1. Introduction

The history behind WLA DX, from the original author, Ville Helin:

I wrote this because I had never written an assembler before and I really needed a macro assembler which could compile the GB-Z80 code I wrote. ;) Gaelan Griffin needed real Z80 support for his SMS projects so I thought I could write WLA to be a little more open and nowadays it supports all the Z80 systems you can think of. You’ll just have to define the memorymap of the destination machine for your project. After fixing some bugs I thought I could add support for 6502 systems so all NES-people would get their share of WLA as well. After finishing that few people said they’d like 65816 support (they had SNES developing in mind) so I added support for that. And then I thought I should write a 6510 version of WLA as well…

This is my ideal GB-Z80 macro assembler (not in final form, not yet). ;) Tastes differ. Thus WLA! Notice that WLA was initially made for Game Boy developers so the GB-Z80 version and the rest differ a little.

Good to know about WLA DX:

Almost all rules that apply to Z80 source code processing with WLA DX apply also to 6502, 65C02, 6510, 65816, HUC6280 and SPC-700.

About the names: WLA DX means all the tools covered in this documentation. So WLA DX includes WLA GB-Z80/Z80/6502/65C02/6510/65816/HUC6280/SPC-700 macro assembler (what a horribly long name), WLAB, and WLALINK GB-Z80/Z80/6502/65C02/6510/65816/HUC6280/SPC-700 linker. We use plain WLA to refer to the macro assembler in this document.

There was WLAD, an GB-Z80 dissassembler, but it has been discontinued and removed from the project and the documentation.

Currently WLA can also be used as a patch tool. Just include the original ROM image into the project with .BACKGROUND and insert e.g., OVERWRITE .SECTION s to patch the desired areas. Output the data into a new ROM image and there you have it. 100% readable (asm coded) patches are reality!

Note that you can directly compile only object and library files. You must use WLALINK to link these (or only one, if you must) into a ROM/program file.

WLA DX’s old homepage: http://www.iki.fi/~vhelin/wla.html
WLA DX’s new homepage: https://github.com/vhelin/wla-dx

2. Assembler Directives

Here’s the order in which the data is placed into the output:

  1. Data and group 3 directives outside sections.
  2. Group 2 directives.
  3. Data and group 3 directives inside sections.
  4. Group 1 directives.
ALL All, GB-Z80, Z80, 6502, 65C02, 6510, 65816, HUC6280 and SPC-700 versions apply.
GB Only the GB-Z80 version applies.
GB8 Only the GB-Z80 and 65816 versions apply.
Z80 Only the Z80 version applies.
658 Only the 65816 version applies.
SPC Only the SPC-700 version applies.
65x Only the 6502, 65C02, 6510, 65816 and HUC6280 versions apply.
!GB Only the Z80, 6502, 65C02, 6510, 65816, HUC6280 and SPC-700 versions apply.

Group 1:

GB .COMPUTEGBCHECKSUM
Z80 .COMPUTESMSCHECKSUM
658 .COMPUTESNESCHECKSUM
Z80 .SDSCTAG 1.0, "DUNGEON MAN", "A wild dungeon exploration game", "Ville Helin"
Z80 .SMSTAG

Group 2:

GB .CARTRIDGETYPE 1
GB .COUNTRYCODE 1
GB .DESTINATIONCODE 1
GB .NINTENDOLOGO
GB .GBHEADER
Z80 .SMSHEADER
GB .COMPUTEGBCOMPLEMENTCHECK
ALL .EMPTYFILL $C9
658 .ENDEMUVECTOR
658 .ENDNATIVEVECTOR
658 .ENDSNES
ALL .EXPORT work_x
658 .FASTROM
658 .HIROM
GB .LICENSEECODENEW "1A"
GB .LICENSEECODEOLD $1A
658 .LOROM
GB8 .NAME "NAME OF THE ROM"
ALL .OUTNAME "other.o"
GB .RAMSIZE 0
GB .ROMDMG
GB .ROMGBC
GB .ROMGBCONLY
GB .ROMSGB
658 .SLOWROM
658 .SMC
658 .SNESEMUVECTOR
658 .SNESHEADER
658 .SNESNATIVEVECTOR

Group 3:

65x .16BIT
658 .24BIT
65x .8BIT
658 .ACCU 8
ALL .ASC "HELLO WORLD!"
ALL .ASCTABLE
ALL .ASCIITABLE
ALL .ASM
ALL .BACKGROUND "parallax.gb"
ALL .BANK 0 SLOT 1
ALL .BASE $80
ALL .BLOCK "Block1"
ALL .BR
ALL .BREAKPOINT
ALL .BYT 100, $30, %1000, "HELLO WORLD!"
ALL .ROMBANKSIZE $4000
ALL .DB 100, $30, %1000, "HELLO WORLD!"
ALL .DBM filtermacro 1, 2, "encrypt me"
ALL .DBCOS 0.2, 10, 3.2, 120, 1.3
ALL .DBRND 20, 0, 10
ALL .DBSIN 0.2, 10, 3.2, 120, 1.3
ALL .DEFINE IF $FF0F
ALL .DEF IF $FF0F
ALL .DS 256, $10
ALL .DSB 256, $10
ALL .DSTRUCT waterdrop INSTANCEOF water DATA "tingle", 40, 120
ALL .DSW 128, 20
ALL .DW 16000, 10, 255
ALL .DWM filtermacro 1, 2, 3
ALL .DWCOS 0.2, 10, 3.2, 1024, 1.3
ALL .DWRND 20, 0, 10
ALL .DWSIN 0.2, 10, 3.2, 1024, 1.3
ALL .ELSE
ALL .ENDA
ALL .ENDASM
ALL .ENDB
ALL .ENDE
ALL .ENDIF
ALL .ENDM
ALL .ENDME
ALL .ENDR
ALL .ENDRO
ALL .ENDS
ALL .ENDST
ALL .ENUM $C000
ALL .EQU IF $FF0F
ALL .FAIL
ALL .FCLOSE FP_DATABIN
ALL .FOPEN "data.bin" FP_DATABIN
ALL .FREAD FP_DATABIN DATA
ALL .FSIZE FP_DATABIN SIZE
ALL .IF DEBUG == 2
ALL .IFDEF IF
ALL .IFDEFM \2
ALL .IFEQ DEBUG 2
ALL .IFEXISTS "main.s"
ALL .IFGR DEBUG 2
ALL .IFGREQ DEBUG 1
ALL .IFLE DEBUG 2
ALL .IFLEEQ DEBUG 1
ALL .IFNDEF IF
ALL .IFNDEFM \2
ALL .IFNEQ DEBUG 2
ALL .INCBIN "sorority.bin"
ALL .INCDIR "/usr/programming/gb/include/"
ALL .INCLUDE "cgb_hardware.i"
658 .INDEX 8
ALL .INPUT NAME
ALL .MACRO TEST
ALL .MEMORYMAP
ALL .ORG $150
ALL .ORGA $150
ALL .PRINTT "Here we are...\n"
ALL .PRINTV DEC DEBUG+1
ALL .RAMSECTION "Vars" BANK 0 SLOT 1 ALIGN 4
ALL .REDEFINE IF $F
ALL .REDEF IF $F
ALL .REPEAT 6
ALL .REPT 6
ALL .ROMBANKMAP
ALL .ROMBANKS 2
ALL .SEED 123
ALL .SECTION "Init" FORCE
ALL .SHIFT
ALL .SLOT 1
ALL .STRUCT enemy_object
ALL .SYM SAUSAGE
ALL .SYMBOL SAUSAGE
ALL .UNBACKGROUND $1000 $1FFF
ALL .UNDEFINE DEBUG
ALL .UNDEF DEBUG
ALL .WORD 16000, 10, 255

Descriptions:

2.1. .8BIT

There are a few mnemonics that look identical, but take different sized arguments. Here’s a list of such 6502 mnemonics:

ADC, AND, ASL, BIT, CMP, CPX, CPY, DEC, EOR, INC, LDA, LDX, LDY, ORA, ROL, SBC, STA, STX and STY.

For example:

LSR 11       ; $46 $0B
LSR $A000    ; $4E $00 $A0

The first one could also be

LSR 11       ; $4E $0B $00

.8BIT is here to help WLA to decide to choose which one of the opcodes it selects. When you give .8BIT (default) no 8-bit address/value is expanded to 16-bits.

By default WLA uses the smallest possible size. This is true also when WLA finds a computation it can’t solve right away. WLA assumes the result will be inside the smallest possible bounds, which depends on the type of the mnemonic.

You can also use the fixed argument size versions of such mnemonics by giving the size with the operand (i.e., operand hinting). Here are few examples:

LSR 11.B   ; $46 $0B
LSR 11.W   ; $4E $0B $00

In WLA-65816 .ACCU/.INDEX/SEP/REP override .8BIT/.16BIT/.24BIT when considering the immediate values, so be careful. Still, operand hints override all of these, so use them to be sure.

This is not a compulsory directive.

2.2. .16BIT

Analogous to .8BIT. .16BIT forces all addresses and immediate values to be expanded into 16-bit range, when possible, that is.

LSR 11       ; $46 $0B

that would be the case, normally, but after .16BIT it becomes

LSR 11       ; $4E $0B $00

This is not a compulsory directive.

2.3. .24BIT

Analogous to .8BIT and .16BIT. .24BIT forces all addresses to be expanded into 24-bit range, when possible, that is.

AND $11       ; $25 $11

that would be the case, normally, but after .24BIT it becomes

AND $11       ; $2F $11 $00 $00

If it is not possible to expand the address into .24BIT range, then WLA tries to expand it into 16-bit range.

This is not a compulsory directive.

2.4. .ACCU 8

Forces WLA to override the accumulator size given with SEP/REP. .ACCU doesn’t produce any code, it only affects the way WLA interprets the immediate values (8 for 8 bit operands, 16 for 16 bit operands) for opcodes dealing with the accumulator.

So after giving .ACCU 8

AND #6

will produce $29 $06, and after giving .ACCU 16

AND #6

will yield $29 $00 $06.

Note that SEP/REP again will in turn reset the accumulator/index register size.

This is not a compulsory directive.

2.5. .INDEX 8

Forces WLA to override the index (X/Y) register size given with SEP/REP. .INDEX doesn’t produce any code, it only affects the way WLA interprets the immediate values (8 for 8 bit operands, 16 for 16 bit operands) for opcodes dealing with the index registers.

So after giving .INDEX 8

CPX #10

will produce $E0 $A0, and after giving .INDEX 16

CPX #10

will yield $E0 $00 $A0.

Note that SEP/REP again will in turn reset the accumulator/index register size.

This is not a compulsory directive.

2.6. .ASM

Tells WLA to start assembling. Use .ASM to continue the work which has been disabled with .ENDASM. .ASM and .ENDASM can be used to mask away big blocks of code. This is analogous to the ANSI C -comments (/*...*/), but .ASM and .ENDASM can be nested, unlike the ANSI C -counterpart.

This is not a compulsory directive.

2.7. .ENDASM

Tells WLA to stop assembling. Use .ASM to continue the work.

This is not a compulsory directive.

2.8. .DBRND 20, 0, 10

Defines bytes, just like .DSB does, only this time they are filled with (pseudo) random numbers. We use the integrated Mersenne Twister to generate the random numbers. If you want to seed the random number generator, use .SEED.

The first parameter (20 in the example) defines the number of random numbers we want to generate. The next two tell the range of the random numbers, i.e. min and max.

