Implement simple binary expressions

Rakefile

@@ -63,3 +63,35 @@ end
 file 'build/stage2b' => ['build/stage2b.s', 'boot/common-boot.s'] do |t|
   sh CROSS_GCC, '-nostdlib', '-o', t.name, *t.prerequisites
 end
+
+desc 'Print remaining lines to rewrite'
+task :statistics do
+  def is_false_positive(word)
+    word.start_with?('(*') ||
+      word.start_with?('*)') ||
+      ('A'..'Z').include?(word[0]) ||
+      /^[[:alpha:]][[:digit:]]$/.match(word)
+  end
+
+  lines = File.read('boot/stage2.elna')
+    .split("\n")
+    .select { |line| line.start_with? "\t" }
+    .map { |line| line.delete_prefix("\t").split(' ') }
+    .reject { |words| is_false_positive(words.first) }
+    .group_by do |words|
+      if words.first.length < 5
+        case words.first
+        when 'goto'
+          'Statements'
+        else
+          words.first
+        end
+      else
+        'Statements'
+      end
+    end
+
+  lines.each do |key, value|
+    puts "#{key}: #{value.count}"
+  end
+end

boot/common-boot.s

@@ -1,4 +1,5 @@
-.global _is_alpha, _is_digit, _is_alnum, _is_upper, _is_lower, _write_out, _read_file, exit, _memcmp, _write_error
+.global _is_alpha, _is_digit, _is_alnum, _is_upper, _is_lower
+.global _write_out, _read_file, exit, _memcmp, _write_error, _put_char
 
 .section .rodata
 
@@ -200,6 +201,29 @@ _read_file:
 	ret
 
 # Terminates the program. a0 contains the return code.
+.type _exit, @function
 exit:
 	li a7, SYS_EXIT
 	ecall
+
+# Writes a character from a0 into the standard output.
+.type _put_char, @function
+_put_char:
+	# Prologue
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	sb a0, 4(sp)
+	li a0, STDOUT
+	addi a1, sp, 4
+	li a2, 1
+	li a7, SYS_WRITE
+	ecall
+
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	add sp, sp, 16
+	ret

boot/stage1.s

@@ -16,6 +16,14 @@ prologue: .ascii "addi sp, sp, -96\nsw ra, 92(sp)\nsw s0, 88(sp)\naddi s0, sp, 9
 .equ PROLOGUE_SIZE, . - prologue
 epilogue: .ascii "lw ra, 92(sp)\nlw s0, 88(sp)\naddi sp, sp, 96\nret\n"
 .equ EPILOGUE_SIZE, . - epilogue
+asm_exit: .ascii "li a0, 0\nli a7, 93\necall\n"
+.equ ASM_EXIT_SIZE, . - asm_exit
+asm_start: .ascii ".type _start, @function\n_start:\n"
+.equ ASM_START_SIZE, . - asm_start
+asm_and_a0_a1: .ascii "and a0, a0, a1\n"
+.equ ASM_AND_A0_A1_SIZE, . - asm_and_a0_a1
+asm_or_a0_a1: .ascii "or a0, a0, a1\n"
+.equ ASM_OR_A0_A1_SIZE, . - asm_or_a0_a1
 
 .section .bss
 .type source_code, @object
@@ -67,6 +75,14 @@ _build_binary_expression:
 	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
@@ -75,11 +91,6 @@ _build_binary_expression:
 	call _token_compare
 	beqz a0, .L_build_binary_expression_equal
 
-	/* DEBUG
-	mv a0, s1
-	li a1, 8
-	call _write_error */
-
 	j .Lbuild_binary_expression_end
 
 .L_build_binary_expression_equal:
@@ -114,16 +125,18 @@ _build_binary_expression:
 	addi s1, s1, 1 # Skip &.
 	li a0, 1
 	call _build_expression
-	li t0, 0x0a3161 # a1\n
+	la a0, asm_and_a0_a1
-	sw t0, 16(sp)
+	li a1, ASM_AND_A0_A1_SIZE
-	li t0, 0x202c3061 # a0,_
+	call _write_out
-	sw t0, 12(sp)
+
-	li t0, 0x202c3061 # a0,_
+	j .Lbuild_binary_expression_end
-	sw t0, 8(sp)
+
-	li t0, 0x20646e61 # and_
+.L_build_binary_expression_or:
-	sw t0, 4(sp)
+	addi s1, s1, 2 # Skip or.
-	addi a0, sp, 4
+	li a0, 1
-	li a1, 15
+	call _build_expression
+	la a0, asm_or_a0_a1
+	li a1, ASM_OR_A0_A1_SIZE
 	call _write_out
 
