(* s1 - Contains the current position in the source text. - The compiler expects valid input, otherwise it will generate invalid assembly or hang. There is no error checking, no semantic analysis, no type checking. - Imports with only a module name without package, e.g. "import dummy", can be parsed, but are ignored. - No loops. Only labels and goto. - Only unsigned number literals are supported (in decimal or hexadecimal format). - Comments are accepted only at the end of a line. - Return can be used only as the last statement of a procedure. It doesn't actually return, but sets a0 to the appropriate value. *) program import dummy const SOURCE_BUFFER_SIZE := 81920 var source_code: [81920]Byte (* Ignores the import. *) proc _compile_import() var loca0: Word begin _advance(6) _skip_spaces() loca0 := _read_token() _advance(loca0) (* Skip the imported module name. *) end proc _build_binary_expression() var loca0, loca4, loca8, loca20: Word begin _build_expression(0) loca4 := 0x2c306120 (* _a0, *) loca8 := 0x0a316120 (* _a1\n *) _skip_spaces() loca20 := _read_token() li t0, 0x26 # & sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_and li t0, 0x726f # or sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_or li t0, 0x3d # = sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_equal li t0, 0x2b # + sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_plus li t0, 0x2d # - sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_minus li t0, 0x2a # * sw t0, 16(sp) mv a0, s1 lw a1, 20(sp) addi a2, sp, 16 call _token_compare beqz a0, .L_build_binary_expression_product goto .Lbuild_binary_expression_end .L_build_binary_expression_equal _advance(1) (* Skip =. *) _build_expression(1) _write_out(@loca0, 3) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) loca0 := 0x7a716573 (* seqz *) _write_out(@loca0, 4) _write_out(@loca4, 4) _write_out(@loca4, 4) goto .Lbuild_binary_expression_end .L_build_binary_expression_and _advance(1) (* Skip &. *) _build_expression(1) loca0 := 0x646e61 (* and *) _write_out(@loca0, 3) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) goto .Lbuild_binary_expression_end .L_build_binary_expression_or _advance(2) (* Skip or. *) _build_expression(1) loca0 := 0x726f (* or *) _write_out(@loca0, 2) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) goto .Lbuild_binary_expression_end .L_build_binary_expression_plus _advance(1) (* Skip +. *) _build_expression(1) loca0 := 0x646461 (* add *) _write_out(@loca0, 3) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) goto .Lbuild_binary_expression_end .L_build_binary_expression_minus _advance(1) (* Skip -. *) _build_expression(1) loca0 := 0x627573 (* sub *) _write_out(@loca0, 3) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) goto .Lbuild_binary_expression_end .L_build_binary_expression_product _advance(1) (* Skip *. *) _build_expression(1) loca0 := 0x6c756d (* mul *) _write_out(@loca0, 3) _write_out(@loca4, 4) _write_out(@loca4, 4) _write_out(@loca8, 4) goto .Lbuild_binary_expression_end .Lbuild_binary_expression_end end (* Evalutes an expression and saves the result in a0. a0 - X in aX, the register number to save the result. *) proc _build_expression() var loca0, loca20, loca28, loca8: Word loca24, loca4: ^Byte begin (* Make the register number to a character and save it. *) loca28 := loca84 + 0x30 (* 0 *) _skip_spaces() loca20 := _read_token() loca24 := _current() lbu a0, (s1) li