Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.sys.apple2.programmer > #5756 > unrolled thread

6502 assembler for Unix with multiple ORG support

Back to article view | Back to comp.sys.apple2.programmer

Contents

  6502 assembler for Unix with multiple ORG support kegs@provalid.com (Kent Dickey) - 2021-03-19 01:15 -0500
    Re: 6502 assembler for Unix with multiple ORG support I am Rob <gids.rs@sasktel.net> - 2021-03-18 23:58 -0700
    Re: 6502 assembler for Unix with multiple ORG support I am Rob <gids.rs@sasktel.net> - 2021-03-19 00:18 -0700
    Re: 6502 assembler for Unix with multiple ORG support Michael J. Mahon <mjmahon@aol.com> - 2021-03-19 03:13 -0500
      Re: 6502 assembler for Unix with multiple ORG support kegs@provalid.com (Kent Dickey) - 2021-03-19 14:36 -0500
        Re: 6502 assembler for Unix with multiple ORG support Michael Pohoreski <michael.pohoreski@gmail.com> - 2021-03-19 13:19 -0700
          Re: 6502 assembler for Unix with multiple ORG support Antoine Vignau <ntn.vignau@gmail.com> - 2021-03-20 08:38 -0700
            Re: 6502 assembler for Unix with multiple ORG support Michael J. Mahon <mjmahon@aol.com> - 2021-03-21 00:13 -0500
    Re: 6502 assembler for Unix with multiple ORG support David Schmidt <schmidtd@my-deja.com> - 2021-03-19 09:57 -0400
    Re: 6502 assembler for Unix with multiple ORG support Michael Pohoreski <michael.pohoreski@gmail.com> - 2021-03-19 07:59 -0700
    Re: 6502 assembler for Unix with multiple ORG support qkumba <peter.ferrie@gmail.com> - 2021-03-19 09:27 -0700
      Re: 6502 assembler for Unix with multiple ORG support kegs@provalid.com (Kent Dickey) - 2021-03-19 15:08 -0500
    Re: 6502 assembler for Unix with multiple ORG support Michael Pohoreski <michael.pohoreski@gmail.com> - 2021-03-19 09:34 -0700
    Re: 6502 assembler for Unix with multiple ORG support ol.sc@web.de (Oliver Schmidt) - 2021-03-20 15:14 +0000

#5756 — 6502 assembler for Unix with multiple ORG support

I would like to know if a 6502 assembler for modern Unix systems supports
multiple ORG requests, so that I can force certain routines at fixed
addresses, and insert 0's to fill any space.  For example:

	org	$300
	jsr	rout320
	rts
	org	$320
rout320
	rts

And this will place the JSR and RTS at $300-303, and then 00's and then place
an RTS at $320.  I want the final binary result to be:

20 20 03 60,00 00 00 00,00 00 00 00,00 00 00 00,60

and be exactly 33 bytes.  If I make an ORG request which cannot work (say,
replace the "org $320" with "org $303", which already contains the previous
RTS) I will accept either a warning or an error, but it cannot be silent.

I'm willing to accept alternative syntax using DS notation:

	org	$300
	jsr	rout320
	rts
	ds	$320-*
rout320
	rts

where DS means insert $320-current_pc number of 00 bytes.  I just need this to
be an error if it overflows, and not 65535 blanks instead.

So, does your favorite 6502 assembler support the above?

Kent

[toc] | [next] | [standalone]

#5758

On Friday, March 19, 2021 at 12:15:43 AM UTC-6, Kent Dickey wrote:
> I would like to know if a 6502 assembler for modern Unix systems supports 
> multiple ORG requests, so that I can force certain routines at fixed 
> addresses, and insert 0's to fill any space. For example: 
> 
> org $300 
> jsr rout320 
> rts 
> org $320 
> rout320 
> rts 
> 
> And this will place the JSR and RTS at $300-303, and then 00's and then place 
> an RTS at $320. I want the final binary result to be: 
> 
> 20 20 03 60,00 00 00 00,00 00 00 00,00 00 00 00,60 
> 
> and be exactly 33 bytes. If I make an ORG request which cannot work (say, 
> replace the "org $320" with "org $303", which already contains the previous 
> RTS) I will accept either a warning or an error, but it cannot be silent. 
> 
> I'm willing to accept alternative syntax using DS notation: 
> 
> org $300 
> jsr rout320 
> rts 
> ds $320-* 
> rout320 
> rts 
> 
> where DS means insert $320-current_pc number of 00 bytes. I just need this to 
> be an error if it overflows, and not 65535 blanks instead. 
> 
> So, does your favorite 6502 assembler support the above? 
> 
> Kent

