:I tried changing the number of CPUs on vkernel from 16 to 4 and reduced
:the -j number given to make command, then I get a panic in a different place:
That's a better panic, for sure. Here is another patch to try:
fetch http://apollo.backplane.com/DFlyMisc/pmap01.patch
* I fixed one place where pmap->pm_generation was not being bumped.
* I fixed another place where it was looping on the wrong npv
under certain circumstances (but it also kprintf()'s something
in that case and your crash dump didn't have the kprintf).
* I added a few more KKASSERT()s closer to the pmap lookup.
* I believe I found a race in pmap_enter()'s call to pmap_remove_pte().
The temporary ptbase can get wiped after the call so we have to reload
the pte pointer.
I'm crossing my fingers. If this fixes the problem I will make the
same changes to the x86_64 and vkernel's pmap code.
-Matt