1021451 - WebRTC Chat between FFOS and Desktop Computer Resulted in Graphics Crash - mozalloc_abort(char const*) | NS_DebugBreak | mozilla::ipc::FatalError(char const*, char const*, int, bool) | mozilla::layers::PSharedBufferManagerChild::SendAllocateGrallocBuffer(

Status: RESOLVED WORKSFORME

(Core :: Graphics: Layers, defect)

Description

[Kate Glazko]

https://crash-stats.mozilla.com/report/index/ee878ae8-ff06-4375-b578-59bb92140606

Here is the crash report, latest build on Flame.

Comment 1

[Sotaro Ikeda [:sotaro]]

Is there a STR to reproduce the crash?

Comment 2

[Milan Sreckovic [:milan] (needinfo for best results)]

Buffer ownership passing?

Comment 3

[Sotaro Ikeda [:sotaro]]

Can we have a logcat log when the crash happens?

Comment 4

[Sotaro Ikeda [:sotaro]]

I tested WebRTC between master flame and Firefox for Windows. It works correctly without crash.

Comment 5

[Sotaro Ikeda [:sotaro]]

(In reply to kglazko from comment #0)
> https://crash-stats.mozilla.com/report/index/ee878ae8-ff06-4375-b578-
> 59bb92140606
> 
> Here is the crash report, latest build on Flame.

The crash just says that a AllocateGrallocBuffer transaction between SharedBufferManagerParent failed by some reasons.

Comment 6

[Kate Glazko]

1.) Start the Web RTC chat on the phone.

2.) Have someone join you on Firefox for Mac on the Computer.

3.) The application crashed and caused an OS restart on my phone.

4.) This is not consistently reproducible, because we tried again after that and it worked fine. However, it might be useful to know that this does happen.

Comment 7

[Sotaro Ikeda [:sotaro]]

(In reply to kglazko from comment #6)
> 1.) Start the Web RTC chat on the phone.
> 
> 2.) Have someone join you on Firefox for Mac on the Computer.
> 
> 3.) The application crashed and caused an OS restart on my phone.

It seems that an actual first crash happened on compositor(b2g process) side and then the transaction failed as in Comment 5.


> 
> 4.) This is not consistently reproducible, because we tried again after that
> and it worked fine. However, it might be useful to know that this does
> happen.

Comment 8

[Kate Glazko]

Should I report this bug to someone else then, and if so, who?

Comment 9

[Sotaro Ikeda [:sotaro]]

(In reply to kglazko from comment #8)
> Should I report this bug to someone else then, and if so, who?

About graphc, milan, nical and I. About webrtc, Randell Jesup seems correct person to notify.

Comment 10

[Sotaro Ikeda [:sotaro]]

And I cc-ed :chiajung. He wrote SharedBufferManager.

Comment 11

[Jason Smith [:jsmith]]

We need to see if we can reproduce this again. This was an OS restart crash, so we really need to get this fixed.

Comment 12

[Chiajung Hung [:chiajung]]

The report says MaybeMagicGrallocBufferHandle deserialing error, which is very strange.
I think it will hard to find a solution without a reliable STR.
The error only occurs when either: 
(1). Too much fds in the fdset
(2). GraphicBuffer::unflatten fail
(3). other internal IPC fail

(1) is quite impossible, as GraphicBuffer usually use up to 2 fds, and we can pass 7 now. (2) and (3) need reproducing to debug.

I'll try it later to see whether I can reproduce the crash.

Comment 13

[Marcia Knous [:marcia]]

I have been hitting intermittent crashes on the homescreen - yesterday I hit https://crash-stats.mozilla.com/report/index/899a2ae0-43de-4d11-93b0-5f0bd2140609. In the app notes of that crash there is a reference to MaybeMagicGrallocBufferHandle.

These crashes seemed to happen very randomly and I have not been able to reproduce them.

Comment 15

[Rachel Pribble [:rpribble]]

Attachment: CrashLogcat?.txt

I encountered this crash on the Flame v2.0 while looking for truncations (connected to data, not wifi, and in Tamil [தமிழ்] language). Attaching the logcat, sorry it's so large.

Repro Steps:
1) Update a Flame to BuildID: 20140630000201
2) Settings > Improve Firefox OS
3) Take a screenshot
4) Tap 'Send Mozilla Feedback', then take a screenshot
5) Tap 'happy', then take a screenshot
6) Continue moving briskly about the 'Improve Firefox OS' settings, taking screenshots

Device went from 'Improve Firefox OS' settings page to the animated fox boot screen (splash animation).


v2.0 Environmental Variables:
Device: Flame v2.0 MOZ ril
BuildID: 20140630000201
Gaia: c0c8ad187c0466285f2580531e09f8322996f561
Gecko: d4dc609bcc8a
Version: 32.0a2
Firmware Version: v122
User Agent: Mozilla/5.0 (Mobile; rv:32.0) Gecko/32.0 Firefox/32

Comment 16

[Sotaro Ikeda [:sotaro]]

This bug might be fixed by Bug 1034294.

Comment 17

[Joshua Mitchell (Inactive)]

comment 15 addressed the steps-wanted request

Comment 18

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

Waiting for bug 1038461 to land before asking for retest.

Comment 19

[Jason Smith [:jsmith]]

(In reply to Joshua Mitchell [:Joshua_M] from comment #17)
> comment 15 addressed the steps-wanted request

Josh - Do we know if comment 15 can be consistently reproducible to use for STR testing?

Comment 20

[Joshua Mitchell (Inactive)]

(In reply to Jason Smith [:jsmith] from comment #19)
> (In reply to Joshua Mitchell [:Joshua_M] from comment #17)
> > comment 15 addressed the steps-wanted request
> 
> Josh - Do we know if comment 15 can be consistently reproducible to use for
> STR testing?

Apparently not - 15 minutes of "moving briskly about the 'Improve Firefox OS' settings, taking screenshots" did not repro this crash (in that reported build) 

Re-adding Steps-wanted

Comment 21

[Joshua Mitchell (Inactive)]

cleaning keywords - please re-tag if this issue is seen again

Comment 22

[Edwin Flores [inactive from 2016-12-01] [:eflores] [:edwin]]

Mass resolving WFM: signature(s) hasn't(/haven't) reported in past 28 days.