diff options
| author | alexmos <alexmos@chromium.org> | 2015-04-09 18:29:04 -0700 |
|---|---|---|
| committer | Commit bot <commit-bot@chromium.org> | 2015-04-10 01:29:34 +0000 |
| commit | 7540d278a1238dfe50d0ba5c98669b260aa34f15 (patch) | |
| tree | 6ec44b11fbbb6d797938055b7b6079e688ac603d /net/cookies/cookie_monster.cc | |
| parent | 9d55de90d77e8a916bcf003b2558ceadc422e0eb (diff) | |
| download | chromium_src-7540d278a1238dfe50d0ba5c98669b260aa34f15.zip chromium_src-7540d278a1238dfe50d0ba5c98669b260aa34f15.tar.gz chromium_src-7540d278a1238dfe50d0ba5c98669b260aa34f15.tar.bz2 | |
Don't crash when calling willCheckAndDispatchMessageEvent with no proxy.
It turns out to be possible to reach
RenderFrameImpl::willCheckAndDispatchMessageEvent when there is no
render_frame_proxy_. This happens when the renderer process is about
to go away, and the top frame has a subframe that tries to send a
postMessage to the top frame in its unload handler.
BUG=475651
TEST=Crashes in the bug report go away. Manual repro: navigate to any
of the URLs in the crash report (e.g.,
http://jsatt.com/blog/class-based-celery-tasks/). Navigate the tab
cross-site to force a process swap and make sure the renderer doesn't
crash.
Review URL: https://codereview.chromium.org/1073963003
Cr-Commit-Position: refs/heads/master@{#324565}
Diffstat (limited to 'net/cookies/cookie_monster.cc')
0 files changed, 0 insertions, 0 deletions