Add stages and a rakefile

1 files changed, 886 insertions, 8 deletions

diff --git a/boot/asm-boot.s b/boot/asm-boot.s
index 594c624..80167a2 100644
--- a/boot/asm-boot.s
+++ b/boot/asm-boot.s
@@ -1,18 +1,897 @@
 # s1 - Contains the current position in the source text.
 
-.data
+.global _start # Program entry point.
+
+.section .rodata
 
-.equ SYS_EXIT, 93
 .equ SOURCE_BUFFER_SIZE, 2048
 
-.bss
+asm_begin: .ascii ".text\n.global _start\n_start:\naddi sp, sp, -64\nsw ra, 60(sp)\nsw s0, 56(sp)\naddi s0, sp, 64\n"
+.equ ASM_BEGIN_SIZE, . - asm_begin
+asm_end: .ascii "addi a0, zero, 0\naddi a7, zero, 93\necall\nlw ra, 60(sp)\nlw s0, 56(sp)\naddi sp, sp, 64\nret\n"
+.equ ASM_END_SIZE, . - asm_end
+asm_program: .ascii ".bss\n"
+.equ ASM_PROGRAM_SIZE, . - asm_program
+asm_type: .ascii ".type "
+.equ ASM_TYPE_SIZE, . - asm_type
+asm_object: .ascii ", @object\n"
+.equ ASM_OBJECT_SIZE, . - asm_object
+asm_size: .ascii ".size "
+.equ ASM_SIZE_SIZE, . - asm_size
+asm_zero: .ascii ".zero "
+.equ ASM_ZERO_SIZE, . - asm_zero
+asm_global: .ascii ".global "
+.equ ASM_GLOBAL_SIZE, . - asm_global
+
+token_begin: .string "begin"
+token_end: .string "end"
+token_import: .string "import"
+token_open_paren: .string "("
+token_close_paren: .string ")"
+token_open_square: .string "["
+token_assign: .string ":="
+token_var: .string "var"
+token_comma: .string ","
+
+space: .ascii " "
+comma: .ascii ","
+new_line: .ascii "\n"
+colon: .ascii ":"
+digit_zero: .ascii "0"
+
+instruction_la: .ascii "la"
+instruction_call: .ascii "call"
+instruction_addi: .ascii "addi"
+instruction_li: .ascii "li"
+instruction_sw: .ascii "sw"
+instruction_lw: .ascii "lw"
+
+register_a0: .ascii "a0"
+register_sp: .ascii "sp"
+register_a: .ascii "a"
+
+.section .bss
 .global source_code
 .type source_code, @object
 .size source_code, SOURCE_BUFFER_SIZE
 source_code: .zero SOURCE_BUFFER_SIZE
 
