mirror of
https://github.com/hsoft/collapseos.git
synced 2024-11-08 17:58:05 +11:00
132 lines
3.2 KiB
NASM
132 lines
3.2 KiB
NASM
; *** Const ***
|
|
; Base of the Return Stack
|
|
.equ RS_ADDR 0xf000
|
|
; Number of bytes we keep as a padding between HERE and the scratchpad
|
|
.equ PADDING 0x20
|
|
; Max length of dict entry names
|
|
.equ NAMELEN 7
|
|
; Offset of the code link relative to the beginning of the word
|
|
.equ CODELINK_OFFSET NAMELEN+3
|
|
|
|
; *** Variables ***
|
|
.equ INITIAL_SP FORTH_RAMSTART
|
|
.equ CURRENT @+2
|
|
.equ HERE @+2
|
|
; Pointer to where we currently are in the interpretation of the current line.
|
|
.equ INPUTPOS @+2
|
|
; Pointer to where compiling words should output. During interpret, it's a
|
|
; moving target in (COMPBUF). During DEFINE, it's (HERE).
|
|
.equ CMPDST @+2
|
|
; Buffer where we compile the current input line. Same size as STDIO_BUFSIZE.
|
|
.equ COMPBUF @+2
|
|
.equ FORTH_RAMEND @+0x40
|
|
|
|
; EXECUTION MODEL
|
|
; After having read a line through stdioReadLine, we want to interpret it. As
|
|
; a general rule, we go like this:
|
|
;
|
|
; 1. read single word from line
|
|
; 2. compile word to atom
|
|
; 3. execute atom
|
|
; 4. goto 1
|
|
;
|
|
; During step 3, it's possible that atom read from input, so INPUTPOS might
|
|
; have moved between 3 and 4.
|
|
;
|
|
; Because the Parameter Stack uses PS, we can't just go around calling routines:
|
|
; This messes with the PS. This is why we almost always jump (unless our call
|
|
; doesn't involve Forth words in any way).
|
|
;
|
|
; This presents a challenge for our interpret loop because step 4, "goto 1"
|
|
; isn't obvious. To be able to do that, we must push a "return routine" to the
|
|
; Return Stack before step 3.
|
|
|
|
; *** Code ***
|
|
forthMain:
|
|
; STACK OVERFLOW PROTECTION:
|
|
; To avoid having to check for stack underflow after each pop operation
|
|
; (which can end up being prohibitive in terms of costs), we give
|
|
; ourselves a nice 6 bytes buffer. 6 bytes because we seldom have words
|
|
; requiring more than 3 items from the stack. Then, at each "exit" call
|
|
; we check for stack underflow.
|
|
push af \ push af \ push af
|
|
ld (INITIAL_SP), sp
|
|
ld hl, LATEST
|
|
ld (CURRENT), hl
|
|
ld hl, FORTH_RAMEND
|
|
ld (HERE), hl
|
|
forthRdLine:
|
|
ld hl, msgOk
|
|
call printstr
|
|
call printcrlf
|
|
call stdioReadLine
|
|
ld ix, RS_ADDR-2 ; -2 because we inc-before-push
|
|
ld (INPUTPOS), hl
|
|
ld hl, COMPBUF
|
|
ld (CMPDST), hl
|
|
forthInterpret:
|
|
call readword
|
|
jr nz, .execute
|
|
call find
|
|
jr nz, .maybeNum
|
|
ex de, hl
|
|
call HLisIMMED
|
|
jr z, .immed
|
|
ex de, hl
|
|
call .writeDE
|
|
jr forthInterpret
|
|
.maybeNum:
|
|
push hl ; --> lvl 1. save string addr
|
|
call parseLiteral
|
|
pop hl ; <-- lvl 1
|
|
jr nz, .undef
|
|
; a valid number in DE!
|
|
ex de, hl
|
|
ld de, NUMBER
|
|
call .writeDE
|
|
ex de, hl ; number in DE
|
|
call .writeDE
|
|
jr forthInterpret
|
|
.undef:
|
|
; When encountering an undefined word during compilation, we spit a
|
|
; reference to litWord, followed by the null-terminated word.
|
|
; This way, if a preceding word expect a string literal, it will read it
|
|
; by calling readLIT, and if it doesn't, the routine will be
|
|
; called, triggering an abort.
|
|
ld de, LIT
|
|
call .writeDE
|
|
ld de, (CMPDST)
|
|
call strcpyM
|
|
ld (CMPDST), de
|
|
jr forthInterpret
|
|
.immed:
|
|
push hl ; --> lvl 1
|
|
ld hl, .retRef
|
|
call pushRS
|
|
pop iy ; <-- lvl 1
|
|
jp executeCodeLink
|
|
.execute:
|
|
ld de, QUIT
|
|
call .writeDE
|
|
ld iy, COMPBUF
|
|
jp compiledWord
|
|
.writeDE:
|
|
push hl
|
|
ld hl, (CMPDST)
|
|
ld (hl), e
|
|
inc hl
|
|
ld (hl), d
|
|
inc hl
|
|
ld (CMPDST), hl
|
|
pop hl
|
|
ret
|
|
|
|
.retRef:
|
|
.dw $+2
|
|
.dw $+2
|
|
call popRS
|
|
jr forthInterpret
|
|
|
|
msgOk:
|
|
.db " ok", 0
|