stack problems after task switch has completed...

[dr_watts]

I recently added a test task switch to my kernel, which switches with no problem.  However, after it gets to the new process and has been running, it has difficulty using the stack, and page faults.  Using simics however, I watch it trace the page tables, find the correct entry, and read the exact value I want it to.  What could be going wrong?  I have tried moving the logical stack address around in memory...  (it is an expand down stack loaded at 0x109000...  I was wanting to map that to 0x900000 (9 megs) and it does not agree.  Switching this to other locations does not work, and I have made sure the new page tables are okay.    As a bad result, the interrupt handlers do not work, meaning a page fault is causing a triple fault  instead of being handled.  Any suggestions?