TDB mutex support breaks CTDB

[Jeremy Allison]

On Wed, Dec 17, 2014 at 10:51:57AM +0100, Stefan (metze) Metzmacher wrote:
> Am 17.12.2014 um 10:03 schrieb Ralph Böhme:
> > Hi Amitay,
> > 
> > On Wed, Dec 17, 2014 at 07:48:33PM +1100, Amitay Isaacs wrote:
> >> I am trying to test samba+ctdb with these changes using nbench torture
> >> test, however smbd is crashing in brlock.  I am not sure if this is related
> >> or something else is going wrong.  I have attached gdb backtrace.  If you
> >> have any ideas on how to debug this, please let me know.
> > 
> > hey, I just ran into this just yesterday too, sounds like the same
> > issue. gdb:
> > 
> > #6  <signal handler called>
> > No symbol table info available.
> > #7  0x00007fcfdc8120b5 in brl_conflict_other (lock=0x0, rw_probe=0x7fffab1a3300) at ../source3/locking/brlock.c:233
> > No locals.
> > #8  0x00007fcfdc814398 in brl_locktest (br_lck=0x7fcfde974870, rw_probe=0x7fffab1a3300) at ../source3/locking/brlock.c:1385
> >         ret = true
> >         i = 0
> >         locks = 0x0
> >         fsp = 0x7fcfde974a20
> >         __FUNCTION__ = "brl_locktest"
> > #9  0x00007fcfdc80e80a in strict_lock_default (fsp=0x7fcfde974a20, plock=0x7fffab1a3300) at ../source3/locking/locking.c:139
> >         br_lck = 0x7fcfde974870
> >         strict_locking = 2
> >         ret = false
> >         __FUNCTION__ = "strict_lock_default"
> > #10 0x00007fcfdc86cd7c in vfswrap_strict_lock (handle=0x7fcfde967060, fsp=0x7fcfde974a20, plock=0x7fffab1a3300) at ../source3/modules/vfs_default.c:2293
> >         __FUNCTION__ = "vfswrap_strict_lock"
> > 
> > For some reason locks is NULL. My band aid wad disabling strict locking.
> 
> I think I found the problem, it's regression introduced with the leases
> code.
> 
> A possible fix is attached...

LGTM. Pushed.

Is there any comment showing a cluster config can call
this will an uninitialized entry ? It seems rather unusual..