; TODO Use memory area given in argument to preserve stack between calls
[bits 64]

; local variables macro
%define stack_size [rbp - 8]
%define stack_base [rbp - 16]
%define brkinit [rbp - 24]
%define args_ptr [rbp - 32]

%define stack_cur r11

; push & pop the RPN stack macros
; check if stack full, then push value
%macro rpn_push 1
	inc stack_cur
	cmp stack_cur, stack_size
	jl %%end
	xor stack_cur, stack_cur
	%%end:
	mov [r10 + stack_cur * 8], %1
%endmacro

; check if stack empty, then pop value
%macro rpn_pop 1
	mov %1, [r10 + stack_cur * 8]
	cmp stack_cur, 0
	jg %%end
	mov stack_cur, stack_size
	%%end:
	dec stack_cur
%endmacro

; define a label with code portion size
; and set global
%macro part_sz 1
	%1_sz: dq $ - %1
	global %1
	global %1_sz
%endmacro

section .data

rpn_exec:
	; unsigned long int rpn_exec(unsigned long int stack_size, unsigned long args)
	; rdi -> stack size (in values (64 bits))
	; rsi -> args pointers
	
	enter 32, 0
	
	mov stack_size, rdi
	mov args_ptr, rsi
	mov r10, rdx ; r10 is stack base

	xor rdx, rdx
	mov brkinit, rdx

	cmp r10, 0
	jne .nobrk
	; if stack is NULL allocate using brk

	mov rax, 0x0C
	xor rdi, rdi ; get brk
	syscall

	mov brkinit, rax
	mov rdi, stack_size
	shl rdi, 3 ; mul 8
	add rdi, rax
	mov rax, 0x0C
	syscall ; new brk

	cmp rax, -1
	jne .cont
	leave
	ret
	.cont:

	mov rcx, stack_size
	mov rdi, brkinit
	xor rax, rax
	.bzero:
		stosq
		loop .bzero
	mov r10, brkinit

	.nobrk:
	xor stack_cur, stack_cur ; r11 is stack item counter
	

part_sz rpn_exec

rpn_exec_ret:
	rpn_pop rax
	mov rdi, brkinit
	cmp rdi, 0
	jne .end
	push rax
	mov rax, 0x0C
	mov rdi, brkinit
	syscall
	pop rax
	.end:
	leave
	ret
part_sz rpn_exec_ret

rpn_value:
	mov rax, strict qword 0x12345678
	rpn_push rax
part_sz rpn_value

rpn_arg:
	mov rax, strict qword 0x12345678
	mov rsi, 8
	mul rsi
	mov rsi, args_ptr
	add rsi, rax
	mov rax, [rsi]
	rpn_push rax
part_sz rpn_arg

rpn_add:
	rpn_pop rax
	rpn_pop rbx
	add rax, rbx
	rpn_push rax
part_sz rpn_add

rpn_sub:
	rpn_pop rbx
	rpn_pop rax
	sub rax, rbx
	rpn_push rax
part_sz rpn_sub

rpn_mul:
	rpn_pop rbx
	rpn_pop rax
	mul rbx
	rpn_push rax
part_sz rpn_mul

rpn_div:
	rpn_pop rbx
	rpn_pop rax
	xor rdx, rdx
	test rbx, rbx
	jz .zerodiv
	div rbx
	jmp .end
	.zerodiv: xor rax, rax
	.end:
	rpn_push rax
part_sz rpn_div

rpn_mod:
	rpn_pop rbx
	rpn_pop rax
	xor rdx, rdx
	test rbx, rbx
	jz .zerodiv
	div rbx
	jmp .end
	.zerodiv: xor rax, rax
	.end:
	rpn_push rdx
part_sz rpn_mod

rpn_neg:
	rpn_pop rax
	neg rax
	rpn_push rax
part_sz rpn_neg

rpn_not:
	rpn_pop rax
	not rax
	rpn_push rax
part_sz rpn_not

rpn_and:
	rpn_pop rbx
	rpn_pop rax
	and rax, rbx
	rpn_push rax
part_sz rpn_and

rpn_or:
	rpn_pop rbx
	rpn_pop rax
	or rax, rbx
	rpn_push rax
part_sz rpn_or

rpn_xor:
	rpn_pop rbx
	rpn_pop rax
	xor rax, rbx
	rpn_push rax
part_sz rpn_xor

rpn_shl:
	rpn_pop rcx
	rpn_pop rax
	cmp rcx, 64
	jge .zero
	shl rax, cl
	jmp .end
	.zero:
	xor rax, rax
	.end:
	rpn_push rax
part_sz rpn_shl

rpn_shr:
	rpn_pop rcx
	rpn_pop rax
	cmp rcx, 64
	jge .zero
	shr rax, cl
	jmp .end
	.zero:
	xor rax, rax
	.end:
	rpn_push rax
part_sz rpn_shr

rpn_xchg:
	rpn_pop rbx
	rpn_pop rax
	rpn_push rbx
	rpn_push rax
part_sz rpn_xchg

rpn_dup:
	rpn_pop rax
	rpn_push rax
	rpn_push rax
part_sz rpn_dup

rpn_pop_op:
	rpn_pop rax
part_sz rpn_pop_op

section .text