path: root/sanctuary.s

; sanctuary

; macros {{{
; TODO: error handling (once i add that)
%macro pspush 1
	lea r15, [r15-8]
	mov qword [r15], %1
%endmacro

%macro pspop 1
	mov %1, qword [r15]
	lea r15, [r15+8]
%endmacro

%define s_latest 0

%macro defdict 3 ; name label flags
	%strlen slen %1
	global lfa_%2
	lfa_%2: dq s_latest
	%define s_latest lfa_%2
	ffa_%2: db %3
	nfa_%2: db slen
		db %1
%endmacro

%macro defcode 3
	defdict %1, %2, %3
	%2:
%endmacro

; this is just taken from jewelforth, and does not correspond
; to how user variables are planned to work in sanctuary
; so todo make better later? i don't know if it really matters
; because it will only apply to builtin variables.
%macro defvar 4
	%2: dq %4
	defdict %1, %2, %3
	pspush qword %2
%endmacro
; }}}

%assign smudge_mask	0x1
%assign immediate_mask	0x2
%assign comp_only_mask	0x4

%assign false 0x0
%assign true (~0x0)

%assign INTERPRET 0x0
%assign COMPILING (~0x0)

%assign __NR_write	1
%assign __NR_mprotect	10
%assign __NR_brk	12
%assign __NR_exit	60

%assign init_brk	0x9c400

section .bss
resq 4091
wstk: resq 1

section .text
global _start
_start:
	lea r15, [wstk]

	call brk@
	pspop r11
	mov qword [dp], r11
	mov qword [dp0], r11
	mov r11, init_brk
	pspush r11
	call grow
	call interpret
	call bye

defcode "brk@", brk@, 0
	xor rdi, rdi
	mov rax, __NR_brk
	syscall
	pspush rax
	ret

defcode "grow", grow, 0
	call brk@
	pspop rdi
	pspop r13
	add rdi, r13
	mov rax, __NR_brk
	syscall
	mov qword [dp$], rax
	; this marks the *whole user memory* as executable.
	; technically redundant but i think it's fine
	mov r14, qword [dp0]
	pspush r14
	sub rax, r14
	pspush rax
	call executable
	ret

defcode "executable", executable, 0
	mov rdx, 0x7 ; PROT_{READ,WRITE,EXEC}
	pspop rsi
	pspop rdi ; addr
	mov rax, __NR_mprotect
	syscall
	ret

defcode "here", here, 0
	mov r11, qword [dp]
	pspush r11
	ret

defcode "bye", bye, 0
	mov rdi, 0
	mov rax, __NR_exit
	syscall
	ret

; input parsing {{{
; r11: string character count
; rsi: input buffer address
; al: char being parsed
; r10: end of input buffer
defcode "parse-name", parse_name, 0
	mov rsi, qword [to_in]
	mov r10, qword [tib]
	add rsi, r10
	add r10, qword [n_tib]
	xor rax, rax

.wsloop:
	cmp rsi, r10
	jge .empty
	lodsb
	cmp al, 0x20
	je .wsloop
	cmp al, 0x09
	je .wsloop
	cmp al, 0x0a
	je .wsloop

	cmp rsi, r10
	jge .empty
	mov r11, 1
	dec rsi ; bring down by one to point to the start
	push rsi ; will become `a`
	inc rsi
.wordloop:
	cmp al, 0x20
	je .wordloop_e
	cmp al, 0x09
	je .wordloop_e
	cmp al, 0x0a
	je .wordloop_e

	cmp rsi, r10
	jge .wordloop_e
	inc r11
	lodsb
	jmp .wordloop

.wordloop_e:
	dec r11
	sub rsi, qword [tib]
	mov qword [to_in], rsi
	pop rsi
	pspush rsi
	pspush r11
	ret

.empty:
	pspush 0
	pspush 0
	ret

; r11: string character count
; rsi: input buffer address
; al: char being parsed
; r10: end of input buffer
defcode "parse", parse, 0
	pspop rbx
	mov rsi, qword [to_in]
	mov r10, qword [tib]
	add rsi, r10
	add r10, qword [n_tib]
	xor rax, rax