[toc] | [prev] | [next] | [standalone]

#5759

On Friday, March 19, 2021 at 12:15:43 AM UTC-6, Kent Dickey wrote:
> I would like to know if a 6502 assembler for modern Unix systems supports 
> multiple ORG requests, so that I can force certain routines at fixed 
> addresses, and insert 0's to fill any space. For example: 
> 
> org $300 
> jsr rout320 
> rts 
> org $320 
> rout320 
> rts 
> 
> And this will place the JSR and RTS at $300-303, and then 00's and then place 
> an RTS at $320. I want the final binary result to be: 
> 
> 20 20 03 60,00 00 00 00,00 00 00 00,00 00 00 00,60 
> 
> and be exactly 33 bytes. If I make an ORG request which cannot work (say, 
> replace the "org $320" with "org $303", which already contains the previous 
> RTS) I will accept either a warning or an error, but it cannot be silent. 
> 
> I'm willing to accept alternative syntax using DS notation: 
> 
> org $300 
> jsr rout320 
> rts 
> ds $320-* 
> rout320 
> rts 
> 
> where DS means insert $320-current_pc number of 00 bytes. I just need this to 
> be an error if it overflows, and not 65535 blanks instead. 
> 
> So, does your favorite 6502 assembler support the above? 

The first way is not possible as there is no way for an assembler to remember what you assembled previously for it to know that if a change to an ORG's address is treading on a pre-existing routine.

The 2nd way "DS $320-*" is the best way, but if you believe the code you want to enter below $320 might over-run $320, then the best way is give yourself some room and start at $340.  Once you get the code below $340 efficient and the way you want, then you can close up the code to the nearest useable fixed address.

The only reason I can think of for doing this is, you want to be able to examine the code afterwards and be able to go to that fixed address.
Instead of doing what you want to do at $300 where space is king of limited, consider writing the code higher up in memory where there is lots of room, say $1000, and spread your fixed code out every $100 bytes.  Once you get the code the way you like and if space permits, then change the ORG's to fit into $300.

[toc] | [prev] | [next] | [standalone]

#5760

Kent Dickey <kegs@provalid.com> wrote:
> I would like to know if a 6502 assembler for modern Unix systems supports
> multiple ORG requests, so that I can force certain routines at fixed
> addresses, and insert 0's to fill any space.  For example:
> 
> 	org	$300
> 	jsr	rout320
> 	rts
> 	org	$320
> rout320
> 	rts
> 
> And this will place the JSR and RTS at $300-303, and then 00's and then place
> an RTS at $320.  I want the final binary result to be:
> 
> 20 20 03 60,00 00 00 00,00 00 00 00,00 00 00 00,60
> 
> and be exactly 33 bytes.  If I make an ORG request which cannot work (say,
> replace the "org $320" with "org $303", which already contains the previous
> RTS) I will accept either a warning or an error, but it cannot be silent.
> 
> I'm willing to accept alternative syntax using DS notation:
> 
> 	org	$300
> 	jsr	rout320
> 	rts
> 	ds	$320-*
> rout320
> 	rts
> 
> where DS means insert $320-current_pc number of 00 bytes.  I just need this to
> be an error if it overflows, and not 65535 blanks instead.
> 
> So, does your favorite 6502 assembler support the above?
> 
> Kent
> 

Hi, Kent!

Merlin is capable of doing what you want in the second way. 

If you want error checking in case the location counter has passed the
desired address, use a macro to issue an “err */$320”, where, for
generality, $320 is replaced by ]1, or the first parameter of the macro. 

So:

morg  mac              ; Org with zero-fill
            err    */]1      ; Error if past location
            ds     ]1-*
            eom

This assumes that it doesn’t matter what the ds does as long as the error
is flagged. Alternatively, you can use the “if” conditional assembly
construct to bracket the ds. 

-- 
-michael - NadaNet 3.1 and AppleCrate II:  http://michaeljmahon.com

[toc] | [prev] | [next] | [standalone]

