Hi, all.

Is PR_URGENT (in subr_pool.c) is really necessary?  (PR_URGENT is like
PR_NOWAIT but it causes panic()).

Yesterday, I got panic ``pool_get: urgent''.  Manually copied
stacktrace is attatched below.  It looks like page allocation for
items succeeded but failed to allocate (separate) page header.

But instead of using PR_URGENT for page header allocation, why we
can't wait or return failure according to the flags passed to allocate
items?

enami.

panic: pool_get: urgent
_pool_get + 0xd3
pool_prime_page + 0x79
_pool_get + 0x115
pool_cache_get + 0x3e
pmap_create + 0xaa
uvmspace_init + 0x45
....