GNOME Bugzilla – Bug 94422
xembed crash possibly related to the notification area
Last modified: 2015-03-24 13:00:35 UTC
Package: gnome-panel Severity: normal Version: 2.0.6 Synopsis: panel crashed on double-clicking on a tasklist item Bugzilla-Product: gnome-panel Bugzilla-Component: Panel BugBuddy-GnomeVersion: 2.0 (2.0.3) Description: Description of Problem: This is in redhat 8.0 Steps to reproduce the problem: 1. 2. 3. Actual Results: Expected Results: How often does this happen? Additional Information: Debugging Information: Backtrace was generated from '/usr/bin/gnome-panel' (no debugging symbols found)...(no debugging symbols found)...[New Thread 8192 (LWP 2474)] 0x420ae169 in wait4 () from /lib/i686/libc.so.6
+ Trace 27985
Thread 1 (Thread 8192 (LWP 2474))
------- Bug moved to this database by unknown@bugzilla.gnome.org 2002-09-28 11:02 ------- Reassigning to the default owner of the component, gnome-panel-maint@bugzilla.gnome.org.
Okay, this seems to be crashing when gtk+ tries to send a focus in message using xembed. Moving to gtk+ in case the stack trace is of any use ...
I can't see this stack trace occuring without something being screwed up elsewhere. In order to get this backtrace, you'd need to have socket->plug_window non-NULL but not a valid GdkWindow. It could be some obscure (more complex) GtkSocket bug, or it could be a bonobo or gnome-panel bug. Hard to get much further without being able to reproduce the bug unfortunately. Have you seen this crash multiple times or just once? What exactly do you mean by "task list icon" - one of the buttons in the task list? Does it matter which application the button is for?
*** Bug 94924 has been marked as a duplicate of this bug. ***
Thanks owen, I'd forgotten about this one
*** Bug 95345 has been marked as a duplicate of this bug. ***
*** Bug 95458 has been marked as a duplicate of this bug. ***
*** Bug 95356 has been marked as a duplicate of this bug. ***
*** Bug 95287 has been marked as a duplicate of this bug. ***
Owen: one thing to note that I hadn't thought of it these are in-proc applets, so I wouldn't have thought we would have been using xembed at all.
*** Bug 95748 has been marked as a duplicate of this bug. ***
*** Bug 97161 has been marked as a duplicate of this bug. ***
*** Bug 97803 has been marked as a duplicate of this bug. ***
Apologies for the spam- I'm removing 'bugsquad' from some keywords via the web interface. This is a one-time only thing before I re-add bugsquad via the SQL interface, which will generate no mail. Apologies again.
What's going on with this bug? Think someone can take a look before 2.0.3?
*** Bug 98304 has been marked as a duplicate of this bug. ***
*** Bug 99030 has been marked as a duplicate of this bug. ***
Ooops. Sorry I marked that last one twice. Since I was able to duplicate this bug, is there any information I might be able to provide to help track it down?
If you can reliably reproduce it, the useful thing to have would be exact information about: - The steps to reproduce - What versions of gnome-panel, gtk+, bonobo, bonoboui, libgnomeui you have installed.
Elijah ?
