prog8/reference.md
2018-01-29 21:36:01 +01:00

29 KiB

IL65 / 'Sick' - Experimental Programming Language for 8-bit 6502/6510 microprocessors

Written by Irmen de Jong (irmen@razorvine.net)

Software license: GNU GPL 3.0, see file LICENSE

This is an experimental programming language for the 8-bit 6502/6510 microprocessor from the late 1970's and 1980's as used in many home computers from that era. IL65 is a medium to low level programming language, which aims to provide many conveniences over raw assembly code (even when using a macro assembler):

  • reduction of source code length
  • easier program understanding (because it's higher level, and more terse)
  • option to automatically run the compiled program in the Vice emulator
  • modularity, symbol scoping, subroutines
  • subroutines have enforced input- and output parameter definitions
  • various data types other than just bytes (16-bit words, floats, strings, 16-bit register pairs)
  • automatic variable allocations, automatic string variables and string sharing
  • constant folding in expressions (compile-time evaluation)
  • automatic type conversions
  • floating point operations
  • optional automatic preserving and restoring CPU registers state, when calling routines that otherwise would clobber these
  • abstracting away low level aspects such as zero page handling, program startup, explicit memory addresses
  • breakpoints, that let the Vice emulator drop into the monitor if execution hits them
  • source code labels automatically loaded in Vice emulator so it can show them in disassembly
  • conditional gotos
  • various code optimizations (code structure, logical and numerical expressions, ...)
  • @todo: loops
  • @todo: memory block operations

It still allows for low level programming however and inline assembly blocks to write performance critical pieces of code, but otherwise compiles fairly straightforwardly into 6502 assembly code. This resulting code is assembled into a binary program by using an external macro assembler, 64tass. It can be compiled pretty easily for various platforms (Linux, Mac OS, Windows) or just ask me to provide a small precompiled executable if you need that. You need Python 3.5 or newer to run IL65 itself.

IL65 is mainly targeted at the Commodore-64 machine, but should be mostly system independent.

Memory Model

Most of the 64 kilobyte address space can be accessed by your program.

type memory area note
Zero page $00 - $ff contains many sensitive system variables
Hardware stack $100 - $1ff is used by the CPU and should normally not be accessed directly
Free RAM or ROM $0200 - $ffff free to use memory area, often a mix of RAM and ROM

A few memory addresses are reserved and cannot be used freely by your own code, because they have a special hardware function, or are reserved for internal use by the compiler:

reserved address
data direction $00
bank select $01
scratch var #1 $02
scratch var #2 $03
NMI vector $fffa
RESET vector $fffc
IRQ vector $fffe

A particular 6502/6510 machine such as the Commodore-64 will have many other special addresses due to:

  • ROMs installed in the machine (BASIC, kernel and character generator roms)
  • memory-mapped I/O registers (for the video and sound chip for example)
  • RAM areas used for screen graphics and sprite data.

Usable Hardware Registers

The following 6502 hardware registers are directly accessible in your code (and are reserved symbols):

  • A, X, Y
  • AX, AY, XY (surrogate registers: 16-bit combined register pairs in LSB byte order lo/hi)
  • SC (status register's Carry flag)
  • SI (status register's Interrupt Disable flag)

Zero Page ("ZP")

The zero page locations $02 - $ff can be regarded as 254 other registers because they take less clock cycles to access and need fewer instruction bytes than access to other memory locations. Theoretically you can use all of them in your program but there are a few limitations:

  • several locations ($02, $03, $fb - $fc, $fd - $fe) are reserved for internal use as scratch registers by IL65
  • most other addresses often are in use by the machine's operating system or kernal, and overwriting them can crash the machine. Your program must take over the entire system to be able to safely use all zero page locations.
  • it's often more convenient to let IL65 allocate the particular locations for you and just use symbolic names in your code.

For the Commodore-64 here is a list of free-to-use zero page locations even when its BASIC and KERNAL are active:

$02; $03; $04; $05; $06; $2a; $52; $f7 - $f8; $f9 - $fa; $fb - $fc; $fd - $fe

The four reserved locations mentioned above are subtracted from this set, leaving you with five 1-byte and two 2-byte usable zero page registers. IL65 knows about all this: it will use the above zero page locations to place its ZP variables in, until they're all used up. You can instruct it to treat your program as taking over the entire machine, in which case (almost) all of the zero page locations are suddenly available for variables. IL65 can generate a special routine that saves and restores the zero page to let your program run and return safely back to the system afterwards - you don't have to take care of that yourself.

