Re: [cc65] Running C program in a cart

From: "Payton Byrd"; on Friday, May 25, 2012; at 05:28 PM -0400
>
> > On Fri, May 25, 2012 at 4:15 PM, Ullrich von Bassewitz
> >
> > On Fri, May 25, 2012 at 03:28:38PM -0500, Payton Byrd wrote:
> > > Now, the cart compiles and loads into VICE; and, when it's loaded
> > > as a Cart, the screen and border turn white, which is as expected
> > > based on the code in the header.  The compiled binary then drops
> > > straight into the CC65 initialization code that is normally at
> > > 2061, as you would expect.  The problem is that somewhere during
> > > initialization the system goes into an infinite loop of displaying
> > > garbage on the screen.
> >
> > I don't know much about C64 carts, but aren't they enabled by
> > modifying $01?  If so, the startup code will probably disable
> > the cart from where the code is loaded because it modifies $01
> > to switch off the BASIC ROM.
>
> Yes, this is my initial thought as well.  I've added code before
> the cc65 init to copy all of the memory from $8000 to $BFFF to RAM
> before performing the init; so, when the cart is disabled, the correct
> code still should be in place.  However, that does not change the
> behavior of the program.
>
> Here is my current header and config. ...
>
> . . . .

The cart header can be written as:

    .segment "ROMLA"

    .addr cold_start
    .addr cold_start
    .byte "CBM80"

cold_start:
    lda #$01
    sta $d020
    sta $d021

The signature text will be translated into proper PetSCII.

>
>     DATA:     load = RAM, type = rw, define = yes;

As Groepaz said, that should be
    DATA:     load = ROM1, run = RAM, type = rw, define = yes;

OK, with those things out of the way, I can give you the bad news:  The
cartridge launcher doesn't initiate the computer -- the cart must do
that job!  Look at a disassembly of the Kernal.  The reset-code, at
$FCE2, sets the stack pointer, then looks for a cart.  If it finds one,
it immediately jumps to that cart.  Only if there is no cart does the
Kernal do a sequence of subroutine-calls that initiates the system.  So,
your cart header must duplicate that list of JSRs.  Therefore, the
header must look like this:

CINT   = $ff81
IOINIT = $ff84
RAMTAS = $ff87
RESTOR = $ff8a

    .segment "ROMLA"

    .addr cold_start
    .addr cold_start
    .byte "CBM80"

cold_start:
    jsr IOINIT
    jsr RAMTAS
    jsr RESTOR
    jsr CINT
    cli

    lda #$01
    sta $d020
    sta $d021

; memcpy code ...

A companion problem is that BASIC doesn't start your code; therefore,
the start-up module mustn't return to BASIC.  So, instead of
"jmp $8200", the header must do:

warm_start:
    jsr startup
    jmp warm_start

startup:
    .end

----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo@musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.