TabManager checks if ServiceWorker exists on the suspending process

TabManager checks if ServiceWorker exists on the suspending process

If the suspending process has a worker thread, currently the worker won't
respond with events any more. This patch is to keep the process running which
has workers.

BUG=671084

[shimazu, 2016-12-13 04:16:35]

shimazu@chromium.org changed reviewers:
+ bashi@chromium.org

[shimazu, 2016-12-13 04:16:36]

ptal. I tried to implement the check, but I found thread hopping looks
necessary.

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
File chrome/browser/memory/tab_manager.cc (right):

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
chrome/browser/memory/tab_manager.cc:470: if
(sw_context->IsServiceWorkerExistsOnProcess(render_process_id))
This is a problem. Do you have any good idea to call this from IO thread?

[nhiroki, 2016-12-13 04:33:58]

nhiroki@chromium.org changed reviewers:
+ nhiroki@chromium.org

[nhiroki, 2016-12-13 04:33:59]

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
File chrome/browser/memory/tab_manager.cc (right):

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
chrome/browser/memory/tab_manager.cc:470: if
(sw_context->IsServiceWorkerExistsOnProcess(render_process_id))
On 2016/12/13 04:16:36, shimazu wrote:
> This is a problem. Do you have any good idea to call this from IO thread?

Are we on the UI thread here? If so, we could use RenderProcessHost's worker
counts:
https://cs.chromium.org/chromium/src/content/public/browser/render_process_ho...

[nhiroki, 2016-12-13 04:34:57]

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
File chrome/browser/memory/tab_manager.cc (right):

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
chrome/browser/memory/tab_manager.cc:470: if
(sw_context->IsServiceWorkerExistsOnProcess(render_process_id))
On 2016/12/13 04:33:59, nhiroki wrote:
> On 2016/12/13 04:16:36, shimazu wrote:
> > This is a problem. Do you have any good idea to call this from IO thread?
> 
> Are we on the UI thread here? If so, we could use RenderProcessHost's worker
> counts:
>
https://cs.chromium.org/chromium/src/content/public/browser/render_process_ho...

(without thread hopping)

[bashi, 2016-12-13 04:40:17]

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
File chrome/browser/memory/tab_manager.cc (right):

https://codereview.chromium.org/2574663002/diff/1/chrome/browser/memory/tab_m...
chrome/browser/memory/tab_manager.cc:470: if
(sw_context->IsServiceWorkerExistsOnProcess(render_process_id))
On 2016/12/13 04:34:56, nhiroki wrote:
> On 2016/12/13 04:33:59, nhiroki wrote:
> > On 2016/12/13 04:16:36, shimazu wrote:
> > > This is a problem. Do you have any good idea to call this from IO thread?
> > 
> > Are we on the UI thread here? If so, we could use RenderProcessHost's worker
> > counts:
> >
>
https://cs.chromium.org/chromium/src/content/public/browser/render_process_ho...
> 
> (without thread hopping)

Thank you for the info. Yes, we may want to use worker_ref_count() in
MemoryCoordinator::CanSuspendRenderer(). Do you have any idea how many renderers
that have worker_ref_count() > 0? If there are many renderers we can't suspend,
MC wouldn't work well...

[shimazu, 2016-12-22 03:11:01]

Description was changed from

==========
TabManager checks if ServiceWorker exists on the suspending process

If the suspending process has a worker thread, currently the worker won't
respond with events any more. This patch is to keep the process running which
has workers.

BUG=671084
==========

to

==========
TabManager checks if ServiceWorker exists on the suspending process

If the suspending process has a worker thread, currently the worker won't
respond with events any more. This patch is to keep the process running which
has workers.

BUG=671084
==========