f9979e9cfc11011ed5c970569ee3d3 Gst_plugins_ugly10: (1.0)c7c911b8320576429e4a4234a1e29ec7436e6814 Gst_plugins_vaapi10: (master)85a60441fa240a0cf884fcb91bb296636a22efe7 2. Testing Steps: ======================================================================== when running command: "gst-launch-1.0 filesrc location=/root/media_tools/decoder/Base_Ext_Main_profile/HPCAMOLQ_BRCM_B.264 '!' h264parse '!' vaapidecode '!' vaapisink sync=false" following error will arise. error log: ...... Setting pipeline to PAUSED ... Pipeline is PREROLLING ... ERROR: from element /GstPipeline:pipeline0/GstH264Parse:h264parse0: GStreamer encountered a general stream error. Additional debug info: gstbaseparse.c(3038): gst_base_parse_loop (): /GstPipeline:pipeline0/GstH264Parse:h264parse0: streaming stopped, reason error ERROR: pipeline doesn't want to preroll. Setting pipeline to NULL ... Freeing pipeline ... 3. Frequency of Occurence: ======================================================================== 100%
The issue has been fixed by Gwenole. Previously the old codec layer doesn't require a YUV 4:0:0 surface for grayscale image, so the driver doesn't support YUV 4:0:0 surface for H.264 decoding as well.
Hi, (In reply to comment #1) > The issue has been fixed by Gwenole. Previously the old codec layer doesn't > require a YUV 4:0:0 surface for grayscale image, so the driver doesn't > support YUV 4:0:0 surface for H.264 decoding as well. Yes, I will push the patches by the end of the week if there is no additional comment. Meanwhile, could the QA team do a test on all existing platforms with: <http://cgit.freedesktop.org/~gb/libva-intel-driver/> (15.grayscale branch)? Thanks. In particular, I am interested in platforms I didn't test, and in order of priority: Baytrail, Sandybridge, Ironlake. I'd suspect a potential issue with the last one if my fallback strategy doesn't work there, but the first two platforms should work just fine.
(In reply to comment #2) > Hi, > > (In reply to comment #1) > > The issue has been fixed by Gwenole. Previously the old codec layer doesn't > > require a YUV 4:0:0 surface for grayscale image, so the driver doesn't > > support YUV 4:0:0 surface for H.264 decoding as well. > > Yes, I will push the patches by the end of the week if there is no > additional comment. Meanwhile, could the QA team do a test on all existing > platforms with: <http://cgit.freedesktop.org/~gb/libva-intel-driver/> > (15.grayscale branch)? > > Thanks. In particular, I am interested in platforms I didn't test, and in > order of priority: Baytrail, Sandybridge, Ironlake. I'd suspect a potential > issue with the last one if my fallback strategy doesn't work there, but the > first two platforms should work just fine. s/in order/in decreasing order/. i.e. Baytrail is the most important one, Sandybridge less, and Ironlake even less. But still worth testing. Thanks.
Hi Gwenole, About "Meanwhile, could the QA team do a test on all existing platforms with: <http://cgit.freedesktop.org/~gb/libva-intel-driver/> (15.grayscale branch)?", I am not sure which test cases(deocde?VPP? or Encode?, or more detail) need to cover for your patch. You know, if all features need to covered on each platform. It will be a huge work. Thanks Focus
verify pass with X11R7.unstable.2014-06-05. X11R7.unstable.2014-06-05 package information: Libva: (staging)35e70cb9b9c77dfb99fb370e319ed501f0c31b17 Libva_intel_driver: (staging)fbbe401aa28a0b3859d587ef08f0df15a2f7c8f2 Gstreamer10: (1.0)4e880d4d1e151ea64f83c28b5c3e1bbc06c57903 Gst_plugins_base10: (1.0)2dd3f028c1e6dea799d7496639f53220818b20b1 Gst_plugins_good10: (1.0)643d425f51f81b56deec16c01162637546708ee5 Gst_plugins_bad10: (1.0)0587ab41b4f9979e9cfc11011ed5c970569ee3d3 Gst_plugins_ugly10: (1.0)c7c911b8320576429e4a4234a1e29ec7436e6814 Gst_plugins_vaapi10: (master)e670e3600709602ddc6a9db651cc18e91dba4a09 Gstreamer010: (0.10)76fc67b18c38e7f6c9cfacc4e4d0ed11d3b2c548 Gst_plugins_base010: (0.10)1e1e6eaf3f0dd11f6618154d9739cbe3e007d206 Gst_plugins_good010: (0.10)5af6f5bfb6c3619a9ccc3b1681579aeb90e8b89a Gst_plugins_bad010: (0.10)cef47d85294a0dca38631f938b81a3f0dd6891bd Gst_plugins_ugly010: (0.10)5ddd97ff27ffeb03298be3a02ed18e8c2674d365 Gst_plugins_vaapi010: (0.4)90713fdc442f27b4ea449b92660a294f7796f83e
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.