performance-aware/decode.c

#include <stdio.h>
#include <stdlib.h>
#include "lib.h"

#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))

enum Instruction
{
    INST_MOV_REG_REG = 0b10001000,
    // INST_MOV_REG_REG = 0b10001000,
    // INST_MOV_REG_REG = 0b10001000,
};

enum Mode
{
    MODE_MEM_NO_DIS = 0b00,
    MODE_MEM_DIS_08 = 0b01,
    MODE_MEM_DIS_16 = 0b10,
    MODE_RGSTR_MODE = 0b11,
};

typedef struct Register
{
    char code;
    char* fullname;
    char* bytename;

    union
    {
        struct
        {
            char low;
            char high;
        };

        u16 full;
    } value;
} Register;

Register registers[8] = {
    {.code = 0b000, .fullname = "ax", .bytename = "al"},
    {.code = 0b001, .fullname = "cx", .bytename = "cl"},
    {.code = 0b010, .fullname = "dx", .bytename = "dl"},
    {.code = 0b011, .fullname = "bx", .bytename = "bl"},
    {.code = 0b100, .fullname = "sp", .bytename = "ah"},
    {.code = 0b101, .fullname = "bp", .bytename = "ch"},
    {.code = 0b110, .fullname = "si", .bytename = "dh"},
    {.code = 0b111, .fullname = "di", .bytename = "bh"},
};

char* memory[65536];

// void inst_mov_rgmm_reg()

int main(int argc, char** argv)
{
    if (argc < 2)
    {
        printf("Usage: Please provide assembled instructions as input\n");
        exit(0);
    }
    unsigned char buf[256];
    FILE *f = fopen(argv[1], "r");
    if (!f)
    {
        perror("fopen\n");
        return EXIT_FAILURE;
    }
    size_t bytes_read;

    printf("; Decoded 8086 Assembly Instructions\n\n");
    printf("bits 16\n\n");

    while ((bytes_read = fread(buf, sizeof(char), 1, f)) > 0)
    {
        char inst = buf[0];
        // Instruction instruction = 0;
        // Register/memory to/from register
        if ((inst & ~0x3) == (char)0b10001000)
        {
            // TODO: We should add some form of error handling here
            bytes_read = fread(buf, sizeof(char), 1, f);
            char next_byte = buf[0];
            char w = inst & 0b00000001;
            char d = inst & 0b00000010;
            char mod = (next_byte & 0b11000000) >> 6;
            char reg = (next_byte & 0b00111000) >> 3;
            char rm = (next_byte & 0b00000111);
            size_t reg_idx = reg;
            size_t rm_idx = rm;
            Register src_reg = d == 0 ? registers[reg_idx] : registers[rm_idx];
            Register dst_reg = d == 0 ? registers[rm_idx] : registers[reg_idx];
            if (mod == MODE_RGSTR_MODE)
            {
                char* src_name = w == 1 ? src_reg.fullname : src_reg.bytename;
                char* dst_name = w == 1 ? dst_reg.fullname : dst_reg.bytename;
                printf("mov %s, %s", dst_name, src_name);
            }
            else
            {
                // char* src_name = w == 1 ? src_reg.fullname : src_reg.bytename;
                char *dst_name;
                if (d)
                {
                    dst_name = w == 1 ? dst_reg.fullname : dst_reg.bytename;
                }
                else
                {
                    dst_name = w == 1 ? src_reg.fullname : src_reg.bytename;
                }
                char* src_name;
                switch (rm)
                {
                case 0b000:
                    src_name = "bx + si";
                    break;
                case 0b001:
                    src_name = "bx + di";
                    break;
                case 0b010:
                    src_name = "bp + si";
                    break;
                case 0b011:
                    src_name = "bp + di";
                    break;
                case 0b100:
                    src_name = "si";
                    break;
                case 0b101:
                    src_name = "di";
                    break;
                case 0b110:
                    src_name = "bp";
                    break;
                case 0b111:
                    src_name = "bx";
                    break;
                }
                if (mod == MODE_MEM_DIS_16)
                {
                    bytes_read = fread(buf, sizeof(char), 2, f);
                    i16 data = (i16)buf[1] << 8 | buf[0];
                    if (d)
                    {
                        printf("mov %s, [%s + %d] ;(mod %d, w %d, d %d)", dst_name, src_name, data, mod, d, w);
                    }
                    else
                    {
                        printf("mov [%s], %s ;(mod %d, w %d, d %d)", src_name, dst_name, mod, d, w);
                    }
                    char disp_lo = buf[0];
                    char disp_hi = buf[1];
                    (void)disp_lo;
                    (void)disp_hi;
                }
                else if (mod == MODE_MEM_DIS_08)
                {
                    bytes_read = fread(buf, sizeof(char), 1, f);
                    if (d)
                    {
                        printf("mov %s, [%s + %d] ;(mod %d, w %d, d %d)", dst_name, src_name, buf[0], mod, d, w);
                    }
                    else
                    {
                        printf("mov [%s], %s ;(mod %d, w %d, d %d)", src_name, dst_name, mod, d, w);
                    }
                }
                else if (mod == MODE_MEM_NO_DIS)
                {
                    if (d)
                    {
                        printf("mov %s, [%s] ;(mod %d, w %d, d %d)", dst_name, src_name, mod, d, w);
                    }
                    else
                    {
                        printf("mov [%s], %s ;(mod %d, w %d, d %d)", src_name, dst_name, mod, d, w);
                    }
                }
            }
        }
        // Immediate to register/memory
        else if ((inst & ~0x1) == (char)0b11000110)
        {
            printf("mov imm to reg/mem");
        }
        // Immediate to register
        else if ((inst & ~0xF) == (char)0b10110000)
        {
            char w   = inst & 0b00001000;
            Register reg = registers[(size_t)inst & 0b00000111];
            char *reg_name = w == 0 ? reg.bytename : reg.fullname;
            if (w)
            {
                bytes_read = fread(buf, sizeof(char), 2, f);
                i16 data = (i16)buf[1] << 8 | buf[0];
                printf("mov %s, %hd     ; Immediate to register", reg_name, data);
            }
            else
            {
                bytes_read = fread(buf, sizeof(char), 1, f);
                printf("mov %s, %d     ; Immediate to register", reg_name, (sbyte)buf[0]);
            }
        }
        // Memory to accumulator
        else if ((inst & ~0x1) == (char)0b10100000)
        {
            printf("mov mem to acc");
        }
        // Accumulator to memory
        else if ((inst & ~0x1) == (char)0b10100010)
        {
            printf("mov acc to mem");
        }
        // Register/memory to segment register
        else if (inst == (char)0b10001110)
        {
            printf("mov regmem to segreg");
        }
        // Segment register to register/memory
        else if (inst == (char)0b10001100)
        {
            printf("mov segreg to regmem");
        }
        else
        {
            fprintf(stderr, "Unrecognized Instruction");
        }

        printf("\n");
    }
}