fix the encoding of vinsertps for disp8N
diff --git a/gen/gen_code.cpp b/gen/gen_code.cpp
index 37877bf..70db4fe 100644
--- a/gen/gen_code.cpp
+++ b/gen/gen_code.cpp
@@ -76,7 +76,7 @@
{ 0xC2, "cmpss", T_0F | T_F3, true, true, 2 },
{ 0x5A, "cvtsd2ss", T_0F | T_F2 | T_EVEX | T_EW1 | T_N8 | T_ER_X, false, true, 2 },
{ 0x5A, "cvtss2sd", T_0F | T_F3 | T_EVEX | T_EW0 | T_N4 | T_SAE_X, false, true, 2 },
- { 0x21, "insertps", T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0, true, true, 2 },
+ { 0x21, "insertps", T_0F3A | T_66 | T_W0 | T_EVEX | T_EW0 | T_N4, true, true, 2 },
{ 0x63, "packsswb", T_0F | T_66 | T_YMM | T_EVEX, false, true, 2 },
{ 0x6B, "packssdw", T_0F | T_66 | T_YMM | T_EVEX | T_EW0 | T_B32, false, true, 2 },
{ 0x67, "packuswb", T_0F | T_66 | T_YMM | T_EVEX, false, true, 2 },
diff --git a/readme.md b/readme.md
index 1604b06..0c622e8 100644
--- a/readme.md
+++ b/readme.md
@@ -392,6 +392,7 @@
http://opensource.org/licenses/BSD-3-Clause
## History
+* 2018/Sep/19 ver 5.72 fix the encoding of vinsertps for disp8N(Thanks to petercaday)
* 2018/Sep/04 ver 5.71 L() returns a new label instance
* 2018/Aug/27 ver 5.70 support setProtectMode() and DontUseProtect for read/exec setting
* 2018/Aug/24 ver 5.68 fix wrong VSIB encoding with vector index >= 16(thanks to petercaday)
diff --git a/readme.txt b/readme.txt
index b9746c8..3b43f1a 100644
--- a/readme.txt
+++ b/readme.txt
@@ -1,5 +1,5 @@
- C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 5.71
+ C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 5.72
-----------------------------------------------------------------------------
◎概要
@@ -373,6 +373,7 @@
-----------------------------------------------------------------------------
◎履歴
+2018/09/19 ver 5.72 fix the encoding of vinsertps for disp8N(Thanks to petercaday)
2018/08/27 ver 5.71 新しいlabelインスタンスを返すL()を追加
2018/08/27 ver 5.70 read/exec設定のためのsetProtectMode()とDontUseProtectの追加
2018/08/24 ver 5.68 indexが16以上のVSIBエンコーディングのバグ修正(thanks to petercaday)
diff --git a/test/make_512.cpp b/test/make_512.cpp
index cf345be..38f989c 100644
--- a/test/make_512.cpp
+++ b/test/make_512.cpp
@@ -1176,7 +1176,7 @@
uint64_t xm;
} tbl[] = {
#ifdef XBYAK64
- { "vinsertps", _XMM, _XMM, _XMM3 | _MEM },
+ { "vinsertps", _XMM3, _XMM, _XMM3 | _MEM },
{ "vshufpd", XMM_KZ, _XMM, M_1to2 | _MEM },
{ "vshufpd", ZMM_KZ, _ZMM, M_1to8 | _MEM },
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index d820b6d..cf52e32 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -105,7 +105,7 @@
enum {
DEFAULT_MAX_CODE_SIZE = 4096,
- VERSION = 0x5710 /* 0xABCD = A.BC(D) */
+ VERSION = 0x5720 /* 0xABCD = A.BC(D) */
};
#ifndef MIE_INTEGER_TYPE_DEFINED
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index 8eadcc5..6111619 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -1,4 +1,4 @@
-const char *getVersionString() const { return "5.71"; }
+const char *getVersionString() const { return "5.72"; }
void adc(const Operand& op, uint32 imm) { opRM_I(op, imm, 0x10, 2); }
void adc(const Operand& op1, const Operand& op2) { opRM_RM(op1, op2, 0x10); }
void adcx(const Reg32e& reg, const Operand& op) { opGen(reg, op, 0xF6, 0x66, isREG32_REG32orMEM, NONE, 0x38); }
@@ -1023,7 +1023,7 @@
void vhsubps(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_F2 | T_0F | T_YMM, 0x7D); }
void vinsertf128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x18, imm); }
void vinserti128(const Ymm& y1, const Ymm& y2, const Operand& op, uint8 imm) { if (!(y1.isYMM() && y2.isYMM() && op.isXMEM())) throw Error(ERR_BAD_COMBINATION); opVex(y1, &y2, op, T_0F3A | T_66 | T_W0 | T_YMM, 0x38, imm); }
-void vinsertps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_66 | T_0F3A | T_W0 | T_EW0 | T_EVEX, 0x21, imm); }
+void vinsertps(const Xmm& x1, const Xmm& x2, const Operand& op, uint8 imm) { opAVX_X_X_XM(x1, x2, op, T_N4 | T_66 | T_0F3A | T_W0 | T_EW0 | T_EVEX, 0x21, imm); }
void vlddqu(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, cvtIdx0(x), addr, T_0F | T_F2 | T_W0 | T_YMM, 0xF0); }
void vldmxcsr(const Address& addr) { opAVX_X_X_XM(xm2, xm0, addr, T_0F, 0xAE); }
void vmaskmovdqu(const Xmm& x1, const Xmm& x2) { opAVX_X_X_XM(x1, xm0, x2, T_0F | T_66, 0xF7); }