Oops, sorry. I forgot to add myself to the cc list. I just tried to duplicate again many times. This bug seems repeatable but not really often--maybe only 5% of the time that I try to repeat it do I actually get a crash. But I do have some more information that may be helpful. I've looked through the some of the duplicates, and it appears that there are some minor variations on how to get the crash (almost all, if not all, appear to be RedHat 8, though). I'm going to try some of those and report back later. Here's how I got the crash. Repeatedly: 1. Log in 2. Remove the Redhat update icon 3. Log out The crash will not occur instantaneously upon removal of the redhat update icon. It occurs when you press the redhat menu selector on the panel in order to try to log out. When the crash happens, the gray background of the menu appears but none of the options do--it's just a gray rectangle. Thus you don't actually get to select 'Log Out'; so perhaps step 3 isn't really relevant. However, I've also tried manually re-adding the rhn applet from a gnome terminal and removing it, but wasn't able to get the crash. I think logging in and out may be required for me to duplicate the bug. Also, the last two times I duplicated this, I got a long stack trace (about 80 functions deep) like the ones found in 95287. I have it saved in a file if it'd be useful at all. I haven't compared it to the others posted other than the first 2 functions or so. Here's some other things I noticed that may or may not be related: 1) One of the times I logged in, as soon as the panel appeared the battstat applet crashed. Bug-buddy didn't find a stack trace, but as soon as I closed that bug-buddy report, the panel itself crashed and bug-buddy got a stacktrace. The latter of these two bugs I reported (bug 99725), but assumed it was unrelated at the time. The thing that makes me believe that bug (or the battstat applet in general) may be related to this is the following: the last time I was able to duplicate this bug, the battstat applet crashed immediately following the panel crash duplicate that I got (note that I did not submit either of these bugs--the panel crash report I simply saved as I mentioned above, and bug-buddy didn't get a stacktrace for the battstat crash). 2) When I started typing in the bug-buddy report from my last duplicate, the repeat delay for the keyboard was super low (which made it hard to type just a single key). That was a problem I had earlier on that laptop, and I tried to fix it by adding 'xset r rate 500 40' to my .cshrc. That normally works fine for me, but the fact that it hadn't taken effect seemed suspect to me. Simplify firing up a gnome-terminal before continuing the bug-buddy report fixed this problem. But perhaps it's related? I may be way off on both of these (feel free to tell me so), but not knowing much (I'm only a wanna-be-hacker) and since this is hard to duplicate, I'm just mentioning everything that looked like it could have been related. And, to answer Owen's question about the versions of software I have: 1 typo:~> rpm -q gnome-panel gtk+ bonobo bonoboui libgnomeui gnome-panel-2.0.6-9 gtk+-1.2.10-22 bonobo-1.0.20-3 package bonoboui is not installed libgnomeui-2.0.3-3 Let me know what else to try or do to try to help with this.
Okay, I've done a ton more repetitions of the above 3 steps (probably 100-125 of them). I got the crash 7 more times. Here's what I've found out: I don't think xset has anything to do with it--I commented it out the the .cshrc and was still able to get a crash. I don't thing the battery applet has anything to do with it--it just also happens to be very unstable (I think I got a few more battstat crashes than panel crashes, and only once more did I get them at the same time). After I removed the battery applet from my panel and tried repeating the three steps, I was still able to get crashes two more times at about the same rate. About 20 of the repetitions were done on a desktop computer (so I could try to rule out problems with just a laptop). This desktop computer was also running Redhat 8.0 with the same versions of the packages. It, of course, had no battery applet, and I also never tried running xset or anything on that machine (in fact, I have no .cshrc file at all there). Yet, I was still able to get the crash. Other possible points of interest: These 7 stack traces are all basically identical except that two of them contain the g_type_check_instance_is_a() function while the other five do not. Each stack trace contained either 80 or 81 function calls (similar to some duplicates of this bug). None of them contained 23 function calls as the stack trace listed in this bug report (yet my bug 99030 stack trace duplicate did...). Off topic: bug-buddy doesn't seem to be able to obtain a stack trace for any of the battery applet crashes. Any one know how to change this? Further off-topic: When I log out, the redhat update icon isn't on the panel. Most the time when I log back in, the redhat icon is put onto the panel--but not everytime (I think about 97% of the time it is). Unless it's meant to be non-deterministic, this sounds like a bug. Do I file it at bugzilla.redhat.com? Back on topic. In addition to the over a hundred reps that I've talked about above, I did 100 repetitions of something somewhat similar to determine whether the crash could be duplicated without logging in and out. These are those steps: 1) Remove the RedHat Update Icon 2) Click on the RedHat Menu Launcher thingy 3) Click on a gnome-terminal (both to make the menu go away and to give focus to the gnome-terminal) 4) Run 'rhn-applet-gui &' to restore the applet Actually, I didn't do step #2 the first 46 times, but I did the last 54. No crashes whatsoever. So it appears that something to do with logging in and logging out seems relevant.. My questions: Should I add any of the stack traces I'm getting? How do I get bug-buddy to be able to get stack traces for the battery applet crashes? Should I file the other bug at redhat.bugzilla.com or here or at all? Is this at all helpful? I hope so, because it sure takes a long time. But if not, I'll be happy to stop. I still haven't looked at the variations reported in some duplicates. I may do that later, but I'm a bit tired of this right now. Let me know if anything I've said is unclear or if there's any specific tests that might be more helpful than others.