 	j .Lbuild_binary_expression_end
@@ -141,20 +154,24 @@ _build_binary_expression:
 .type _build_expression, @function
 _build_expression:
 	# Prologue.
-	addi sp, sp, -32
+	addi sp, sp, -40
-	sw ra, 28(sp)
+	sw ra, 36(sp)
-	sw s0, 24(sp)
+	sw s0, 32(sp)
-	addi s0, sp, 32
+	addi s0, sp, 40
 
 	addi a0, a0, '0' # Make the register number to a character.
-	sw a0, 20(sp) # And save it.
+	sw a0, 28(sp) # And save it.
 
 	call _skip_spaces
 	call _read_token
-	sw s1, 16(sp)
+	sw s1, 24(sp)
-	sw a0, 12(sp)
+	sw a0, 20(sp)
+
+	addi a0, s1, 0
+	lbu a0, (a0)
+	li t0, '@'
+	beq a0, t0, .Lbuild_expression_address
 
-	# Integer literal.
 	addi a0, s1, 0
 	lbu a0, (a0)
 	call _is_digit
@@ -162,7 +179,7 @@ _build_expression:
 
 	addi a0, s1, 0
 	lbu a0, (a0)
-	li t0, 0x5f # _
+	li t0, '_'
 	beq a0, t0, .Lbuild_expression_call
 
 	li t0, 0x61636f6c # loca
@@ -174,7 +191,7 @@ _build_expression:
 	beqz a0, .Lbuild_expression_identifier
 
 	# Named identifier.
-	lw t1, 20(sp)
+	lw t1, 28(sp)
 	li t0, 0x00202c00 # \0,_
 	or t0, t0, t1
 	sw t0, 8(sp)
@@ -184,48 +201,80 @@ _build_expression:
 	li a1, 7
 	call _write_out
 
-	lw a0, 16(sp)
+	lw a0, 24(sp)
-	lw a1, 12(sp)
+	lw a1, 20(sp)
+	call _write_out
+
+	li a0, '\n'
+	call _put_char
+
+	j .Lbuild_expression_advance
+
+.Lbuild_expression_address:
+	lw t1, 28(sp)
+	li t0, 0x20 # _
+	sw t0, 16(sp)
+	li t0, 0x2c707320 # _sp,
+	sw t0, 12(sp)
+	li t0, 0x2c006120 # _a\0,
+	sw t0, 8(sp)
+	sb t1, 10(sp)
+	li t0, 0x69646461 # addi
+	sw t0, 4(sp)
+	addi a0, sp, 4
+	li a1, 13
+	call _write_out
+
+	addi s1, s1, 1 # Skip @.
+	call _skip_spaces
+	call _read_token
+	sw s1, 24(sp)
+	sw a0, 20(sp)
+
+	lw a0, 24(sp)
+	lw a1, 20(sp)
+	addi a0, a0, 4 # Skip the "loca" variable prefix.
+	addi a1, a1, -4 # Skip the "loca" variable prefix.
 	call _write_out
 
 	li t0, 0x0a # \n
-	sw t0, 8(sp)
+	sw t0, 16(sp)
-	addi a0, sp, 8
+	addi a0, sp, 16
 	li a1, 1
 	call _write_out
 
 	j .Lbuild_expression_advance
 
 .Lbuild_expression_identifier:
-	lw t1, 20(sp)
+	lw t1, 28(sp)
 	li t0, 0x00202c00 # \0,_
 	or t0, t0, t1
-	sw t0, 8(sp)
+	sw t0, 16(sp)
 	li t0, 0x6120776c # lw a
-	sw t0, 4(sp)
+	sw t0, 12(sp)
-	addi a0, sp, 4
+	addi a0, sp, 12
 	li a1, 7
 	call _write_out
 
