1
0

update decoder to decode from memory

This commit is contained in:
Rokas Puzonas 2023-05-07 14:01:04 +03:00
parent 5ba1f1f948
commit 7c0024436d
4 changed files with 115 additions and 49 deletions

View File

@ -8,6 +8,7 @@
#include "sim8086.h"
#include "sim8086.c"
#include "sim8086_memory.c"
#include "sim8086_decoder.c"
#include "sim8086_simulator.c"
@ -86,39 +87,51 @@ err:
int dissassemble(FILE *src, FILE *dst) {
fprintf(dst, "bits 16\n\n");
struct memory mem = { .mem = { 0xFF } };
int byte_count = load_mem_from_stream(&mem, src, 0);
if (byte_count == -1) {
fprintf(stderr, "ERROR: Failed to load file to memory\n");
return -1;
}
char buff[256];
struct instruction inst;
int counter = 1;
while (true) {
enum decode_error err = decode_instruction(src, &inst);
u32 inst_address = 0;
while (inst_address < byte_count) {
enum decode_error err = decode_instruction(&mem, &inst_address, &inst);
if (err == DECODE_ERR_EOF) break;
if (err != DECODE_OK) {
fprintf(stderr, "ERROR: Failed to decode %d instruction: %s\n", counter, decode_error_to_str(err));
fprintf(stderr, "ERROR: Failed to decode instruction at 0x%08x: %s\n", inst_address, decode_error_to_str(err));
return -1;
}
instruction_to_str(buff, sizeof(buff), &inst);
fprintf(dst, buff);
fprintf(dst, "\n");
counter += 1;
}
return 0;
}
int simulate(FILE *src) {
struct memory mem;
int byte_count = load_mem_from_stream(&mem, src, 0);
if (byte_count == -1) {
fprintf(stderr, "ERROR: Failed to load file to memory\n");
return -1;
}
struct cpu_state state = { 0 };
struct instruction inst;
int counter = 1;
while (true) {
enum decode_error err = decode_instruction(src, &inst);
u32 inst_address = 0;
while (inst_address < byte_count) {
enum decode_error err = decode_instruction(&mem, &inst_address, &inst);
if (err == DECODE_ERR_EOF) break;
if (err != DECODE_OK) {
fprintf(stderr, "ERROR: Failed to decode %d instruction: %s\n", counter, decode_error_to_str(err));
fprintf(stderr, "ERROR: Failed to decode instruction at 0x%08x: %s\n", inst_address, decode_error_to_str(err));
return -1;
}
execute_instruction(&state, &inst);
counter += 1;
}
printf("Final registers:\n");
@ -134,7 +147,7 @@ int simulate(FILE *src) {
}
void print_usage(const char *program) {
fprintf(stderr, "Usage: %s <test-dump|dump|run> <file>\n", program);
fprintf(stderr, "Usage: %s <test-dump|dump|sim> <file>\n", program);
}
int test_decoder(const char *asm_file) {
@ -257,7 +270,7 @@ int main(int argc, char **argv) {
} else if (strequal(argv[1], "dump")) {
return dump_decompilation(argv[2]);
} else if (strequal(argv[1], "run")) {
} else if (strequal(argv[1], "sim")) {
return run_simulation(argv[2]);
} else {

View File

@ -11,7 +11,8 @@
#define panic(...) fprintf(stderr, "PANIC(%s:%d): ", __FILE__, __LINE__); fprintf(stderr, __VA_ARGS__); abort()
#define todo(...) fprintf(stderr, "TODO(%s:%d): ", __FILE__, __LINE__); fprintf(stderr, __VA_ARGS__); abort()
#define ARRAY_LEN(arr) sizeof(arr) / sizeof(arr[0])
#define ARRAY_LEN(arr) (sizeof(arr) / sizeof(arr[0]))
#define MEMORY_SIZE (1024 * 1024) // 1 MiB
enum operation {
OP_MOV,
@ -101,6 +102,10 @@ struct instruction {
i8 jmp_offset;
};
struct memory {
u8 mem[MEMORY_SIZE];
};
struct cpu_state {
u16 ax;
u16 bx;

