This looks quite a bit like the error I had recently
which you can see here in the archives:
http://mailman.mit.edu/pipermail/ker...ry/012940.html
my solution was to not include the krb5support lib in the link. YMMV.

John


> #0 0xb7f7a402 in ?? ()
> #1 0x00b86c00 in raise () from /lib/libc.so.6
> #2 0x00b88451 in abort () from /lib/libc.so.6
> #3 0x00b8024b in __assert_fail () from /lib/libc.so.6
> #4 0x00878e7d in krb5int_init_fac () from /usr/lib/libkrb5support.so.
> 0
> #5 0x00876c4f in krb5int_pthread_loaded () from /usr/lib/
> libkrb5support.so.0
> #6 0x00876d9b in krb5int_call_thread_support_init ()
> from /usr/lib/libkrb5support.so.0
> #7 0x0087a039 in krb5int_set_error_info_callout_fn ()
> from /usr/lib/libkrb5support.so.0
> #8 0x007f127a in krb5int_cleanup_library () from /usr/lib/libkrb5.so.
> 3
> #9 0x007f143b in krb5int_initialize_library () from /usr/lib/
> libkrb5.so.3
> #10 0x00836ae8 in krb5_set_default_in_tkt_ktypes () from /usr/lib/
> libkrb5.so.3
> #11 0xb7ecfeb8 in pam_sm_open_session () from /lib/security/
> pam_krb5.so
> #12 0xb7ecdfce in pam_sm_open_session () from /lib/security/
> pam_krb5.so
> #13 0x007ad42e in _pam_dispatch () from /lib/libpam.so.0
> #14 0xb7f75796 in ?? () from /lib/security/pam_stack.so
> #15 0x007ad42e in _pam_dispatch () from /lib/libpam.so.0
> #16 0x007b08b3 in pam_open_session () from /lib/libpam.so.0
> #17 0x0806fef8 in do_pam_session () at auth-pam.c:1048
> #18 0x0805456e in main (ac=5, av=Variable "av" is not available.
> ) at sshd.c:1899
>
>
> On Jan 22, 4:01 pm, Ken Raeburn wrote:
> > On Jan 22, 2008, at 15:40, psulliva...@gmail.com wrote:
> >
> > > Has anyone seen this error before:

> >
> > > sshd: ../../include/k5-thread.h:704: k5_mutex_finish_init_1: Assertion
> > > `(&(&m->os)->n)->initialized != K5_MUTEX_DEBUG_INITIALIZED' failed.

> >
> > > We've encountered this error on RHEL 5 using MIT Kerberos and Openssh
> > > 4.7p1.

> >
> > Looks like some bit of code is trying to initialize a mutex (with the
> > krb5 wrapper functions) that's already been initialized, and not
> > destroyed in between. Can you get a stack trace?
> >
> > Ken

>
> ________________________________________________
> Kerberos mailing list Kerberos@mit.edu
> https://mailman.mit.edu/mailman/listinfo/kerberos
>