content: Remove switches::kSkipGpuDataLoading from content_shell

content: Remove switches::kSkipGpuDataLoading from content_shell

switches::kSkipGpuDataLoading causes the content_shell to not load
the blacklist and driver bug list.  This is preventing another cl
from landing that enables GPU rasterization.

Note for sherrifs, this flag was introduced to fix flake with
GpuTabTest, if we see this test flaking out, it may be as a
result to this change.

BUG=445500
Committed: https://crrev.com/ace5b877610adc14182677cca9c6b038ea161ccf
Cr-Commit-Position: refs/heads/master@{#310132}

[hendrikw, 2014-12-30 19:36:36]

hendrikw@chromium.org changed reviewers:
+ piman@chromium.org

[hendrikw, 2014-12-30 19:36:37]

PTAL, thanks!

[piman, 2015-01-05 23:35:18]

LGTM

[hendrikw, 2015-01-06 19:53:02]

The CQ bit was checked by hendrikw@chromium.org

[commit-bot: I haz the power, 2015-01-06 19:54:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/827293002/1

[commit-bot: I haz the power, 2015-01-06 20:21:11]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-01-06 20:22:27]

Patchset 1 (id:??) landed as
https://crrev.com/ace5b877610adc14182677cca9c6b038ea161ccf
Cr-Commit-Position: refs/heads/master@{#310132}

[jochen (gone - plz use gerrit), 2015-01-07 10:15:33]

jochen@chromium.org changed reviewers:
+ jochen@chromium.org

[jochen (gone - plz use gerrit), 2015-01-07 10:15:34]

we're still skipping the black list, right? Because ffmpeg used to be
blacklisted, but we need it for layout tests.

also, loading the blacklist takes a few hundred ms. giving that we not need it,
it's a shame to waste those cycles for every single test...

[hendrikw, 2015-01-07 15:44:10]

On 2015/01/07 10:15:34, jochen (slow) wrote:
> we're still skipping the black list, right? Because ffmpeg used to be
> blacklisted, but we need it for layout tests.
> 
> also, loading the blacklist takes a few hundred ms. giving that we not need
it,
> it's a shame to waste those cycles for every single test...

We're not skipping the blacklist anymore.  Did I break ffmpeg tests?

If this is a problem, we should add "--skip-gpu-data-loading" to the tests
rather than hard code it into the content shell, but we'll also need to pass
"--disable-gpu-rasterization", since the black list turns off gpu rasterization,
and we can't use it for the layout tests.

[jochen (gone - plz use gerrit), 2015-01-07 15:53:00]

On 2015/01/07 at 15:44:10, hendrikw wrote:
> On 2015/01/07 10:15:34, jochen (slow) wrote:
> > we're still skipping the black list, right? Because ffmpeg used to be
> > blacklisted, but we need it for layout tests.
> > 
> > also, loading the blacklist takes a few hundred ms. giving that we not need
it,
> > it's a shame to waste those cycles for every single test...
> 
> We're not skipping the blacklist anymore.  Did I break ffmpeg tests?
> 

sorry, i meant "osmesa". in line 150 we have --use-gl=osmesa - is this still
honored for layout tests?

> If this is a problem, we should add "--skip-gpu-data-loading" to the tests
rather than hard code it into the content shell, but we'll also need to pass
"--disable-gpu-rasterization", since the black list turns off gpu rasterization,
and we can't use it for the layout tests.

yeah, but the default for layout tests should be osmesa (otherwise we don't get
reliable pixel results)

[hendrikw, 2015-01-07 16:10:56]

On 2015/01/07 15:53:00, jochen (slow) wrote:
> On 2015/01/07 at 15:44:10, hendrikw wrote:
> > On 2015/01/07 10:15:34, jochen (slow) wrote:
> > > we're still skipping the black list, right? Because ffmpeg used to be
> > > blacklisted, but we need it for layout tests.
> > > 
> > > also, loading the blacklist takes a few hundred ms. giving that we not
need
> it,
> > > it's a shame to waste those cycles for every single test...
> > 
> > We're not skipping the blacklist anymore.  Did I break ffmpeg tests?
> > 
> 
> sorry, i meant "osmesa". in line 150 we have --use-gl=osmesa - is this still
> honored for layout tests?
> 
> > If this is a problem, we should add "--skip-gpu-data-loading" to the tests
> rather than hard code it into the content shell, but we'll also need to pass
> "--disable-gpu-rasterization", since the black list turns off gpu
rasterization,
> and we can't use it for the layout tests.
> 
> yeah, but the default for layout tests should be osmesa (otherwise we don't
get
> reliable pixel results)

--use-gl=osmesa is still used for layout tests.  Are you suggesting that mesa
might be blacklisted on some systems?  hmm.

It does appear to blacklist mesa in a few cases, older versions of the driver or
linux. It doesn't look like we'd run into any of these for our tests.  Still,
it's something to watch for.

[jsbell, 2015-01-07 22:58:45]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/836373003/ by jsbell@chromium.org.

The reason for reverting is: Speculative revert for crashes on WinXP bots. See
crbug.com/446834
I will re-land if this does not fix the crashes..