fix vpblendvb
diff --git a/gen/gen_code.cpp b/gen/gen_code.cpp
index bf79542..fbbcd67 100644
--- a/gen/gen_code.cpp
+++ b/gen/gen_code.cpp
@@ -1542,14 +1542,19 @@
}
// 4-op
{
- printf("void vblendvpd(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4B, true, -1, x4.getIdx() << 4); }\n");
- printf("void vblendvpd(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4B, true, -1, x4.getIdx() << 4); }\n");
-
- printf("void vblendvps(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4A, true, -1, x4.getIdx() << 4); }\n");
- printf("void vblendvps(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4A, true, -1, x4.getIdx() << 4); }\n");
-
- printf("void vpblendvb(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4C, false, -1, x4.getIdx() << 4); }\n");
- printf("void vpblendvb(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4C, false, -1, x4.getIdx() << 4); }\n");
+ const struct Tbl {
+ const char *name;
+ uint8 code;
+ } tbl[] = {
+ { "vblendvpd", 0x4B },
+ { "vblendvps", 0x4A },
+ { "vpblendvb", 0x4C },
+ };
+ for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
+ const Tbl& p = tbl[i];
+ printf("void %s(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x%02X, true, -1, x4.getIdx() << 4); }\n", p.name, p.code);
+ printf("void %s(const Xmm& x1, const Operand& op, const Xmm& x4) { %s(x1, x1, op, x4); }\n", p.name, p.name);
+ }
}
// mov
{
diff --git a/readme.md b/readme.md
index 8931aef..bb5d2f0 100644
--- a/readme.md
+++ b/readme.md
@@ -1,5 +1,5 @@
-Xbyak 4.88 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
+Xbyak 4.89 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
=============
Abstract
@@ -285,6 +285,7 @@
History
-------------
+* 2016/Jan/30 ver 4.89 vpblendvb supports ymm reg(thanks to John Funnell)
* 2016/Jan/24 ver 4.88 lea, cmov supports 16-bit register(thanks to whyisthisfieldhere)
* 2015/Oct/05 ver 4.87 support segment selectors
* 2015/Aug/18 ver 4.86 fix [rip + label] addressing with immediate value(thanks to whyisthisfieldhere)
diff --git a/readme.txt b/readme.txt
index e488668..8cd2e64 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 4.88
+ C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 4.89
-----------------------------------------------------------------------------
◎概要
@@ -301,6 +301,7 @@
-----------------------------------------------------------------------------
◎履歴
+2016/01/30 ver 4.89 vpblendvbがymmレジスタをサポートしていなかった(thanks to John Funnell)
2016/01/24 ver 4.88 lea, cmovの16bitレジスタ対応(thanks to whyisthisfieldhere)
2015/08/16 ver 4.87 セグメントセレクタに対応
2015/08/16 ver 4.86 [rip + label]アドレッシングで即値を使うと壊れる(thanks to whyisthisfieldhere)
diff --git a/test/make_nm.cpp b/test/make_nm.cpp
index 4a1cc03..05fef8f 100644
--- a/test/make_nm.cpp
+++ b/test/make_nm.cpp
@@ -1805,7 +1805,7 @@
} tbl[] = {
{ "vblendvpd", true },
{ "vblendvps", true },
- { "vpblendvb", false },
+ { "vpblendvb", true },
};
for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
const Tbl& p = tbl[i];
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index 5e7fc17..49c8f53 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -101,7 +101,7 @@
enum {
DEFAULT_MAX_CODE_SIZE = 4096,
- VERSION = 0x4880 /* 0xABCD = A.BC(D) */
+ VERSION = 0x4890 /* 0xABCD = A.BC(D) */
};
#ifndef MIE_INTEGER_TYPE_DEFINED
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index 376c4c8..068fef4 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -1389,11 +1389,11 @@
void vpsrlq(const Xmm& x1, const Xmm& x2, uint8 imm) { opAVX_X_X_XM(x1.isYMM() ? ym2 : xm2, x1, x2, MM_0F | PP_66, 0x73, true, -1, imm); }
void vpsrlq(const Xmm& x, uint8 imm) { opAVX_X_X_XM(x.isYMM() ? ym2 : xm2, x, x, MM_0F | PP_66, 0x73, true, -1, imm); }
void vblendvpd(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4B, true, -1, x4.getIdx() << 4); }
-void vblendvpd(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4B, true, -1, x4.getIdx() << 4); }
+void vblendvpd(const Xmm& x1, const Operand& op, const Xmm& x4) { vblendvpd(x1, x1, op, x4); }
void vblendvps(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4A, true, -1, x4.getIdx() << 4); }
-void vblendvps(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4A, true, -1, x4.getIdx() << 4); }
-void vpblendvb(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4C, false, -1, x4.getIdx() << 4); }
-void vpblendvb(const Xmm& x1, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x1, op, MM_0F3A | PP_66, 0x4C, false, -1, x4.getIdx() << 4); }
+void vblendvps(const Xmm& x1, const Operand& op, const Xmm& x4) { vblendvps(x1, x1, op, x4); }
+void vpblendvb(const Xmm& x1, const Xmm& x2, const Operand& op, const Xmm& x4) { opAVX_X_X_XM(x1, x2, op, MM_0F3A | PP_66, 0x4C, true, -1, x4.getIdx() << 4); }
+void vpblendvb(const Xmm& x1, const Operand& op, const Xmm& x4) { vpblendvb(x1, x1, op, x4); }
void vmovd(const Xmm& x, const Reg32& reg) { opAVX_X_X_XM(x, xm0, Xmm(reg.getIdx()), MM_0F | PP_66, 0x6E, false, 0); }
void vmovd(const Xmm& x, const Address& addr) { opAVX_X_X_XM(x, xm0, addr, MM_0F | PP_66, 0x6E, false, 0); }
void vmovd(const Reg32& reg, const Xmm& x) { opAVX_X_X_XM(x, xm0, Xmm(reg.getIdx()), MM_0F | PP_66, 0x7E, false, 0); }