-	lw a0, 16(sp)
+	lw a0, 24(sp)
-	lw a1, 12(sp)
+	lw a1, 20(sp)
 	addi a0, a0, 4 # Skip the "loca" variable prefix.
 	addi a1, a1, -4 # Skip the "loca" variable prefix.
 	call _write_out
 
 	li t0, 0x0a # \n
-	sw t0, 8(sp)
+	sw t0, 16(sp)
 	li t0, 0x29707328 # (sp)
-	sw t0, 4(sp)
+	sw t0, 12(sp)
-	addi a0, sp, 4
+	addi a0, sp, 12
 	li a1, 5
 	call _write_out
 
 	j .Lbuild_expression_advance
 
 .Lbuild_expression_call:
-	lw a0, 16(sp)
+	lw a0, 24(sp)
-	lw a1, 12(sp)
+	lw a1, 20(sp)
 	add s1, s1, a1
 	addi s1, s1, 1
 	call _compile_call
@@ -233,37 +282,34 @@ _build_expression:
 	j .Lbuild_expression_end
 
 .Lbuild_expression_literal:
-	lw t1, 20(sp)
+	lw t1, 28(sp)
 	li t0, 0x00202c00 # \0,_
 	or t0, t0, t1
-	sw t0, 8(sp)
+	sw t0, 16(sp)
 	li t0, 0x6120696c # li a
-	sw t0, 4(sp)
+	sw t0, 12(sp)
-	addi a0, sp, 4
+	addi a0, sp, 12
 	li a1, 7
 	call _write_out
 
-	lw a0, 16(sp)
+	lw a0, 24(sp)
-	lw a1, 12(sp)
+	lw a1, 20(sp)
 	call _write_out
 
-	li t0, 0x0a # \n
+	li a0, '\n'
-	sw t0, 8(sp)
+	call _put_char
-	addi a0, sp, 8
-	li a1, 1
-	call _write_out
 
 	j .Lbuild_expression_advance
 
 .Lbuild_expression_advance:
-	lw a0, 12(sp)
+	lw a0, 20(sp)
 	add s1, s1, a0
 
 .Lbuild_expression_end:
 	# Epilogue.
-	lw ra, 28(sp)
+	lw ra, 36(sp)
-	lw s0, 24(sp)
+	lw s0, 32(sp)
-	addi sp, sp, 32
+	addi sp, sp, 40
 	ret
 
 # Compiles a statement beginning with an identifier.
@@ -554,6 +600,9 @@ _read_token:
 
 	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)
@@ -638,8 +687,50 @@ _skip_indentation:
 # a0 - Line length.
 .type _skip_comment, @function
 _skip_comment:
-	add s1, s1, a0
+	# Prologue.
-	addi s1, s1, 1 # Skip the new line.
+	addi sp, sp, -16
+	sw ra, 12(sp)
+	sw s0, 8(sp)
+	addi s0, sp, 16
+
+	/* DEBUG
+	mv a0, s1
+	li a1, 8
+	call _write_error */
+
+	# 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
+
+	addi s1, s1, 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
+
+	addi s1, s1, 1
+
+	j .Lskip_comment_loop
+
+.Lskip_comment_close:
+	addi s1, s1, 2 # Skip *).
+
+.Lskip_comment_end:
+	# Epilogue.
+	lw ra, 12(sp)
+	lw s0, 8(sp)
+	addi sp, sp, 16
 	ret
 
 # Parameters:
@@ -662,11 +753,8 @@ _compile_assembly:
 	lw t0, 4(sp)
 	add s1, s1, t0
 
-	li t0, '\n'
+	li a0, '\n'
-	sb t0, 0(sp)
+	call _put_char
-	addi a0, sp, 0
-	li a1, 1
-	call _write_out
 
 	addi s1, s1, 1 # Skip the new line.
 
@@ -742,7 +830,6 @@ _compile_constant:
 	call _write_out
 
 	call _skip_spaces
-	call _read_token
 	addi s1, s1, 2 # Skip the assignment sign.
 