View File

@ -1,5 +1,3 @@
#include <stdio.h>
#define dbg(...) printf("; "); printf(__VA_ARGS__); printf("\n")
// TODO: find a way to merge "to/from register" with "to/from accumulator" branches into a single code path
@ -73,24 +71,31 @@ static enum mem_base decode_mem_base(u8 rm) {
}
// Table 4-10. R/M (Register/Memory) Field Encoding
static void decode_reg_or_mem(struct reg_or_mem_value *value, FILE *src, u8 rm, u8 mod, bool wide) {
static void decode_reg_or_mem(
struct reg_or_mem_value *value,
struct memory *mem,
u32 *addr,
u8 rm,
u8 mod,
bool wide
) {
if (mod == 0b11) { // Mod = 0b11, register
value->is_reg = true;
value->reg = decode_reg(rm, wide);
} else if (mod == 0b10) { // Mod = 0b10, memory with i16 displacement
i16 displacement = fgetc(src) | (fgetc(src) << 8);
i16 displacement = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
value->is_reg = false;
value->mem.base = decode_mem_base(rm);
value->mem.disp = displacement;
} else if (mod == 0b01) { // Mod = 0b01, memory with i8 displacement
i8 displacement = fgetc(src);
i8 displacement = pull_byte_at(mem, addr);
value->is_reg = false;
value->mem.base = decode_mem_base(rm);
value->mem.disp = extend_sign_bit(displacement);
} else if (mod == 0b00) { // Mod = 0b00, memory no displacement (most of the time)
value->is_reg = false;
if (rm == 0b110) { // Direct address
u16 address = fgetc(src) | (fgetc(src) << 8);
u16 address = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
value->mem.base = MEM_BASE_DIRECT_ADDRESS;
value->mem.disp = address;
} else {
@ -102,9 +107,16 @@ static void decode_reg_or_mem(struct reg_or_mem_value *value, FILE *src, u8 rm,
}
}
static void deocde_reg_or_mem_to_src(struct src_value *value, FILE *src, u8 rm, u8 mod, bool wide) {
static void deocde_reg_or_mem_to_src(
struct src_value *value,
struct memory *mem,
u32 *addr,
u8 rm,
u8 mod,
bool wide
) {
struct reg_or_mem_value reg_or_mem;
decode_reg_or_mem(&reg_or_mem, src, rm, mod, wide);
decode_reg_or_mem(&reg_or_mem, mem, addr, rm, mod, wide);
if (reg_or_mem.is_reg) {
value->variant = SRC_VALUE_REG;
value->reg = reg_or_mem.reg;
@ -117,13 +129,12 @@ static void deocde_reg_or_mem_to_src(struct src_value *value, FILE *src, u8 rm,
// TODO: change to readinf from a byte buffer
// TODO: add handling for 'DECODE_ERR_MISSING_BYTES'
// Handy reference: Table 4-12. 8086 Instruction Encoding
enum decode_error decode_instruction(FILE *src, struct instruction *output) {
u8 byte1 = fgetc(src);
if (feof(src)) return DECODE_ERR_EOF;
enum decode_error decode_instruction(struct memory *mem, u32 *addr, struct instruction *output) {
u8 byte1 = pull_byte_at(mem, addr);
// MOVE: Register memory to/from register
if ((byte1 & 0b11111100) == 0b10001000) {
u8 byte2 = fgetc(src);
u8 byte2 = pull_byte_at(mem, addr);
bool wide = byte1 & 0b1;
bool direction = (byte1 & 0b10) >> 1;
@ -135,11 +146,11 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
if (direction) {
output->dest.is_reg = true;
output->dest.reg = decode_reg(reg, wide);
deocde_reg_or_mem_to_src(&output->src, src, rm, mod, wide);
deocde_reg_or_mem_to_src(&output->src, mem, addr, rm, mod, wide);
} else {
output->src.variant = SRC_VALUE_REG;
output->src.reg = decode_reg(reg, wide);
decode_reg_or_mem(&output->dest, src, rm, mod, wide);
decode_reg_or_mem(&output->dest, mem, addr, rm, mod, wide);
}
// MOVE: Immediate to register
@ -153,29 +164,30 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
if (wide) {
output->src.variant = SRC_VALUE_IMMEDIATE16;
output->src.immediate = fgetc(src) | (fgetc(src) << 8);
output->src.immediate = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
} else {
output->src.variant = SRC_VALUE_IMMEDIATE8;
output->src.immediate = fgetc(src);
output->src.immediate = pull_byte_at(mem, addr);
}
// MOVE: Immediate to register/memory
} else if ((byte1 & 0b11111110) == 0b11000110) {
u8 byte2 = fgetc(src);
u8 byte2 = pull_byte_at(mem, addr);
bool wide = byte1 & 0b1;
u8 mod = (byte2 & 0b11000000) >> 6;
u8 rm = byte2 & 0b00000111;
output->op = OP_MOV;
decode_reg_or_mem(&output->dest, src, rm, mod, wide);
decode_reg_or_mem(&output->dest, mem, addr, rm, mod, wide);
if (wide) {
output->src.variant = SRC_VALUE_IMMEDIATE16;
output->src.immediate = fgetc(src) | (fgetc(src) << 8);
output->src.immediate = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
} else {
output->src.variant = SRC_VALUE_IMMEDIATE8;
output->src.immediate = fgetc(src);
output->src.immediate = pull_byte_at(mem, addr);
