cc: Fix resource provider fences and add DCHECKs.

cc: Fix resource provider fences and add DCHECKs.

When I added the resource provider fence concept, I didn't
extend the read-fence long enough (it currently only protects
the current frame). This extends the read-lock for one frame.

BUG=172995
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=182162

[epenner, 2013-02-11 19:18:42]

Ptal.

What's confusing to me is why the original method fixed the tearing. Are we
perhaps uploading twice in the same frame? Unfortunately it's not easy to DCHECK
we don't do that, without tracking all resources allocated from the pool.

[nduca, 2013-02-12 05:41:45]

I wonder, should we split out the bugfix from the debug stuff? The debug stuff
adds some probes that I'm leery of but I dont want to hold up the fix...

[epenner, 2013-02-12 06:24:01]

On 2013/02/12 05:41:45, nduca wrote:
> I wonder, should we split out the bugfix from the debug stuff? The debug stuff
> adds some probes that I'm leery of but I dont want to hold up the fix...

Yeah I was kind of feeling the same, but I'm also worried about this going wrong
and exploding our memory usage. But yes, for sure I can make it two patches.  If
you have a better idea in mind for the verification, let's chat.

[nduca, 2013-02-12 06:44:45]

I didn't have any good ideas, it was sort of that I maybe getting distrcted by
the verification stuff and then other stuff yanked me away. I'll try to fit this
in my head in a bit.

[epenner, 2013-02-12 18:43:46]

On 2013/02/12 06:44:45, nduca wrote:
> I didn't have any good ideas, it was sort of that I maybe getting distrcted by
> the verification stuff and then other stuff yanked me away. I'll try to fit
this
> in my head in a bit.

Yeah, this code almost entirely validation. The actual change is tiny so let me
put it in another CL.

[nduca, 2013-02-12 19:00:32]

Yeah, much easier to see the change. Very Lgtm.

[epenner, 2013-02-12 19:03:40]

That's better, PTAL. Maybe we can have some DCHECKS based on our memory limit.

Also, since this allows us to double our memory budget, a drastic measure to get
memory back would be to have a fence->shallowWait() function that we can call
from the resource pool. Given our current fence locations, this could be a
"glShallowFinish" if we had such a function, or just glFinish.  It's drastic,
but if we *all* half our textures in one frame, I doubt a GL finish will matter.

[epenner, 2013-02-12 19:05:01]

rather -> "if we upload *all* our textures in one frame, I doubt a glFinish will
matter"

[nduca, 2013-02-12 19:23:46]

still lgtm :)

[commit-bot: I haz the power, 2013-02-12 19:25:19]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/12220095/6001

[commit-bot: I haz the power, 2013-02-12 19:25:25]

Presubmit check for 12220095-6001 failed and returned exit status 1.

INFO:root:Found 3 file(s).

Running presubmit commit checks ...
Running /b/commit-queue/workdir/chromium/PRESUBMIT.py
Running /b/commit-queue/workdir/chromium/cc/PRESUBMIT.py

** Presubmit ERRORS **
Missing LGTM from an OWNER for files in these directories:
    content/renderer/gpu

Presubmit checks took 1.0s to calculate.

[apatrick_chromium, 2013-02-12 19:30:26]

content/renderer/gpu/ LGTM

[commit-bot: I haz the power, 2013-02-12 23:07:53]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/12220095/6001

[commit-bot: I haz the power, 2013-02-13 00:01:21]

Sorry for I got bad news for ya.
Compile failed with a clobber build on win7_aura.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win7_aura&...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-02-13 02:33:32]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/12220095/10003

[commit-bot: I haz the power, 2013-02-13 03:42:17]

Sorry for I got bad news for ya.
Compile failed with a clobber build on win.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win&number...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-02-13 04:22:43]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/epenner@chromium.org/12220095/5005

[commit-bot: I haz the power, 2013-02-13 08:21:33]

Change committed as 182162

[joth, 2013-04-05 05:30:32]

https://chromiumcodereview.appspot.com/12220095/diff/5005/content/renderer/gp...
File content/renderer/gpu/compositor_output_surface.cc (right):

https://chromiumcodereview.appspot.com/12220095/diff/5005/content/renderer/gp...
content/renderer/gpu/compositor_output_surface.cc:155: base::PlatformThreadId
g_last_thread = base::PlatformThread::CurrentId();
removing the static here makes the dcheck below pointless? (And, g_ prefix does
not make sense as it's a plain local var)

can probably write this as:
  static NonThreadSafe g_nts;
  DCHECK(g_nts.CalledOnValidThread());
(no need for ifdef)