Fixed YUV Video quads when their visible_content_rect != their rect.

cc/output/gl_renderer.cc

Index: cc/output/gl_renderer.cc
diff --git a/cc/output/gl_renderer.cc b/cc/output/gl_renderer.cc
index 3813e1ddb31b824a2d46234bd7f1a043aac6c281..b49ac8eefec9d79361aab1583bbc9e807bc61d0a 100644
--- a/cc/output/gl_renderer.cc
+++ b/cc/output/gl_renderer.cc
@@ -2073,7 +2073,7 @@ void GLRenderer::DrawYUVVideoQuad(const DrawingFrame* frame,
   // un-antialiased quad should have and which vertex this is and the float
   // quad passed in via uniform is the actual geometry that gets used to draw
   // it. This is why this centered rect is used and not the original quad_rect.
-  gfx::RectF tile_rect = quad->visible_rect;
+  gfx::RectF tile_rect = quad->rect;

[danakj, 2015/03/16 23:08:49]

This means we overdraw when the video is occluded tho? Maybe it's a corner
case.. but would it be difficult to still use the visible rect?

[awoloszyn, 2015/03/16 23:15:54]

This matches the original behavior before my splitting patch went in, (as in we
used to use the quad->rect instead of the visible_rect).
It may be possible to use the visible_rect instead, but that would require a
change to how the shader operates. 

[enne (OOO), 2015/03/16 23:16:55]

I think this is fine as-is.

You'd have to scale both the uv and and the ya texture coordinates differently,
I think.  It's somewhat complicated, so possible to get wrong (or break easily).
 Watching videos that are partially offscreen seems like a bit of a corner case
to me.  If I were going to optimize videos, I'd look at a number of other things
first.

   GLC(gl_, gl_->UniformMatrix3fv(yuv_matrix_location, 1, 0, yuv_to_rgb));
   GLC(gl_, gl_->Uniform3fv(yuv_adj_location, 1, yuv_adjust));