t0, '-' beq a0, t0, .Lbuild_expression_negate lbu a0, (s1) li t0, '@' beq a0, t0, .Lbuild_expression_address loca0 := _front() if _is_digit(loca0) then goto .Lbuild_expression_literal end lbu a0, (s1) li t0, '_' beq a0, t0, .Lbuild_expression_call li t0, 0x61636f6c # loca sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 4 call _memcmp beqz a0, .Lbuild_expression_identifier (* Named identifier. *) loca8 := 0x6120616c (* la a *) _write_out(@loca8, 4) loca8 := 0x00202c00 or loca28 _write_out(@loca8, 3) _write_out(loca24, loca20) _put_char(0x0a) goto .Lbuild_expression_advance .Lbuild_expression_negate _advance(1) (* Skip the -. *) _build_expression(0) loca8 := 0x2067656e (* neg_ *) _write_out(@loca8, 4) loca8 := 0x202c3061 (* a0,_ *) _write_out(@loca8, 4) loca8 := 0x0a3061 (* a0,_ *) _write_out(@loca8, 3) goto .Lbuild_expression_advance .Lbuild_expression_address loca8 := 0x69646461 (* addi *) _write_out(@loca8, 4) loca8 := 0x6120 (* _a *) _write_out(@loca8, 2) _put_char(loca28) loca8 := 0x7073202c (* , sp *) _write_out(@loca8, 4) loca8 := 0x202c (* ,_ *) _write_out(@loca8, 2) _advance(1) (* Skip @. *) _skip_spaces() loca24 := _current() loca20 := _read_token() (* Skip the "loca" variable prefix. *) loca4 := loca24 + 4 loca0 := loca20 - 4 _write_out(loca4, loca0) _put_char(0xa) goto .Lbuild_expression_advance .Lbuild_expression_identifier loca8 := 0x6120776c (* lw a *) _write_out(@loca8, 4) loca8 := 0x00202c00 or loca28 # \0,_ _write_out(@loca8, 3) loca4 := loca24 + 4 loca0 := loca20 - 4 _write_out(loca4, loca0) (* Skip the "loca" variable prefix. *) loca8 := 0x29707328 (* (sp) *) _write_out(@loca8, 4) _put_char(0x0a) (* \n *) goto .Lbuild_expression_advance .Lbuild_expression_call _advance(loca20) _advance(1) _compile_call(loca24, loca20) goto .Lbuild_expression_end .Lbuild_expression_literal loca8 := 0x6120696c (* li a *) _write_out(@loca8, 4) loca8 := 0x00202c00 or loca28 (* \0,_ *) _write_out(@loca8, 3) _write_out(loca24, loca20) _put_char(0x0a) (* \n *) goto .Lbuild_expression_advance .Lbuild_expression_advance _advance(loca20) .Lbuild_expression_end end (* Compiles a statement beginning with an identifier. Left values should be variables named "loca n", where n is the offset of the variable on the stack, like loca8 or loca4. *) proc _compile_identifier() var loca0, loca16, loca8: Word loca20, loca12: ^Byte begin (* Save the pointer to the identifier and its length on the stack. *) loca20 := _current() loca16 := _read_token() _advance(loca16) _skip_spaces() (* Save the pointer and the length of the token following the identifier. *) loca12 := _current() loca8 := _read_token() _advance(loca8) # Skip that token. _skip_spaces() li t0, 0x3d3a # := sw t0, 4(sp) lw a0, 12(sp) lw a1, 8(sp) addi a2, sp, 4 call _token_compare beqz a0, .Lcompile_identifier_assign lw t0, 12(sp) lbu t0, (t0) li t1, 0x28 # ( beq t0, t1, .Lcompile_identifier_call goto .Lcompile_identifier_end .Lcompile_identifier_call _compile_call(loca20, loca16) goto .Lcompile_identifier_end .Lcompile_identifier_assign _build_binary_expression() loca0 := 0x61207773 (* sw a *) _write_out(@loca0, 4) loca0 := 0x202c30 (* 0,_ *) _write_out(@loca0, 3) loca20 := loca20 + 4 loca16 := loca16 - 4 _write_out(loca20, loca16) loca0 := 0x29707328 (* (sp) *) _write_out(@loca0, 4) _put_char(0x0a) (* \n *) goto .Lcompile_identifier_end .Lcompile_identifier_end end (* 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. *) proc _compile_call(loca84: ^Byte, loca80: Word) var loca0, loca4, loca12: Word begin loca12 := 0 (* Argument count for a procedure call. *) .Lcompile_call_paren _skip_spaces() lbu t0, (s1) li t1, 0x29 # ) beq t0, t1, .Lcompile_call_complete .Lcompile_call_argument _build_expression(0) loca0 := 0x61207773 (* sw a *) _write_out(@loca0, 4) loca0 := 0x202c30 (* 0,_ *) _write_out(@loca0, 3) (* Only 6 arguments are supported with a0-a5. Save all arguments on the stack so they aren't overriden afterwards. *) loca0 := -4 * loca12 loca0 := loca0 + 60 _printi(loca0) loca0 := 0x29707328 (* (sp) *) _write_out(@loca0, 4) _put_char(0x0a) (* \n *) _skip_spaces() lbu t0, (s1) li t1, ',' bne t0, t1, .Lcompile_call_paren loca12 := loca12 + 1 (* Argument count for a procedure call. *) _advance(1) (* Skip the comma between the arguments. *) goto .Lcompile_call_argument .Lcompile_call_complete loca12 := 0 .Lcompile_call_restore (* Just go through all a0-a5 registers and read them from stack. If this stack value contains garbage, the procedure just shouldn't use it. *) loca0 := 0x6120776c (* lw a *) (* lw a0, 60(sp) *) _write_out(@loca0, 4) loca4 := 0x36202c30 (* 0, 6 *) _write_out(@loca4, 4) loca4 := 0x70732830 (* 0(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) (* lw a1, 56(sp) *) _write_out(@loca0, 4) loca4 := 0x35202c31 (* 1, 5 *) _write_out(@loca4, 4) loca4 := 0x70732836 (* 6(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) (* lw a2, 52(sp) *) _write_out(@loca0, 4) loca4 := 0x35202c32 (* 2, 5 *) _write_out(@loca4, 4) loca4 := 0x70732832 (* 2(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) (* lw a3, 48(sp) *) _write_out(@loca0, 4) loca4 := 0x34202c33 (* 3, 4 *) _write_out(@loca4, 4) loca4 := 0x70732838 (* 8(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) (* lw a4, 44(sp) *) _write_out(@loca0, 4) loca4 := 0x34202c34 (* 4, 4 *) _write_out(@loca4, 4) loca4 := 0x70732834 (* 4(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) (* lw a5, 40(sp) *) _write_out(@loca0, 4) loca4 := 0x34202c35 (* 5, 4 *) _write_out(@loca4, 4) loca4 := 0x70732830 (* 0(sp *) _write_out(@loca4, 4) loca4 := 0x0a29 (* )\n *) _write_out(@loca4, 2) loca0 := 0x6c6c6163 (* call *) _write_out(@loca0, 4) _put_char(0x20) (* _ *) _write_out(loca84, loca80) _put_char(0x0a) (* \n *) _skip_spaces() _advance(1) (* Skip the close paren. *) end (* 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. *) proc _read_token() var loca4: Word begin lbu t0, (s1) # t0 = Current character. loca4 := 0 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 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 li t1, '*' beq t0, t1, .Ltoken_character_single li t1, '@' beq t0, t1, .Ltoken_character_single (* Expect an identifier or a number. *) .Ltoken_character_loop_do lw t6, 4(sp) add t1, s1, t6 lbu a0, (t1) # a0 = Current character. call _is_alnum beqz a0, .Ltoken_character_end loca4 := loca4 + 1 goto .Ltoken_character_loop_do .Ltoken_character_single loca4 := loca4 + 1 goto .Ltoken_character_end .Ltoken_character_colon lbu t0, 1(s1) # t0 = The character after the colon. loca4 := loca4 + 1 li t1, '=' beq t0, t1, .Ltoken_character_single goto .Ltoken_character_end .Ltoken_character_end lw a0, 4(sp) end (* Skips the spaces till the next non space character. *) proc _skip_spaces() begin .