diff --git a/src/xbyak/xbyak.h b/src/xbyak/xbyak.h
index 939ffee..63efccd 100644
--- a/src/xbyak/xbyak.h
+++ b/src/xbyak/xbyak.h
@@ -9,14 +9,9 @@
 	@note modified new BSD license
 	http://opensource.org/licenses/BSD-3-Clause
 */
-#if !defined(XBYAK_USE_OP_NAMES) && !defined(XBYAK_NO_OP_NAMES)
+#if (not +0) && !defined(XBYAK_NO_OP_NAMES) // trick to detect whether 'not' is operator or not
 	#define XBYAK_NO_OP_NAMES
 #endif
-#ifndef XBYAK_NO_OP_NAMES
-	#if not +0 // trick to detect whether 'not' is operator or not
-		#error "use -fno-operator-names option if you want to use and(), or(), xor(), not() as function names, Or define XBYAK_NO_OP_NAMES and use and_(), or_(), xor_(), not_()."
-	#endif
-#endif
 
 #include <stdio.h> // for debug print
 #include <assert.h>
@@ -29,7 +24,9 @@
 
 // #define XBYAK_DISABLE_AVX512
 
-//#define XBYAK_USE_MMAP_ALLOCATOR
+#if !defined(XBYAK_USE_MMAP_ALLOCATOR) && !defined(XBYAK_DONT_USE_MMAP_ALLOCATOR)
+	#define XBYAK_USE_MMAP_ALLOCATOR
+#endif
 #if !defined(__GNUC__) || defined(__MINGW32__)
 	#undef XBYAK_USE_MMAP_ALLOCATOR
 #endif
@@ -83,9 +80,12 @@
 	#include <sys/mman.h>
 	#include <stdlib.h>
 #endif
-#if defined(__APPLE__) && defined(MAP_JIT)
+#if defined(__APPLE__) && !defined(XBYAK_DONT_USE_MAP_JIT)
 	#define XBYAK_USE_MAP_JIT
 	#include <sys/sysctl.h>
+	#ifndef MAP_JIT
+		#define MAP_JIT 0x800
+	#endif
 #endif
 #if !defined(_MSC_VER) || (_MSC_VER >= 1600)
 	#include <stdint.h>
@@ -120,7 +120,7 @@ namespace Xbyak {
 
 enum {
 	DEFAULT_MAX_CODE_SIZE = 4096,
-	VERSION = 0x5850 /* 0xABCD = A.BC(D) */
+	VERSION = 0x5912 /* 0xABCD = A.BC(D) */
 };
 
