vasm relocation error

[BippyM]

I am getting a few

error 39: illegal relocation

errors wih vasm.


The errors are with the following lines (note, these lines are taken from the whole routine, and are not listed like this in the code, I can post the full code if needed, but also see the attached screenshot)

move.w     #PIPEBLIT,d0
move.w     #BLITDMOD,d0

move.w     #sprheight<<6,d2; #sprheight*64 = 32*64 = 2048
 move.w     #sprwidth>>4,d3; Spritewidth / 8 = 32/8 = 4BITPLANE_SIZE_B
add.l      #SCREEN_WIDTH/8,a0; Next source Bitplane

The errors are showing on the following constants:
PIPEBLIT, BLITMOD, sprheight, sprwidth, SCREEN_WIDTH


Here is the setup for the above constants

sprheight        equ 32
sprwidth         equ 32
sprsize          equ (sprheight*4)+8; Bytes to skip to next sprite (sprheight*bytesperline+controlwords)
sprskip          equ sprsize*4

SPRxSPD          equ 1
SPRySPD          equ 2


BLITDMOD          equ (BITPLANE_WORDS*SCREEN_DEPTH)-(sprwidth/8)
PIPEBLIT          equ (PIPE_TILE_WIDTH-sprwidth)/8

[jotd]

happens when the name is not defined as equ.

[Galahad/FLT]

Are you using any includes in place of using hardware registers directly?

BLITDMOD or BLTDMOD is usually to describe $dff066, is VASM being told elsewhere that BLTDMOD is actually $dff066 and so the move.w wont work because $dff066 is being declared as a longword?

[phx]

Most likely one or more of your Equates are unknown. Maybe as a result of a typo or simply wrong case?

These relocation errors appear when you use an unknown symbol, which the assembler then assumes as being externally defined, in complex arithmetic operations. Because Relocs only allow a simple addend with a referenced symbol.

EDIT:
The equates you have shown seem correct. But also check if BITPLANE_WORDS, SCREEN_DEPTH and PIPE_TILE_WIDTH are defined. Otherwise you have equates based on unknown symbols, and the arithmetic operations with them will lead to the relocation errors.

[BippyM]

Originally Posted by phx

Most likely one or more of your Equates are unknown. Maybe as a result of a typo or simply wrong case?

These relocation errors appear when you use an unknown symbol, which the assembler then assumes as being externally defined, in complex arithmetic operations. Because Relocs only allow a simple addend with a referenced symbol.

EDIT:
The equates you have shown seem correct. But also check if BITPLANE_WORDS, SCREEN_DEPTH and PIPE_TILE_WIDTH are defined. Otherwise you have equates based on unknown symbols, and the arithmetic operations with them will lead to the relocation errors.

From what I can see everything is defined. Most of it is done in an external file that is included early in my code.


sprheight is defined in my constants.asm file and then only used above. I have't used it anywhere else.

I have added my code for the whole project (it is messy as I am learning).

Originally Posted by Galahad/FLT

Are you using any includes in place of using hardware registers directly?

BLITDMOD or BLTDMOD is usually to describe $dff066, is VASM being told elsewhere that BLTDMOD is actually $dff066 and so the move.w wont work because $dff066 is being declared as a longword?

I don't think so, as mentioned above, i get the error just with the sprheight declaration and that is declared and used once.

Originally Posted by jotd

happens when the name is not defined as equ.

Everything is defined that I can see, but maybe I cannot see the wood for the trees!

[JoeJoe]

So with vasm 1.9 the whole thing can be compiled and started. But the includes are needed, otherwise the code will not run.

[BippyM]

Originally Posted by JoeJoe

So with vasm 1.9 the whole thing can be compiled and started. But the includes are needed, otherwise the code will not run.

You'll need all the data files too to run it. I'll upload it all later as I'm out with the kids now

[JoeJoe]

Yes I know it. You have post it on other side

[BippyM]

Originally Posted by JoeJoe

Yes I know it. You have post it on other side

Yeah differe t issue that I thiught was related to the plug in. I deleted that one but if you still have it feel free to include it here if it is needed

[JoeJoe]

Sorry, then I expressed myself stupidly. Your sources can be compiled without problems. I have simply copied the missing includes from the "other" project.
So I can't reproduce the error.

[phx]

Originally Posted by BippyM

Everything is defined that I can see, but maybe I cannot see the wood for the trees!

Phew... your example source needed a lot of changes to make it assemble on a Unix system (differing case in file names). All INCBINs were missing, so I commented them out. funcdef.i. The includes from "I" are missing, so I took my own "custom.i", and I'm not sure which SDK you were using. NDK3.9? Devpac-includes?