-.text
-.global _start # Program entry point.
+.section .text
+
+.type _skip_spaces, @function
+_skip_spaces:
+.Lspace_loop_do:
+	lbu t0, (s1) # t0 = Current character.
+
+	li t1, ' '
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\t'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\n'
+	beq t0, t1, .Lspace_loop_repeat
+	li t1, '\r'
+	beq t0, t1, .Lspace_loop_repeat
+
+	j .Lspace_loop_end
+.Lspace_loop_repeat:
+	addi s1, s1, 1
+	j .Lspace_loop_do
+
+.Lspace_loop_end:
+	ret
+
+# Compares two string, which of one has a length, the other one is null-terminated.
+#
+# a0 - The address of the token string.
+# a1 - The length of the string in a0.
+# a2 - The address of the null-terminated string.
+#
+# If the strings match sets a0 to 0, otherwise sets it to 1.
+.type _token_compare, @function
+_token_compare:
+	addi t0, a0, 0
+	addi t1, a1, 0
+	addi t2, a2, 0
+
+.Ltoken_compare_loop:
+	lbu t3, (t2)
+
+	# Will only be 0 if the current character in the null terminated string is \0 and the remaining length of the
+	# another string is 0.
+	or t4, t3, t1
+	beqz t4, .Ltoken_compare_equal
+
+	beqz t1, .Ltoken_compare_not_equal
+	beqz t3, .Ltoken_compare_not_equal
+
+	lbu t4, (t0)
+	bne t3, t4, .Ltoken_compare_not_equal
+
+	addi t0, t0, 1
+	addi t1, t1, -1
+	addi t2, t2, 1
+	j .Ltoken_compare_loop
+
+.Ltoken_compare_not_equal:
+	li a0, 1
+	j .Ltoken_compare_end
+
+.Ltoken_compare_equal:
+	li a0, 0
+
+.Ltoken_compare_end:
+	ret
+
+# Reads a token and returns its length in a0.
+# _read_token doesn't change s1, it finds the length of the token s1 is pointing to.
+.type _read_token, @function
+_read_token:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	lbu t0, (s1) # t0 = Current character.
+	sw zero, 4(sp)
+
+	li t1, '.'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ','
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ':'
+	beq t0, t1, .Ltoken_character_colon
+
+	li t1, ';'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '('
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ')'
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, '['
+	beq t0, t1, .Ltoken_character_single
+
+	li t1, ']'
+	beq t0, t1, .Ltoken_character_single
+
+.Ltoken_character_loop_do: # Expect an identifier or a number.
+	lw t6, 4(sp)
+	add t1, s1, t6
+	lbu a0, (t1) # a0 = Current character.
+
+	call is_alnum
+
+	beqz a0, .Ltoken_character_end
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_loop_do
+
+.Ltoken_character_single:
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+	j .Ltoken_character_end
+
+.Ltoken_character_colon:
+	lbu t0, 1(s1) # t0 = The character after the colon.
+	lw t6, 4(sp)
+	addi t6, t6, 1
+	sw t6, 4(sp)
+
+	li t1, '='
+	beq t0, t1, .Ltoken_character_single
+	j .Ltoken_character_end
+
+.Ltoken_character_end:
+	lw a0, 4(sp)
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+# Generate entry point symbol.
+.type _compile_begin, @function
+_compile_begin:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	# Write initial assembler.
+	la a0, asm_begin
+	addi a1, zero, ASM_BEGIN_SIZE
+	call write_out
+
+	addi s1, s1, 5
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Generate program termination code.
+.type _compile_end, @function
+_compile_end:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	# Write closing assembler.
+	la a0, asm_end
+	addi a1, zero, ASM_END_SIZE
+	call write_out
+
+	addi s1, s1, 3
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Ignores the import.
+.type _compile_import, @function
+_compile_import:
+	# Prologue.
+	addi sp, sp, -8
+	sw ra, 4(sp)
+	sw s0, 0(sp)
+	addi s0, sp, 8
+
+	addi s1, s1, 6
+	call _skip_spaces
+	call _read_token
+	add s1, s1, a0 # Skip the imported module name.
+
+	# Epilogue.
+	lw ra, 4(sp)
+	lw s0, 0(sp)
+	addi sp, sp, 8
+	ret
+
+# Compiles a procedure call. Expects s1 to point to the first argument.
+# a0 - Pointer to the procedure name.
+# a1 - Length of the procedure name.
+#
+# Returns the procedure result in a0.
+.type _compile_call, @function
+_compile_call:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp)
+	sw a1, 16(sp)
+	sw zero, 12(sp) # Argument count for a procedure call.
+
+.Lcompile_call_paren:
+	call _skip_spaces
+	call _read_token
+	addi a1, a0, 0
+	addi a0, s1, 0
+	la a2, token_close_paren
+	call _token_compare
+	beqz a0, .Lcompile_call_complete
+
+.Lcompile_call_argument:
+	call _build_expression
+
+	la a0, instruction_sw
+	li a1, 2
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	la a0, register_a0
+	li a1, 2
