T0b1/fadec
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

decode: Faster prefix decoder

Browse Source 
This improves decoding performance by ~5%.
This commit is contained in:
Alexis Engelke
2023-01-21 12:29:10 +01:00
parent b48495805e
commit 2c21073379
1 changed files with 48 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

113
decode.c
View File

@@ -122,71 +122,54 @@ fd_decode(const uint8_t* buffer, size_t len_sz, int mode_int, uintptr_t address,
int off = 0;
uint8_t vex_operand = 0;
unsigned prefix_rep = 0;
bool prefix_lock = false;
bool prefix_66 = false;
uint8_t addr_size = mode == DECODE_64 ? 3 : 2;
unsigned prefix_rex = 0;
int rex_off = -1;
unsigned vexl = 0;
unsigned prefix_evex = 0;
instr->segment = FD_REG_NONE;
if (mode == DECODE_32) {
while (LIKELY(off < len))
{
uint8_t prefix = buffer[off];
switch (UNLIKELY(prefix))
{
default: goto prefix_end;
// From segment overrides, the last one wins.
case 0x26: instr->segment = FD_REG_ES; break;
case 0x2e: instr->segment = FD_REG_CS; break;
case 0x36: instr->segment = FD_REG_SS; break;
case 0x3e: instr->segment = FD_REG_DS; break;
case 0x64: instr->segment = FD_REG_FS; break;
case 0x65: instr->segment = FD_REG_GS; break;
case 0x66: prefix_66 = true; break;
case 0x67: addr_size = 1; break;
case 0xf0: prefix_lock = true; break;
case 0xf3: prefix_rep = 2; break;
case 0xf2: prefix_rep = 3; break;
}
off++;
}
}
if (mode == DECODE_64) {
while (LIKELY(off < len))
{
uint8_t prefix = buffer[off];
switch (UNLIKELY(prefix))
{
default: goto prefix_end;
// ES/CS/SS/DS overrides are ignored.
case 0x26: case 0x2e: case 0x36: case 0x3e: break;
// From segment overrides, the last one wins.
case 0x64: instr->segment = FD_REG_FS; break;
case 0x65: instr->segment = FD_REG_GS; break;
case 0x66: prefix_66 = true; break;
case 0x67: addr_size = 2; break;
case 0xf0: prefix_lock = true; break;
case 0xf3: prefix_rep = 2; break;
case 0xf2: prefix_rep = 3; break;
case 0x40: case 0x41: case 0x42: case 0x43: case 0x44: case 0x45:
case 0x46: case 0x47: case 0x48: case 0x49: case 0x4a: case 0x4b:
case 0x4c: case 0x4d: case 0x4e: case 0x4f:
prefix_rex = prefix;
rex_off = off;
break;
}
off++;
}
}
enum {
PF_SEG1 = 1|8,
PF_SEG2 = 1,
PF_66 = 3,
PF_67 = 4,
PF_LOCK = 5,
PF_REP = 6,
PF_REX = 8,
};
prefix_end:
// REX prefix is only considered if it is the last prefix.
if (rex_off != off - 1)
prefix_rex = 0;
static const uint8_t pflut[256] = {
[0x26] = PF_SEG1, [0x2e] = PF_SEG1, [0x36] = PF_SEG1, [0x3e] = PF_SEG1,
[0x64] = PF_SEG2, [0x65] = PF_SEG2, [0x66] = PF_66, [0x67] = PF_67,
[0xf0] = PF_LOCK, [0xf2] = PF_REP, [0xf3] = PF_REP,
[0x40] = PF_REX, [0x41] = PF_REX, [0x42] = PF_REX, [0x43] = PF_REX,
[0x44] = PF_REX, [0x45] = PF_REX, [0x46] = PF_REX, [0x47] = PF_REX,
[0x48] = PF_REX, [0x49] = PF_REX, [0x4a] = PF_REX, [0x4b] = PF_REX,
[0x4c] = PF_REX, [0x4d] = PF_REX, [0x4e] = PF_REX, [0x4f] = PF_REX,
};
uint8_t prefixes[12] = {0};
uint8_t lutmask = mode == DECODE_64 ? 0xff : 0x7;
while (LIKELY(off < len)) {
uint8_t prefix = buffer[off];
uint8_t lut = pflut[prefix] & lutmask;
if (LIKELY(!lut))
break;
prefixes[lut] = prefix;
off++;
}
if (off) {
if (UNLIKELY(prefixes[PF_SEG2])) {
if (prefixes[PF_SEG2] & 0x02)
instr->segment = prefixes[PF_SEG2] >> 3 & 3;
else
instr->segment = prefixes[PF_SEG2] & 7;
}
if (UNLIKELY(prefixes[PF_67]))
addr_size--;
if (buffer[off - 1] == prefixes[PF_REX])
prefix_rex = prefixes[PF_REX];
}
uint8_t prefix_rep = prefixes[PF_REP];
if (UNLIKELY(off >= len))
return FD_ERR_PARTIAL;
@@ -207,7 +190,7 @@ prefix_end:
// If there is no REP/REPNZ prefix offer 66h as mandatory prefix. If
// there is a REP prefix, then the 66h prefix is ignored here.
mandatory_prefix = prefix_rep ? prefix_rep : !!prefix_66;
mandatory_prefix = prefix_rep ? prefix_rep ^ 0xf1 : !!prefixes[PF_66];
}
else if (UNLIKELY((unsigned) buffer[off] - 0xc4 < 2 || buffer[off] == 0x62))
{
@@ -221,7 +204,7 @@ prefix_end:
// VEX/EVEX + 66/F3/F2/REX will #UD.
// Note: REX is also here only respected if it immediately precedes the
// opcode, in this case the VEX/EVEX "prefix".
if (prefix_66 || prefix_rep || prefix_rex)
if (prefixes[PF_66] || prefixes[PF_REP] || prefix_rex)
return FD_ERR_UD;
uint8_t byte = buffer[off + 1];
@@ -328,8 +311,8 @@ prefix_end:
instr->type = desc->type;
instr->addrsz = addr_size;
instr->flags = prefix_rep == 2 ? FD_FLAG_REP :
prefix_rep == 3 ? FD_FLAG_REPNZ : 0;
instr->flags = prefix_rep == 0xf3 ? FD_FLAG_REP :
prefix_rep == 0xf2 ? FD_FLAG_REPNZ : 0;
if (mode == DECODE_64)
instr->flags |= FD_FLAG_64;
instr->address = address;
@@ -384,11 +367,11 @@ prefix_end:
op_size = 1;
else if (mode == DECODE_64)
op_size = ((prefix_rex & PREFIX_REXW) || DESC_OPSIZE(desc) == 3) ? 4 :
UNLIKELY(prefix_66 && !DESC_IGN66(desc)) ? 2 :
UNLIKELY(prefixes[PF_66] && !DESC_IGN66(desc)) ? 2 :
DESC_OPSIZE(desc) ? 4 :
3;
else
op_size = UNLIKELY(prefix_66 && !DESC_IGN66(desc)) ? 2 : 3;
op_size = UNLIKELY(prefixes[PF_66] && !DESC_IGN66(desc)) ? 2 : 3;
} else {
op_size = 5 + vexl;
op_size_alt = op_size - (DESC_OPSIZE(desc) & 3);
@@ -694,7 +677,7 @@ prefix_end:
}
skip_modrm:
if (UNLIKELY(prefix_lock)) {
if (UNLIKELY(prefixes[PF_LOCK])) {
if (!DESC_LOCK(desc) || instr->operands[0].type != FD_OT_MEM)
return FD_ERR_UD;
instr->flags |= FD_FLAG_LOCK;