601097 - crash [@ SocketConnect ]

Status: RESOLVED INCOMPLETE

(Core :: Networking, defect)

Description

[Scoobidiver (away)]

This is a residual crash signature that exists in 3.0, 3.6 and trunk builds.
It is #3 top crasher in b7pre/20100930 build.

Comments say that it is a Flash plugin issue.

Signature	SocketConnect
UUID	24f8eec2-b509-4d29-b608-997532100930
Time 	2010-09-30 20:27:02.184859
Uptime	364
Last Crash	37659 seconds (10.5 hours) before submission
Install Age	48856 seconds (13.6 hours) since version was first installed.
Product	Firefox
Version	4.0b7pre
Build ID	20100930041305
Branch	2.0
OS	Windows NT
OS Version	5.1.2600 Service Pack 3
CPU	x86
CPU Info	GenuineIntel family 6 model 23 stepping 10
Crash Reason	EXCEPTION_ACCESS_VIOLATION_READ
Crash Address	0x43dfffc
App Notes 	AdapterVendorID: 8086, AdapterDeviceID: 29c2

Frame 	Module 	Signature [Expand] 	Source
0 		@0x43f006b 	
1 	nspr4.dll 	SocketConnect 	nsprpub/pr/src/io/prsocket.c:286
2 	xul.dll 	nsSocketTransport::InitiateSocket 	netwerk/base/src/nsSocketTransport2.cpp:1184
3 	xul.dll 	nsSocketTransport::OnSocketEvent 	netwerk/base/src/nsSocketTransport2.cpp:1454
4 	xul.dll 	nsSocketEvent::Run 	netwerk/base/src/nsSocketTransport2.cpp:98
5 	xul.dll 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:547
6 	xul.dll 	NS_ProcessPendingEvents_P 	obj-firefox/xpcom/build/nsThreadUtils.cpp:200
7 	xul.dll 	nsSocketTransportService::Run 	netwerk/base/src/nsSocketTransportService2.cpp:579
8 	xul.dll 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:547
9 	xul.dll 	nsThread::ThreadFunc 	xpcom/threads/nsThread.cpp:263
10 	nspr4.dll 	_PR_NativeRunThread 	nsprpub/pr/src/threads/combined/pruthr.c:426
11 	nspr4.dll 	pr_root 	nsprpub/pr/src/md/windows/w95thred.c:122
12 	mozcrt19.dll 	_callthreadstartex 	obj-firefox/memory/jemalloc/crtsrc/threadex.c:348
13 	mozcrt19.dll 	_threadstartex 	obj-firefox/memory/jemalloc/crtsrc/threadex.c:326
14 	kernel32.dll 	BaseThreadStart 	

More reports at :
http://crash-stats.mozilla.com/report/list?product=Firefox&query_search=signature&query_type=exact&query=&range_value=1&range_unit=weeks&hang_type=any&process_type=any&plugin_field=&plugin_query_type=&plugin_query=&do_query=1&admin=&signature=SocketConnect

Comment 1

[chris hofmann]

not very many comments but of the ones received most are non-english

Correlation to releases

checking --- SocketConnect 20100930-crashdata.csv
found in: 3.6.10 4.0b7pre 4.0b6 3.0.19 3.6.9 3.6.8 3.0.4
release total-crashes
              SocketConnect crashes
                         pct.
all     336929  94      0.000278991
3.6.10  195929  39      0.000199052
4.0b7pre        2278    29      0.0127305
4.0b6   23602   17      0.000720278
3.0.19  9344    4       0.000428082
3.6.9   4111    2       0.0004865
3.6.8   20011   2       9.9945e-05
3.0.4   455     1       0.0021978

os breakdown
SocketConnectTotal 88
Win5.1  0.92
Win6.0  0.07
Win6.1  0.01

looks to be pretty close to start up for the valid times we are getting.

Correlation to startup or time of session
94 total crashes for SocketConnect on 20100930-crashdata.csv
46 startup crashes inside 30 sec.
62 startup crashes inside 3 min.
42 repeated crashes inside 3 min. of last crash
_ null start times
_13 null time since last crash
_13 bogus big or null times

mostly null, missing or google search start page as url confirms proximity to start.