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 goto .Lspace_loop_end .Lspace_loop_repeat _advance(1) goto .Lspace_loop_do .Lspace_loop_end end (* Skips tabs at the line beginning. *) proc _skip_indentation() begin .Lskip_indentation_do lbu t0, (s1) li t1, '\t' beq t0, t1, .Lskip_indentation_skip goto .Lskip_indentation_end .Lskip_indentation_skip _advance(1) goto .Lskip_indentation_do .Lskip_indentation_end end (* Parameters: a0 - Line length. *) proc _skip_comment(loca84: Word) begin (* Check whether this is a comment. *) li t0, 0x2a28 # (* sw t0, 4(sp) addi a0, sp, 4 mv a1, s1 li a2, 2 call _memcmp bnez a0, .Lskip_comment_end _advance(2) (* Skip (*. *) li t0, 0x292a # *) sw t0, 4(sp) .Lskip_comment_loop addi a0, sp, 4 mv a1, s1 li a2, 2 call _memcmp beqz a0, .Lskip_comment_close _advance(1) goto .Lskip_comment_loop .Lskip_comment_close _advance(2) (* Skip "*" and ")". *) .Lskip_comment_end end (* Parameters: a0 - Line length. *) proc _compile_assembly(loca84: Word) var loca0: ^Byte begin (* Write the source to the standard output. *) loca0 := _current() _write_out(loca0, loca84) _advance(loca84) _put_char(0xa) (* \n *) _advance(1) (* Skip the new line. *) end proc _compile_program() var loca0: Word begin (* .global _start *) loca0 := 0x6f6c672e (* .glo *) _write_out(@loca0, 4) loca0 := 0x206c6162 (* bal_ *) _write_out(@loca0, 4) loca0 := 0x6174735f (* _sta *) _write_out(@loca0, 4) loca0 := 0x0a7472 (* rt\n *) _write_out(@loca0, 3) _advance(8) (* program\n. *) end proc _compile_constant_section() var loca0: Word begin (* .section .rodata *) loca0 := 0x6365732e (* .sec *) _write_out(@loca0, 4) loca0 := 0x6e6f6974 (* tion *) _write_out(@loca0, 4) loca0 := 0x6f722e20 (* _.ro *) _write_out(@loca0, 4) loca0 := 0x61746164 (* data *) _write_out(@loca0, 4) loca0 := 0x0a (* \n *) _write_out(@loca0, 1) _advance(6) (* const\n. *) .Lcompile_constant_section_item _skip_spaces() lbu a0, (s1) call _is_upper beqz a0, .Lcompile_constant_section_end _compile_constant() goto .Lcompile_constant_section_item .Lcompile_constant_section_end end proc _compile_constant() var loca0, loca4: Word loca8: ^Byte begin loca4 := _read_token() loca8 := _current() (* Save the identifier pointer before advancing it. *) _write_out(loca8, loca4) _advance(loca4) _skip_spaces() _advance(2) (* Skip the assignment sign. *) (* : .long *) loca0 := 0x6c2e203a (* : .l *) _write_out(@loca0, 4) loca0 := 0x20676e6f (* ong_ *) _write_out(@loca0, 4) _skip_spaces() loca4 := _read_token() loca8 := _current() (* Save the literal pointer before advancing it. *) _write_out(loca8, loca4) _advance(loca4) _put_char(0x0a) (* \n *) end proc _compile_variable_section() var loca0: Word begin (* .section .bss *) loca0 := 0x6365732e (* .sec *) _write_out(@loca0, 4) loca0 := 0x6e6f6974 (* tion *) _write_out(@loca0, 4) loca0 := 0x73622e20 (* _.bs *) _write_out(@loca0, 4) loca0 := 0x0a73 (* s\n *) _write_out(@loca0, 2) _advance(4) (* var\n. *) .Lcompile_variable_section_item _skip_spaces() lbu a0, (s1) call _is_lower beqz a0, .Lcompile_variable_section_end _compile_variable() goto .Lcompile_variable_section_item .Lcompile_variable_section_end end proc _compile_variable() var loca28, loca16: ^Byte loca0, loca24, loca20: Word begin (* Save the identifier on the stack since it should emitted multiple times. *) loca24 := _read_token() loca28 := _current() _advance(loca24) _skip_spaces() _advance(1) (* Skip the colon in front of the type. *) _skip_spaces() _advance(1) (* Skip the opening bracket. *) (* Save the array size on the stack