| Summary: | memory leak in gnome-settings-daemon | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Joe Wright <jwright> | ||||||
| Component: | gnome-settings-daemon | Assignee: | Rui Matos <rmatos> | ||||||
| Status: | CLOSED WONTFIX | QA Contact: | Desktop QE <desktop-qa-list> | ||||||
| Severity: | medium | Docs Contact: | |||||||
| Priority: | unspecified | ||||||||
| Version: | 7.2 | CC: | cww, jwright, tpelka | ||||||
| Target Milestone: | rc | Keywords: | Desktop, Performance | ||||||
| Target Release: | --- | ||||||||
| Hardware: | x86_64 | ||||||||
| OS: | Linux | ||||||||
| Whiteboard: | |||||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||||
| Doc Text: | Story Points: | --- | |||||||
| Clone Of: | Environment: | ||||||||
| Last Closed: | 2016-11-15 22:46:19 UTC | Type: | Bug | ||||||
| Regression: | --- | Mount Type: | --- | ||||||
| Documentation: | --- | CRM: | |||||||
| Verified Versions: | Category: | --- | |||||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||||
| Attachments: |
|
||||||||
After installing the debuginfo packages for gnome-settings-daemon with debuginfo-install, run gnome-settings-daemon under valgrind: valgrind /usr/libexec/gnome-settings-daemon --replace Let it run a little while (from the logs, about 20 minutes should be enough to capture a part of the leak), and run: /usr/libexec/gnome-settings-daemon --replace To make the original gnome-settings-daemon stop, and valgrind to report all the leaked memory. Created attachment 1163353 [details]
valgrind data
see attached
Those 2: ==13846== 159,170 (32,736 direct, 126,434 indirect) bytes in 22 blocks are definitely lost in loss record 14,329 of 14,331 ==13846== at 0x4C29BFD: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13846== by 0x77D225E: g_malloc (gmem.c:97) ==13846== by 0x77BC632: g_hash_table_get_keys_as_array (ghash.c:1703) ==13846== by 0x66C6D26: g_settings_schema_source_list_schemas (gsettingsschema.c:803) ==13846== by 0x4049E1: is_schema (gnome-settings-manager.c:208) ==13846== by 0x4049E1: _load_file (gnome-settings-manager.c:261) ==13846== by 0x4049E1: _load_dir (gnome-settings-manager.c:314) ==13846== by 0x4049E1: _load_all (gnome-settings-manager.c:330) ==13846== by 0x4049E1: gnome_settings_manager_start (gnome-settings-manager.c:424) ==13846== by 0x403F4A: start_settings_manager (main.c:149) ==13846== by 0x403F4A: name_acquired_handler (main.c:304) ==13846== by 0x66E5365: do_call (gdbusnameowning.c:215) ==13846== by 0x66E5597: request_name_cb (gdbusnameowning.c:327) ==13846== by 0x6681F46: g_simple_async_result_complete (gsimpleasyncresult.c:763) ==13846== by 0x66DD321: g_dbus_connection_call_done (gdbusconnection.c:5497) ==13846== by 0x6681F46: g_simple_async_result_complete (gsimpleasyncresult.c:763) ==13846== by 0x6681FA8: complete_in_idle_cb (gsimpleasyncresult.c:775) and ==16821== 15,334 (2,640 direct, 12,694 indirect) bytes in 22 blocks are definitely lost in loss record 11,293 of 11,321 ==16821== at 0x4C29BFD: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==16821== by 0x77D225E: g_malloc (gmem.c:97) ==16821== by 0x77BC632: g_hash_table_get_keys_as_array (ghash.c:1703) ==16821== by 0x66C6D4A: g_settings_schema_source_list_schemas (gsettingsschema.c:809) ==16821== by 0x4049E1: is_schema (gnome-settings-manager.c:208) ==16821== by 0x4049E1: _load_file (gnome-settings-manager.c:261) ==16821== by 0x4049E1: _load_dir (gnome-settings-manager.c:314) ==16821== by 0x4049E1: _load_all (gnome-settings-manager.c:330) ==16821== by 0x4049E1: gnome_settings_manager_start (gnome-settings-manager.c:424) ==16821== by 0x403F4A: start_settings_manager (main.c:149) ==16821== by 0x403F4A: name_acquired_handler (main.c:304) ==16821== by 0x66E5365: do_call (gdbusnameowning.c:215) ==16821== by 0x66E5597: request_name_cb (gdbusnameowning.c:327) ==16821== by 0x6681F46: g_simple_async_result_complete (gsimpleasyncresult.c:763) ==16821== by 0x66DD321: g_dbus_connection_call_done (gdbusconnection.c:5497) ==16821== by 0x6681F46: g_simple_async_result_complete (gsimpleasyncresult.c:763) ==16821== by 0x6681FA8: complete_in_idle_cb (gsimpleasyncresult.c:775) Are https://bugzilla.gnome.org/show_bug.cgi?id=754681 But they're one-time (at startup), so not our problem. Those look like leaks inside the CUPS library: ==13846== 8,368 bytes in 1 blocks are definitely lost in loss record 14,257 of 14,331 ==13846== at 0x4C2B974: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13846== by 0x233508C5: cupsFileOpenFd (file.c:1220) ==13846== by 0x23350B92: cupsFileOpen (file.c:1170) ==13846== by 0x233785A3: _cupsSetDefaults (usersys.c:836) ==13846== by 