mykola/blackjack
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

implement opcode matching as well as opreg

Browse source
This commit is contained in:
mykola2312 2024-08-16 12:21:58 +03:00
parent b636bc8891
commit 64f44d9a44
2 changed files with 40 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
include/rtdisasm.h
View file

@ -6,7 +6,7 @@
// code should point to place with machine instructions, and size
// limits the area of analyze, so no segfaults would be triggered on
// page boundaries.
// returns -1 when no instruction was found, 0 when size limit reached
// returns 0 when no instruction was found, -1 when size limit reached
// and non-negative-non-zero number of actual instruction size
int rtdisasm_analyze_single(const uint8_t* code, uint8_t size);

39
src/rtdisasm.c
View file

@ -80,7 +80,10 @@ static int test_rex_prefix(const uint8_t rex)
return -1;
}
const instruction_t* find_instruction(const uint8_t* cur, unsigned type, int vex, int rex)
// so we can ignore register encoded in opcode
#define OPREG_MASK 0b11111000
static const instruction_t* find_instruction(const uint8_t* cur, unsigned type, int vex, int rex)
{
for (unsigned i = 0; i < rtdisasm_table_len; i++)
{
@ -95,12 +98,33 @@ const instruction_t* find_instruction(const uint8_t* cur, unsigned type, int vex
continue;
}
// compare opcodes
if (ins->config.has_modrm)
{
// instruction encoding employs register embedded into last opcode byte
// so we need to apply bit mask
// plain means opcode bytes that are not affected
// by opcode register encoding
uint16_t plain_len = ins->opcode_len - 1;
if (plain_len)
{
if (memcmp(cur, &ins->opcode, plain_len))
continue;
}
// now let's match the opreg encoded byte
if (cur[plain_len] & OPREG_MASK != ins->opcode[plain_len])
continue;
}
else
{
// just compare opcodes
if (memcmp(cur, &ins->opcode, ins->opcode_len))
{
// opcodes don't match up, skip
continue;
}
}
// for now, everything looks good, so that's our instruction
return ins;
@ -113,7 +137,7 @@ int rtdisasm_analyze_single(const uint8_t* code, uint8_t size)
{
const uint8_t* cur = code;
const uint8_t* const end = code + size;
if (cur == end) return 0;
if (cur == end) return -1;
// skip standard prefixes
while (is_std_prefix(*cur))
@ -143,4 +167,13 @@ int rtdisasm_analyze_single(const uint8_t* code, uint8_t size)
}
const instruction_t* ins = find_instruction(cur, type, vex, rex);
if (!ins) return 0; // no instruction
// since we now instruction, we need advance past opcode bytes
cur += ins->opcode_len;
if (cur >= end) return -1;
// if instruction has ModRM, we need to analyze it,
// since it can lead to SIB byte
// if (ins->config.has_modrm)
}