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

decode: Store instruction descriptors separately

Browse Source
This commit is contained in:
Alexis Engelke
2020-11-22 22:27:43 +01:00
parent bb8510d77f
commit fd80706f54
2 changed files with 40 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
decode.c
View File

@@ -9,12 +9,6 @@
#define LIKELY(x) __builtin_expect((x), 1) #define LIKELY(x) __builtin_expect((x), 1)
#define UNLIKELY(x) __builtin_expect((x), 0) #define UNLIKELY(x) __builtin_expect((x), 0)
#define FD_DECODE_TABLE_DATA
static __attribute__((aligned(16))) const uint16_t _decode_table[] = {
#include <fadec-table.inc>
};
#undef FD_DECODE_TABLE_DATA
// Defines FD_TABLE_OFFSET_32 and FD_TABLE_OFFSET_64, if available // Defines FD_TABLE_OFFSET_32 and FD_TABLE_OFFSET_64, if available
#define FD_DECODE_TABLE_DEFINES #define FD_DECODE_TABLE_DEFINES
#include <fadec-table.inc> #include <fadec-table.inc>
@@ -39,6 +33,11 @@ typedef enum DecodeMode DecodeMode;
static inline unsigned static inline unsigned
table_walk(unsigned cur_idx, unsigned entry_idx, unsigned* out_kind) { table_walk(unsigned cur_idx, unsigned entry_idx, unsigned* out_kind) {
static __attribute__((aligned(16))) const uint16_t _decode_table[] = {
#define FD_DECODE_TABLE_DATA
#include <fadec-table.inc>
#undef FD_DECODE_TABLE_DATA
};
unsigned entry = _decode_table[cur_idx + entry_idx]; unsigned entry = _decode_table[cur_idx + entry_idx];
*out_kind = entry & ENTRY_MASK; *out_kind = entry & ENTRY_MASK;
return (entry & ~ENTRY_MASK) >> 1; return (entry & ~ENTRY_MASK) >> 1;
@@ -181,8 +180,7 @@ decode_modrm(const uint8_t* buffer, int len, DecodeMode mode, FdInstr* instr,
struct InstrDesc struct InstrDesc
{ {
uint16_t type; uint16_t type;
uint8_t operand_indices; uint16_t operand_indices;
uint8_t immediate;
uint16_t operand_sizes; uint16_t operand_sizes;
uint16_t reg_types; uint16_t reg_types;
} __attribute__((packed)); } __attribute__((packed));
@@ -195,11 +193,11 @@ struct InstrDesc
#define DESC_VEXREG_IDX(desc) ((((desc)->operand_indices >> 4) & 3) ^ 3) #define DESC_VEXREG_IDX(desc) ((((desc)->operand_indices >> 4) & 3) ^ 3)
#define DESC_HAS_IMPLICIT(desc) (((desc)->operand_indices & (3 << 6)) != 0) #define DESC_HAS_IMPLICIT(desc) (((desc)->operand_indices & (3 << 6)) != 0)
#define DESC_IMPLICIT_IDX(desc) ((((desc)->operand_indices >> 6) & 3) ^ 3) #define DESC_IMPLICIT_IDX(desc) ((((desc)->operand_indices >> 6) & 3) ^ 3)
#define DESC_IMM_CONTROL(desc) (((desc)->immediate >> 4) & 0x7) #define DESC_IMM_CONTROL(desc) (((desc)->operand_indices >> 12) & 0x7)
#define DESC_IMM_IDX(desc) (((desc)->immediate & 3) ^ 3) #define DESC_IMM_IDX(desc) ((((desc)->operand_indices >> 8) & 3) ^ 3)
#define DESC_IMPLICIT_VAL(desc) (((desc)->immediate >> 2) & 1) #define DESC_IMPLICIT_VAL(desc) (((desc)->operand_indices >> 10) & 1)
#define DESC_LOCK(desc) (((desc)->immediate >> 3) & 1) #define DESC_LOCK(desc) (((desc)->operand_indices >> 11) & 1)
#define DESC_VSIB(desc) (((desc)->immediate >> 7) & 1) #define DESC_VSIB(desc) (((desc)->operand_indices >> 15) & 1)
#define DESC_OPSIZE(desc) (((desc)->operand_sizes >> 8) & 3) #define DESC_OPSIZE(desc) (((desc)->operand_sizes >> 8) & 3)
#define DESC_SIZE_FIX1(desc) (((desc)->operand_sizes >> 10) & 7) #define DESC_SIZE_FIX1(desc) (((desc)->operand_sizes >> 10) & 7)
#define DESC_SIZE_FIX2(desc) (((desc)->operand_sizes >> 13) & 3) #define DESC_SIZE_FIX2(desc) (((desc)->operand_sizes >> 13) & 3)
@@ -391,7 +389,12 @@ prefix_end:
if (UNLIKELY(kind != ENTRY_INSTR)) if (UNLIKELY(kind != ENTRY_INSTR))
return kind == 0 ? FD_ERR_UD : FD_ERR_PARTIAL; return kind == 0 ? FD_ERR_UD : FD_ERR_PARTIAL;
struct InstrDesc* desc = (struct InstrDesc*) &_decode_table[table_idx]; static __attribute__((aligned(16))) const struct InstrDesc descs[] = {
#define FD_DECODE_TABLE_DESCS
#include <fadec-table.inc>
#undef FD_DECODE_TABLE_DESCS
};
const struct InstrDesc* desc = &descs[table_idx >> 2];
instr->type = desc->type; instr->type = desc->type;
instr->flags = prefix_rep == 2 ? FD_FLAG_REP : instr->flags = prefix_rep == 2 ? FD_FLAG_REP :

