513334 - Crash at [@nsEventListenerManager::Release() ]

Status: RESOLVED DUPLICATE of bug 500879

(Firefox :: General, defect, P2)

Description

[[:Cww]]

Crash #30 on crash-stats

http://crash-stats.mozilla.com/report/list?product=Firefox&query_search=signature&query_type=exact&query=&date=&range_value=1&range_unit=weeks&do_query=1&signature=nsEventListenerManager%3A%3ARelease%28%29

Comment 1

[[:Cww]]

CCing people for help.

Comment 2

[Jesse Ruderman]

More specifically, this is topcrash #12 for Firefox 3.5.2.

Comment 3

[Johnny Stenback (:jst)]

Smaug, any thoughts? Based on some of the comments I saw this might have been a regression from 3.5.1 to 3.5.2, but there's not necessarily enough data to back that up yet.

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

So something is probably not addrefing some event target when it should;
a missing NS_ADDREF for an element somewhere (before or during event dispatch).
Or perhaps it is a window which isn't addrefed
http://crash-stats.mozilla.com/report/index/4f5fdbd0-abcf-48e2-a2c6-975912090904
Or, it could be an extra release somewhere too.

I wonder if some binary extension is doing something bad.
I couldn't see any common .dll, though.

I don't recall any changes to event dispatching which could have caused these
problems. (In fact the last change to nsEventDispatcher itself on 1.9.1 branch happened almost a year ago.)

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

Seems like the crash isn't event type specific.
So in some case it is C++ dispatching a commandupdate event, but
in other case it is JS creating and dispatching an event.
Or occasionally it is the focus event...

Comment 6

[Olli Pettay [:smaug][[e-mail]]]

Hmm, or could it be a missing addref or release to event listener manager

Comment 7

[[:Cww]]

Anything we should tell users with this crash?

Comment 8

[Jesse Ruderman]

Just the usual: "Mozilla is aware that this is a common type of crash, and is trying to figure out what causes it. Please add information to bug N."

Can you run your module-listing thing on this crash?  Any obvious correlations between this crash and having certain modules installed?

Comment 9

[[:Cww]]

Attachment: analysis results take 1

Right now, all it seems to do is correlate with being windows only and 3.5 specific.

Comment 10

[Jesse Ruderman]

Half of the crash reports have rlls.dll and rlxg.dll ("Relevant Knowledge is a variant of the MarketScore spyware"?).

Another quarter have have pmxg.dll, which just happens to have the same version number (1.3.323.1) as rlxg.dll.  ("PremierOpinion adware"?).

So, mostly caused by unwanted software hooking into Firefox.

Comment 11

[Jesse Ruderman]

Thanks Cww, your new module analysis made that pretty easy to figure out.

Comment 12

[Mike Beltzner [:beltzner, not reading bugmail]]

Need to fix this, either by blocklisting or by fixing. Is there a better component than Firefox::General?

Comment 13

[Johnny Stenback (:jst)]

(In reply to comment #6)
> Hmm, or could it be a missing addref or release to event listener manager

Smaug, this is looking like it's directly related to the cycle collector topcrash bugs, in particular bug 500879 comment 14, and bug 504392. I can't say for sure that this is entirely due to third party code, but that's what it's looking like right now.

Comment 14

[Johnny Stenback (:jst)]

Duping against bug 500879 per todays crashkill meeting as this is due to the same reasons.