514481 - crash [@nsThreadManager::GetMainThread(nsIThread**)] [Mac]

Status: RESOLVED WORKSFORME

(Core :: XPCOM, defect)

Description

[Ludovic Hirlimann [:Usul]]

0  	libxpcom_core.dylib  	nsThreadManager::GetMainThread  	 xpcom/threads/nsThreadManager.cpp:262
1 	libxpcom_core.dylib 	NS_GetMainThread_P 	nsThreadUtils.cpp:109
2 	libxpcom_core.dylib 	nsProxyObjectManager::GetProxyForObject 	nsThreadUtils.h:208
3 	libxpcom_core.dylib 	NS_GetProxyForObject 	xpcom/proxy/src/nsProxyObjectManager.cpp:340
4 	libxpcom_core.dylib 	nsThreadPool::ShutdownThread 	xpcom/threads/nsThreadPool.cpp:144
5 	libxpcom_core.dylib 	nsThreadPool::Run 	xpcom/threads/nsThreadPool.cpp:228
6 	libxpcom_core.dylib 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:510
7 	libxpcom_core.dylib 	NS_ProcessNextEvent_P 	nsThreadUtils.cpp:227
8 	libxpcom_core.dylib 	nsThread::ThreadFunc 	xpcom/threads/nsThread.cpp:254
9 	libnspr4.dylib 	_pt_root 	nsprpub/pr/src/pthreads/ptthread.c:228
10 	libSystem.B.dylib 	_pthread_start 	
11 	libSystem.B.dylib 	thread_start

Comment 1

[Benjamin Smedberg]

This signature is oddly common: it indicates that either mMainThread is null (very unlikely!) or that it has been destroyed (perhaps by being over-released somewhere?)

Any common extensions or modules?

Comment 2

[timeless]

http://crash-stats.mozilla.com/report/index/bp-c47d65a9-912a-422f-ad29-f67bf2090901

shows the main thread is still alive

Comment 3

[Benjamin Smedberg]

Yes, that was clear (also there's a null-check for mMainThread immediately above). I strongly suspect over-releasing here, but I suppose some other form of memory corruption is possible.

Comment 4

[Wayne Mery (:wsmwk)]

bsmedberg in comment #1
> Any common extensions or modules?

none mentioned in crash-stats tabs

~3/4 of crashes are nsProxyObjectManager::GetProxyForObject (xref timeless' bug 422231). The rest have nsImapProtocol::SetupWithUrl, nominally similar stack to Bug 558452 (which is not Mac-only)


increaseed to #49 crash in Thunderbird v3.0.4 from #99 in v3.0.3 (which is pretty significant, being apparently a Mac-only crash), for reasons that I can't explain

Comment 5

[Wayne Mery (:wsmwk)]

As a Mac crash, this doesn't exist starting in version 5.
Exist in 3.1 branch bp-73f96678-e78f-47c5-9513-cf4072120116 3.1.17