30
parseinstrs.py
View File

@@ -213,7 +213,7 @@ class EntryKind(Enum):
class TrieEntry(NamedTuple): class TrieEntry(NamedTuple):
kind: EntryKind kind: EntryKind
items: Tuple[Optional[str]] items: Tuple[Optional[str]]
payload: Tuple[Union[int, str]] descidx: Optional[int]
TABLE_LENGTH = { TABLE_LENGTH = {
EntryKind.TABLE256: 256, EntryKind.TABLE256: 256,
@@ -227,19 +227,19 @@ class TrieEntry(NamedTuple):
def table(cls, kind): def table(cls, kind):
return cls(kind, (None,) * cls.TABLE_LENGTH[kind], ()) return cls(kind, (None,) * cls.TABLE_LENGTH[kind], ())
@classmethod @classmethod
def instr(cls, payload): def instr(cls, descidx):
return cls(EntryKind.INSTR, (), payload) return cls(EntryKind.INSTR, (), descidx)
@property @property
def encode_length(self): def encode_length(self):
return len(self.payload) + len(self.items) return len(self.items)
def encode(self, encode_item) -> Tuple[Union[int, str]]: def encode(self, encode_item) -> Tuple[Union[int, str]]:
enc_items = (encode_item(item) if item else 0 for item in self.items) enc_items = (encode_item(item) if item else 0 for item in self.items)
return self.payload + tuple(enc_items) return tuple(enc_items)
def map(self, map_func): def map(self, map_func):
mapped_items = (map_func(i, v) for i, v in enumerate(self.items)) mapped_items = (map_func(i, v) for i, v in enumerate(self.items))
return TrieEntry(self.kind, tuple(mapped_items), self.payload) return TrieEntry(self.kind, tuple(mapped_items), self.descidx)
def update(self, idx, new_val): def update(self, idx, new_val):
return self.map(lambda i, v: new_val if i == idx else v) return self.map(lambda i, v: new_val if i == idx else v)
@@ -347,6 +347,7 @@ class Table:
self.roots = ["root%d"%i for i in range(root_count)] self.roots = ["root%d"%i for i in range(root_count)]
for i in range(root_count): for i in range(root_count):
self.data["root%d"%i] = TrieEntry.table(EntryKind.TABLE_ROOT) self.data["root%d"%i] = TrieEntry.table(EntryKind.TABLE_ROOT)
self.descs = []
self.offsets = {} self.offsets = {}
self.annotations = {} self.annotations = {}
@@ -375,7 +376,10 @@ class Table:
raise Exception("{}, have {}, want {}".format( raise Exception("{}, have {}, want {}".format(
name, self.data[tn].kind, kind)) name, self.data[tn].kind, kind))
self._update_table(tn, opcode[-1][1], name, TrieEntry.instr(instr_encoding)) if instr_encoding not in self.descs:
self.descs.append(instr_encoding)
desc_idx = self.descs.index(instr_encoding)
self._update_table(tn, opcode[-1][1], name, TrieEntry.instr(desc_idx))
def deduplicate(self): def deduplicate(self):
synonyms = True synonyms = True
@@ -395,6 +399,9 @@ class Table:
def calc_offsets(self): def calc_offsets(self):
current = 0 current = 0
for name, entry in self.data.items(): for name, entry in self.data.items():
