ksherlock / qasm Goto Github PK

View Code? Open in Web Editor NEW

This project forked from marketideas/qasm

1.0 1.0 0.0 9.18 MB

Merlin 8/16/16+ Compatible 6502/65816 assembler/linker suite in C++ for linux

License: Other

CMake 0.14% Makefile 0.10% C++ 49.45% C 20.83% Assembly 23.95% Shell 0.10% HTML 5.40% Batchfile 0.01% Clarion 0.02%

qasm's People

Contributors

Stargazers

Watchers

qasm's Issues

qlink and DS \ and ERR \

The DS \ handling needs work. Current logic:

LNK:

loads the file
stores handle, etc, into a file table
handles for DS \ and ERR \ ops
builds symbol table entries
merging handles together handled by SAV opcode I guess

The DS \ / ERR \ checking:

scan reloc table for DS \ and ERR \
DS: increases the handle size, moves the relocation records, and inserts fill bytes
fill bytes are based on size of the current file, not the segment [WRONG]
ends processing after DS \ opcode [WRONG - could be ERR \ after it]
Block move to move the relocation table moves count=fill byte count instead of relocation table size [WRONG]

Need to keep track of

current segment size
current file size (for LEN) [ignores DS adjustments]
current position (for POS), which is similar to segment size but can be reset and ignores DS adjustments.

DS \ should do fill bytes based on the current segment size, move relocation size byte

bad address mode with comma

Bad address mode in line: 1.
                     1              pea  #','

qlink size warnings

No warning when segment size > $ffff

entry symbols

shell -

    ENT symbol1,symbol2,...
    ...
symbol1 ...

```
symbols not being put in the REL file if ENT before the symbol.

label size limits?

qasm I.UDP.S

n.b. - only compiles if PLS macro (from RJBUTILS.MAC.S) is commented out...

TCP.EQU.S:


Assembling I.UDP.S

Assembling.


Duplicate label in line: 37 >134.

Duplicate label in line: 37 >164.

Duplicate label in line: 37 >165.

Duplicate label in line: 37 >166.

Misalignment in line: 37 >164.
                     >164 tcpDGMSICMPECHORQOUTADRL0    ;ICMP ECHORQ replies sent out


Misalignment in line: 37 >165.
                     >165 tcpDGMSICMPECHORPADRL0       ;ICMP ECHORPs in


Misalignment in line: 37 >166.
                     >166 tcpDGMSICMPECHORPBADIDADRL0  ;ICMP ECHORPs unclaimed


Misalignment in line: 37 >168.
                     >168 tcpDGMSUDP ADRL 0            ;UDPs OK (added to UDP queue)


Object saved as I.UDP.L,A$0185,L$0204,LNK

End of QuickASM assembly. 516 bytes, 8 errors, 2467 lines, 617 symbols.

Elapsed time = < 1 second.

```

label equ *

label equ * is equivalent to label, however label equ * doesn't introduce a new global scope (for :lables) in qasm.

bleh equ *
:local
another equ *
:local   ; duplicate label in qasm, ok in merlin

err in link script

should asm error bump the link error count?

da `*`

* needs to be reset after each da/dw/db/etc argument.

merlin:

                     1 
008000: 14 00 12 00  2              da   offset-*,offset-* 
008004: 00 00 00 00  3              ds   16           
008008: 00 00 00 00 00 00 00 00 
008010: 00 00 00 00 
                     4 offset                         
                     5

qasm:

                     1 
008000: 14 00 14 00  2              da   offset-*,offset-* 
008004: 00 00 00 00  3              ds   16           
008008: 00 00 00 00 00 00 00 00 
008010: 00 00 00 00 
                     4 offset                         
                     5

Marinetti issues

HOSTNAME ENT
 STR 'appleiigs'
 DS 1+30-*+HOSTNAME

Illegal relative address in line: 91.
                     91             DS   1+30-*+HOSTNAME

RASTAN source incompatibilities

rastan source code uses dw ^$00

merlin apparently treats that as the unary ^ operator. qasm treats it as a label character.

dw #^$00 is a workaround for qasm

^ is a legitimate label character (thanks, merlin).

spurious misalignment error

	rel
*	ext a
	jmp	a
b

Undefined label in line: 3.
                     3              jmp  a            
Misalignment in line: 4.
                     4 b

adding an a label (real or ext) eliminates the misalignment error.

input files > 65,535 bytes

seems to cause a hang.

dum segments lose bank information?

(QA Tools)

qasetcmdtbl
                 lda            7,s
                 stal           cmdtbl
                 lda            9,s
                 stal           cmdtbl+2
                 lda            5,s
                 sta            5+4,s
                 lda            3,s
                 sta            3+4,s
                 lda            1,s
                 sta            1+4,s
                 pla
                 pla
                 brl            noerror
...

DS_Start
                 dum            *                           ;allocated by loader

cmdline          ds             256
cmdhandle        ds             4                           ;handle to command table
cmdtbl           ds             4                           ;ptr to the internal cmd table
...

generated, loaded code:

12/558e: a3 07                          LDA $07,s
12/5590: 8f e6 c3 00                    STA $00c3e6
12/5594: a3 09                          LDA $09,s
12/5596: 8f e8 c3 00                    STA $00c3e8
12/559a: a3 05                          LDA $05,s
12/559c: 83 09                          STA $09,s

pls macro

PLS MAC
 do ]0
 LUP ]1
 PLA
 else
 PLA
 fin
 --^
 <<<

PLS : error (with do ]0)
PLS 0: error
PLS 1: generates 1 PLA
PLS 2: generates 3 PLAs

PLS x: generate 2x -1 PLAs (??)

file names

errors display a breadcrumb of line number (24 > 136). would be nice if it printed the current file name.

handling @

merlin has special treatment for the @ character within a LUP - it's replaced with A-Z. (if > 26 loops, will generate a label error).

Outside of a LUP, @ by itself is a hidden character. (need to verify again as my recollection isn't complete)

duplicate absolute labels

merlin seems to allow duplicate absolute labels if the value match.

bleh = 1
bleh = 1

macro expansion of quoted parameters

getvars in asm.cond.s

' and " are both macro parameter delimiters, but they can't be embedded within each other since it screws up the parameter counting mechanism.

first pass errors

errors on the first pass print the line number but not the line, maybe it should print the line as well.

ksherlock / qasm Goto Github PK

qasm's People

Contributors

Stargazers

Watchers

qasm's Issues

Recommend Projects

Recommend Topics

Recommend Org