1351327 - Crash in mozalloc_abort | Abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::dom::PContentChild::Read

Status: RESOLVED WONTFIX

(Core :: DOM: Service Workers, defect, P3)

Description

[Marcia Knous [:marcia]]

This bug was filed from the Socorro interface and is 
report bp-02c5ca94-847a-4fc5-b3cc-2764a2170328.
=============================================================

This crash is new in 53 beta 5 and showed up on the explosiveness report: http://bit.ly/2mMvJsV

All the crashes seem to be confined to Mac OS and affects all versions of Mac OS.

Not sure which component this belongs in - I just made my best guess.

Comment 1

[Benjamin Smedberg]

Does anyone know why this doesn't have a crash reason annotation?

Comment 2

[Marcia Knous [:marcia]]

As phillip pointed out this has not been seen after Beta 5, so we can close it out if it doesn't reappear.

Comment 3

[Andrew McCreight [:mccr8]]

IPC was a reasonable start, but I looked at a half dozen of these and they all have stuff like mozilla::dom::PContentChild::Read(nsTArray<mozilla::dom::ServiceWorkerRegistrationData>*, IPC::Message const*, PickleIterator*) in the stack, so this is or was some service worker issue, I think.

Comment 4

[Andrew McCreight [:mccr8]]

From the array, I'd guess this is the PContent InitServiceWorkers message.

Comment 5

[Eric Rahm [:erahm]]

(In reply to Benjamin Smedberg [:bsmedberg] from comment #1)
> Does anyone know why this doesn't have a crash reason annotation?

It looks like we only annotate in the parent process, but this was from the child process [1].

[1] https://hg.mozilla.org/releases/mozilla-beta/annotate/62467dd7218f/ipc/glue/ProtocolUtils.cpp#l302

Comment 6

[Benjamin Smedberg]

Ooh, we should fix that! That probably dates from the days where crash annotations in children were main-thread-only.

Comment 7

[Trevor Skywalker]

(In reply to Eric Rahm [:erahm] (please no mozreview requests) from comment #5)
> (In reply to Benjamin Smedberg [:bsmedberg] from comment #1)
> > Does anyone know why this doesn't have a crash reason annotation?
> 
> It looks like we only annotate in the parent process, but this was from the
> child process [1].
> 
> [1]
> https://hg.mozilla.org/releases/mozilla-beta/annotate/62467dd7218f/ipc/glue/
> ProtocolUtils.cpp#l302

(In reply to Benjamin Smedberg from comment #6)
> Ooh, we should fix that! That probably dates from the days where crash
> annotations in children were main-thread-only.

Therefore, this bug Depends on: 

Bug 1169641
AnnoteCrashReporter like API so that child processes annotations show up in the parent reports
NEW Unassigned (NEW bug with no priority) 

Reported: 3 years ago
Modified: 3 years ago

(In reply to Milan Sreckovic [:milan] from comment #0)
> From a side conversation with :ted.mielczarek and :jimm :
> 
> When you call AnnotateCrashReport in a child process the annotation is sent
> via IPC to the parent process, but it's not stored in the global annotations
> hashtable there, there's a hashtable that hangs off of
> PCrashReporterParent(?) that holds them, and when the child crashes it
> provides them to the crashreporter code to write out the .extra file. When
> the parent crashes it just writes out the global hashtable.
> 
> ... we could special-case certain annotations to make their way into the
> parent's global hashtable, or we can add an extra API for this (we have
> special APIs for other kinds of annotations).

Comment 8

[BugBot [:suhaib / :marco/ :calixte]]

Closing because no crashes reported for 12 weeks.