358 lines
8.4 KiB
NASM
358 lines
8.4 KiB
NASM
;; This assembler is a 2-pass compiler, permitting forward and backward label references.
|
|
;; It's also the first stage in this bootstrapping experiment that will
|
|
;; consume proper mnemonic assembly, rather than hex input.
|
|
|
|
section .data
|
|
_start_lbl db "_start", 0
|
|
|
|
;; supported instructions:
|
|
;; mov
|
|
;; test
|
|
;; xor
|
|
;; and
|
|
;; shl
|
|
;; shr
|
|
;; inc
|
|
;; dec
|
|
;; neg
|
|
;; ret
|
|
;; not
|
|
;; add
|
|
;; sub
|
|
;; imul
|
|
;; cmp
|
|
;; lea
|
|
;; push, pop
|
|
;; cmovcc
|
|
;; xchg
|
|
;; bts
|
|
;; bsf
|
|
;; bt
|
|
|
|
;; syscall
|
|
;; jcc rel32
|
|
;; jmp rel32
|
|
;; call rel32
|
|
|
|
section .text
|
|
instr_table:
|
|
; mnemonic, num_operands, handler
|
|
_mov:
|
|
dq "mov"
|
|
db 2
|
|
dq instr_table - _mov_handler
|
|
|
|
;; @param lhs: (rdi, rsi)
|
|
;; @param rhs: (rdx, rcx)
|
|
;; @return al
|
|
strcmp:
|
|
cmp rcx, rsi
|
|
cmovb rsi, rcx ; if rhs is shorter, use its length for the loop
|
|
xor eax, eax
|
|
.strcmp_loop:
|
|
cmp rsi, rax
|
|
je .strcmp_equal
|
|
movzx ecx, byte [rdx + rax]
|
|
cmp byte [rdi + rax], cl
|
|
lea rax, [rax + 1]
|
|
je .strcmp_loop
|
|
seta al ; al = lhs > rhs
|
|
sbb al, 0 ; al = al - CF
|
|
ret
|
|
.strcmp_equal:
|
|
xor eax, eax
|
|
ret
|
|
|
|
memcpy:
|
|
xor rax, rax
|
|
.memcpy_loop:
|
|
cmp rax, rdx
|
|
je .memcpy_done
|
|
mov al, byte [rdi + rax]
|
|
mov byte [rsi + rax], al
|
|
lea rax, [rax + 1]
|
|
jmp .memcpy_loop
|
|
.memcpy_done:
|
|
ret
|
|
|
|
;; read from input file
|
|
read_file:
|
|
|
|
; ;; returns the next byte in the input stream without advancing the read position.
|
|
; peekc:
|
|
|
|
; ;; returns the next byte in the input stream and advances the read position.
|
|
; getc:
|
|
extern peekc
|
|
extern getc
|
|
|
|
;; converts char $dil to a digit with radix $rsi, returning it in $edx. $al is set to 1 if the char is a valid digit, and 0 otherwise.
|
|
to_digit:
|
|
lea eax, [rsi - 2]
|
|
cmp eax, 35
|
|
jae .invalid
|
|
movzx rdi, dil
|
|
lea edx, [rdi - 65] ; 'A' = 65
|
|
and edx, -33 ; convert to uppercase
|
|
add edx, 10 ; 'A' should map to 10
|
|
lea eax, [rdi - 48] ; '0' = 48
|
|
cmp esi, 11
|
|
cmovb edx, eax ; if radix <= 10, then take the difference from '0'
|
|
cmp edi, 58
|
|
cmovb edx, eax ; or if char < '9', then take the difference from '0'
|
|
xor eax, eax
|
|
cmp edx, esi
|
|
setb al ; al = edx < radix
|
|
ret
|
|
.invalid:
|
|
xor eax, eax
|
|
ret
|
|
|
|
;; reads from $rdi and parses digits with radix $rsi until a non-digit is encountered.