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	lw t0, 12(sp) # Argument count for a procedure call.
+
+	# Only 8 arguments are supported with a0-a7.
+	# Save all arguments on the stack so they aren't overriden afterwards.
+	# The offset on the stack always has two digits in this case.
+	li t1, -4
+	mul t1, t0, t1
+	addi t1, t1, 52
+	li t2, 10
+	div t3, t1, t2
+	rem t4, t1, t2
+	addi t3, t3, '0'
+	addi t4, t4, '0'
+
+	sw t3, 8(sp)
+	sw t4, 4(sp)
+
+	addi a0, sp, 8
+	li a1, 1
+	call write_out
+
+	addi a0, sp, 4
+	li a1, 1
+	call write_out
+
+	la a0, token_open_paren
+	li a1, 1
+	call write_out
+
+	la a0, register_sp
+	li a1, 2
+	call write_out
+
+	la a0, token_close_paren
+	li a1, 1
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	call _skip_spaces
+	call _read_token
+	addi a1, a0, 0
+	addi a0, s1, 0
+	la a2, token_comma
+	call  _token_compare
+
+	/* DEBUG. Write the current token to stderr.
+	addi a0, zero, STDERR
+	addi a1, s1, 0
+	li a2, 4 #(sp)
+	addi a7, zero, SYS_WRITE
+	ecall
+	addi a0, zero, STDERR
+	la a1, token_open_square
+	li a2, 1
+	addi a7, zero, SYS_WRITE
+	ecall
+	DEBUG. End */
+
+	bnez a0, .Lcompile_call_paren
+
+	lw t0, 12(sp) # Argument count for a procedure call.
+	addi t0, t0, 1
+	sw t0, 12(sp)
+
+	addi s1, s1, 1 # Skip the comma between the arguments.
+	j .Lcompile_call_argument
+
+.Lcompile_call_complete:
+	sw zero, 12(sp)
+
+.Lcompile_call_restore:
+	# Just go through all a0-a7 registers and read them from stack.
+	# If this stack value contains garbage, the procedure just shouldn't use it.
+	lw t0, 12(sp)
+	li t1, 7
+	bgt t0, t1, .Lcompile_call_perform
+
+	la a0, instruction_lw
+	li a1, 2
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	la a0, register_a
+	li a1, 1
+	call write_out
+
+	lw t0, 12(sp)
+	addi t0, t0, '0'
+	sw t0, 8(sp)
+
+	addi a0, sp, 8
+	li a1, 1
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	lw t0, 12(sp) # Argument count for a procedure call.
+
+	li t1, -4
+	mul t1, t0, t1
+	addi t1, t1, 52
+	li t2, 10
+	div t3, t1, t2
+	rem t4, t1, t2
+	addi t3, t3, '0'
+	addi t4, t4, '0'
+
+	sw t3, 8(sp)
+	sw t4, 4(sp)
+
+	addi a0, sp, 8
+	li a1, 1
+	call write_out
+
+	addi a0, sp, 4
+	li a1, 1
+	call write_out
+
+	la a0, token_open_paren
+	li a1, 1
+	call write_out
+
+	la a0, register_sp
+	li a1, 2
+	call write_out
+
+	la a0, token_close_paren
+	li a1, 1
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	lw t0, 12(sp) # Increment.
+	addi t0, t0, 1
+	sw t0, 12(sp)
+
+	j .Lcompile_call_restore
+
+.Lcompile_call_perform:
+	la a0, instruction_call
+	li a1, 4
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	call _skip_spaces
+	addi s1, s1, 1 # Skip the close paren.
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile, @function
+compile:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	# Write .bss section header for global variables.
+	la a0, asm_program
+	addi a1, zero, ASM_PROGRAM_SIZE
+	call write_out
+
+	call _skip_spaces
+	addi s1, s1, 7 # Skip "program" keyword.
+
+.Lcharacter_loop_do:
+	call _skip_spaces
+
+	lbu t0, (s1) # t0 = Current character.
+	beqz t0, .Lcharacter_loop_end
+
+	call _read_token
+	sw a0, 4(sp) # Save the token length on the stack.
+	beqz a0, .Lcharacter_loop_end # No token read, there is unrecognized input.
+
+	lw a0, 4(sp)
+	call _handle_token
+
+	j .Lcharacter_loop_do
+.Lcharacter_loop_end:
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+# Evalutes an expression and saves the result in a0.
+.type _build_expression, @function
+_build_expression:
+	# Prologue.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	call _skip_spaces
+	call _read_token
+	sw s1, 4(sp)
+	sw a0, 0(sp)
+
+	# Integer literal.
+	addi a0, s1, 0
+	lb a0, (a0)
+	call is_digit
+	bnez a0, .Lbuild_expression_number_literal
+
+	# Named identifier.
+	la a0, instruction_la
+	li a1, 2
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	la a0, register_a0
+	li a1, 2
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	lw a0, 4(sp)
+	lw a1, 0(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	j .Lbuild_expression_end
+
+.Lbuild_expression_number_literal:
+	la a0, instruction_li
+	li a1, 2
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	la a0, register_a0
+	li a1, 2
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	lw a0, 4(sp)
+	lw a1, 0(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	j .Lbuild_expression_end
+
+.Lbuild_expression_end:
+	lw a0, 0(sp)
+	add s1, s1, a0
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
+	ret
+
+# Compiles a statement beginning with an identifier.
+.type _compile_identifier, @function
