538767 - Crash in nsSocketTransport::InitiateSocket

Status: RESOLVED WONTFIX

(Core :: Networking, defect, P3)

Description

[chris hofmann]

in very early 3.6 RC1 data the #15 crash doesn't appear to have been noticed before.   Could be a single or few people influencing the ranking.

http://crash-stats.mozilla.com/report/index/d553cf86-ad54-4f95-9f13-3a5b52100108

Frame  	Module  	Signature [Expand]  	Source
0 		@0x1f5006b 	
1 	nspr4.dll 	_PR_MD_CONNECT 	nsprpub/pr/src/md/windows/w95sock.c:261
2 	nspr4.dll 	SocketConnect 	nsprpub/pr/src/io/prsocket.c:286
3 	xul.dll 	nsSocketTransport::InitiateSocket 	netwerk/base/src/nsSocketTransport2.cpp:1177
4 	xul.dll 	nsSocketTransport::OnSocketEvent 	netwerk/base/src/nsSocketTransport2.cpp:1447
5 	xul.dll 	nsSocketEvent::Run 	netwerk/base/src/nsSocketTransport2.cpp:98
6 	xul.dll 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:527
7 	xul.dll 	NS_ProcessPendingEvents_P 	obj-firefox/xpcom/build/nsThreadUtils.cpp:200
8 	xul.dll 	nsSocketTransportService::Run 	netwerk/base/src/nsSocketTransportService2.cpp:575
9 	xul.dll 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:527
10 	xul.dll 	NS_ProcessNextEvent_P 	obj-firefox/xpcom/build/nsThreadUtils.cpp:250
11 	xul.dll 	nsThread::ThreadFunc 	xpcom/threads/nsThread.cpp:254
12 	nspr4.dll 	_PR_NativeRunThread 	nsprpub/pr/src/threads/combined/pruthr.c:426
13 	nspr4.dll 	pr_root 	nsprpub/pr/src/md/windows/w95thred.c:122
14 	mozcrt19.dll 	_callthreadstartex 	obj-firefox/memory/jemalloc/crtsrc/threadex.c:348
15 	mozcrt19.dll 	_threadstartex 	obj-firefox/memory/jemalloc/crtsrc/threadex.c:326
16 	kernel32.dll 	BaseThreadStart

more reports at
http://crash-stats.mozilla.com/report/list?range_value=2&range_unit=weeks&signature=_PR_MD_CONNECT&version=Firefox%3A3.6

not much to go on to start the analysis.

looks like a crash very near startup.

the only recent source file changes on the stack are related to 
Jeremias Bosch - Bug 495674: Internet connection should be initiated if needed [r=biesi]  

http://hg.mozilla.org/releases/mozilla-1.9.2/log/672df9e9c441/netwerk/base/src/nsSocketTransport2.cpp

but not sure these changes are related.

Comment 1

[chris hofmann]

32 reports for this signature on the 1st day of 3.6 rc1 and we really don't see it elsewhere in other releases, by comparison.

checking --- 20100108-crashdata.csv _PR_MD_CONNECT
release total-crashes
              _PR_MD_CONNECT crashes
                         pct.
all     212980  124     0.000582214
3.0.15  2056            0
3.0.16  8335            0
3.5.5   5677    7       0.00123305
3.5.6   26670   21      0.000787402
3.6     1758    32      0.0182025
3.6b5   22218   9       0.000405077
3.6b4   1646    1       0.000607533
3.6b3   627     1       0.0015949
3.6b2   731             0
3.6b1   2047            0

Comment 2

[chris hofmann]

looks like it might be at least 3 people or systems hitting the crash on rc1

#crashes sig  firefox_vers  os
   1 _PR_MD_CONNECT 3.6 Windows NT 5.1.2600 Dodatek Service Pack 3
  10 _PR_MD_CONNECT 3.6 Windows NT 5.1.2600 Service Pack 2
  10 _PR_MD_CONNECT 3.6 Windows NT 5.1.2600 Service Pack 3

Comment 3

[chris hofmann]

I'm guessing this might decline in ranking, but we can keep an eye out over the next few days.

Comment 4

[timeless]

can you see if this is idmmzcc.dll and friends?

it should be an LSP based on where/what, but i haven't really had time to figure out the best way to identify them.

Comment 5

[chris hofmann]

Attachment: idmmzcc.dll version numbers where that .dll is present

idmmzcc.dll is around in some of the the crashes but not always.

Comment 6

[timeless]

:(

http://crash-stats.mozilla.com/report/index/74879e58-9c40-4f21-9ef9-ea6ad2100108

has nothing that i can spot as foreign, but we're definitely talking about either jumping into an unloaded library (lsp?) or somehow giving windows a structure so bad that the winsock code returns/jumps into a bogus address.

I wonder how much harm would come from us refusing to unload libraries :).

Comment 7

[chris hofmann]

looks like this problem also goes by the signature [@ connect ] in firefox 3.6.3 and its currently ranked #48 in early 3.6.3 data.  the combined signature counts would rank it higher.  the user comments follow a similar pattern of junk entries so a few people might be influencing the early ranking.

Comment 8

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 9

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 10

[Sylvestre Ledru [:Sylvestre]]

Closing because no crashes reported for 12 weeks.