|
|
parse_digits:
|
|
sub rsp, 24
|
|
mov qword [rsp + 16], rsi ; save the radix
|
|
mov qword [rsp + 8], rdi ; save the source iterator
|
|
mov qword [rsp], 0
|
|
.loop:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
mov dil, al
|
|
mov rsi, qword [rsp + 16] ; restore the radix
|
|
call to_digit
|
|
test al, al
|
|
jz .done
|
|
mov rax, qword [rsp]
|
|
mov rsi, qword [rsp + 16] ; restore the radix
|
|
imul rax, rsi
|
|
add rax, rdx
|
|
mov qword [rsp], rax
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc
|
|
jmp .loop
|
|
|
|
.done:
|
|
mov rax, [rsp]
|
|
add rsp, 24
|
|
ret
|
|
|
|
;; enum Register {
|
|
; A = 0,
|
|
; B,
|
|
; C,
|
|
; D,
|
|
; Src,
|
|
; Dst,
|
|
; Sp,
|
|
; Bp,
|
|
; R8,
|
|
; R9,
|
|
; R10,
|
|
; R11,
|
|
; R12,
|
|
; R13,
|
|
; R14,
|
|
; R15,
|
|
;; }
|
|
|
|
;; parses a non-extended GPR operand (e.g. a register name that isn't r8-r15)
|
|
;; returns the register number in rax and the width in rdx
|
|
;; if $rdx == 0, then parsing failed (invalid register name)
|
|
;; $rdi = source
|
|
;; $rsi = prefix char (e.g. 'r' or 'e' or 0)
|
|
global parse_gpr
|
|
parse_gpr:
|
|
sub rsp, 24
|
|
mov qword [rsp + 16], -1 ; register number, initialized to -1 (invalid)
|
|
mov qword [rsp + 8], rdi ; source iterator
|
|
mov qword [rsp], rsi ; prefix char
|
|
call getc
|
|
|
|
; check for a,b,c,d,s
|
|
movzx rax, al
|
|
sub rax, 'a' ; convert 'a' to 0, 'b' to 1, ..., 's' to 4
|
|
cmp rax, 's' - 'a'
|
|
ja .invalid ; if it's greater than 's', it's invalid
|
|
mov rcx, 4
|
|
cmovz rax, rcx ; if it's 's', set rax to 4
|
|
cmp rax, 4
|
|
ja .invalid ; if it's greater than 4, it's invalid
|
|
lea rcx, [rel .jt] ; jump table for a,b,c,d
|
|
movsxd rdx, dword [rcx + rax * 4] ; get the offset of the handler if a-d
|
|
add rcx, rdx ; calculate the address of the handler
|
|
jmp rcx
|
|
|
|
.suffix:
|
|
mov byte [rsp + 16], al ; save the register number for a,b,c,d,s
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
cmp al, 'l'
|
|
jne .not_l
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc ; consume the suffix
|
|
mov rdx, 1 ; width = 1 for al, bl, cl, dl, sil, dil
|
|
jmp .done
|
|
.not_l: ; x is only valid for a,b,c,d, not sp, bp, si or di
|
|
mov rdx, 2
|
|
cmp byte [rsp + 16], 4
|
|
jae .done ; for sp, bp, si, di, the width is 2 if not 'l'
|
|
cmp al, 'x'
|
|
jne .not_x
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc ; consume the suffix
|
|
mov rdx, 2 ; width = 2 for ax, bx, cx, dx
|
|
jmp .done
|
|
.not_x: ; x is only valid for a,b,c,d
|
|
cmp al, 'h'
|
|
jne .invalid
|
|
mov rdx, 3 ; width = 3 for ah, bh, ch, dh
|
|
.done:
|
|
movzx rax, byte [rsp + 16] ; move the register number into rax
|
|
add rsp, 24
|
|
ret
|
|
.invalid:
|
|
xor rax, rax ; rax = 0 indicates invalid register
|
|
