Description of problem: Fairly regularly, xfce4-panel will segfaults during a vnc server startup. It can be started afterwards without problem. Not sure if this is really a xfce4-panel issue or a library issue, but here goes. Stack trace (may or may not be valid): Version-Release number of selected component (if applicable): #0 0x0000003f2ea3ac58 in IA__g_ascii_strcasecmp (s1=0x4165d6 "xfce", s2=0x21 <Address 0x21 out of bounds>) at gstrfuncs.c:1753 #1 0x00000000004116d6 in notify_cb ( name=0x220000002e <Address 0x220000002e out of bounds>, channel_name=Variable "channel_name" is not available. ) at mcs_client.c:109 #2 0x0000003ae3302bf6 in read_settings (client=0x5e3370, channel_name=Variable "channel_name" is not available. ) at mcs-client.c:250 #3 0x0000003ae3303132 in mcs_client_process_event (client=Variable "client" is not available. ) at mcs-client.c:792 #4 0x0000000000411747 in client_event_filter (xevent=Variable "xevent" is not available. ) at mcs_client.c:129 #5 0x0000003a06d3d326 in gdk_event_apply_filters (xevent=0x7fffffd86850, event=0x6ecc00, filters=Variable "filters" is not available. ) at gdkevents-x11.c:328 #6 0x0000003a06d3e8bc in gdk_event_translate (display=0x537b30, event=0x6ecc00, xevent=0x7fffffd86850, return_exposes=0) at gdkevents-x11.c:872 #7 0x0000003a06d40078 in _gdk_events_queue (display=0x537b30) at gdkevents-x11.c:2193 #8 0x0000003a06d400df in gdk_event_dispatch (source=Variable "source" is not available. ) at gdkevents-x11.c:2253 #9 0x0000003f2ea24a9e in IA__g_main_context_dispatch (context=0x546c80) at gmain.c:1934 #10 0x0000003f2ea27734 in g_main_context_iterate (context=0x546c80, block=1, dispatch=1, self=Variable "self" is not available. ) at gmain.c:2565 #11 0x0000003f2ea27c20 in IA__g_main_loop_run (loop=0x6f89f0) at gmain.c:2769 #12 0x0000003a06802145 in IA__gtk_main () at gtkmain.c:1009 #13 0x00000000004113d2 in main (argc=1, argv=0x7fffffd86cd8) at main.c:501 (gdb) up #1 0x00000000004116d6 in notify_cb ( name=0x220000002e <Address 0x220000002e out of bounds>, channel_name=Variable "channel_name" is not available. ) at mcs_client.c:109 109 if (g_ascii_strcasecmp (CHANNEL, channel_name)) (gdb) list 104 /* event handling */ 105 static void 106 notify_cb (const char *name, const char *channel_name, 107 McsAction action, McsSetting * setting, void *data) 108 { 109 if (g_ascii_strcasecmp (CHANNEL, channel_name)) 110 return; 111 112 switch (action) 113 { How reproducible: Fairly. Steps to Reproduce: 1. startup vncserver with the following xstartup: #!/bin/sh ulimit -c unlimited if [ -f $HOME/.Xresources ] then xrdb $HOME/.Xresources fi xsetroot -solid grey xterm -geometry 80x24+10+10 -ls -title "$VNCDESKTOP Desktop" & #mwm xfwm4 & xfce4-panel &
Greetings. I don't have a x86_64 test box, so it might be difficult to duplicate this here. ;( Can you try a few things: - rpm -ivh http://download.fedora.redhat.com/pub/fedora/linux/extras/4/x86_64/debug/xfce4-panel-debuginfo-4.2.3-1.fc4.x86_64.rpm (assuming you are running the latest version from extras). Then re-run your gdb traces? This should hopefully provide more information. - Running xfce4-panel & in your xstartup is not the prefred way to start xfce, (although it should work). can you try: exec startxfce4 and see if it makes any diffrence?
The debuginfo package is installed (otherwise you wouldn't see line number info). We'll try startxfce4 and see if that helps.
Just got access to a fc4 x86_64 box to test with. I used your ~/.vnc/xstartup file from above. The first time I ran vncserver everything was fine, but after that I was able to see xfce4-panel die like you were. ;) It looks like it's a race condition if you try and start xfwm4 and xfce4-panel at the same time. If I put a 'sleep 2' between them I can't get it to fail. It also always works for me if I use the 'exec startxfce4' (which needs xfce-utils installed, BTW). I did some tracing on xfce4-panel, and it dies when it's trying to talk to xfce-mcs-manager it looks like. I will mention this issue on the upstream lists, although they may well say "don't do that then". If you could confirm if startxfce4 works that would be good.
Any news on if using startxfce4 works around the issue for you? I posted to the upstream list, but didn't get any response. I can file an upstream bug on it...
So far no problems with startxfce4. I think that will work for us.