Here’s how it works:

.DBRND A, B, C

for (i = 0; i < A; i++)
  output_data((rand() % (C-B+1)) + B);

This is not a compulsory directive.

2.9. .DWRND 20, 0, 10

Analogous to .DBRND (but defines words).

This is not a compulsory directive.

2.10. .DBCOS 0.2, 10, 3.2, 120, 1.3

Defines bytes just like .DSB does, only this time they are filled with cosine data. .DBCOS takes five arguments.

The first argument is the starting angle. Angle value ranges from 0 to 359.999…, but you can supply WLA with values that are out of the range - WLA fixes them ok. The value can be integer or float.

The second one descibes the amount of additional angles. The example will define 11 angles.

The third one is the adder value which is added to the angle value when next angle is calculated. The value can be integer or float.

The fourth and fifth ones can be seen from the pseudo code below, which also describes how .DBCOS works. The values can be integer or float.

Remember that cos (and sin) here returns values ranging from -1 to 1.

.DBCOS A, B, C, D, E

for (B++; B > 0; B--) {
  output_data((D * cos(A)) + E)
  A = keep_in_range(A + C)
}

This is not a compulsory directive.

2.11. .DBSIN 0.2, 10, 3.2, 120, 1.3

Analogous to .DBCOS, but does sin() instead of cos().

This is not a compulsory directive.

2.12. .DWCOS 0.2, 10, 3.2, 1024, 1.3

Analogous to .DBCOS (but defines words).

This is not a compulsory directive.

2.13. .DWSIN 0.2, 10, 3.2, 1024, 1.3

Analogous to .DBCOS (but defines words and does sin() instead of cos()).

This is not a compulsory directive.

2.14. .NAME "NAME OF THE ROM"

If .NAME is used with WLA-GB then the 16 bytes ranging from $0134 to $0143 are filled with the provided string. WLA-65816 fills the 21 bytes from $FFC0 to $FFD4 in HiROM- and from $7FC0 to $7FD4 in LoROM-mode with the name string (SNES ROM title).

If the string is shorter than 16/21 bytes the remaining space is filled with $00.

This is not a compulsory directive.

2.15. .ROMBANKS 2

Indicates the size of the ROM in rombanks. This value is converted to a standard Gameboy ROM size indicator value found at $148 in a Gameboy ROM, and there this one is put into.

This is a compulsory directive unless .ROMBANKMAP is defined.

You can redefine .ROMBANKS as many times as you wish as long as the old and the new ROM bank maps match as much as possible. This way you can enlarge the size of the project on the fly.

2.16. .RAMSIZE 0

Indicates the size of the RAM. This is a standard Gameboy RAM size indicator value found at $149 in a Gameboy ROM, and there this one is put to also.

This is not a compulsory directive.

2.17. .EMPTYFILL $C9

This byte is used in filling the unused areas of the ROM file. EMPTYFILL defaults to $00.

This is not a compulsory directive.

2.18. .CARTRIDGETYPE 1

Indicates the type of the cartridge (mapper and so on). This is a standard Gameboy cartridge type indicator value found at $147 in a Gameboy ROM, and there this one is put to also.

This is not a compulsory directive.

2.19. .COUNTRYCODE 1

Indicates the country code located at $14A of a Gameboy ROM.

This is not a compulsory directive.

2.20. .DESTINATIONCODE 1

.DESTINATIONCODE is an alias for .COUNTRYCODE.

This is not a compulsory directive.

2.22. .GBHEADER

This begins the GB header definition, and automatically defines .COMPUTEGBCHECKSUM. End the header definition with .ENDGB. Here’s an example:

.GBHEADER
    NAME "TANKBOMBPANIC"  ; identical to a freestanding .NAME.
    LICENSEECODEOLD $34   ; identical to a freestanding .LICENSEECODEOLD.
    LICENSEECODENEW "HI"  ; identical to a freestanding .LICENSEECODENEW.
    CARTRIDGETYPE $00     ; identical to a freestanding .CARTRIDGETYPE.
    RAMSIZE $09           ; identical to a freestanding .RAMSIZE.
    COUNTRYCODE $01       ; identical to a freestanding .COUNTRYCODE.
    DESTINATIONCODE $01   ; identical to a freestanding .DESTINATIONCODE.
    NINTENDOLOGO          ; identical to a freestanding .NINTENDOLOGO.
    ROMDMG                ; identical to a freestanding .ROMDMG.
                          ; Alternatively, ROMGBC or ROMGBCONLY can be used
.ENDGB

This is not a compulsory directive.

2.23. .SMSHEADER

.SMSHEADER
    PRODUCTCODE 26, 70, 2 ; 2.5 bytes
    VERSION 1             ; 0-15
    REGIONCODE 4          ; 3-7
    RESERVEDSPACE 0, 0    ; 2 bytes
.ENDSMS

The REGIONCODE also defines the system:

3 SMS Japan
4 SMS Export
5 GG Japan
6 GG Export
7 GG International

When .SMSHEADER is defined, also the checksum is calculated, and TMR SEGA, two reserved bytes and ROM size are defined.

See http://www.smspower.org/Development/ROMHeader for more information about SMS header.

This is not a compulsory directive.

2.24. .LICENSEECODEOLD $1A

This is a standard old licensee code found at $14B in a Gameboy ROM, and there this one is put to also. .LICENSEECODEOLD cannot be defined with .LICENSEECODENEW.

This is not a compulsory directive.

2.25. .LICENSEECODENEW "1A"

This is a standard new licensee code found at $144 and $145 in a Gameboy ROM, and there this one is put to also. .LICENSEECODENEW cannot be defined with .LICENSEECODEOLD. $33 is inserted into $14B, as well.

This is not a compulsory directive.

2.26. .COMPUTEGBCHECKSUM

When this directive is used WLA computes the ROM checksum found at $14E and $14F in a Gameboy ROM. Note that this directive can only be used with WLA-GB.

Note that you can also write .COMPUTECHECKSUM (the old name for this directive), but it’s not recommended.

This is not a compulsory directive.

2.27. .COMPUTESMSCHECKSUM

When this directive is used WLA computes the ROM checksum found at $7FFA and $7FFB (or $3FFA - $3FFB is the ROM is 16KBs, or $1FFA - $1FFB for 8KB ROMs) in a SMS/GG ROM. Note that this directive can only be used with WLA-z80. Also note that the ROM size must be at least 8KBs. The checksum is calculated using bytes 0x0000 - 0x1FEF/0x3FEF/0x7FEF.

This is not a compulsory directive.

2.28. .COMPUTESNESCHECKSUM

When this directive is used WLA computes the SNES ROM checksum and inverse checksum found at $7FDC - $7FDF (LoROM) or $FFDC-$FFDF (HiROM). Note that this directive can only be used with WLA-65816. Also note that the ROM size must be at least 32KB for LoROM images and 64KB for HiROM images.

.LOROM or .HIROM must be issued before .COMPUTESNESCHECKSUM.

This is not a compulsory directive.

2.29. .SMSTAG

.SMSTAG forces WLA to write an ordinary SMS/GG ROM tag to the ROM file. Currently only the string TMR SEGA and ROM checksum are written (meaning that .SMSTAG also defines .COMPUTESMSCHECKSUM). The ROM size must be at least 8KBs.

This is not a compulsory directive.

2.30. .SDSCTAG 1.0, "DUNGEON MAN", "A wild dungeon exploration game", "Ville Helin"

.SDSCTAG adds SDSC tag to your SMS/GG ROM file. The ROM size must be at least 8KB just like with .COMPUTESMSCHECKSUM and .SMSTAG. For more information about this header take a look at http://www.smspower.org/dev/sdsc/. Here’s an explanation of the arguments:

.SDSCTAG {version number}, {program name}, {program release notes}, {program author}

Note that program name, release notes and program author can also be pointers to strings instead of being only strings (which WLA terminates with zero, and places them into suitable locations inside the ROM file). So

.SDSCTAG 0.8, PRGNAME, PRGNOTES, PRGAUTHOR
...
PRGNAME:  .DB "DUNGEON MAN", 0
PRGNOTES: .DB "A wild and totally crazy dungeon exploration game", 0
PRGAUTHOR:.DB "Ville Helin", 0

works also. All strings supplied explicitly to .SDSCTAG are placed somewhere in .BANK 0 SLOT 0.

.SDSCTAG 1.0, "", "", ""
.SDSCTAG 1.0, 0, 0, 0

are also valid, here 0 and "" mean the user doesn’t want to use any descriptive strings. Version number can also be given as an integer, but then the minor version number defaults to zero.

.SDSCTAG also defines .SMSTAG (as it’s part of the SDSC ROM tag specification).

This is not a compulsory directive.

2.31. .COMPUTEGBCOMPLEMENTCHECK

When this directive is used WLA computes the ROM complement check found at $14D in a Gameboy ROM.

Note that you can still use .COMPUTECOMPLEMENTCHECK (the old name for this directive), but it’s not recommended.

This is not a compulsory directive.

2.32. .INCDIR "/usr/programming/gb/include/"

Changes the current include root directory. Use this to specify main directory for the following .INCLUDE and .INCBIN directives. If you want to change to the current working directory (WLA also defaults to this), use

.INCDIR ""

If the INCDIR is specified in the command line, that directory will be searched before the .INCDIR in the file. If the file is not found, WLA will then silently search the specified .INCDIR.

This is not a compulsory directive.

2.33. .INCLUDE "cgb_hardware.i"

Includes the specified file to the source file. If the file’s not found in the .INCDIR directory, WLA tries to find it in the current working directory. If the INCDIR is specified in the command line, WLA will first try to find the file specified in that directory. Then proceed as mentioned before if it is not found.

This is not a compulsory directive.

2.34. .INCBIN "sorority.bin"

Includes the specified data file into the source file. .INCBIN caches all files into memory, so you can .INCBIN any data file millions of times, but it is loaded from hard drive only once.

You can optionally use SWAP after the file name, e.g.,

.INCBIN "kitten.bin" SWAP

.INCBIN data is divided into blocks of two bytes, and inside every block the bytes are exchanged (like SWAP r does to nibbles). This requires that the size of the file is even.

You can also force WLA to skip n bytes from the beginning of the file by writing for example:

.INCBIN "kitten.bin" SKIP 4

Four bytes are skipped from the beginning of kitten.bin and the rest is incbinned.

It is also possible to incbin only n bytes from a file:

.INCBIN "kitten.bin" READ 10

Will read ten bytes from the beginning of kitten.bin.

You can also force WLA to create a definition holding the size of the file:

.INCBIN "kitten.bin" FSIZE size_of_kitten

Want to circulate all the included bytes through a filter macro? Do this:

.INCBIN "kitten.bin" FILTER filtermacro

The filter macro is executed for each byte of the included data, data byte being the first argument, and offset from the beginning being the second parameter, just like in the case of .DBM and .DWM.

And you can combine all these four commands:

.INCBIN "kitten.bin" SKIP 10 READ 8 SWAP FSIZE size_of_kitten FILTER filtermacro

This example shows how to incbin eight bytes (swapped) after skipping 10 bytes from the beginning of file kitten.bin, and how to get the size of the file into a definition label size_of_kitten. All the data bytes are circulated through a filter macro.

Here’s an example of a filter macro that increments all the bytes by one:

.macro filtermacro    ; the input byte is \1, the output byte is in "_out"
.redefine _out \1+1
.endm

Note that the order of the extra commands is important.

