The following happened (twice in a row before taking the
previous kernel) on my NetBSD/mac68k with a kernel built of
today's sup (31 Aug).

The kernel was config'ed with a fresh config followed by
make depend&&make, but no make clean--but I did not recall
anything from the cvs commit messages calling explicitly
for a complete rebuild, but I'll do that overnight...

...
root file system type: ffs
panic: pool_get: must have NOWAIT

The traceback is:

panic
pool_get(e345c,2,332000,333000,df990)+22
uvm_map_clip_start(df98c,793078,332000)+24
uvm_unmap_remove(df98c,332000,333000,1,79adbc,df990,2)+88
uvm_km_free_wakeup
pmap_remove_mapping
pmap_remove
uvm_unmap_remove
uvm_km_free_wakeup
pmap_release
pmap_destroy
uvmspace_free
cpu_exit
exit1
sys_exit
syscall(1) + de
trap0()+e

If this isn't immediately recognized by some of the central people playing
with (u)vm and the memory pools and if the problem persists over a couple
of days, I'll of course send a PR.

- Erik