#5768

In article <roedncHijdqBwMn9nZ2dnUU7-cPNnZ2d@giganews.com>,
Michael J. Mahon  <mjmahon@aol.com> wrote:
>Kent Dickey <kegs@provalid.com> wrote:
>> I would like to know if a 6502 assembler for modern Unix systems supports
>> multiple ORG requests, so that I can force certain routines at fixed
>> addresses, and insert 0's to fill any space.  For example:
[ snip ]
>> I'm willing to accept alternative syntax using DS notation:
>> 
>> 	org	$300
>> 	jsr	rout320
>> 	rts
>> 	ds	$320-*
>> rout320
>> 	rts
>> 
>> where DS means insert $320-current_pc number of 00 bytes.  I just need this to
>> be an error if it overflows, and not 65535 blanks instead.
>> 
>> So, does your favorite 6502 assembler support the above?
>> 
>> Kent
>> 
>
>Hi, Kent!
>
>Merlin is capable of doing what you want in the second way. 
>
>If you want error checking in case the location counter has passed the
>desired address, use a macro to issue an “err */$320”, where, for
>generality, $320 is replaced by ]1, or the first parameter of the macro. 
>
>So:
>
>morg  mac              ; Org with zero-fill
>            err    */]1      ; Error if past location
>            ds     ]1-*
>            eom
>
>This assumes that it doesn’t matter what the ds does as long as the error
>is flagged. Alternatively, you can use the “if” conditional assembly
>construct to bracket the ds. 
>
>-- 
>-michael - NadaNet 3.1 and AppleCrate II:  http://michaeljmahon.com

OK, I could make the "MORG" work with the following macro:

morg    mac
        err     * > ]1
        ds      ]1 - *
        eom

I need "ds 0" to work, in case the code just happened to line up perfectly.
The above passes my tests so far, so it looks fine.

Also, Merlin32 has a bug where it core dumps on the following:

	org	$300
	jmp	$fded
	ds	$302 - *
	jmp	$300

The ds being effectively -1 causes the core dump

I like Merlin since I used Merlin-16 on my Apple IIgs, but I don't love the
lack of support for parentheses in expressions.

My intention is to use this to get my version of ROMs to have the ROM
routines to be at the right address.  COUT has to be at $FDED, I cannot move
it.  And now the source will force it there.

Kent

[toc] | [prev] | [next] | [standalone]

#5770

On Friday, March 19, 2021 at 12:37:03 PM UTC-7, Kent Dickey wrote:
> Also, Merlin32 has a bug where it core dumps on the following: 
> 
> org $300 
> jmp $fded 
> ds $302 - * 
> jmp $300 
> 
> The ds being effectively -1 causes the core dump 

Hmm, strange. I get no crash but I don't see the binary or the listing file either. It stops after:
    o Build Object Code...

And doesn't get to the Link stage.
    o Build Object Code...
  + Link project files...

Guess we should ping Antoinne and/or Olivier in a new thread.

m.

[toc] | [prev] | [next] | [standalone]

#5773

On Friday, March 19, 2021 at 9:19:08 PM UTC+1, michael....@gmail.com wrote:
> On Friday, March 19, 2021 at 12:37:03 PM UTC-7, Kent Dickey wrote: 
> > Also, Merlin32 has a bug where it core dumps on the following: 
> > 
> > org $300 
> > jmp $fded 
> > ds $302 - * 
> > jmp $300 
> > 
> > The ds being effectively -1 causes the core dump
> Hmm, strange. I get no crash but I don't see the binary or the listing file either. It stops after: 
> o Build Object Code... 
> 
> And doesn't get to the Link stage. 
> o Build Object Code... 
> + Link project files... 
> 
> Guess we should ping Antoinne and/or Olivier in a new thread. 
> 
> m.

Hi There,
This is what I have on my mac:
./Merlin32 -V Library/ sources/test/ds.s 
./Merlin32 v1.0 64-bit universal, (c) Brutal Deluxe 2011-2021
  + Assemble project files...
    o Loading Sources files...
        - ds.s
    o Loading Macro files...
    o Check for duplicated Macros...
    o Decoding lines types...
    o Process local/variable Labels...
    o Process Asterisk lines...
    o Build External table...
    o Build Equivalence table...
    o Build Variable table...
    o Process Equivalence values...
    o Replace Macros with Code...
    o Replace Lup with code...
    o Process MX directives...
    o Process Conditional directives...
    o Build Label table...
    o Check for duplicated Labels...
    o Check for unknown Source lines...
    o Check for Dum lines...
    o Compute Operand Code size...
    o Compute Operand Data size...
    o Compute Line address...
      => [Error] Error : Evaluation of DS data size ends up as negative value (-1) : '$302-ozunid_1' (line 7, file 'ds.s').
     => Creating Output file 'sources/test/error_output.txt'

[toc] | [prev] | [next] | [standalone]