If the file’s not found in the .INCDIR directory, WLA tries to find it in the current working directory. If the INCDIR is specified in the command line, WLA will first search for the file in that directory. If not found, it will then proceed as aforementioned.

This is not a compulsory directive.

2.35. .INPUT NAME

.INPUT is much like any Basic-language input: .INPUT asks the user for a value or string. After .INPUT is the variable name used to store the data.

.INPUT works like .REDEFINE, but the user gets to type in the data.

Here are few examples how to use input:

.PRINTT "The name of the ROM? "
.INPUT NAME
.NAME NAME

...

.PRINTT "Give the .DB amount.\n"
.INPUT S
.PRINTT "Give .DB data one at a time.\n"
.REPEAT S
  .INPUT B
  .DB B
.ENDR

...

This is not a compulsory directive.

2.36. .BACKGROUND "parallax.gb"

This chooses an existing ROM image (parallax.gb in this case) as a background data for the project. You can overwrite the data with OVERWRITE sections only, unless you first clear memory blocks with .UNBACKGROUND after which there’s room for other sections as well.

Note that .BACKGROUND can be used only when compiling an object file.

.BACKGROUND is useful if you wish to patch an existing ROM image with new code or data.

This is not a compulsory directive.

2.37. .UNBACKGROUND $1000 $1FFF

After issuing .BACKGROUND you might want to free some parts of the backgrounded ROM image for e.g., FREE sections. With .UNBACKGROUND you can define such regions. In the example a block starting at $1000 and ending at $1FFF was released (both ends included). You can issue .UNBACKGROUND as many times as you wish.

This is not a compulsory directive.

2.38. .FAIL

Terminates the compiling process.

This is not a compulsory directive.

2.39. .FCLOSE FP_DATABIN

Closes the filehandle FP_DATABIN.

This is not a compulsory directive.

2.40. .FOPEN "data.bin" FP_DATABIN

Opens the file data.bin for reading and associates the filehandle with name FP_DATABIN.

This is not a compulsory directive.

2.41. .FREAD FP_DATABIN DATA

Reads one byte from FP_DATABIN and creates a definition called DATA to hold it. DATA is an ordinary definition label, so you can .UNDEFINE it.

Here’s an example on how to use .FREAD:

.fopen "data.bin" fp
.fsize fp t
.repeat t
.fread fp d
.db d+26
.endr
.undefine t, d

This is not a compulsory directive.

2.42. .FSIZE FP_DATABIN SIZE

Creates a definition called SIZE, which holds the size of the file associated with the filehandle FP_DATABIN. SIZE is an ordinary definition label, so you can .UNDEFINE it.

This is not a compulsory directive.

2.43. .MACRO TEST

Begins a macro called TEST.

You can use \@ inside a macro to e.g., separate a label from the other macro TEST occurrences. \@ is replaced with an integer number indicating the amount of times the macro has been called previously so it is unique to every macro call. \@ can also be used inside strings inside a macro or just as a plain value. Look at the following examples for more information.

You can also type \! to get the name of the source file currently being parsed.

Also, if you want to use macro arguments in e.g., calculation, you can type \X where X is the number of the argument. Another way to refer to the arguments is to use their names given in the definition of the macro (see the examples for this).

Remember to use .ENDM to finish the macro definition. Note that you cannot use .INCLUDE inside a macro. Note that WLA’s macros are in fact more like procedures than real macros, because WLA doesn’t substitute macro calls with macro data. Instead WLA jumps to the macro when it encounters a macro call at compile time.

You can call macros from inside a macro. Note that the preprocessor does not expand the macros. WLA traverses through the code according to the macro calls.

Here are some examples:

.MACRO NOPMONSTER
       .REPT 32         ; evil...
       NOP
       .ENDR
.ENDM

.MACRO LOAD_ABCD
       LD A, \1
       LD B, \2
       LD C, \3
       LD D, \4
       NOPMONSTER
       LD HL, 1<<\1
.INCBIN \5
.ENDM

.MACRO QUEEN

QUEEN\@:
    LD   A, \1
    LD   B, \1
    CALL QUEEN\@

    .DB  "\@", 0     ; will translate into a zero terminated string
                         ; holding the amount of macro QUEEN calls.
        .DB  "\\@", 0    ; will translate into a string containing
                         ; \@.
        .DB  \@          ; will translate into a number indicating
                         ; the amount of macro QUEEN calls.

.ENDM

.MACRO LOAD_ABCD_2 ARGS ONE, TWO, THREE, FOUR, FIVE
       LD A, ONE
       LD B, TWO
       LD C, THREE
       LD D, FOUR
       NOPMONSTER
       LD HL, 1<<ONE
.INCBIN FIVE
.ENDM

.MACRO TEST NARGS 3
       .DB \1, \2, \3
.ENDM

And here’s how they can be used:

NOPMONSTER
LOAD_ABCD $10, $20, $30, XYZ, "merman.bin"
QUEEN 123
LOAD_ABCD_2 $10, $20, $30, XYZ, "merman.bin"
TEST 1, 2, 3

Note that you must separate the arguments with commas.

If you want to give names to the macro’s arguments you can do that by listing them in order after supplying ARGS after the macro’s name.

Every time a macro is called a definition NARGS is created. It shows only inside the macro and holds the number of arguments the macro was called with. So don’t have your own definition called NARGS. Here’s an example:

.MACRO LUPIN
  .IF NARGS != 1
    .FAIL
  .ENDIF

  .PRINTT "Totsan! Ogenki ka?\n"
.ENDM

This is not a compulsory directive.

2.44. .ENDM

Ends a .MACRO.

This is not a compulsory directive, but when .MACRO is used this one is required to terminate it.

2.45. .SHIFT

Shifts the macro arguments one down (\2 becomes \1, \3 -> \2, etc.). .SHIFT can thus only be used inside a .MACRO.

This is not a compulsory directive.

2.46. .FASTROM

Sets the ROM memory speed bit in $FFD5 (.HIROM) or $7FD5 (.LOROM) to indicate that the SNES ROM chips are 120ns chips.

This is not a compulsory directive.

2.47. .SLOWROM

Clears the ROM memory speed bit in $FFD5 (.HIROM) or $7FD5 (.LOROM) to indicate that the SNES ROM chips are 200ns chips.

This is not a compulsory directive.

2.48. .SMC

Forces WLALINK to compute a proper SMC header for the ROM file.

SMC header is a chunk of 512 bytes. WLALINK touches only its first three bytes, and sets the rest to zeroes. Here’s what will be inside the first three bytes:

Byte Description
0 low byte of 8KB page count.
1 high byte of 8KB page count.
2
  • Bit 7: 0
  • Bit 6: 0
  • Bit 5: 0 = LoROM, 1 = HiROM
  • Bit 4: 0 = LoROM, 1 = HiROM
  • Bit 3 and 2: SRAM size (00 = 256Kb, 01 = 64Kb, 10 = 16Kb, 11 = 0Kb)
  • Bit 1: 0
  • Bit 0: 0

This is not a compulsory directive.

2.49. .HIROM

With this directive you can define the SNES ROM mode to be HiROM. Issuing .HIROM will override the user’s ROM bank map when WLALINK computes 24-bit addresses and bank references. If no .HIROM or .LOROM are given then WLALINK obeys the banking defined in .ROMBANKMAP.

.HIROM also sets the ROM mode bit in $FFD5.

This is not a compulsory directive.

2.50. .LOROM

With this directive you can define the SNES ROM mode to be LoROM. Issuing .LOROM will override the user’s ROM bank map when WLALINK computes 24-bit addresses and bank references. If no .HIROM or .LOROM are given then WLALINK obeys the banking defined in .ROMBANKMAP.

WLA defaults to .LOROM.

This is not a compulsory directive.

2.51. .BASE $80

Defines the base value for the bank number (used only in 24-bit addresses and when getting a label’s bank number with :). Here are few examples of how to use .BASE (both examples assume the label resides in the first ROM bank):

.BASE $00
label1:
.BASE $80
label2:

  JSL label1   ; if label1 address is $1234, this will assemble into
               ; JSL $001234
  JSL label2   ; label2 is also $1234, but this time the result will be
               ; JSL $801234

.BASE defaults to $00. Note that the address of the label will also contribute to the bank number (bank number == .BASE + ROM bank of the label).

On 65816, use .LOROM or .HIROM to define the ROM mode.

This is not a compulsory directive.

2.52. .BLOCK "Block1"

Begins a block (called Block1 in the example). These blocks have only one function: to display the number of bytes they contain. When you embed such a block into your code, WLA displays its size when it assembles the source file.

Use .ENDB to terminate a .BLOCK. Note that you can nest .BLOCK s.

This is not a compulsory directive.

2.53. .ENDB

Terminates .BLOCK.

This is not a compulsory directive, but when .BLOCK is used this one is required to terminate it.

2.54. .BANK 0 SLOT 1

Defines the ROM bank and the slot it is inserted into in the memory. You can also type the following:

.BANK 0

This tells WLA to move into BANK 0 which will be put into the DEFAULTSLOT of .MEMORYMAP.

Every time you use .BANK, supply .ORG/.ORGA as well, just to make sure WLA calculates addresses correctly.

This is a compulsory directive.

2.55. .SLOT 1

Changes the currently active memory slot. This directive is meant to be used with SUPERFREE sections, where only the slot number is constant when placing the sections.

This is not a compulsory directive.

2.56. .ROMBANKSIZE $4000

Defines the ROM bank size. Old syntax is .BANKSIZE x.

This is a compulsory directive unless .ROMBANKMAP is defined.

2.57. .ORG $150

Defines the starting address. The value supplied here is relative to the ROM bank given with .BANK.

When WLA starts to parse a source file, .ORG is set to $0, but it’s always a good idea to explicitly use .ORG, for clarity.

This is a compulsory directive.

2.58. .ORGA $150

Defines the starting address. The value supplied here is absolute and used directly in address computations. WLA computes the right position in ROM file. By using .ORGA you can instantly see from the source file where the following code is located in the 16-bit memory.

Here’s an example:

.MEMORYMAP
SLOTSIZE $4000
DEFAULTSLOT 0
SLOT 0 $0000
SLOT 1 $4000
.ENDME

.ROMBANKMAP
BANKSTOTAL 2
BANKSIZE $4000
BANKS 2
.ENDRO

.BANK 0 SLOT 1
.ORGA $4000

MAIN:       JP      MAIN

Here MAIN is at $0000 in the ROM file, but the address for label MAIN is $4000. By using .ORGA instead of .ORG, you can directly see from the value the address where you want the code to be as .ORG is just an offset to the SLOT.

2.59. .DS 256, $10

.DS is an alias for .DSB.

This is not a compulsory directive.

2.60. .DSB 256, $10

Defines 256 bytes of $10.

This is not a compulsory directive.

2.61. .DSTRUCT waterdrop INSTANCEOF water DATA "tingle", 40, 120

Defines an instance of struct water, called waterdrop, and fills it with the given data. Before calling .DSTRUCT we must have defined the structure, and in this example it could be like:

.STRUCT water
name   ds 8
age    db
weight dw
.ENDST

Note that the keywords INSTANCEOF and DATA are optional, so

.DSTRUCT waterdrop, water, "tingle", 40, 120

also works. And one can define instances without supplying values to all struct members:

.DSTRUCT waterdrop, water, "somedrop"

Note that WLA fills the missing bytes with the data defined with .EMPTYFILL, or $00 if no .EMPTYFILL has been issued.

