Match html canvas which is transferred to OffscreenCanvas to CSS style

cc/layers/layer_impl.cc

Index: cc/layers/layer_impl.cc
diff --git a/cc/layers/layer_impl.cc b/cc/layers/layer_impl.cc
index be233f65d512fd6ebe4e958db2c06235e3400dd5..b8bd94a1ee982159e0c2789ace973cbd36f90c1a 100644
--- a/cc/layers/layer_impl.cc
+++ b/cc/layers/layer_impl.cc
@@ -179,12 +179,18 @@ void LayerImpl::PopulateSharedQuadState(SharedQuadState* state) const {
 
 void LayerImpl::PopulateScaledSharedQuadState(SharedQuadState* state,

[danakj, 2016/11/28 22:26:22]

I don't think we need 2 methods really. How about just having existing callers
call the one below, passing the same value twice?

[xlai (Olivia), 2016/12/13 17:29:17]

Done.

                                               float scale) const {
+  PopulateScaledSharedQuadState(state, scale, scale);
+}
+
+void LayerImpl::PopulateScaledSharedQuadState(SharedQuadState* state,
+                                              float scale_x,

[danakj, 2016/11/28 22:26:22]

can you rename these input vars to all be like "layer_to_content_scale" instead
of just "scale"?

[xlai (Olivia), 2016/12/13 17:29:17]

Done.

+                                              float scale_y) const {
   gfx::Transform scaled_draw_transform =
       draw_properties_.target_space_transform;
-  scaled_draw_transform.Scale(SK_MScalar1 / scale, SK_MScalar1 / scale);
-  gfx::Size scaled_bounds = gfx::ScaleToCeiledSize(bounds(), scale);
+  scaled_draw_transform.Scale(SK_MScalar1 / scale_x, SK_MScalar1 / scale_y);
+  gfx::Size scaled_bounds = gfx::ScaleToCeiledSize(bounds(), scale_x, scale_y);
   gfx::Rect scaled_visible_layer_rect =
-      gfx::ScaleToEnclosingRect(visible_layer_rect(), scale);
+      gfx::ScaleToEnclosingRect(visible_layer_rect(), scale_x, scale_y);
   scaled_visible_layer_rect.Intersect(gfx::Rect(scaled_bounds));
 
   state->SetAll(scaled_draw_transform, scaled_bounds, scaled_visible_layer_rect,