Parse header and first instruction

.clang-format

@@ -0,0 +1 @@
+UseTab: Always

.editorconfig

@@ -0,0 +1,9 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+trim_trailing_whitespace = true
+insert_final_newline = true
+indent_style = tab
+indent_size = tab

Makefile

@@ -1,9 +1,10 @@
 NAME = dasm
 CFLAGS = -Wall -Wextra -Wshadow
-LDFLAGS = 
+LDFLAGS =
 
 SRC = src/main.c \
-	  src/dasm.c
+	  src/dasm.c \
+	  src/instructions.c
 
 OBJ = $(SRC:%.c=%.o)
 

src/dasm.c

@@ -2,19 +2,51 @@
 #include <stdio.h>
 #include "dasm.h"
 
-void print_instruction(unsigned addr)
+void print_instruction(unsigned addr, instruction_t inst, unsigned raw)
 {
-	printf("%X\n", addr);
+	printf("%04x: %-14x %s\n", addr, raw, inst.name);
 }
 
-int dasm(char *binary)
+unsigned read_size(u_int8_t *binary, unsigned size)
+{
+	unsigned ret = binary[0];
+
+	for (unsigned i = 1; i < size; i++) {
+		ret |= binary[i] << (8 * i);
+	}
+	return ret;
+}
+
+instruction_t parse_inst(u_int8_t *binary)
+{
+	for (int i = 0; instructions[i].name; i++) {
+		if (instructions[i].opcode == *binary)
+			return instructions[i];
+	}
+	return (const instruction_t){
+		.opcode = 0xFF, .name = "unknown", .mode = {END}, .size = 1
+	};
+}
+
+int dasm(u_int8_t *binary, unsigned long size)
 {
 	int pc = 0;
+	int header_size = 0;
 
-	while (*binary) {
-		instruction_t inst = *binary;
-		print_instruction(pc);
+	if (binary[0] == 0xEB && binary[1] == 0x0E)
+		header_size = 16;
+	else if (binary[0] == 0x01 && binary[1] == 0x03)
+		header_size = binary[4];
+
+	printf("Header size: %d\n", header_size);
+	binary += header_size;
+
+	while (pc < size - header_size) {
+		instruction_t inst = parse_inst(binary);
+		unsigned long raw = read_size(binary, inst.size);
+		print_instruction(pc, inst, raw);
 		pc += inst.size;
+		binary += inst.size;
 	}
 	return 0;
 }

src/dasm.h

@@ -1,6 +1,38 @@
-int dasm(char *binary);
+#include <sys/types.h>
+
+int dasm(u_int8_t *binary, unsigned long size);
+
+// typedef enum addressing_mode {
+// 	/// Param is the data (MOV AL, ~68FE~)
+// 	IMMEDIATE,
+// 	REGISTER,
+// 	DIRECT,
+// 	REGISTER_INDIRECT,
+// 	BASED,
+// 	INDEXED,
+// 	BASED_INDEX,
+// 	BASED_INDEX_DISPLACEMENT,
+// } addressing_mode_t;
+
+typedef enum addressing_mode {
+	// Register or 8bit of memory
+	R_M8,
+	// 8bit register
+	REG8,
+	// 8bit immediate data
+	IMM8,
+	// 16bit immediate data
+	IMM16,
+	// Special item just to mark the end
+	END,
+} addressing_mode_t;
 
 typedef struct instruction {
+	u_int8_t opcode;
 	const char *name;
 	unsigned size;
+	addressing_mode_t mode[5];
 } instruction_t;
+
+extern const instruction_t instructions[];
+

src/instructions.c

@@ -0,0 +1,24 @@
+#include <stddef.h>
+#include "dasm.h"
+
+const instruction_t instructions[] = {
+	{.opcode = 0x00, .name = "add", .mode = {R_M8, REG8, END}, .size = 2},
+	{.opcode = 0xBB, .name = "mov bf", .mode = {R_M8, REG8, END}, .size = 2},
+	{.opcode = 0xB0, .name = "mov al", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB1, .name = "mov cl", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB2, .name = "mov dl", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB3, .name = "mov bl", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB4, .name = "mov ah", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB5, .name = "mov ch", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB6, .name = "mov dh", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB7, .name = "mov bh", .mode = {IMM8}, .size = 2},
+	{.opcode = 0xB8, .name = "mov ax", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xB9, .name = "mov cx", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBA, .name = "mov dx", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBB, .name = "mov bx", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBC, .name = "mov sp", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBD, .name = "mov bp", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBE, .name = "mov si", .mode = {IMM16}, .size = 3},
+	{.opcode = 0xBF, .name = "mov di", .mode = {IMM16}, .size = 3},
+	{.name = NULL}
+};

src/main.c

@@ -2,33 +2,33 @@
 #include <string.h>
 #include <stdlib.h>
 #include "dasm.h"
+#include "sys/types.h"
 
-char *open_and_read(char *path)
+unsigned long open_and_read(char *path, u_int8_t **out)
 {
+	*out = NULL;
 	FILE *file = fopen(path, "r");
 	if (!file) {
 		dprintf(2, "Invalid path: %s. File does not exist.\n", path);
-		return NULL;
+		return 0;
 	}
 
 	int rc = fseek(file, 0L, SEEK_END);
 	if (rc != 0) {
 		fclose(file);
-		return NULL;
+		return 0;
 	}
 	long size = ftell(file);
 	if (size == -1) {
 		fclose(file);
-		return NULL;
+		return 0;
 	}
 	rewind(file);
 
-	char *ret = malloc((size + 1) * sizeof(char));
-	size = fread(ret, sizeof(char), size, file);
-	ret[size] = '\0';
-
+	*out = malloc(size * sizeof(u_int8_t));
+	size = fread(*out, sizeof(u_int8_t), size, file);
 	fclose(file);
-	return ret;
+	return size;
 }
 
 int main(int argc, char **argv)
@@ -38,13 +38,14 @@ int main(int argc, char **argv)
 		return 2;
 	}
 
-	char *binary = open_and_read(argv[2]);
+	u_int8_t *binary;
+	unsigned long size = open_and_read(argv[2], &binary);
 	if (!binary) {
 		puts("Could not read binary file.");
 		return 2;
 	}
 
-	int ret = dasm(binary);
+	int ret = dasm(binary, size);
 	free(binary);
 	return ret;
 }