[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] 040/060 MMU tree builder (fwd)



 Ok, I managed to mess with the mail addy...

---------- Forwarded message ----------
Date: Fri, 21 May 1999 21:44:40 +0000
From: Odd Skancke <oskancke@online.no>
To: mint@fishpoo.com
Subject: Re: [MiNT] 040/060 MMU tree builder

On Thu, 20 May 1999, Michael Schwingen wrote:

> On Wed, May 19, 1999 at 11:18:21PM +0000, Odd Skancke wrote:
> > 
> > up until then anyway.. hmm.. trying to set the mmu up before a warm boot
> > is another thing I haven't tested....
> 
> TOS *does* re-initialize all hardware (including the memory controller) on a
> warm boot, so I don't see why it should not re-initialize the PMMU, too.

 Well, I agree that all hardware should be reset, ofcourse, and the Hades
TOS does that. But I thought that the memory was kept intact trough a warm
boot. By this, I mean that the memory layout/mapping should not be changed
by a warm-boot, since some things are resident, and would thus fail if
memory mapping changed. And since TOS does allow for resident programs to
survive a warm boot, the pmmu must not be changed during them. I might be
on thin ice here, and this is more a question than anything :) I honestly
thought this was the major difference between a cold and warm boot.

> 
> The original TOS 3.06 *does* turn off and re-initialize the PMMU on every
> reset, although I am not sure now how that was modified on the hades.

 To be absolutely honest, I don't know if  the pmmu is reset by TOS on
warm boots, as I set up the tree after splitram.prg. I still haven't
tested this out yet, but if it doesn't, it would be much easier setting up
ram as I wish.

> 
> possible to use one "layout"
> 
> No. It is the ROMs task to get the system up to a stable state, no matter if
> it is a warm or a cold boot, and this includes setting up a good PMMU
> mapping if necessary on the machine.

 Yes, this would be the absolutely best solution. I guess Milan TOS lets
you configure this to some extent?

> 
> This may be bad for you if you want to patch it, but IMHO this is
> preferrable if the TOS *does* init the PMMU the right way.
> 

Yes, agreed. But since it doesn't, I'm looking for other ways to do it :)

> > until the next cold-boot. I mean, during a cold boot, everything hardware
> > related is set up, and the memory map is created. Then on sucessive
> > warm-boots, the current "hardware" configuration should be preserved. Does
> > this sound reasonable?
> 
> No. The only diffeence between cold and warm start is that the memory test
> and size detection is skipped on warm starts. I want the system to come up
> clean after a reset, and a reset is not meant as a way to half-restart the
> OS after patching around.

 I thought it kinda was, since it lets resident stuff survive a warm-boot,
it should not alter the current memory-map either.. I think. But, as I
said before, I don't really know :)

 I don't mean that hardware should not reset, as it always should be. I
only mean the memory layout or mapping.




 Odd Skancke - ozk@atari.org