Support headless content_shell for linux/aura.

Support headless content_shell for linux/aura.

When running blink layout tests, we shouldn't create native windows. We should
rather just create the virtual content window for render of content. Headless
mode of content shell support is already present on other ports and needs to be
added for aura. It is important in terms of performance (when running layout tests).

BUG=316712
R=jochen

TEST=None; no behavior change;

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=236040

[r.kasibhatla, 2013-11-13 07:16:00]

PTAL.

[jochen (gone - plz use gerrit), 2013-11-13 11:47:36]

can you verify that if you run a regular pixel test followed by a W3C-SVG test
(on the same content_shell instance) both pass as expected?

[r.kasibhatla, 2013-11-13 13:18:45]

On 2013/11/13 11:47:36, jochen wrote:
> can you verify that if you run a regular pixel test followed by a W3C-SVG test
> (on the same content_shell instance) both pass as expected?

The output of running a pixel test (a test inside compositing/ folder) followed
by W3C-SVG test can be seen @ http://pastebin.ca/2475828.
I think both are working normally, but I am not sure on the interpretation of
the result, so someone needs to correct me if I am wrong.

Couple of other points which I wanted to check:
1. When running the regular layout tests (with --pixel-tests option), I am
seeing lots of image diff failures (~850). I have verified most of the image
diff and all are same - the expected image has a different scrollbar than the
actual scrollbar. It seems the expected image is a GTK-type scrollbar but the
actual scrollbar is more blink drawn (like the ones which can be seen for div
elements). Is this expected result and the tests needs to be re-baselined or the
patch needs to be corrected?

2. I am getting some timeout failures (~30 cases), all testcases related to
plugin. I am not sure that this is a regression with my change (I am trying to
verify by disabling my patch and testing again), but is this something being
seen on bots/by others as well?

[jochen (gone - plz use gerrit), 2013-11-13 13:46:55]

the initial sizes in your output look correct.

We still don't have baselines for aura, so the image diffs are expected.

No idea about the plugin timeouts. Do you also get them without your patch?

[r.kasibhatla, 2013-11-13 15:55:19]

On 2013/11/13 13:46:55, jochen wrote:
> the initial sizes in your output look correct.
> 
> We still don't have baselines for aura, so the image diffs are expected.
> 
> No idea about the plugin timeouts. Do you also get them without your patch?

I have checked again without my patch and the plugin cases are getting timed out
even without the patch - the same 34 plugin test cases.
The list of the cases is copied @ http://pastebin.ca/2475851. Not sure, if these
are seen at bots as well.

[jochen (gone - plz use gerrit), 2013-11-13 15:58:45]

lgtm

[r.kasibhatla, 2013-11-13 16:06:18]

On 2013/11/13 15:58:45, jochen wrote:
> lgtm
Please wait!!!
With the change, I have noticed that there is a random crash. I am working on
it. Will update the CL after I fix it.

[Dirk Pranke, 2013-11-13 17:26:50]

Yes, plugins appear broken w/ Linux Aura, and we need lots of new baselines. 850
failures seems low, actually; I'm seeing more like 1300 on my machine. I'll be
working on these issues today.

[r.kasibhatla, 2013-11-19 05:09:09]

I noticed in the latest trunk, we have reverted back to gtk builds instead of
aura. Any specific reason for that?
There were crashes in popup window related test cases in earlier patch and this
patch corrects it. PTAL.

[Dirk Pranke, 2013-11-19 06:14:25]

On 2013/11/19 05:09:09, r.kasibhatla wrote:
> I noticed in the latest trunk, we have reverted back to gtk builds instead of
> aura. Any specific reason for that?

Yes, there were a lot of bugs and performance issues we needed to address, so we
reverted back. We still want this change, however.

[jochen (gone - plz use gerrit), 2013-11-19 13:43:38]

lgtm

[commit-bot: I haz the power, 2013-11-19 18:00:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/r.kasibhatla@samsung.com/71243002/130001

[commit-bot: I haz the power, 2013-11-19 20:32:13]

Change committed as 236040