Elijah: thanks so much for going to so much bother. Sounds very painful. Owen, Havoc: this is increasingly sounding like a bug with the RedHat up2date thingy. What is that? Is it an icon on the system tray ? That would explain the xembed traces .... Have seen many dups of this on bugzilla.redhat.com ?
*** Bug 100362 has been marked as a duplicate of this bug. ***
Probably not an up2date bug, perhaps an eggtrayicon bug. Hard to know, though. I haven't seen any dups of this on redhat.com, no.
It can't (strictly speaking) be an up2date bug, because the systray shouldn't crash no matter what the icons in it do. I suspect it has nothing to do with the particular icons in the system tray; it's just that the two Red Hat 8 icons are going to be the majority of what people are using with the XEMBED-based system tray.
Okay, moving to the notification area for the moment.
Um. Yeah.
*** Bug 102421 has been marked as a duplicate of this bug. ***
*** Bug 102597 has been marked as a duplicate of this bug. ***
The crash occurred for me after I had removed the RHN Update Applet. I went to run Red Hat Network after removing it and that is when the crash occurs. The versions of the above mentioned packages are as follows: gnome-panel-2.0.6-9 gtk+-1.2.10-22 bonobo-1.0.20-3 package bonoboui is not installed libgnomeui-2.0.3-3
*** Bug 103363 has been marked as a duplicate of this bug. ***
*** Bug 104575 has been marked as a duplicate of this bug. ***
*** Bug 105545 has been marked as a duplicate of this bug. ***
*** Bug 105578 has been marked as a duplicate of this bug. ***
*** Bug 107412 has been marked as a duplicate of this bug. ***
*** Bug 107705 has been marked as a duplicate of this bug. ***
Okay, every single one of these is with gnome-panel 2.0.6. Assuming its fixed in 2.2.x since we have no dups. Re-open if there's any reports of this bug with 2.2.x
*** Bug 108217 has been marked as a duplicate of this bug. ***
*** Bug 108851 has been marked as a duplicate of this bug. ***
*** Bug 109027 has been marked as a duplicate of this bug. ***
*** Bug 109455 has been marked as a duplicate of this bug. ***
*** Bug 110537 has been marked as a duplicate of this bug. ***
*** Bug 111721 has been marked as a duplicate of this bug. ***
*** Bug 112043 has been marked as a duplicate of this bug. ***
*** Bug 110658 has been marked as a duplicate of this bug. ***
*** Bug 109739 has been marked as a duplicate of this bug. ***
*** Bug 113863 has been marked as a duplicate of this bug. ***
*** Bug 114782 has been marked as a duplicate of this bug. ***
*** Bug 114250 has been marked as a duplicate of this bug. ***
*** Bug 118542 has been marked as a duplicate of this bug. ***
*** Bug 119283 has been marked as a duplicate of this bug. ***
*** Bug 119344 has been marked as a duplicate of this bug. ***
*** Bug 120879 has been marked as a duplicate of this bug. ***
*** Bug 121349 has been marked as a duplicate of this bug. ***
*** Bug 110637 has been marked as a duplicate of this bug. ***
*** Bug 121540 has been marked as a duplicate of this bug. ***
*** Bug 122594 has been marked as a duplicate of this bug. ***
*** Bug 123603 has been marked as a duplicate of this bug. ***
I have a strong suspicion that this actually was the problem described in bug 123570 and bug 123585, though I don't know why it stopped crashing like this with GNOME-2.2. In some cases, the *same* tray icon can get docked in the notification area multiple times; the symptom I was tracking down was a sizing problem for the RHN icon (https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=75225) but it makes a lot of sense to me that it would cause this crash as well - if their are multiple sockets that all think they own the RHN icon, then removing the RHN icon will only clean up one of them; the rest will be left with a stale pointer to the icon window.
*** Bug 124380 has been marked as a duplicate of this bug. ***
*** Bug 128335 has been marked as a duplicate of this bug. ***
*** Bug 133987 has been marked as a duplicate of this bug. ***
*** Bug 146181 has been marked as a duplicate of this bug. ***
*** Bug 146770 has been marked as a duplicate of this bug. ***
*** Bug 161995 has been marked as a duplicate of this bug. ***
*** Bug 167978 has been marked as a duplicate of this bug. ***
*** Bug 316318 has been marked as a duplicate of this bug. ***