Small IRA Tutorial

[phx]

Originally Posted by hop

There seems to be a problem with the latest changes.

Indeed. I should always test all cases after a modifications.
I have updated the archive.

[hop]

Originally Posted by phx

I have updated the archive.

All working as expected now thanks.

[hop]

Originally Posted by phx

The current V2.11 beta source is here:
http://sun.hasenbraten.de/~frank/TEST/ira_211_beta.lha

I added this new option:

        -LABEL=<style>    Label style 0=LAB_index, 1=LAB_addr.

The default is still the old label-style 0.

Do you think it would make sense to apply this option to EXT symbols too?

Before:

EXT_0007	EQU	$400
EXT_0008	EQU	$402
EXT_0009	EQU	$8002
EXT_000a	EQU	$56660
EXT_000d	EQU	$5666A
EXT_000e	EQU	$7FFFE

After:

EXT_400		EQU	$400
EXT_402		EQU	$402
EXT_8002	EQU	$8002
EXT_56660	EQU	$56660
EXT_5666A	EQU	$5666A
EXT_7FFFE	EQU	$7FFFE

[phx]

Originally Posted by hop

Do you think it would make sense to apply this option to EXT symbols too?

The EXT-symbols are equates and no labels, but why not? Ok, I updated the beta source archive.

[hop]

Originally Posted by phx

The EXT-symbols are equates and no labels, but why not? Ok, I updated the beta source archive.

Thanks very much for making this change. It is a definite improvement, and makes the disassembly easier to understand at a glance. It's especially useful for game bootblocks and loaders that use fixed addresses.

Do you think it is worth mentioning this feature in -LABEL=<style> string in the Usage in FormatError(), or would it just be clutter?

On a related note, is it possible to configure ira to use a single equate for all of the $400 values in the following code? It would be great if the equate could be renamed too.

EXT_400		EQU	$400
...
; ------------------------------------------------------------------------------
; Load Splash Screen Sequence to $400
; ------------------------------------------------------------------------------
	BSR.W	SeekTrackZero		;70106: 610009a4
	BSR.W	MotorOn			;7010a: 61000a02
	MOVEA.L	#$00000400,A0		;7010e: 207c00000400
	MOVEQ	#3,D0			;70114: 7003
	MOVE.W	#$0006,D1		;70116: 323c0006
LAB_7011A:
	BSR.W	ReadTrack		;7011a: 61000916
	ADDA.L	#$00001800,A0		;7011e: d1fc00001800
	ADDI.L	#$00000001,D0		;70124: 068000000001
	DBF	D1,LAB_7011A		;7012a: 51c9ffee
	BSR.W	MotorOff		;7012e: 61000a0a
	MOVEA.L	#$00000400,A0		;70132: 207c00000400
	MOVEA.L	#$00000400,A1		;70138: 227c00000400
	BSR.W	Unpack			;7013e: 61000722
; Execute loaded code at $400 [Splash screens sequence plays]
	JSR	EXT_400.W		;70142: 4eb80400

Ideally the output would look like:

SPLASH_ADDR		EQU	$400
...
; ------------------------------------------------------------------------------
; Load Splash Screen Sequence to $400
; ------------------------------------------------------------------------------
	BSR.W	SeekTrackZero		;70106: 610009a4
	BSR.W	MotorOn			;7010a: 61000a02
	MOVEA.L	#SPLASH_ADDR,A0		;7010e: 207c00000400
	MOVEQ	#3,D0			;70114: 7003
	MOVE.W	#$0006,D1		;70116: 323c0006
LAB_7011A:
	BSR.W	ReadTrack		;7011a: 61000916
	ADDA.L	#$00001800,A0		;7011e: d1fc00001800
	ADDI.L	#$00000001,D0		;70124: 068000000001
	DBF	D1,LAB_7011A		;7012a: 51c9ffee
	BSR.W	MotorOff		;7012e: 61000a0a
	MOVEA.L	#SPLASH_ADDR,A0		;70132: 207c00000400
	MOVEA.L	#SPLASH_ADDR,A1		;70138: 227c00000400
	BSR.W	Unpack			;7013e: 61000722
; Execute loaded code at $400 [Splash screens sequence plays]
	JSR	SPLASH_ADDR.W		;70142: 4eb80400

[phx]

Originally Posted by hop

Do you think it is worth mentioning this feature in -LABEL=<style> string in the Usage in FormatError(), or would it just be clutter?

It's a little bit tight, but I tried to put it there. Source is updated.

On a related note, is it possible to configure ira to use a single equate for all of the $400 values in the following code? It would be great if the equate could be renamed too.

That's not so trivial and would require some more time to do - which I currently don't have, being focussed on other projects.

[hop]

Originally Posted by phx

That's not so trivial and would require some more time to do - which I currently don't have, being focussed on other projects.

No problem at all. I just wanted to make sure I wasn't missing something. Thanks very much for your support!

[hop]

Hi. I'm a bit confused by the intent of the

ExtendFileName

function in init.c. It is used to generate the output filenames:

targetName

(.asm),

configName

(.cnf), as well as

binaryName

and

labelName

.

It looks like

ExtendFileName

is designed to append a file extension, and also strip the volume/drive label and directories from a path. For example: DH0:games/game -> game.asm, game.cnf etc

So it would seem that ira is designed to output the .asm/.cnf file in the current working directory by default, rather than next to the source file.

