WIP Refactor: Instruction parser using a struct

decode.c

@@ -1,10 +1,11 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include "lib.h"
+#include <sys/stat.h>
 
 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
 
-enum Instruction
+enum InstructionType
 {
     INST_MOV_REG_REG = 0b10001000,
     // INST_MOV_REG_REG = 0b10001000,
@@ -24,15 +25,11 @@ typedef struct Register
     char code;
     char* fullname;
     char* bytename;
-
+    union {
-    union
+        struct {
-    {
-        struct
-        {
             char low;
             char high;
         };
-
         u16 full;
     } value;
 } Register;
@@ -48,9 +45,23 @@ Register registers[8] = {
     {.code = 0b111, .fullname = "di", .bytename = "bh"},
 };
 
-char* memory[65536];
+typedef struct Instruction
+{
+    u16 id;
+    char *name;
+    u16_opt data;
+    u16_opt displacement;
+    u8_opt w;
+    u8_opt d;
+    u8_opt s;
+    u8_opt mod;
+    u8_opt reg;
+    u8_opt rm;
+    u8_opt SR;
+    u8 bytes_read;
+} Instruction;
 
-// void inst_mov_rgmm_reg()
+char *memory[65536];
 
 /// Get Effective Address Calculation Registers
 char* get_eac_registers(char rm)
@@ -58,47 +69,118 @@ char* get_eac_registers(char rm)
     char* reg_name;
     switch (rm)
     {
-    case 0b000:
+    case 0b000: reg_name = "bx + si"; break;
-        reg_name = "bx + si";
+    case 0b001: reg_name = "bx + di"; break;
-        break;
+    case 0b010: reg_name = "bp + si"; break;
-    case 0b001:
+    case 0b011: reg_name = "bp + di"; break;
-        reg_name = "bx + di";
+    case 0b100: reg_name = "si";      break;
-        break;
+    case 0b101: reg_name = "di";      break;
-    case 0b010:
+    case 0b110: reg_name = "bp";      break;
-        reg_name = "bp + si";
+    case 0b111: reg_name = "bx";      break;
-        break;
+    default: perror("Invalid R/M value"); exit(1);
-    case 0b011:
-        reg_name = "bp + di";
-        break;
-    case 0b100:
-        reg_name = "si";
-        break;
-    case 0b101:
-        reg_name = "di";
-        break;
-    case 0b110:
-        reg_name = "bp";
-        break;
-    case 0b111:
-        reg_name = "bx";
-        break;
-    default:
-        perror("Invalid R/M value");
-        exit(1);
     }
     return reg_name;
 }
 
-static inline char* reg_name(Register reg, char wide)
+static char* reg_name(Register reg, char wide)
 {
     return wide == 1 ? reg.fullname : reg.bytename;
 }
 
