GitLab

When forwarding a frame to Video Encode Accelerator, RTCVideoEncoder
would almost always create a frame copy unless the incoming
media::VideoFrame was backed by STORAGE_SHMEM. The copy would be backed
by STORAGE_SHMEM, which results in the video decode accelerator wrapping
the VideoFrame's memory in a CVPixelBufferRef.

Changes:
1) Break out code in media::VideoFrame::WrapIOSurface() that creates
   an IOSurface from a GMB handle.
2) Code is added to WrapVideoFrameInCVPixelBuffer to create a
   CVPixelBufferRef via GpuMemoryBuffer::CloneHandle().
3) RTCVideoEncoder is changed to pass on GMB-backed video frames.

Bug: 1127791, 1125879
Change-Id: I59dd1eae59ae4dffb591830152f44e470fb3cce5
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2410244Reviewed-by: ccameron <ccameron@chromium.org>
Reviewed-by: Miguel Casas <mcasas@chromium.org>
Reviewed-by: Dale Curtis <dalecurtis@chromium.org>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Commit-Queue: Markus Handell <handellm@google.com>
Cr-Commit-Position: refs/heads/master@{#809450}