Direct composite canvas background if possible.

Direct composite canvas background if possible.

This avoids allocating backing for the containter GraphicsLayer.
If the background is opaque, canvas texture can be drawn with
GL blending disabled. If the canvas has global opacity, it also
avoids drawing to an intermediate render-surface.

BUG=236982
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=154786

[alokp, 2013-07-19 06:04:07]

I would like to add a similar layout test for 2D canvas, but this is only
meaningful for accelerated canvas. Is there a way to add a
accelerated-canvas-only test? Perhaps a setting in the test-runner?

[enne (OOO), 2013-07-19 16:23:46]

On 2013/07/19 06:04:07, Alok Priyadarshi wrote:
> I would like to add a similar layout test for 2D canvas, but this is only
> meaningful for accelerated canvas. Is there a way to add a
> accelerated-canvas-only test? Perhaps a setting in the test-runner?

All canvases inside of compositing/ are accelerated, assuming that they meet the
minimum size threshold.  Although, you'll probably have to disable the test for
the virtual/softwarecompositing directory if you're checking the layer tree.

[jamesr, 2013-07-19 17:01:37]

On Jul 18, 2013 11:04 PM, <alokp@chromium.org> wrote:
>
> Reviewers: jamesr, enne,
>
> Message:
> I would like to add a similar layout test for 2D canvas, but this is only
> meaningful for accelerated canvas. Is there a way to add a
> accelerated-canvas-only test? Perhaps a setting in the test-runner?

Put the test in fast/canvas and the compositor expectations in
virtual/gpu/fast/canvas.

>
> Description:
> Direct composite canvas background if possible.
>
> This avoids allocating backing for the containter GraphicsLayer.
> If the background is opaque, canvas texture can be drawn with
> GL blending disabled. If the canvas has global opacity, it also
> avoids drawing to an intermediate render-surface.
>
> BUG=236982
>
> Please review this at https://codereview.chromium.org/19543014/
>
> SVN Base: svn://svn.chromium.org/blink/trunk
>
> Affected files:
>   A LayoutTests/compositing/webgl/webgl-simple-background-expected.html
>   A LayoutTests/compositing/webgl/webgl-simple-background.html
>   M Source/core/platform/graphics/chromium/Canvas2DLayerBridge.cpp
>   M Source/core/platform/graphics/gpu/DrawingBuffer.cpp
>   M Source/core/rendering/RenderLayerBacking.cpp
>   M Source/core/rendering/RenderVideo.h
>   M public/platform/WebExternalTextureLayer.h
>
>

[alokp, 2013-07-19 23:27:07]

Included canvas-2d by detecting if canvas is accelerated. If layer-tree-string
for canvas element is empty, I assume that canvas is not composited.

Also added test cases for contents-box containing background-box and box
decorations.

[jamesr, 2013-07-19 23:28:41]

On 2013/07/19 23:27:07, Alok Priyadarshi wrote:
> Included canvas-2d by detecting if canvas is accelerated. If layer-tree-string
> for canvas element is empty, I assume that canvas is not composited.

Have you tested that this test actually tests the accelerated canvas 2d path? 
Looking at the paths I doubt that it really does.

> 
> Also added test cases for contents-box containing background-box and box
> decorations.

[alokp, 2013-07-19 23:39:35]

> 
> Have you tested that this test actually tests the accelerated canvas 2d path? 
> Looking at the paths I doubt that it really does.
> 

I have verified that running virtual/softwarecompositing/draws-content, I get
empty layer-tree string. Else I get proper layer-tree.

What makes you think it will not be accelerated? canvas sizes? canvas size based
switch for acceleration is disabled when running layout tests.
See
https://code.google.com/p/chromium/codesearch#chromium/src/content/shell/comm...

[alokp, 2013-07-22 18:07:12]

I realized that I only need pixel (ref) test for the case where background is
direct-composited with the content layer. So I separated the tests into text and
ref tests.

The text test makes sure that in the simple background case, the backing layer
does not have (drawsContent 1) , but other cases do. The pixel test makes sure
that the simple background is indeed painted into the content layer.

I had to add an additional expectation for canvas-background-layer into
virtual/softwarecompositing which does not accelerate canvas. The layer-tree in
that case is empty.

[enne (OOO), 2013-07-22 19:56:55]

Could you upload this patch again? I'm seeing "error: old chunk mismatch" when
looking at diffs.

[alokp, 2013-07-22 20:22:46]

On 2013/07/22 19:56:55, enne wrote:
> Could you upload this patch again? I'm seeing "error: old chunk mismatch" when
> looking at diffs.

Can you check if it works for you now?

[enne (OOO), 2013-07-22 21:57:48]

https://codereview.chromium.org/19543014/diff/24002/LayoutTests/compositing/d...
File LayoutTests/compositing/draws-content/canvas-simple-background.html
(right):

https://codereview.chromium.org/19543014/diff/24002/LayoutTests/compositing/d...
LayoutTests/compositing/draws-content/canvas-simple-background.html:19:
context.clearRect(0, 0, canvas.width, canvas.height);
Could you draw something in this canvas too, even if it's just a non-green solid
color rectangle in the upper left?

[alokp, 2013-07-22 23:40:19]

https://codereview.chromium.org/19543014/diff/24002/LayoutTests/compositing/d...
File LayoutTests/compositing/draws-content/canvas-simple-background.html
(right):

https://codereview.chromium.org/19543014/diff/24002/LayoutTests/compositing/d...
LayoutTests/compositing/draws-content/canvas-simple-background.html:19:
context.clearRect(0, 0, canvas.width, canvas.height);
On 2013/07/22 21:57:48, enne wrote:
> Could you draw something in this canvas too, even if it's just a non-green
solid
> color rectangle in the upper left?

DONE. Did the same for WebGL.

[enne (OOO), 2013-07-23 01:43:26]

Thanks.  lgtm

[commit-bot: I haz the power, 2013-07-23 17:51:28]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/alokp@chromium.org/19543014/38001

[commit-bot: I haz the power, 2013-07-23 18:03:21]

Retried try job too often on blink_presubmit for step(s) presubmit
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=blink_pres...

[alokp, 2013-07-23 18:05:47]

On 2013/07/23 18:03:21, I haz the power (commit-bot) wrote:
> Retried try job too often on blink_presubmit for step(s) presubmit
>
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=blink_pres...

jamesr: Needs owners blessing.

[jamesr, 2013-07-23 18:36:13]

public/ lgtm

[commit-bot: I haz the power, 2013-07-23 18:42:28]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/alokp@chromium.org/19543014/38001

[commit-bot: I haz the power, 2013-07-23 19:59:37]

Retried try job too often on mac_layout_rel for step(s) webkit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_layout...

[commit-bot: I haz the power, 2013-07-23 20:38:54]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/alokp@chromium.org/19543014/56001

[commit-bot: I haz the power, 2013-07-23 22:40:35]

Change committed as 154786