since it has to be emitted multiple times. *) loca16 := _read_token() loca20 := _current() _advance(loca16) _skip_spaces() _advance(1) (* Skip the closing bracket. *) _skip_spaces() loca0 := _read_token() _advance(loca0) (* Skip the type. *) (* .type identifier, @object *) loca0 := 0x7079742e (* .typ *) _write_out(@loca0, 4) loca0 := 0x2065 (* e_ *) _write_out(@loca0, 2) _write_out(loca28, loca24) loca0 := 0x6f40202c (* , @o *) _write_out(@loca0, 4) loca0 := 0x63656a62 (* bjec *) _write_out(@loca0, 4) loca0 := 0x0a74 (* t\n *) _write_out(@loca0, 2) (* .size identifier, size *) loca0 := 0x7a69732e (* .siz *) _write_out(@loca0, 4) loca0 := 0x2065 (* e_ *) _write_out(@loca0, 2) _write_out(loca28, loca24) loca0 := 0x202c (* ,_ *) _write_out(@loca0, 2) _write_out(loca20, loca16) _put_char(0x0a) (* \n *) (* identifier: .zero size *) _write_out(loca28, loca24) loca0 := 0x7a2e203a (* : .z *) _write_out(@loca0, 4) loca0 := 0x206f7265 (* ero_ *) _write_out(@loca0, 4) _write_out(loca20, loca16) _put_char(0x0a) (* \n *) end proc _compile_procedure() var loca0, loca4, loca8, loca12, loca16: Word loca20: ^Char begin _advance(5) (* Skip proc_ *) loca16 := _read_token() loca20 := _current() _advance(loca16) (* .type identifier, @function *) loca0 := 0x7079742e (* .typ *) _write_out(@loca0, 4) loca0 := 0x2065 (* e_ *) _write_out(@loca0, 2) _write_out(loca20, loca16) loca0 := 0x6640202c (* , @f *) _write_out(@loca0, 4) loca0 := 0x74636e75 (* unct *) _write_out(@loca0, 4) loca0 := 0x0a6e6f69 (* ion\n *) _write_out(@loca0, 4) _write_out(loca20, loca16) loca0 := 0x0a3a (* :\n *) _write_out(@loca0, 2) _skip_spaces() _advance(1) (* Skip opening argument paren. *) _skip_spaces() _advance(1) (* Skip closing argument paren. *) loca12 := 0x6e (* n *) loca8 := 0x69676562 (* begi *) (* Skip all declarations until we find the "begin" keyword, denoting the beginning of the procedure body. *) .Lcompile_procedure_begin _skip_spaces() call _read_token mv a1, a0 mv a0, s1 addi a2, sp, 8 add s1, s1, a1 call _token_compare bnez a0, .Lcompile_procedure_begin (* Generate the procedure prologue with a predefined stack size. *) loca0 := 0x69646461 (* addi *) _write_out(@loca0, 4) loca0 := 0x2c707320 (* _sp, *) _write_out(@loca0, 4) _write_out(@loca0, 4) loca0 := 0x0a36392d (* -96\n *) _write_out(@loca0, 4) loca0 := 0x72207773 (* sw r *) _write_out(@loca0, 4) loca0 := 0x39202c61 (* a, 9 *) _write_out(@loca0, 4) loca0 := 0x70732832 (* 2(sp *) _write_out(@loca0, 4) loca0 := 0x0a29 (* )\n *) _write_out(@loca0, 2) loca0 := 0x73207773 (* sw s *) _write_out(@loca0, 4) loca0 := 0x38202c30 (* 0, 8 *) _write_out(@loca0, 4) loca0 := 0x70732838 (* 8(sp *) _write_out(@loca0, 4) loca0 := 0x0a29 (* )\n *) _write_out(@loca0, 2) loca0 := 0x69646461 (* addi *) _write_out(@loca0, 4) loca0 := 0x2c307320 (* _s0, *) _write_out(@loca0, 4) loca0 := 0x2c707320 (* _sp, *) _write_out(@loca0, 4) loca0 := 0x0a363920 (* _96\n *) _write_out(@loca0, 4) (* Save passed arguments on the stack. *) loca0 := 0x61207773 (* sw a *) _write_out(@loca0, 4) loca4 := 0x38202c30 (* 0, 8 *) _write_out(@loca4, 4) loca8 := 0x70732834 (* 4(sp *) _write_out(@loca8, 4) loca12 := 0x0a29 (* )\n *) _write_out(@loca12, 2) _write_out(@loca0, 4) loca4 := 0x38202c31 (* 1, 8 *) _write_out(@loca4, 4) loca8 := 0x70732830 (* 0(sp *) _write_out(@loca8, 4) _write_out(@loca12, 2) _write_out(@loca0, 4) loca4 := 0x37202c32 (* 2, 7 *) _write_out(@loca4, 4) loca8 := 