accept k0 mask register (it means no mask)

7 files changed, 24 insertions, 8 deletions

diff --git a/readme.md b/readme.md
index 5c61387..139ebca 100644
--- a/readme.md
+++ b/readme.md
@@ -1,6 +1,6 @@
 [![Build Status](https://travis-ci.org/herumi/xbyak.png)](https://travis-ci.org/herumi/xbyak)
 
-# Xbyak 5.90 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
+# Xbyak 5.91 ; JIT assembler for x86(IA32), x64(AMD64, x86-64) by C++
 
 ## Abstract
 
@@ -151,6 +151,8 @@ vfpclassps k5{k3}, [rax+64]{1to4}, 5    --> vfpclassps(k5|k3, yword_b [rax+64],
 ```
 ### Remark
 * `k1`, ..., `k7` are opmask registers.
+  - `k0` is dealt as no mask.
+  - e.g. `vmovaps(zmm0|k0, ptr[rax]);` and `vmovaps(zmm0|T_z, ptr[rax]);` are same to `vmovaps(zmm0, ptr[rax]);`.
 * use `| T_z`, `| T_sae`, `| T_rn_sae`, `| T_rd_sae`, `| T_ru_sae`, `| T_rz_sae` instead of `,{z}`, `,{sae}`, `,{rn-sae}`, `,{rd-sae}`, `,{ru-sae}`, `,{rz-sae}` respectively.
 * `k4 | k3` is different from `k3 | k4`.
 * use `ptr_b` for broadcast `{1toX}`. X is automatically determined.
@@ -423,6 +425,7 @@ modified new BSD License
 http://opensource.org/licenses/BSD-3-Clause
 
 ## History
+* 2020/Apr/20 ver 5.91 accept mask register k0 (it means no mask)
 * 2020/Apr/09 ver 5.90 kmov{b,d,w,q} throws exception for an unsupported register
 * 2020/Feb/26 ver 5.891 fix typo of type
 * 2020/Jan/03 ver 5.89 fix error of vfpclasspd
diff --git a/readme.txt b/readme.txt
index 981b295..042826c 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 5.90

+    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 5.91

 

 -----------------------------------------------------------------------------

 ◎概要

@@ -371,6 +371,7 @@ sample/{echo,hello}.bfは http://www.kmonos.net/alang/etc/brainfuck.php から
 -----------------------------------------------------------------------------

 ◎履歴

 

+2020/04/20 ver 5.91 マスクレジスタk0を受け入れる(マスクをしない)

 2020/04/09 ver 5.90 kmov{b,w,d,q}がサポートされないレジスタを受けると例外を投げる

 2020/02/26 ver 5.891 zm0のtype修正

 2020/01/03 ver 5.89 vfpclasspdの処理エラー修正

diff --git a/test/make_512.cpp b/test/make_512.cpp
index d29adeb..004fa16 100644
--- a/test/make_512.cpp
+++ b/test/make_512.cpp
@@ -537,7 +537,7 @@ public:
 			printf("vaddpd(%s%s%s, %s, %s%s); dump();\n", r1, pk, pz, r2, r3, saeTblXbyak[sae]);
 		} else {
 			if (kIdx) CYBOZU_SNPRINTF(pk, sizeof(pk), "{k%d}", kIdx);
-			if (z) pz = "{z}";
+			if (z && kIdx) pz = "{z}";
 			printf("vaddpd %s%s%s, %s, %s%s\n", r1, pk, pz, r2, r3, saeTblNASM[sae]);
 		}
 	}
diff --git a/test/make_nm.cpp b/test/make_nm.cpp