In this example you would also get the following labels:

waterdrop
waterdrop.name
waterdrop.age
waterdrop.weight

This is not a compulsory directive.

2.62. .DSW 128, 20

Defines 128 words (two bytes) of 20.

This is not a compulsory directive.

2.63. .DB 100, $30, %1000, "HELLO WORLD!"

Defines bytes.

This is not a compulsory directive.

2.64. .BYT 100, $30, %1000, "HELLO WORLD!"

.BYT is an alias for .DB.

This is not a compulsory directive.

2.65. .DBM filtermacro 1, 2, "encrypt me"

Defines bytes using a filter macro. All the data is passed to filtermacro in the first argument, one byte at a time, and the byte that actually gets defined is the value of definition _OUT (_out works as well). The second macro argument holds the offset from the beginning (the first byte) in bytes (the series being 0, 1, 2, 3, …).

Here’s an example of a filter macro that increments all the bytes by one:

.macro increment
.redefine _out \1+1
.endm

This is not a compulsory directive.

2.66. .SYM SAUSAGE

WLA treats symbols (SAUSAGE in this example) like labels, but they only appear in the symbol files WLALINK outputs. Useful for finding out the location where WLALINK puts data.

This is not a compulsory directive.

2.67. .SYMBOL SAUSAGE

.SYMBOL is an alias for .SYM.

This is not a compulsory directive.

2.68. .BR

Inserts a breakpoint that behaves like a .SYM without a name. Breakpoints can only be seen in WLALINK’s symbol file.

This is not a compulsory directive.

2.69. .BREAKPOINT

.BREAKPOINT is an alias for .BR.

This is not a compulsory directive.

2.70. .ASCIITABLE

.ASCIITABLE’s only purpose is to provide character mapping for .ASC. Take a look at the example:

.ASCIITABLE
MAP "A" TO "Z" = 0
MAP "!" = 90
.ENDA

Here we set such a mapping that character A is equal to 0, B is equal to 1, C is equal to 2, and so on, and ! is equal to 90.

After you’ve given the .ASCIITABLE, use .ASC to define bytes using this mapping (.ASC is an alias for .DB, but with .ASCIITABLE mapping). For example, .ASC "ABZ" would define bytes 0, 1 and 25.

Note that the following works as well:

.ASCIITABLE
MAP 'A' TO 'Z' = 0
MAP 65 = 90          ; 65 is the decimal for ASCII 'A'
.ENDA

Also note that the characters that are not given any mapping in .ASCIITABLE map to themselves (i.e., A maps to A, etc.).

This is not a compulsory directive.

2.71. .ENDA

Ends the ASCII table.

This is not a compulsory directive, but when .ASCIITABLE or .ASCTABLE are used this one is required to terminate them.

2.72. .ASCTABLE

.ASCTABLE is an alias for .ASCIITABLE.

This is not a compulsory directive.

2.73. .ASC "HELLO WORLD!"

.ASC is an alias for .DB, but if you use .ASC it will remap the characters using the mapping given via .ASCIITABLE.

This is not a compulsory directive.

2.74. .DW 16000, 10, 255

Defines words (two bytes each). .DW takes only numbers and characters as input, not strings.

This is not a compulsory directive.

2.75. .WORD 16000, 10, 255

.WORD is an alias for .DW.

This is not a compulsory directive.

2.76. .DWM filtermacro 1, 2, 3

Defines 16-bit words using a filter macro. All the data is passed to filtermacro in the first argument, one word at a time, and the word that actually gets defined is the value of definition _OUT (_out works as well). The second macro argument holds the offset from the beginning (the first word) in bytes (the series being 0, 2, 4, 6, …).

Here’s an example of a filter macro that increments all the words by one:

.macro increment
.redefine _out \1+1
.endm

This is not a compulsory directive.

2.77. .DEFINE IF $FF0F

Assigns a number or a string to a definition label.

By default all defines are local to the file where they are presented. If you want to make the definition visible to all the files in the project, use .EXPORT.

WARNING: Please declare your definition lexically before using it as otherwise the assembler might make incorrect assumptions about its value and size and choose e.g. wrong opcodes and generate binary that doesn’t run properly.

Here are some examples:

.DEFINE X 1000
.DEFINE FILE "level01.bin"
.DEFINE TXT1 "hello and welcome", 1, "to a new world...", 0
.DEFINE BYTES 1, 2, 3, 4, 5
.DEFINE COMPUTATION X+1
.DEFINE DEFAULTV

All definitions with multiple values are marked as data strings, and .DB is about the only place where you can later on use them.

.DEFINE BYTES 1, 2, 3, 4, 5
.DB 0, BYTES, 6

is the same as

.DB 0, 1, 2, 3, 4, 5, 6

If you omit the definition value (in our example DEFAULTV), WLA will default to 0.

Note that you must do your definition before you use it, otherwise WLA will use the final value of the definition. Here’s an example of this:

.DEFINE AAA 10
.DB AAA            ; will be 10.
.REDEFINE AAA 11

but

.DB AAA            ; will be 11.
.DEFINE AAA 10
.REDEFINE AAA 11

You can also create definitions on the command line. Here’s an example of this:

wla-gb -vl -DMOON -DNAME=john -DPRICE=100 -DADDRESS=$100 math.s

MOON’s value will be 0, NAME is a string definition with value john, PRICE’s value will be 100, and ADDRESS’s value will be $100.

Note that

.DEFINE AAA = 10   ; the same as ".DEFINE AAA 10".

works as well.

This is not a compulsory directive.

2.78. .DEF IF $FF0F

.DEF is an alias for .DEFINE.

This is not a compulsory directive.

2.79. .EQU IF $FF0F

.EQU is an alias for .DEFINE.

This is not a compulsory directive.

2.80. .REDEFINE IF $0F

Assigns a new value or a string to an old definition. If the definition doesn’t exist, .REDEFINE performs .DEFINE’s work.

When used with .REPT REDEFINE helps creating tables:

.DEFINE CNT 0

.REPT 256
.DB CNT
.REDEFINE CNT CNT+1
.ENDR

This is not a compulsory directive.

2.81. .REDEF IF $0F

.REDEF is an alias for .REDEFINE.

This is not a compulsory directive.

2.82. .IF DEBUG == 2

If the condition is fulfilled the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Operands must be immediate values or strings.

The following operators are supported:

< less than
<= less or equal to
> greater than
>= greater or equal to
== equals to
!= doesn’t equal to

All IF directives (yes, including .IFDEF, .IFNDEF, etc) can be nested. They can also be used within ENUM s, RAMSECTION s, STRUCT s, ROMBANKMAP s, and most other directives that occupy multiple lines.

This is not a compulsory directive.

2.83. .IFDEF IF

If IF is defined, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped.

This is not a compulsory directive.

2.84. .IFEXISTS "main.s"

If main.s file can be found, then the following piece of code is acknowledged until .ENDIF/.LESE occurs in the text, otherwise it is skipped.

By writing the following few lines you can include a file if it exists without breaking the compiling loop if it doesn’t exist.

.IFEXISTS FILE
.INCLUDE FILE
.ENDIF

This is not a compulsory directive.

2.85. .UNDEFINE DEBUG

Removes the supplied definition label from system. If there is no such label as given no error is displayed as the result would be the same.

You can undefine as many definitions as you wish with one .UNDEFINE:

.UNDEFINE NUMBER, NAME, ADDRESS, COUNTRY
.UNDEFINE NAME, AGE

This is not a compulsory directive.

2.86. .UNDEF DEBUG

.UNDEF is an alias for .UNDEFINE.

This is not a compulsory directive.

2.87. .IFNDEF IF

If IF is not defined, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped.

This is not a compulsory directive.

2.88. .IFDEFM \2

If the specified argument is defined (argument number two, in the example), then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the macro, otherwise it is skipped.

This is not a compulsory directive. .IFDEFM works only inside a macro.

2.89. .IFNDEFM \2

If the specified argument is not defined, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the macro, otherwise it is skipped.

This is not a compulsory directive. .IFNDEFM works only inside a macro.

2.90. .IFEQ DEBUG 2

If the value of DEBUG equals to 2, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.91. .IFNEQ DEBUG 2

If the value of DEBUG doesn’t equal to 2, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.92. .IFLE DEBUG 2

If the value of DEBUG is less than 2, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.93. .IFLEEQ DEBUG 2

If the value of DEBUG is less or equal to 2, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.94. .IFGR DEBUG 2

If the value of DEBUG is greater than 2, then the following piece of code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.95. .IFGREQ DEBUG 2

If the value of DEBUG is greater or equal to 2, then the following pieceof code is acknowledged until .ENDIF/.ELSE occurs in the text, otherwise it is skipped. Both arguments can be computations, defines or immediate values.

This is not a compulsory directive.

2.96. .ELSE

If the previous .IFxxx failed then the following text until .ENDIF is acknowledged.

This is not a compulsory directive.

2.97. .ENDIF

This terminates any .IFxxx directive.

This is not a compulsory directive, but if you use any .IFxxx then you need also to apply this.

2.98. .REPEAT 6

Repeats the text enclosed between .REPEAT x and .ENDR x times (6 in this example). You can use .REPEAT s inside .REPEAT s. x must be bigger or equal than 0.

It’s also possible to have the repeat counter/index in a definition:

.REPEAT 6 INDEX COUNT
.DB COUNT
.ENDR

This would define bytes 0, 1, 2, 3, 4 and 5.

This is not a compulsory directive.

2.99. .REPT 6

.REPT is an alias for .REPEAT.

This is not a compulsory directive.

2.100. .ENDR

Ends the repetition.

This is not a compulsory directive, but when .REPEAT is used this one is required to terminate it.

2.101. .ENUM $C000

Starts enumeration from $C000. Very useful for defining variables.

To start a descending enumeration, put DESC after the starting address. WLA defaults to ASC (ascending enumeration).

You can also add EXPORT after these if you want to export all the generated definitions automatically.

Here’s an example of .ENUM:

...
.STRUCT mon                ; check out the documentation on
name ds 2                  ; .STRUCT
age  db
.ENDST

.ENUM $A000
_scroll_x DB               ; db  - define byte (byt and byte work also)
_scroll_y DB
player_x: DW               ; dw  - define word (word works also)
player_y: DW
map_01:   DS  16           ; ds  - define size (bytes)
map_02    DSB 16           ; dsb - define size (bytes)
map_03    DSW  8           ; dsw - define size (words)
monster   INSTANCEOF mon 3 ; three instances of structure mon
dragon    INSTANCEOF mon   ; one mon
.ENDE
...

Previous example transforms into following definitions:

.DEFINE _scroll_x      $A000
.DEFINE _scroll_y      $A001
.DEFINE player_x       $A002
.DEFINE player_y       $A004
.DEFINE map_01         $A006
.DEFINE map_02         $A016
.DEFINE map_03         $A026
.DEFINE monster        $A036
.DEFINE monster.name   $A036
.DEFINE monster.age    $A038
.DEFINE monster.1      $A036
.DEFINE monster.1.name $A036
.DEFINE monster.1.age  $A038
.DEFINE monster.2      $A039
.DEFINE monster.2.name $A039
.DEFINE monster.2.age  $A03B
.DEFINE monster.3      $A03C
.DEFINE monster.3.name $A03C
.DEFINE monster.3.age  $A03E
.DEFINE dragon         $A03F
.DEFINE dragon.name    $A03F
.DEFINE dragon.age     $A041

DB, DW, DS, DSB, DSW and INSTANCEOF can also be in lowercase. You can also use a dotted version of the symbols, but it doesn’t advance the memory address. Here’s an example:

.ENUM $C000 DESC EXPORT
bigapple_h db
bigapple_l db
bigapple:  .dw
.ENDE

And this is what is generated:

.DEFINE bigapple_h $BFFF
.DEFINE bigapple_l $BFFE
.DEFINE bigapple   $BFFE
.EXPORT bigapple, bigapple_l, bigapple_h

This way you can generate a 16-bit variable address along with pointers to its parts.

