Closed Bug 1277961 Opened 9 years ago Closed 9 years ago

Crash in mozalloc_abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::net::PNeckoChild::Write

Categories

(Core :: Networking: HTTP, defect, P1)

Unspecified
Windows 10
defect

Tracking

()

RESOLVED DUPLICATE of bug 1277582
Tracking Status
e10s + ---
firefox49 + fix-optional

People

(Reporter: mccr8, Assigned: valentin)

References

Details

(Keywords: crash, topcrash, Whiteboard: [necko-active], e10st?)

Crash Data

Attachments

(1 obsolete file)

This bug was filed from the Socorro interface and is report bp-3c46782a-5ef5-4d5e-a799-3635a2160603. ============================================================= Right now, this is a #8 crash for the last 7 days, when looking at build date. I'm not sure which fatal error this is. I think the two for this method are: FatalError("NULL actor value passed to non-nullable param"); FatalError("actor has been |delete|d");
Oh wow, this is a beacon that is being initiated as a result of a page being hidden. Specifically, the page is being hidden as a result of TabChild::ActorDestroy, so guess which tab child we try to send along with the new HTTP request?
Crash Signature: [@ mozalloc_abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::net::PNeckoChild::Write] → [@ mozalloc_abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::net::PNeckoChild::Write] [@ mozalloc_abort | NS_DebugBreak | mozilla::ipc::FatalError | mozilla::net::PNeckoChild::FatalError | mozilla::net::PNeckoChild::Write]
Correction, this signature is number #7, #2 is bug 1277582.
See Also: → 1277582
Assignee: nobody → valentin.gosu
Tracking this top crash - 49+
Priority: -- → P1
Whiteboard: [necko-active]
Depends on: 1268559
Whiteboard: [necko-active] → [necko-active], e10st?
This happens to me when shutting down the system without closing firefox. Crash report: https://crash-stats.mozilla.com/report/index/1ed4ba6b-5e83-4762-a898-629062160617
From the crash stack in report bp-3c46782a-5ef5-4d5e-a799-3635a2160603, we see mozilla::ipc::MessageChannel::OnNotifyMaybeChannelError(), which indicates that the child process observes IPC channel breakage (due to parent process shutdown). From the raw dump, the crash reason is "actor has been |delete|d". So we need to prevent reusing the dead/dying PBrowserChild by PNeckoChild deep inside the stack of PBrowserChild::ActorDestroy().
Maybe HttpChannelChild::ConnectParent() also needs this check?
Attachment #8765853 - Flags: feedback?(josh)
Noting this is still the #2 topcrash in 49, so if we do get an effective fix into m-c please request uplift to aurora.
Hi Valentin, are you currently working on this bug? If not I'd like to proceed with the current patch. Thanks.
Flags: needinfo?(valentin.gosu)
Hi Cervantes. Yes, I have a patch in bug 1277582.
Depends on: 1277582
Flags: needinfo?(valentin.gosu)
Comment on attachment 8765853 [details] [diff] [review] Don't use a destoryed TabChild in HttpChannelChild::ContinueAsyncOpen(). Obsolete this patch since bug 1277582 now has covered this bug with a new revision.
Attachment #8765853 - Attachment is obsolete: true
Attachment #8765853 - Flags: feedback?(josh)
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: