681211 - crash PL_DHashTableEnumerate in the DOM memory reporter

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Ludovic Hirlimann [:Usul]]

This bug was filed from the Socorro interface and is 
report bp-e4666f13-6df5-4e1e-9149-029df2110823 .
============================================================= 

Got there while trying to open about:memory

Comment 1

[Boris Zbarsky [:bzbarsky]]

It's a null-deref; is it possible for |windows| to come back null?  I'd think so, if the IPC message races shutdown.

Comment 2

[Ludovic Hirlimann [:Usul]]

(In reply to Boris Zbarsky (:bz) from comment #1)
> It's a null-deref; is it possible for |windows| to come back null?  I'd
> think so, if the IPC message races shutdown.

I wasn't closing firefox on the mobile. Just trying to open about:memory.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Yes, but I'm talking about shutdown of the content process, not the chrome process.  I wouldn't expect it to be happening in this case yet, so presumably the null is for a different reason, but once we move to process-per-tab content processes will be shutting down all the time.

Comment 4

[Mounir Lamouri (:mounir)]

Attachment: Patch v1

Let's try that patch.

I see only two reasons why |windows| could be null: whether nsGlobalWindow ctor has never been called or nsGlobalWindow::Shutdown has been called. I have a patch adding a ::Init method to nsGlobalWindow which would remove the requirement of nsGlobalWindow ctor being called. Though, it seems unlikely that the ctor isn't called... but I don't know that much our e10s architecture...

Anyway, this patch should not hurt I believe ;)

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 555353 [details] [diff] [review]
Patch v1

r=me

Comment 6

[Mounir Lamouri (:mounir)]

I don't know how, I didn't see this review has been done... Thanks for the reminder Boris :)

Comment 7

[Marco Bonardo [:mak]]

http://hg.mozilla.org/mozilla-central/rev/80a4a3d420ea