IRQ and the Zero page:

The normal IRQ routine in the C-64's kernal will read and write several locations in the zero page:

$a0 - $a2; $91; $c0; $c5; $cb; $f5 - $f6

These locations will not be used by the compiler for zero page variables, so your variables will not interfere with the IRQ routine and vice versa. This is true for the normal zp mode but also for the mode where the whole zp has been taken over. So the normal IRQ vector is still running when your program is entered, even when you use %zp clobber.

@todo: some global way (in ZP block) to promote certian other blocks/variables from that block or even subroutine to the zeropage. Don't do this in the block itself because it's a global optimization and if blocks require it themselves you can't combine various modules anymore once ZP runs out.

Data Types

IL65 supports the following data types:

type size type identifier example
(unsigned) byte 8 bits .byte $8f
(unsigned) integer 16 bits .word $8fee
boolean 1 byte true, false (aliases for the numeric values 1 and 0)
character 1 byte '@' (converted to a numeric byte)
floating-point 40 bits .float 1.2345 (stored in 5-byte cbm MFLPT format)
string variable .text, .stext "hello." (implicitly terminated by a 0-byte)
pascal-string variable .ptext, .pstext "hello." (implicit first byte = length, no 0-byte
address-of 16 bits #variable
indirect variable [ address ]

Strings can be writen in your code as CBM PETSCII or as C-64 screencode variants, these will be translated by the compiler. PETSCII is the default, if you need screencodes you have to use the s variants of the type identifier. If you write a string with just one character in it, it is always considered to be a BYTE instead with that character's PETSCII value. So if you really need a string of length 1 you must declare it explicitly as a variable of type .text, you cannot put "x" as a subroutine argument where the subroutine expects (the address of) a string. IL65's type system is unfortunately not strict enough to avoid this mistake, but it does print a warning if the situation is detected.

For many floating point operations, the compiler has to use routines in the C-64 BASIC and KERNAL ROMs. So they will only work if the BASIC ROM (and KERNAL ROM) are banked in, and your code imports the c654lib.ill.

The largest 5-byte MFLPT float that can be stored is: 1.7014118345e+38 (negative: -1.7014118345e+38)

The # prefix is used to take the address of something. This is sometimes useful, for instance when you want to manipulate the address of a memory mapped variable rather than the value it represents. You can take the address of a string as well, but the compiler already treats those as a value that you manipulate via its address, so the # is ignored here. For most other types this prefix is not supported.

Indirect addressing: The [address] syntax means: the contents of the memory at address, or "indirect addressing". By default, if not otherwise known, a single byte is assumed. You can add the .byte or .word or .float type identifier, inside the bracket, to make it clear what data type the address points to. For instance: [address .word] (notice the space, to distinguish this from a dotted symbol name). For an indirect goto call, the 6502 CPU has a special instruction (jmp indirect) and an indirect subroutine call (jsr indirect) is emitted using a couple of instructions.

Program Structure

In IL65 every line in the source file can only contain one statement or definitons. Compilation is done on one main source code file, but other files can be imported. A source file can start with global directives (starting with %) and continues with imports and block definitions.

Comments

Everything after a semicolon ';' is a comment and is ignored. If the comment is the only thing on the line, it is copied into the resulting assembly source code. This makes it easier to understand and relate the generated code.

Output Mode

The default format of the generated program is a "raw" binary where code starts at $c000. You can generate other types of programs as well, by telling IL65 via an output mode directive at the beginning of your program:

mode directive meaning
%output raw no load address bytes
%output prg include the first two load address bytes, (default is $0801), no BASIC program
%output prg,basic as 'prg', but include a BASIC start program with SYS call, default code start is immediately after the BASIC program at $081d, or after that.
%address $0801 override program start address (default is set to $c000 for raw mode and $0801 for C-64 prg mode). Cannot be used if output mode is prg,basic because BASIC programs always have to start at $0801.

ZeroPage Options

You can tell the compiler how to treat the zero page. Normally it is considered a 'no-go' area except for the frew free locations mentioned under "Memory Model". However you can specify some options globally in your program by using the zp directive to change this behavior:

  • %zp clobber
    Use the whole zeropage for variables. It is not possible to exit your program correctly back to BASIC, other than resetting the machine. It does make your program smaller and faster because many more variables can be stored in the ZP, which is more efficient.
  • %zp clobber, restore Use the whole zeropage, but make a backup copy of the original values at program start. When your program exits, the original ZP is restored (except for the software jiffy clock in $a0 - $a2) and you drop back to the BASIC prompt. Not that the default IRQ routine is still enabled when your program is entered! See the paragraph on the zero page for more info about this.

