Revert "Give a canvas/video test more time."

Revert "Give a canvas/video test more time."

The cause of the timeouts wasn't just that the
virtual/gpu/fast/canvas/yuv-video-on-accelerated-canvas.html test
was slow because it still timeout in try bots, just taking longer
to do so.

This reverts commit 60fdfab472ffa17fad5f12a9f389ccf85094e9f3.

BUG=450699
R=junov@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=189176

[Daniel Bratell, 2015-01-28 10:40:43]

Please take a look.

[dshwang, 2015-01-28 12:23:36]

On 2015/01/28 10:40:43, Daniel Bratell wrote:
> Please take a look.

Which platform are still failed? I guess Win.
could you mark it to Timeout for Win in TestExpectations, and remain it in
SlowTests for other platform?

[Daniel Bratell, 2015-01-28 12:31:57]

On 2015/01/28 12:23:36, dshwang wrote:
> On 2015/01/28 10:40:43, Daniel Bratell wrote:
> > Please take a look.
> 
> Which platform are still failed? I guess Win.
> could you mark it to Timeout for Win in TestExpectations, and remain it in
> SlowTests for other platform?

I you search for virtual/gpu/fast/canvas/yuv-video-on-accelerated-canvas.html at
http://build.chromium.org/f/chromium/flakiness/ (not sure how to make a direct
link though I think it's possible) you will see the results.

It has timeouts (the yellow "30") for WebKit Linux, WebKit Mac, WebKit Win7 and
WebKit XP. The only bots that don't timeout are WebKit Mac Oilpan (it has other
errors though), WebKit Linux ASAN and WebKit Linux MSAN. That ASAN and MSAN
makes the test reliable is probably a clue but I don't know what it indicates.

Anyway, all platforms have the timeouts in /virtual/gpu/*, not just Windows. The
only reason this focused on Windows is because the error rate for all tests is
too high there. Actually, most of the fixes I've done have not been related to
Windows at all.

[dshwang, 2015-01-28 18:11:34]

On 2015/01/28 12:31:57, Daniel Bratell wrote:
> On 2015/01/28 12:23:36, dshwang wrote:
> > On 2015/01/28 10:40:43, Daniel Bratell wrote:
> > > Please take a look.
> > 
> > Which platform are still failed? I guess Win.
> > could you mark it to Timeout for Win in TestExpectations, and remain it in
> > SlowTests for other platform?
> 
> I you search for virtual/gpu/fast/canvas/yuv-video-on-accelerated-canvas.html
at
> http://build.chromium.org/f/chromium/flakiness/ (not sure how to make a direct
> link though I think it's possible) you will see the results.
> 
> It has timeouts (the yellow "30") for WebKit Linux, WebKit Mac, WebKit Win7
and
> WebKit XP. The only bots that don't timeout are WebKit Mac Oilpan (it has
other
> errors though), WebKit Linux ASAN and WebKit Linux MSAN. That ASAN and MSAN
> makes the test reliable is probably a clue but I don't know what it indicates.
> 
> Anyway, all platforms have the timeouts in /virtual/gpu/*, not just Windows.
The
> only reason this focused on Windows is because the error rate for all tests is
> too high there. Actually, most of the fixes I've done have not been related to
> Windows at all.

thank you for explaining. IMO you should mark it to Timeout in TestExpectations
for all platform.

[Justin Novosad, 2015-01-28 18:16:29]

I agree with Dongseong. If marking it as slow does not work, we should then mark
it as flaky to eliminate red noise on the bots.

[Justin Novosad, 2015-01-28 18:17:21]

On 2015/01/28 18:16:29, junov wrote:
> I agree with Dongseong. If marking it as slow does not work, we should then
mark
> it as flaky to eliminate red noise on the bots.

That being said, lgtm for reverting the slow expectation

[Daniel Bratell, 2015-01-29 09:33:52]

The CQ bit was checked by bratell@opera.com

[commit-bot: I haz the power, 2015-01-29 09:34:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/885603002/1

[commit-bot: I haz the power, 2015-01-29 09:37:29]

Committed patchset #1 (id:1) as
https://src.chromium.org/viewvc/blink?view=rev&revision=189176