Re: [MiNT] FreeMiNT MMU usage define

[xavier.joubert@free.fr]

Hello,

----- "Jean-François Lemaire" <jflemaire@skynet.be> a écrit :
> On Tuesday 14 Jun 2011 22:45:30 Jo Even Skarstein wrote:
> > This is also the case with the Afterburner and the Hades (and maybe
> the
> > CT60?)
> Indeed it is. And MP is working fine despite this.

In fact, no. The CT60 does not need any special MMU setup.

When we designed the CT60 memory map with Rodolphe Czuba and Didier Mequignon, we had the CT2 in mind. This (68 030) board needed some MMU tricks to get a Falcon compatible memory map. The TT-Ram was not located at 0x1000000 ; the low memory access was not supervisor only, etc. This gave problems like faulty programs running fine on CT2 and crashing on a stock Falcon. It was a nightmare for programmers with CT2 as they weren't able to see their bugs.

And the Centek developpers choose to use the MMU to hide their own setup. So it was impossible to look at their firmware, or at the MMU tree without modifying the sait MMU tree... And it was impossible to setup a correct MMU tree without a deep knowledge of the CT2 memory map... Quite tricky but not impossible, as I debbuged the CT60 initialization code, and modified the Falcon TOS on my CT2.

So we decided to implement everything needed to have a full Falcon memory map in hardware. Rodolphe did the hard work to put this in the logic while keeping the performance very good.

So the MMU setup used by the CT60 TOS is the same as the one used by the original Falcon TOS... with some adaptations for the 060 MMU, of course.

Regards,

Xavier