Biju Reporter
	Description • 14 years ago

Mozilla/5.0 (Windows NT 6.0; rv:2.0b9pre) Gecko/20101219 Firefox/4.0b9pre
This issue is similar to bug 613800. But work around for this issue is not resolving bug 613800. Hence creating a new bug...


New tab-modal alerts can make Firefox unable to quit.
But this is not issue if we do an additional step, see step 6 in StR. 

Steps:-
Phase I
1. Open 2 tabs
2. in one tab navigate to attachment alert_issues.html
3. Click "Start Test"
4. wait 5 seconds
5. when alerts popup dismiss first 2 of them
6. now navigate to "about:blank" on same tab
7. Close the same tab by clicking "x" on the Tab
8. Exit Firefox
9. Restart Firefox
(Firefox restarts without an issue)

Phase II
10. Again open 2 tabs
11. in one tab navigate to attachment alert_issues.html
12. Click "Start Test"
13. wait 5 seconds
14. when alerts popup dismiss first 2 of them
15. Close the same tab by clicking "x" on the Tab
16. Exit Firefox
17. Restart Firefox again

Result:-
on Phase II, Firefox is unable to restart, user get message

      Firefox is already running, but is not responding. 
      To open a new window, you must first close the existing Firefox process, 
      or restart your system.

So cant we do a programmatic action similar to step 6 when user close a tab
to avoid similar issues?

	:Gavin Sharp [email: gavin@gavinsharp.com]
	Comment 1 • 14 years ago

This sounds like something we need to investigate - it seems like we're probably not exiting the dialog event loops for some reason.

	Alan Thomas
	Comment 2 • 14 years ago

Simpler STR can work for me (on Windows 7 32-bit Gecko/20101221).

1. open the attached testcase, with another tab open (any page)
2. click "Start Tests", and click OK for first two alerts
3. close testcase tab (by clicking X button)
4. close Firefox
 * firefox process stays resident in memory
5. attempt to re-launch Firefox
 * not possible

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 3 • 14 years ago

I can't reproduce this on tip.

	Biju Reporter
	Comment 4 • 14 years ago

(In reply to comment #3)
> I can't reproduce this on tip.

I am able to always reproduce this in Win XP and Vista. 
On Ubuntu I see firefox-bin in "futex_wait_queue_me" waiting channel after following the phase II STR

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 5 • 14 years ago

Can you post the full stack?

	Nochum Sossonko [:Natch]
	Comment 6 • 14 years ago

I can repro. I'm not sure what to break on to get a useful stack though, if you can point me to a useful Shutdown function to break on, I'd appreciate it.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 7 • 14 years ago

Just attach and dump the stack once we're hung?

	Nochum Sossonko [:Natch]
	Comment 8 • 14 years ago

(In reply to comment #7)
> Just attach and dump the stack once we're hung?

It's a useless windows stack.

0:033> kp
ChildEBP RetAddr  
08cefc2c 7776c9a0 ntdll!DbgBreakPoint
08cefc5c 76e7d0e9 ntdll!DbgUiRemoteBreakin+0x3c
08cefc68 777219bb kernel32!BaseThreadInitThunk+0xe
08cefca8 7772198e ntdll!__RtlUserThreadStart+0x23
08cefcc0 00000000 ntdll!_RtlUserThreadStart+0x1b

	Nochum Sossonko [:Natch]
	Comment 9 • 14 years ago

Didn't get the other threads, here's a full stack trace.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 10 • 14 years ago

Thanks!