.wsloop:
	cmp rsi, r10
	jge .empty
	lodsb
	cmp al, bl
	je .wsloop
	cmp al, 0x0a
	je .wsloop

	cmp rsi, r10
	jge .empty
	mov r11, 1
	dec rsi ; bring down by one to point to the start
	push rsi ; will become `a`
	inc rsi
.wordloop:
	cmp al, bl
	je .wordloop_e
	cmp al, 0x0a
	je .wordloop_e

	cmp rsi, r10
	jge .wordloop_e
	inc r11
	lodsb
	jmp .wordloop

.wordloop_e:
	dec r11
	sub rsi, qword [tib]
	mov qword [to_in], rsi
	pop rsi
	pspush rsi
	pspush r11
	ret

.empty:
	pspush 0
	pspush 0
	ret
; }}}

defcode "type", type, 0
	pspop rsi
	pspop rdx
	mov rdi, 1
	mov rax, __NR_write
	syscall
	ret

; r9:  processing temporary value
; r10: input size
; r11: input addr
; r12: pointer into currently processing word
; r13: same as r12 but kept at xt
defcode "find", find, 0
	pspop r10 ; u
	pspop r11 ; a
	mov r12, qword [latest]
	mov r13, r12

.check_smudge:
	add r12, 8
	mov r9b, byte [r12]
	test r9b, smudge_mask
	jnz .no

	inc r12
	mov r9b, byte [r12]
	cmp r9b, r10b
	jne .no

	mov rsi, r12
	inc rsi
	mov rdi, r11
	mov rcx, r10
	repe cmpsb
	jnz .no

	pspush r13
	mov r13, true
	pspush r13
	ret

.no:
	mov r12, qword [r13]
	mov r13, r12
	cmp r12, 0
	jne .check_smudge

	pspush r11
	pspush r10
	mov r13, false
	pspush r13
	ret

; interpret {{{
; r11: word found flag
; r12: state
; TODO respect comp-only flag (do this once error handling is impld)
; TODO it doesnt work right
defcode "interpret", interpret, 0
.loop:
	call parse_name
	cmp qword [r15], 0
	je .eof
	call find
	mov r12, qword [state]
	cmp r12, COMPILING
	je .compl
 
	pspop r11
	cmp r11, false
	je .intrpnum

.callw: ; label here for immed jump
	call to_body
	pspop r13
	call r13
	jmp .loop

.intrpnum:
	call number
	pspop r11
	cmp r11, false
	je .notfound
	jmp .loop

.compl:
	; call dots
	pspop r11
	cmp r11, false
	je .complnum

	; dup immediate? if [imm] else normal then
	; i think this is broken (; not working)
	call dup
	; call dots
	call immediate_q
	; call dots
	pspop r13
	cmp r13, true
	je .callw

	call to_body
	call compile_comma
	jmp .loop

.complnum:
	; call dots
	call number
	; call dots
	pspop r11
	cmp r11, false
	je .notfound
	call literal
	jmp .loop

.notfound:
	; error handling should go here
	jmp .loop

.eof:
	lea r15, [r15+16] ; drop a u
	ret
; }}}

defcode "immediate?", immediate_q, 0
	pspop r11
	add r11, 8
	mov r12b, byte [r11]
	xor r13, r13
	test r12b, immediate_mask
	setnz r13b
	neg r13
	pspush r13
	ret

defcode ">body", to_body, 0
	pspop r11
	add r11, 9
	xor r12, r12
	mov r12b, byte [r11]
	inc r11
	add r11, r12
	pspush r11
	ret

defcode "literal", literal, immediate_mask|comp_only_mask
	; 4d 8d 7f f8
	; 49 bb VAL
	; 4d 89 1f
	pspop r11
	mov r12, qword [dp]

	mov dword [r12], 0xf87f8d4d
	add r12, 4
	mov word [r12], 0xbb49
	add r12, 2
	mov qword [r12], r11
	add r12, 8
	mov word [r12], 0x894d
	add r12, 2
	mov byte [r12], 0x1f
	inc r12

	mov qword [dp], r12
	ret

defcode "compile,", compile_comma, 0
	pspop r11
	mov r12, [dp]