If you use %zp clobber, you can no longer use most BASIC or KERNAL routines, because these depend on most of the locations in the ZP. This includes the floating-point logic and several utility routines that do I/O, such as print_string.

Importing, Including and Binary-Including other files or raw assembly code

  • %import "filename[.ill]" Must be used after any global option directives, and outside of a block, but otherwise can be placed anywhere in the program. Reads and compiles the named IL65 file and stores the resulting definitions as if they were a part of your current program.
  • %asminclude "filename.txt", scopelabel This directive can only be used inside a block. The assembler will include the file as raw assembly source text at this point, il65 will not process this at all. The scopelabel will be used as a prefix to access the labels from the included source code, otherwise you would risk symbol redefinitions or duplications.
  • %asmbinary "filename.bin" [, <offset>[, <length>]] This directive can only be used inside a block. The assembler will include the file as binary bytes at this point, il65 will not process this at all. The optional offset and length can be used to select a particular piece of the file.
  • %asm { [raw assembly code lines] } This directive includes raw unparsed assembly code at that exact spot in the program. The %asm { and } start and end markers each have to be on their own unique line.

Program Start and Entry Point

Every program has to have one entry point where code execution begins. The compiler looks for the start label in the main block for this. For proper program termination, this block has to end with a return statement (or a goto call). Blocks and other details are described below.

The initial values of your variables will be restored automatically when the program is (re)started, except for string variables. It is assumed they are unchanged by your program. If you do modify them in-place, you should take care yourself that they work as expected when the program is restarted.

Blocks

    ~ blockname [address] {
            [directives...]
            [statements...]
    }

Blocks form the separate pieces of code and data of your program. They are combined and arranged to a single output program. No code or data can occur outside a block. A block is also a scope in your program so the symbols in the block don't clash with symbols of the same name defined elsewhere. You can refer to the symbols in a particular block by using a dotted name: blockname.symbolname.

Block names must be unique in your entire program, except "ZP": the contents of every block with that name are merged into one. This block name refers to the zero page. Its start address is always set to $04, because $00/$01 are used by the hardware and $02/$03 are reserved as general purpose scratch registers. Block address must be >= $0200 (because $00-$fff is the ZP and $100-$200 is the cpu stack)

You can omit the blockname but then you can only refer to the contents of the block via its absolute address, which is required in this case. If you omit both, the block is ignored altogether (and a warning is displayed). This is a way to quickly "comment out" a piece of code that is unfinshed or may contain errors that you want to work on later, because the contents of the ignored block are not fully parsed.

Assignments

Assignment statements assign a single value to one or more variables or memory locations. If you know that you have to assign the same value to more than one thing at once, it is more efficient to write it as a multi-assign instead of several separate assignments. The compiler tries to detect this situation however and optimize it itself if it finds the case.

    target = value-expression
    target1 = target2 = target3 [,...] = value-expression

Augmented Assignments

A special assignment is the augmented assignment where the value is modified in-place. Several assignment operators are available: +=, -=, &=, |=, ^=, <<=, >>=

Expressions

In most places where a number or other value is expected, you can use just the number, or a full constant expression. The expression is parsed and evaluated by Python itself at compile time, and the (constant) resulting value is used in its place. Ofcourse the special il65 syntax for hexadecimal numbers ($xxxx), binary numbers (%bbbbbb), and the address-of (#xxxx) is supported. Other than that it must be valid Python syntax. Expressions can contain function calls to the math library (sin, cos, etc) and you can also use all builtin functions (max, avg, min, sum etc). They can also reference idendifiers defined elsewhere in your code, if this makes sense.

Subroutine Definition

Subroutines are parts of the code that can be repeatedly invoked using a subroutine call from elsewhere. Their definition, using the sub statement, includes the specification of the required input- and output parameters. For now, only register based parameters are supported (A, X, Y and paired registers, the carry status bit SC and the interrupt disable bit SI as specials). For subroutine return values, the special SZ register is also available, it means the zero status bit.

