Simtel MSDOS 1992 December/simtel1292_SIMTEL_1292_Walnut_Creek.iso/msdos/crossasm/svasm02.arc/INFORMAL.DOC

Text File | 1988-05-24 | 10KB | 159 lines

                                S V A S M 0 2

                    SOME INFORMAL COMMENTS ON VERSION 1.0

                                  May, 1988


I had been using an 8-bit (CP/M) commercial assembler for developing 6502 code
for an amateur radio 2 meter repeater system I had built.  As the code grew, I
needed  an assembler that would give me a cross reference listing of  symbols.
I also wanted to try the 65C02 chip,  which I had no assembler for.  The price
for  a new commercial product was ludicrous,  and GEnie had nothing that would
suit my needs.   The solution was obvious:  write one.   I started in the  old
TRS-80 (CP/M) using 8080 assembly language,  but soon outgrew the system.   It
was time to upgrade anyway, so I switched to an IBM XT compatible.  Then I had
to learn 8086 assembly and rewrite what I had thus far.  After about a year of
work,  I  had a two pass assembler that would produce loadable hex code for  a
6502 or 65C02, and give me a cross referenced symbol table.

SVASM02 is written in good tight 8086 assembly for efficiency.  I once wrote a
cross  referencing  assembler in 8-bit CP/M BASIC.   The compiled version  was
over 27K long,  and was still slow.   The SVASM02 assembler is around 12K, and
runs  quite rapidly.   A single 64K segment is sufficient for  the  code,  I/O
buffers,  and the symbol table.  No "temporary" files are generated, since the
symbol  table  is constructed in RAM.   The second pass reads the source  code
from  the disk a second time,  and undefined first pass symbols are  resolved.
This  approach  is much more efficient than writing temporary  information  to
disk  for later read back,  and significantly reduces disk and  CPU  overhead.
There are no support files required for SVASM02.

The program was developed by first writing all of the support routines  before
adding  the  instruction set decoder.   This modular approach will  allow  the
generation  of other assemblers by merely writing the instruction set  decoder
for the chip in question.  I plan to write one for the Motorola 6805 series (I
have  an application for it),  using the SVASM02 code as a foundation.   Other
chips,  like the 6800 or 6809 (for which I don't have an application) may also
be done if the demand (and rewards) are forthcoming.

The nice part about writing an assembler is that you get to define the  syntax
of the code you are assembling.   I had used a couple of commercial assemblers
for various chips,  and liked some features of each.   For example,  I use the
asterisk (*) as the symbol for the program counter in preference to the dollar
sign.   I also don't like the colon as a label delimiter.   In SVASM02, labels
are the first thing on a line other than an instruction, pseudo-op, comment or
an  equate,  without the dumb colon.  Future versions may permit the use of  a
colon but not require it.   Also,  I don't like DB, DW, ORG, and similar Intel
type syntax.  I use the  pseudo-ops .BYTE and .WORD for bytes and words and "*
=  " for defining the program counter and setting aside uninitialized storage.
The equal sign is used to define equates.

SVASM02 is easy to run.   Once the program name is entered,  prompts are given
for  the  filenames  for source,  object,  and listing files.   There  are  no
switches  or  modifiers.    The program directives will do  about  anything  a
switch  would do,  with less confusion and more options.   Selection of object
and listing output is optional,  and the listing output may be specified to go
to  the  screen  or the printer by entering the destination  at  the  filename
prompt.

Now a word or two about case.   I don't like lower case for assembly language.
In fact,  I can't STAND lower case for assembly language.   This is probably a
quirk  on  my  part,  but  I find lower case  almost  impossible  to  read  in
expressions  like "start1 = $a3c6".   Perhaps it wouldn't be too bad if we had
lower case numbers.  Except for filename entry during invocation,  the SVASM02
assembler will not convert lower to upper,  and is totally based on upper case
characters only.   Lower case is fully acceptable within a quoted string or as
commentary information, however, as it should be.

There  are a few other things SVASM02 will and will not do.   It will assemble
6502  and/or 65C02 mnemonics and produce both a line-numbered  source  listing
with  a  cross referenced symbol table,  and a HEX load file in Intel  format.
Input is from disk only.   The HEX file normally will go to disk.  The  source
listing output may be directed to disk,  the printer, or the screen.  Both the
source  listing and HEX files are optional.  While DOS supports the use of the
Reserved Device Names, SVASM02 may not perform as expected if they are used in
place  of the usual disk name.   Drivers have been included to permit  correct
console (CON) and printer (LPT) output if desired.   See the SVASM02.DOC  file
for details.

