I've being looking at kern_fork.c and kern_exit.c wrt fiddling
with the p_hash lings....

In my travels I've noticed some possible locking anomolies.
It looks as though the code in current is largely MP clean.

In particular the main process lists (via p_hash and p_link)
seem ok.

However there doesn't seem to be any locking of the process
p_sibling chain, p_pglist nor of some other minor items [1].

Has someone missed these? or is the MP code only partially
merged?

	David

[1] p2->p_limit->p_refcnt++ (kern_fork.c line 402)
--p->p_cred->p_refcnt	(kern_exit.c line 510)
nprocs--		(kern_exit.c line 510)

-- 
David Laight: david@l8s.co.uk