Squashed 'externals/xbyak/' changes from 913099fc8..88f2f771f

88f2f771f Merge branch 'dev'
8623616e0 v6.61
1b911598d mingw uses __thread instead of __declspec(thread)
d224701f4 fix movzx eax, ah on 64bit and test it
1c35e34ab Merge branch 'dev'
4fb0e58aa test of cpuid
913c798ff update doc
8f6dee319 v6.60.2
21419721e Merge branch 'Wunkolo-standalone-cpuid' into dev
1b2e7c7da Fix detection of `GFNI`,`VAES`, and `VPCLMULQDQ`

git-subtree-dir: externals/xbyak
git-subtree-split: 88f2f771f179a3d856ecd0b07a20598bd06ba21d

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 2.6...3.0.2)
 
-project(xbyak LANGUAGES CXX VERSION 6.60.1)
+project(xbyak LANGUAGES CXX VERSION 6.61)
 
 file(GLOB headers xbyak/*.h)
 

doc/changelog.md

@@ -1,5 +1,7 @@
 # History
 
+* 2022/Jul/29 ver 6.61 fix exception of movzx eax, ah in 64-bit mode
+* 2022/Jun/16 ver 6.60.2 fix detection of GFNI, VAES, and VPCLMULQDQ
 * 2022/Jun/15 ver 6.60.1 fix link error of Xbyak::util::Cpu on Visual Studio with /O0 option
 * 2022/Jun/06 ver 6.60 change the version format to avoid it going backward
 * 2022/Jun/01 ver 6.06 refactor Cpu::Type class and improve MmapAllocator when XBYAK_USE_MEMFD is defined.

meson.build

@@ -5,7 +5,7 @@
 project(
 	'xbyak',
 	'cpp',
-	version: '6.60.1',
+	version: '6.61',
 	license: 'BSD-3-Clause',
 	default_options: 'b_ndebug=if-release'
 )

readme.md

@@ -1,5 +1,5 @@
 
-# Xbyak 6.60.1 [![Badge Build]][Build Status]
+# Xbyak 6.61 [![Badge Build]][Build Status]
 
 *A C++ JIT assembler for x86 (IA32), x64 (AMD64, x86-64)*
 

readme.txt

@@ -1,5 +1,5 @@
 
-    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 6.60.1
+    C++用x86(IA-32), x64(AMD64, x86-64) JITアセンブラ Xbyak 6.61
 
 -----------------------------------------------------------------------------
 ◎概要
@@ -400,6 +400,8 @@ sample/{echo,hello}.bfは http://www.kmonos.net/alang/etc/brainfuck.php から
 -----------------------------------------------------------------------------
 ◎履歴
 
+2022/07/29 ver 6.61 movzx eax, ahがエラーになるのを修正
+2022/06/16 ver 6.60.2 GFNI, VAES, VPCLMULQDQの判定修正
 2022/06/15 ver 6.60.1 Visual Studio /O0でXbyak::util::Cpuがリンクエラーになるのに対応
 2022/06/06 ver 6.60 バージョンのつけ方を数値が戻らないように変更
 2022/06/01 ver 6.06 Cpu::TypeクラスのリファクタリングとXBYAK_USE_MEMFDが定義されたときのMmapAllocatorの改善

sample/cpuid/adl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx_vnni waitpkg clflushopt cldemote movdiri movdir64b

sample/cpuid/bdw.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap f16c movbe

sample/cpuid/clx.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap f16c movbe avx512f avx512dq avx512cd avx512bw avx512vl avx512_vnni clflushopt

sample/cpuid/cnl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi clflushopt

sample/cpuid/cpuid.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+make -C ../ test_util64
+
+cpus=(p4p mrm pnr nhm wsm snb ivb hsw bdw slt slm glm glp tnt skl cnl icl skx clx cpx icx knl knm tgl adl spr)
+for cpu in ${cpus[@]} ; do
+  echo $cpu
+  ~/bin/sde -$cpu -- ../test_util64 -cpuid > tmp.txt
+  diff tmp.txt $cpu.txt
+done
+

sample/cpuid/cpx.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap f16c movbe avx512f avx512dq avx512cd avx512bw avx512vl avx512_vnni avx512_bf16 clflushopt

sample/cpuid/glm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq enh_rep rdrand rdseed smap sha movbe clflushopt

sample/cpuid/glp.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq enh_rep rdrand rdseed smap sha movbe clflushopt

sample/cpuid/hsw.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt enh_rep rdrand f16c movbe

sample/cpuid/icl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq clflushopt

sample/cpuid/icx.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq clflushopt

sample/cpuid/ivb.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx enh_rep rdrand f16c

sample/cpuid/knl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed prefetchwt1 f16c movbe avx512f avx512pf avx512er avx512cd

sample/cpuid/knm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed prefetchwt1 f16c movbe avx512f avx512pf avx512er avx512cd avx512_4vnniw avx512_4fmaps avx512_vpopcntdq

sample/cpuid/mrm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3

sample/cpuid/nhm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt rdtscp

sample/cpuid/p4p.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3

sample/cpuid/pnr.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41

sample/cpuid/skl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap f16c movbe clflushopt

sample/cpuid/skx.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap f16c movbe avx512f avx512dq avx512cd avx512bw avx512vl clflushopt