If you want more flexible variable positioning, take a look at .RAMSECTION s.

This is not a compulsory directive.

2.102. .ENDE

Ends the enumeration.

This is not a compulsory directive, but when .ENUM is used this one is required to terminate it.

2.103. .STRUCT enemy_object

Begins the definition of a structure. These structures can be placed inside RAMSECTION s and ENUM s. Here’s an example:

.STRUCT enemy_object
id    dw               ; the insides of a .STRUCT are 1:1 like in .ENUM
x     db               ; except that no structs inside structs are
y     db               ; allowed.
data  ds  10
info  dsb 16
stats dsw  4
.ENDST

This also creates a definition _sizeof_[struct name], in our example this would be _sizeof_enemy_object, and the value of this definition is the size of the object, in bytes (2+1+1+10+16+4*2 = 38 in the example).

You’ll get the following definitions as well:

enemy_object.id    (== 0)
enemy_object.x     (== 2)
enemy_object.y     (== 3)
enemy_object.data  (== 4)
enemy_object.info  (== 14)
enemy_object.stats (== 30)

After defining a .STRUCT you can create an instance of it in a .RAMSECTION / .ENUM by typing

<instance name> INSTANCEOF <struct name> [optional, the number of structures]

Here’s an example:

.RAMSECTION "enemies" BANK 4 SLOT 4
enemies   INSTANCEOF enemy_object 4
enemyman  INSTANCEOF enemy_object
enemyboss INSTANCEOF enemy_object
.ENDS

This will create labels like enemies, enemies.id, enemies.x, enemies.y and so on. Label enemies is followed by four enemy_object structures, and only the first one is labeled. After there four come enemyman and enemyboss instances.

Take a look at the documentation on .RAMSECTION & .ENUM, they have more examples of how you can use .STRUCT s.

A WORD OF WARNING: Don’t use labels b, B, w and W inside a structure as e.g., WLA sees enemy.b as a byte sized reference to enemy. All other labels should be safe.

lda enemy1.b  ; load a byte from zeropage address enemy1 or from the address
              ; of enemy1.b??? i can't tell you, and WLA can't tell you...

This is not a compulsory directive.

2.104. .ENDST

Ends the structure definition.

This is not a compulsory directive, but when .STRUCT is used this one is required to terminate it.

2.105. .MEMORYMAP

Begins the memory map definition. Using .MEMORYMAP you must first describe the target system’s memory architecture to WLA before it can start to compile the code. .MEMORYMAP gives you the freedom to use WLA Z80/6502/65C02/6510/65816/HUC6280/SPC-700 to compile data for numerous different real Z80/6502/65C02/6510/65816/HUC6280/SPC-700 based systems.

Examples:

.MEMORYMAP
DEFAULTSLOT 0
SLOTSIZE $4000
SLOT 0 $0000
SLOT 1 $4000
.ENDME

.MEMORYMAP
DEFAULTSLOT 0
SLOT 0 $0000 $4000
SLOT 1 $4000 $4000
.ENDME

.MEMORYMAP
DEFAULTSLOT 0
SLOT 0 START $0000 SIZE $4000
SLOT 1 START $4000 SIZE $4000
.ENDME

.MEMORYMAP
DEFAULTSLOT 1
SLOTSIZE $6000
SLOT 0 $0000
SLOTSIZE $2000
SLOT 1 $6000
SLOT 2 $8000
.ENDME

Here’s a real life example from Adam Klotblixt. It should be interesting for all the ZX81 coders:

...

.MEMORYMAP
DEFAULTSLOT 1
SLOTSIZE $2000
SLOT 0 $0000
SLOTSIZE $6000
SLOT 1 $2000
.ENDME

.ROMBANKMAP
BANKSTOTAL 2
BANKSIZE $2000
BANKS 1
BANKSIZE $6000
BANKS 1
.ENDRO

.BANK 1 SLOT 1
.ORGA $2000

...

SLOTSIZE defines the size of the following slots, unless you explicitly specify the size of the slot, like in the second and third examples. You can redefine SLOTSIZE as many times as you wish.

DEFAULTSLOT describes the default slot for banks which aren’t explicitly inserted anywhere. Check .BANK definition for more information.

SLOT defines a slot and its starting address. SLOT numbering starts at 0 and ends to 255 so you have 256 slots at your disposal.

This is a compulsory directive, and make sure all the object files share the same .MEMORYMAP or you can’t link them together.

Note that both START and SIZE are optional!

2.106. .ENDME

Terminates .MEMORYMAP.

This is not a compulsory directive, but when .MEMORYMAP is used this one is required to terminate it.

2.107. .ROMBANKMAP

Begins the ROM bank map definition. You can use this directive to describe the project’s ROM banks. Use .ROMBANKMAP when not all the ROM banks are of equal size. Note that you can use .ROMBANKSIZE and .ROMBANKS instead of .ROMBANKMAP, but that’s only when the ROM banks are equal in size. Some systems based on a real Z80 chip, 6502/65C02/6510/65816/HUC6280/SPC-700 CPUs and Pocket Voice cartridges for Game Boy require the usage of this directive.

Examples:

.ROMBANKMAP
BANKSTOTAL 16
BANKSIZE $4000
BANKS 16
.ENDRO

.ROMBANKMAP
BANKSTOTAL 510
BANKSIZE $6000
BANKS 1
BANKSIZE $2000
BANKS 509
.ENDRO

The first one describes an ordinary ROM image of 16 equal sized banks. The second one defines a 4MB Pocket Voice ROM image. In the PV ROM image the first bank is $6000 bytes and the remaining 509 banks are smaller ones, $2000 bytes each.

BANKSTOTAL tells the total amount of ROM banks. It must be defined prior to anything else.

BANKSIZE tells the size of the following ROM banks. You can supply WLA with BANKSIZE as many times as you wish.

BANKS tells the amount of banks that follow and that are of the size BANKSIZE which has been previously defined.

This is not a compulsory directive when .ROMBANKSIZE and .ROMBANKS are defined.

You can redefine .ROMBANKMAP as many times as you wish as long as the old and the new ROM bank maps match as much as possible. This way you can enlarge the size of the project on the fly.

2.108. .ENDRO

Ends the rom bank map.

This is not a compulsory directive, but when .ROMBANKMAP is used this one is required to terminate it.

2.109. .SEED 123

Seeds the random number generator.

This is not a compulsory directive. The random number generator is initially seeded with the output of time(), which is, according to the manual, the time since the Epoch (00:00:00 UTC, January 1, 1970), measured in seconds. So if you don’t .SEED the random number generator yourself with a constant value, .DBRND and .DWRND give you different values every time you run WLA.

In WLA DX 9.4a and before we used the stdlib’s srand() and rand() functions making the output differ on different platforms. Since v9.4 WLA DX contains its own Mersenne Twister pseudo random number generator.

2.110. .SECTION "Init" FORCE

Section is a continuous area of data which is placed into the output file according to the section type and .BANK and .ORG directive values.

The example begins a section called Init. Before a section can be declared, .BANK and .ORG must be used unless WLA is in library file output mode. Library file’s sections must all be FREE ones. .BANK tells the bank number where this section will be later relocated into. .ORG tells the offset for the relocation from the beginning of .BANK.

You can put sections inside a namespace. For instance, if you put a section into a namespace called bank0, then labels in that section can be accessed with bank0.label. This is not necessary inside the section itself. The namespace directive should immediately follow the name.

.SECTION "Init" NAMESPACE "bank0"

You can give the size of the section the following way:

.SECTION "Init" SIZE 100 FREE

It’s possible to force WLALINK to align the FREE, SEMIFREE and SUPERFREE sections by giving the alignment as follows:

.SECTION "Init" SIZE 100 ALIGN 4 FREE

And if you want that WLA returns the ORG to what it was before issuing the section, put RETURNORG at the end of the parameter list:

.SECTION "Init" SIZE 100 ALIGN 4 FREE RETURNORG

By default WLA advances the ORG, so, for example, if your ORG was $0 before a section of 16 bytes, then the ORG will be 16 after the section.

Note also that if your section name begins with double underlines (e.g., __UNIQUE_SECTION!!!) the section will be unique in the sense that when WLALINK recieves files containing sections which share the same name, WLALINK will save only the first of them for further processing, all others are deleted from memory with corresponding labels, references and calculations.

If a section name begins with an exclamation mark (!) it tells WLALINK to not to drop it, even if you use WLALINK’s ability to discard all unreferenced sections and there are no references to the section.

FORCE after the name of the section tells WLA that the section must be inserted so it starts at .ORG. FORCE can be replaced with FREE which means that the section can be inserted somewhere in the defined bank, where there is room. You can also use OVERWRITE to insert the section into the memory regardless of data collisions. Using OVERWRITE you can easily patch an existing ROM image just by .BACKGROUND’ing the ROM image and inserting OVERWRITE sections into it. SEMIFREE sections are also possible and they behave much like FREE sections. The only difference is that they are positioned somewhere in the bank starting from .ORG. SEMISUBFREE sections on the other hand are positioned somewhere in the bank starting from $0 and ending to .ORG.

SUPERFREE sections are also available, and they will be positioned into the first suitable place inside the first suitable bank (candidates for these suitable banks have the same size with the slot of the section, no other banks are considered). You can also leave away the type specifier as the default type for the section is FREE.

You can name the sections as you wish, but there is one special name. A section called BANKHEADER is placed in the front of the bank where it is defined. These sections contain data that is not in the memory map of the machine, so you can’t refer to the data of a BANKHEADER section, but you can write references to outside. So no labels inside BANKHEADER sections. These special sections are useful when writing e.g., MSX programs. Note that library files don’t take BANKHEADER sections.

Here’s an example of a BANKHEADER section:

.BANK 0
.ORG 0
.SECTION "BANKHEADER"
    .DW MAIN
    .DW VBI
.ENDS

.SECTION "Program"
MAIN:        CALL   MONTY_ON_THE_RUN
VBI:         PUSH   HL
     ...
     POP    HL
     RETI
.ENDS

Here’s an example of an ordinary section:

.BANK 0
.ORG $150
.SECTION "Init" FREE
        DI
        LD  SP, $FFFE
        SUB A
        LD  ($FF00+R_IE), A
.ENDS

This tells WLA that a FREE section called Init must be located somewhere in bank 0. If you replace FREE with SEMIFREE the section will be inserted somewhere in the bank 0, but not in the $0 - $14F area. If you replace FREE with SUPERFREE the section will be inserted somewhere in any bank with the same size as bank 0.

Here’s the order in which WLA writes the sections:

  1. FORCE
  2. SEMISUBFREE
  3. SEMIFREE & FREE
  4. SUPERFREE
  5. OVERWRITE

Before the sections are inserted into the output file, they are sorted by size, so that the biggest section gets processed first and the smallest last.

You can also create a RAM section. For more information about them, please read the .RAMSECTION directive explanation.

