From 3e9df7670078bf3b3f9ab02ed0a096bf4285d95f Mon Sep 17 00:00:00 2001
From: Zoe Roux <zoe.roux@zoriya.dev>
Date: Wed, 31 May 2023 16:21:22 +0900
Subject: [PATCH] Add imediate memory support

---
 src/dasm.c         | 28 +++++++++++++++++++++++-----
 src/instructions.c | 33 ++++++++++++++++-----------------
 src/main.c         |  5 +++++
 3 files changed, 44 insertions(+), 22 deletions(-)

diff --git a/src/dasm.c b/src/dasm.c
index e70c255..8c134e1 100644
--- a/src/dasm.c
+++ b/src/dasm.c
@@ -4,7 +4,20 @@
 
 void print_instruction(unsigned addr, instruction_t inst, unsigned raw)
 {
-	printf("%04x: %-14x %s\n", addr, raw, inst.name);
+	printf("%04x: %-14x %s", addr, raw, inst.name);
+	for (int i = 0; inst.mode[i] != END; i++) {
+		switch (inst.mode[i]) {
+		case IMM8:
+			printf(", %02x", raw & 0xFF);
+			break;
+		case IMM16:
+			printf(", %04x", raw & 0xFFFF);
+			break;
+		default:
+			printf(", undefined");
+		}
+	}
+	printf("\n");
 }
 
 unsigned read_size(u_int8_t *binary, unsigned size)
@@ -12,7 +25,8 @@ 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);
+		ret <<= 8;
+		ret |= binary[i];
 	}
 	return ret;
 }
@@ -30,7 +44,7 @@ instruction_t parse_inst(u_int8_t *binary)
 
 int dasm(u_int8_t *binary, unsigned long size)
 {
-	int pc = 0;
+	unsigned long pc = 0;
 	int header_size = 0;
 
 	if (binary[0] == 0xEB && binary[1] == 0x0E)
@@ -38,11 +52,15 @@ int dasm(u_int8_t *binary, unsigned long size)
 	else if (binary[0] == 0x01 && binary[1] == 0x03)
 		header_size = binary[4];
 
-	printf("Header size: %d\n", header_size);
 	binary += header_size;
+	size -= header_size;
 
-	while (pc < size - header_size) {
+	while (pc < size) {
 		instruction_t inst = parse_inst(binary);
+		if (pc + inst.size > size) {
+			printf("Invalid file. Missing operand for instruction: '%s'. (pc: %lx)\n", inst.name, pc);
+			return 1;
+		}
 		unsigned long raw = read_size(binary, inst.size);
 		print_instruction(pc, inst, raw);
 		pc += inst.size;
diff --git a/src/instructions.c b/src/instructions.c
index 50642d8..144c2d4 100644
--- a/src/instructions.c
+++ b/src/instructions.c
@@ -3,22 +3,21 @@
 
 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},
+	{.opcode = 0xB0, .name = "mov al", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB1, .name = "mov cl", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB2, .name = "mov dl", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB3, .name = "mov bl", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB4, .name = "mov ah", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB5, .name = "mov ch", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB6, .name = "mov dh", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB7, .name = "mov bh", .mode = {IMM8, END}, .size = 2},
+	{.opcode = 0xB8, .name = "mov ax", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xB9, .name = "mov cx", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBA, .name = "mov dx", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBB, .name = "mov bx", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBC, .name = "mov sp", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBD, .name = "mov bp", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBE, .name = "mov si", .mode = {IMM16, END}, .size = 3},
+	{.opcode = 0xBF, .name = "mov di", .mode = {IMM16, END}, .size = 3},
 	{.name = NULL}
 };
diff --git a/src/main.c b/src/main.c
index 981d7b9..968bb93 100644
--- a/src/main.c
+++ b/src/main.c
@@ -44,6 +44,11 @@ int main(int argc, char **argv)
 		puts("Could not read binary file.");
 		return 2;
 	}
+	// Smaller than the header size.
+	if (size < 32) {
+		puts("Binary too small.");
+		return 2;
+	}
 
 	int ret = dasm(binary, size);
 	free(binary);