So we're in a nested event loop with JS on the stack.  Sounds like comment 3 is correct.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 11 • 14 years ago

er, comment 1 of course

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 12 • 14 years ago

I can reproduce now.  Investigating.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 13 • 14 years ago

When closing the tab, I get 

************************************************************
* Call to xpconnect wrapped JSObject produced this error:  *
[Exception... "'JavaScript component does not have a method named: "handleEvent"' when calling metho
d: [nsIDOMEventListener::handleEvent]"  nsresult: "0x80570030 (NS_ERROR_XPC_JSOBJECT_HAS_NO_FUNCTION
_NAMED)"  location: "native frame :: <unknown filename> :: <TOP_LEVEL> :: line 0"  data: no]
************************************************************

with the following stack

>	xul.dll!nsXPCWrappedJSClass::CheckForException(XPCCallContext & ccx, const char * aPropertyName, const char * anInterfaceName, int aForceReport)  Line 1174	C++
 	xul.dll!nsXPCWrappedJSClass::CallMethod(nsXPCWrappedJS * wrapper, unsigned short methodIndex, const XPTMethodDescriptor * info, nsXPTCMiniVariant * nativeParams)  Line 1735 + 0x21 bytes	C++
 	xul.dll!nsXPCWrappedJS::CallMethod(unsigned short methodIndex, const XPTMethodDescriptor * info, nsXPTCMiniVariant * params)  Line 588 + 0x13 bytes	C++
 	xul.dll!PrepareAndDispatch(nsXPTCStubBase * self, unsigned int methodIndex, unsigned int * args, unsigned int * stackBytesToPop)  Line 114 + 0x1a bytes	C++
 	xul.dll!SharedStub()  Line 142	C++
 	xul.dll!nsEventListenerManager::HandleEventSubType(nsListenerStruct * aListenerStruct, nsIDOMEventListener * aListener, nsIDOMEvent * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aPhaseFlags, nsCxPusher * aPusher)  Line 1114 + 0x9 bytes	C++
 	0045a958()	
 	xul.dll!nsXPConnect::Push(JSContext * cx)  Line 2511 + 0xb bytes	C++
 	xul.dll!nsCxPusher::DoPush(JSContext * cx)  Line 2995 + 0x9 bytes	C++
 	xul.dll!nsCxPusher::Push(JSContext * cx, int aRequiresScriptContext)  Line 2978 + 0xb bytes	C++
 	xul.dll!nsCxPusher::Push(nsPIDOMEventTarget * aCurrentTarget)  Line 2925 + 0xa bytes	C++
 	xul.dll!nsEventListenerManager::HandleEventInternal(nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aFlags, nsEventStatus * aEventStatus, nsCxPusher * aPusher)  Line 1211 + 0x1a bytes	C++
 	xul.dll!nsEventListenerManager::HandleEvent(nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aFlags, nsEventStatus * aEventStatus, nsCxPusher * aPusher)  Line 146 + 0x18 bytes	C++
 	xul.dll!nsEventTargetChainItem::HandleEvent(nsEventChainPostVisitor & aVisitor, unsigned int aFlags, int aMayHaveNewListenerManagers, nsCxPusher * aPusher)  Line 213	C++
 	xul.dll!nsEventTargetChainItem::HandleEventTargetChain(nsEventChainPostVisitor & aVisitor, unsigned int aFlags, nsDispatchingCallback * aCallback, int aMayHaveNewListenerManagers, nsCxPusher * aPusher)  Line 343	C++
