Force a swap on resize to work around Intel driver issues with PostSubBuffer.

Force a swap on resize to work around Intel driver issues with PostSubBuffer.

BUG=chrome-os-partner:8223
TEST=run aura chrome with --ui-enable-partial-swap on device, plug in external screen.

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=125956

[jonathan.backer, 2012-03-09 19:51:38]

On 2012/03/09 19:36:38, piman wrote:

I'm concerned that this will also be called on desktop Linux
(http://code.google.com/p/chromium/issues/detail?id=116187), where we probably
don't want to swap --- the joy of driver bugs.

Maybe the proper thing is to set a flag, so that on the first PostSubBuffer
after a resize, we actually swap?

[piman, 2012-03-09 20:35:44]

On 2012/03/09 19:51:38, jonathan.backer wrote:
> On 2012/03/09 19:36:38, piman wrote:
> 
> I'm concerned that this will also be called on desktop Linux
> (http://code.google.com/p/chromium/issues/detail?id=116187), where we probably
> don't want to swap --- the joy of driver bugs.
> 
> Maybe the proper thing is to set a flag, so that on the first PostSubBuffer
> after a resize, we actually swap?

That's too late. The problem is that we call Resize at the beginning of the
frame, then draw the entire thing, then call PostSubBuffer. If we don't do the
swap on resize, then we don't get the updated back buffer for this frame, so it
will be incomplete, and after the swap, the new back buffer will be garbage too
(but we did a post so we expect it to be correct). So we'd need a way to signal
upstream to all the clients that the post didn't actually do a post, and that we
lost both that frame, and the back buffer, so we need to redraw everything
again. All those clients will need to add retry logic for that. That seems
unnecessarily complex.

With this, we avoid almost all the problems.
First we only do that swap on drivers that support MESA post, which are those
that would have the problem at the first place.
Second, this happens on Resize, where we already lost the contents of the
window. So swapping "garbage" won't really be worse. I tried on desktop with
nvidia driver (and disabled the test for the extension to always do the swap),
and didn't get any extra artifact.

[jonathan.backer, 2012-03-09 20:47:20]

LGTM

I agree that swapping garbage is no worse.

We need this driver bug fixed.

[commit-bot: I haz the power, 2012-03-09 20:48:30]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/9669006/1

[commit-bot: I haz the power, 2012-03-09 23:58:16]

Commit queue had an internal error.
Something went really wrong, probably a crash, a hickup or
simply the monkeys went out for diner.
The relevant dude was pinged already.

[commit-bot: I haz the power, 2012-03-10 00:07:52]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/piman@chromium.org/9669006/1