ThreadBaseStart on the stack makes me wonder if this is connected to bug 524944 and these crashes just run a bit longer before Firefox dies

Comment 2

[Marcia Knous [:marcia]]

chofmann: The 3.6.9 comments that are in English mention flash being involved. Looks like one person crashed several times on Facebook and there is a youtube.com crash. In the few reports that I looked at they appeared to all be using 10.1.82.76.

Comment 3

[timeless]

these tend to be buggy or evil LSPs..., someone should try to find an unexpected library...

Comment 4

[Bjarne (:bjarne)]

Similar symptoms as bug #598007 ?

Comment 5

[chris hofmann]

could be, but comments on these reports appear to be asian languages as opposed to russian, and the sharp spike point doesn't appear on sept 10 as it did in in the other signatures -- 
HeapDestroy   bug 597960   
nsDiskCacheDevice::OpenDiskCache bug 595957 
nsDiskCacheMap::Open.nsILocalFile..  bug 598007

date     crashes at
         SocketConnect
20100901 23
20100902 22
20100903 24
20100904 22
20100905 26
20100906 35
20100907 36
20100908 39
20100909 56
20100910 49
20100911 63
20100912 48
20100913 45
20100914 44
20100915 5
20100916 22
20100917 30
20100918 38
20100919 59
20100920 45
20100921 54
20100922 52
20100923 112
20100924 61
20100925 97
20100926 80
20100927 58
20100928 102
20100929 92
20100930 94
20101001 86
20101002 71
20101003 63
20101004 100
20101005 74
20101006 99

Comment 6

[Wayne Mery (:wsmwk)]

Is @0x0 | SocketConnect a different issue?
0		@0x0	
1	nspr4.dll	SocketConnect	nsprpub/pr/src/io/prsocket.c:286
2	nspr4.dll	PR_NewTCPSocketPair	nsprpub/pr/src/io/prsocket.c:1541
3	nspr4.dll	PR_NewPollableEvent	nsprpub/pr/src/io/prpolevt.c:178
4	xul.dll	nsSocketTransportService::Init	netwerk/base/src/nsSocketTransportService2.cpp:399
5	xul.dll	nsSocketTransportServiceConstructor	netwerk/build/nsNetModule.cpp:97
6	xul.dll	mozilla::GenericFactory::CreateInstance	obj-firefox/xpcom/build/GenericFactory.cpp:48
7	xul.dll	nsComponentManagerImpl::CreateInstanceByContractID	xpcom/components/nsComponentManager.cpp:1303
8	xul.dll	nsComponentManagerImpl::GetServiceByContractID	xpcom/components/nsComponentManager.cpp:1664
9	xul.dll	nsCOMPtr_base::assign_from_gs_contractid_with_error	obj-firefox/xpcom/build/nsCOMPtr.cpp:141
10	xul.dll	nsIOService::Init	netwerk/base/src/nsIOService.cpp:201 

only 1 comment of 11 in the past 3 months is English - bp-af22bbfd-0588-4087-ad9a-cda782101026 "Flash caused some sort of crash" - most of the rest are German

Comment 7

[timeless]

oehook.dll	sounds suspicious

since these are basically lsp triggered things, it's a matter of correctly sorting to find lumps

Comment 8

[Robert Kaiser]

This has risen significantly on 4.0* in the last two days, now #131 on yesterday's 4.0* topcrash list. On 3.6* the picture is very much the same, significant rise since 2011-05-01, now #138 on there. Not too high, but rising at the same time on both stable versions probably means something not directly in our code triggering it.

Comment 9

[Marcia Knous [:marcia]]

This is showing up on the explosive report for Firefox 7 today: https://crash-analysis.mozilla.com/rkaiser/2011-10-19/2011-10-19.firefox.7.explosiveness.html.

Most of the comments related to GMail sign in.

Comment 11

[Scoobidiver (away)]

SocketConnect is now in the skiplist. I close it as incomplete.