Re: C64 MMU POC

On 9/30/2019 4:21 PM, Nejat Dilek wrote:
> On Mon, Sep 30, 2019 at 11:08 PM Jim Brain <brain_at_jbrain.com> wrote:
>> Special timed writes/reads are fine to unlock the register space, but it
>> would not make sense to set all of the registers in this MMU via that
>> mechanism, due to the number of addresses to change and the frequency of
>> changing them.
>>
> You could have a state machine where you only react to command writes
> when the state machine is in unlocked state otherwise ignore them.
I intend to.
> After the unlocking, calling program may start issuing commands to the
> registers of the hardware. You could have a control register to
> provide shut off mechanism for the calling program too when in
> unlocked state.
THis is the plan.  The question was where to put the registers when 
unlocked.
>
> By the way you don't have to deal with bit transfers to use such a
> unlocking mechanism, in my case I was constrained to 1 bit input so I
> designed a bit of time taking procedure for the unlocking mechanism.
> If you can capture full byte writes then the whole process could
> complete in just 30-40 cycles too. Of course then you need a litle bit
> of memory buffer in the hardware. Say, getting 8 consecutive writes of
> specific data like "TURN ON!" to a specified address.

I doubt I'll use writes.  In some of my other projects, I use reads from 
locations in a specific 256 byte address space.  The low byte of the 
address is used as the "magic" value.  So, to unlock using 
$55,$aa,$ff,$00,$c0, for example::

lda $c055

lda $c0aa

lda $c0ff

lda $c000

lda $c0c0

THe advantage is that the operation is "safe" at least in RAM space (a 
read in IO space is not always safe, but it always safe in RAM)

Jim