Allow IPCs during RenderFrameObserver::FrameDetached()

Allow IPCs during RenderFrameObserver::FrameDetached()

Currently IPCs are disallowed to disincentivize observers from
manipulating the DOM after the frame is detached, which may have
unpredictable consequences. However this shouldn't resrict observers
which want to send IPCs that don't require maniuplating or inspecting
the frame. This resitiction is now removed and the restriction on
manipulating the frame is more explicitly stated.

BUG=450806
Committed: https://crrev.com/1937042392cdbd33b9c1d6f26fa165d44709c664
Cr-Commit-Position: refs/heads/master@{#320838}

[Garrett Casto, 2015-03-16 23:00:22]

gcasto@chromium.org changed reviewers:
+ nasko@chromium.org

[Garrett Casto, 2015-03-16 23:00:23]

nasko@: PTAL. I didn't update the documentation in RenderViewObserver since in
general functions there don't have documentation and I'm assuming that it's
going away soon anyway. It's obviously trivial to add if you'd like.

[nasko, 2015-03-16 23:11:19]

+site-isolation-reviews@chromium.org

Not updating render_view_observer is fine, since that file has pretty much no
comments.

I do have one concern with the CL. Once that is fixed, I think it will be good
to go.

https://codereview.chromium.org/1014703003/diff/1/content/renderer/render_fra...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/1014703003/diff/1/content/renderer/render_fra...
content/renderer/render_frame_impl.cc:2083: Send(new
FrameHostMsg_Detach(routing_id_));
I would move the sending of this message after the observer notifications.
Otherwise it is possible that the browser side will have deleted the host object
or other state related to this frame, prior to the observers finishing their
work.

[Garrett Casto, 2015-03-16 23:15:58]

https://codereview.chromium.org/1014703003/diff/1/content/renderer/render_fra...
File content/renderer/render_frame_impl.cc (right):

https://codereview.chromium.org/1014703003/diff/1/content/renderer/render_fra...
content/renderer/render_frame_impl.cc:2083: Send(new
FrameHostMsg_Detach(routing_id_));
On 2015/03/16 23:11:18, nasko wrote:
> I would move the sending of this message after the observer notifications.
> Otherwise it is possible that the browser side will have deleted the host
object
> or other state related to this frame, prior to the observers finishing their
> work.

Done.

[nasko, 2015-03-16 23:22:22]

LGTM

[Garrett Casto, 2015-03-16 23:23:29]

The CQ bit was checked by gcasto@chromium.org

[commit-bot: I haz the power, 2015-03-16 23:24:02]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1014703003/20001

[commit-bot: I haz the power, 2015-03-17 01:02:21]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-03-17 01:03:04]

Patchset 2 (id:??) landed as
https://crrev.com/1937042392cdbd33b9c1d6f26fa165d44709c664
Cr-Commit-Position: refs/heads/master@{#320838}