The syntax is:

    sub   <identifier>  ([proc_parameters]) -> ([proc_results])  {
            ... statements ...
    }

proc_parameters = comma separated list of ":" pairs specifying the input parameters. You can omit the parameter names as long as the arguments "line up". (actually, the Python parameter passing rules apply, so you can also mix positional and keyword arguments, as long as the keyword arguments come last)

proc_results = comma separated list of names specifying in which register(s) the output is returned. If the register name ends with a '?', that means the register doesn't contain a real return value but is clobbered in the process so the original value it had before calling the sub is no longer valid. This is not immediately useful for your own code, but the compiler needs this information to emit the correct assembly code to preserve the cpu registers if needed when the call is made. For convenience: a single '?' als the result spec is shorthand for A?, X?, Y? ("I don't know what the changed registers are, assume the worst")

Subroutines that are pre-defined on a specific memory location (usually routines from ROM), can also be defined using the 'sub' statement. But in this case you don't supply a block with statements, but instead assign a memory address to it:

    sub  <identifier>  ([proc_parameters]) -> ([proc_results])  = <address>

example:

    sub  CLOSE  (logical: A) -> (A?, X?, Y?)  = $FFC3"

Subroutine Calling

You call a subroutine like this:

    subroutinename_or_address ( [arguments...] )

or:

    subroutinename_or_address ![register(s)] ( [arguments...] )

If the subroutine returns one or more values as results, you must use an assignment statement to store those values somewhere:

    outputvar1, outputvar2  =  subroutine ( arg1, arg2, arg3 )

The output variables must occur in the correct sequence of return registers as specified in the subroutine's definiton. It is possible to not specify any of them but the compiler will issue a warning then if the result values of a subroutine call are discarded. If you don't have a variable to store the output register in, it's then required to list the register itself instead as output variable.

Arguments should match the subroutine definition. You are allowed to omit the parameter names. If no definition is available (because you're directly calling memory or a label or something else), you can freely add arguments (but in this case they all have to be named).

To jump to a subroutine (without returning), prefix the subroutine call with the word 'goto'. Unlike gotos in other languages, here it take arguments as well, because it essentially is the same as calling a subroutine and only doing something different when it's finished.

Register preserving calls: use the ! followed by a combination of A, X and Y (or followed by nothing, which is the same as AXY) to tell the compiler you want to preserve the origial value of the given registers after the subroutine call. Otherwise, the subroutine may just as well clobber all three registers. Preserving the original values does result in some stack manipulation code to be inserted for every call like this, which can be quite slow.

Calling Convention

Subroutine arguments and results are passed via registers (and sometimes implicitly via certain memory locations).
@todo support call non-register args (variable parameter passing)

In IL65 the "caller saves" principle applies to registers used in a subroutine. This means the code that calls a subroutine or performs some function that clobber certain registers (A, X or Y), is responsible for storing and restoring the original values if that is required.

You should assume that the 3 hardware registers A, X and Y are volatile and their contents cannot be depended upon, unless you make sure otherwise.

Normally, the registers are NOT preserved when calling a subroutine or when a certian operations are performed. Most calls will be simply a few instructions to load the values in the registers and then a JSR or JMP.

By using the %saveregisters directive (globally or in a block) you can tell the compiler to preserve all registers. This does generate a lot of extra code that puts original values on the stack and gets them off the stack again once the subroutine is done. In this case however you don't have to worry about A, X and Y losing their original values and you can essentially treat them as three local variables instead of scratch data.

You can also use a ! on a single subroutine call to preserve register values, instead of setting this behavior for the entire block or even program. See the paragraph on Subroutine Calling for more info.

Conditional Execution Flow

Conditional execution flow means that the flow of execution changes based on certiain conditions, rather than having fixed gotos or subroutine calls. IL65 has a conditional goto statement for this, that is translated into a comparison (if needed) and then a conditional branch instruction:

    if[_XX] [<expression>] goto <label>

The if-status XX is one of: [cc, cs, vc, vs, eq, ne, true, not, zero, pos, neg, lt, gt, le, ge] It defaults to 'true' (=='ne', not-zero) if omitted. ('pos' will translate into 'pl', 'neg' into 'mi') @todo signed: lts==neg?, gts==eq+pos?, les==neg+eq?, ges==pos?

The is optional. If it is provided, it will be evaluated first. Only the [true] and [not] and [zero] if-statuses can be used when such a comparison expression is used. An example is:

    if_not  A > 55  goto  more_iterations

Conditional jumps are compiled into 6502's branching instructions (such as bne and bcc) so the rather strict limit on how far it can jump applies. The compiler itself can't figure this out unfortunately, so it is entirely possible to create code that cannot be assembled successfully. You'll have to restructure your gotos in the code (place target labels closer to the branch) if you run into this type of assembler error.

Debugging (with Vice)

The %breakpoint directive instructs the compiler to put a breakpoint at that position in the code. It's a logical breakpoint instead of a physical BRK instruction because that will usually halt the machine altogether instead of breaking execution. Instead, a NOP instruction is generated and in a special output file the list of breakpoints is written.