I've been using ira on Windows, which often uses backslashes for path separators.

ExtendFileName

does not account for this - it only searches for forward slashes. Fixing this has exposed the behaviour described above, which I don't think is very desirable. I think it would be better if

ExtendFileName

did not strip the preceding path from a filename, so for example: DH0:games/game -> DH0:games/game.asm, DH0:games/game.cnf etc and similarly for Windows C:\Games\Game -> C:\Games\Game.asm, C:\Games\Game.cnf

What are your thoughts on this? Please could you let me know if I have misunderstood anything here? Thanks

[phx]

Originally Posted by hop

I think it would be better if

ExtendFileName

did not strip the preceding path from a filename, so for example: DH0:games/game -> DH0:games/game.asm, DH0:games/game.cnf etc and similarly for Windows C:\Games\Game -> C:\Games\Game.asm, C:\Games\Game.cnf

Absolutely! I can only agree. Although I use IRA under AmigaOS/MorphOS and NetBSD I never noticed the problem, as I'm always working with files in my current directory only.

Fixed and updated the test-archive. Only the

ExtendFileName

function in

init.c

was modified.

[hop]

Originally Posted by phx

Absolutely! I can only agree. Although I use IRA under AmigaOS/MorphOS and NetBSD I never noticed the problem, as I'm always working with files in my current directory only.

Fixed and updated the test-archive. Only the

ExtendFileName

function in

init.c

was modified.

Thanks very much. Where can I find the test-archive? I am only aware of https://aminet.net/package/dev/asm/ira

[phx]

It's still the same URL we used previously in this thread:
http://sun.hasenbraten.de/~frank/TEST/ira_211beta.lha

[hop]

Originally Posted by phx

It's still the same URL we used previously in this thread:
http://sun.hasenbraten.de/~frank/TEST/ira_211beta.lha

Thanks. Sorry, I'd forgotten about that.

I've just been disassembling WB1.3 Calculator with ira as a test case, and I am not sure if I have encountered a bug or if this is user error. Can you please advise?

LAB_67A:
	JMP	LAB_67C+2(PC,D0.W)	;067a: 4efb0000
LAB_67E:
	UNLK	A5			;067e: 4e5d
	RTS				;0680: 4e75

The disassembly does not assemble because LAB_67C is not defined. However, 67C+2=LAB_67E, which is defined.

[phx]

I would guess this is a problem in your private ira compilation, as I saw exactly the same problem in your still secret project and sent you a screenshot. I don't see it with ira under NetBSD or AmigaOS.

[hop]

Originally Posted by phx

I would guess this is a problem in your private ira compilation, as I saw exactly the same problem in your still secret project and sent you a screenshot. I don't see it with ira under NetBSD or AmigaOS.

Ah. Sorry. I misunderstood. I'll investigate this. I have only made very minor changes to ira. I'll look into this (test vanilla ira and also on Linux) and let you know what I find. Many thanks.

[hop]

Originally Posted by phx

I would guess this is a problem in your private ira compilation, as I saw exactly the same problem in your still secret project and sent you a screenshot. I don't see it with ira under NetBSD or AmigaOS.

I don't seem to have any local changes to ira. I can repro the issue with your latest archive compiling with

nmake -f Makefile.win32

. It looks this problem may be related the

-LABEL=1

option.

For Calculator with

-LABEL=0

the resultant disassembly is valid - the JMP dest is relative to a defined label:

LAB_0052:
	JMP	LAB_0052+2(PC,D0.W)	;067a: 4efb0000
LAB_0053:
	UNLK	A5			;067e: 4e5d

For Calculator with

-LABEL=1

the resultant disassembly is invalid - the JMP dest is relative to an undefined label and the offset is effectively applied twice:

LAB_67A:
	JMP	LAB_67C+2(PC,D0.W)	;067a: 4efb0000
LAB_67E:
	UNLK	A5			;067e: 4e5d

I'll see if I can debug this. The command line is:

ira -a -label=1 -radix=0 -compat=bi -config Calculator

Thanks

[phx]

Originally Posted by hop

IIt looks this problem may be related the

-LABEL=1

option.

Confirmed. I missed that you use that option, while trying to reproduce it.
It happens when offsetting a label (like

LAB_xyz+2

).

[hop]

Originally Posted by phx

Confirmed. I missed that you use that option, while trying to reproduce it.
It happens when offsetting a label (like

LAB_xyz+2

).

Thanks. I think I've found the point in the code that appends the label and the +, but not sure quite why it's going wrong yet.

[phx]

Fixed it. I have to use

LabelAdr2[r]

instead of

label.labelAdr[r]

in

CatLabelAdr()

. Update with the beta archive from the previous link.

[hop]

Originally Posted by phx

Fixed it. I have to use

LabelAdr2[r]

instead of

label.labelAdr[r]

in

CatLabelAdr()

. Update with the beta archive from the previous link.

Haha! I have literally this minute just found this myself Many thanks for confirming!

[phx]

Two hours later I fixed a similar bug with offsetting a symbol defined at section start (replacing

SECSTRT_n

). Make sure

ira_2.c

has date 13:58 CEST. I reuploaded the archive.