sample/cpuid/slm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp pclmulqdq prefetchw enh_rep rdrand movbe

sample/cpuid/slt.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 movbe

sample/cpuid/snb.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx

sample/cpuid/spr.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq avx512_bf16 avx512_vp2intersect amx(tile) amx(int8) amx(bf16) avx_vnni avx512_fp16 waitpkg clflushopt cldemote movdiri movdir64b

sample/cpuid/tgl.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq avx512_vp2intersect clflushopt movdiri movdir64b

sample/cpuid/tmp.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq avx fma avx2 bmi1 bmi2 lzcnt prefetchw enh_rep rdrand adx rdseed smap sha f16c movbe avx512f avx512dq avx512_ifma avx512cd avx512bw avx512vl avx512_vbmi avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq avx512_bf16 avx512_vp2intersect amx(tile) amx(int8) amx(bf16) avx_vnni avx512_fp16 waitpkg clflushopt cldemote movdiri movdir64b

sample/cpuid/tnt.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp osxsave(xgetvb) pclmulqdq enh_rep rdrand rdseed smap sha movbe clflushopt

sample/cpuid/update-txt.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+make -C ../ test_util64
+
+cpus=(p4p mrm pnr nhm wsm snb ivb hsw bdw slt slm glm glp tnt skl cnl icl skx clx cpx icx knl knm tgl adl spr)
+for cpu in ${cpus[@]} ; do
+  echo $cpu
+  ~/bin/sde -$cpu -- ../test_util64 -cpuid > $cpu.txt
+done
+

sample/cpuid/wsm.txt

@@ -0,0 +1,2 @@
+vendor intel
+ mmx mmx2 cmov sse sse2 sse3 ssse3 sse41 sse42 popcnt aesni rdtscp pclmulqdq

test/make_nm.cpp

@@ -1317,6 +1317,7 @@ class Test {
 				put(p, REG64, REG16|REG8|MEM8|MEM16);
 				put(p, REG32, REG16|REG8|MEM8|MEM16);
 				put(p, REG16, REG8|MEM8);
+				put(p, "eax, ah");
 			}
 		}
 #ifdef XBYAK64

xbyak/xbyak.h

@@ -77,7 +77,11 @@
 	#endif
 	#include <windows.h>
 	#include <malloc.h>
-	#define XBYAK_TLS __declspec(thread)
+	#ifdef _MSC_VER
+		#define XBYAK_TLS __declspec(thread)
+	#else
+		#define XBYAK_TLS __thread
+	#endif
 #elif defined(__GNUC__)
 	#include <unistd.h>
 	#include <sys/mman.h>
@@ -144,7 +148,7 @@ namespace Xbyak {
 
 enum {
 	DEFAULT_MAX_CODE_SIZE = 4096,
-	VERSION = 0x6601 /* 0xABCD = A.BC(.D) */
+	VERSION = 0x6610 /* 0xABCD = A.BC(.D) */
 };
 
 #ifndef MIE_INTEGER_TYPE_DEFINED
@@ -2187,9 +2191,6 @@ private:
 	{
 		if (op.isBit(32)) XBYAK_THROW(ERR_BAD_COMBINATION)
 		int w = op.isBit(16);
-#ifdef XBYAK64
-		if (op.isHigh8bit()) XBYAK_THROW(ERR_BAD_COMBINATION)
-#endif
 		bool cond = reg.isREG() && (reg.getBit() > op.getBit());
 		opModRM(reg, op, cond && op.isREG(), cond && op.isMEM(), 0x0F, code | w);
 	}

xbyak/xbyak_mnemonic.h

@@ -1,4 +1,4 @@
-const char *getVersionString() const { return "6.60.1"; }
+const char *getVersionString() const { return "6.61"; }
 void adc(const Operand& op, uint32_t 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); }

xbyak/xbyak_util.h

@@ -506,9 +506,6 @@ public:
 						if (EBX & (1U << 31)) type_ |= tAVX512VL;
 						if (ECX & (1U << 1)) type_ |= tAVX512_VBMI;
 						if (ECX & (1U << 6)) type_ |= tAVX512_VBMI2;
-						if (ECX & (1U << 8)) type_ |= tGFNI;
-						if (ECX & (1U << 9)) type_ |= tVAES;
-						if (ECX & (1U << 10)) type_ |= tVPCLMULQDQ;
 						if (ECX & (1U << 11)) type_ |= tAVX512_VNNI;
 						if (ECX & (1U << 12)) type_ |= tAVX512_BITALG;
 						if (ECX & (1U << 14)) type_ |= tAVX512_VPOPCNTDQ;
@@ -537,6 +534,9 @@ public:
 			if (EBX & (1U << 29)) type_ |= tSHA;
 			if (ECX & (1U << 0)) type_ |= tPREFETCHWT1;
 			if (ECX & (1U << 5)) type_ |= tWAITPKG;
+			if (ECX & (1U << 8)) type_ |= tGFNI;
+			if (ECX & (1U << 9)) type_ |= tVAES;
+			if (ECX & (1U << 10)) type_ |= tVPCLMULQDQ;
 			if (ECX & (1U << 25)) type_ |= tCLDEMOTE;
 			if (ECX & (1U << 27)) type_ |= tMOVDIRI;
 			if (ECX & (1U << 28)) type_ |= tMOVDIR64B;