Any known issues or experiences where a byte or 2 bytes get corrupted
randomly at different addresses but always on a 4k page boundary. This
happens in a different user process and the address where the
corruption happens is in the heap - malloc'd area, malloc using mmap's
of pagesize which is 4k (NetBSD-1.6.1 - ppc750 -1g of physical memory
- no swap). We have scoured over application code/library code etc. no
clues yet to point to anything. The bug always manifests itself on a
4k page boundary and a byte or two gets whacked after a day or two of
testing.

We have tried various methods mprotect, logging allocations, function
call trace etc..dont seem to lead to any results since any
instrumentation makes it hard to reproduce.

Appreciate any reply's thanks in advance.

-rahul