On Thu, Mar 30, 2006, Claus Assmann wrote: > application behaves fine. Hence I'm wondering whether there is a > memory leak in some library or the OS, which also could be triggered > by the way my application uses it (see the recent thread about
Thanks to all of you who offered suggestion how to track down this problem! Summary: it was a bug in my application that showed up only on OpenBSD as it is the only OS of those four I tried that use malloc() in pthread_mutex_init(). There was one place in one of my libraries where pthread_mutex_destroy() was not called when a struct was freed that contained a mutex. Special thanks to Anthony Howe who gave me his debugging malloc replacement that "intercepts" the libc version, Ted Unangst for pointing out that OpenBSD uses malloc() for mutexes, and Hannah Schroeter for explaining her approach to the problem (I added those write() calls to Anthony's program and wrote a simple awk script to analyze the output and print those pointers that haven't been free()d.) Thanks again to everyone who answered, your help is much appreciated!