 	# : .long
@@ -762,11 +849,8 @@ _compile_constant:
 	add s1, s1, a1
 	call _write_out
 
-	li t0, '\n'
+	li a0, '\n'
-	sw t0, 4(sp)
+	call _put_char
-	addi a0, sp, 4
-	li a1, 1
-	call _write_out
 
 	# Epilogue.
 	lw ra, 12(sp)
@@ -1234,7 +1318,7 @@ _compile_line:
 	beqz a0, .Lcompile_line_empty # Skip an empty line.
 
 	lbu t0, (s1)
-	li t1, '#'
+	li t1, '('
 	beq t0, t1, .Lcompile_line_comment
 
 	li t0, 0x0a6d6172 # ram\n
@@ -1396,6 +1480,11 @@ _compile_line:
 	mv a0, zero
 
 .Lcompile_line_end:
+	sw a0, 12(sp)
+	call _skip_spaces
+	call _skip_comment
+	lw a0, 12(sp)
+
 	# Epilogue.
 	lw ra, 28(sp)
 	lw s0, 24(sp)
@@ -1406,94 +1495,61 @@ _compile_line:
 .type _compile_text_section, @function
 _compile_text_section:
 	# Prologue.
-	addi sp, sp, -24
+	addi sp, sp, -8
-	sw ra, 20(sp)
+	sw ra, 4(sp)
-	sw s0, 16(sp)
+	sw s0, 0(sp)
-	addi s0, sp, 24
+	addi s0, sp, 8
 
 	la a0, section_text
 	li a1, SECTION_TEXT_SIZE
 	call _write_out
 
 	# Epilogue.
-	lw ra, 20(sp)
+	lw ra, 4(sp)
-	lw s0, 16(sp)
+	lw s0, 0(sp)
-	addi sp, sp, 24
+	addi sp, sp, 8
 	ret
 
 .type _compile_entry_point, @function
 _compile_entry_point:
 	# Prologue.
-	addi sp, sp, -64
+	addi sp, sp, -8
-	sw ra, 60(sp)
+	sw ra, 4(sp)
-	sw s0, 56(sp)
+	sw s0, 0(sp)
-	addi s0, sp, 64
+	addi s0, sp, 8
 
 	# .type _start, @function
-	li t0, 0x0a3a7472 # rt:\n
+	la a0, asm_start
-	sw t0, 52(sp)
+	li a1, ASM_START_SIZE
-	li t0, 0x6174735f # _sta
-	sw t0, 48(sp)
-	li t0, 0x0a6e6f69 # ion\n
-	sw t0, 44(sp)
-	li t0, 0x74636e75 # unct
-	sw t0, 40(sp)
-	li t0, 0x6640202c # , @f
-	sw t0, 36(sp)
-	li t0, 0x74726174 # tart
-	sw t0, 32(sp)
-	li t0, 0x735f2065 # e _s
-	sw t0, 28(sp)
-	li t0, 0x7079742e # .typ
-	sw t0, 24(sp)
-	addi a0, sp, 24
-	li a1, 32
 	call _write_out
 
 	addi s1, s1, 6 # Skip begin\n.
 
 	# Epilogue.
-	lw ra, 60(sp)
+	lw ra, 4(sp)
-	lw s0, 56(sp)
+	lw s0, 0(sp)
-	addi sp, sp, 64
+	addi sp, sp, 8
 	ret
 
 .type _compile_exit, @function
 _compile_exit:
 	# Prologue.
-	addi sp, sp, -64
+	addi sp, sp, -8
-	sw ra, 60(sp)
+	sw ra, 4(sp)
-	sw s0, 56(sp)
+	sw s0, 0(sp)
-	addi s0, sp, 64
+	addi s0, sp, 8
 
-	# li a0, 0
+	la a0, asm_exit
-	# li a7, SYS_EXIT
+	li a1, ASM_EXIT_SIZE
-	# ecall
-	li t0, 0x0a # \n
-	sw t0, 52(sp)
-	li t0, 0x6c6c6163 # call
-	sw t0, 48(sp)
-	li t0, 0x650a3339 # 93\ne
-	sw t0, 44(sp)
-	li t0, 0x202c3761 # a7,_
-	sw t0, 40(sp)
-	li t0, 0x20696c0a # \nli_
-	sw t0, 36(sp)
-	li t0, 0x30202c30 # 0, 0
-	sw t0, 32(sp)
-	li t0, 0x6120696c # li a
-	sw t0, 28(sp)
-	addi a0, sp, 28
-	li a1, 25
 	call _write_out
 
 	addi s1, s1, 4 # Skip end.
 	call _skip_spaces # Read the possible new line at the end of the file.
 
 	# Epilogue.
-	lw ra, 60(sp)
+	lw ra, 4(sp)
-	lw s0, 56(sp)
+	lw s0, 0(sp)
-	addi sp, sp, 64
+	addi sp, sp, 8
 	ret
 
 # Finds the end of the line and returns its length in a0.