fix for remrq from Mike

p_regs moves to p_md.md_regs

avoid tracing panics at exit

fix to previous changes

added ifdef PSL_ALLCC around old wait system call code.

leave curproc set until last sleep, then clear before state is invalid
(so clock, etc. don't poke around stuff that's been freed)

unmap (user) address space early enough to allow sleep (if releasing
unlinked files); have to keep credentials until wait(); cpu_wait may be needed
on other architectures (no #ifdef i386)

two minor fixes for 386; others unaffected

mv vmspace_free and kmem_free of kernel stack to cpu_exit;
marc's fix to do pgsignal on exit of session leader only if still attached

rm user.h; rm unneeded proc flags; noproc is now !curproc

more-or-less working with new proc & user structs

add dynamically allocated file descriptors

update for new VM

fix jobc adjustments so we increment before decrementing, avoiding spurious
signals when we join parent's pgrp; move common code into subr (from torek)

from chris: exit doesn't return

new copyright notice

RETURN => return, remove syscontext.h

expunge p_cursig, using p_xstat instead for stop signals;
expunge u_code

pass proc pointer to acct()

update to new system call convention; expunge more u.u_procp

only HUP processes with ctty on session leader exit

fix posix semantics for exit of controlling process

merge in support for hp300 from Utah

no longer have per user quota structures

[us]time's maintained in proc. copy to rusage as necessary