	; compile mov r11, [cfa]
	mov word [r12], 0xbb49
	add r12, 2
	mov qword [r12], r11
	add r12, 8

	; compile call r11
	mov word [r12], 0xff41
	add r12, 2
	mov byte [r12], 0xd3
	inc r12

	mov qword [dp], r12
	ret

defcode "[", lbrac, immediate_mask
	mov qword [state], INTERPRET
	ret

defcode "]", rbrac, immediate_mask
	mov qword [state], COMPILING
	ret

defcode "(header)", brac_header, 0
	pspop r9
	pspop r10
	mov r11, qword [latest]
	mov r12, qword [dp]
	push r12

	mov qword [r12], r11
	add r12, 8
	mov byte [r12], 0
	inc r12
	mov byte [r12], r9b
	inc r12

	mov rcx, r9
	mov rsi, r10
	mov rdi, r12
	rep movsb
	add r12, r9

	mov qword [dp], r12
	pop r12
	pspush r12
	ret

; fix to follow ans: yielding colon-sys
defcode ":", colon, 0
	call parse_name
	; todo check zero
	call brac_header
	pspop r11
	mov qword [latest], r11
	call smudge
	mov qword [state], COMPILING
	ret

; fix to follow ans: reading from colon-sys
; this will not work with :noname or i think does>.
defcode ";", semicolon, immediate_mask
	mov r12, [latest]
	add r12, 8
	mov r13b, [r12]
	and r13b, ~smudge_mask ; does this syntax work? equiv SMUDGEMASK NOT
	mov byte [r12], r13b

	mov r11, qword [dp]
	mov byte [r11], 0xc3 ; ret
	inc r11
	mov qword [dp], r11
	mov qword [state], INTERPRET
	ret

defcode "smudge", smudge, 0
	mov r12, [latest]
	add r12, 8
	mov r13b, [r12]
	xor r13b, smudge_mask
	mov byte [r12], r13b
	ret

; number {{{
defcode "number", number, 0 ; ( c-addr u -- ?n flag )
	pspop r11 ; u
	pspop r12 ; c-addr
	xor r13, r13 ; r13: result
	xor r14, r14 ; r14b: current char
	xor r10, r10 ; r10: negative flag
	mov r9, qword [base]

	cmp r11, 0
	je .no

	mov r14b, byte [r12]
	cmp r14b, '-'
	jnz .enterloop
	mov r10, true
	inc r12
	dec r11

.loop:
	mov r14b, byte [r12]

.enterloop:
	; non numeral = goodbye
	cmp r14b, 48
	jl .no

	sub r14b, 48
	cmp r14b, 10 ; 48+10: < ':', <= '9'
	jl .basecmp
	cmp r14b, 17 ; ':' - '@'
	jl .no
	sub r14b, 7 ; keep 10 so 'A' = 10
	cmp r14b, 36 ; < '[' <= 'Z'
	jl .basecmp
	cmp r14b, 42 ; < 'a'
	jl .no
	sub r14b, 32
	cmp r14b, 36 ; < '{' <= 'z'
	jl .basecmp
	jmp .no

.basecmp:
	cmp r14, r9
	jge .no

	imul r13, r9
	add r13, r14

	inc r12
	dec r11

	cmp r11, 0
	jne .loop

	test r10, r10
	jz .bye
	neg r13

.bye:
	pspush r13
	mov r13, true
	pspush r13
	ret

.no:
	mov r13, false
	pspush r13
	ret
; }}}

; syscall {{{
defcode "syscall0", syscall0, 0
	pspop rax
	syscall
	pspush rax
	ret

defcode "syscall1", syscall1, 0
	pspop rax
	pspop rdi
	syscall
	pspush rax
	ret

defcode "syscall2", syscall2, 0
	pspop rax
	pspop rdi
	pspop rsi
	syscall
	pspush rax
	ret

defcode "syscall3", syscall3, 0
	pspop rax
	pspop rdi
	pspop rsi
	pspop rdx
	syscall
	pspush rax
	ret
; }}}