-static inline i16 get_data(unsigned char* buf, char wide)
+static i16 get_data(unsigned char* buf, char wide)
 {
     // Cast buf[0] to sbyte if not the conversion to i16 won't detect signedness
     return wide == 1 ? (i16)buf[1] << 8 | buf[0] : (sbyte)buf[0];
 }
+#define IS_INST(value, m, inst) ((value & ~m) == (char)inst)
+
+typedef struct InstMask {char *name; u8 mask; u8 code;} InstMask;
+
+typedef struct InstFormat
+{
+    u16 id;
+    char *name;
+    u8 inst_enc;
+    u8 mask_inst;
+    u8 mask_w;
+    u64 mask_reg;
+    bool has_operands;
+    bool has_displacement;
+    bool has_data;
+    bool has_d;
+    bool has_w;
+    bool has_reg;
+    bool has_mod;
+    bool has_rm;
+    bool has_s;
+    bool has_SR;
+} InstFormat;
+
+InstFormat inst_formats[] =
+{
+    ////////
+    // MOV
+    ////////
+    // Register/memory to/from register
+    {.id=1, .name="mov", .inst_enc=0b10001000, .mask_inst=0x3, .has_operands=true, .has_displacement=true,
+     .has_d=true, .has_w=true, .has_reg=true, .has_mod=true, .has_rm=true},
+    // Immediate to register/memory
+    {.id=2, .name="mov", .inst_enc=0b11000110, .mask_inst=0x1, .mask_w=0x1, .has_operands=true,
+     .has_displacement=true, .has_data=true, .has_w=true, .has_mod=true, .has_rm=true},
+    // Immediate to register
+    {.id=3, .name="mov", .inst_enc=0b10110000, .mask_inst=0xF, .mask_w=0x8,
+     .mask_reg=0x0b00000111, .has_data=true, .has_w=true},
+    // Memory to accumulator | Accumulator to memory using the `d` bit
+    // even though the manual doesn't specify it
+    {.id=4, .name="mov", .inst_enc=0b10100000, .mask_inst=0x3, .mask_w=0x1,
+     .has_data=true, .has_w=true, .has_d=true},
+    // Register/memory to segment register and inverse using the `d` bit
+    {.id=5, .name="mov", .inst_enc=0b10001100, .mask_inst=0x3, .has_SR=true, .has_d=true,
+     .has_displacement=true, .has_mod=true, .has_rm=true},
+    ////////
+    // ADD
+    ////////
+    // Reg/memory with register or either
+    {.id=6, .name="add", .inst_enc=0b00000000, .mask_inst=0x3, .mask_w=0x1,
+     .has_operands=true, .has_displacement=true, .has_w=true,
+     .has_d=true, .has_reg=true, .has_mod=true, .has_rm=true},
+    // Immediate to register/memory
+    {.id=7, .name="add", .inst_enc=0b10000000, .mask_inst=0x3, .mask_w=0x1, .has_w=true,
+     .has_operands=true, .has_displacement=true, .has_data=true, .has_mod=true, .has_rm=true},
+    {.id=8, .name="add", .inst_enc=0b00000100, .mask_inst=0x1, .mask_w=0x1, .has_data=true, .has_w=true},
+};
+
+Instruction parse_instruction(u8* buf)
+{
+    u8 inst = buf[0];
+    InstFormat format;
+    bool matched_inst = false;
+    // TODO: This might be a good time to learn how to make a hashtable in C
+    for (u16 i = 0; i < sizeof(inst_formats) / sizeof(InstFormat); i++)
+    {
+        if ((inst & ~inst_formats[i].mask_inst) == inst_formats[i].inst_enc)
+        {
+            format = inst_formats[i];
+            matched_inst = true;
+            break;
+        }
+    }
+    if (!matched_inst)
+        return (Instruction){.bytes_read = 0};
+    return (Instruction) {
+        .id = format.id,
+        .name = format.name,
+        .data = none_u16(),
+        .displacement = none_u16(),
+        .w = none_u8(),
+        .d = none_u8(),
+        .s = none_u8(),
+        .mod = none_u8(),
+        .reg = none_u8(),
+        .rm = none_u8(),
+        .bytes_read = 2,
+    };
+}
 
 bool mov_inst(FILE* f, unsigned char* buf, char inst)
 {
@@ -262,27 +344,58 @@ int main(int argc, char** argv)
         printf("Usage: Please provide assembled instructions as input\n");
         exit(0);
     }
-    unsigned char buf[256];
+
+    struct stat st;
+    if (stat(argv[1], &st) == -1)
+    {
+        perror("Unable to get file size\n");
+        return EXIT_FAILURE;
+    }
+
+    unsigned char* buffer = malloc(st.st_size);
+    if (!buffer)
+    {
+        perror("Unable to allocate memory for binary file");
+        return EXIT_FAILURE;
+    }
+
     FILE *f = fopen(argv[1], "r");
     if (!f)
     {
         perror("fopen\n");
+        free(buffer);
         return EXIT_FAILURE;
     }
 
-    size_t bytes_read;
+    size_t bytes_read = fread(buffer, sizeof(unsigned char), st.st_size, f);
+    if (bytes_read != (size_t)st.st_size)
+    {
+        fprintf(stderr, "Read of binary file to memory incomplete.\n");
+        free(buffer);
+        fclose(f);
+        return EXIT_FAILURE;
+    }
+
+    fclose(f);
 
     printf("; Decoded 8086 Assembly Instructions\n\n");
     printf("bits 16\n\n");
 
-    while ((bytes_read = fread(buf, sizeof(char), 1, f)) > 0)
+    u32 bytes_processed = 0;
+    while (bytes_processed < bytes_read)
     {
-        char inst = buf[0];
+        Instruction inst = parse_instruction(buffer+bytes_processed);
-        if (mov_inst(f, buf, inst)) goto handled;
+        bytes_processed += inst.bytes_read;
-        if (add_inst(f, buf, inst)) goto handled;
+        // char inst = buffer[0];
+        // if (mov_inst(f, buffer, inst)) goto handled;
+        // if (add_inst(f, buffer, inst)) goto handled;
 
-        fprintf(stderr, "___Unrecognized Instruction___");
+        if (inst.name != NULL)
-    handled:
+            printf("%s ;%d", inst.name, inst.id);
+        else
+            fprintf(stderr, "___Unrecognized Instruction___");
+    // handled:
         printf("\n");
     }
+    free(buffer);
 }

lib.h

@@ -50,3 +50,4 @@ enum OptionTag {NONE, SOME};
 
 
 OPTION(u8)
+OPTION(u16)