0x70732836 (* 6(sp *) _write_out(@loca8, 4) _write_out(@loca12, 2) _write_out(@loca0, 4) loca4 := 0x37202c33 (* 3, 7 *) _write_out(@loca4, 4) loca8 := 0x70732832 (* 2(sp *) _write_out(@loca8, 4) _write_out(@loca12, 2) _write_out(@loca0, 4) loca4 := 0x36202c34 (* 4, 6 *) _write_out(@loca4, 4) loca8 := 0x70732838 (* 8(sp *) _write_out(@loca8, 4) _write_out(@loca12, 2) _write_out(@loca0, 4) loca4 := 0x36202c35 (* 5, 6 *) _write_out(@loca4, 4) loca8 := 0x70732838 (* 4(sp *) _write_out(@loca8, 4) _write_out(@loca12, 2) (* Generate the body of the procedure. *) .Lcompile_procedure_body _skip_indentation() call _read_line sw a0, 12(sp) li t0, 0x0a646e65 # end\n sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 4 call _memcmp beqz a0, .Lcompile_procedure_end lw a0, 12(sp) call _compile_line goto .Lcompile_procedure_body .Lcompile_procedure_end _advance(4) (* Skip end\n. *) (* Generate the procedure epilogue with a predefined stack size. *) loca0 := 0x7220776c # lw r _write_out(@loca0, 4) loca0 := 0x39202c61 # a, 9 _write_out(@loca0, 4) loca0 := 0x70732832 # 2(sp _write_out(@loca0, 4) loca0 := 0x0a29 # )\n _write_out(@loca0, 2) loca0 := 0x7320776c (* lw s *) _write_out(@loca0, 4) loca0 := 0x38202c30 (* 0, 8 *) _write_out(@loca0, 4) loca0 := 0x70732838 (* 8(sp *) _write_out(@loca0, 4) loca0 := 0x0a29 (* )\n *) _write_out(@loca0, 2) loca0 := 0x69646461 (* addi *) _write_out(@loca0, 4) loca0 := 0x2c707320 (* _sp, *) _write_out(@loca0, 4) _write_out(@loca0, 4) loca0 := 0x0a3639 (* 96\n *) _write_out(@loca0, 4) loca0 := 0x0a746572 (* ret\n *) _write_out(@loca0, 4) end (* 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. *) proc _token_compare() begin 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 goto .Ltoken_compare_loop .Ltoken_compare_not_equal li a0, 1 goto .Ltoken_compare_end .Ltoken_compare_equal li a0, 0 .Ltoken_compare_end end proc _compile_goto() var loca0: Word loca8: ^Byte begin _advance(4) (* Skip the goto keyword. *) loca0 := 0x206a (* j_ *) _write_out(@loca0, 2) _skip_spaces() loca8 := _current() (* We should be on dot the label is beginning with. *) _advance(1) loca0 := _read_token() _advance(loca0) loca0 := loca0 + 1 (* Label length and the dot. *) _write_out(loca8, loca0) _advance(1) (* Skip the new line. *) _put_char(0x0a) (* \n *) end (* a0 - Line length. *) proc _compile_label(loca84: Word) var loca0: Word begin (* Write the whole line as is. *) loca0 := _current() _write_out(loca0, loca84) lw t0, 84(sp) # Line length. mv t1, s1 # Line start. add t1, t1, t0 addi t1, t1, -1 # Last character on the line. lbu t1, (t1) li t2, ':' beq t1, t2, .Lcompile_label_colon _put_char(0x3a) (* : *) .Lcompile_label_colon _put_char(0x0a) (* \n *) _advance(loca84) end proc _compile_return() begin _advance(6) (* Skip return. *) _skip_spaces() _build_binary_expression() end proc _compile_if() var loca8, loca12, loca16, loca20: Word loca4: ^Byte begin _advance(2) (* Skip the if. *) _skip_spaces() _build_binary_expression() _skip_spaces() _advance(4) (* Skip the then. *) loca20 := 0x00646e65 (* if end marker and newline. *) loca16 := 0x66694c2e (* Label prefix ".Lif". *) loca12 := 0x7a716562 (* beqz *) _write_out(@loca12, 4) loca12 := 0x2c306120 (* _a0, *) _write_out(@loca12, 4) _put_char(0x20) (* _ *) (* Write the label *) _write_out(@loca16, 4) mv a0, s2 call _printi _put_char(0x0a) (* \n *) .Lcompile_if_loop _skip_spaces() loca12 := _read_token() loca4 := _current() loca8 := _token_compare(loca4, loca12, @loca20) if loca8 then loca12 := _read_line() _compile_line(loca12, 1) goto .Lcompile_if_loop end (* Write the label *) _write_out(@loca16, 4) mv a0, s2 call _printi loca12 := 0x0a3a0a3a (* :\n:\n *) _write_out(@loca12, 2) addi s2, s2, 1 # Increment the label counter. _advance(4) (* Skip the end with newline. *) end (* Parameters: a0 - Line length. a1 - Whether the section header was already emitted. If not it should be emitted before any code is written. Returns 1 in a0 if the parsed line contained a text section element such a procedure or the program entry point. Otherwise sets a0 to 0. *) proc _compile_line(loca84: Word, loca80: Bool) begin beqz a0, .Lcompile_line_empty # Skip an empty line. lbu t0, (s1) li t1, '(' beq t0, t1, .Lcompile_line_comment li t0, 0x0a6d6172 # ram\n sw t0, 12(sp) li t0, 0x676f7270 # prog sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 8 call _memcmp beqz a0, .Lcompile_line_program li t0, 0x0a74 # t\n sw t0, 12(sp) li t0, 0x736e6f63 # cons sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 6 call _memcmp beqz a0, .Lcompile_line_const li t0, 0x0a726176 # var\n sw t0, 12(sp) mv a0, s1 addi a1, sp, 12 li a2, 4 call _memcmp beqz a0, .Lcompile_line_var li t0, 0x20 # _ sw t0, 12(sp) li t0, 0x636f7270 # proc sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 5 call _memcmp beqz a0, .Lcompile_line_procedure li t0, 0x0a6e # n\n sw t0, 12(sp) li t0, 0x69676562 # begi sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 6 call _memcmp beqz a0, .Lcompile_line_begin li t0, 0x2e646e65 # end. sw t0, 12(sp) mv a0, s1 addi a1, sp, 12 li a2, 4 call _memcmp beqz a0, .Lcompile_line_exit li t0, 0x61636f6c # loca sw t0, 12(sp) mv a0, s1 addi a1, sp, 12 li a2, 4 call _memcmp beqz a0, .Lcompile_line_identifier li t0, 0x7472 # rt sw t0, 12(sp) li t0, 0x6f706d69 # impo sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 6 call _memcmp beqz a0, .Lcompile_line_import li t0, 0x6f746f67 # goto sw t0, 12(sp) mv a0, s1 addi a1, sp, 12 li a2, 4 call _memcmp beqz a0, .Lcompile_line_goto li t0, 0x6e72 # rn sw t0, 12(sp) li t0, 0x75746572 # retu sw t0, 8(sp) mv a0, s1 addi a1, sp, 8 li a2, 6 call _memcmp beqz a0, .Lcompile_line_return li t0, 0x6669 # if sw t0, 12(sp) mv a0, s1 addi a1, sp, 12 li a2, 2 call _memcmp beqz a0, .Lcompile_line_if lbu t0, (s1) li t1, '.' beq t0, t1, .Lcompile_line_label li t1, '_' beq t0, t1, .Lcompile_line_identifier goto .Lcompile_line_unchanged (* Else. *) .Lcompile_line_if: _compile_if() goto .Lcompile_line_section .Lcompile_line_label _compile_label(loca84) goto .Lcompile_line_section .Lcompile_line_return (* DEBUG mv a0, s1 li a1, 8 call _write_error *) _compile_return() goto .Lcompile_line_section .Lcompile_line_goto _compile_goto() goto .Lcompile_line_section .Lcompile_line_import _compile_import() goto .Lcompile_line_section .Lcompile_line_identifier _compile_identifier() goto .Lcompile_line_section .Lcompile_line_exit _compile_exit() goto .Lcompile_line_section .Lcompile_line_begin lw a1, 80(sp) bnez a1, .Lcompile_line_compile_entry _compile_text_section() .Lcompile_line_compile_entry _compile_entry_point() li a0, 1 goto .Lcompile_line_end .Lcompile_line_const _compile_constant_section() goto .Lcompile_line_section .Lcompile_line_procedure lw a1, 80(sp) bnez a1, .Lcompile_line_compile_procedure _compile_text_section() .Lcompile_line_compile_procedure _compile_procedure() li