Fix windows gpu bots

Fix windows gpu bots.

http://codereview.chromium.org/8498036 broke it, because I didn't realize that AsyncPresentAndAcknowledge called the callback from a background thread.
I simply reverted part of that change which just meant to simplify the code, and it should go back to the expected behavior.

BUG=107083
TEST=gpu bots

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=114160

[piman, 2011-12-12 22:02:48]

+darin for OWNERS

[apatrick_chromium, 2011-12-12 22:45:32]

lgtm

[darin (slow to review), 2011-12-12 23:54:32]

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
File content/browser/renderer_host/render_widget_host_view_win.cc (right):

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:197:
BrowserThread::PostTask(BrowserThread::IO, FROM_HERE, task);
perhaps you meant to forward |from_here|?

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:2056: new
AcceleratedSurfaceMsg_BuffersSwappedACK(params.route_id));
it seems like you should be using the new Passed() / .Pass() feature that ajwong
just posted to chromium-dev about.  that way if the acknowledge_task fails to
run, you won't incur a memory leak.

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:2061:
base::Bind(PostTaskOnIOThread, FROM_HERE, acknowledge_task));
hmm, seems like it might be nice to add PostTaskOnIOThread to browser_thread.h.

[piman, 2011-12-13 00:35:41]

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
File content/browser/renderer_host/render_widget_host_view_win.cc (right):

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:197:
BrowserThread::PostTask(BrowserThread::IO, FROM_HERE, task);
On 2011/12/12 23:54:32, darin wrote:
> perhaps you meant to forward |from_here|?

Probably... I was just reverting part of the CL, so this was the previous code.
In this version, the function has disappeared, so problem goes away.

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:2056: new
AcceleratedSurfaceMsg_BuffersSwappedACK(params.route_id));
On 2011/12/12 23:54:32, darin wrote:
> it seems like you should be using the new Passed() / .Pass() feature that
ajwong
> just posted to chromium-dev about.  that way if the acknowledge_task fails to
> run, you won't incur a memory leak.

Done.

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:2061:
base::Bind(PostTaskOnIOThread, FROM_HERE, acknowledge_task));
On 2011/12/12 23:54:32, darin wrote:
> hmm, seems like it might be nice to add PostTaskOnIOThread to
browser_thread.h.

I removed that function.
We could argue whether or not we want to add that to the public content
interface (BrowserThread is public), but we should do that in another CL that
involves appropriate OWNERS.
This CL is a partial revert to fix the GPU bots, I would like to avoid bike
shedding here, so that we can land it quickly and unblock people.

[darin (slow to review), 2011-12-13 00:40:19]

LGTM

[commit-bot: I haz the power, 2011-12-13 00:45:26]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/8921017/6002

[piman, 2011-12-13 00:45:32]

On 2011/12/13 00:35:41, piman wrote:
>
http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
> File content/browser/renderer_host/render_widget_host_view_win.cc (right):
> 
>
http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
> content/browser/renderer_host/render_widget_host_view_win.cc:197:
> BrowserThread::PostTask(BrowserThread::IO, FROM_HERE, task);
> On 2011/12/12 23:54:32, darin wrote:
> > perhaps you meant to forward |from_here|?
> 
> Probably... I was just reverting part of the CL, so this was the previous
code.
> In this version, the function has disappeared, so problem goes away.
> 
>
http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
> content/browser/renderer_host/render_widget_host_view_win.cc:2056: new
> AcceleratedSurfaceMsg_BuffersSwappedACK(params.route_id));
> On 2011/12/12 23:54:32, darin wrote:
> > it seems like you should be using the new Passed() / .Pass() feature that
> ajwong
> > just posted to chromium-dev about.  that way if the acknowledge_task fails
to
> > run, you won't incur a memory leak.
> 
> Done.
> 
>
http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
> content/browser/renderer_host/render_widget_host_view_win.cc:2061:
> base::Bind(PostTaskOnIOThread, FROM_HERE, acknowledge_task));
> On 2011/12/12 23:54:32, darin wrote:
> > hmm, seems like it might be nice to add PostTaskOnIOThread to
> browser_thread.h.
> 
> I removed that function.
> We could argue whether or not we want to add that to the public content
> interface (BrowserThread is public), but we should do that in another CL that
> involves appropriate OWNERS.
> This CL is a partial revert to fix the GPU bots, I would like to avoid bike
> shedding here, so that we can land it quickly and unblock people.

http://codereview.chromium.org/8917028/ for the CL that adds it.

[commit-bot: I haz the power, 2011-12-13 00:47:04]

Try job failure for 8921017-6002 on win_rel for steps "update_scripts, update".
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

Step "update" is always a major failure.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2011-12-13 00:54:06]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/8921017/6002

[piman, 2011-12-13 01:04:50]

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
File content/browser/renderer_host/render_widget_host_view_win.cc (right):

http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/r...
content/browser/renderer_host/render_widget_host_view_win.cc:2061:
base::Bind(PostTaskOnIOThread, FROM_HERE, acknowledge_task));
On 2011/12/13 00:35:41, piman wrote:
> On 2011/12/12 23:54:32, darin wrote:
> > hmm, seems like it might be nice to add PostTaskOnIOThread to
> browser_thread.h.
> 
> I removed that function.

Turns out, it's actually needed for overload resolution. So it's back (with
FROM_HERE->from_here).

[commit-bot: I haz the power, 2011-12-13 01:11:27]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/8921017/3005

[commit-bot: I haz the power, 2011-12-13 01:14:17]

Try job failure for 8921017-3005 on win_rel for step "update".
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

Step "update" is always a major failure.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2011-12-13 01:38:33]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/8921017/3005

[commit-bot: I haz the power, 2011-12-13 02:47:17]

Change committed as 114160

[darin (slow to review), 2011-12-13 05:18:25]

On Mon, Dec 12, 2011 at 5:04 PM, <piman@chromium.org> wrote:

>
> http://codereview.chromium.**org/8921017/diff/1/content/**
>
browser/renderer_host/render_**widget_host_view_win.cc<http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/render_widget_host_view_win.cc>
> File content/browser/renderer_host/**render_widget_host_view_win.cc
> (right):
>
> http://codereview.chromium.**org/8921017/diff/1/content/**
>
browser/renderer_host/render_**widget_host_view_win.cc#**newcode2061<http://codereview.chromium.org/8921017/diff/1/content/browser/renderer_host/render_widget_host_view_win.cc#newcode2061>
> content/browser/renderer_host/**render_widget_host_view_win.**cc:2061:
> base::Bind(PostTaskOnIOThread, FROM_HERE, acknowledge_task));
> On 2011/12/13 00:35:41, piman wrote:
>
>> On 2011/12/12 23:54:32, darin wrote:
>> > hmm, seems like it might be nice to add PostTaskOnIOThread to
>> browser_thread.h.
>>
>
>  I removed that function.
>>
>
> Turns out, it's actually needed for overload resolution. So it's back
> (with FROM_HERE->from_here).
>

OK


>
>
http://codereview.chromium.**org/8921017/<http://codereview.chromium.org/8921...
>