Since around September 17th the OS X nightly trunk builds have crashed on launch if the Xmarks add-on is enabled. I emailed Xmarks and this is what they said: > The next step, if you haven't already done so, is to file a bug > w/Mozilla, as it sounds like recent changes to the JIT Javascript engine > are a likely culprit. There's also this, which seems vaguely related: > https://bugzilla.mozilla.org/show_bug.cgi?id=510996 I installed last night's shark build and reproduced the crash - signature is here: http://crash-stats.mozilla.com/report/index/14d38e3d-aa88-4c41-971b-8a2712090922?p=1

Your stacktrace doesn't have any symbols for one reason or another. Just another [@js_Interpret] though. Crashes on Linux too: bp-7a551050-3c7a-46fb-97b7-836eb2090922 If I disable chrome and content JIT it still crashes.

bp-7a551050-3c7a-46fb-97b7-836eb2090922 0 libmozjs.so js_Interpret js/src/jsobj.h:266 1 libmozjs.so js_Execute js/src/jsinterp.cpp:1613 2 libmozjs.so JS_ExecuteScript js/src/jsapi.cpp:4960 3 libxul.so mozJSComponentLoader::GlobalForLocation js/src/xpconnect/loader/mozJSComponentLoader.cpp:1393 4 libxul.so mozJSComponentLoader::LoadModule js/src/xpconnect/loader/mozJSComponentLoader.cpp:695 5 libxul.so nsFactoryEntry::GetFactory xpcom/components/nsComponentManager.cpp:3600 6 libxul.so nsComponentManagerImpl::CreateInstanceByContractID xpcom/components/nsComponentManager.cpp:1679 7 libxul.so nsComponentManagerImpl::GetServiceByContractID xpcom/components/nsComponentManager.cpp:2251 8 libxul.so CallGetService nsComponentManagerUtils.cpp:94 9 libxul.so nsGetServiceByContractIDWithError::operator const nsComponentManagerUtils.cpp:288 10 libxul.so nsCOMPtr_base::assign_from_gs_contractid_with_error nsCOMPtr.cpp:141 11 libxul.so nsAppStartupNotifier::Observe nsCOMPtr.h:1031 12 libxul.so XRE_main toolkit/xre/nsAppRunner.cpp:3246 13 firefox-bin main browser/app/nsBrowserApp.cpp:156 14 libc-2.7.so libc-2.7.so@0x1644f

This seems to be fixed now. I've enabled Xmarks again and I didn't see any startup crashes since a few days (tested on Windows and Mac).

I'm running the the 24 October build of Minefield on Windows 7 Pro x64 and I don't see any crashes with Xmarks enabled.

Yeah, crash looks fixed. Closing as WFM.

I think I'm seeing this crash again on trunk - at least, I can reproduce a crash if xmarks is enabled. Seems like it regressed in last night's build? http://crash-stats.mozilla.com/report/index/80638831-2c68-495f-8514-756052091030?p=1

Xmarks + Minefield build 20091030060356 on Windows 7 Professional x64 works for me. No crashes whatsoever.

I'm also seeing crashes on Mac. http://crash-stats.mozilla.com/report/index/535f18d9-6dba-4007-9405-bcb812091103?p=1 http://crash-stats.mozilla.com/report/index/13e703ac-8e36-467d-b551-bcfe02091103?p=1

@js_Interpret is a topcrash, and XMarks is a really popular add-on; I think we should figure out if this is reproducible and if so, figure out the cause. Neil: can you work with someone in the Toronto office to catch this in a debugger?

I haven't seen this crash since the weekend builds so I'm guessing this is fixed. I'll reopen if it reoccurs.

No bug or patch referenced as the fix.