T0b1/test-repo
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Output more available information in formatter

Browse Source 
The formatter now includes the following information:

- Segment overrides
- Address-size overrides
- REP/REPNZ prefixes
- LOCK prefix
- High-byte registers (determined using presence of REX prefix)
This commit is contained in:
Alexis Engelke
2019-01-13 20:38:15 +01:00
parent 2d1a9582b8
commit 8fdfe53822
9 changed files with 121 additions and 180 deletions
Download Patch File Download Diff File Expand all files Collapse all files

195
format.c
View File

@@ -1,6 +1,7 @@
#include <stddef.h>
#include <stdint.h>
#include <stdio.h>
#include <decode.h>
@@ -17,152 +18,90 @@ static const uint16_t _mnemonic_offs[] = {
};
#undef DECODE_TABLE_STRTAB2
static
void
instr_format_decimal(char** cur, uint32_t value)
{
char buffer[32];
size_t buf_idx = sizeof(buffer) - 1;
if (value == 0)
{
buffer[buf_idx] = '0';
}
else
{
while (value > 0)
{
uint32_t digit = value % 10;
buffer[buf_idx--] = '0' + digit;
value /= 10;
}
buf_idx++;
}
size_t length = sizeof(buffer) - buf_idx;
__builtin_memcpy(*cur, buffer + buf_idx, length);
*cur += length;
}
static
void
instr_format_hex(char** cur, size_t value)
{
char buffer[32];
size_t buf_idx = sizeof(buffer) - 1;
if (value == 0)
{
buffer[buf_idx] = '0';
}
else
{
while (value > 0)
{
uint32_t nibble = value & 0xf;
buffer[buf_idx--] = "0123456789abcdef"[nibble];
value >>= 4;
}
buf_idx++;
}
buffer[--buf_idx] = 'x';
buffer[--buf_idx] = '0';
size_t length = sizeof(buffer) - buf_idx;
__builtin_memcpy(*cur, buffer + buf_idx, length);
*cur += length;
}
#define FMT_CONCAT(buf, end, ...) do { \
buf += snprintf(buf, end - buf, __VA_ARGS__); \
if (buf > end) \
buf = end; \
} while (0)
void
instr_format(const Instr* instr, char buffer[128])
{
char* cur = buffer;
*(cur++) = '[';
char* buf = buffer;
char* end = buffer + 128;
const char* mnemonic = &_mnemonic_str[_mnemonic_offs[instr->type]];
while (*mnemonic)
{
*(cur++) = *(mnemonic++);
}
FMT_CONCAT(buf, end, "[");
if (INSTR_HAS_REP(instr))
FMT_CONCAT(buf, end, "rep:");
if (INSTR_HAS_REPNZ(instr))
FMT_CONCAT(buf, end, "repnz:");
if (INSTR_SEGMENT(instr) < 6)
FMT_CONCAT(buf, end, "%cs:", "ecsdfg"[INSTR_SEGMENT(instr)]);
if (INSTR_IS64(instr) && INSTR_ADDRSZ(instr) == 4)
FMT_CONCAT(buf, end, "addr32:");
if (!INSTR_IS64(instr) && INSTR_ADDRSZ(instr) == 2)
FMT_CONCAT(buf, end, "addr16:");
if (INSTR_HAS_LOCK(instr))
FMT_CONCAT(buf, end, "lock:");
FMT_CONCAT(buf, end, "%s", &_mnemonic_str[_mnemonic_offs[instr->type]]);
if (INSTR_WIDTH(instr))
{
*(cur++) = '_';
instr_format_decimal(&cur, INSTR_WIDTH(instr));
}
FMT_CONCAT(buf, end, "_%u", INSTR_WIDTH(instr));
for (int i = 0; i < 4; i++)
{
const struct Operand* operand = &instr->operands[i];
if (operand->type == OT_NONE)
{
break;
}
__builtin_memcpy(cur, " REG IMM MEM" + operand->type * 4 - 4, 4);
cur += 4;
instr_format_decimal(&cur, operand->size);
*(cur++) = ':';
const char* op_type_name = "reg\0imm\0mem" + operand->type * 4 - 4;
FMT_CONCAT(buf, end, " %s%u:", op_type_name, operand->size);
switch (operand->type)
{
size_t immediate;
case OT_REG:
instr_format_decimal(&cur, reg_index(operand->reg));
break;
case OT_IMM:
immediate = instr->immediate;
if (operand->size == 1)
{
immediate &= 0xff;
}
else if (operand->size == 2)
{
immediate &= 0xffff;
}
#if defined(ARCH_X86_64)
else if (operand->size == 4)
{
immediate &= 0xffffffff;
}
#endif
instr_format_hex(&cur, immediate);
break;
case OT_MEM:
if (!reg_is_none(operand->reg))
{
instr_format_decimal(&cur, reg_index(operand->reg));
*(cur++) = ':';
}
if (instr->scale != 0)
{
uint8_t scale = 1 << (instr->scale - 1);
instr_format_decimal(&cur, scale);
*(cur++) = '*';
instr_format_decimal(&cur, reg_index(instr->sreg));
*(cur++) = ':';
}
if (instr->disp < 0)
{
*(cur++) = '-';
instr_format_hex(&cur, -instr->disp);
}
else
{
instr_format_hex(&cur, instr->disp);
}
break;
case OT_NONE:
default:
break;
size_t immediate;
case OT_REG:
if (operand->size == 1 && !INSTR_HAS_REX(instr) &&
operand->reg >= 4 && operand->reg < 8)
FMT_CONCAT(buf, end, "r%uh", operand->reg - 4);
else
FMT_CONCAT(buf, end, "r%u", operand->reg);
break;
case OT_IMM:
immediate = instr->immediate;
if (operand->size == 1)
immediate &= 0xff;
else if (operand->size == 2)
immediate &= 0xffff;
else if (operand->size == 4)
immediate &= 0xffffffff;
FMT_CONCAT(buf, end, "0x%lx", immediate);
break;
case OT_MEM:
if (!reg_is_none(operand->reg))
{
FMT_CONCAT(buf, end, "r%u", operand->reg);
if (instr->scale != 0 || instr->disp > 0)
FMT_CONCAT(buf, end, "+");
}
if (instr->scale != 0)
{
FMT_CONCAT(buf, end, "%u*r%u", 1 << (instr->scale - 1),
instr->sreg);
if (instr->disp > 0)
FMT_CONCAT(buf, end, "+");
}
if (instr->disp < 0)
FMT_CONCAT(buf, end, "-0x%lx", -instr->disp);
else if ((reg_is_none(operand->reg) && instr->scale == 0) ||
instr->disp > 0)
FMT_CONCAT(buf, end, "0x%lx", instr->disp);
break;
case OT_NONE:
default:
break;
}
}
*(cur++) = ']';
*(cur++) = '\0';
#ifndef NDEBUG
if (cur - buffer > 128)
{
__builtin_trap();
}
#endif
FMT_CONCAT(buf, end, "]");
}