422027 - topcrash on quit [@ jsds_ScriptHookProc]

Status: VERIFIED DUPLICATE of bug 421303

(Core :: XPConnect, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Crashes at jsds_ScriptHookProc have been the #1 topcrash on trunk since they started suddenly in the 2008-02-29 builds and are #5 on beta 4.  (That's a clear start date on all of Windows, Mac, and Linux.

The most likely checkins are probably those in JS and XPConnect that day, since nobody touched jsd:
http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=mozilla%2Fjs%2Fsrc+mozilla%2Fjs%2Fjsd&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2008-02-28&maxdate=2008-02-29+05%3A00&cvsroot=%2Fcvsroot

For the trunk crashes, see:
http://crash-stats.mozilla.com/report/list?range_unit=weeks&version=Firefox%3A3.0b5pre&range_value=2&signature=jsds_ScriptHookProc
for windows and
http://crash-stats.mozilla.com/report/list?range_unit=weeks&version=Firefox%3A3.0b5pre&range_value=2&signature=jsds_ScriptHookProc(JSDContext*%2C+JSDScript*%2C+int%2C+void*)
for mac and linux.

Many comments refer to quitting or restarting for an update.

Typical stacks:
0  	jsds_ScriptHookProc  	 mozilla/js/jsd/jsd_xpc.cpp:720
1 	jsd_DestroyScriptHookProc 	mozilla/js/jsd/jsd_scpt.c:630
2 	js3250.dll@0x4c7aa 	
3 	JS_GC 	mozilla/js/src/jsapi.c:2418
4 	nsXPConnect::Collect() 	mozilla/js/src/xpconnect/src/nsXPConnect.cpp:526
5 	nsCycleCollector::Collect(unsigned int) 	mozilla/xpcom/base/nsCycleCollector.cpp:2192
6 	nsCycleCollector_shutdown() 	mozilla/xpcom/base/nsCycleCollector.cpp:2850
7 	NS_ShutdownXPCOM_P 	mozilla/xpcom/build/nsXPComInit.cpp:783

so bug 414977 seems the most likely cause.

Comment 1

[Mike Shaver (:shaver emeritus)]

Should we just null-check gJsds, or should we try to figure out why the shutdown case can cause us to have the hook installed even when the global has been torn down?

Comment 2

[timeless]

is this still a problem? i'm hoping bug 421303 fixed this (and bug 421303 comment 2 should explain this).

Bug 414977 is almost certainly what started my headaches.

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Yeah, my search didn't find it because it was marked fixed a few hours earlier.

Comment 4

[timeless]

heh, that's what i get for resolving bugs as fixed :)