It is also possible to merge two or more sections using APPENDTO:

.SECTION "Base"
.DB 0
.ENDS

.SECTION "AppendToBase" FREE RETURNORG APPENDTO "Base"
.DB 1
.ENDS

This is not a compulsory directive.

2.111. .RAMSECTION "Vars" BANK 0 SLOT 1 ALIGN 4

RAMSECTION s accept only variable labels and variable sizes, and the syntax to define these is identical to .ENUM (all the syntax rules that apply to .ENUM apply also to .RAMSECTION). Additionally you can embed structures (.STRUCT) into a RAMSECTION. Here’s an example:

.RAMSECTION "Some of my variables" BANK 0 SLOT 1
vbi_counter:   db
player_lives:  db
.ENDS

RAMSECTION s behave like FREE sections, but instead of filling any banks RAM sections will occupy area inside slots. You can fill different slots with different variable labels. It’s recommend that you create separate slots for holding variables (as ROM and RAM don’t usually overlap). Here’s another example:

.MEMORYMAP
SLOTSIZE $4000
DEFAULTSLOT 0
SLOT 0 $0000   ; ROM slot 0.
SLOT 1 $4000   ; ROM slot 1.
SLOT 2 $A000   ; variable RAM is here!
.ENDME

.STRUCT game_object
x DB
y DB
.ENDST

.RAMSECTION "vars 1" BANK 0 SLOT 2
moomin1   DW
phantom   DB
nyanko    DB
enemy     INSTANCEOF game_object
.ENDS

.RAMSECTION "vars 2" BANK 1 SLOT 2
moomin2   DW
.ENDS

.RAMSECTION "vars 3" BANK 1 SLOT 2
moomin3   DW
.ENDS

If no other RAM sections are used, then this is what you will get:

.DEFINE moomin1 $A000
.DEFINE phantom $A002
.DEFINE nyanko  $A003
.DEFINE enemy   $A004
.DEFINE enemy.x $A004
.DEFINE enemy.y $A005
.DEFINE moomin2 $A000
.DEFINE moomin3 $A002

BANK in .RAMSECTION is optional so you can leave it away if you don’t switch RAM banks, or the target doesn’t have them.

It is also possible to merge two or more sections using APPENDTO:

.RAMSECTION "RAMSection1" BANK 0 SLOT 0
label1    DB
.ENDS

.RAMSECTION "RAMSection2" APPENDTO "RAMSection1"
label2    DB
.ENDS

This is not a compulsory directive.

2.112. .ENDS

Ends the section.

This is not a compulsory directive, but when .SECTION is used this one is required to terminate it.

2.113. .ROMGBC

Inserts data into the specific ROM location to mark the ROM as a dual-mode ROM ($80 -> $0143, so ROM name is max. 15 characters long). It will run in either DMG or GBC mode.

This is not a compulsory directive.

2.114. .ROMGBCONLY

Inserts data into the specific ROM location to mark the ROM as a Gameboy Color ROM ($C0 -> $0143, so ROM name is max. 15 characters long). It will only run in GBC mode.

This is not a compulsory directive.

2.115. .ROMDMG

Inserts data into the specific ROM location to mark the ROM as a DMG (Gameboy) ROM ($00 -> $0146). It will only run in DMG mode.

This is not a compulsory directive. .ROMDMG cannot be used with .ROMSGB.

2.116. .ROMSGB

Inserts data into the specific ROM location to mark the ROM as a Super Gameboy enhanced ROM ($03 -> $0146).

This is not a compulsory directive. .ROMSGB cannot be used with .ROMDMG.

2.117. .EXPORT work_x

Exports the definition work_x to outside world. Exported definitions are visible to all object files and libraries in the linking procedure. Note that you can only export value definitions, not string definitions.

You can export as many definitions as you wish with one .EXPORT:

.EXPORT NUMBER, NAME, ADDRESS, COUNTRY
.EXPORT NAME, AGE

This is not a compulsory directive.

2.118. .PRINTT "Here we are...\n"

Prints the given text into stdout. Good for debugging stuff. PRINTT takes only a string as argument, and the only supported formatting symbol is \n (line feed).

This is not a compulsory directive.

2.119. .PRINTV DEC DEBUG+1

Prints the value of the supplied definition or computation into stdout. Computation must be solvable at the time of printing (just like definitions values). PRINTV takes two parameters. The first describes the type of the print output. DEC means decimal, HEX means hexadecimal.

Use PRINTV with PRINTT as PRINTV doesn’t print linefeeds, only the result. Here’s an example:

.PRINTT "Value of \"DEBUG\" = $"
.PRINTV HEX DEBUG
.PRINTT "\n"

This is not a compulsory directive.

2.120. .OUTNAME "other.o"

Changes the name of the output file. Here’s and example:

wla-gb -o test.o test.s

would normally output test.o, but if you had written

.OUTNAME "new.o"

somewhere in the code WLA would write the output to new.o instead.

This is not a compulsory directive.

2.121. .SNESHEADER

This begins the SNES header definition, and automatically defines .COMPUTESNESCHECKSUM. From here you may define any of the following:

  • ID "ABCD" - inserts a one to four letter string starting at $7FB2 (lorom) or $FFB2 (hirom).

  • NAME "Hello World!" - identical to a freestanding .NAME.

  • LOROM - identical to a freestanding .LOROM.

  • HIROM - identical to a freestanding .HIROM.

  • SLOWROM - identical to a freestanding .SLOWROM.

  • FASTROM - identical to a freestanding .FASTROM.

  • CARTRIDGETYPE $00 - Places the given 8-bit value in $7FD6 ($FFD6 in HiROM). Some possible values I’ve come across but cannot guarantee the accuracy of:

    $00 ROM    
    $01 ROM RAM  
    $02 ROM SRAM  
    $03 ROM   DSP1
    $04 ROM RAM DSP1
    $05 ROM SRAM DSP1
    $13 ROM   Super FX
  • ROMSIZE $09 - Places the given 8-bit value in $7FD7 ($FFD7 in HiROM). Possible values include (but may not be limited to):

    $08 2 Megabits
    $09 4 Megabits
    $0A 8 Megabits
    $0B 16 Megabits
    $0C 32 Megabits
  • SRAMSIZE $01 - Places the given 8-bit value into $7FD8 ($FFD8 in HiROM). I believe these are the only possible values:

    $00 0 kilobits
    $01 16 kilobits
    $02 32 kilobits
    $03 64 kilobits
  • COUNTRY $00 - Places the given 8-bit value into $7FD9 ($FFD9 in HiROM). $00 is Japan and $01 is the United States, and there several more for other regions that I cannot recall off the top of my head.

  • LICENSEECODE $00 - Places the given 8-bit value into $7FDA ($FFDA in HiROM.) You must find the legal values yourself as there are plenty of them. ;)

  • VERSION $01 - Places the given 8-bit value into $7FDB ($FFDB in HiROM) This is supposedly interpreted as version 1.byte, so a $01 here would be version 1.01.

This is not a compulsory directive.

2.122. .ENDSNES

This ends the SNES header definition.

This is not a compulsory directive, but when .SNESHEADER is used this one is required to terminate it.

2.123. .SNESNATIVEVECTOR

Begins definition of the native mode interrupt vector table.

.SNESNATIVEVECTOR
COP    COPHandler
BRK    BRKHandler
ABORT  ABORTHandler
NMI    VBlank
UNUSED $0000
IRQ    IRQHandler
.ENDNATIVEVECTOR

These can be defined in any order, but they will be placed into memory starting at $7FE4 ($FFE4 in HiROM) in the order listed above. All the vectors default to $0000.

This is not a compulsory directive.

2.124. .ENDNATIVEVECTOR

Ends definition of the native mode interrupt vector table.

This is not a compulsory directive, but when .SNESNATIVEVECTOR is used this one is required to terminate it.

2.125. .SNESEMUVECTOR

Begins definition of the emulation mode interrupt vector table.

.SNESEMUVECTOR
COP    COPHandler
UNUSED $0000
ABORT  BRKHandler
NMI    VBlank
RESET  Main
IRQBRK IRQBRKHandler
.ENDEMUVECTOR

These can be defined in any order, but they will be placed into memory starting at $7FF4 ($FFF4 in HiROM) in the order listed above. All the vectors default to $0000.

This is not a compulsory directive.

2.126. .ENDEMUVECTOR

Ends definition of the emulation mode interrupt vector table.

This is not a compulsory directive, but when .SNESEMUVECTOR is used this one is required to terminate it.

3. Assembler Syntax

3.1. Case Sensitivity

WLA is case sensitive, so be careful.

3.2. Comments

Comments begin with ; or * and end along with the line. ; can be used anywhere, but * can be placed only at the beginning of a new line.

Version 4.1 of WLA introduced ANSI C -like commenting. This means you can start a multiline comment with /* and end it with */.

Version 6.0 of WLA introduced .ASM and .ENDASM directives. These function much like ANSI C comments, but unlike the ANSI C comments these can be nested.

3.3. Labels

Labels are ordinary strings (which can also end to a :). Labels starting with _ are considered to be local labels and do not show outside sections where they were defined, or outside object files, if they were not defined inside a section.

Here are few examples of different labels:

VBI_IRQ:
VBI_IRQ2
_VBI_LOOP:
main:

Labels starting with @ are considered to be child labels. They can only be referenced within the scope of their parent labels, unless the full name is specified. When there is more than one @, the label is considered to be a child of a child.

Here are some examples of child labels:

PARENT1:
@CHILD:
@@SUBCHILD

PARENT2:
@CHILD:

This is legal, since each of the @CHILD labels has a different parent. You can specify a parent to be explicit, like so:

jr PARENT1@CHILD@SUBCHILD

Note that when you place : in front of the label string when referring to it, you’ll get the bank number of the label, instead of the label’s address. Here’s an example:

LD A, :LOOP
.BANK 2 SLOT 0
LOOP:

Here LD A, :LOOP will be replaced with LD A, 2 as the label LOOP is inside the bank number two.

When you are referring to a label and you are adding something to its address (or subtracting, any arithmetics apply) the result will always be bytes.

.org 20
DATA:  .dw 100, 200, 300
       ld  a, DATA+1
              ^^^^^^ = r

So here the result r will be the address of DATA plus one, here 21. Some x86 assemblers would give here 22 as the result r as DATA points to an array or machine words, but WLA isn’t that smart (and some people including me think this is the better solution).

Note that each CPU WLA supports contains opcodes that either generate an absolute reference or a relative reference to the given label. For example,

.org 20
DATA:  ld  a, DATA   ; DATA becomes 20 (absolute)
       jr  DATA      ; DATA becomes -4 (relative)

Check out section 14 for the list of opcodes that generate relative references.

You can also use -, --, ---, +, ++, +++, … as un-named labels. Labels consisting of - are meant for reverse jumps and labels consisting of + are meant for forward jumps. You can reuse un-named labels as much as you wish inside your source code. Here’s an example of this:

    dec e
    beq ++      ; jump -> ?
    dec e
    beq +       ; jump -> %
    ld d, 14
--- ld a, 10    ; !
--  ld b, c     ; #
-   dec b       ; *
    jp nz, -    ; jump -> *
    dec c
    jp nz, --   ; jump -> #
    dec d
    jp nz, ---  ; jump -> !
    ld a, 20
-   dec a       ; $
    jp nz, -    ; jump -> $
+   halt        ; %
++  nop         ; ?

