OOPIF: Don't crash when running layout tests that have an OOP iframe.

OOPIF: Don't crash when running layout tests that have an OOP iframe.

This CLs fixes a few places in layout test code that were assuming
that the top-level frame is local, which was causing renderer crashes
when running a layout test containing a cross-site frame with
--site-per-process.  For example, most tests under
http/tests/security/postMessage/ ran into this.

BUG=474792
TEST=http/tests/security/postMessage/* tests don't crash with --site-per-process

Committed: https://crrev.com/297eab63774e6b75d2fccef92c1991707be16246
Cr-Commit-Position: refs/heads/master@{#324311}

[alexmos, 2015-04-08 17:45:58]

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

[alexmos, 2015-04-08 17:45:59]

Charlie, can you please take a look?  These bailouts get layout tests to not
crash with cross-site iframes.  I hope all of these are ok to just skip for
remote frames, since they should also happen in the process rendering the actual
main frame.

[Charlie Reis, 2015-04-08 21:46:20]

[+jochen to CC for WebKitTestRunner]

LGTM as a short term workaround to get layout tests to run in
--site-per-process, but WebKitTestRunner is likely going to need a redesign to
handle OOPIFs properly.  Let's chat with Jochen about it separately.

[alexmos, 2015-04-08 21:55:08]

On 2015/04/08 21:46:20, Charlie Reis wrote:
> [+jochen to CC for WebKitTestRunner]
> 
> LGTM as a short term workaround to get layout tests to run in
> --site-per-process, but WebKitTestRunner is likely going to need a redesign to
> handle OOPIFs properly.  Let's chat with Jochen about it separately.

Thanks.  Agreed that there's a lot more to do in testRunner for OOPIF.  For
example, DumpFramesAsText and similar functions in web_test_proxy.cc aren't
going to work.

[alexmos, 2015-04-08 21:55:19]

The CQ bit was checked by alexmos@chromium.org

[commit-bot: I haz the power, 2015-04-08 21:56:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1073533002/1

[commit-bot: I haz the power, 2015-04-08 23:58:30]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-04-08 23:59:22]

Patchset 1 (id:??) landed as
https://crrev.com/297eab63774e6b75d2fccef92c1991707be16246
Cr-Commit-Position: refs/heads/master@{#324311}