 	xul.dll!nsEventDispatcher::Dispatch(nsISupports * aTarget, nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * aDOMEvent, nsEventStatus * aEventStatus, nsDispatchingCallback * aCallback, nsCOMArray<nsPIDOMEventTarget> * aTargets)  Line 628 + 0x16 bytes	C++
 	xul.dll!nsEventDispatcher::DispatchDOMEvent(nsISupports * aTarget, nsEvent * aEvent, nsIDOMEvent * aDOMEvent, nsPresContext * aPresContext, nsEventStatus * aEventStatus)  Line 691 + 0x14 bytes	C++
 	xul.dll!nsDocument::DispatchPageTransition(nsPIDOMEventTarget * aDispatchTarget, const nsAString_internal & aType, int aPersisted)  Line 7324 + 0xe bytes	C++
 	xul.dll!nsDocument::OnPageHide(int aPersisted, nsIDOMEventTarget * aDispatchStartTarget)  Line 7435 + 0x2a bytes	C++
 	xul.dll!DocumentViewerImpl::PageHide(int aIsUnload)  Line 1269	C++
 	xul.dll!nsDocShell::FirePageHideNotification(int aIsUnload)  Line 1521	C++
 	xul.dll!nsDocShell::Destroy()  Line 4488	C++
 	xul.dll!nsFrameLoader::Finalize()  Line 424	C++
 	xul.dll!nsDocument::MaybeInitializeFinalizeFrameLoaders()  Line 5471 + 0x17 bytes	C++
 	xul.dll!nsDocument::EndUpdate(unsigned int aUpdateType)  Line 3978 + 0x7 bytes	C++
 	xul.dll!nsXULDocument::EndUpdate(unsigned int aUpdateType)  Line 3320	C++
 	xul.dll!mozAutoDocUpdate::~mozAutoDocUpdate()  Line 66 + 0x12 bytes	C++
 	xul.dll!nsINode::doRemoveChildAt(unsigned int aIndex, int aNotify, nsIContent * aKid, nsAttrAndChildArray & aChildArray, int aMutationEvent)  Line 3684 + 0x2a bytes	C++
 	xul.dll!nsGenericElement::RemoveChildAt(unsigned int aIndex, int aNotify, int aMutationEvent)  Line 3624 + 0x14 bytes	C++
 	xul.dll!nsXULElement::RemoveChildAt(unsigned int aIndex, int aNotify, int aMutationEvent)  Line 1010	C++
 	xul.dll!nsINode::RemoveChild(nsINode * aOldChild)  Line 485 + 0xc bytes	C++
 	xul.dll!nsIDOMNode_RemoveChild(JSContext * cx, unsigned int argc, jsval_layout * vp)  Line 6094	C++
 	mozjs.dll!js::CallJSNative(JSContext * cx, int (JSContext *, unsigned int, js::Value *)* native, unsigned int argc, js::Value * vp)  Line 684 + 0x7 bytes	C++
 	mozjs.dll!js::Interpret(JSContext * cx, JSStackFrame * entryFrame, unsigned int inlineCallCount, JSInterpMode interpMode)  Line 4744 + 0x14 bytes	C++
 	mozjs.dll!js::RunScript(JSContext * cx, JSScript * script, JSStackFrame * fp)  Line 657 + 0xb bytes	C++
 	mozjs.dll!js::Invoke(JSContext * cx, const js::CallArgs & argsRef, unsigned int flags)  Line 737 + 0x8 bytes	C++
 	mozjs.dll!js::ExternalInvoke(JSContext * cx, const js::Value & thisv, const js::Value & fval, unsigned int argc, js::Value * argv, js::Value * rval)  Line 858 + 0xd bytes	C++
 	mozjs.dll!JS_CallFunctionValue(JSContext * cx, JSObject * obj, jsval_layout fval, unsigned int argc, jsval_layout * argv, jsval_layout * rval)  Line 5029	C++