; stack {{{
defcode "dup", dup, 0
	mov r11, [r15]
	pspush r11
	ret

defcode "2dup", twodup, 0
	mov r11, [r15]
	mov r12, [r15+8]
	pspush r12
	pspush r11
	ret

defcode "swap", swap, 0
	pspop r11
	pspop r12

	pspush r11
	pspush r12
	ret

defcode "over", over, 0
	mov r11, [r15+8]
	pspush r11
	ret

defcode "rot", rot, 0
	pspop r11
	pspop r12
	pspop r13
	pspush r12
	pspush r11
	pspush r13
	ret

defcode "-rot", dash_rot, 0
	pspop r11
	pspop r12
	pspop r13
	pspush r11
	pspush r13
	pspush r12
	ret

defcode "r>", from_r, 0
	pop r12 ; keep return address
	pop r11
	pspush r11
	push r12
	ret

defcode ">r", to_r, 0
	pop r12 ; ret addr
	pspop r11
	push r11
	push r12
	ret

defcode "drop", drop, 0
	pspop r11
	ret

defcode "2drop", twodrop, 0
	pspop r11
	pspop r11
	ret

defcode "rdrop", rdrop, 0
	pop r12 ; ret addr
	pop r11
	push r12
	ret
; }}}

; math + comparison {{{
; i believe some of these could be improved by direct accesses to [r15]
defcode "+", plus, 0
	pspop r11
	pspop r12
	add r11, r12
	pspush r11
	ret

defcode "-", minus, 0
	pspop r11
	pspop r12
	sub r12, r11
	pspush r12
	ret

defcode "*", _times, 0
	pspop r11
	pspop r12
	imul r11, r12
	pspush r11
	ret

defcode "/mod", divmod, 0
	xor rdx, rdx
	pspop r11
	pspop rax
	idiv r11
	pspush rdx
	pspush rax
	ret

defcode "and", _and, 0
	pspop r11
	and [r14], r11
	ret

defcode "or", _or, 0
	pspop r11
	or [r14], r11
	ret

defcode "xor", _xor, 0
	pspop r11
	xor [r14], r11
	ret

defcode "invert", invert, 0
	not qword [r14]
	ret

defcode "*/mod", starslashmod, 0
	pspop r15 ; n3
	pspop r13 ; n2
	pspop rax ; n1

	imul r13
	idiv r15

	pspush rdx
	pspush rax
	ret
; }}}

; TEMPORARY WONKY DEBUGGING FUNCTIONS {{{
; .s {{{
defcode ".s", dots, 0
	push r11
	push r12

	mov r12, r15
.loop:
	cmp r12, wstk
	jge .done

	mov [.space], r12
	mov rdx, 8 ; qword
	mov rsi, .space
	mov rdi, 1
	mov rax, __NR_write
	syscall

	mov rdx, 8 ; qword
	mov rsi, r12
	mov rdi, 1
	mov rax, __NR_write
	syscall
	lea r12, [r12+8]
	jmp .loop

.done:
	mov rdx, 16 ; 2 qword
	mov rsi, .dmsg
	mov rdi, 1
	mov rax, __NR_write
	syscall

	pop r12
	pop r11
	ret
.space: resq 1
.dmsg: db "DONEDONEYIPPEEEE"
; }}}

defcode "HEREDUMP", heredump, 0
	mov rdx, [dp]
	mov r11, [dp0]
	sub rdx, r11
	mov rsi, [dp0]
	mov rdi, 1
	mov rax, __NR_write
	syscall
	ret
; }}}

defvar "state", state, 0, INTERPRET
defvar "base", base, 0, 10
defvar "dp", dp, 0, 0
defvar "dp0", dp0, 0, 0
defvar "dp$", dp$, 0, 0
defvar "tib", tib, 0, initfile
defvar "#tib", n_tib, 0, initlen
defvar ">in", to_in, 0, 0
defvar "latest", latest, 0, lfa_latest

initfile: incbin "sanctuary.fs"
initlen equ $ - initfile