1140115 - crash in `anonymous namespace''::ChildImpl::OpenProtocolOnMainThread

Status: RESOLVED WORKSFORME

(Core :: DOM: Content Processes, defect, P1)

Description

[Trevor Rowbotham [:rowbot]]

Unfortunately, I don't have any steps to reproduce this.  I keep having crash reports with this signature show up in my list of crash reports, however, the browser never crashes and the crash reporter never shows up.

bp-6738679b-95fd-4342-bdef-5d2b32150220
bp-6738679b-95fd-4342-bdef-5d2b32150220
bp-17e71f73-1294-4cd4-a07c-2ea502150305

Comment 1

[Benjamin Smedberg]

These are content-process crashes, so if a tab wasn't open with that content process there would be no UI to submit them. Do you have e10s enabled?

Comment 2

[Trevor Rowbotham [:rowbot]]

Yes, I have e10s enabled.

Comment 3

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

Blake or Ben, do you guys have any idea what's going on here?

Comment 4

[u279076]

I'm not sure if this is the same but there are legitimate crash reports in Socorro with this signature (238 over the last month to be exact):
https://crash-stats.mozilla.com/report/list?product=Firefox&range_unit=days&range_value=28&signature=%60anonymous+namespace%27%27%3A%3AChildImpl%3A%3AOpenProtocolOnMainThread%28nsIEventTarget*%29#tab-reports

Looking at this further:
* all of these are on Windows with Windows 7 accounting for 68%
* Firefox 39.0a1 has the most share with 25% but there are reports going back to Firefox 31.0
* almost all of the top URLs are on a yahoo.com domain

Comment 5

[Ben Turner (not reading bugmail, use the needinfo flag!)]

It looks like we're hitting http://hg.mozilla.org/releases/mozilla-release/diff/e75583a0cbbb/ipc/glue/BackgroundImpl.cpp

Comment 6

[Ben Turner (not reading bugmail, use the needinfo flag!)]

Er. http://hg.mozilla.org/releases/mozilla-release/diff/e75583a0cbbb/ipc/glue/BackgroundImpl.cpp#l1.1736

Comment 7

[Tracy Walker [:tracy]]

There have been no reports of this crash signature after 2015040303. Marking this wfm as it appears to be fixed with no known source of that fix.

Comment 8

[Trevor Rowbotham [:rowbot]]

Looks like this is still kicking around.  I haven't seen this for a few weeks prior to this new report[1], so if this was fixed by some unknown source like suggested in comment 7, that may have regressed.  I'm going to reopen this and restore the needinfo that was removed in comment 7.

[1] bp-a4dcb37f-ae22-42f4-8a14-a95ee2150502

Comment 9

[u279076]

I don't think this was fixed and regressed as I'm seeing reports for basically *all* Firefox versions since Firefox 33. That said, volume remains really low - 64 reports over the last month (down 73% since March). Unless it's immediately evident to :mrbkap what's causing this I do not think this is important enough to be prioritized.

Comment 10

[Blake Kaplan (:mrbkap) (inactive)]

Yeah, it's really unclear what's happening here. If this isn't a top crasher, I'm going to let it be for the moment.

Comment 11

[Tracy Walker [:tracy]]

It's definitely not a top crasher.  I'll plus it, and we can take a look at how it's doing when we triage that list.

Comment 12

[Jim Mathies [:jimm]]

Currently the #3 top crash in our beta e10s experiment.

Comment 13

[Brad Lassey [:blassey] (use needinfo?)]

renom'ing. Top crash in beta experiment. Poiru, can you indicate where this is in the top crash list?

Comment 14

[Biru [:poiru]]

(In reply to Brad Lassey [:blassey] (use needinfo?) from comment #13)
> renom'ing. Top crash in beta experiment. Poiru, can you indicate where this
> is in the top crash list?

This is #10 (1.22%). See bug 1249209 comment 2 for full list.

Comment 15

[Blake Kaplan (:mrbkap) (inactive)]

Contrary to comment 5, these crashes all appear to be failing to open the actual pipe to the parent process (and we shouldn't be in shutdown, otherwise the line numbers would be different). Looking at the code, the most likely culprit would be if we failed to create a HANDLE for the pipe; but talking to jimm, that should be very, very rare.

Furthermore, this seems to have disappeared entirely, [1] shows that there are only 18 instances of this, all on beta, and all on WINNT. I'm going to close this as WFM.

[1] https://crash-stats.mozilla.com/signature/?signature=%60anonymous+namespace%27%27%3A%3AChildImpl%3A%3AOpenProtocolOnMainThread&_columns=date&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=reason&_columns=address&page=1#aggregations

Comment 16

[florian.billebault]

Hello, it look like it regressed again (or reappeared as the status has been changed without any intervention (hope of luck ?)). It happens each time i open my firefox a lot of time (developper edition, 49.0a2) and sometimes after it.

Comment 17

[Andrew McCreight [:mccr8]]

See bug 1278443 for a newer incarnation of this signature.