958133 - Delayed installation of add-ons can cause ASSERTION: Uh, EnterModalState() called w/o a reachable top window?: 'Error', file ../../../dom/base/nsGlobalWindow.cpp, line 8220

Status: RESOLVED INACTIVE

(Toolkit :: Add-ons Manager, defect)

Description

[Henrik Skupin [:whimboo][⌚️UTC+1]]

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:29.0) Gecko/20100101 Firefox/29.0 ID:20140108030203 CSet: cf2d1bd796ea

Today I have seen this assertion in a debug build and also got a crash in an opt build when I wanted to install a locally hosted add-on by a Mozmill test. The machine was connected to the router but the router had no internet connection. So installing that add-on caused a request to a remote site given via the preference 'extensions.getAddons.get.url'. There was no response and Mozmill was shutting down the browser. In the moment when the main window gets closed the assertion and crash appear.

Steps:
1. Interrupt network connection
2. Download the latest mozmill version for your platform from https://mozqa.com/mozmill-env/
3. Unzip and execute mozmill-env/run
4. Execute 'hg clone http://hg.mozilla.org/qa/mozmill-tests'
4. Run 'mozmill -b %path_to_firefox% -t mozmill-tests/firefox/tests/functional/restartTests/testAddons_enableDisableExtension/'

You will see that the notification popup appears for downloading the extension. It reaches 100% but nothing happens. It still sits around. Then 30s later Mozmill will shutdown Firefox. At this time the install dialog will appear and causes the crash and assertion.

The assertion details I get locally from the downloaded debug build don't look that helpful:

JSPrincipals::dump()+0x0009f29d [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x011405ad]
JSPrincipals::dump()+0x00ed6aed [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x01f77dfd]
JSPrincipals::dump()+0x00edb0f6 [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x01f7c406]
JSPrincipals::dump()+0x00ed8254 [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x01f79564]
NS_InvokeByIndex+0x0000021d [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x000a426d]
XRE_XPCShellMain+0x0001a058 [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x0106ae68]
XRE_XPCShellMain+0x000184dd [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x010692ed]
XRE_XPCShellMain+0x00022195 [/Volumes/data/code/mozmill/FirefoxNightlyDebug.app/Contents/MacOS/XUL +0x01072fa5]

Comment 1

[Blair McBride [:Unfocused] (UNAVAILABLE)]

Huh. How does Mozmill shutdown the browser?

Comment 2

[Edmund Wong (:ewong)]

Is this bug somewhat similar to bug 719719?

Comment 3

[Henrik Skupin [:whimboo][⌚️UTC+1]]

(In reply to Blair McBride [:Unfocused] from comment #1)
> Huh. How does Mozmill shutdown the browser?

Sure. To allow us testing a successful installation of an add-on we have to restart the browser. In cases as above the installation prompt gets drastically delayed (see bug 958126), and our test times out. So Firefox gets restarted to proceed with the next test.

Comment 4

[BMO Automation]

Per policy at https://wiki.mozilla.org/Bug_Triage/Projects/Bug_Handling/Bug_Husbandry#Inactive_Bugs. If this bug is not an enhancement request or a bug not present in a supported release of Firefox, then it may be reopened.