[Bug 338737] New: gnome-panel leakage
https://bugzilla.novell.com/show_bug.cgi?id=338737 Summary: gnome-panel leakage Product: openSUSE 11.0 Version: unspecified Platform: i586 OS/Version: openSUSE 10.3 Status: NEW Severity: Normal Priority: P5 - None Component: GNOME AssignedTo: bnc-team-gnome@forge.provo.novell.com ReportedBy: mtgordon@novell.com QAContact: qa@suse.de Found By: System Test While confirming that Bug 78329 was fixed, I found a new generation of panel leakage. Valgrind log at https://bugzilla.novell.com/attachment.cgi?id=181818 IMHO, this will need to be evaluated to determine a good idea of the true severity. For now, I'm going to leave the severity at Normal. The summary after maybe half an hour is: ==17469== LEAK SUMMARY: ==17469== definitely lost: 42,195 bytes in 268 blocks. ==17469== indirectly lost: 79,100 bytes in 3,901 blocks. ==17469== possibly lost: 270,809 bytes in 2,026 blocks. ==17469== still reachable: 3,474,048 bytes in 42,821 blocks. ==17469== suppressed: 0 bytes in 0 blocks. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=338737
JP Rosevear
https://bugzilla.novell.com/show_bug.cgi?id=338737
Magnus Boman
https://bugzilla.novell.com/show_bug.cgi?id=338737
Vincent Untz
https://bugzilla.novell.com/show_bug.cgi?id=338737
User vuntz@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=338737#c1
Vincent Untz
https://bugzilla.novell.com/show_bug.cgi?id=338737
User mtgordon@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=338737#c2
Mark Gordon
https://bugzilla.novell.com/show_bug.cgi?id=338737
User vuntz@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=338737#c3
--- Comment #3 from Vincent Untz
AFAIK, the breakpad noise is a function of exception handler initialization code
Ah maybe. Apologies.
, i.e. "I'm ready to handle exceptions" code, distinct from "I'm actually handling an exception" code. If gnome-panel crashed, it crashed because I hit Ctrl-C, which is how I terminated the valgrind session. I sincerely believe that if a program can't clean up its memory properly in response to SIGINT, it's a bug
I'm nearly sure that 99% of the programs out there don't properly clean up the memory on SIGINT. The reason is that the memory will be returned back anyway since the process is exiting and the exit path is not the normal one.
but if you have alternative advice on how to detect memory leakage in gnome-panel, I'm interested in hearing it.
Just run "gnome-session-remove gnome-panel" to have the panel properly exit :-)
In any case, I'm unable to generate a valgrind log that's devoid of breakpad spewage. If that renders this bug report INVALID, then I guess there's not much I can do.
Updated valgrind log with G_SLICE=always-malloc (and the same old breakpad spewage) attached.
Thanks. There's at least one leak in the multiload applet. I'll send a patch for this. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=338737
User vuntz@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=338737#c4
Vincent Untz
https://bugzilla.novell.com/show_bug.cgi?id=338737
JP Rosevear
participants (1)
-
bugzilla_noreply@novell.com