 	xul.dll!nsJSContext::CallEventHandler(nsISupports * aTarget, void * aScope, void * aHandler, nsIArray * aargv, nsIVariant * * arv)  Line 2177 + 0x1c bytes	C++
 	xul.dll!nsJSEventListener::HandleEvent(nsIDOMEvent * aEvent)  Line 228 + 0x34 bytes	C++
 	xul.dll!nsXBLPrototypeHandler::ExecuteHandler(nsPIDOMEventTarget * aTarget, nsIDOMEvent * aEvent)  Line 333	C++
 	xul.dll!nsXBLEventHandler::HandleEvent(nsIDOMEvent * aEvent)  Line 90	C++
 	xul.dll!nsEventListenerManager::HandleEventSubType(nsListenerStruct * aListenerStruct, nsIDOMEventListener * aListener, nsIDOMEvent * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aPhaseFlags, nsCxPusher * aPusher)  Line 1114 + 0x9 bytes	C++
 	xul.dll!nsEventListenerManager::HandleEventInternal(nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aFlags, nsEventStatus * aEventStatus, nsCxPusher * aPusher)  Line 1211 + 0x1a bytes	C++
 	xul.dll!nsEventListenerManager::HandleEvent(nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * * aDOMEvent, nsPIDOMEventTarget * aCurrentTarget, unsigned int aFlags, nsEventStatus * aEventStatus, nsCxPusher * aPusher)  Line 146 + 0x18 bytes	C++
 	xul.dll!nsEventTargetChainItem::HandleEvent(nsEventChainPostVisitor & aVisitor, unsigned int aFlags, int aMayHaveNewListenerManagers, nsCxPusher * aPusher)  Line 213	C++
 	xul.dll!nsEventTargetChainItem::HandleEventTargetChain(nsEventChainPostVisitor & aVisitor, unsigned int aFlags, nsDispatchingCallback * aCallback, int aMayHaveNewListenerManagers, nsCxPusher * aPusher)  Line 366	C++
 	xul.dll!nsEventDispatcher::Dispatch(nsISupports * aTarget, nsPresContext * aPresContext, nsEvent * aEvent, nsIDOMEvent * aDOMEvent, nsEventStatus * aEventStatus, nsDispatchingCallback * aCallback, nsCOMArray<nsPIDOMEventTarget> * aTargets)  Line 628 + 0x16 bytes	C++
 	xul.dll!nsTransitionManager::WillRefresh(mozilla::TimeStamp aTime)  Line 1010 + 0x12 bytes	C++
 	xul.dll!nsRefreshDriver::Notify(nsITimer * __formal)  Line 256	C++
 	xul.dll!nsTimerImpl::Fire()  Line 441	C++
 	xul.dll!nsTimerEvent::Run()  Line 519	C++
 	xul.dll!nsThread::ProcessNextEvent(int mayWait, int * result)  Line 626 + 0xe bytes	C++
 	xul.dll!NS_InvokeByIndex_P(nsISupports * that, unsigned int methodIndex, unsigned int paramCount, nsXPTCVariant * params)  Line 103	C++
 	xul.dll!CallMethodHelper::Invoke()  Line 3064 + 0x11 bytes	C++
 	xul.dll!CallMethodHelper::Call()  Line 2328	C++
 	xul.dll!XPCWrappedNative::CallMethod(XPCCallContext & ccx, XPCWrappedNative::CallMode mode)  Line 2290 + 0x10 bytes	C++
 	xul.dll!XPC_WN_CallMethod(JSContext * cx, unsigned int argc, jsval_layout * vp)  Line 1588 + 0xa bytes	C++
 	051fff68()	
 	mozjs.dll!js::ExecuteTrace(JSContext * cx, nanojit::Fragment * f, js::TracerState & state)  Line 6420	C++
 	mozjs.dll!js::ExecuteTree(JSContext * cx, js::TreeFragment * f, unsigned int & inlineCallCount, js::VMSideExit * * innermostNestedGuardp, js::VMSideExit * * lrp)  Line 6520 + 0xf bytes	C++
