Don't crash the browser when a webui page misbehaves

Don't crash the browser when a webui page misbehaves

Just ignore the call (which is what would happen anyways), and upload a
minidump

BUG=516690
R=creis@chromium.org,dbeam@chromium.org,estade@chromium.org

Committed: https://crrev.com/b2a49780f87f37649ddb706cff538b7b228b5d17
Cr-Commit-Position: refs/heads/master@{#347875}

[jochen (gone - plz use gerrit), 2015-09-08 11:01:08]

ptal

this would keep the crash reports coming in, avoids leaking information, but
also wouldn't crash any longer..

[Charlie Reis, 2015-09-08 18:47:22]

On 2015/09/08 11:01:08, jochen wrote:
> ptal
> 
> this would keep the crash reports coming in, avoids leaking information, but
> also wouldn't crash any longer..

I don't see how this is much different from dbeam@'s proposal to drop the JS if
the target frame isn't ready.  estade@ wasn't happy with that approach, per
comment 23 on the bug.

I agree that it's a minimal step toward not crashing Chrome for users, but we
might as well skip the DumpWithoutCrashing calls (which are quite noticeably
slow) if we know this is going to happen.  The additional crash reports don't
tell us much new.

(Note: As with my comment on https://codereview.chromium.org/1266273006/, it
still seems like WebUI code risks doing other dangerous things if it isn't
accounting for when the frame is ready, but dbeam@ and estade@ have said that
the JS execution is really the only behavior to worry about.  From that
perspective, I'm fine with dropping the JS for now and eventually having a WebUI
refactor that gets it right.)

[jochen (gone - plz use gerrit), 2015-09-08 20:17:05]

we still keep getting new crash reports, so I think keeping the dump call is
useful.

I'd also rather go with a small fix for now which we can merge back

[Evan Stade, 2015-09-08 21:06:04]

lgtm

[Charlie Reis, 2015-09-08 21:17:25]

On 2015/09/08 20:17:05, jochen wrote:
> we still keep getting new crash reports, so I think keeping the dump call is
> useful.
> 
> I'd also rather go with a small fix for now which we can merge back

If the motivation for DumpWithoutCrashing is to give developers an incentive to
fix the broken WebUI pages (without disrupting the user too much), then I guess
this will do.  LGTM.

[Dan Beam, 2015-09-09 01:14:10]

lgtm if you're waiting for me

[jochen (gone - plz use gerrit), 2015-09-09 07:40:38]

The CQ bit was checked by jochen@chromium.org

[commit-bot: I haz the power, 2015-09-09 07:41:08]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1332513002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1332513002/1

[commit-bot: I haz the power, 2015-09-09 09:10:02]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-09-09 09:11:15]

Patchset 1 (id:??) landed as
https://crrev.com/b2a49780f87f37649ddb706cff538b7b228b5d17
Cr-Commit-Position: refs/heads/master@{#347875}