Web setting for canvas2d msaa

Web setting for setting canvas2d msaa sample count.

Corresponding Chromium change to add a command line switch: https://codereview.chromium.org/26601007

BUG=308277
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=160361

[jamesr, 2013-10-18 19:24:17]

public/ and Source/web/ lgtm, I defer to Stephen on Source/core/

[Stephen White, 2013-10-18 20:13:11]

LGTM. If allocation w/MSAA set fails, should HTMLCanvasElement be doing a retry
w/MSAA 0?

https://codereview.chromium.org/26541008/diff/8001/Source/core/html/HTMLCanva...
File Source/core/html/HTMLCanvasElement.cpp (right):

https://codereview.chromium.org/26541008/diff/8001/Source/core/html/HTMLCanva...
Source/core/html/HTMLCanvasElement.cpp:501: if (Accelerated == renderingMode &&
document().settings()->antialiased2dCanvasEnabled())
Nit: blink style doesn't reverse the == operands (although it doesn't disallow
it either).

https://codereview.chromium.org/26541008/diff/8001/Source/core/platform/graph...
File Source/core/platform/graphics/ImageBuffer.h (right):

https://codereview.chromium.org/26541008/diff/8001/Source/core/platform/graph...
Source/core/platform/graphics/ImageBuffer.h:151: ImageBuffer(const IntSize&,
float resolutionScale, RenderingMode, OpacityMode, int acceleratedSampleCount,
bool& success);
Ubernit: this is called acceleratedSampleCount here, acceleratedMSAASampleCount
above, and msaaSampleCount in the .cpp file. Any reason for the differences?

[bsalomon, 2013-10-21 13:47:51]

On 2013/10/18 20:13:11, Stephen White wrote:
> LGTM. If allocation w/MSAA set fails, should HTMLCanvasElement be doing a
retry
> w/MSAA 0?

Probably. Skia internally will fallback to the largest non-zero supported sample
count supported if the requested sample count exceeds the max. However, it won't
fallback to non-MSAA. Skia also doesn't check for OOM or other errors when used
by blink so we won't fail in that case, though later drawing will obviously not
render correctly. So I think the only case that needs addressing is the
requested sample count is > 0 but MSAA isn't supported. I'll address that in a
follow on patch.

[bsalomon, 2013-10-21 13:48:47]

https://codereview.chromium.org/26541008/diff/8001/Source/core/html/HTMLCanva...
File Source/core/html/HTMLCanvasElement.cpp (right):

https://codereview.chromium.org/26541008/diff/8001/Source/core/html/HTMLCanva...
Source/core/html/HTMLCanvasElement.cpp:501: if (Accelerated == renderingMode &&
document().settings()->antialiased2dCanvasEnabled())
On 2013/10/18 20:13:12, Stephen White wrote:
> Nit: blink style doesn't reverse the == operands (although it doesn't disallow
> it either).

Done, to be more consistent other blink code.

https://codereview.chromium.org/26541008/diff/8001/Source/core/platform/graph...
File Source/core/platform/graphics/ImageBuffer.h (right):

https://codereview.chromium.org/26541008/diff/8001/Source/core/platform/graph...
Source/core/platform/graphics/ImageBuffer.h:151: ImageBuffer(const IntSize&,
float resolutionScale, RenderingMode, OpacityMode, int acceleratedSampleCount,
bool& success);
On 2013/10/18 20:13:12, Stephen White wrote:
> Ubernit: this is called acceleratedSampleCount here,
acceleratedMSAASampleCount
> above, and msaaSampleCount in the .cpp file. Any reason for the differences?

nope, changed the cpp to match the header.

[commit-bot: I haz the power, 2013-10-21 13:48:59]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bsalomon@google.com/26541008/138001

[commit-bot: I haz the power, 2013-10-21 15:22:35]

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

[bsalomon, 2013-10-22 17:51:57]

There was an issue in Canvas2DLayerBridge::isValid(). The call to
accessRenderTarget() would actually recurse back to
Canvas2DLayerBridge::isValid() causing infinite recursion. I now store the
sample count on the Canvas2DLayerBridge so that the surface can recreated
without accessing the old render target.

Justin, can you have a look at the latest patch?

[Justin Novosad, 2013-10-23 16:43:08]

On 2013/10/22 17:51:57, bsalomon wrote:
> There was an issue in Canvas2DLayerBridge::isValid(). The call to
> accessRenderTarget() would actually recurse back to
> Canvas2DLayerBridge::isValid() causing infinite recursion. I now store the
> sample count on the Canvas2DLayerBridge so that the surface can recreated
> without accessing the old render target.
> 
> Justin, can you have a look at the latest patch?

lgtm

[commit-bot: I haz the power, 2013-10-23 17:26:31]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bsalomon@google.com/26541008/358001

[commit-bot: I haz the power, 2013-10-23 19:16:42]

Change committed as 160361