if entry.kind == EntryKind.INSTR:
self.offsets[name] = entry.descidx << 2
else:
self.annotations[current] = "%s(%d)" % (name, entry.kind.value) self.annotations[current] = "%s(%d)" % (name, entry.kind.value)
self.offsets[name] = current self.offsets[name] = current
current += (entry.encode_length + 3) & ~3 current += (entry.encode_length + 3) & ~3
@@ -406,7 +413,7 @@ class Table:
def compile(self): def compile(self):
self.calc_offsets() self.calc_offsets()
ordered = sorted((off, self.data[k]) for k, off in self.offsets.items()) ordered = sorted((off, self.data[k]) for k, off in self.offsets.items() if self.data[k].encode_length)
data = () data = ()
for off, entry in ordered: for off, entry in ordered:
@@ -414,7 +421,7 @@ class Table:
stats = dict(Counter(entry.kind for entry in self.data.values())) stats = dict(Counter(entry.kind for entry in self.data.values()))
print("%d bytes" % (2*len(data)), stats) print("%d bytes" % (2*len(data)), stats)
return data, self.annotations, [self.offsets[k] for k in self.roots] return data, self.annotations, [self.offsets[k] for k in self.roots], self.descs
def bytes_to_table(data, notes): def bytes_to_table(data, notes):
strdata = tuple(d+"," if type(d) == str else "%#04x,"%d for d in data) strdata = tuple(d+"," if type(d) == str else "%#04x,"%d for d in data)
@@ -425,6 +432,8 @@ def bytes_to_table(data, notes):
template = """// Auto-generated file -- do not modify! template = """// Auto-generated file -- do not modify!
#if defined(FD_DECODE_TABLE_DATA) #if defined(FD_DECODE_TABLE_DATA)
{hex_table} {hex_table}
#elif defined(FD_DECODE_TABLE_DESCS)
{descs}
#elif defined(FD_DECODE_TABLE_STRTAB1) #elif defined(FD_DECODE_TABLE_STRTAB1)
{mnemonic_cstr} {mnemonic_cstr}
#elif defined(FD_DECODE_TABLE_STRTAB2) #elif defined(FD_DECODE_TABLE_STRTAB2)
@@ -611,7 +620,7 @@ if __name__ == "__main__":
table.add_opcode(opcode_path, desc.encode(ign66), i) table.add_opcode(opcode_path, desc.encode(ign66), i)
table.deduplicate() table.deduplicate()
table_data, annotations, root_offsets = table.compile() table_data, annotations, root_offsets, descs = table.compile()
mnemonic_tab = [0] mnemonic_tab = [0]
for name in mnemonics: for name in mnemonics:
@@ -622,6 +631,7 @@ if __name__ == "__main__":
decode_table = template.format( decode_table = template.format(
hex_table=bytes_to_table(table_data, annotations), hex_table=bytes_to_table(table_data, annotations),
descs="\n".join("{{{0},{1},{2},{3}}},".format(*desc) for desc in descs),
mnemonic_cstr=mnemonic_cstr, mnemonic_cstr=mnemonic_cstr,
mnemonic_offsets=",".join(str(off) for off in mnemonic_tab), mnemonic_offsets=",".join(str(off) for off in mnemonic_tab),
defines="\n".join("#define " + line for line in defines), defines="\n".join("#define " + line for line in defines),