1143339 - [Mac] crash ~15 seconds after closing a youtube tab [@ CoreFoundation@0x188b7] or [@ CoreFoundation@0x18827] or [@ mozilla::AppleVDADecoder::Shutdown()]

Status: RESOLVED FIXED

(Core :: Audio/Video, defect)

Description

[Benoit Girard (:BenWa)]

This bug was filed from the Socorro interface and is 
report bp-e4c986c6-ee41-4e68-ab88-b48ef2150314.
=============================================================

I got this crash after experiencing bug 1142901.

Comment 1

[Benoit Girard (:BenWa)]

Page:
https://www.youtube.com/watch?v=MBaCMj2XlS0

Comment 2

[Benoit Girard (:BenWa)]

This has been happening to me be about 10-30 seconds having closing a youtube tab after seeing bug 1142901.

Comment 3

[Nickolay_Ponomarev]

I got a similar crash at a different address [@ CoreFoundation@0x19217], also after watching youtube:
https://crash-stats.mozilla.com/report/index/357c1fe4-e99e-4693-ac93-58fb32150315 

The most popular address seems to be [@ CoreFoundation@0x188b7] with >100 crashes:
https://crash-stats.mozilla.com/search/?topmost_filenames=AppleVDADecoder.cpp&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-signature

Seems new in 20150313030231 build.

Pasting the whole stack to make this easier to find.
  CoreFoundation  CoreFoundation@0x18827  
1     CoreFoundation  CoreFoundation@0x28793  
2     CoreFoundation  CoreFoundation@0x18a0d  
3     VideoToolbox    VideoToolbox@0x9183     
4       @0x1197eabe8    
5   XUL     mozilla::AppleVDADecoder::Shutdown()    dom/media/fmp4/apple/AppleVDADecoder.cpp
6   XUL     mozilla::SharedDecoderManager::Shutdown()   dom/media/fmp4/SharedDecoderManager.cpp
7   XUL     mozilla::MediaSourceReader::ContinueShutdown()  dom/media/mediasource/MediaSourceReader.cpp
8   XUL     mozilla::MediaPromise<bool, bool, false>::ThenValue<mozilla::MediaTaskQueue, mozilla::MediaSourceReader, void (mozilla::MediaSourceReader::*)(), void (mozilla::MediaSourceReader::*)()>::DoResolve(bool)   obj-firefox/x86_64/dist/include/MediaPromise.h
9   XUL     mozilla::MediaPromise<bool, bool, false>::ThenValueBase::ResolveRunnable::Run()     dom/media/MediaPromise.h
10  XUL     mozilla::MediaTaskQueue::Runner::Run()  dom/media/MediaTaskQueue.cpp
11  XUL     nsThreadPool::Run()     xpcom/threads/nsThreadPool.cpp
12  XUL     _ZThn8_N12nsThreadPool3RunEv    obj-firefox/x86_64/xpcom/threads/Unified_cpp_xpcom_threads0.cpp
13  XUL     nsThread::ProcessNextEvent(bool, bool*)     xpcom/threads/nsThread.cpp
14  XUL     NS_ProcessNextEvent(nsIThread*, bool)   xpcom/glue/nsThreadUtils.cpp
15  XUL     mozilla::ipc::MessagePumpForNonMainThreads::Run(base::MessagePump::Delegate*)   ipc/glue/MessagePump.cpp
16  XUL     MessageLoop::Run()  ipc/chromium/src/base/message_loop.cc
17  XUL     nsThread::ThreadFunc(void*)     xpcom/threads/nsThread.cpp
18  libnss3.dylib   _pt_root    nsprpub/pr/src/pthreads/ptthread.c
19    libsystem_pthread.dylib     libsystem_pthread.dylib@0x32fb  
20    libsystem_pthread.dylib     libsystem_pthread.dylib@0x3278  
21    libsystem_pthread.dylib     libsystem_pthread.dylib@0x14b0  
22  libnss3.dylib   libnss3.dylib@0x208cdf

Comment 4

[Jean-Yves Avenard [:jya]]

Attachment: Ensure successive calls to Shutdown() won't potentially crash

Can't think of another reason on why it would crash now that it's managed by SharedDecoderManager but didn't before. So likely multiple calls to Shutdown() (and so likely similar to 1142527: we can't have an unbalanced call to Shutdown() and ReleaseMediaResources())

Comment 5

[Jean-Yves Avenard [:jya]]

(In reply to Nickolay_Ponomarev from comment #3)
> I got a similar crash at a different address [@ CoreFoundation@0x19217],
> also after watching youtube:
> https://crash-stats.mozilla.com/report/index/357c1fe4-e99e-4693-ac93-
> 58fb32150315 
> 
> The most popular address seems to be [@ CoreFoundation@0x188b7] with >100
> crashes:
> https://crash-stats.mozilla.com/search/?topmost_filenames=AppleVDADecoder.
> cpp&_facets=signature&_columns=date&_columns=signature&_columns=product&_colu
> mns=version&_columns=build_id&_columns=platform#facet-signature
> 
> Seems new in 20150313030231 build.
> 
> Pasting the whole stack to make this easier to find.
>   CoreFoundation  CoreFoundation@0x18827  
> 1     CoreFoundation  CoreFoundation@0x28793  
> 2     CoreFoundation  CoreFoundation@0x18a0d  
> 3     VideoToolbox    VideoToolbox@0x9183     
> 4       @0x1197eabe8    
> 5   XUL     mozilla::AppleVDADecoder::Shutdown()   
> dom/media/fmp4/apple/AppleVDADecoder.cpp
> 6   XUL     mozilla::SharedDecoderManager::Shutdown()  
> dom/media/fmp4/SharedDecoderManager.cpp

hold a sec...

