yuzu/xbyak
1
0
Fork 0
mirror of https://github.com/herumi/xbyak synced 2024-11-20 16:06:14 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

unify opAESKL and opSHA

Browse source
This commit is contained in:
MITSUNARI Shigeo 2023-12-20 15:52:42 +09:00
parent bfd14244ad
commit 3427be2988
3 changed files with 22 additions and 31 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
gen/gen_code.cpp
View file

@ -1433,9 +1433,9 @@ void put()
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl *p = &tbl[i];
printf("void %s(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0x%02X, 0x%02X); }\n", p->name, p->code, p->code2);
printf("void %s(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0x%02X, T_MUST_EVEX, 0x%02X); }\n", p->name, p->code, p->code2);
}
puts("void sha1rnds4(const Xmm& x, const Operand& op, uint8_t imm) { opSHA(x, op, T_0F3A, 0xCC, 0xD4, imm); }");
puts("void sha1rnds4(const Xmm& x, const Operand& op, uint8_t imm) { opSSE_APX(x, op, T_0F3A, 0xCC, T_MUST_EVEX, 0xD4, imm); }");
}
// (m, x), (m, y)
{
@ -2048,9 +2048,9 @@ void put64()
std::string s1 = type2String(p->type1);
std::string s2 = type2String(p->type2);
if (p->idx == 8) {
printf("void %s(const Xmm& x, const Address& addr) { opAESKL(&x, addr, %s, %s, 0x%02X); }\n", p->name, s1.c_str(), s2.c_str(), p->code);
printf("void %s(const Xmm& x, const Address& addr) { opSSE_APX(x, addr, %s, 0x%02X, %s, 0x%02X); }\n", p->name, s1.c_str(), p->code, s2.c_str(), p->code);
} else {
printf("void %s(const Address& addr) { opAESKL(&xmm%d, addr, %s, %s, 0x%02X); }\n", p->name, p->idx, s1.c_str(), s2.c_str(), p->code);
printf("void %s(const Address& addr) { opSSE_APX(xmm%d, addr, %s, 0x%02X, %s, 0x%02X); }\n", p->name, p->idx, s1.c_str(), p->code, s2.c_str(), p->code);
}
}
}

15
xbyak/xbyak.h
View file

@ -2738,15 +2738,6 @@ private:
if (opROO(Reg(), *p2, *p1, T_MAP1|type, code)) return;
opVex(static_cast<const Reg&>(*p1), 0, *p2, T_L0|T_0F|type, code);
}
void opAESKL(const Xmm *x, const Address& addr, uint64_t type1, uint64_t type2, uint8_t code)
{
if (x && x->getIdx() >= 16) XBYAK_THROW(ERR_INVALID_REG_IDX)
if (addr.hasRex2()) {
opROO(Reg(), addr, *x, type2, code);
return;
}
opRO(*x, addr, type1, code);
}
void opEncodeKey(const Reg32& r1, const Reg32& r2, uint8_t code1, uint8_t code2)
{
if (r1.getIdx() < 8 && r2.getIdx() < 8) {
@ -2755,13 +2746,13 @@ private:
}
opROO(Reg(), r2, r1, T_MUST_EVEX|T_F3, code2);
}
void opSHA(const Xmm& x, const Operand& op, uint64_t type, uint8_t code1, uint8_t code2, int imm = NONE)
void opSSE_APX(const Xmm& x, const Operand& op, uint64_t type1, uint8_t code1, uint64_t type2, uint8_t code2, int imm = NONE)
{
if (x.getIdx() <= 15 && op.hasRex2() && opROO(Reg(), op, x, T_MUST_EVEX, code2, imm != NONE ? 1 : 0)) {
if (x.getIdx() <= 15 && op.hasRex2() && opROO(Reg(), op, x, type2, code2, imm != NONE ? 1 : 0)) {
if (imm != NONE) db(imm);
return;
}
opSSE(x, op, type, code1, isXMM_XMMorMEM, imm);
opSSE(x, op, type1, code1, isXMM_XMMorMEM, imm);
}
public:
unsigned int getVersion() const { return VERSION; }

