[WebGL-blink] Add test function to force fail the webgl-context

[Webgl-blink] Add test function to force fail the webgl-context.

In order to test whether the  GPU(Vendor,Renderer,Driver info)
is getting fetched properly when creating the context,
we need to provide a function to force fail the context on 
creation, which can be used by test-runner. 

This patch implements the forceNextWebGLContextCreationToFail.
Test-runner can use this further.

test-runner-cl which uses forceNextWebGLContextCreationToFail
@https://codereview.chromium.org/706193003/

BUG=412440
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=185272

[sivag, 2014-11-05 14:48:11]

siva.gunturi@samsung.com changed reviewers:
+ kbr@chromium.org

[sivag, 2014-11-05 18:06:13]

@kbr, please take a look at this approach.

As i can't see any direct way to make the context creation fail by using the
canvas js apis.

I tried using LayoutTestSupport::isRunningLayoutTest()conditions in
WebGLRenderingContext::create but  this could effect other layout tests. 

So I am planning to expose the validateWebglContextCreationStatusMessage
function to js
and test the validity there? One more way of doing the same thing is to  provide
shared context as input parameter to WebGLRenderingContext::create() and use it
by passing a sharedcontext whose context is lost.
As of now we pass 0 for this.

Please share your views.

[Ken Russell (switch to Gerrit), 2014-11-06 07:09:55]

Sorry, this isn't what I had in mind. What I'm suggesting is to add a new method
to TestRunner like forceNextWebGLContextCreationToFail(). That should call a new
public static method in WebGLRenderingContext like
"forceNextContextCreationToFail". This should set a flag causing the next call
to WebGLRenderingContext::create to fail. Then a new layout test can be written
which calls this method, listens for the webglcontextcreationerror event on the
canvas, and does some simple verification of its error message.

See src/content/shell/renderer/test_runner/test_runner.cc and
capturePixelsAsyncThen for an example of adding a new method to the test runner.

[sivag, 2014-11-09 11:51:27]

Patchset #5 (id:80001) has been deleted

[sivag, 2014-11-10 13:44:49]

ptal..

[Ken Russell (switch to Gerrit), 2014-11-10 19:55:28]

The new patch set looks good overall. A couple of comments.

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
File Source/core/html/canvas/WebGLRenderingContext.cpp (right):

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:64: static bool
contextShouldFail = false;
Please name this something like shouldFailContextCreationForTesting. (The
"ForTesting" is the important thing.)

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:92: if (!context ||
contextShouldFail) {
This code path must set the "should fail" flag back to false.

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:187: void
WebGLRenderingContext::forceNextWebGLContextCreationToFail(bool failContext)
Per comments on the Chromium side CL, this shouldn't take a boolean argument. It
should just set the internal "should fail" flag to true.

[sivag, 2014-11-11 12:54:38]

siva.gunturi@samsung.com changed reviewers:
+ tkent@chromium.org

[sivag, 2014-11-11 12:55:15]

ptal..

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
File Source/core/html/canvas/WebGLRenderingContext.cpp (right):

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:64: static bool
contextShouldFail = false;
On 2014/11/10 19:55:28, Ken Russell wrote:
> Please name this something like shouldFailContextCreationForTesting. (The
> "ForTesting" is the important thing.)

Done.

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:92: if (!context ||
contextShouldFail) {
On 2014/11/10 19:55:28, Ken Russell wrote:
> This code path must set the "should fail" flag back to false.

Done.

https://codereview.chromium.org/703903002/diff/100001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:187: void
WebGLRenderingContext::forceNextWebGLContextCreationToFail(bool failContext)
On 2014/11/10 19:55:28, Ken Russell wrote:
> Per comments on the Chromium side CL, this shouldn't take a boolean argument.
It
> should just set the internal "should fail" flag to true.

Done.

[sivag, 2014-11-12 18:47:51]

ptal..