#5774

Antoine Vignau <ntn.vignau@gmail.com> wrote:
> On Friday, March 19, 2021 at 9:19:08 PM UTC+1, michael....@gmail.com wrote:
>> On Friday, March 19, 2021 at 12:37:03 PM UTC-7, Kent Dickey wrote: 
>>> Also, Merlin32 has a bug where it core dumps on the following: 
>>> 
>>> org $300 
>>> jmp $fded 
>>> ds $302 - * 
>>> jmp $300 
>>> 
>>> The ds being effectively -1 causes the core dump
>> Hmm, strange. I get no crash but I don't see the binary or the listing
>> file either. It stops after: 
>> o Build Object Code... 
>> 
>> And doesn't get to the Link stage. 
>> o Build Object Code... 
>> + Link project files... 
>> 
>> Guess we should ping Antoinne and/or Olivier in a new thread. 
>> 
>> m.
> 
> Hi There,
> This is what I have on my mac:
> ./Merlin32 -V Library/ sources/test/ds.s 
> ./Merlin32 v1.0 64-bit universal, (c) Brutal Deluxe 2011-2021
>   + Assemble project files...
>     o Loading Sources files...
>         - ds.s
>     o Loading Macro files...
>     o Check for duplicated Macros...
>     o Decoding lines types...
>     o Process local/variable Labels...
>     o Process Asterisk lines...
>     o Build External table...
>     o Build Equivalence table...
>     o Build Variable table...
>     o Process Equivalence values...
>     o Replace Macros with Code...
>     o Replace Lup with code...
>     o Process MX directives...
>     o Process Conditional directives...
>     o Build Label table...
>     o Check for duplicated Labels...
>     o Check for unknown Source lines...
>     o Check for Dum lines...
>     o Compute Operand Code size...
>     o Compute Operand Data size...
>     o Compute Line address...
>       => [Error] Error : Evaluation of DS data size ends up as negative
> value (-1) : '$302-ozunid_1' (line 7, file 'ds.s').
>      => Creating Output file 'sources/test/error_output.txt'
> 

In (at least) the 8-bit Merlin manuals, in the discussion of ORG, it
specifies that ORG *-n is not a supported construct, and DS -n should be
used to “back up” the location counter.   

I’m surprised that Glen changed this on later versions—or did he?

-- 
-michael - NadaNet 3.1 and AppleCrate II:  http://michaeljmahon.com

[toc] | [prev] | [next] | [standalone]

#5761

On 3/19/21 2:15 AM, Kent Dickey wrote:
> [...]
> So, does your favorite 6502 assembler support the above?

Multiple orgs are more of a linker construct in the ca65 sense.  You aim 
bytes at memory using a linker config with defined sections of memory. 
That said...

Here's how I complain if I try to get this one thing bigger than 256 
bytes, as an example:

;.org $a000
...
;.align 256
;.assert        * = $a100, error, "Code got too big to fit in a page!"

Reference:
https://cc65.github.io/doc/ca65.html#s11

[toc] | [prev] | [next] | [standalone]

#5763

On Thursday, March 18, 2021 at 11:15:43 PM UTC-7, Kent Dickey wrote:
> I would like to know if a 6502 assembler for modern Unix systems supports 
> multiple ORG requests, so that I can force certain routines at fixed 
> addresses, and insert 0's to fill any space. For example: 
> 
> org $300 
> jsr rout320 
> rts 
> org $320 
> rout320 
> rts 

ORG does NOT generate bytes.  All it does it is update the assembler what the _next_ output address to use.

> I'm willing to accept alternative syntax using DS notation: 

That's not an alternative syntax, that is the primary syntax. ;-)

If you aren't using a linker then you need to tell the assembler to Define Space and fill in padding.

m.

[toc] | [prev] | [next] | [standalone]

#5764

ACME can do something like that, though not using the "ORG" syntax.
Instead, it uses the current address via "*", so
*=$300
...
*=$320
...
etc.
and it will zero-fill in between (and also complain if you have too much code and already occupy the requested address).

[toc] | [prev] | [next] | [standalone]

#5769

In article <4dbdfb4b-0e25-4187-8c77-56884501a101n@googlegroups.com>,
qkumba  <peter.ferrie@gmail.com> wrote:
>ACME can do something like that, though not using the "ORG" syntax.
>Instead, it uses the current address via "*", so
>*=$300
>...
>*=$320
>...
>etc.
>and it will zero-fill in between (and also complain if you have too much
>code and already occupy the requested address).

I tried ACME.  It has an unusual directive syntax.  It also has no support
for low/high ascii strings (as far as I could tell).


