Diagnose renderer-side crash with browser-side stack trace.

Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.

Committed: https://crrev.com/20eb8c2a7547867470e7af030f87e6c2e9bf68ce
Cr-Commit-Position: refs/heads/master@{#371336}

[Charlie Reis, 2016-01-25 19:06:26]

Description was changed from

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.
==========

to

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.
==========

[Charlie Reis, 2016-01-25 19:06:26]

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

[Charlie Reis, 2016-01-25 20:29:15]

Nasko, can you take a look?  If my https://codereview.chromium.org/1618063003/
check is correct, this will help us track down how it happens in the browser
process.  (Unfortunately, I haven't seen any evidence one way or the other yet
in the crashes.)

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
File content/browser/renderer_host/render_view_host_impl.cc (right):

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
content/browser/renderer_host/render_view_host_impl.cc:303: proxy_route_id !=
MSG_ROUTING_NONE);
Note: There's an existing set of crashes on this line (with no bug filed yet):
https://crash.corp.google.com/browse?q=product.name%3D%27Chrome%27%20AND%20%2...

The new ones I'm adding should be easy to distinguish, since their magic
signature will start with "[Dump without crash]".

[nasko, 2016-01-25 21:31:48]

LGTM with a nit.

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
File content/browser/renderer_host/render_view_host_impl.cc (right):

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
content/browser/renderer_host/render_view_host_impl.cc:312:
base::debug::DumpWithoutCrashing();
Multiline if statement requires {}.

[Charlie Reis, 2016-01-25 22:02:33]

Patchset #2 (id:20001) has been deleted

[Charlie Reis, 2016-01-25 22:07:11]

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
File content/browser/renderer_host/render_view_host_impl.cc (right):

https://codereview.chromium.org/1624413002/diff/1/content/browser/renderer_ho...
content/browser/renderer_host/render_view_host_impl.cc:312:
base::debug::DumpWithoutCrashing();
On 2016/01/25 21:31:48, nasko wrote:
> Multiline if statement requires {}.

As we chatted about, I think it's optional in this case (since the statement is
one line), plus we'll be reverting it soon anyway.  I'll go ahead with this
patchset to make sure it's in tomorrow's canary.

Reference:
https://google.github.io/styleguide/cppguide.html#Conditionals

[Charlie Reis, 2016-01-25 22:07:35]

The CQ bit was checked by creis@chromium.org

[commit-bot: I haz the power, 2016-01-25 22:11:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1624413002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1624413002/1

[commit-bot: I haz the power, 2016-01-25 22:21:54]

Description was changed from

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.
==========

to

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.
==========

[commit-bot: I haz the power, 2016-01-25 22:21:55]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-01-25 22:22:57]

Description was changed from

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.
==========

to

==========
Diagnose renderer-side crash with browser-side stack trace.

This will log a crash dump without crashing the browser process, letting
us see how a params object is unexpectedly created with both a main
frame routing ID and a proxy routing ID.

BUG=575245
TEST=Crash dumps in RenderViewHostImpl::CreateRenderView.

Committed: https://crrev.com/20eb8c2a7547867470e7af030f87e6c2e9bf68ce
Cr-Commit-Position: refs/heads/master@{#371336}
==========

[commit-bot: I haz the power, 2016-01-25 22:22:58]

Patchset 1 (id:??) landed as
https://crrev.com/20eb8c2a7547867470e7af030f87e6c2e9bf68ce
Cr-Commit-Position: refs/heads/master@{#371336}