368289 - ASSERTION: Expected only one call!: '!gCallCount++', file xptiInterfaceInfoManager.cpp, line 229

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[Martin Mokrejs]

User-Agent:       Opera/9.10 (X11; Linux i686; U; en)
Build Identifier: 

I have jsut started seamonkey (browser window) and run ctrl+q to immediately quit. Here is the result.

(gdb) where
#0  Break (aMsg=0xbfbd3050 "###!!! ASSERTION: Expected only one call!: '!gCallCount++', file xptiInterfaceInfoManager.cpp, line 229") at nsDebugImpl.cpp:472
#1  0xb7ea96e9 in NS_DebugBreak_P (aSeverity=1, aStr=0xb7ef2c98 "Expected only one call!", aExpr=0xb7ef2c8a "!gCallCount++", aFile=0xb7ef23ed "xptiInterfaceInfoManager.cpp", aLine=229) at nsDebugImpl.cpp:350
#2  0xb7ebef17 in xptiInterfaceInfoManager::BuildFileSearchPath (aPath=0xbfbd34e0) at xptiInterfaceInfoManager.cpp:229
#3  0xb7ebf7a9 in xptiInterfaceInfoManager::GetInterfaceInfoManagerNoAddRef () at xptiInterfaceInfoManager.cpp:64
#4  0xb7eb8b2d in xptiInterfaceInfoManager::GetInfoMonitor (self=0x0) at xptiprivate.h:904
#5  0xb7eb6a06 in xptiInterfaceInfo::Release (this=0x813e548) at xptiInterfaceInfo.cpp:790
#6  0xb6e0f77a in ~nsCOMPtr (this=0x813a658) at ../../../../dist/include/xpcom/nsCOMPtr.h:583
#7  0xb6e47a92 in ~XPCNativeInterface (this=0x813a658) at xpcprivate.h:1298
#8  0xb6e44d99 in XPCNativeInterface::DestroyInstance (cx=0x8cbd8b0, rt=0x812ef18, inst=0x813a658) at xpcwrappednativeinfo.cpp:504
#9  0xb6e273e1 in NativeInterfaceSweeper (table=0x812f730, hdr=0x87d3928, number=13, arg=0xbfbd370c) at xpcjsruntime.cpp:138
#10 0xb7d1fe1f in JS_DHashTableEnumerate (table=0x812f730, etor=0xb6e2737a <NativeInterfaceSweeper>, arg=0xbfbd370c) at jsdhash.c:674
#11 0xb6e27a35 in IID2NativeInterfaceMap::Enumerate (this=0x812ede8, f=0xb6e2737a <NativeInterfaceSweeper>, arg=0xbfbd370c) at xpcmaps.h:326
#12 0xb6e27040 in XPCJSRuntime::GCCallback (cx=0x8cbd8b0, status=JSGC_FINALIZE_END) at xpcjsruntime.cpp:454
#13 0xb6c91b70 in jsds_GCCallbackProc (cx=0x8cbd8b0, status=JSGC_FINALIZE_END) at jsd_xpc.cpp:519
#14 0xb66d2f43 in DOMGCCallback (cx=0x8cbd8b0, status=JSGC_FINALIZE_END) at nsJSEnvironment.cpp:3251
#15 0xb7d4485a in js_GC (cx=0x8cbd8b0, gckind=GC_NORMAL) at jsgc.c:3162
#16 0xb7d13cd9 in js_DestroyContext (cx=0x8cbd8b0, mode=JSDCM_FORCE_GC) at jscntxt.c:407
#17 0xb7d01c84 in JS_DestroyContext (cx=0x8cbd8b0) at jsapi.c:979
#18 0xb6e2e25e in ~XPCJSContextStack (this=0x80bcea8) at xpcthreadcontext.cpp:61
#19 0xb6e2d68b in XPCPerThreadData::Cleanup (this=0x80bcfd8) at xpcthreadcontext.cpp:460
#20 0xb6e2d81f in ~XPCPerThreadData (this=0x80bcfd8) at xpcthreadcontext.cpp:469
#21 0xb6e2d90b in xpc_ThreadDataDtorCB (ptr=0x80bcfd8) at xpcthreadcontext.cpp:504
#22 0xb7cb7056 in _PR_DestroyThreadPrivate (self=0x8089548) at prtpd.c:265
#23 0xb7cd3726 in _pt_thread_death (arg=0x8089548) at ptthread.c:825
#24 0xb7cd3a9b in _PR_Fini () at ptthread.c:999
#25 0xb7cabe08 in __do_global_dtors_aux () from ./libnspr4.so
#26 0xb7cd9120 in _fini () from ./libnspr4.so
#27 0xb7f52f67 in _dl_fini () from /lib/ld-linux.so.2
#28 0xb747ba74 in exit () from /lib/libc.so.6
#29 0xb746682c in __libc_start_main () from /lib/libc.so.6
#30 0x0804a455 in _start ()
(gdb)

I have configured few hours old cvs HEAD checkout tree with:
./configure --disable-optimize --enable-debug='-g3 -O0 -ggdb' --enable-debug-modules=all --enable-debugger-info-modules --enable-detect-webshell-leaks --disable-svg --enable-image-decoders=all --with-qtdir=/usr/qt/3 --enable-application=suite --disable-freetype2 --enable-jprof --enable-default-toolkit=gtk2 --enable-xft --disable-gssapi

Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Martin Mokrejs]

Attachment: full gdb backtrace

Comment 2

[Andrew Schultz]

Right (firefox too).  This started happening when bug 362768 landed.  After the !gCallCount++ assertion, I also see these, which also started appearing at the same time:

ASSERTION: can't get xpt search path!: 'Error', file mozilla/xpcom/reflect/xptinfo/src/xptiInterfaceInfoManager.cpp, line 67
ASSERTION: null monitor: 'mMonitor', file ../../dist/include/xpcom/nsAutoLock.h, line 284

Comment 3

[Wan-Teh Chang]

Brendan, could you ask the appropriate JavaScript developer
to look at this bug?  It is caused by my NSPR in bug 362768
to always clean up the PRThread structure for the primordial
thread on process exit.  Part of the cleanup is to run the
thread-private data destructors, and we are getting assertion
failures in the destructor xpc_ThreadDataDtorCB registered by
JavaScript.

I'm willing to change my patch to not call thread-private data
destructors when NSPR cleans up the primordial thread, but I'd
like a JavaScript developer to look at these assertion failures
first.

Comment 4

[Jesse Ruderman]

I see this when I quit Firefox, but only when there are large leaks (domwindows?).

Comment 5

[Wan-Teh Chang]

I've changed NSPR's _PR_Fini function to not call thread-private
data destructors when cleaning up the primordial thread, so this
bug has been fixed.

Comment 6

[Wan-Teh Chang]

The fix was committed in bug 383977 comment 10.