 #ifndef MIE_INTEGER_TYPE_DEFINED
@@ -469,9 +469,8 @@ 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;
 	}
@@ -555,7 +554,7 @@ inline void Operand::setBit(int bit)
 {
 	if (bit != 8 && bit != 16 && bit != 32 && bit != 64 && bit != 128 && bit != 256 && bit != 512) goto ERR;
 	if (isBit(bit)) return;
-	if (is(MEM)) {
+	if (is(MEM | OPMASK)) {
 		bit_ = bit;
 		return;
 	}
@@ -1667,6 +1666,7 @@ 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));
@@ -1780,6 +1780,7 @@ private:
 			db(longCode); dd(disp - longJmpSize);
 		}
 	}
+	bool isNEAR(LabelType type) const { return type == T_NEAR || (type == T_AUTO && isDefaultJmpNEAR_); }
 	template<class T>
 	void opJmp(T& label, LabelType type, uint8 shortCode, uint8 longCode, uint8 longPref)
 	{
@@ -1789,7 +1790,7 @@ private:
 			makeJmp(inner::VerifyInInt32(offset - size_), type, shortCode, longCode, longPref);
 		} else {
 			int jmpSize = 0;
-			if (type == T_NEAR) {
+			if (isNEAR(type)) {
 				jmpSize = 4;
 				if (longPref) db(longPref);
 				db(longCode); dd(0);
@@ -1804,7 +1805,7 @@ private:
 	void opJmpAbs(const void *addr, LabelType type, uint8 shortCode, uint8 longCode, uint8 longPref = 0)
 	{
 		if (isAutoGrow()) {
-			if (type != T_NEAR) throw Error(ERR_ONLY_T_NEAR_IS_SUPPORTED_IN_AUTO_GROW);
+			if (!isNEAR(type)) throw Error(ERR_ONLY_T_NEAR_IS_SUPPORTED_IN_AUTO_GROW);
 			if (size_ + 16 >= maxSize_) growMemory();
 			if (longPref) db(longPref);
 			db(longCode);
@@ -2258,7 +2259,7 @@ public:
 	const Zmm zmm0, zmm1, zmm2, zmm3, zmm4, zmm5, zmm6, zmm7;
 	const Xmm &xm0, &xm1, &xm2, &xm3, &xm4, &xm5, &xm6, &xm7;
 	const Ymm &ym0, &ym1, &ym2, &ym3, &ym4, &ym5, &ym6, &ym7;
-	const Ymm &zm0, &zm1, &zm2, &zm3, &zm4, &zm5, &zm6, &zm7;
+	const Zmm &zm0, &zm1, &zm2, &zm3, &zm4, &zm5, &zm6, &zm7;
 	const Reg32 eax, ecx, edx, ebx, esp, ebp, esi, edi;
 	const Reg16 ax, cx, dx, bx, sp, bp, si, di;
 	const Reg8 al, cl, dl, bl, ah, ch, dh, bh;
@@ -2298,6 +2299,9 @@ public:
 #ifndef XBYAK_DISABLE_SEGMENT
 	const Segment es, cs, ss, ds, fs, gs;
 #endif
+private:
+	bool isDefaultJmpNEAR_;
+public:
 	void L(const std::string& label) { labelMgr_.defineSlabel(label); }
 	void L(Label& label) { labelMgr_.defineClabel(label); }
 	Label L() { Label label; L(label); return label; }
@@ -2317,6 +2321,8 @@ public:
 	void putL(std::string label) { putL_inner(label); }
 	void putL(const Label& label) { putL_inner(label); }
 
