359488 - Random crash [@ [NSWindow sendEvent:]][@ 0xfffeff18][@ 0xfffeff20][@ libSystem.B.dylib]

Status: RESOLVED WORKSFORME

(Core :: Widget: Cocoa, defect)

Description

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Our topcrasher for 1.1a1 (and 3 other high crasher stacks) seems to match up with the stacks for the random crashes many of us have seen lately.  (Some of those stacks are catch-all, but talkback is matching some of them up with these crashes).

Some of the crashes seem to involve opening or closing tabs, but beyond that...we've got nothing :(

Comment 1

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Attachment: Example crashlog

Comment 2

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Sam's using an ancient build (10/13) and seeing this crash, so it's been going on for some time.

It often seems to happen when switching tabs, as well.

There is some speculation that it might be tooltips, although smorgan says his implementation is different than the older one (which did crash).

Comment 3

[Håkan Waara]

This is happening almost every day to me. 

I just crashed, and an interesting exception was thrown in the system console when the crash happened:

2006-11-09 01:15:50.315 Camino[206] *** -[NSCFString mouseEntered:]: selector not recognized [self = 0x299902c0]
2006-11-09 01:15:50.345 Camino[206] *** -[NSCFString mouseEntered:]: selector not recognized [self = 0x299902c0]

I wonder why we're trying to send an event to a string...

Comment 4

[Stuart Morgan]

We need more crashlogs here to try to get a sense of how many different problems we are actually dealing with.  We may be lumping more than one crash into a big bucket, which is going to be counterproductive.

Comment 5

[Stuart Morgan]

> I wonder why we're trying to send an event to a string...

Do you always see that? If we are dealing with something past its actual life, you may have just ended up with a string in that memory location by accident.

Comment 6

[Samuel Sidler (old account; do not CC)]

Attachment: Two Sample Crashlogs

Here are two sample crash logs from today. I was using Version 2006101301 (1.0+).

Comment 7

[Simon Fraser [no longer active]]

Someone who sees this should run with NSZombieEnabled=YES.

Comment 8

[philippe (part-time)]

1. I've seen the log entries in comment #3 once, but in a different context: Camino had stopped loading pages completely. Was stuck at 'loading....' for several minutes. At the same time those entries were logged. After several minutes, I tried to load anotehr page (dev server @localhost) with the same result (loading...), and the same log entries.
Camino didn't crash, as everything else was functional.
I had posted about this in the forums:
<http://forums.mozillazine.org/viewtopic.php?t=480082>

2. I've seen these types of crashes sometimes during the month of September (with trunk builds), and more frequently in October. Earliest entry in the crash log I can find is dated 2006.08.30, I'll attach it next. The most recent one: 2006.10.27. Since then: no crashes (knock on wood).
At first I thought those where related to the use of CaminoSession extension (it turned out not the case, in the end).
Iirc, most of those crashes happened when using the mouse to scroll a page using the scrollbar.

Comment 9

[philippe (part-time)]

Attachment: crash log, 20060830

Comment 10

[philippe (part-time)]

*** Bug 361936 has been marked as a duplicate of this bug. ***

Comment 11

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

Setting the most-closely analogous flag (we need this fixed for Camino 1.1)

Comment 12

[Daniel Veditz [:dveditz]]

missed 1.8.1.1, moving request out.

Comment 13

[Håkan Waara]

Is anyone still seeing this?

Comment 14

[philippe (part-time)]

(In reply to comment #13)
> Is anyone still seeing this?
> 

For Camino (trunk), nothing since my comment #8 above.
For Minefield: After posting the duped bug 361936 above, I experienced a few similar crashes until 1203 (when I duped that bugged). Since then nothing anymore, despite trying hard.

Comment 15

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

This isn't showing up in our branch Talkback post-a1 (there is no trunk Talkback still), and I haven't seen it for a while.

I'm OK-ish with closing this WFM for now, but I'd really prefer to have some stats from builds with higher usage/wider testing (i.e., 1.1a2) first, to ensure it's really gone, and we won't have those before the end of the month....

Comment 16

[Håkan Waara]

How about some stats from the gecko a1? 

Comment 17

[Smokey Ardisson (offline for a while; not following bugs - do not email)]

We're not seeing this in 1.1a2 Talkback or in any anecdotal reports, so I'll go ahead and close this.

As I mentioned in bug 361936 comment 4 and 6, the Firefox one is very different looking (in stack and in trigger), and if those stacks continue, that bug should be re-opened.