valgrind invalid reads error on unloading

[Sebastien Bacher]

Using the current natty version, valgrind display those service errors when the running a GNOME session and unloading the indicator-applet from the configuration

"==16099== Invalid read of size 1
==16099== at 0x402734E: strcmp (mc_replace_strmem.c:426)
==16099== by 0x424E703: g_str_equal (gstring.c:116)
==16099== by 0x42191B1: g_hash_table_remove_internal (ghash.c:330)
==16099== by 0x4048E16: unwatch_core (indicator-service.c:558)
==16099== by 0x4101B41: actually_do_call (gdbusnamewatching.c:166)
==16099== by 0x4101C36: do_call (gdbusnamewatching.c:215)
==16099== by 0x4101E49: on_name_owner_changed (gdbusnamewatching.c:305)
==16099== by 0x40F126F: emit_signal_instance_in_idle_cb (gdbusconnection.c:3400)
==16099== by 0x4226460: g_idle_dispatch (gmain.c:4536)
==16099== by 0x422AC17: g_main_context_dispatch (gmain.c:2440)
==16099== by 0x422B3DF: g_main_context_iterate.clone.5 (gmain.c:3091)
==16099== by 0x422BAA2: g_main_loop_run (gmain.c:3299)
==16099== by 0x804DADA: main (messages-service.c:1474)
==16099== Address 0x50d55e1 is 1 bytes inside a block of size 8 free'd
==16099== at 0x40259E0: free (vg_replace_malloc.c:366)
==16099== by 0x4231DE5: g_free (gmem.c:263)
==16099== by 0x410151F: client_unref (gdbusnamewatching.c:105)
==16099== by 0x4048E04: unwatch_core (indicator-service.c:413)
==16099== by 0x4101B41: actually_do_call (gdbusnamewatching.c:166)
==16099== by 0x4101C36: do_call (gdbusnamewatching.c:215)
==16099== by 0x4101E49: on_name_owner_changed (gdbusnamewatching.c:305)
==16099== by 0x40F126F: emit_signal_instance_in_idle_cb (gdbusconnection.c:3400)
==16099== by 0x4226460: g_idle_dispatch (gmain.c:4536)
==16099== by 0x422AC17: g_main_context_dispatch (gmain.c:2440)
==16099== by 0x422B3DF: g_main_context_iterate.clone.5 (gmain.c:3091)
==16099== by 0x422BAA2: g_main_loop_run (gmain.c:3299)
==16099== by 0x804DADA: main (messages-service.c:1474)"

[Sebastien Bacher (seb128)]

Setting as low since unloading is not a frequent usecase, still would be nice to fix those errors

[Sebastien Bacher (seb128)]

ok, seems to not been happening with the current versions, let's close it

[Sebastien Bacher (seb128)]

in fact it's still an issue, getting it while running indicator-datetime under valgrind and restarting the unity-panel-service

[Charles Kerr (charlesk)]

The valgrind log there looks more like a glib error than a libindicator one.

I'm not certain about this, but this <https://bugzilla.gnome.org/show_bug.cgi?id=650459> glib bug from May 2011 seems similar -- is anyone still seeing this issue in 11.10 or 12.04?

[Charles Kerr (charlesk)]

Ignore my last comment. I'm glad I said "I'm not certain about this" because it was wrong. :)

Fixed by ted in http://bazaar.launchpad.net/~indicator-applet-developers/libindicator/trunk.0.5/revision/455

[Launchpad Janitor (janitor)]

This bug was fixed in the package libindicator - 0.4.93-0ubuntu1

---------------
libindicator (0.4.93-0ubuntu1) precise; urgency=low

  * New upstream release.
    * Fix branching past initialization of variables (LP: #939061)
    * Support new FD.o Desktop Actions (LP: #942042)
    * Fix valgrind error on unload (LP: #719457)
 -- Ted Gould <email address hidden> Wed, 07 Mar 2012 14:14:07 -0600