GNOME Bugzilla – Bug 351027
Crashed when changing folders in the mailer
Last modified: 2013-09-14 16:49:47 UTC
Memory status: size: 441614336 vsize: 0 resident: 441614336 share: 0 rss: 183721984 rss_rlim: 0 CPU usage: start_time: 1155377223 rtime: 0 utime: 19495 stime: 0 cutime:14879 cstime: 0 timeout: 4616 it_real_value: 0 frequency: 5 Backtrace was generated from '/usr/bin/evolution-2.8' Using host libthread_db library "/lib/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1209178416 (LWP 15767)] [New Thread -1448084576 (LWP 16021)] [New Thread -1380988000 (LWP 16018)] [New Thread -1363154016 (LWP 15855)] [New Thread -1352664160 (LWP 15854)] [New Thread -1282942048 (LWP 15821)] [New Thread -1272452192 (LWP 15820)] [New Thread -1261962336 (LWP 15816)] 0xb7f09402 in __kernel_vsyscall ()
+ Trace 70295
Thread 5 (Thread -1352664160 (LWP 15854))
#0 0xb7f09
looks like gtkhtml
I think this happened from two different points in gtkhtml so it could be a gdk/gtk+ problem I reckon. Here's the other trace I got:
+ Trace 70311
Moving to e-d-s since the real problem seems to be in search_match_all() in camel.
Tried getting more data with valgrind and found these in my log: ==15585== Invalid free() / delete / delete[] ==15585== at 0x4004FEA: free (vg_replace_malloc.c:233) ==15585== by 0x5635E2: free_mem (dl-libc.c:235) ==15585== by 0x563202: __libc_freeres (set-freeres.c:47) ==15585== by 0x40011F6: _vgnU_freeres (vg_preloaded.c:60) ==15585== by 0x4E96D3: _Exit (in /lib/libc-2.4.90.so) ==15585== by 0x8A53A9: g_spawn_async_with_pipes (gspawn.c:584) ==15585== by 0x8A5449: g_spawn_async (gspawn.c:102) ==15585== by 0x6A13D3B: gconf_activate_server (gconf-internals.c:3046) ==15585== by 0x6A1FF51: gconf_get_config_server (gconf.c:2207) ==15585== by 0x6A20CDA: gconf_engine_connect (gconf.c:324) ==15585== by 0x6A20F2D: gconf_engine_get_database (gconf.c:399) ==15585== by 0x6A2263F: gconf_engine_set (gconf.c:1248) ==15585== by 0x6A2272C: error_checked_set (gconf.c:3321) ==15585== by 0x6A271D9: gconf_client_set_int (gconf-client.c:1712) ==15585== by 0x805B134: store_window_size (e-shell-window.c:910) ==15585== by 0x87A915: g_timeout_dispatch (gmain.c:3420) ==15585== by 0x87A341: g_main_context_dispatch (gmain.c:2043) ==15585== by 0x87D31E: g_main_context_iterate (gmain.c:2675) ==15585== by 0x87D6C8: g_main_loop_run (gmain.c:2879) ==15585== by 0x6B03A02: bonobo_main (bonobo-main.c:311) ==15585== by 0x805DB95: main (main.c:614) ==15585== Address 0x4033350 is not stack'd, malloc'd or (recently) free'd Though I'm not sure this is related to the other problem.
Could it be something to do with file size/message count? I see gmem.c is bailing out with a message that it can't allocate memory in the trace, and the folder causing it most often is fairly large (170k messages and ~700 MB)
We have a new lead on this thanks to Matthias Clasen: http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=202309#c24 Looks like it's only affecting those using a _very_ recent glibc.
This was confirmed to be a glibc problem.