Help needed with a bit of code...

daveysludge · October 2014

My minds gone blank again I'm afraid, I'm having trouble with this routine that loads variables into memory then calls the corresponding routine.

Theres 3 variables which makes it awkward. I'm sure theres an easy way, but I just cant see the wood for the trees at the minute...

Thanks in advance...

ld b,3
	ld hl, cor
	ld de,adds
loop
	ld a,(de)
	ld (caller+1),a
	inc de
	ld a,(de)
	ld (caller+2),a; loads address to be called in de
	inc de

	;ld a,(hl)	;hl stores variables
			;now i need to load the variables into the address stored in de ???

	push af
	push de
	push bc
caller
	call 56003	;routine to be  called 
	pop bc
	pop de
	pop af
	djnz loop
	ret


adds
	defw 58239
	defw 56003
	defw 51173	;start adress of routines

cor 
	defb 2,2,4
	defb 10,10,4
	defb 12,14,6	;variables for routine 3 bytes prior to start

Battle Bunny · October 2014

My first thought is that using CALL ... JP (HL) would be better than poking the call address. That is, calling a fixed point (there's one in the ROM somewhere, which I'll find in a minute) then using HL to redirect to the required address.

There's a JP (HL) at $1715, but it only takes one byte, so it might as well just be in your own code. Still thinking about the variables question ...

--

OK, here's my other thought. Store each sub-routine address followed by its three variables: address, byte, byte, byte. Point HL to the one you want, then:

LD E,(HL) : INC HL : LD D,(HL) : INC HL : EX DE,HL : CALL $1715

then when the sub-routine is entered DE will be pointing to its first variable.

joefish · October 2014

I note you're not PUSHing HL to keep it safe when you do the CALL.

If you need that 'adds' table to point to your routines, why not have them point three bytes earlier? And instead of programming the addresses, just refer to a label for each routine like:

defw routine_1 - 3
defw routine_2 - 3
defw routine_3 - 3

Then you load the address you get from the table into DE, point HL at your numerical data, then do LDI three times to copy three bytes from (HL) to (DE). PUSH everything you want to preserve, then EX HL,DE to swap the pointers over and CALL (HL), since it will now be pointing to the start of your routine. When you POP those values again, DE and HL will be back as they were.

And don't forget, if you run out of address pointers, you can use IX and IY in most places where you use HL, it'll just be a fraction slower. They can be very convenient though to use as table pointers, freeing up DE and HL for different tasks. And EX DE,HL can come in very handy when using one address to look up another. And don't forget you can always PUSH DE and later POP the value into HL instead later, for example.

daveysludge · October 2014

Still flummoxed!

The poking the call routine isn't the problem, that works fine, its the variables, particularly this bit:

ld a,(HL)
ld (address stored in DE (less 3)),a
inc hl
ld a,(HL)
ld (address stored in DE (less 2)),a
inc hl
ld a,(HL)
ld (address stored in DE (less 1)),a
inc hl

I need to load a register with the value of the contents of DE and then load that address with the contents of HL.

I cant seem to find the right command,

AndyC · October 2014

If that's really what you mean, then why not:

 ex hl, de
 ld c,(hl)
 inc hl
 ld b,(hl)
 dec hl ; assuming you need to preserve 'DE'
 ex hl, de
 dec bc
 dec bc
 dec bc
 ld a,(hl)
 ld (bc),a
 inc hl
 inc bc
 ld a,(hl)
 ld (bc),a
 inc hl
 inc bc
 ld a,(hl)
 ld (bc),a
 inc hl

Although that doesn't seem terribly efficient and looks like it would trash your address table (unless there is some in-between code missing). I can't help but feel if it were clearer what you were trying to achieve overall, there'd be a better approach. Calling a parameterized jump block for example.

daveysludge · October 2014

@ AndyC...

I'm basically trying to avoid typing out loads of this...

ld a,10
ld (56000),a
ld a,14
ld (56001),a
ld a,6
ld (56002),a
call 56003

ld a,12
ld (56000),a
ld a,24
ld (56001),a
ld a,3
ld (56002),a
call 56003

;ad infinitum!

Paradigm Shifter · October 2014

Use a macro to save typing? Although jp (hl) is probably what you want.

AndyC · October 2014

 call 56003
 defb 10,14,6
 call 56003
 defb 12,24,3
  
 56003:
 pop hl
 ld de,56000
 ldi
 ldi
 ldi
 push hl

Stack abuse FTW. :-)

daveysludge · October 2014

Sorted!

37 bytes, can anyone improve on it?

ld b,2
	ld hl,dat
loop	ld e,(hl)
	inc hl
	ld d,(hl) 
	inc hl
	ld a,(hl)
	ld (de),a
	inc hl
	inc de
	ld a,(hl)
	ld (de),a
	inc hl
	inc de
	ld a,(hl)
	ld (de),a
	inc hl
	inc de
	ld (caller+1),de
	push de
	push hl
	push bc
caller	call 56000
	pop bc
	pop hl
	pop de
	djnz loop
	ret

dat	defw 56000
	defb 14,20,3

	defw 56000
	defb 5,5,6

Morkin · October 2014

If you do your PUSH BC at the start of the loop, you can (as AndyC indicated) probably replace:

ld a,(hl)
ld (de),a
inc hl
inc de

with

ldi

http://z80-heaven.wikidot.com/instructions-set:ldi

spud · October 2014

31 bytes, assuming the caller function doesn't wreck the shadow registers:

org 32000
	ld b,2
	ld hl,dat
loop	exx
	ld e,(hl)
	inc hl
	ld d,(hl) 
	inc hl
	ldi
	ldi
	ldi
	ld (caller+1),de
	push de
	push hl
caller	call 56000
	pop hl
	pop de
	exx
	djnz loop
	ret

dat	defw 56000
	defb 14,20,3

	defw 56000
	defb 5,5,6

daveysludge · October 2014

Cheers guys!

Memory is getting tight so any saving is essential!

spud · October 2014

Actually Ignore mine, duno what i was smoking. :)

You could use the af/'af for the count, that was in my mind originally then went off on a merry road of insanity with the exx.

	ld a,2
	ld hl,dat
loop	
        ex af,af'
	ld e,(hl)
	inc hl
	ld d,(hl) 
	inc hl
	ldi
	ldi
	ldi
	ld (caller+1),de
	push de
	push hl
caller	call 56000
	pop hl
	pop de
	ex af,af'
	dec a
        jr nz,  loop
	ret

dat	defw 56000
	defb 14,20,3

	defw 56000
	defb 5,5,6

Help needed with a bit of code...

Comments