}
// MOVE: Memory to accumulator
@ -188,9 +200,9 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
bool wide = byte1 & 0b1;
if (wide) {
output->src.mem.disp = fgetc(src) | (fgetc(src) << 8);
output->src.mem.disp = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
} else {
output->src.mem.disp = fgetc(src);
output->src.mem.disp = pull_byte_at(mem, addr);
}
// MOVE: Accumulator to memory
@ -204,9 +216,9 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
output->dest.mem.base = MEM_BASE_DIRECT_ADDRESS;
if (wide) {
output->dest.mem.disp = fgetc(src) | (fgetc(src) << 8);
output->dest.mem.disp = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
} else {
output->dest.mem.disp = fgetc(src);
output->dest.mem.disp = pull_byte_at(mem, addr);
}
// ADD/SUB/CMP: Reg/memory with register to either
@ -223,7 +235,7 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
bool wide = byte1 & 0b01;
bool direction = (byte1 & 0b10) >> 1;
u8 byte2 = fgetc(src);
u8 byte2 = pull_byte_at(mem, addr);
u8 mod = (byte2 & 0b11000000) >> 6;
u8 reg = (byte2 & 0b00111000) >> 3;
u8 rm = byte2 & 0b00000111;
@ -231,16 +243,16 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
if (direction) {
output->dest.is_reg = true;
output->dest.reg = decode_reg(reg, wide);
deocde_reg_or_mem_to_src(&output->src, src, rm, mod, wide);
deocde_reg_or_mem_to_src(&output->src, mem, addr, rm, mod, wide);
} else {
output->src.variant = SRC_VALUE_REG;
output->src.reg = decode_reg(reg, wide);
decode_reg_or_mem(&output->dest, src, rm, mod, wide);
decode_reg_or_mem(&output->dest, mem, addr, rm, mod, wide);
}
// ADD/SUB/CMP: immediate with register/memory
} else if ((byte1 & 0b11111100) == 0b10000000) {
u8 byte2 = fgetc(src);
u8 byte2 = pull_byte_at(mem, addr);
u8 variant = (byte2 & 0b00111000) >> 3;
if (variant == 0b000) {
@ -256,19 +268,19 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
u8 mod = (byte2 & 0b11000000) >> 6;
u8 rm = byte2 & 0b00000111;
decode_reg_or_mem(&output->dest, src, rm, mod, wide);
decode_reg_or_mem(&output->dest, mem, addr, rm, mod, wide);
if (wide) {
output->src.variant = SRC_VALUE_IMMEDIATE16;
if (sign_extend) {
output->src.immediate = fgetc(src);
output->src.immediate = pull_byte_at(mem, addr);
output->src.immediate = extend_sign_bit(output->src.immediate);
} else {
output->src.immediate = fgetc(src) | (fgetc(src) << 8);
output->src.immediate = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
}
} else {
output->src.variant = SRC_VALUE_IMMEDIATE8;
output->src.immediate = fgetc(src);
output->src.immediate = pull_byte_at(mem, addr);
}
// ADD/SUB/CMP: immediate with accumulator
@ -289,22 +301,22 @@ enum decode_error decode_instruction(FILE *src, struct instruction *output) {
if (wide) {
output->src.variant = SRC_VALUE_IMMEDIATE16;
output->src.immediate = fgetc(src) | (fgetc(src) << 8);
output->src.immediate = pull_byte_at(mem, addr) | (pull_byte_at(mem, addr) << 8);
} else {
output->src.variant = SRC_VALUE_IMMEDIATE8;
output->src.immediate = fgetc(src);
output->src.immediate = pull_byte_at(mem, addr);
}
// Conditional jumps
} else if ((byte1 & 0b11110000) == 0b01110000) {
i8 jmp_offset = fgetc(src);
i8 jmp_offset = pull_byte_at(mem, addr);
u8 opcode = byte1 & 0b00001111;
output->op = cond_jmp_lookup[opcode];
output->jmp_offset = jmp_offset;
// Conditional loop jumps
} else if ((byte1 & 0b11111100) == 0b11100000) {
i8 jmp_offset = fgetc(src);
i8 jmp_offset = pull_byte_at(mem, addr);
u8 opcode = byte1 & 0b00000011;
output->op = cond_loop_jmp_lookup[opcode];
output->jmp_offset = jmp_offset;

36
src/sim8086_memory.c Normal file
View File

@ -0,0 +1,36 @@
#include <stdio.h>
// TODO: add error codes
int load_mem_from_stream(struct memory *mem, FILE *stream, u32 start) {
u32 offset = 0;
while (true) {
u8 byte = fgetc(stream);
if (feof(stream)) break;
if (start + offset >= MEMORY_SIZE) return -1;
mem->mem[start + offset] = byte;
offset++;
}
return offset;
}
int load_mem_from_file(struct memory *mem, const char *filename, u32 start) {
FILE *stream = fopen(filename, "rb");
if (stream == NULL) {
return -1;
}
int byte_count = load_mem_from_stream(mem, stream, start);
fclose(stream);
return byte_count;
}
// TODO: Make this error some kind of error, when reading past end
u8 read_byte_at(struct memory *mem, u32 address) {
return mem->mem[address % MEMORY_SIZE];
}
u8 pull_byte_at(struct memory *mem, u32 *address) {
u8 byte = read_byte_at(mem, *address);
(*address)++;
return byte;
}