Fix test-running in --single-process mode.

Fix test-running in --single-process mode.

* Avoid destroying the ContentRendererClient, since that asserts we're on the renderer's thread
* Avoid asserting that all RenderProcessHosts are destroyed, since r170644 avoided destroying them in single-process mode.


Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=173543

[Jeffrey Yasskin, 2012-12-14 22:43:19]

I've tested with `xvfb-run  gtk_out/Debug/browser_tests --single-process
--gtest_filter=Declarative*` and `xvfb-run  gtk_out/Debug/browser_tests
--single_process --gtest_filter=Declarative*`, as well as the trybots you see
running.

@mad, could you double-check the profile_destroyer change? I don't see a
directory owner who's touched this recently, but you added the DCHECK. Joth
suggested using "if (content::RenderProcessHost::run_renderer_in_process()
return;" instead.

[Jeffrey Yasskin, 2012-12-14 22:48:07]

On 2012/12/14 22:43:19, Jeffrey Yasskin wrote:
> Joth
> suggested using "if (content::RenderProcessHost::run_renderer_in_process()
> return;" instead.

This failed because a ConfigurationPolicyProvider::observer_list_ winds up
getting destroyed without being empty, so the current change is probably close
to right. Wording suggestions for the comment are welcome.

[Paweł Hajdan Jr., 2012-12-14 22:50:09]

Running more than one test with --single_process is not supported.

[Jeffrey Yasskin, 2012-12-14 22:53:31]

On 2012/12/14 22:50:09, Paweł Hajdan Jr. wrote:
> Running more than one test with --single_process is not supported.

Yes, that's fine. This patch fixes running a *single* test with
--single-process.

[jam, 2012-12-14 22:59:33]

lgtm

[Paweł Hajdan Jr., 2012-12-14 23:11:21]

LGTM

[commit-bot: I haz the power, 2012-12-14 23:14:49]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jyasskin@chromium.org/11567053/1

[MAD, 2012-12-15 00:11:21]

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
File chrome/browser/profiles/profile_destroyer.cc (right):

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
chrome/browser/profiles/profile_destroyer.cc:40:
DCHECK(content::RenderProcessHost::run_renderer_in_process() ||
Can you please add a comment stating why we need to do this?

[Jeffrey Yasskin, 2012-12-15 00:31:31]

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
File chrome/browser/profiles/profile_destroyer.cc (right):

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
chrome/browser/profiles/profile_destroyer.cc:40:
DCHECK(content::RenderProcessHost::run_renderer_in_process() ||
On 2012/12/15 00:11:22, MAD wrote:
> Can you please add a comment stating why we need to do this?

Sure. Could you explain why off-the-record profiles leave RenderProcessHosts
around so the resulting comment is symmetric?

[commit-bot: I haz the power, 2012-12-15 01:09:41]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jyasskin@chromium.org/11567053/6002

[MAD, 2012-12-15 04:05:30]

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
File chrome/browser/profiles/profile_destroyer.cc (right):

https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
chrome/browser/profiles/profile_destroyer.cc:40:
DCHECK(content::RenderProcessHost::run_renderer_in_process() ||
On 2012/12/15 00:31:31, Jeffrey Yasskin wrote:
> On 2012/12/15 00:11:22, MAD wrote:
> > Can you please add a comment stating why we need to do this?
> 
> Sure. Could you explain why off-the-record profiles leave RenderProcessHosts
> around so the resulting comment is symmetric?

Good point, this is because we don't want to delay destruction of off the record
profile, since they contain private information that has be destroyed as quickly
as possible.

[commit-bot: I haz the power, 2012-12-15 04:24:10]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jyasskin@chromium.org/11567053/13001

[Jeffrey Yasskin, 2012-12-15 04:24:13]

On 2012/12/15 04:05:30, MAD wrote:
>
https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
> File chrome/browser/profiles/profile_destroyer.cc (right):
> 
>
https://chromiumcodereview.appspot.com/11567053/diff/1/chrome/browser/profile...
> chrome/browser/profiles/profile_destroyer.cc:40:
> DCHECK(content::RenderProcessHost::run_renderer_in_process() ||
> On 2012/12/15 00:31:31, Jeffrey Yasskin wrote:
> > On 2012/12/15 00:11:22, MAD wrote:
> > > Can you please add a comment stating why we need to do this?
> > 
> > Sure. Could you explain why off-the-record profiles leave RenderProcessHosts
> > around so the resulting comment is symmetric?
> 
> Good point, this is because we don't want to delay destruction of off the
record
> profile, since they contain private information that has be destroyed as
quickly
> as possible.

Incorporated. Thanks!

[commit-bot: I haz the power, 2012-12-15 08:06:04]

Retried try job too often on win_rel for step(s) browser_tests

[commit-bot: I haz the power, 2012-12-15 08:52:31]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jyasskin@chromium.org/11567053/13001

[commit-bot: I haz the power, 2012-12-15 12:15:58]

Retried try job too often on win_rel for step(s) browser_tests

[commit-bot: I haz the power, 2012-12-17 18:26:02]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jyasskin@chromium.org/11567053/13001

[commit-bot: I haz the power, 2012-12-17 21:57:22]

Change committed as 173543