 	mozjs.dll!js::RecordLoopEdge(JSContext * cx, unsigned int & inlineCallCount)  Line 7069 + 0x28 bytes	C++
 	mozjs.dll!js::Interpret(JSContext * cx, JSStackFrame * entryFrame, unsigned int inlineCallCount, JSInterpMode interpMode)  Line 2907 + 0x142 bytes	C++
 	mozjs.dll!js::RunScript(JSContext * cx, JSScript * script, JSStackFrame * fp)  Line 657 + 0xb bytes	C++
 	mozjs.dll!js::Invoke(JSContext * cx, const js::CallArgs & argsRef, unsigned int flags)  Line 737 + 0x8 bytes	C++
 	mozjs.dll!js::mjit::stubs::SlowCall(js::VMFrame & f, unsigned int argc)  Line 198 + 0x24 bytes	C++
 	mozjs.dll!SlowCallFromIC(js::VMFrame & f, js::mjit::ic::CallICInfo * ic)  Line 426	C++
 	mozjs.dll!js::mjit::EnterMethodJIT(JSContext * cx, JSStackFrame * fp, void * code, js::Value * stackLimit)  Line 745 + 0x24 bytes	C++
 	mozjs.dll!js::mjit::JaegerShot(JSContext * cx)  Line 787 + 0x1b bytes	C++
 	mozjs.dll!js::RunScript(JSContext * cx, JSScript * script, JSStackFrame * fp)  Line 654 + 0x6 bytes	C++
 	mozjs.dll!js::Invoke(JSContext * cx, const js::CallArgs & argsRef, unsigned int flags)  Line 737 + 0x8 bytes	C++
 	mozjs.dll!js::ExternalInvoke(JSContext * cx, const js::Value & thisv, const js::Value & fval, unsigned int argc, js::Value * argv, js::Value * rval)  Line 858 + 0xd bytes	C++
 	mozjs.dll!JS_CallFunctionValue(JSContext * cx, JSObject * obj, jsval_layout fval, unsigned int argc, jsval_layout * argv, jsval_layout * rval)  Line 5029	C++
 	xul.dll!nsXPCWrappedJSClass::CallMethod(nsXPCWrappedJS * wrapper, unsigned short methodIndex, const XPTMethodDescriptor * info, nsXPTCMiniVariant * nativeParams)  Line 1699	C++
 	xul.dll!nsXPCWrappedJS::CallMethod(unsigned short methodIndex, const XPTMethodDescriptor * info, nsXPTCMiniVariant * params)  Line 588 + 0x13 bytes	C++
 	xul.dll!PrepareAndDispatch(nsXPTCStubBase * self, unsigned int methodIndex, unsigned int * args, unsigned int * stackBytesToPop)  Line 114 + 0x1a bytes	C++
 	xul.dll!SharedStub()  Line 142	C++
 	xul.dll!nsGlobalWindow::Alert(const nsAString_internal & aString)  Line 4638 + 0x1a bytes	C++
 	006a0048()	
 	xul.dll!nsGlobalWindow::Alert(const nsAString_internal & aString)  Line 4580 + 0xa bytes	C++
 	xul.dll!NS_InvokeByIndex_P(nsISupports * that, unsigned int methodIndex, unsigned int paramCount, nsXPTCVariant * params)  Line 103	C++
 	xul.dll!CallMethodHelper::Invoke()  Line 3064 + 0x11 bytes	C++
 	xul.dll!CallMethodHelper::Call()  Line 2328	C++
 	xul.dll!XPCWrappedNative::CallMethod(XPCCallContext & ccx, XPCWrappedNative::CallMode mode)  Line 2290 + 0x10 bytes	C++
 	xul.dll!XPC_WN_CallMethod(JSContext * cx, unsigned int argc, jsval_layout * vp)  Line 1588 + 0xa bytes	C++
 	02681041()	
 	xul.dll!nsJSContext::CallEventHandler(nsISupports * aTarget, void * aScope, void * aHandler, nsIArray * aargv, nsIVariant * * arv)  Line 2177 + 0x1c bytes	C++

