Turn on AreCrossProcessFramesPossible by default.

This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

[Closed in favor of https://crrev.com/448380 and https://crrev.com/448761]

[Charlie Reis, 2017-01-31 23:22:41]

The CQ bit was checked by creis@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-01-31 23:23:31]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-01-31 23:50:07]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-01-31 23:50:08]

Dry run: Try jobs failed on following builders:
  linux_chromium_tsan_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[Charlie Reis, 2017-01-31 23:58:51]

Description was changed from

==========
Turn on AreCrossProcessFramesPossible by default.

BUG=545200
TEST=No visible behavior change.
==========

to

==========
Turn on AreCrossProcessFramesPossible by default.

This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
==========

[Charlie Reis, 2017-02-01 02:04:12]

Description was changed from

==========
Turn on AreCrossProcessFramesPossible by default.

This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
==========

to

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
==========

[Charlie Reis, 2017-02-03 01:11:05]

Description was changed from

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
==========

to

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

[Charlie Reis, 2017-02-03 01:11:54]

The CQ bit was checked by creis@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-03 01:12:29]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Charlie Reis, 2017-02-03 01:14:03]

creis@chromium.org changed reviewers:
+ alexmos@chromium.org

[Charlie Reis, 2017-02-03 01:14:04]

Alex, can you take a look at some of these TODOs and possibly take over this CL
while I'm out?  Lucas confirmed that we should change the guest one, we should
confirm with James about the input event one and I'm worried about the RFHM one.

Otherwise, I think we should be set for turning this on, and the sooner the
better for spotting any regressions.  :)

Thanks!

[commit-bot: I haz the power, 2017-02-03 02:34:56]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-03 02:34:56]

Dry run: This issue passed the CQ dry run.

[Charlie Reis, 2017-02-03 03:34:36]

(We might also consider landing the unrelated fixes before flipping on the
policy function, in case the policy part needs to be reverted.)

[wjmaclean, 2017-02-03 17:01:53]

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

[wjmaclean, 2017-02-03 17:01:53]

https://codereview.chromium.org/2668013005/diff/20001/content/browser/rendere...
File content/browser/renderer_host/render_widget_host_view_event_handler.cc
(right):

https://codereview.chromium.org/2668013005/diff/20001/content/browser/rendere...
content/browser/renderer_host/render_widget_host_view_event_handler.cc:843: //
TODO(creis): Is this right?
Yes, I think this is correct. (Although these days I think the mouse-wheels get
converted to GestureScrolls, and not the other way around).

We should start to look at the cases where there is no delegate, or the delegate
has no input event router, and see if they still exist.

There's already a bug filed to make |disable_input_event_router_for_testing_|
unnecessary.

[alexmos, 2017-02-04 02:03:01]

On 2017/02/03 01:14:04, Charlie Reis (OOO til Feb 13) wrote:
> Alex, can you take a look at some of these TODOs and possibly take over this
CL
> while I'm out?  Lucas confirmed that we should change the guest one, we should
> confirm with James about the input event one and I'm worried about the RFHM
one.
> 
> Otherwise, I think we should be set for turning this on, and the sooner the
> better for spotting any regressions.  :)
> 
> Thanks!
>
> (We might also consider landing the unrelated fixes before flipping on the
> policy function, in case the policy part needs to be reverted.)

Sure.  I'm spinning up two CLs from this one:
https://codereview.chromium.org/2673963003 for the unrelated fixes
https://codereview.chromium.org/2676823003 for the actual flip

Also see my responses about the TODOs.  I'll discuss the RFHM one with Nasko 
and others to see whether it's worth fixing.

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
File content/browser/frame_host/render_frame_host_manager.cc (right):

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
content/browser/frame_host/render_frame_host_manager.cc:1102: //
--process-per-tab.
Filed https://bugs.chromium.org/p/chromium/issues/detail?id=688617.  It does
appear that we are currently breaking cross-process navigations under
--single-process if we are also in an OOPIF mode.

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
File content/browser/frame_host/render_widget_host_view_guest.cc (right):

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
content/browser/frame_host/render_widget_host_view_guest.cc:358: // TODO(creis):
This is wrong.  Should check for OOPIF webview mode instead.
I actually think this is ok.  The description in the CL that introduced this,
https://codereview.chromium.org/1845083002, seems to imply that this should
trigger based on whether the input event router is installed, which James
confirmed should be triggered by AreCrossProcessFramesPossible().  I chatted
with Lucas, and he also agreed that this call is ok as is.

[Charlie Reis, 2017-02-13 22:02:49]

Thanks for landing these while I was out!  I'll close this and will take a look
at https://crbug.com/688617.  (Also surprised to see the Android regression in
https://crbug.com/690229, which we should investigate.)

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
File content/browser/frame_host/render_widget_host_view_guest.cc (right):

https://codereview.chromium.org/2668013005/diff/20001/content/browser/frame_h...
content/browser/frame_host/render_widget_host_view_guest.cc:358: // TODO(creis):
This is wrong.  Should check for OOPIF webview mode instead.
On 2017/02/04 02:03:01, alexmos wrote:
> I actually think this is ok.  The description in the CL that introduced this,
> https://codereview.chromium.org/1845083002, seems to imply that this should
> trigger based on whether the input event router is installed, which James
> confirmed should be triggered by AreCrossProcessFramesPossible().  I chatted
> with Lucas, and he also agreed that this call is ok as is.

Acknowledged.

https://codereview.chromium.org/2668013005/diff/20001/content/browser/rendere...
File content/browser/renderer_host/render_widget_host_view_event_handler.cc
(right):

https://codereview.chromium.org/2668013005/diff/20001/content/browser/rendere...
content/browser/renderer_host/render_widget_host_view_event_handler.cc:843: //
TODO(creis): Is this right?
On 2017/02/03 17:01:53, wjmaclean wrote:
> Yes, I think this is correct. (Although these days I think the mouse-wheels
get
> converted to GestureScrolls, and not the other way around).
> 
> We should start to look at the cases where there is no delegate, or the
delegate
> has no input event router, and see if they still exist.
> 
> There's already a bug filed to make |disable_input_event_router_for_testing_|
> unnecessary.

Acknowledged.

[Charlie Reis, 2017-02-13 22:02:58]

Description was changed from

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

to

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

[Charlie Reis, 2017-02-13 22:03:54]

Description was changed from

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

to

==========
This turns on all of the plumbing required to create out-of-process
iframes, without enabling any policies that create them in practice.
This minimizes the difference between OOPIF and non-OOPIF modes.

Behavior changes include:
 - Creating proxies for child frames and replicating state to them
 - Enabling cross-process policy checks for child frame navigations
 - More complex fullscreen toggling logic
 - Slightly different event routing and focus logic

This CL also corrects a few test expectations, which were using the
wrong policy function to detect --site-per-process behavior.

BUG=545200
TEST=No visible behavior change.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation

[Closed in favor of https://crrev.com/448380 and https://crrev.com/448761]
==========