966326 - ###!!! ASSERTION: Uh, IsInModalState() called w/o a reachable top window?

Status: NEW

(Core :: DOM: Core & HTML, defect, P5)

Description

[06needhamt]

Attachment: Screenshot of error message

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:29.0) Gecko/20100101 Firefox/29.0 (Beta/Release)
Build ID: 20140131161412

Steps to reproduce:

Go on the WebGL globe at http://www.chromeexperiments.com/detail/webgl-globe/

Move the globe around a bit 

Close the browser by clicking on the 'X' button at the top of the window


Actual results:

###!!! ASSERTION: Uh, IsInModalState() called w/o a reachable top window?


Expected results:

The browser should of exited sucessfully

Comment 1

[06needhamt]

added correct url

http://workshop.chromeexperiments.com/globe-search/

Comment 2

[Paul Silaghi, QA [:pauly]]

Can't repro on Nightly 2014-02-03-mozilla-central-debug, win 7 x64

Comment 3

[06needhamt]

Still occurs for me with same str

Comment 4

[Paul Silaghi, QA [:pauly]]

I can't reproduce this having only 1 tab with the above URL opened.
But I can reproduce the assertion by just closing FF with 2 empty tabs opened.
Could you please check again?

Comment 5

[06needhamt]

It happens to me on both of those occasions that you mentioned hope this helps

Comment 6

[:Irving Reid (No longer working on Firefox)]

Attachment: Additional NSPR logging in nsGlobalWindow

Bobby, is there any chance this is related to https://hg.mozilla.org/mozilla-central/rev/72bc1aebb5d0?

I'm seeing the same assertion failure but with completely different steps to reproduce; I've figured out (roughly) why I'm seeing it, so I wanted to record it here in case it helps...

I was trying to track down the cause of bug 920978, and running that single test on my debug trunk build with the attached additional NSPR logging showed the following sequence:

Test opens about:addons tab
(event loop spins)
Test opens modal doorhanger on about:addons tab
(event loop spins, lots of test activity)
Test cancels modal doorhanger
...nsGlobalWindow.LeaveModalState() creates an nsPendingTimeoutRunner and schedules it on the event loop
Test closes about:addons tab
(event loop spins)
nsPendingTimeoutRunner.Run() is invoked, and it tries to access the now closed about:addons tab


Modifying my test to let the event loop spin once between cancelling the modal doorhanger and closing the tab makes the assertion go away.


Here's a slightly edited log of my test output with the assertion failure:
... (close modal dialog here)
 0:31.64 2114982272[10065b070]: LeaveModalState 1272d2c00 (topwin 1272d2c00) depth 0
 0:31.64 2114982272[10065b070]: nsPendingTimeoutRunner 1130fb840 created for window 1272d2c00
... (next lines are where we close about:addons)
 0:31.68 TEST-PASS | chrome://mochitests/content/browser/toolkit/mozapps/extensions/test/browser/browser_dragdrop.js | Should have an add-ons manager window to close
 0:31.68 TEST-PASS | chrome://mochitests/content/browser/toolkit/mozapps/extensions/test/browser/browser_dragdrop.js | Should be closing window with correct URI
...
 0:31.87 2114982272[10065b070]: nsPendingTimeoutRunner 1130fb840 running
 0:31.87 [47461] ###!!! ASSERTION: Uh, IsInModalState() called w/o a reachable top window?: 'Error', file /Users/ireid/tbird/mozilla-central-PB/dom/base/nsGlobalWindow.cpp, line 8417
 0:31.89 nsGlobalWindow::IsInModalState()+0x0000004C [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x02501C3C]
 0:31.89 nsGlobalWindow::RunTimeout(nsTimeout*)+0x00000026 [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x025029B6]
 0:31.90 nsGlobalWindow::RunPendingTimeoutsRecursive(nsGlobalWindow*, nsGlobalWindow*)+0x00000061 [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x025026E1]
 0:31.90 nsPendingTimeoutRunner::Run()+0x00000071 [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x025432E1]
 0:31.90 nsThread::ProcessNextEvent(bool, bool*)+0x00000619 [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x00134139]
 0:31.91 NS_ProcessPendingEvents(nsIThread*, unsigned int)+0x000000AB [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x000357BB]
 0:31.91 nsBaseAppShell::NativeEventCallback()+0x000000C9 [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x02280459]
 0:31.91 nsAppShell::ProcessGeckoEvents(void*)+0x000001AC [/Users/ireid/tbird/obj/mozilla-central-PB/dist/NightlyDebug.app/Contents/MacOS/XUL +0x0220C17C]

Comment 7

[Bobby Holley (:bholley)]

(In reply to :Irving Reid from comment #6)
> Created attachment 8385870 [details] [diff] [review]
> Additional NSPR logging in nsGlobalWindow
> 
> Bobby, is there any chance this is related to
> https://hg.mozilla.org/mozilla-central/rev/72bc1aebb5d0?

I doubt it.

I think this assertion is just bogus. GetScriptableTop can return null (via GetParent returning null once the window has been disconnected from the docshell). We should just remove it.

Comment 8

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.