Fix 30s pause when starting MinuteMaid WebView

Fix 30s pause when starting MinuteMaid WebView

The app can hog SERIAL_EXECUTOR; we should not assume it is free.

internal bug 20913700

Committed: https://crrev.com/475b52d88628c08a530933d093f39e577648edf6
Cr-Commit-Position: refs/heads/master@{#329178}

[paulmiller, 2015-05-08 23:06:46]

paulmiller@chromium.org changed reviewers:
+ yfriedman@chromium.org

[paulmiller, 2015-05-08 23:07:08]

PTAL

[boliu, 2015-05-08 23:10:07]

fwiw, SERIAL_EXECUTOR isn't really a bulletproof way either, as that just means
a pool of #cpus + 1 threads, which in theory can get clogged up too, but it
works in this case \o/

Long term, chrome probably want its own thread pool implementation (possibly
reusing the native ones that already exist)

[Yaron, 2015-05-11 13:41:08]

lgtm

[Torne, 2015-05-11 15:18:29]

On 2015/05/08 23:10:07, boliu wrote:
> fwiw, SERIAL_EXECUTOR isn't really a bulletproof way either, as that just
means
> a pool of #cpus + 1 threads, which in theory can get clogged up too, but it
> works in this case \o/

you mean THREAD_POOL_EXECUTOR?

> Long term, chrome probably want its own thread pool implementation (possibly
> reusing the native ones that already exist)

Yeah, maybe we should just wrap this across the codebase to use our own
ThreadPoolExecutor that isn't the default one? That would mean webview apps gain
even more threads, though.

[boliu, 2015-05-11 15:52:22]

On 2015/05/11 15:18:29, Torne wrote:
> On 2015/05/08 23:10:07, boliu wrote:
> > fwiw, SERIAL_EXECUTOR isn't really a bulletproof way either, as that just
> means
> > a pool of #cpus + 1 threads, which in theory can get clogged up too, but it
> > works in this case \o/
> 
> you mean THREAD_POOL_EXECUTOR?

Yep..

> 
> > Long term, chrome probably want its own thread pool implementation (possibly
> > reusing the native ones that already exist)
> 
> Yeah, maybe we should just wrap this across the codebase to use our own
> ThreadPoolExecutor that isn't the default one? That would mean webview apps
gain
> even more threads, though.

[paulmiller, 2015-05-11 16:54:38]

The CQ bit was checked by paulmiller@chromium.org

[commit-bot: I haz the power, 2015-05-11 16:56:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1129033005/1

[commit-bot: I haz the power, 2015-05-11 18:00:18]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-05-11 18:02:05]

Patchset 1 (id:??) landed as
https://crrev.com/475b52d88628c08a530933d093f39e577648edf6
Cr-Commit-Position: refs/heads/master@{#329178}