Revert 222105 "Enable GPU blacklist in tests."

Revert 222105 "Enable GPU blacklist in tests."

Failed on mac-asan bot:
http://build.chromium.org/p/chromium.memory/builders/Mac%20ASAN%20Tests%20%281%29/builds/14510

AcceleratedCompositingBlockedTest.AcceleratedCompositingBlocked: 
[7719:3847:0909/155847:1828989999085:INFO:gpu_control_list.cc(1043)] Control list match for rule #48.
[7719:3847:0909/155847:1829419815783:INFO:gpu_control_list.cc(1043)] Control list match for rule #48.
[7719:3847:0909/155847:1829591550844:INFO:gpu_control_list.cc(1043)] Control list match for rule #.
../../chrome/test/gpu/gpu_feature_browsertest.cc:143: Failure
Value of: 0U
Actual: 0
Expected: event_count
Which is: 1

> Enable GPU blacklist in tests.
> 
> Partial re-land of https://codereview.chromium.org/22198004/) -- Always enable FCM on Windows -- Part 1/3.
> 
> On top of fixing some of the issues there; only lands the part that enables the GPU blacklist in tests as the former CL is too hard to land all at once.
> 
> Also keeping --skip-gpu-data-loading around for now to be able to deal with failures caused in layout_tests by this on their own later (this needs to land ASAP and layout_tests don't need this ASAP).
> 
> Another CL will follow to always enable FCM on non-blacklisted Windows machines.
> 
> This part re-enables loading the blacklist in tests (and adds a content_browsertest to make sure that the configuration we expect to be testing is indeed the one we are testing -- this uncertainty is basically the only reason the blacklist was explicitly disabled before).
> 
> This CL also cleans up compositor_util.cc which was enforcing the blacklist twice.
> 
> The original plan was to do this only for Windows as Mac/Linux was causing trouble, but it turns out to be harder to do it only on Windows; so taking care of http://crbug.com/277242 in this CL too after all... 
> 
> BUG=233830, 267038, 190942, 277242
> TBR=jcivelli, piman 
> 
> Originally Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=219132 
> Then Reverted: https://src.chromium.org/viewvc/chrome?view=rev&revision=219159
> 
> Then re-committed (part 1/3): https://src.chromium.org/viewvc/chrome?view=rev&revision=221114
> Reverted in: https://src.chromium.org/viewvc/chrome?view=rev&revision=221145
> 
> Review URL: https://chromiumcodereview.appspot.com/23534006

TBR=gab@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=222129

[please use gerrit instead, 2013-09-09 23:38:52]

Committed patchset #1 manually as r222129.

[gab, 2013-09-10 02:54:42]

On 2013/09/10 02:54:20, gab wrote:

Piman: I saw you disabled GpuFeatureTest.AcceleratedCompositingAllowed
for OS_WIN || ADDRESS_SANITIZER || USE_AURA.

Would it make sense to do the same thing for
AcceleratedCompositingBlockedTest.AcceleratedCompositingBlocked?

This CL makes it flake 50% of the time and I have no idea why a SwapBuffers
would happen more with this CL than it did before this CL...

Is the test just broken? Or should the SwapBuffers really never happen (seems
only 1 SwapBuffers event happens)?
This test flakes the same way locally on my Windows machine.

Thanks,
Gab

[piman, 2013-09-10 03:23:08]

On 2013/09/10 02:54:42, gab wrote:
> On 2013/09/10 02:54:20, gab wrote:
> 
> Piman: I saw you disabled GpuFeatureTest.AcceleratedCompositingAllowed
> for OS_WIN || ADDRESS_SANITIZER || USE_AURA.
> 
> Would it make sense to do the same thing for
> AcceleratedCompositingBlockedTest.AcceleratedCompositingBlocked?
> 
> This CL makes it flake 50% of the time and I have no idea why a SwapBuffers
> would happen more with this CL than it did before this CL...
> 
> Is the test just broken? Or should the SwapBuffers really never happen (seems
> only 1 SwapBuffers event happens)?

For AcceleratedCompositingBlocked, the SwapBuffers for the renderer should not
happen.

I'm not sure if you may be picking up the SwapBuffers for the browser - see
https://code.google.com/p/chromium/issues/detail?id=280675

I disabled AcceleratedCompositingAllowed for another reason (namely, with
Übercompositor, the renderer doesn't SwapBuffers).

> This test flakes the same way locally on my Windows machine.
> 
> Thanks,
> Gab