This file is called "yourprogramname.vice-mon-list" and is meant to be used by the Vice C-64 emulator. It contains a series of commands for Vice's monitor, this includes source labels and the breakpoint settings. If you use the vice autostart feature of the compiler, it will be automatically processed by Vice. If you launch Vice manually, you can use a command line option to load the file: x64 -moncommands yourprogramname.vice-mon-list

Vice will use the label names in memory disassembly, and will activate the breakpoints as well so if your program runs and it hits a breakpoint, Vice will halt execution and drop into the monitor.

@Todo

IF_XX:

if[_XX] [] { ... } [ else { ... ; evaluated when the condition is not met } ]

==> DESUGARING ==>

(no else:)

            if[_!XX] [<expression>] goto il65_if_999_end          ; !XX being the conditional inverse of XX
            .... (true part)

il65_if_999_end ; code continues after this

(with else): if[_XX] [] goto il65_if_999 ... (else part) goto il65_if_999_end il65_if_999 ... (true part) il65_if_999_end ; code continues after this

IF X Y:

==> DESUGARING ==> compare X, Y if_XX goto .... XX based on .

While

while[_XX] { ... continue break }

==> DESUGARING ==>

goto il65_while_999_check    ; jump to the check

il65_while_999 ... (code) goto il65_while_999 ;continue goto il65_while_999_end ;break il65_while_999_check if[_XX] goto il65_while_999 ; loop condition il65_while_999_end ; code continues after this

Repeat

repeat { ... continue break } until[_XX]

==> DESUGARING ==>

il65_repeat_999 ... (code) goto il65_repeat_999 ;continue goto il65_repeat_999_end ;break if[_!XX] goto il65_repeat_999 ; loop condition via conditional inverse of XX il65_repeat_999_end ; code continues after this

For

for = <from_expression> to <to_expression> [step <step_expression>] { ... break continue }

@todo how to do signed integer loopvars?

==> DESUGARING ==>

    loopvar = <from_expression>
    compare loopvar, <to_expression>
    if_ge goto il65_for_999_end       ; loop condition
    step = <step_expression>        ; (store only if step < -1 or step > 1)

il65_for_999 goto il65_for_999_end ;break goto il65_for_999_loop ;continue .... (code) il65_for_999_loop loopvar += step ; (if step > 1 or step < -1) loopvar++ ; (if step == 1) loopvar-- ; (if step == -1) goto il65_for_999 ; continue the loop il65_for_999_end ; code continues after this

Macros

@todo macros are meta-code (written in Python syntax) that actually runs in a preprecessing step during the compilation, and produces output value that is then replaced on that point in the input source. Allows us to create pre calculated sine tables and such. Something like:

    var .array sinetable ``[sin(x) * 10 for x in range(100)]``

Memory Block Operations

@todo matrix,list,string memory block operations:

  • matrix type operations (whole matrix, per row, per column, individual row/column) operations: set, get, copy (from another matrix with the same dimensions, or list with same length), shift-N (up, down, left, right, and diagonals, meant for scrolling) rotate-N (up, down, left, right, and diagonals, meant for scrolling) clear (set whole matrix to the given value, default 0)

  • list operations (whole list, individual element) operations: set, get, copy (from another list with the same length), shift-N(left,right), rotate-N(left,right) clear (set whole list to the given value, default 0)

  • list and matrix operations ofcourse work identical on vars and on memory mapped vars of these types.

  • strings: identical operations as on lists.

  • matrix with row-interleave can only be a memory mapped variable and can be used to directly access a rectangular area within another piece of memory - such as a rectangle on the (character) screen

these should call (or emit inline) optimized pieces of assembly code, so they run as fast as possible

Bitmap Definition (for Sprites and Characters)

to define CHARACTERS (8x8 monochrome or 4x8 multicolor = 8 bytes) --> PLACE in memory on correct address (???k aligned) and SPRITES (24x21 monochrome or 12x21 multicolor = 63 bytes) --> PLACE in memory on correct address (base+sprite pointer, 64-byte aligned)

More Datatypes

@todo pointers/addresses? (as opposed to normal WORDs) @todo signed integers (byte and word)?

Some support for simple arithmetic

    A *= Y
    A = X * Y
    A /= Y
    A = Y / Y

Troubleshooting

Getting assembler error about undefined symbols such as not defined 'c64flt'? This happens when your program uses floating point values, and you forgot to import the c64lib. If you use floating points, the program will need routines from that library. Fix it by adding an %import c64lib.