+_compile_identifier:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# Save the pointer to the identifier and its length on the stack.
+	sw a0, 20(sp)
+	sw a1, 16(sp)
+
+	add s1, s1, a1
+	call _skip_spaces
+	call _read_token
+
+	# Save the pointer and the length of the token following the identifier.
+	sw s1, 12(sp)
+	sw a0, 8(sp)
+
+	add s1, s1, a0 # Skip that token.
+	call _skip_spaces
+
+	lw a0, 12(sp)
+	lw a1, 8(sp)
+	la a2, token_assign
+	call _token_compare
+	beqz a0, .Lcompile_identifier_assign
+
+	lw a0, 12(sp)
+	lw a1, 8(sp)
+	la a2, token_open_paren
+	call _token_compare
+	beqz a0, .Lcompile_identifier_call
+
+	j .Lcompile_identifier_end
+
+.Lcompile_identifier_call:
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call _compile_call
+
+	j .Lcompile_identifier_end
+
+.Lcompile_identifier_assign:
+	call _build_expression
+
+	la a0, instruction_addi
+	li a1, 4
+	call write_out
+
+	la a0, space
+	li a1, 1
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	la a0, register_a0
+	li a1, 2
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	la a0, digit_zero
+	li a1, 1
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	j .Lcompile_identifier_end
+
+.Lcompile_identifier_end:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _compile_var, @function
+_compile_var:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	# Variable name.
+	addi s1, s1, 3
+	call _skip_spaces
+	call _read_token
+	sw s1, 20(sp)
+	sw a0, 16(sp)
+	add s1, s1, a0
+
+	# Skip the colon.
+	call _skip_spaces
+	call _read_token
+	add s1, s1, a0
+
+	call _skip_spaces
+	call _read_token
+	sw a0, 12(sp)
+
+	addi a0, s1, 0
+	lw a1, 12(sp)
+	la a2, token_open_square
+	call _token_compare
+	beqz a0, .Lcompile_var_array
+
+	j .Lcompile_var_end
+
+.Lcompile_var_array:
+	call _skip_spaces
+	add s1, s1, 1 # Skip the opening square bracket.
+
+	call _skip_spaces
+	call _read_token
+	sw a0, 8(sp)
+
+	la a0, asm_type
+	li a1, ASM_TYPE_SIZE
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, asm_object
+	li a1, ASM_OBJECT_SIZE
+	call write_out
+
+	la a0, asm_size
+	li a1, ASM_SIZE_SIZE
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, comma
+	li a1, 1
+	call write_out
+
+	addi a0, s1, 0
+	lw a1, 8(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, colon
+	li a1, 1
+	call write_out
+
+	la a0, asm_zero
+	li a1, ASM_ZERO_SIZE
+	call write_out
+
+	addi a0, s1, 0
+	lw a1, 8(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	la a0, asm_global
+	li a1, ASM_GLOBAL_SIZE
+	call write_out
+
+	lw a0, 20(sp)
+	lw a1, 16(sp)
+	call write_out
+
+	la a0, new_line
+	li a1, 1
+	call write_out
+
+	lw a0, 8(sp)
+	add s1, s1, a0
+
+	call _skip_spaces
+	add s1, s1, 1 # Skip the closing square bracket.
+
+	call _skip_spaces
+	call _read_token
+
+	sw a0, 12(sp)
+	j .Lcompile_var_end
+
+.Lcompile_var_end:
+	lw a0, 12(sp)
+	add s1, s1, a0
+
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
+
+.type _handle_token, @function
+_handle_token:
+	# Prologue.
+	addi sp, sp, -32
+	sw ra, 28(sp)
+	sw s0, 24(sp)
+	addi s0, sp, 32
+
+	sw a0, 20(sp)
+
+	# Detect what token has been read.
+	addi a0, s1, 0
+	lw a1, 20(sp)
+	la a2, token_begin
+	call _token_compare
+	beqz a0, .Lhandle_token_begin
+
+	addi a0, s1, 0
+	lw a1, 20(sp)
+	la a2, token_end
+	call _token_compare
+	beqz a0, .Lhandle_token_end
+
+	addi a0, s1, 0
+	lw a1, 20(sp)
+	la a2, token_import
+	call _token_compare
+	beqz a0, .Lhandle_token_import
+
+	addi a0, s1, 0
+	lw a1, 20(sp)
+	la a2, token_var
+	call _token_compare
+	beqz a0, .Lhandle_token_var
+
+	# If the first symbol in the token is a character, assume an identifier.
+	addi a0, s1, 0
+	lb a0, (a0)
+	call is_alpha
+	bnez a0, .Lhandle_token_identifier
+
+	# Ignore the unknown token.
+	lw t0, 20(sp)
+	add s1, s1, t0
+	j .Lhandle_token_return
+
+.Lhandle_token_begin:
+	call _compile_begin
+	j .Lhandle_token_return
+
+.Lhandle_token_end:
+	call _compile_end
+	j .Lhandle_token_return
+
+.Lhandle_token_import:
+	call _compile_import
+	j .Lhandle_token_return
+
+.Lhandle_token_var:
+	call _compile_var
+	j .Lhandle_token_return
+
+.Lhandle_token_identifier:
+	addi a0, s1, 0
+	lw a1, 20(sp)
+	call _compile_identifier
+	j .Lhandle_token_return
+
+.Lhandle_token_return:
+	# Epilogue.
+	lw ra, 28(sp)
+	lw s0, 24(sp)
+	addi sp, sp, 32
+	ret
 
 _start:
 	# Read the source from the standard input.
@@ -24,6 +903,5 @@ _start:
 	call compile
 
 	# Call exit.
-	addi a0, zero, 0   # Use 0 return code.
-	addi a7, zero, SYS_EXIT
-	ecall
+	li a0, 0   # Use 0 return code.
+	call exit