Could you make your patches against the staging branch first ? On the one hand, the nightly testing can check whether the patches result in regression first, one the other hand, the hw info has been moved to a separated file on the staging branch, your patches might introduce a lot of conflict when we merge (or cherry-pick) the patches from staging to master.
Thanks Haihao > Hi, > > This patch series fixes and optimizes support for H.grayscale streams > encoded in H.264 (Patch8). This is backwards compatible with solutions > that incorrectly request for YUV 4:2:0 formats instead of the > appropriate Y800 one. > > Tested with gstreamer-vaapi and FFmpeg/vaapi on Ivybridge. No > regression. > > v2: split into smaller patches, while maintaining regression > testability, drop dead code in the Y800 rendering code path. > > Regards, > Gwenole Beauchesne (8): > surface: fix geometry (size, layout) of grayscale surfaces. > surface: factor out release of surface buffer storage. > surface: fix vaDeriveImage() for grayscale. > config: fix vaGetConfigAttributes() to validate profile/entrypoint. > config: fix vaCreateConfig() to not override user chroma format. > config: fix supported set of chroma formats for JPEG decode. > decoder: h264: factor out allocation of reconstructed surfaces. > decoder: h264: optimize support for grayscale surfaces. > > src/gen6_mfd.c | 20 ++- > src/gen75_mfd.c | 18 +-- > src/gen7_mfd.c | 18 +-- > src/gen8_mfd.c | 18 +-- > src/gen8_render.c | 3 + > src/i965_decoder_utils.c | 75 +++++++++++- > src/i965_decoder_utils.h | 8 ++ > src/i965_drv_video.c | 301 > +++++++++++++++++++++++++++++++--------------- > src/i965_drv_video.h | 14 +++ > src/i965_output_dri.c | 6 +- > src/i965_render.c | 3 + > 11 files changed, 326 insertions(+), 158 deletions(-) > _______________________________________________ Libva mailing list Libva@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/libva