How can the AppleVDADecoder not be wrapped in an AVCCMediaDataDecoder ???

Comment 6

[Jean-Yves Avenard [:jya]]

The only way this can happen is if the mimetype isn't video/avc

cpearce, what mimetype can we get for h264 content ?

Comment 7

[Jean-Yves Avenard [:jya]]

Attachment: Part2. Also wrap mimetype video/mp4 in avcc wrapper

Also wrap video/mp4 mimetype. I'm no fan of using the mimetype, according to RFC: http://www.rfc-editor.org/rfc/rfc4337.txt anything with video will be video/mp4 as MP4 can use more than h264 as video codec it's not restrictive enough.

Comment 8

[Jean-Yves Avenard [:jya]]

Attachment: Part3. FFmpeg decoder can play video/mp4 mimetype

Same deal with FFmpeg

Comment 9

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

Comment on attachment 8577690 [details] [diff] [review]
Part3. FFmpeg decoder can play video/mp4 mimetype

Review of attachment 8577690 [details] [diff] [review]:
-----------------------------------------------------------------

Lovely. :/

Comment 10

[Justin Dolske [:Dolske]]

FWIW, I'm hitting this extremely frequently, same stack, after closing a tab playing YouTube. No Flash installed.

Comment 11

[Ed Lee :Mardak]

https://crash-stats.mozilla.com/search/?signature=~CoreFoundation&version=39.0a1

70% of reports are with CoreFoundation@0x188b7

0 	CoreFoundation 	CoreFoundation@0x188b7 	
1 	CoreFoundation 	CoreFoundation@0x28823 	
2 	CoreFoundation 	CoreFoundation@0x18a9d 	
3 	VideoToolbox 	VideoToolbox@0x9033 	
4 		@0x10c58abe8 	
5 	XUL 	mozilla::AppleVDADecoder::Shutdown() 	dom/media/fmp4/apple/AppleVDADecoder.cpp
6 	XUL 	mozilla::SharedDecoderManager::Shutdown() 	dom/media/fmp4/SharedDecoderManager.cpp
7 	XUL 	mozilla::MediaSourceReader::ContinueShutdown() 	dom/media/mediasource/MediaSourceReader.cpp

Comment 12

[Justin Dolske [:Dolske]]

(Attempting to autolinkify from Socorro...)

Comment 13

[Chris Pearce [:cpearce (Not reading bugmail)]]

(In reply to Jean-Yves Avenard [:jya] from comment #6)
> The only way this can happen is if the mimetype isn't video/avc
> 
> cpearce, what mimetype can we get for h264 content ?

WMFDecoderModule::SupportsVideoMimeType() returns true for "video/mp4" or "video/avc", so maybe you need that for the Apple PDM?

Comment 14

[Jean-Yves Avenard [:jya]]

remote:   https://hg.mozilla.org/integration/mozilla-inbound/rev/6c7e269c8707
remote:   https://hg.mozilla.org/integration/mozilla-inbound/rev/1c8639568d8b
remote:   https://hg.mozilla.org/integration/mozilla-inbound/rev/ce5ecfd4fe1c

Comment 16

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/6c7e269c8707
https://hg.mozilla.org/mozilla-central/rev/1c8639568d8b
https://hg.mozilla.org/mozilla-central/rev/ce5ecfd4fe1c

Comment 17

[(not currently active) Ted Mielczarek]

I'm assuming the crash I just hit [@ mozilla::AVCCMediaDataDecoder::Shutdown() ] is the same root cause:
https://crash-stats.mozilla.com/report/index/e838f7a2-5777-4d17-a2d4-972122150318

Comment 18

[David Bolter [:davidb] (NeedInfo me for attention)]

(In reply to Ted Mielczarek [:ted.mielczarek] from comment #17)
> I'm assuming the crash I just hit [@
> mozilla::AVCCMediaDataDecoder::Shutdown() ] is the same root cause:
> https://crash-stats.mozilla.com/report/index/e838f7a2-5777-4d17-a2d4-
> 972122150318

Yeah that was the assumption in closing as DUPE bug 1143687...

Comment 19

[Jean-Yves Avenard [:jya]]

(In reply to Ted Mielczarek [:ted.mielczarek] from comment #17)
> I'm assuming the crash I just hit [@
> mozilla::AVCCMediaDataDecoder::Shutdown() ] is the same root cause:
> https://crash-stats.mozilla.com/report/index/e838f7a2-5777-4d17-a2d4-
> 972122150318

no.. this one is a different root cause (and different backtrace) bug 1143687 had the same backtrace as this bug, so they were dupes. 
The backtrace doesn't make much sense however. It would appear that mPDM is no longer pointing to valid memory (but it's a refptr)

Can you please open another bug with details on how it happened?

Will have to monitor that one.

thank you

Comment 20

[(not currently active) Ted Mielczarek]

Filed bug 1145101.