Does anyone want to look into this for next68k?

Darrin

------- Start of forwarded message -------
	by dbj.ne.mediaone.net with POP3 (fetchmail-4.7.5)
	for dbj@localhost (single-drop); Mon, 06 Sep 1999 15:52:36 -0400 (EDT)
  by redmail.netbsd.org with SMTP; 6 Sep 1999 19:20:08 -0000
	by chuq.com (8.8.8/8.8.8) id MAA25895
	for tech-kern@netbsd.org; Mon, 6 Sep 1999 12:20:07 -0700 (PDT)
Date: Mon, 6 Sep 1999 12:20:07 -0700
From: Chuck Silvers <chuq@chuq.com>
To: tech-kern@netbsd.org
Subject: PMAP_NEW for all platforms?
Message-ID: <19990906122006.B25692@chuq.chuq.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: tech-kern-owner@netbsd.org
Lines: 27
Xref: dbj.ne.mediaone.net mail.netbsd.tech-kern:5412

hi folks,

one thing that I'd like to get done for 1.5 is to enable PMAP_NEW for all
platforms.  now to really do this right requires a bunch of very platform-
specific work to do all the pre-allocation of page-table pages (or TLB entries
for platforms with just a software-managed TLB) and pretty much requires
having the hardware available to test with, so I'd like to cheat somewhat.
what I'd like to do is create pmap_kenter*() and pmap_kremove() as wrappers
around pmap_enter() and pmap_remove() for the non-PMAP_NEW platforms.
pmap_k*() are really only important for SMP locking issues, and most of the
non-PMAP_NEW platforms don't have MP hardware anyway (and the ones that do
have MP hardware need much more work on their pmaps before they'd be MP-ready).

the point of this exercise is to get all platforms using the same pmap API
(or at least closer to the same API).  having multiple versions of this API
present in the tree for extended periods of time seems to me to be a bad idea,
so I'd like to fix that.

any objections?  if not, I'll commit this stuff next weekend.
(by the way, I've only tested these changes on i386, alpha, sparc, mac68k
sun3 and vax since that's all the hardware I have.  if other people could
apply http://www.chuq.com/netbsd/diff.byebyepmapnew and test on other
non-PMAP_NEW platforms, that would be great.)

-Chuck


------- End of forwarded message -------