30
xbyak/xbyak_mnemonic.h
View file

@ -988,13 +988,13 @@ void setpo(const Operand& op) { if (opROO(Reg(), op, Reg(), T_APX|T_ZU|T_F2, 0x4
void sets(const Operand& op) { if (opROO(Reg(), op, Reg(), T_APX|T_ZU|T_F2, 0x40 | 8)) return; opRext(op, 8, 0, T_0F, 0x90 | 8); }//-V524
void setz(const Operand& op) { if (opROO(Reg(), op, Reg(), T_APX|T_ZU|T_F2, 0x40 | 4)) return; opRext(op, 8, 0, T_0F, 0x90 | 4); }//-V524
void sfence() { db(0x0F); db(0xAE); db(0xF8); }
void sha1msg1(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xC9, 0xD9); }
void sha1msg2(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xCA, 0xDA); }
void sha1nexte(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xC8, 0xD8); }
void sha1rnds4(const Xmm& x, const Operand& op, uint8_t imm) { opSHA(x, op, T_0F3A, 0xCC, 0xD4, imm); }
void sha256msg1(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xCC, 0xDC); }
void sha256msg2(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xCD, 0xDD); }
void sha256rnds2(const Xmm& x, const Operand& op) { opSHA(x, op, T_0F38, 0xCB, 0xDB); }
void sha1msg1(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xC9, T_MUST_EVEX, 0xD9); }
void sha1msg2(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xCA, T_MUST_EVEX, 0xDA); }
void sha1nexte(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xC8, T_MUST_EVEX, 0xD8); }
void sha1rnds4(const Xmm& x, const Operand& op, uint8_t imm) { opSSE_APX(x, op, T_0F3A, 0xCC, T_MUST_EVEX, 0xD4, imm); }
void sha256msg1(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xCC, T_MUST_EVEX, 0xDC); }
void sha256msg2(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xCD, T_MUST_EVEX, 0xDD); }
void sha256rnds2(const Xmm& x, const Operand& op) { opSSE_APX(x, op, T_0F38, 0xCB, T_MUST_EVEX, 0xDB); }
void shl(const Operand& op, const Reg8& _cl) { opShift(op, _cl, 12); }
void shl(const Operand& op, int imm) { opShift(op, imm, 12); }
void shl(const Reg& d, const Operand& op, const Reg8& _cl) { opShift(op, _cl, 12, &d); }
@ -1926,14 +1926,14 @@ void cmpoxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opRRO(r
void cmppxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opRRO(r1, r2, addr, T_APX|T_66|T_0F38, 0xEA); }
void cmpsxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opRRO(r1, r2, addr, T_APX|T_66|T_0F38, 0xE8); }
void cmpzxadd(const Address& addr, const Reg32e& r1, const Reg32e& r2) { opRRO(r1, r2, addr, T_APX|T_66|T_0F38, 0xE4); }
void aesdec128kl(const Xmm& x, const Address& addr) { opAESKL(&x, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xDD); }
void aesdec256kl(const Xmm& x, const Address& addr) { opAESKL(&x, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xDF); }
void aesdecwide128kl(const Address& addr) { opAESKL(&xmm1, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xD8); }
void aesdecwide256kl(const Address& addr) { opAESKL(&xmm3, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xD8); }
void aesenc128kl(const Xmm& x, const Address& addr) { opAESKL(&x, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xDC); }
void aesenc256kl(const Xmm& x, const Address& addr) { opAESKL(&x, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xDE); }
void aesencwide128kl(const Address& addr) { opAESKL(&xmm0, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xD8); }
void aesencwide256kl(const Address& addr) { opAESKL(&xmm2, addr, T_F3|T_0F38, T_F3|T_MUST_EVEX, 0xD8); }
void aesdec128kl(const Xmm& x, const Address& addr) { opSSE_APX(x, addr, T_F3|T_0F38, 0xDD, T_F3|T_MUST_EVEX, 0xDD); }
void aesdec256kl(const Xmm& x, const Address& addr) { opSSE_APX(x, addr, T_F3|T_0F38, 0xDF, T_F3|T_MUST_EVEX, 0xDF); }
void aesdecwide128kl(const Address& addr) { opSSE_APX(xmm1, addr, T_F3|T_0F38, 0xD8, T_F3|T_MUST_EVEX, 0xD8); }
void aesdecwide256kl(const Address& addr) { opSSE_APX(xmm3, addr, T_F3|T_0F38, 0xD8, T_F3|T_MUST_EVEX, 0xD8); }
void aesenc128kl(const Xmm& x, const Address& addr) { opSSE_APX(x, addr, T_F3|T_0F38, 0xDC, T_F3|T_MUST_EVEX, 0xDC); }
void aesenc256kl(const Xmm& x, const Address& addr) { opSSE_APX(x, addr, T_F3|T_0F38, 0xDE, T_F3|T_MUST_EVEX, 0xDE); }
void aesencwide128kl(const Address& addr) { opSSE_APX(xmm0, addr, T_F3|T_0F38, 0xD8, T_F3|T_MUST_EVEX, 0xD8); }
void aesencwide256kl(const Address& addr) { opSSE_APX(xmm2, addr, T_F3|T_0F38, 0xD8, T_F3|T_MUST_EVEX, 0xD8); }
void encodekey128(const Reg32& r1, const Reg32& r2) { opEncodeKey(r1, r2, 0xFA, 0xDA); }
void encodekey256(const Reg32& r1, const Reg32& r2) { opEncodeKey(r1, r2, 0xFB, 0xDB); }
void ldtilecfg(const Address& addr) { if (opROO(Reg(), addr, tmm0, T_APX|T_0F38|T_W0, 0x49)) return; opVex(tmm0, &tmm0, addr, T_0F38|T_W0, 0x49); }