GitLab

The coded_size of input VideoFrame is always aligned by 2. For
instance, in the case the dimension to be encoded is 641x361,
the input VideoFrame's coded_size is 642x362 and visible_rect is
0, 0, 641x361.

This is against the expectation on which crrev.com/c/1997426 was
based; The coded size of input VideoFrame is the same as the
dimension to be encoded.

Additionally, this corrects the va surface size mismatched bugs
introduced in crrev.com/c/1997426.

Bug: 982201
Bug: b:148744040
Test: video.EncodeAccel.* on hatch, grunt and eve.
Test: VP8 Simulcast.
Test: https://appr.tc/?debug=loopback&vsc=vp8&video=maxWidth=361,maxHeight=481,minWidth=361,minHeight=481 on hatch
Change-Id: I68fe77b4e4f16879d76a7e84c3b06b73b07660b1
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2037279
Commit-Queue: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Andres Calderon Jaramillo <andrescj@chromium.org>
Reviewed-by: Dan Sanders <sandersd@chromium.org>
Cr-Commit-Position: refs/heads/master@{#739313}