GNOME Bugzilla – Bug 542040
netspeed_applet2 0.14 leaks memory
Last modified: 2010-05-05 10:49:11 UTC
Please describe the problem: Cut-n-pasting from gentoo report since literature isn't my thing: netspeed_applet continuously leaks small amounts of memory. If left over a week, it grows into a 200-400 Meg beast. Growth seems to be close to linear. Steps to reproduce: 1. Add netspeed_applet to the gnome toolbar (or whatever that thing is called) 2. Use net for a while 3. Look at top, sort by memory 4. Observe existence of netspeed_applet in top 3 positions. Actual results: netspeed_applet continues to accrue DRAM real estate Expected results: netspeed_applet is a non-issue wrt memory as it doesn't do anything that should accumulate it Does this happen every time? Yes Other information: These same bug was posted in gentoo http://bugs.gentoo.org/show_bug.cgi?id=229969 Arun Raghavan valgrinded the applet and found some definite leaks. Valgrind output is attached to the gentoo bug report.
Created attachment 114180 [details] valgrind output Valgrind output created by Arun Raghavan
I have never seen this behaviour. Possibly leaking 120KB ==14990== 124,364 (38,656 direct, 85,708 indirect) bytes in 136 blocks are definitely lost in loss record 185 of 190 ==14990== at 0x4023BAE: realloc (in /usr/lib/valgrind/x86-linux/vgpreload_memcheck.so) ==14990== by 0x4E7557F: (within /usr/lib/libfontconfig.so.1.3.0) ==14990== by 0x4E7601F: (within /usr/lib/libfontconfig.so.1.3.0) ==14990== by 0x4E7668C: (within /usr/lib/libfontconfig.so.1.3.0) ==14990== by 0x4E72601: FcFontRenderPrepare (in /usr/lib/libfontconfig.so.1.3.0) ==14990== by 0x4E33D6E: (within /usr/lib/libpangoft2-1.0.so.0.2002.1) ==14990== by 0x454D66F: pango_font_map_load_fontset (in /usr/lib/libpango-1.0.so.0.2002.1) ==14990== by 0x454B5C1: (within /usr/lib/libpango-1.0.so.0.2002.1) ==14990== by 0x454B9A6: pango_itemize_with_base_dir (in /usr/lib/libpango-1.0.so.0.2002.1) ==14990== by 0x4554A98: (within /usr/lib/libpango-1.0.so.0.2002.1) ==14990== by 0x4555E7C: (within /usr/lib/libpango-1.0.so.0.2002.1) ==14990== by 0x4266737: (within /usr/lib/libgtk-x11-2.0.so.0.1200.10) doesn't seem to be enough to move netspeed into top3 memory usage. Which gtk theme are you using ? Would you be able to re-run valgrind with a bigger num-callers ?
Please also display reachables.
Oh and what about "Running ~x86. This has been happening for a long time (over a year). It does not happen on Ubuntu." It very looks like a theme issue.
Not sure what theme I was in. Preferences=>Appearance seemed to suggest I was using ClearLooks. I've now switched it to Glossy and restarted netspeed_applet to see if the same thing happens. I'll report back in a couple of days. (If I'm changing gnome themes instead of gtk themes in case they are not the same thing, please advise. I've got zero idea about themes)
Created attachment 114714 [details] Improved valgrind output This is from a longer run (several hours) of Valgrind with --leak-check=full and --show-reachable=yes Not sure there are any obvious culprits here (at least not from the applet itself).
Changing to Glossy seems not to have helped. netspeed_applet currently second on top memory users here at 118Megs resident after running for around a week. Any ideas about what I could try?
==17213== LEAK SUMMARY: ==17213== definitely lost: 26,414 bytes in 95 blocks. ==17213== indirectly lost: 57,816 bytes in 2,858 blocks. ==17213== possibly lost: 127,370 bytes in 152 blocks. ==17213== still reachable: 2,413,182 bytes in 15,608 blocks. No obvious leak. Could you run valgrind in massif mode please ?
0.15.2 has a leak also (Gentoo/~amd64).
Problem disappears when Assistive Technologies is disabled and gnome is restarted.
Dear reporter, could you please provide the information requested in comment #8? Benoît, could you please tell us how to obtain the requested information?
Hm. If there'll be no activity soonish, I'd say we have to close this bug report as INCOMPLETE. Would be a pity for all the obtained valgrind logs...
Closing this bug report as no further information has been provided. Please feel free to reopen this bug if you can provide the information asked for. Thanks!