refactor masm to add codegen step

1 files changed, 99 insertions, 41 deletions

diff --git a/masm/parse.h b/masm/parse.h
index 9e0e928..61036cd 100644
--- a/masm/parse.h
+++ b/masm/parse.h
@@ -6,89 +6,147 @@
 #include "lex.h"
 
 #include <mlimits.h>
-#include <mips.h>
 #include <stdint.h>
 
-/* mips directive types */
-enum mips_directive_type {
-	MIPS_DIRECTIVE_ALIGN,
-	MIPS_DIRECTIVE_SPACE,
-	MIPS_DIRECTIVE_WORD,
-	MIPS_DIRECTIVE_HALF,
-	MIPS_DIRECTIVE_BYTE,
-	MIPS_DIRECTIVE_SECTION,
-	MIPS_DIRECTIVE_EXTERN,
-	MIPS_DIRECTIVE_GLOBL,
-	MIPS_DIRECTIVE_ASCII,
-	MIPS_DIRECTIVE_ASCIIZ,
+/// the type to a direcive
+enum expr_directive_type {
+	EXPR_DIRECTIVE_ALIGN,
+	EXPR_DIRECTIVE_SPACE,
+	EXPR_DIRECTIVE_WORD,
+	EXPR_DIRECTIVE_HALF,
+	EXPR_DIRECTIVE_BYTE,
+	EXPR_DIRECTIVE_SECTION,
+	EXPR_DIRECTIVE_EXTERN,
+	EXPR_DIRECTIVE_GLOBL,
+	EXPR_DIRECTIVE_ASCII,
+	EXPR_DIRECTIVE_ASCIIZ,
 };
 
-/* mip32 directive */
-struct mips_directive {
-	enum mips_directive_type type;
-	uint32_t len; // used for words, halfs, bytes
+/// holds a directive
+struct expr_directive {
+	// the type of the directive
+	enum expr_directive_type type;
+
+	// lengh of .word, .half, or .byte directive
+	uint32_t len;
+
+	// directive data
 	union {
+		// e.g. align 2
 		uint16_t align;
+		// e.g. space 4096
 		uint16_t space;
+		// e.g. .word 0x1 0x2
 		uint32_t words[MAX_ARG_LENGTH];
 		uint16_t halfs[MAX_ARG_LENGTH];
-		uint8_t bytes[MAX_ARG_LENGTH];
-		char name[MAX_ARG_LENGTH];
+		uint8_t  bytes[MAX_ARG_LENGTH];
+		// e.g. .ascii "hello world!"
+		struct string string;
+		// e.g. .globl main
+		struct string label;
+		// e.g. .text
+		struct string section;
 	};
 };
 
-struct reference {
-	// ELF relocate type
-	unsigned char type;
+/// holds a constant expression
+struct expr_const {
+	// the name of the constant
+	struct string name;
+
+	// the value of the constant
+	uint32_t num;
+};
+
+/// the type to a right
+/// hand side argument to an
+/// instruction
+enum expr_ins_arg_type {
+	// e.g. $ra
+	EXPR_INS_ARG_REGISTER,
+
+	// e.g. 0x80
+	EXPR_INS_ARG_IMMEDIATE,
 
-	/// symbol name
-	char name[MAX_LEX_LENGTH];
+	// e.g. main
+	EXPR_INS_ARG_LABEL,
 
-	/// integer addend
-	int64_t addend;
+	// e.g. 4($sp)
+	EXPR_INS_ARG_OFFSET,
 };
 
-struct const_expr {
-	char name[MAX_LEX_LENGTH];
-	uint32_t value;
+/// a right hand argument
+/// to an instruction
+struct expr_ins_arg {
+	enum expr_ins_arg_type type;
+
+	union {
+		// register
+		struct string reg;
+
+		// immediate
+		uint64_t immd;
+
+		// label
+		struct string label;
+
+		// offset
+		struct expr_ins_offset {
+			// immediate
+			uint64_t immd;
+			// register
+			struct string reg;
+		} offset;
+	};
 };
 
-struct ins_expr {
+/// holds a instruction
+struct expr_ins {
 	/// pesudo instructions can return
 	/// more than one instruction
-	size_t ins_len;
-	struct mips_instruction ins[2];
+	struct string name;
 
-	/// instructions can reference symbols.
-	/// instruction `n` will be paried with reference `n`
-	struct reference ref[2];
+	// the arguments of the instruction
+	uint32_t args_len;
+	struct expr_ins_arg args[MAX_ARG_LENGTH];
 };
 
 enum expr_type {
+	// e.g. .align 2
 	EXPR_DIRECTIVE,
+	// e.g. SIZE = 8
 	EXPR_CONSTANT,
+	// e.g. li $t0, 17
 	EXPR_INS,
+	// e.g. _start:
 	EXPR_LABEL,
 };
 
 struct expr {
 	enum expr_type type;
+
+	uint32_t line_no;
+	uint32_t byte_start;
+	uint32_t byte_end;
+
 	union {
 		// directive
-		struct mips_directive directive;
+		struct expr_directive directive;
 		// constant
-		struct const_expr constant;
+		struct expr_const constant;
 		// instruction
-		struct ins_expr ins;
+		struct expr_ins instruction;
 		// label
-		char label[MAX_LEX_LENGTH];
+		struct string label;
 	};
 };
 
+void expr_free(struct expr *expr);
+
 struct parser {
 	// the lexer
 	// *weak* ponter, we do not own this
-	struct lexer *lexer;
+	struct lexer lexer;
 	// the last token peeked
 	struct token peek;
 };
@@ -97,7 +155,7 @@ struct parser {
 int parser_next(struct parser *parser, struct expr *expr);
 
 /* initalize the base parser */
-int parser_init(struct lexer *lexer, struct parser *parser);
+int parser_init(const char *file, struct parser *parser);
 
 /* free the base parser */
 void parser_free(struct parser *parser);