In the end I could assemble it, and the only error I saw was this:

vasm 1.9a (c) in 2002-2022 Volker Barthelmann
vasm M68k/CPU32/ColdFire cpu backend 2.5c (c) 2002-2022 Frank Wille
vasm motorola syntax module 3.16 (c) 2002-2022 Frank Wille
vasm test output module 1.0 (c) 2002 Volker Barthelmann

error 39 in line 43 of "src/Copper.asm": illegal relocation
        included from line 32 of "Overflow.s"
>        move.l     #(BPL0PTH<<16),d0                           ; Bitplane high pointer to $00E00000

That's because BPL0PTH does not exist, and it cannot shift an unknown symbol (hence, illegal relocation). Officially, they are called BPL1PT .. BPL6PT. No BPL0PT.

Do you have something which reproduces the other errors?

EDIT: When trying to link an executable I also notice that Sprite1a and Sprite2a are undefined.

[JoeJoe]

The symbole "BPL0PTH" is added at line 20 (include "i/custom.i") in the Overflow.s file.

[phx]

But there was no "i/custom.i" in the archive.

@bippy: Can you exclude a problem with VisualCode? Did you also try to assemble your project manually from the shell?

[BippyM]

Okay, so maybe I dived in abit too deep


well this is the header of my custom.i

*
* Custom chips hardware registers
*
* Written by Frank Wille in 2013, 2016.
*

Anyway I seem to be using a mixture of includes, I think I should probably replace them with a consistent set. I have some files from release 40.13 and some from an unknown release with intuition.i from release 1.3.. So I will sort this tonight I think. Any advice on which version of includes would be best? NDK3.9 maybe??




I will also go back to my source and try and clean it up. I might even restart it so that it is better structured. The issue with the sprites is the code started off just creating a screen, and then I took the agony sprite (which is an attached sprite 32px wide (so 4 sprites as it is 16 colour)) and I moved it around the screen. I then literally replaced the sprite and changed my code to get a gamescreen running.


The sprite data is actually in another directory (bin/). I probably should have included everything in my archive.


Thanks for the help. I will go and look at what you and others have pointed out and see what I can figure out further

edit: attached the custom.i I am using and bpl0pth is defined

Originally Posted by phx

Phew... your example source needed a lot of changes to make it assemble on a Unix system (differing case in file names). All INCBINs were missing, so I commented them out. funcdef.i. The includes from "I" are missing, so I took my own "custom.i", and I'm not sure which SDK you were using. NDK3.9? Devpac-includes?

In the end I could assemble it, and the only error I saw was this:

vasm 1.9a (c) in 2002-2022 Volker Barthelmann
vasm M68k/CPU32/ColdFire cpu backend 2.5c (c) 2002-2022 Frank Wille
vasm motorola syntax module 3.16 (c) 2002-2022 Frank Wille
vasm test output module 1.0 (c) 2002 Volker Barthelmann

error 39 in line 43 of "src/Copper.asm": illegal relocation
        included from line 32 of "Overflow.s"
>        move.l     #(BPL0PTH<<16),d0                           ; Bitplane high pointer to $00E00000

That's because BPL0PTH does not exist, and it cannot shift an unknown symbol (hence, illegal relocation). Officially, they are called BPL1PT .. BPL6PT. No BPL0PT.

Do you have something which reproduces the other errors?

EDIT: When trying to link an executable I also notice that Sprite1a and Sprite2a are undefined.

[BippyM]

So... I decided to move all the constants out of the constants.asm file and included them at the bottom of the overflow.s file after all includes. The errors have now cleared

[Galahad/FLT]

Originally Posted by BippyM

So... I decided to move all the constants out of the constants.asm file and included them at the bottom of the overflow.s file after all includes. The errors have now cleared

It might be that the file you've now moved your constants into, is now loaded/referenced earlier than the other file, which means the assembler now knows all your equates before the source uses them.

[BippyM]

Originally Posted by Galahad/FLT

It might be that the file you've now moved your constants into, is now loaded/referenced earlier than the other file, which means the assembler now knows all your equates before the source uses them.

Well i have a file that loads in all the includes and source files and jumps to my progstart: lable. The first include after they system includes was my constants include which had well.. You know Anyway all I have done is removed that include and instead copied the contents of it into the file. That does the including, only it's at the end... So no idea!

[phx]

Originally Posted by BippyM

edit: attached the custom.i I am using and bpl0pth is defined

Your custom.i has my header, but I am quite sure that somebody modified it and changed BPL1..6PT to BPL0..5PT. It wasn't me. The custom.i in all my games uses the official 1..6 numbering.

Such things happen, when you release your source into the public domain...

Glad that you found the issue, BTW.

[BippyM]

Originally Posted by phx

Your custom.i has my header, but I am quite sure that somebody modified it and changed BPL1..6PT to BPL0..5PT. It wasn't me. The custom.i in all my games uses the official 1..6 numbering.

Such things happen, when you release your source into the public domain...

Glad that you found the issue, BTW.

Thanks, I have no idea where I got your modified includes from. I'll source some proper updated ones