Note that __ (that’s two underline characters) serves also as a un-named label. You can refer to this label from both directions. Use _b when you are jumping backwards and _f when you are jumping forwards label __.

Example:

   dec e
   jp z, _f     ; jump -> *
   dec e
__ ldi a, (hl)  ; *
   dec e
   jp nz, _b    ; jump -> *

CAVEAT! CAVEAT! CAVEAT!

The following code doesn’t work as it would if WLA would determine the distance lexically (but in practice it’s WLALINK that does all the calculations and sees only the preprocessed output of WLA):

.macro dummy
-  dec a        ; #
   jp nz, -     ; jump -> #
.endm

   ...
-  nop          ; *
   dummy
   dec e
   jp nz, -     ; i'd like to jump to *, but i'll end up jumping
                ; to # as it's closest to me in the output WLA produces
                ; for WLALINK (so it's better to use \@ with labels inside
                ; a macro).

WLALINK will also generate _sizeof_[label] defines that measure the distance between two consecutive labels. These labels have the same scope as the labels they describe. Here is an example:

Label1:
    .db 1, 2, 3, 4
Label2:

In this case you’ll get a definition _sizeof_Label1 that will have value 4.

WLA will skip over any child labels when calculating _sizeof. So, in this example:

Label1:
.db 1, 2
@child:
    .db 3, 4
Label2:

The value of _sizeof_Label1 will still have a value of 4.

3.4. Number Types

1000 decimal
$100 hexadecimal
100h hexadecimal
%100 binary
'x' character

Remember that if you use the suffix h to give a hexadecimal value, and the value begins with an alphabet, you must place a zero in front of it so WLA knows it’s not a label (e.g., 0ah instead of ah).

3.5. Strings

Strings begin with and end to ". Note that no 0 is inserted to indicate the termination of the string like in e.g., ANSI C. You’ll have to do it yourself. You can place quotation marks inside strings the way C preprocessors accept them.

Here are some examples of strings:

"Hello world!"
"He said: \"Please, kiss me honey.\""

3.6. Mnemonics

You can give the operand size with the operand itself (and this is highly recommended) in WLA 6502/65C02/6510/HUC6280/65816:

and #20.b
and #20.w
bit loop.b
bit loop.w

3.7. Brackets?

Brackets are also supported in the GB-Z80/Z80/6502/65C02/HUC6280/6510 syntax. So you can write

LDI (HL), A

or

LDI [HL], A

Yes, you could write

LDI [HL), A

but I don’t recommend that. ;)

Note that brackets have special meaning when dealing with a 65816/SPC-700 system so you can’t use

AND [$65]

instead of

AND ($65)

as they mean different things.

4. Error Messages

There are quite a few of them in WLA, but most of them are not as informative as I would like them to be. This will be fixed in the future. Mean while, be careful. ;)

5. Supported ROM/RAM/Cartridge Types (WLA-GB)

5.1. ROM Size

GB-Z80 version of WLA supports the following ROM bank sizes. There’s no such limit in the Z80/6502/65C02/6510/65816/HUC6280/SPC-700 version of WLA. Supply one of the following values to .ROMBANKS.

$00 256Kbit 32KByte 2 banks
$01 512Kbit 64KByte 4 banks
$02 1Mbit 128KByte 8 banks
$03 2Mbit 256KByte 16 banks
$04 4Mbit 512KByte 32 banks
$05 8Mbit 1MByte 64 banks
$06 16Mbit 2MByte 128 banks
$52 9Mbit 1.1MByte 72 banks
$53 10Mbit 1.2MByte 80 banks
$54 12Mbit 1.5MByte 96 banks

5.2. RAM Size

Supply one of the following hex values to .RAMSIZE in the GB-Z80 version of WLA.

$00 None None None
$01 16kbit 2kByte 1 bank
$02 64kbit 8kByte 1 bank
$03 256kbit 32kByte 4 banks
$04 1Mbit 128kByte 16 banks

5.3. Cartridge Type

It’s up to the user to check that the cartridge type is valid and can be used combined with the supplied ROM and RAM sizes. Give one the the following values to .CARTRIDGETYPE in the GB-Z80 version of WLA.

$00 ROM          
$01 ROM MBC1        
$02 ROM MBC1 RAM      
$03 ROM MBC1 RAM BATTERY    
$05 ROM MBC2        
$06 ROM MBC2   BATTERY    
$08 ROM   RAM      
$09 ROM   RAM BATTERY    
$0B ROM MMM01        
$0C ROM MMM01 SRAM      
$0D ROM MMM01 SRAM BATTERY    
$0F ROM MBC3   BATTERY TIMER  
$10 ROM MBC3 RAM BATTERY TIMER  
$11 ROM MBC3        
$12 ROM MBC3 RAM      
$13 ROM MBC3 RAM BATTERY    
$19 ROM MBC5        
$1A ROM MBC5 RAM      
$1B ROM MBC5 RAM BATTERY    
$1C ROM MBC5       RUMBLE
$1D ROM MBC5 SRAM     RUMBLE
$1E ROM MBC5 SRAM BATTERY   RUMBLE
$1F Pocket Camera          
$BE Pocket Voice          
$FD Bandai TAMA5          
$FE Hudson HuC-3          
$FF Hudson HuC-1          

6. Bugs

If you find bugs, please let us know about them via GitHub: https://github.com/vhelin/wla-dx/issues

7. Files

7.1. examples

The main purpose of the files in the examples directory is to test that WLA and WLALINK can assemble and link the tiny project correctly. You can also take a look at the code and syntax in the files, but beware: if you run the rom files you probably don’t see anything on screen.

include directory under gb-z80 could be very useful as the six include files there have all the Game Boy hardware register address and memory definitions you could ever need and more.

7.2. examples/gb-z80/lib

This folder holds few very useful libraries for you to use in your Game Boy projects. Instead of reinventing the wheel, use the stuff found in here. Remember to compile the libraries right after you’ve installed WLA by executing make in the lib directory.

7.3. memorymaps

Here you can find default memory maps (see .MEMORYMAP) for various computers and video game consoles.

8. Temporary Files

Note that WLA will generate two temporary files while it works. Both files are placed into the current working directory.

System File A File B
Amiga wla_a.tmp wla_b.tmp
MSDOS wla_a.tmp wla_b.tmp
Win32 .wla%PID%a .wla%PID%b
Unix .wla%PID%a .wla%PID%b
AmigaOS4 .wla%PID%a .wla%PID%b

(%PID% is the process id in cases where it is included)

When WLA finishes its work these two files are deleted as they serve of no further use.

9. Compiling

9.1. Compiling Object Files

To compile an object file use the -o [OUT] option on the command line.

These object files can be linked together (or with library files) later with WLALINK.

Name object files so that they can be recognized as object files. Normal suffix is .o (WLA default). This can also be changed with .OUTNAME.

With object files you can reduce the amount of compiling when editing small parts of the program. Note also the possibility of using local labels (starting with _).

Note: When you compile objects, group 1 directives are saved for linking time, when they are all compared and if they differ, an error message is shown. It is advisable to use something like an include file to hold all the group 1 directives for that particular project and include it to every object file.

If you are interested in the WLA object file format, take a look at the file txt/wla_file_formats.txt which is included in the release archive.

Here are some examples of definitions:

  • -D IEXIST
  • -D DAY=10
  • -D BASE = $10
  • -D NAME=elvis

And here’s an WLA example creating definitions on the command line:

wla-gb -D DEBUG -D VERBOSE=5 -D NAME = "math v1.0" -o math.o math.s

DEBUG’s value will be 0, VERBOSE’s 5 and NAME is a string definition with value math v1.0. Note that -D always needs a space after it, but the rest of the statement can be optionally stuck inside one word.

9.2. Compiling Library Files

To compile a library file use the -l [OUT] option on the command line.

Name these files so that they can be recognized as library files. Normal suffix is .lib (WLA default).

With library files you can reduce the amount of compiling. Library files are meant to hold general functions that can be used in different projects. Note also the possibility of using local labels (starting with _). Library files consist only of FREE sections.

10. Linking

After you have produced one or more object files and perhaps some library files, you might want to link them together to produce a ROM image / program file. WLALINK is the program you use for that. Here’s how you use it:

wlalink [OPTIONS] <LINK FILE> <OUTPUT FILE>

Choose the option -b [OUT] for program file or -r [OUT] for ROM image linking.

Link file is a text file that contains information about the files you want to link together. Here’s the format:

  1. You must define the group for the files. Put the name of the group inside brackets. Valid group definitions are

    [objects]
    [libraries]
    [header]
    [footer]
    [definitions]
    
  2. Start to list the file names.

    [objects]
    main.o
    vbi.o
    level_01.o
    ...
    
  3. Give parameters to the library files:

    [libraries]
    bank 0 slot 1 speed.lib
    bank 4 slot 2 map_data.lib
    ...
    

    Here you can also use base to define the 65816 CPU bank number (like .BASE works in WLA):

    [libraries]
    bank 0 slot 1 base $80 speed.lib
    bank 4 slot 2 base $80 map_data.lib
    ...
    

    You must tell WLALINK the bank and the slot for the library files.

  4. If you want to use header and/or footer in your project, you can type the following:

    [header]
    header.dat
    [footer]
    footer.dat
    
  5. If you want to make value definitions, here’s your chance:

    [definitions]
    debug 1
    max_str_len 128
    start $150
    ...
    

If flag v is used, WLALINK displays information about ROM file after a succesful linking.

If flag s is used, WLALINK will produce a NO$GMB symbol file. It’s useful when you work under MSDOS (NO$GMB is a very good Game Boy emulator for MSDOS/Windows) as it contains information about the labels in your project.

If flag S is used, WLALINK will create a WLA symbol file, that is much like NO$GMB symbol file, but shows also symbols and breakpoints, not just labels and definitions.

If flag d is used, WLALINK discards all unreferenced FREE and SEMIFREE sections. This way you can link big libraries to your project and WLALINK will choose only the used sections, so you won’t be linking any dead code/data.

If flag i is given, WLALINK will write list files. Note that you must compile the object and library files with -i flag as well. Otherwise WLALINK has no extra information it needs to build list files. Here is an example of a list file: Let’s assume you’ve compiled a source file called main.s using the i flag. After you’ve linked the result also with the i flag WLALINK has created a list file called main.lst. This file contains the source text and the result data the source compiled into. List files are good for debugging.

If flag L is given after the above options, WLALINK will use the directory specified after the flag for including libraries. If WLALINK cannot find the library in the specified directory, it will then silently search the current working directory. This is useful when using WLA in an SDK environment where a global path is needed.

Make sure you don’t create duplicate labels in different places in the memory map as they break the linking loop. Duplicate labels are allowed when they overlap each other in the destination machine’s memory. Look at the following example:

...
.BANK 0
.ORG $150

    ...
    LD      A, 1
    CALL    LOAD_LEVEL
    ...

LOAD_LEVEL:
    LD      HL, $2000
    LD      (HL), A
    CALL    INIT_LEVEL
    RET

.BANK 1
.ORG 0

INIT_LEVEL:
    ...
    RET

.BANK 2
.ORG $0

INIT_LEVEL:
    ...
    RET
...

Here duplicate INIT_LEVEL labels are accepted as they both point to the same memory address (in the program’s point of view).

