Hi,

> -----Original Message-----
> Okay, then could you please check which part is really required?
> 
> - protection by _cpu_exception_suspend()/_cpu_exception_resume() pairs
>   on accesses to TLB settings in mmu*.c

Say this is (a)

> - protection by splvm()/splx() pairs on accesses to PTE settings in
>   pmap.c

(b)

> - passing protection bits to flags too on pmap_enter(9) in
>   vm_machdep.c:vmapbuf()

(c)


<nag>
The reboots should also happen on your site if you remove the
patches. So you could also check yourself :-)
</nag>


Am I correct, that for performance reasons, it would be best if (c),
else (b), else (a) fixes the problem?

Then I will try the patches in the following order:

(c) applied
(b) applied
(a) applied
(c) and (b) applied
(c) and (a) applied
(b) and (a) applied

I will test by unpacking the distribution tarballs.

Without the patches the reboots happened most of the time when
unpacking base.tgz. I had one occurrence of a successful extraction of
base.tgz, but then the reboot happened while extracting comp.tgz.

If someone else also wants to help testing I suggest to start from the
end of the list. Only unpack base.tgz and comp.tgz and if this works,
assume the combination is working and go to the next one.

I will send a short notification to the list for each of my test runs.

I'm using the netbsd-2-0 branch from a week ago or so.

Maybe someone joins in....  :-)

regards,
chris