GNOME applications, like Videos (Totem), do not start with nvidia driver


GNOME applications seem to not start with nvidia driver. Particularly Videos (Totem), and I also suspect Cheese is also the case. The application neither start from the desktop environment nor command line.

Totem developers, based on GNOME/Videos stacktrace, suggest it is a bug in nvidia libGL implementation, could you please verify that?

GNOME/Videos stacktrace used:
Thread 1 (Thread 0x7ffff7fb0a40 (LWP 3521)):
#0 0x000000318280b17e in __pthread_mutex_unlock_usercnt (decr=1,
mutex=0x3182220908 <_rtld_local+2312>) at pthread_mutex_unlock.c:57
#1 __GI___pthread_mutex_unlock (mutex=0x3182220908 <_rtld_local+2312>)
at pthread_mutex_unlock.c:310
#2 0x0000003182001099 in tls_get_addr_tail (ti=0x3184620f58,
dtv=0x7ffff7fb1350, the_map=0x7ffff7fc8560) at dl-tls.c:730
#3 0x0000003184410d0e in getprocattrcon_raw () from /lib64/
#4 0x000000318440aa6e in is_selinux_enabled () from /lib64/
#5 0x000000318aca5da6 in ?? () from /usr/lib64/nvidia-304xx/
#6 0x000000318ac8505b in ?? () from /usr/lib64/nvidia-304xx/
#7 0x000000318200f29d in call_init (l=0x7ffff7fbf4e0, argc=argc@entry=1,
argv=argv@entry=0x7fffffffe218, env=env@entry=0x7fffffffe228)
at dl-init.c:67
#8 0x000000318200f3c3 in call_init (env=,
argv=, argc=, l=)
at dl-init.c:34
#9 _dl_init (main_map=0x3182221168, argc=1, argv=0x7fffffffe218,
env=0x7fffffffe228) at dl-init.c:130
#10 0x000000318200122a in _dl_start_user () from /lib64/
#11 0x0000000000000001 in ?? ()
#12 0x00007fffffffe509 in ?? ()
—Type to continue, or q to quit—
#13 0x0000000000000000 in ?? ()

Related bugreport can be found in

Thank you for any help.

nvidia-bug-report.log.gz (73.1 KB)

Thanks for reporting this. It’s being tracked in bug 1357836.

Thank you for paying attention. I am sorry I couldn’t be more helpful.

I am really sorry I have just realised that I have got an incorrect (legacy) driver installed (it is also shown in the bug report log). After “normal” driver has been installed the GNOME applications work/start well.

I am sorry once more.

Thanks, peartown. This is actually an issue that was debugged and fixed in later releases, but we’re looking into the feasibility of backporting the fix to the legacy branch.

Thank you for explanation. IMHO that would be good news for the users of legacy driver(s).