Examples:

[seravy@localhost tbp]# wlalink -r linkfile testa.sfc
[seravy@localhost tbp]# wlalink -d -i -b linkfile testb.sfc
[seravy@localhost tbp]# wlalink -v -S -L ../../lib linkfile testc.sfc

11. Arithmetics

WLA is able to solve really complex calculations like

-((HELLO / 2) | 3)
skeletor_end-skeletor
10/2.5

so you can write something like

LD HL, data_end-data
LD A, (pointer + 1)
CP (TEST + %100) & %10101010

WLALINK also has this ability so it can compute the pending calculations WLA wasn’t able to solve.

The following operators are valid:

, comma
| or
& and
^ power
<< shift left
>> shift right
+ plus
- minux
# modulo
~ xor
* multiply
/ divide
< get the low byte
> get the high byte

Note that you can do NOT using XOR:

  • VALUE_A ~ $FF is 8-bit NOT
  • VALUE_B ~ $FFFF is 16-bit NOT

WLA computes internally with real numbers so (5/2)*2 produces 5, not 4.

12. Binary to DB Conversion

WLAB converts binary files to WLA’s byte definition strings. Here’s how you use it:

wlab -[ap]{bdh} <BIN FILE>

Give it the binary file and WLAB will output the WLA DB formatted data of it into stdout. Here’s an example from real life:

wlab -da gayskeletor.bin > gayskeletor.s

WLAB has three command flags of which one must be given to WLAB:

-b Output data in binary format.
-d Output data in decimal format.
-h Output data in hexadecimal format.

WLAB has also two option flags:

-a Print the address (relative to the beginning of the data).
-p Don’t print file header.

Examples:

[seravy@localhost src]# wlab -bap iscandar.bin > iscandar.s
[seravy@localhost src]# wlab -h starsha.bin > starsha.s

13. Things you should know about coding for…

Please check out the source code examples (in examples directory) for quick target system specific information.

13.1. Z80

Check the Z80 specific directives. All SMS/GG coders should find .SMSTAG, .SDSCTAG and .COMPUTESMSCHECKSUM very useful…

Opcodes that make relative label references:

JR *
DJNZ

13.2. 6502

For example mnemonics ADC, AND, ASL, etc… cause problems to WLA, because they take different sized arguments. Take a look at this:

LSR 11       ; $46 $0B
LSR $A000    ; $4E $00 $A0

The first one could also be

LSR 11       ; $4E $0B $00

To really get what you want, use .8BIT, .16BIT and .24BIT directives. Or even better, supply WLA the size of the argument:

LSR 11.W     ; $4E $0B $00

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS

13.3. 65C02

Read the subsection 6502 as the information applies also to 65C02 coding…

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS
BRA
BBR*
BBS*

13.4. 6510

Read the subsection 6502 as the information applies also to 6510 coding…

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS

13.5. 65816

Read the subsection 6502 as the information applies also to 65816 coding…

WLA-65816 has also few SNES specific directives which are all very helpful. Remember that when you use .LOROM, .HIROM, .SLOWROM and .FASTROM WLA automatically writes the information into the output. .COMPUTESNESCHECKSUM, .SNESHEADER and few others could also be useful.

Use .BASE to set the upmost eight bits of 24-bit addresses.

If possible, use operand hints to specify the size of the operand. WLA is able to deduce the accumulator/index mode to some extent from REP/SEP-mnemonics and .ACCU and .INDEX-directives, but just to be sure, terminate the operand with .B, .W or .L.

AND #10     ; can be two different things, depending on the size of the accu.
AND #10.B   ; forces 8-bit immediate value.
AND #10.W   ; forces 16-bit immediate value.

Or if you must, these work as well:

AND.B #10   ; the same as "AND #10.B".
AND.W #10   ; the same as "AND #10.W".

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS
BRA
BRL
PER

13.6. HUC6280

Read the subsection 6502 as the information applies also to HUC6280 coding…

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS
BSR
BBR*
BBS*

13.7. SPC-700

Note that you’ll have to put an exclamation mark before a 16-bit value. For example,

CALL !Main
AND  A, !$1000

Opcodes that make relative label references:

BCC
BCS
BEQ
BMI
BNE
BPL
BVC
BVS
BRA
BBS
BBC
CBNE *
DBNZ *

13.8. Pocket Voice (GB-Z80)

Pocket Voice uses its own MBC. You can enable Pocket Voice mode by selecting Pocket Voice cartridge type ($BE in $0147) and defining correct .ROMBANKMAP and .MEMORYMAP. In PV mode bank 0 is 24KB and the rest are 8KB.

Note that WLA assumes that ROM offset is all the time 0. If you use something else as the offset, make sure to compute the jumps by hand as WLA cannot do that.

Check out examples/gb-z80/include/pocket_voice.i for more information.

13.9. GB-Z80

WLA outputs only $10 when it decodes STOP. Often it’s necessary to put an extra NOP ($00) after a STOP, and sometimes something else, but that’s left entirely to the user.

Opcodes that make relative label references:

JR *

14. WLA Flags

Here are short descriptions for the flags you can give to WLA:

You can supply WLA with some (or all or none) of the following option flags.

-i Add list file information. Adds extra information to the output so WLALINK can produce list files.
-M WLA generates makefile rules describing the dependencies of the main source file. Use only with flags o and l.
-q Quiet mode. .PRINT* -directives output nothing.
-t Test compile. Doesn’t output any files.
-v Verbose mode. Shows a lot of information about the compiling process.
-x Extra compile time definitions. WLA does extra work by creating few helpful definitions on the fly.

One (and only one) of the following command flags must be defined.

-l Output a library file.
-o Output an object file.

You may also use an extra option to specify the include directory. WLA will search this directory for included files before defaulting to the specified .INCDIR or current working directory.

-I Directory to include files.

Examples:

[seravy@localhost tbp]# wla -v -i -o testa.o testa.s
[seravy@localhost tbp]# wla -M -o testa.o testa.s
[seravy@localhost tbp]# wla -l testb.lib testb.s
[seravy@localhost tbp]# wla -I ../../include -l testb.lib testb.s
[seravy@localhost tbp]# wla -M -I myfiles -l testa.lib testa.s

Note that the first example produces file named testa.o.

15. Extra compile time definitions

When you supply WLA with the flag x it will maintain few useful definitions while compiling your source codes. Please use the enhanced error reporting engine (so don’t use flag f) in conjunction with flag x as some of the definitions require extra information about the flow of the data which isn’t available when using the old, crippled error reporting engine.

Here’s a list of what you get when you use flag x:

WLA_FILENAME A string definition holding the file name WLA is currently processing.
WLA_TIME A string definition holding the calendar time (obtained using C’s ctime()).
WLA_VERSION A string definition holding the version number of WLA.

So you can do for example something like

.DB WLA_TIME

to store the time when the build process started into the ROM file you are compiling.

Definition CADDR, which is present without supplying the flag x, contains the current 16-bit memory address. So

LD HL, CADDR

will load the address of the operand data into registers H and L.

CAVEAT: Remember when using defines that CADDR gets the address of the place where the definition is used, not the address of the definition, which contains the CADDR.

Note that you’ll also get all these definitions in lower case (e.g., wla_filename).

16. Good things to know about WLA

  • Is 255 bytes too little for a string (file names, labels, definition labels, etc)? Check out MAX_NAME_LENGTH in shared.h.
  • WLA preprocessor doesn’t expand macros and repetitions. Those are actually traversed in the assembling phase.
  • WLA’s source code is mainly a huge mess, but WLALINK is quite well structured and written. So beware!
  • Do not write .E into your sources as WLA uses it internally to mark the end of a file.

17. WLA DX’s architectural overview

The two most important executables inside WLA DX are WLA (the assembler) and WLALINK (the linker).

17.1. WLA

WLA has four separate phases, called a little bit incorrectly passes:

  1. pass_1.c: pass_1():
    • The biggest data processor in WLA.
    • Includes the include files: every time time happens the file is read in, white space is removed, lines formatted, etc.
    • Macros are processed along with directives
    • All textual data, code, etc. are transformed into WLA’s internal byte code that gets written into a tmp (TMP) file, and after this phase the assembler or the linker has no idea of target CPU’s opcodes - all is just pure WLA byte code.
    • The first and the only pass that handles the assembly source files supplied by the user.
    • The parser in this pass starts from the first byte of the first source file, then moves forward parsing everything that it encounters, but when a macro is called, the parser jumps to the beginning of the macro, and continues parsing from there.
  2. pass_2.c: pass_2():
    • If the user has issued directives like .SDSCTAG, here we generate the needed data and write that into TMP.
  3. pass_3.c: pass_3():
    • Here we read in TMP and do some sanity checks for the data, give labels addresses (if possible), generate internal structures for labels and sections.
  4. pass_4.c: pass_4():
    • Again we read in TMP.
    • Now we check that if there is a reference to a calculation, and that calculation has been succesfully calculated, then we can replace the reference with the result.
    • This phase writes out object and library files, i.e., transforms TMP to final output files (this write out could actually be pass_5)…

2. Manpage: WLA-CPU

2.1. SYNOPSIS

wla-gb [OPTIONS] SRC_FILE
wla-65c02 [OPTIONS] SRC_FILE
wla-6502 [OPTIONS] SRC_FILE
wla-6510 [OPTIONS] SRC_FILE
wla-65816 [OPTIONS] SRC_FILE
wla-huc6280 [OPTIONS] SRC_FILE
wla-spc700 [OPTIONS] SRC_FILE
wla-z80 [OPTIONS] SRC_FILE

2.2. OPTIONS

-D __DEFINE(=VAR)__ Define DEFINE with value VAR (VAR is optional)

-i Add list file information
-I DIR Add Include directory
-q Quiet mode (.PRINT*-directives output nothing)
-v Test compile (Don’t output any files)
-x Extra compile time definitions

Choose one:

-o OUT Output an object file
-l OUT Output an library file

2.3. DESCRIPTION

Assemble a BIN_FILE to an object file (-o) or to an library file (-l).

These object files can be linked together (or with library files) later with wlalink(1).

Name object files so that they can be recognized as object files. Normal suffix is .o (WLA default). This can also be changed with .OUTNAME.

Name these files so that they can be recognized as library files. Normal suffix is .lib (WLA default).

With object files you can reduce the amount of compiling when editing small parts of the program. Note also the possibility of using local labels (starting with _).

With library files you can reduce the amount of compiling. Library files are meant to hold general functions that can be used in different projects. Note also the possibility of using local labels (starting with _). Library files consist only of FREE sections.

Note: When you compile objects, group 1 directives are saved for linking time, when they are all compared and if they differ, an error message is shown. It is advisable to use something like an include file to hold all the group 1 directives for that particular project and include it to every object file.

If you are interested in the WLA object file format, take a look at the file txt/wla_file_formats.txt which is included in the release archive.

2.4. EXAMPLES

wla-gb -D DEBUG -D VERBOSE=5 -D NAME = "math v1.0" -o math.o math.s
  • -D IEXIST
  • -D DAY=10
  • -D BASE = $10
  • -D NAME=elvis

3. Manpage: WLAB

3.1. SYNOPSIS

wlab -[ap]{bdh} BIN_FILE

3.2. OPTIONS

-a Print the address (relative to the beginning of the data).
-p Don’t print file header.

Choose one:

-b Output data in binary format.
-d Output data in decimal format.
-h Output data in hexadecimal format.

3.3. DESCRIPTION

wla(1) converts binary files to WLA’s byte definition strings and print it to the standard output.

3.4. EXAMPLES

wlab -da gayskeletor.bin > gayskeletor.s
wlab -bap iscandar.bin > iscandar.s
wlab -h starsha.bin > starsha.s