Red Hat Bugzilla – Bug 205731
gam_server segfault in gam_queue_event
Last modified: 2007-11-30 17:11:42 EST
On my freshly updated Rawhide VM, found attached core file from
/usr/libexec/gam_server. This is gamin 0.1.7-6.fc6.
#0 0x08050eda in gam_queue_event (conn=0x1, reqno=7, event=32,
path=0x8bb8368 "fedora-soundconverter.desktop;45012a5b", len=38) at
#1 0x0804b3b9 in gam_server_emit_one_event (
node_is_dir=1, event=GAMIN_EVENT_CREATED, sub=0x8bb0ec0, force=1) at
#2 0x08052a47 in gam_inotify_event_callback (
subdata=0x8bb0ec0) at gam_inotify.c:135
#3 0x08053310 in ih_event_callback (event=0x8bb5340, sub=0x8bb0e50) at
#4 0x08055433 in ip_event_dispatch (dir_list=0x8bae520, pair_dir_list=0x0,
#5 0x080555a3 in ip_event_callback (event=0x8bb5340) at inotify-path.c:375
#6 0x08054563 in ik_process_eq_callback (user_data=0x0) at inotify-kernel.c:673
#7 0x0093c916 in g_source_get_current_time () from /lib/libglib-2.0.so.0
#8 0x0093c342 in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#9 0x0093f31f in g_main_context_check () from /lib/libglib-2.0.so.0
#10 0x0093f6c9 in g_main_loop_run () from /lib/libglib-2.0.so.0
#11 0x0804bafd in main (argc=1, argv=0xbfe1ee24) at gam_server.c:647
#12 0x0077ff2c in __libc_start_main () from /lib/libc.so.6
#13 0x0804a311 in _start ()
The path argument ("soundconverter.desktop") was from a built RPM I was
installing at the time. Will try to reproduce and post more info.
Created attachment 135831 [details]
Core file from /usr/libexec/gam_server
Seems inotify related. John, got any idea about this one?
It fails on the line:
But it doesn't actually call into the assert, so i guess the problem is
dereferenceing conn. Maybe we can actually believe the conn=0x1 part from gdb.
Actually, it seems like sub->listener points to some bogus memory area in
gam_server_emit_one_event(). (Although sub itself seems to be ok):
running gam_server in valgrind immediately turned up this:
gchar *ih_sub_get_filename (gchar *pathname)
- gchar *out;
- // FIXME: return filename here
- return out;
+ return g_path_get_basename (pathname);
I dunno if that is the cause of the problem though.
There were some other bugs too. I fixed them in upstream cvs and in
This fixed a lot more of the gamin test suite too...
Thanks for the quick response.
Wow, I totally missed that when I committed. Thanks for the quick fix!