hmm. So the trap is always E (page fault), but is it always 9? It always says page fault: 9?
Usually when I've had this problem, which you seem to be getting some time after processes are running, it's been problems with caches not being flushed correctly after context switch, and thus processes returning from kernel mode with a stack which is some other procs stack. Or the tlb flush failing which can result in similar issues. I'm hardly the expert here however. Any chance you can get this on serial console? It might help. thanks ron
