Hello, I was lucky to have a reproducer for the crash crash. It happens that it was a mere error in the call of proc_getprocinfo, thrashing the stack, and thus asking for a core dump to itself... and thus letting aborted processes indefinitely wait. There were other calls in other places, I believe that was at least one of the reasons for the ps command to crash sometimes.
BTW, this was all debugged actually very easily with mere gdb: a breakpoint, backtrace, step-by-step execution, etc. Samuel
