On Wed, 8 Aug 2007, Matthias Scheler wrote:

> > If I understand this correctly, some operating systems' threading 
> > implementations ignore the error of calling pthread_mutex_unlock() on a 
> > data area which is not a properly initialized mutex. But NetBSD doesn't 
> > allow it.
> 
> NetBSD can be convinced to allow that, see "man 3 pthread".

I read through it yesterday and before. And I have missed it. Can someone 
please tell me how?

Yesterday, I also used PTHREAD_DIAGASSERT=Ae and got more details:

mem.c:814: fatal error: RUNTIME_CHECK(((__libc_mutex_unlock(((&lock))) 
== 0) ? 0 : 34) == 0) failed

... if that helps. I didn't post it because I posted:

> #3  0x081606d4 in isc_mem_createx2 (init_max_size=0, target_size=0, 
>     memalloc=0x815e944 <default_memalloc>, 
>     memfree=0x815da18 <default_memfree>, arg=0x0, ctxp=0x81b32e0, flags=2)
>     at mem.c:814


  Jeremy C. Reed