index 67c523b..fc46eb6 100644
--- a/test/make_nm.cpp
+++ b/test/make_nm.cpp
@@ -2628,7 +2628,7 @@ public:
 			printf("vaddpd(%s%s%s, %s, %s%s); dump();\n", r1, pk, pz, r2, r3, saeTblXbyak[sae]);
 		} else {
 			if (kIdx) CYBOZU_SNPRINTF(pk, sizeof(pk), "{k%d}", kIdx);
-			if (z) pz = "{z}";
+			if (z && kIdx) pz = "{z}";
 			printf("vaddpd %s%s%s, %s, %s%s\n", r1, pk, pz, r2, r3, saeTblNASM[sae]);
 		}
 	}
diff --git a/test/misc.cpp b/test/misc.cpp
index 4a769a3..23a6a2b 100644
--- a/test/misc.cpp
+++ b/test/misc.cpp
@@ -119,6 +119,18 @@ CYBOZU_TEST_AUTO(kmask)
 			CYBOZU_TEST_EXCEPTION(kmovd(k1, rax), std::exception);
 			CYBOZU_TEST_NO_EXCEPTION(kmovq(k1, rax));
 #endif
+			CYBOZU_TEST_NO_EXCEPTION(vmovaps(xm0|k0, ptr[eax]));
+			checkT_z();
+		}
+		void checkT_z()
+		{
+			const uint8_t *p1 = getCurr();
+			vmovaps(zm0, ptr[eax]);
+			const uint8_t *p2 = getCurr();
+			vmovaps(zm0|T_z, ptr[eax]);
+			const uint8_t *end = getCurr();
+			CYBOZU_TEST_EQUAL(p2 - p1, end - p2);
+			CYBOZU_TEST_EQUAL_ARRAY(p1, p2, end - p2);
 		}
 	} c;
 }
diff --git a/xbyak/xbyak.h b/xbyak/xbyak.h
index 81ffce6..41409ec 100644
--- a/xbyak/xbyak.h
+++ b/xbyak/xbyak.h
@@ -115,7 +115,7 @@ namespace Xbyak {
 
 enum {
 	DEFAULT_MAX_CODE_SIZE = 4096,
-	VERSION = 0x5900 /* 0xABCD = A.BC(D) */
+	VERSION = 0x5910 /* 0xABCD = A.BC(D) */
 };
 
 #ifndef MIE_INTEGER_TYPE_DEFINED
@@ -464,9 +464,8 @@ public:
 	}
 	// err if MMX/FPU/OPMASK/BNDREG
 	void setBit(int bit);
-	void setOpmaskIdx(int idx, bool ignore_idx0 = false)
+	void setOpmaskIdx(int idx, bool /*ignore_idx0*/ = true)
 	{
-		if (!ignore_idx0 && idx == 0) throw Error(ERR_K0_IS_INVALID);
 		if (mask_) throw Error(ERR_OPMASK_IS_ALREADY_SET);
 		mask_ = idx;
 	}
@@ -1662,6 +1661,7 @@ private:
 		bool Vp = !((v ? v->isExtIdx2() : 0) | Hi16Vidx);
 		bool z = reg.hasZero() || base.hasZero() || (v ? v->hasZero() : false);
 		if (aaa == 0) aaa = verifyDuplicate(base.getOpmaskIdx(), reg.getOpmaskIdx(), (v ? v->getOpmaskIdx() : 0), ERR_OPMASK_IS_ALREADY_SET);
+		if (aaa == 0) z = 0; // clear T_z if mask is not set
 		db(0x62);
 		db((R ? 0x80 : 0) | (X ? 0x40 : 0) | (B ? 0x20 : 0) | (Rp ? 0x10 : 0) | (mm & 3));
 		db((w == 1 ? 0x80 : 0) | ((vvvv & 15) << 3) | 4 | (pp & 3));
diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
index c9cc9af..f0ecb14 100644
--- a/xbyak/xbyak_mnemonic.h
+++ b/xbyak/xbyak_mnemonic.h
@@ -1,4 +1,4 @@
-const char *getVersionString() const { return "5.90"; }
+const char *getVersionString() const { return "5.91"; }
 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); }