We're firing onPageHide, which is one of the things nsPrompter.js listens for to close the tab-modal prompt and exit the nested event loop.  That event gets lost so we never shut down the prompts.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 14 • 14 years ago

Somehow the <tabmodalprompt> is getting destroyed before pagehide fires.  If I add a call to shutdown the prompt in the destructor this bug goes away.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 16 • 14 years ago

Comment on attachment 505276 [details] [diff] [review]
Make double-extra-sure that we shutdown the prompt.

I can't reproduce this on tip anymore ....

That said, this is the patch I have in my queue for this.  It's purely defensive programming, so it wouldn't hurt to take this.

I don't think we can test this.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 17 • 14 years ago

Biju, can you still reproduce this on tip?

	Justin Dolske [:Dolske]
	Comment 18 • 14 years ago

Comment on attachment 505276 [details] [diff] [review]
Make double-extra-sure that we shutdown the prompt.

Urk. Part of me doesn't want to do this, lest we wallpaper over something that shouldn't be failing anyway (and since XBL destructors apparently don't always fire, we get a hard-to-reproduce hang).

OTOH, we're close to ship, and better to only hang when 2 rare failures happen instead of when 1 rare failure happens.

	Biju Reporter
	Comment 19 • 14 years ago

I can always reproduce this on WinXP and Vista following steps in "Phase II"
Mozilla/5.0 (Windows NT 6.0; rv:2.0b10pre) Gecko/20110121 Firefox/4.0b10pre

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 20 • 14 years ago

OK, I'll check this in next time I push then.

	Biju Reporter
	Comment 21 • 14 years ago

Kyle, can you please check whether Bug 613800 also get resolved with this patch

	Dão Gottwald [:dao]
	Comment 22 • 14 years ago

http://hg.mozilla.org/mozilla-central/rev/42847b167826

	Biju Reporter
	Comment 23 • 14 years ago

Mozilla/5.0 (Windows NT 6.0; rv:2.0b11pre) Gecko/20110126 Firefox/4.0b11pre

Tested on WinXP and Vista it still FAILS on Comment 0 phase II
What should we reopen this bug or create a new one ?


(In reply to comment #21)
STR at Bug 613800 Comment 5 - still fails
STR at Bug 613800 Comment 0 - this was already passing Vista 
                              before landing attachment 505276 [details] [diff] [review]

	:Paolo Amadini
	Comment 24 • 14 years ago

I reproduced this on Windows XP. It looks like the call that sets the flag to
make us exit the event loop was missing.

If it may be of interest, it seems to me that the prompt that is shut down
abnormally in the XBL destructor is not the last prompt that is displayed,
but a new prompt that is triggered by the fact that when the tab is closed
we exit modal state and thus we run pending timeouts, causing a new "alert"
call to be made immediately while we are closing the tab.

	Justin Dolske [:Dolske]
	Comment 25 • 14 years ago

Comment on attachment 508052 [details] [diff] [review]
Addition

Sigh, I suck for missing this.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 26 • 14 years ago

http://hg.mozilla.org/mozilla-central/rev/10c381463850

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 27 • 14 years ago

Backed out due to test failures http://hg.mozilla.org/mozilla-central/rev/10c381463850

	Justin Dolske [:Dolske]
	Comment 28 • 14 years ago

I think I see what might be happening...

During a normal prompt shutdown [ie, a call to shutdownPrompt()], we execute onCloseCallback() and then set isLive=false.

The problem is that onCloseCallback removes the prompt from the DOM (indeed, that's the whole reason we added the callback). If the destructor is run then, isLive is still true, and so it calls abortPrompt(). Bam.

Indeed, the tinderbox failure is:

[SimpleTest/SimpleTest.js, window.onerror] An error occurred - uncaught exception: [Exception... "prompt aborted by user" nsresult: "0x80040111 (NS_ERROR_NOT_AVAILABLE)" location: "JS frame :: resource://gre/components/nsPrompter.js :: openTabPrompt :: line 468" data: no] at :0

Should be able to fix this by setting isLive to false before invoking the callback.

	(no longer active)
	Comment 29 • 14 years ago

Removing the misleading checkin-needed.

	:Paolo Amadini
	Comment 30 • 14 years ago

Sorry, I should have said from the start that this was just a drive-by
suggestion. I guess you I assumed I ran the various test suites because
I attached a patch, but I didn't intend to take the bug, as I'm working
on other things at the moment. I hope this hasn't caused too much trouble.

	Kyle Huey (Exited; not receiving bugmail, old account, do not use) Assignee
	Comment 31 • 14 years ago

It's not a problem Paolo.  Thanks for finding the issue in the original patch.

I'll finish this up later this week.

	Justin Dolske [:Dolske]
	Comment 32 • 14 years ago

I made (and tested ;) the small change I suggested in comment 28 and relanded...

Pushed http://hg.mozilla.org/mozilla-central/rev/76b361ae1c3a

	Biju Reporter
	Comment 33 • 14 years ago

Comment 0 phase II
Mozilla/5.0 (Windows NT 6.0; rv:2.0b12pre) Gecko/20110202 Firefox/4.0b12pre
- PASS

Mozilla/5.0 (Windows NT 6.0; rv:2.0b11pre) Gecko/20110201 Firefox/4.0b11pre
- FAIL