Subject: Re: Are there any known problems with uvm or pmap in -current as of about Oct. 16?
To: Markus W Kilbinger <kilbi@rad.rwth-aachen.de>
From: Chuck Silvers <chuq@chuq.com>
List: port-macppc
Date: 11/07/2001 10:12:41
yea, that looks like it.  the manual appears to be wrong.

the old pmap looks more like the "general case" section right before that.
we should note that the section 7.6.3.2.2 code is only for clearing
the referenced bit, not the modified bit "because the R bit need not
be maintained exactly."  if we want to optimize pmap_clear_reference()
to use this faster algorithm, it can't share code with pmap_clear_modify().

-Chuck


On Wed, Nov 07, 2001 at 12:30:36PM -0500, Allen Briggs wrote:
> On Wed, Nov 07, 2001 at 09:10:28AM -0800, Chuck Silvers wrote:
> > I don't know what document that reference about "As shown in Section
> > 7.6.3.2.2" comes from, is that document wrong or did you just mis-read
> > it, matt?
> 
> That looks to be from the 6xx Programming Environments Manual (PEM), pdf
> page 333 / 7-65 / section "7.6.3.2.2 Clearing the Referenced (R) Bit":
> 
> 	lock(PTE)
> 	oldR <-- PTE[R]
> 	if oldR = 1, then
> 		PTE[R] <-- 0
> 		tlbie(PTE)
> 		eieio
> 		tlbsync
> 		sync
> 	unlock(PTE)
> 
> -allen
> 
> -- 
>  Allen Briggs                     briggs@wasabisystems.com
>  http://www.wasabisystems.com/    Quality NetBSD CDs, Sales, Support, Service
> NetBSD dev. for _your_ Alpha, ARM, M68K, MIPS, PowerPC, SH3, Sparc, x86, etc...