initial mips32 (r2000ish mips32r6) assembler

1 files changed, 156 insertions, 0 deletions

diff --git a/masm/parse.h b/masm/parse.h
new file mode 100644
index 0000000..2aea0be
--- /dev/null
+++ b/masm/parse.h
@@ -0,0 +1,156 @@
+/* Copyright (c) 2024 Freya Murphy */
+
+#ifndef __PARSE_H__
+#define __PARSE_H__
+
+#include "lex.h"
+
+#include <mlimits.h>
+#include <mips.h>
+#include <stdint.h>
+
+struct const_expr {
+	char name[MAX_LEX_LENGTH];
+	uint32_t value;
+};
+
+enum expr_type {
+	EXPR_INS,
+	EXPR_DIRECTIVE,
+	EXPR_CONSTANT,
+	EXPR_SEGMENT,
+	EXPR_LABEL,
+};
+
+struct expr {
+	enum expr_type type;
+	union {
+		// instruction
+		union mips_instruction ins;
+		// directive
+		union mips_directive directive;
+		// constant
+		struct const_expr constant;
+		// segment or label
+		char text[MAX_LEX_LENGTH];
+	};
+};
+
+enum symbol_flag {
+	SYM_LOCAL,
+	SYM_GLOBAL,
+	SYM_EXTERNAL,
+};
+
+struct symbol {
+	char name[MAX_LEX_LENGTH];
+	uint32_t position;
+	enum symbol_flag flag;
+
+};
+
+struct symbol_table {
+	uint32_t count;
+	uint32_t len;
+	struct symbol *symbols;
+};
+
+int symtbl_init(struct symbol_table *sym_tbl);
+void symtbl_free(struct symbol_table *sym_tbl);
+
+int symtbl_push(struct symbol_table *sym_tbl, struct symbol sym);
+int symtbl_find(struct symbol_table *sym_tbl, struct symbol **sym,
+		const char name[MAX_LEX_LENGTH]);
+
+struct section {
+	uint32_t count;
+	uint32_t len;
+	uint32_t start;
+	uint32_t alignment;
+	union mips_instruction *ins;
+	char name[MAX_LEX_LENGTH];
+};
+
+struct section_table {
+	uint32_t count;
+	uint32_t len;
+	struct section *sections;
+	struct section *current;
+	char name[MAX_LEX_LENGTH];
+	uint32_t total_ins;
+};
+
+int sectbl_init(struct section_table *sec_tbl);
+void sectbl_free(struct section_table *sec_tbl);
+
+int sectbl_alloc(struct section_table *sec_tbl, struct section **sec,
+		 const char name[MAX_LEX_LENGTH]);
+int sectbl_push(struct section_table *sec_tbl, struct section *section,
+		union mips_instruction ins);
+int sectbl_get(struct section_table *sec_tbl, struct section **sec,
+	       const char name[MAX_LEX_LENGTH]);
+
+enum reference_type {
+	REF_OFFESET,
+	REF_TARGET,
+};
+
+struct reference {
+	enum reference_type type;
+	struct section *section;
+	uint32_t index;
+	char name[MAX_LEX_LENGTH];
+};
+
+struct reference_table {
+	uint32_t count;
+	uint32_t len;
+	struct reference *references;
+};
+
+int reftbl_init(struct reference_table *ref_tbl);
+void reftbl_free(struct reference_table *ref_tbl);
+int reftbl_push(struct reference_table *ref_tbl, struct reference reference);
+
+struct parser {
+	struct lexer *lexer;
+	struct token peek;
+
+	// sections
+	struct section_table sec_tbl;
+
+	// symbols
+	struct symbol_table sym_tbl;
+
+	// references
+	struct reference_table ref_tbl;
+
+	int (*parse_instruction)(struct parser *, union mips_instruction *,
+			  struct token);
+	int (*parse_directive)(struct parser *, union mips_directive *);
+	int (*is_instruction)(const char *ident);
+};
+
+/* get the next token in the parser */
+int next_token(struct parser *parser, struct token *tok);
+
+/* peek the next token in the parser */
+int peek_token(struct parser *parser, struct token *tok);
+
+/* assert the next token is a specific type */
+int assert_token(struct parser *parser, enum token_type type,
+		 struct token *tok);
+
+/* assert the next token is EOF or NL */
+int assert_eol(struct parser *parser);
+
+/* get the next expression in the parser */
+int parser_next(struct parser *parser, struct expr *expr);
+
+/* initalize the base parser */
+int parser_init(struct lexer *lexer, struct parser *parser);
+
+/* free the base parser */
+void parser_free(struct parser *parser);
+
+#endif /* __PARSE_H__ */