xor rdx, rdx ; rdx = 0 indicates invalid register
|
|
jmp .done
|
|
|
|
.a:
|
|
mov rax, 0
|
|
jmp .suffix
|
|
.b:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
cmp al, 'p'
|
|
mov rax, 1
|
|
jne .suffix
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc
|
|
mov rax, 5
|
|
jmp .suffix
|
|
.c:
|
|
mov rax, 2
|
|
jmp .suffix
|
|
.d:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
cmp al, 'i'
|
|
mov rax, 3
|
|
jne .suffix
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc
|
|
mov rax, 7
|
|
jmp .suffix
|
|
.s:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call getc
|
|
mov cl, al
|
|
cmp cl, 'i'
|
|
mov rax, 6
|
|
je .suffix
|
|
cmp cl, 'p'
|
|
mov rax, 4
|
|
je .suffix
|
|
jmp .invalid
|
|
|
|
.jt:
|
|
dd .a-.jt
|
|
dd .b-.jt
|
|
dd .c-.jt
|
|
dd .d-.jt
|
|
dd .s-.jt
|
|
|
|
|
|
|
|
|
|
;; parses an extended GPR operand (e.g. r8-r15)
|
|
;; $rdi = source, leading r is already consumed
|
|
parse_egpr:
|
|
sub rsp, 16
|
|
mov qword [rsp + 8], rdi ; source iterator
|
|
mov rsi, 10
|
|
call parse_digits
|
|
cmp rax, 8
|
|
jb .invalid
|
|
cmp rax, 15
|
|
ja .invalid
|
|
mov qword [rsp], rax
|
|
.suffix:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
mov rcx, 1
|
|
cmp al, 'b'
|
|
je .done_getc
|
|
mov rcx, 2
|
|
cmp al, 'w'
|
|
je .done_getc
|
|
mov rcx, 4
|
|
cmp al, 'd'
|
|
je .done_getc
|
|
mov rcx, 8 ; default width is 8 for r8-r15
|
|
jmp .done
|
|
.done_getc:
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
push rcx
|
|
call getc
|
|
pop rcx
|
|
.done:
|
|
mov rdx, rcx
|
|
mov rax, qword [rsp] ; move the register number into rax
|
|
add rsp, 16
|
|
ret
|
|
.invalid:
|
|
mov qword [rsp], 0
|
|
xor rcx, rcx ; rdx = 0 indicates invalid register
|
|
jmp .done
|
|
|
|
;; (rdi, rsi): source string
|
|
global parse_reg
|
|
parse_reg:
|
|
sub rsp, 16
|
|
mov qword [rsp + 8], rdi ; source iterator
|
|
mov byte [rsp], 0 ; x1
|
|
call peekc
|
|
cmp al, 'r'
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
je .r
|
|
cmp al, 'e'
|
|
jne .gpr
|
|
mov byte [rsp], 1 ; x2
|
|
call getc ; consume the 'e'
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
jmp .gpr
|
|
.r:
|
|
mov byte [rsp], 2 ; x4
|
|
call getc ; consume the 'r'
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
call peekc
|
|
mov dil, al
|
|
mov rsi, 10
|
|
call to_digit
|
|
test al, al
|
|
mov rdi, qword [rsp + 8] ; restore the source iterator
|
|
jz .gpr
|
|
call parse_egpr
|
|
jmp .done
|
|
.gpr:
|
|
call parse_gpr
|
|
mov cl, byte [rsp]
|
|
shl rdx, cl
|
|
.done:
|
|
add rsp, 16
|
|
ret
|
|
|
|
|
|
;; enum Operand {
|
|
;; Reg(Reg, width: u8),
|
|
;; Imm(u32, width: u8),
|
|
;; Mem(width: u8, base: Option<Reg>, index: Option<(Reg, scale: u8)>, disp: Option<u32>),
|
|
;; }
|
|
|
|
;; mov r/m, r
|
|
;; mov r, r/m
|
|
;; mov r/m, imm
|
|
_mov_handler:
|
|
ret
|
|
|
|
|