Stephen Jones <smj@cirr.com> writes:
> Are you saying that just by using the newer version of trap.c
> you no longer have lock_mgr locking on itself panics and vnlock
> dead locks?
For my particular test case, yes.  However, I don't recall ever
seeing vnlock dead locks when running the test program.  Perhaps
we are seeing two different bugs?

FWIW, I even compiled apache2, with the --with-mpm=worker
settings instead of the default --with-mpm=prefork, and I could
run the web server as

  env PTHREAD_CONCURRENCY=2 /usr/pkg/etc/rc.d/apache forcestart

and at times see 10s of LWPs per apache process.  So, from my
perspective the new trap.c is a giant step in the right
direction.
					-jarle
-- 
"True happiness is when the Programmer admits what the Beta Tester has known
 all along. Bugs are romping in the code."
					-- Marge Robbins