+	// set default type of `jmp` of undefined label to T_NEAR
+	void setDefaultJmpNEAR(bool isNear) { isDefaultJmpNEAR_ = isNear; }
 	void jmp(const Operand& op) { opR_ModM(op, BIT, 4, 0xFF, NONE, NONE, true); }
 	void jmp(std::string label, LabelType type = T_AUTO) { opJmp(label, type, 0xEB, 0xE9, 0); }
 	void jmp(const char *label, LabelType type = T_AUTO) { jmp(std::string(label), type); }
@@ -2575,6 +2581,7 @@ public:
 #ifndef XBYAK_DISABLE_SEGMENT
 		, es(Segment::es), cs(Segment::cs), ss(Segment::ss), ds(Segment::ds), fs(Segment::fs), gs(Segment::gs)
 #endif
+		, isDefaultJmpNEAR_(false)
 	{
 		labelMgr_.set(this);
 	}
diff --git a/src/xbyak/xbyak_mnemonic.h b/src/xbyak/xbyak_mnemonic.h
index 457a641..2de6ec2 100644
--- a/src/xbyak/xbyak_mnemonic.h
+++ b/src/xbyak/xbyak_mnemonic.h
@@ -1,4 +1,4 @@
-const char *getVersionString() const { return "5.85"; }
+const char *getVersionString() const { return "5.912"; }
 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); }
@@ -1671,17 +1671,17 @@ void kandnw(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r
 void kandq(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r2, r3, T_L1 | T_0F | T_W1, 0x41); }
 void kandw(const Opmask& r1, const Opmask& r2, const Opmask& r3) { opVex(r1, &r2, r3, T_L1 | T_0F | T_W0, 0x41); }
 void kmovb(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_66 | T_W0, 0x91); }
-void kmovb(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W0, 0x90); }
+void kmovb(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) throw Error(ERR_BAD_COMBINATION); opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W0, 0x90); }
 void kmovb(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_66 | T_W0, 0x92); }
 void kmovb(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_66 | T_W0, 0x93); }
 void kmovd(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_66 | T_W1, 0x91); }
-void kmovd(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W1, 0x90); }
+void kmovd(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) throw Error(ERR_BAD_COMBINATION); opVex(k, 0, op, T_L0 | T_0F | T_66 | T_W1, 0x90); }
 void kmovd(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_F2 | T_W0, 0x92); }
 void kmovd(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_F2 | T_W0, 0x93); }
 void kmovq(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_W1, 0x91); }
-void kmovq(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_W1, 0x90); }
+void kmovq(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) throw Error(ERR_BAD_COMBINATION); opVex(k, 0, op, T_L0 | T_0F | T_W1, 0x90); }
 void kmovw(const Address& addr, const Opmask& k) { opVex(k, 0, addr, T_L0 | T_0F | T_W0, 0x91); }
-void kmovw(const Opmask& k, const Operand& op) { opVex(k, 0, op, T_L0 | T_0F | T_W0, 0x90); }
+void kmovw(const Opmask& k, const Operand& op) { if (!op.isMEM() && !op.isOPMASK()) throw Error(ERR_BAD_COMBINATION); opVex(k, 0, op, T_L0 | T_0F | T_W0, 0x90); }
 void kmovw(const Opmask& k, const Reg32& r) { opVex(k, 0, r, T_L0 | T_0F | T_W0, 0x92); }
 void kmovw(const Reg32& r, const Opmask& k) { opVex(r, 0, k, T_L0 | T_0F | T_W0, 0x93); }
 void knotb(const Opmask& r1, const Opmask& r2) { opVex(r1, 0, r2, T_0F | T_66 | T_W0, 0x44); }
diff --git a/src/xbyak/xbyak_util.h b/src/xbyak/xbyak_util.h
index eefd152..4f79d8f 100644
--- a/src/xbyak/xbyak_util.h
+++ b/src/xbyak/xbyak_util.h
@@ -1,5 +1,6 @@
 #ifndef XBYAK_XBYAK_UTIL_H_
 #define XBYAK_XBYAK_UTIL_H_
+#include <string.h>
 
 /**
 	utility class and functions for Xbyak
@@ -146,6 +147,11 @@ class Cpu {
 					numCores_[level - 1] = extractBit(data[1], 0, 15);
 				}
 			}
+			/*
+				Fallback values in case a hypervisor has 0xB leaf zeroed-out.
+			*/
+			numCores_[SmtLevel - 1] = (std::max)(1u, numCores_[SmtLevel - 1]);
+			numCores_[CoreLevel - 1] = (std::max)(numCores_[SmtLevel - 1], numCores_[CoreLevel - 1]);
 		} else {
 			/*
 				Failed to deremine num of cores without x2APIC support.
@@ -754,7 +760,7 @@ public:
 	};
 	Profiler()
 		: mode_(None)
-		, suffix_(0)
+		, suffix_("")
 		, startAddr_(0)
 #ifdef XBYAK_USE_PERF
 		, fp_(0)
@@ -828,7 +834,16 @@ public:
 #ifdef XBYAK_USE_PERF
 		if (mode_ == Perf) {
 			if (fp_ == 0) return;
-			fprintf(fp_, "%llx %zx %s%s\n", (long long)startAddr, funcSize, funcName, suffix_);
+			fprintf(fp_, "%llx %zx %s%s", (long long)startAddr, funcSize, funcName, suffix_);
+			/*
+				perf does not recognize the function name which is less than 3,
+				so append '_' at the end of the name if necessary
+			*/
+			size_t n = strlen(funcName) + strlen(suffix_);
+			for (size_t i = n; i < 3; i++) {
+				fprintf(fp_, "_");
+			}
+			fprintf(fp_, "\n");
 			fflush(fp_);
 		}
 #endif