MMU changes wasRe: [Coco] Re: Coco Repack

[jdaggett at gate.net]

Willard

I  have no plans to change from the 8K pages. And yes I have t 
hought of ways to increase beyond two tasks. Four  would be nice. 
Four would also require another 16 bytes or a means of multiplexing 
the existing 16 bytes. 

james

On 10 Aug 2004 at 3:13, Willard Goosey wrote:

Date sent:      	Tue, 10 Aug 2004 03:13:52 -0600
From:           	Willard Goosey <goosey at virgo.sdc.org>
To:             	coco at maltedmedia.com
Subject:        	MMU changes wasRe: [Coco] Re: Coco 
Repack
Send reply to:  	CoCoList for Color Computer Enthusiasts 
<coco at maltedmedia.com>
	<mailto:coco-
request at maltedmedia.com?subject=unsubscribe>
	<mailto:coco-
request at maltedmedia.com?subject=subscribe>


> >From: KnudsenMJ at aol.com

> >Date: Mon, 9 Aug 2004 23:24:34 EDT

> >

> >In a message dated 8/8/04 1:55:34 PM Eastern Daylight Time,

> >jdaggett at gate.net writes:

> >

> >> Yes you can establish page size to any size block of memory that

> >> you want. 

> >1K 

> 

> Rather than changing the block size of the GIME's MMU, I'd add more

> contexts.  It only has two.  4 would be cool, more even better.

> 

> IIRC, to print a character on the screen, an OS-9 program goes through

> 3 context switchs: user program -> kernel -> grfdrv.  The first

> task-switch is nice and quick, the kernel lives in context #0 and the

> user process in #1.  But that third switch has to save the user

> context somewhere and load grfdrv's.  Then it has to switch back.

> 

> >Maybe that would be OK.  But given the number of good apps (and OS-9

> >as a whole) written with 8K blocks, do we really need to screw with

> >this now?

> > --Mike K.

> 

> Gotta agree.  Even my idea would probably break lots of software. :-(

> 

> Willard

> -- 

> Willard Goosey  goosey at sdc.org

> Socorro, New Mexico, USA

> "I've never been to Contempt!  Isn't that somewhere in New Mexico?"

>    --- Yacko

> 

> -- 

> Coco mailing list

> Coco at maltedmedia.com

> http://five.pairlist.net/mailman/listinfo/coco