0x233787BA: cupsServer (usersys.c:183) ==13846== by 0x23126556: gsd_print_notifications_manager_start_idle (gsd-print-notifications-manager.c:1324) ==13846== by 0x77CC799: g_main_dispatch (gmain.c:3109) ==13846== by 0x77CC799: g_main_context_dispatch (gmain.c:3708) ==13846== by 0x77CCAE7: g_main_context_iterate.isra.24 (gmain.c:3779) ==13846== by 0x77CCDB9: g_main_loop_run (gmain.c:3973) ==13846== by 0x5A8A044: gtk_main (gtkmain.c:1207) ==13846== by 0x4037C0: main (main.c:427) ==13846== ==13846== 8,368 bytes in 1 blocks are definitely lost in loss record 14,258 of 14,331 ==13846== at 0x4C2B974: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13846== by 0x233508C5: cupsFileOpenFd (file.c:1220) ==13846== by 0x23350B92: cupsFileOpen (file.c:1170) ==13846== by 0x233785A3: _cupsSetDefaults (usersys.c:836) ==13846== by 0x23378779: cupsEncryption (usersys.c:100) ==13846== by 0x23371D37: _cupsConnect (request.c:1034) ==13846== by 0x23372BC6: cupsDoIORequest (request.c:156) ==13846== by 0x233480F2: _cupsGetDests (dest.c:1467) ==13846== by 0x2334925B: cupsGetDests2 (dest.c:1688) ==13846== by 0x231265A4: gsd_print_notifications_manager_start_idle (gsd-print-notifications-manager.c:1325) ==13846== by 0x77CC799: g_main_dispatch (gmain.c:3109) ==13846== by 0x77CC799: g_main_context_dispatch (gmain.c:3708) ==13846== by 0x77CCAE7: g_main_context_iterate.isra.24 (gmain.c:3779) ==13846== ==13846== 8,368 bytes in 1 blocks are definitely lost in loss record 14,259 of 14,331 ==13846== at 0x4C2B974: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13846== by 0x233508C5: cupsFileOpenFd (file.c:1220) ==13846== by 0x23350B92: cupsFileOpen (file.c:1170) ==13846== by 0x233785A3: _cupsSetDefaults (usersys.c:836) ==13846== by 0x23378779: cupsEncryption (usersys.c:100) ==13846== by 0x231248DF: renew_subscription (gsd-print-notifications-manager.c:1076) ==13846== by 0x231265D4: renew_subscription_timeout_enable (gsd-print-notifications-manager.c:1192) ==13846== by 0x231265D4: gsd_print_notifications_manager_start_idle (gsd-print-notifications-manager.c:1328) ==13846== by 0x77CC799: g_main_dispatch (gmain.c:3109) ==13846== by 0x77CC799: g_main_context_dispatch (gmain.c:3708) ==13846== by 0x77CCAE7: g_main_context_iterate.isra.24 (gmain.c:3779) ==13846== by 0x77CCDB9: g_main_loop_run (gmain.c:3973) ==13846== by 0x5A8A044: gtk_main (gtkmain.c:1207) ==13846== by 0x4037C0: main (main.c:427) Please file a separate bug against CUPS for those. If the problem is in the gnome-settings-daemon plugin, the bug will get reassigned. That looks like a bug in gnome-desktop: ==13846== 4,064 (3,840 direct, 224 indirect) bytes in 48 blocks are definitely lost in loss record 14,166 of 14,331 ==13846== at 0x4C29BFD: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==13846== by 0x77D225E: g_malloc (gmem.c:97) ==13846== by 0x77E845D: g_slice_alloc (gslice.c:1007) ==13846== by 0x7802402: g_variant_iter_new (gvariant.c:2912) ==13846== by 0x780430B: g_variant_valist_get_nnp (gvariant.c:4734) ==13846== by 0x780430B: g_variant_valist_get_leaf (gvariant.c:4913) ==13846== by 0x780430B: g_variant_valist_get (gvariant.c:5094) ==13846== by 0x78041C2: g_variant_valist_get (gvariant.c:5129) ==13846== by 0x7805199: g_variant_get_va (gvariant.c:5356) ==13846== by 0x7805581: g_variant_get (gvariant.c:5303) ==13846== by 0x56774A4: crtc_initialize (gnome-rr.c:1825) ==13846== by 0x56774A4: fill_screen_info_from_resources (gnome-rr.c:399) ==13846== by 0x56774A4: fill_out_screen_info (gnome-rr.c:439) ==13846== by 0x56774A4: screen_info_new (gnome-rr.c:456) ==13846== by 0x5677BBE: on_proxy_acquired (gnome-rr.c:624) ==13846== by 0x6681F46: g_simple_async_result_complete (gsimpleasyncresult.c:763) ==13846== by 0x6681FA8: complete_in_idle_cb (gsimpleasyncresult.c:775) Again please file a separate bug against the gnome-desktop3 package. There's also a fair number of gdbus and dbus-glib calls without cleanups, but there's not enough data to find the root causes. I don't think that whatever function it is that leaks had time to run in your tests. You might want to try disabling the "housekeeping" plugin with: gsettings set org.gnome.settings-daemon.plugins.housekeeping active false and see whether you can reproduce the problem. |
Created attachment 1141277 [details] memory usage chart over time Description of problem: - memory leak observed for gnome-settings-daemon Version-Release number of selected component (if applicable): - currentrelease How reproducible: - run gnome Steps to Reproduce: 1. 2. 3. Actual results: - slow memory leak Expected results: Additional info: - See attachments showing slow memory usage increase