Assembly  time errors (all 25 of them) are shown in English words,  and not  a
single  letter  code  that  one has to waste time fumbling  through  a  manual
looking for to get an explanation.  Regardless of the I/O routes selected, all
errors  are written to the screen as well as any other chosen  output  medium.
With regard to disk I/O, the present version does not support directory paths.
Valid disk designations (A:, B:, etc.) may be freely used.

SVASM02 Version 1.0 will not do macros, but I am looking at the possibility of
including  macro  assembly  in  a  future  version.    SVASM02  will  not   do
parenthetical  or  hierarchical  arithmetic,  like "SCRADR  =  ((REW+WER)/ONE-
TWO)/DSF".   For  the  present  I recommend BASIC.   SVASM02  will  do  simple
arithmetic,  evaluating  an  expression from left to right as  the  terms  are
encountered.   For  example,  "TABADR  =  BASADR+OFFSET*2" will  be  correctly
evaluated  from left to right.   The value of OFFSET is added to the value  of
BASADR  and  the sum is multiplied by 2.   The four arithmetic  operators  are
supported as well as Boolean AND,  OR,  NOT,  and XOR; also SHL and SHR (shift
left  and  right);  MOD;  and byte extraction operators.   The left  to  right
restriction applies to the logical,  MOD, and shift operators also.  (A future
version  may  include  the capability to  do  hierarchical  and  parenthetical
evaluations,  as  well as relational operations like less than,  greater than,
etc.)

Labels and symbols are limited to six characters.  I know some assemblers will
use up to 40 or even more, but why take up a lot of file and memory space, and
CPU  time to evaluate them when a few characters will do the same  thing?   If
you  need a lot of characters to describe a variable or code sequence,  use  a
comment.  The CPU overhead is a lot less.  I could be talked into 8 characters
for a future version.

There are several pseudo-ops in SVASM02,  like .PAGE, .WIDTH, .INCLUDE, .CMOS,
conditional assembly, .LIST and .NOLIST, .PRINTC, and others.  These are fully
described in the DOC file.

The  TEST02.ASM file contains all of the 6502 and 65C02 instruction  mnemonics
and addressing modes,  as well as examples of syntax,  errors, etc.  This file
should be assembled and the output referred to to answer additional  questions
the programmer may have regarding the program or its operation.

SVASM02  has  been used to assemble 6502 code for the control of  two  amateur
radio repeaters,  at KM2H and N2CEH.  There are no known bugs.  That is not to
say  a bug may not be present.   Other programmers,  using their own style  of
programming technique,  may uncover a hidden bug.   If I am informed of a bona
fide  bug,  I  will fix it and return a free corrected version to the  sender,
whether he or she is a registered user or not.   This offer does not apply  to
users who use the program for profit unless they were originally registered.

A  software  project  of this magnitude is never really  complete.   There  is
always  something  more  one can add,  and I  plan  to  do  so.   Hierarchical
and parenthetical arithmetic (a bear to write!); relational operators like EQ,
NE, GT, LT, etc. (I've started looking at this one); more than 6 characters in
labels  and symbols (will require more RAM and disk space and slow the program
a  trifle);  and  maybe even macros (good-bye 64K memory use)  are  all  being
considered.   User comments and input will all be considered,  but frivolities
and user-specific functions will be rejectd.

Now  the  tough  part.   While  writing  SVASM02  was  a  labor  of  love,  it
nevertheless  was a labor.   I offer it to GEnie users and others for private,
non-commercial, non-profit use only.  If you find it useful, a contribution of
$25.00  would be appreciated.   If you intend to profit from its use,  I  must
also.   Upon  written  application  and receipt  of  $100.00,  a  license  for
commercial use of SVASM02 will be issued.   In either case, you will receive a
disk containing the latest versions of the code and documentation,  and future
notification of any new versions.  No future versions, even to fix a bug, will
be  released  to the public domain.  Future assemblers referred to  previously
will be released in their initial versions only.   At no time will the  source
code  be  released for any reason to anyone.   Please contact me for terms  of
commercial distribution.  I may be reached by amateur packet radio (rapid), by
U.S.  mail (less rapid), or by GE mail on GEnie (slowest, since I don't get to
log on as often as I would like).   Comments,  constructive or otherwise,  are
welcome and will be appreciated.  Please, no phone calls.


                U.S. Mail:    John Michne
                              10 Evergreen Ave.
                              Clifton Park, New York   12065

     Amateur Packet Radio:    KM2H @ WA2WNI-4
                          or  KM2H @ WA2PVV

       GEnie Mail Address:    KM2H