a0, 1 goto .Lcompile_line_end .Lcompile_line_var _compile_variable_section() goto .Lcompile_line_section .Lcompile_line_program _compile_program() goto .Lcompile_line_section .Lcompile_line_comment _skip_comment(loca84) goto .Lcompile_line_section .Lcompile_line_empty _advance(1) goto .Lcompile_line_section .Lcompile_line_unchanged _compile_assembly(loca84) goto .Lcompile_line_section .Lcompile_line_section mv a0, zero .Lcompile_line_end sw a0, 12(sp) _skip_spaces() _skip_comment() lw a0, 12(sp) end (* Prints ".section .text" and exits. *) proc _compile_text_section() var loca0: Word begin (* .section .text *) loca0 := 0x6365732e (* .sec *) _write_out(@loca0, 4) loca0 := 0x6e6f6974 (* tion *) _write_out(@loca0, 4) loca0 := 0x65742e20 (* _.te *) _write_out(@loca0, 4) loca0 := 0x0a7478 (* xt\n *) _write_out(@loca0, 3) end proc _compile_entry_point() var loca0: Word begin (* .type _start, @function *) loca0 := 0x7079742e (* .typ *) _write_out(@loca0, 4) loca0 := 0x735f2065 (* e _s *) _write_out(@loca0, 4) loca0 := 0x74726174 (* tart *) _write_out(@loca0, 4) loca0 := 0x6640202c (* , @f *) _write_out(@loca0, 4) loca0 := 0x74636e75 (* unct *) _write_out(@loca0, 4) loca0 := 0x0a6e6f69 (* ion\n *) _write_out(@loca0, 4) loca0 := 0x6174735f (* _sta *) _write_out(@loca0, 4) loca0 := 0x0a3a7472 (* rt:\n *) _write_out(@loca0, 4) _advance(6) (* Skip begin\n. *) end proc _compile_exit() var loca0: Word begin (* li a0, 0 li a7, SYS_EXIT ecall *) loca0 := 0x6120696c (* li a *) _write_out(@loca0, 4) loca0 := 0x30202c30 (* 0, 0 *) _write_out(@loca0, 4) loca0 := 0x20696c0a (* \nli_ *) _write_out(@loca0, 4) loca0 := 0x202c3761 (* a7,_ *) _write_out(@loca0, 4) loca0 := 0x650a3339 (* 93\ne *) _write_out(@loca0, 4) loca0 := 0x6c6c6163 (* call *) _write_out(@loca0, 4) loca0 := 0x0a (* \n *) _write_out(@loca0, 1) _advance(4) (* Skip end. *) _skip_spaces() (* Read the possible new line at the end of the file. *) end (* Finds the end of the line and returns its length in a0. *) proc _read_line() begin loca0 := _current() (* Local position in the source text. *) .Lread_line_do lw t0, 0(sp) lbu t1, (t0) # t1 = Current character. beqz t1, .Lread_line_end # Exit the loop on the NUL character. li t2, '\n' beq t1, t2, .Lread_line_end # Exit the loop on the new line. loca0 := loca0 + 1 goto .Lread_line_do .Lread_line_end loca4 := _current() return loca0 - loca4 (* Return the line length. *) end proc _compile() var loca0, loca4: Word loca8: Bool begin loca4 := 0 (* Whether the text section header was already emitted. *) .Lcompile_do lbu t0, (s1) # t0 = Current character. beqz t0, .Lcompile_end # Exit the loop on the NUL character. _skip_indentation() loca0 := _read_line() loca8 := _compile_line(loca0, loca4) beqz a0, .Lcompile_do (* Update whether the text section header was already emitted. *) loca4 := loca4 or loca8 goto .Lcompile_do .Lcompile_end end (* Returns the pointer to the current position in the source text in a0. *) proc _current() begin mv a0, s1 end (* a0 is the number of bytes to advance in the source text. *) proc _advance() begin add s1, s1, a0 end (* Returns the first character in the remaining source text. *) proc _front() begin lbu a0, (s1) end (* Entry point. *) begin (* Read the source from the standard input. *) la a0, source_code la a1, SOURCE_BUFFER_SIZE # Buffer size. lw a1, (a1) call _read_file la s1, source_code # s1 = Source code position. li s2, 1 _compile() end.