Here's my test file (try1.a):

        !cpu 65c02
        * =     $d000
        jmp     rout_d234
        rts

        * =     $d108
rout_d108
        nop
        nop
        jmp     rout_d234

        * =     $d234
rout_d234
        jmp     *
        rts
        rts
        * =     $d238
rout_d238
        bra     *
        bra     *
        nop

Where rout_d238 doesn't fit since * is already at $d239.  And this was a
warning.  But it forced the PC back a byte, and wiped out the RTS at $d238.
This is not desirable--if it is going to wipe out code, it should be an error.
And the exit code from the acme assembler command is 0, so to catch this, I'd
have to watch the output for the Warning and make it an error.

I like the syntax of expressions, but all the ! commands looks
odd to me.  If you use ACME, do you make macros so DFB, DS, etc. work,
or do you stick with the ACME native !byte, etc. pseudo-ops?

Kent

[toc] | [prev] | [next] | [standalone]

#5765

On Thursday, March 18, 2021 at 11:15:43 PM UTC-7, Kent Dickey wrote:

Another solution is to reserve a page for entry points.  You have two options:

1. Use an array of 4C xx xx bytes like this.

    ORG $BF00
Foo JMP doFoo
Bar  JMP doBar
Qux JMP doQux

2. Do away with wasting one byte per function entry point and just have an array of function pointers
Again here you have to sub-options depending if you are using RTS or JMP.

a) RTS uses words-1
b) JMP use addresses

e.g.
DW Foo-1
DW Bar-1
DW Qux-1

DA Foo
DA Bar
DA Qux

This places the burden on the caller though since you are no longer using a simpler JSR. You need to specify which function to call.

 i.e
   LDX #1  ; Foo
   JSR Trampoline
   LDX #2   ; Bar
   JSR Trampoline
   LDX #3   ; Qux
   JSR Trampoline

Trampoline
   TXA
   ASL
   LDA FuncTable+0,X ; low
   LDY FuncTable+1,X ; high
  STA FuncCall+0 ; ***SELF-MODIFIES
  STX FuncCall+1 ; *** SELF-MODIFIES
FuncCall
  JMP $C0DE ; *** SELF-MODIFIED

FuncTable
 DA Foo
 DA Bar
 DA Qux


> where DS means insert $320-current_pc number of 00 bytes. I just need this to 
> be an error if it overflows, and not 65535 blanks instead. 
> 
> So, does your favorite 6502 assembler support the above? 

I did this on Nox Archaist. We were using Page 3 for font rendering code and didn't want to overflow into the $3F0 vector.
Assembler: sbasm3

Page3End
	.DO Page3End>$3EF
		.ER	F,***Overflow into the IRQ vector!!!
	.FI

In case the whitespace gets butchered there should be indentation for .DO and .FI and another indentation for the .ER.

m.

[toc] | [prev] | [next] | [standalone]

#5772

Hi,

>I would like to know if a 6502 assembler for modern Unix systems supports
>multiple ORG requests, so that I can force certain routines at fixed
>addresses, and insert 0's to fill any space.  For example:
>
>	org	$300
>	jsr	rout320
>	rts
>	org	$320
>rout320
>	rts
>
>And this will place the JSR and RTS at $300-303, and then 00's and then place
>an RTS at $320.  I want the final binary result to be:
>
>20 20 03 60,00 00 00 00,00 00 00 00,00 00 00 00,60
>
>and be exactly 33 bytes.  If I make an ORG request which cannot work (say,
>replace the "org $320" with "org $303", which already contains the previous
>RTS) I will accept either a warning or an error, but it cannot be silent.

The seasoned cc65 user would accomplish this with...

demo.s contains:

	.segment "s300"
	jsr	rout320
	rts
	.segment "s320"
rout320:
	rts

demo.cfg contains:

memory {
	m: file = %O, start = $300, size = $FFFF;
}
segments {
	s300: load m, start = $300;
	s320: load m, start = $320;
}

Build it with:
cl65 -t none -C demo.cfg demo.s

It for sure complains if things don't fit.

>I'm willing to accept alternative syntax using DS notation:
>
>	org	$300
>	jsr	rout320
>	rts
>	ds	$320-*
>rout320
>	rts
>
>where DS means insert $320-current_pc number of 00 bytes.  I just need this to
>be an error if it overflows, and not 65535 blanks instead.

The casual ca65 user would accomplish this with...

demo.s contains:

	.org	$300
	jsr	rout320
	rts
	.res	$320-*
rout320:
	rts

Built it with:
cl65 -t none demo.s

It for sure complains if things don't fit.

>So, does your favorite 6502 assembler support the above?

Yes, see above.

Regards,
Oliver

[toc] | [prev] | [standalone]

Back to top | Article view | comp.sys.apple2.programmer

csiph-web