[Ken Russell (switch to Gerrit), 2014-11-12 21:26:02]

Again looks good in general, but there's one logic bug.

https://codereview.chromium.org/703903002/diff/140001/Source/core/html/canvas...
File Source/core/html/canvas/WebGLRenderingContext.cpp (right):

https://codereview.chromium.org/703903002/diff/140001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:101:
shouldFailContextCreationForTesting = false;
This should be moved up to the beginning of the block, before the dispatchEvent
call. Otherwise if the event listener tries to create a WebGL context it'll fail
too.

[Ken Russell (switch to Gerrit), 2014-11-12 21:27:02]

Also: please edit the CL description. This CL now only adds the
forceNextWebGLContextCreationToFail API.

[tkent, 2014-11-13 00:25:44]

lgtm.  Please update the description.

[tkent, 2014-11-13 00:27:11]

https://codereview.chromium.org/703903002/diff/140001/public/web/WebView.h
File public/web/WebView.h (right):

https://codereview.chromium.org/703903002/diff/140001/public/web/WebView.h#ne...
public/web/WebView.h:494: virtual void forceNextWebGLContextCreationToFail() =
0;
This is not for 'i18n'.  Maybe this should be in the following 'Testing
functionality' section.

[sivag, 2014-11-13 03:29:16]

Description Updated.

https://codereview.chromium.org/703903002/diff/140001/public/web/WebView.h
File public/web/WebView.h (right):

https://codereview.chromium.org/703903002/diff/140001/public/web/WebView.h#ne...
public/web/WebView.h:494: virtual void forceNextWebGLContextCreationToFail() =
0;
On 2014/11/13 00:27:11, tkent wrote:
> This is not for 'i18n'.  Maybe this should be in the following 'Testing
> functionality' section.
> 

Done.

[sivag, 2014-11-13 03:32:15]

The CQ bit was checked by siva.gunturi@samsung.com

[commit-bot: I haz the power, 2014-11-13 03:32:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/703903002/160001

[sivag, 2014-11-13 03:34:06]

The CQ bit was unchecked by siva.gunturi@samsung.com

[tkent, 2014-11-13 03:37:24]

https://codereview.chromium.org/703903002/diff/160001/public/web/WebWidget.h
File public/web/WebWidget.h (right):

https://codereview.chromium.org/703903002/diff/160001/public/web/WebWidget.h#...
public/web/WebWidget.h:276: virtual void forceNextWebGLContextCreationToFail() {
}
Do you need this one?

[sivag, 2014-11-13 03:40:32]

@kbr ptal..

https://codereview.chromium.org/703903002/diff/140001/Source/core/html/canvas...
File Source/core/html/canvas/WebGLRenderingContext.cpp (right):

https://codereview.chromium.org/703903002/diff/140001/Source/core/html/canvas...
Source/core/html/canvas/WebGLRenderingContext.cpp:101:
shouldFailContextCreationForTesting = false;
On 2014/11/12 21:26:02, Ken Russell wrote:
> This should be moved up to the beginning of the block, before the
dispatchEvent
> call. Otherwise if the event listener tries to create a WebGL context it'll
fail
> too.

Done.

[sivag, 2014-11-13 05:26:04]

@tkent, kbr
ptal..

https://codereview.chromium.org/703903002/diff/160001/public/web/WebWidget.h
File public/web/WebWidget.h (right):

https://codereview.chromium.org/703903002/diff/160001/public/web/WebWidget.h#...
public/web/WebWidget.h:276: virtual void forceNextWebGLContextCreationToFail() {
}
On 2014/11/13 03:37:23, tkent wrote:
> Do you need this one?

Not needed.Removed.

https://codereview.chromium.org/703903002/diff/160001/public/web/WebWidget.h#...
public/web/WebWidget.h:276: virtual void forceNextWebGLContextCreationToFail() {
}
On 2014/11/13 03:37:23, tkent wrote:
> Do you need this one?

Done.

[tkent, 2014-11-13 05:51:21]

The CQ bit was checked by tkent@chromium.org

[tkent, 2014-11-13 05:51:22]

lgtm

[commit-bot: I haz the power, 2014-11-13 05:52:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/703903002/200001

[commit-bot: I haz the power, 2014-11-13 07:18:08]

Committed patchset #10 (id:200001) as 185272