I am trying to make my first GFX assembler code

[meynaf]

Originally Posted by StingRay

You can use local labels. Though, the approach to use the stack to obtain the (string) addresses can indeed be quite useful.

Well, it's a real pain to disassemble (and to debug).
This program was just full of this and if i could kick the author's ass i'd have done it.

[litwr]

Originally Posted by meynaf

Well, it's a real pain to disassemble (and to debug).
This program was just full of this and if i could kick the author's ass i'd have done it.

Don't hack!

[meynaf]

Originally Posted by litwr

Don't hack!

Tell that to people who got happy with my game ports

[StingRay]

Originally Posted by meynaf

Well, it's a real pain to disassemble (and to debug).
This program was just full of this and if i could kick the author's ass i'd have done it.

Well, nothing a simple MACRO in ReSource couldn't solve! Quite a few games actually use constructs like this (check the Wizkid code for a prime example) but I do agree, they are painful to debug/disassemble.

[litwr]

I have the next sequence in my x86 code

     test [data],3
     jne label

    ...
data dw 0

How to do something similar with 68k? I could find out

    btst.w #0,data   ;or btst.b #0,data+1
    bne label

    btst.w #1,data
    bne label

And a bit shorter code

    move data,ccr
    bcc label
    bvc label

It is a bit odd that MOVE to CCR takes a word not a byte - it is not important, indeed.
Have I missed a better way? Thanks.

[meynaf]

Originally Posted by litwr

Have I missed a better way? Thanks.

Why not the obvious :

 moveq #3,d0
 and.w data,d0
 bne label

[hooverphonique]

Originally Posted by litwr

    btst.w #0,data   ;or btst.b #0,data+1

btst is always byte-sized when operating on memory!

[StingRay]

And it doesn't have a size spezifier either, i.e. btst.w is wrong and any assembler which accepts that is broken IMHO.

[litwr]

I have found out that Text call preserves A1 with A500 but corrupts it with A1200.

[Leffmann]

D0-D1/A0-A1 are scratch registers, this has always been the case for system calls.

[litwr]

Originally Posted by Leffmann

D0-D1/A0-A1 are scratch registers, this has always been the case for system calls.

Thank you. I hoped that A1200 remains more compatible with A500 and was surprised when my code failed to work with A1200. I am going to follow documentation more strictly.

[Don_Adan]

Originally Posted by litwr

Thank you. I hoped that A1200 remains more compatible with A500 and was surprised when my code failed to work with A500. I am going to follow documentation more strictly.

Remember some Amiga ROM-s routines were reworked a few times. D0-D1/A0-A1 are always Amiga scratch registers, except a few math and exec routines (info is stored inside docs). Of course not always all scratch registers are used/changed.