From c1a44ed41dcf92d1567ab927d7f1675d6918aff9 Mon Sep 17 00:00:00 2001
From: Freya Murphy <freya@freyacat.org>
Date: Wed, 11 Sep 2024 12:47:27 -0400
Subject: [PATCH] add ascii and asciiz directives, fix symtab bug

---
 include/mips.h | 26 --------------------------
 masm/asm.c     | 20 ++++++++++++++++++++
 masm/asm.h     |  3 +++
 masm/parse.c   | 20 ++++++++++++++++++++
 masm/parse.h   | 30 +++++++++++++++++++++++++++---
 masm/symtab.c  |  2 +-
 masm/test.asm  |  5 +++++
 7 files changed, 76 insertions(+), 30 deletions(-)

diff --git a/include/mips.h b/include/mips.h
index 8d76d12..1233693 100644
--- a/include/mips.h
+++ b/include/mips.h
@@ -475,30 +475,4 @@ MIPS_INS(XORI, I, .op = MIPS_OP_XORI)
 
 #undef MIPS_INS
 
-/* 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,
-};
-
-/* mip32 directive */
-struct mips_directive {
-	enum mips_directive_type type;
-	uint32_t len; // used for words, halfs, bytes
-	union {
-		uint16_t align;
-		uint16_t space;
-		uint32_t words[MAX_ARG_LENGTH];
-		uint16_t halfs[MAX_ARG_LENGTH];
-		uint8_t bytes[MAX_ARG_LENGTH];
-		char name[MAX_ARG_LENGTH];
-	};
-};
-
 #endif /* __MIPS_H__ */
diff --git a/masm/asm.c b/masm/asm.c
index 54d766b..a9cee99 100644
--- a/masm/asm.c
+++ b/masm/asm.c
@@ -165,6 +165,26 @@ static int handle_directive(struct assembler *assembler,
 
 		break;
 	}
+
+        case MIPS_DIRECTIVE_ASCII: {
+		struct section_entry entry;
+		entry.type = ENT_STR;
+		entry.size = strlen(directive->name);
+		memcpy(entry.str, directive->name, entry.size);
+		if (sec_push(assembler->sectab.current, entry))
+			return M_ERROR;
+		break;
+	}
+
+        case MIPS_DIRECTIVE_ASCIIZ: {
+		struct section_entry entry;
+		entry.type = ENT_STR;
+		entry.size = strlen(directive->name) + 1;
+		memcpy(entry.str, directive->name, entry.size);
+		if (sec_push(assembler->sectab.current, entry))
+			return M_ERROR;
+		break;
+	}
         }
 
         return M_SUCCESS;
diff --git a/masm/asm.h b/masm/asm.h
index c8a6394..267aa1c 100644
--- a/masm/asm.h
+++ b/masm/asm.h
@@ -7,6 +7,7 @@
 #include <elf.h>
 #include <mips.h>
 
+#include "mlimits.h"
 #include "parse.h"
 #include "lex.h"
 
@@ -102,6 +103,7 @@ enum section_entry_type {
 	ENT_WORD,
 	ENT_HALF,
 	ENT_BYTE,
+	ENT_STR,
 	ENT_NO_DATA,
 };
 
@@ -117,6 +119,7 @@ struct section_entry {
 
 		// data
 		struct mips_instruction ins;
+		char str[MAX_LEX_LENGTH];
 		int32_t word;
 		int16_t half;
 		int8_t byte;
diff --git a/masm/parse.c b/masm/parse.c
index ccabf41..2a84a91 100644
--- a/masm/parse.c
+++ b/masm/parse.c
@@ -875,6 +875,20 @@ static int parse_directive_globl(struct parser *parser,
 	return M_SUCCESS;
 }
 
+static int parse_directive_ascii(struct parser *parser,
+				 struct mips_directive *directive,
+				 enum mips_directive_type type)
+{
+	struct token token;
+	if (assert_token(parser, TOK_STRING, &token))
+		return M_ERROR;
+
+	directive->type = type;
+	strcpy(directive->name, token.text);
+
+	return M_SUCCESS;
+}
+
 static int parse_section(struct mips_directive *directive,
 			 char name[MAX_LEX_LENGTH])
 {
@@ -909,6 +923,12 @@ static int parse_directive(struct parser *parser,
 		return parse_directive_extern(parser, directive);
 	else if (strcmp(token.text, "globl") == 0)
 		return parse_directive_globl(parser, directive);
+	else if (strcmp(token.text, "ascii") == 0)
+		return parse_directive_ascii(parser, directive,
+			       MIPS_DIRECTIVE_ASCII);
+	else if (strcmp(token.text, "asciiz") == 0)
+		return parse_directive_ascii(parser, directive,
+			       MIPS_DIRECTIVE_ASCIIZ);
 	else
 		return parse_section(directive, token.text);
 }
diff --git a/masm/parse.h b/masm/parse.h
index 9181899..3052d51 100644
--- a/masm/parse.h
+++ b/masm/parse.h
@@ -9,9 +9,33 @@
 #include <mips.h>
 #include <stdint.h>
 
-///
-/// reference
-///
+/* 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,
+};
+
+/* mip32 directive */
+struct mips_directive {
+	enum mips_directive_type type;
+	uint32_t len; // used for words, halfs, bytes
+	union {
+		uint16_t align;
+		uint16_t space;
+		uint32_t words[MAX_ARG_LENGTH];
+		uint16_t halfs[MAX_ARG_LENGTH];
+		uint8_t bytes[MAX_ARG_LENGTH];
+		char name[MAX_ARG_LENGTH];
+	};
+};
 
 enum reference_type {
 	REF_NONE,
diff --git a/masm/symtab.c b/masm/symtab.c
index 7d40609..897c4f9 100644
--- a/masm/symtab.c
+++ b/masm/symtab.c
@@ -61,7 +61,7 @@ int symtab_find(struct symbol_table *symtab, Elf32_Sym **ptr,
 
                         ptrdiff_t diff = sym - symtab->symbols;
                         if (idx != NULL)
-                                *idx = diff / sizeof(Elf32_Sym);
+                                *idx = diff;
 
                         return M_SUCCESS;
                 }
diff --git a/masm/test.asm b/masm/test.asm
index 1820b62..3dea7c9 100644
--- a/masm/test.asm
+++ b/masm/test.asm
@@ -1,3 +1,8 @@
+.data
+
+str:
+.ascii "hello world"
+
 .text
 .align 2