Closed Bug 511831 Opened 15 years ago Closed 15 years ago

Crash [@ js_Invoke] playing videos at Comcast.net

Categories

(Core :: JavaScript Engine, defect)

1.9.2 Branch
x86
macOS
defect
Not set
critical

Tracking

()

RESOLVED DUPLICATE of bug 516009

People

(Reporter: stephend, Assigned: dmandelin)

References

()

Details

(Keywords: crash)

Crash Data

STR:

1. Using a Namoroka nightly (I'm on Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.2a2pre) Gecko/20090820 Namoroka/3.6a2pre), load http://www.comcast.net
2. Acknowledge the warning about browser compatibility
3. Click on any of the videos, and just wait -- eventually you'll crash (timer?):

0  	libmozjs.dylib  	js_Invoke  	 js/src/jsinterp.cpp:1130
1 	libmozjs.dylib 	js_Interpret 	js/src/jsinterp.cpp:5229
2 	libmozjs.dylib 	js_Invoke 	js/src/jsinterp.cpp:1379
3 	libmozjs.dylib 	js_InternalInvoke 	js/src/jsinterp.cpp:1451
4 	libmozjs.dylib 	JS_CallFunctionValue 	js/src/jsapi.cpp:5176
5 	XUL 	nsJSContext::CallEventHandler 	dom/base/nsJSEnvironment.cpp:2097
6 	XUL 	nsGlobalWindow::RunTimeout 	dom/base/nsGlobalWindow.cpp:7935
7 	XUL 	nsGlobalWindow::TimerCallback 	dom/base/nsGlobalWindow.cpp:8269
8 	XUL 	nsTimerImpl::Fire 	xpcom/threads/nsTimerImpl.cpp:427
9 	XUL 	nsTimerEvent::Run 	xpcom/threads/nsTimerImpl.cpp:519
10 	XUL 	nsThread::ProcessNextEvent 	xpcom/threads/nsThread.cpp:527
Summary: Crash in [js_Invoke] playing videos at Comcast.net → Crash [@ js_Invoke] playing videos at Comcast.net
I just tried this with a debug TM build and it blinked for a while, showing the URL 'http://www.comcast.net/browserupgrade/' in the location bar while spewing a lot of stuff like this to the console, and then stopped with "Unable to connect".

--DOMWINDOW == 22 (0x1574ae80) [serial = 358] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 21 (0x163befc0) [serial = 359] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 20 (0x1893e190) [serial = 360] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 19 (0x13998bf0) [serial = 361] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 18 (0x18933720) [serial = 362] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 17 (0x15f35ce0) [serial = 363] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 16 (0x10dc42c0) [serial = 364] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 15 (0x10d615c0) [serial = 365] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 14 (0x15f42530) [serial = 366] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
--DOMWINDOW == 13 (0x1634c020) [serial = 367] [outer = 0x1356ee60] [url = http://www.comcast.net/browserupgrade/]
I got the same thing (minus console spew) with today's nightly.
(In reply to comment #4)
> I got the same thing (minus console spew) with today's nightly.

Comcast is doing user-agent sniffing now, so you'll have to use general.useragent.override (type: string) with something like "Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1.2) Gecko/20090729 Firefox/3.5.2"

Crash report: http://crash-stats.mozilla.com/report/index/b3f015f6-81a9-4d87-9a7f-26c272090905?p=1
I can replicate with today's trunk and TM nightlies. It's crashing on trace so it will not be easy to debug. Could someone please get us a regression window?
Whiteboard: regression-window-wanted
3.6a1pre, it could be between 6/15 and 6/16. I left the build from 6/15 play several videos in a span of a couple of hours and it didn't crash, whereas the build from 6/16 crashes eventually, although not right away.

http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=a012b6f9858b&tochange=ca8799e74642
Whiteboard: regression-window-wanted
I don't see any JS changes in that window.
It could be that I didn't wait long enough for the early June builds to crash, but in at least two early June builds from 6/3 and 6/12, for example, I let it play videos for several minutes, and the one from 6/15 for at least two hours with no crash. I'll try checking these again, with fresher profiles, since I tried to find the regression with the same one.
Starting from the range in comment 7, I rebisected using the TM tree and found the problem to start with this rev:

changeset:   28946:a6f9df8c33a9
user:        David Mandelin <dmandelin@mozilla.com>
date:        Fri Jun 05 16:19:06 2009 -0700
summary:     Bug 496270: correct use of C++ reference returned from function, r=gal
Assignee: general → dmandelin
I believe this is a dup of bug 516009, which has already been fixed (in TM and m-c). I also tried these videos again on today's TM nightly, and they did not crash. The fix has not to my knowledge been landed to 1.9.2 yet but is marked blocking.
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → DUPLICATE
Crash Signature: [@ js_Invoke]
You need to log in before you can comment on or make changes to this bug.