[X86] Add ENQCMD intrinsics.

[tianqing]

Summary

For more details about these instructions, please refer to the latest ISE document: https://software.intel.com/en-us/download/intel-architecture-instruction-set-extensions-programming-reference.

[RKSimon]

lib/Headers/enqcmdintrin.h

22

Please add doxygen comments describing the intrinsics

[tianqing]

Added doxygen comments of intrinsics in enqcmdintrin.h.

[craig.topper]

LGTM

[RKSimon]

In D62282#1528631, @tianqing wrote:

Added doxygen comments of intrinsics in enqcmdintrin.h.

Thanks, LGTM as well.

docs/ClangCommandLineReference.rst

 .. option:: -mclflushopt, -mno-clflushopt
 
 .. option:: -mclwb, -mno-clwb
 
 .. option:: -mclzero, -mno-clzero
 
 .. option:: -mcx16, -mno-cx16
 
+.. option:: -menqcmd, -mno-enqcmd
+
 .. option:: -mf16c, -mno-f16c
 
 .. option:: -mfma, -mno-fma
 
 .. option:: -mfma4, -mno-fma4
 
 .. option:: -mfsgsbase, -mno-fsgsbase
 

include/clang/Basic/BuiltinsX86.def

 TARGET_BUILTIN(__builtin_ia32_movdir64b, "vv*vC*", "n", "movdir64b")
 
 // PTWRITE
 TARGET_BUILTIN(__builtin_ia32_ptwrite32, "vUi", "n", "ptwrite")
 
 // INVPCID
 TARGET_BUILTIN(__builtin_ia32_invpcid, "vUiv*", "nc", "invpcid")
 
+// ENQCMD
+TARGET_BUILTIN(__builtin_ia32_enqcmd, "Ucv*vC*", "n", "enqcmd")
+TARGET_BUILTIN(__builtin_ia32_enqcmds, "Ucv*vC*", "n", "enqcmd")
+
 // MSVC
 TARGET_HEADER_BUILTIN(_BitScanForward, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(_BitScanReverse, "UcUNi*UNi", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 
 TARGET_HEADER_BUILTIN(_ReadWriteBarrier, "v", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(_ReadBarrier,      "v", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 TARGET_HEADER_BUILTIN(_WriteBarrier,     "v", "nh", "intrin.h", ALL_MS_LANGUAGES, "")
 

include/clang/Driver/Options.td

 def mclwb : Flag<["-"], "mclwb">, Group<m_x86_Features_Group>;
 def mno_clwb : Flag<["-"], "mno-clwb">, Group<m_x86_Features_Group>;
 def mwbnoinvd : Flag<["-"], "mwbnoinvd">, Group<m_x86_Features_Group>;
 def mno_wbnoinvd : Flag<["-"], "mno-wbnoinvd">, Group<m_x86_Features_Group>;
 def mclzero : Flag<["-"], "mclzero">, Group<m_x86_Features_Group>;
 def mno_clzero : Flag<["-"], "mno-clzero">, Group<m_x86_Features_Group>;
 def mcx16 : Flag<["-"], "mcx16">, Group<m_x86_Features_Group>;
 def mno_cx16 : Flag<["-"], "mno-cx16">, Group<m_x86_Features_Group>;
+def menqcmd : Flag<["-"], "menqcmd">, Group<m_x86_Features_Group>;
+def mno_enqcmd : Flag<["-"], "mno-enqcmd">, Group<m_x86_Features_Group>;
 def mf16c : Flag<["-"], "mf16c">, Group<m_x86_Features_Group>;
 def mno_f16c : Flag<["-"], "mno-f16c">, Group<m_x86_Features_Group>;
 def mfma : Flag<["-"], "mfma">, Group<m_x86_Features_Group>;
 def mno_fma : Flag<["-"], "mno-fma">, Group<m_x86_Features_Group>;
 def mfma4 : Flag<["-"], "mfma4">, Group<m_x86_Features_Group>;
 def mno_fma4 : Flag<["-"], "mno-fma4">, Group<m_x86_Features_Group>;
 def mfsgsbase : Flag<["-"], "mfsgsbase">, Group<m_x86_Features_Group>;
 def mno_fsgsbase : Flag<["-"], "mno-fsgsbase">, Group<m_x86_Features_Group>;

lib/Basic/Targets/X86.h

class LLVM_LIBRARY_VISIBILITY X86TargetInfo : public TargetInfo {
   bool HasRetpolineExternalThunk = false;
   bool HasLAHFSAHF = false;
   bool HasWBNOINVD = false;
   bool HasWAITPKG = false;
   bool HasMOVDIRI = false;
   bool HasMOVDIR64B = false;
   bool HasPTWRITE = false;
   bool HasINVPCID = false;
+  bool HasENQCMD = false;
 
 protected:
   /// Enumeration of all of the X86 CPUs supported by Clang.
   ///
   /// Each enumeration represents a particular CPU supported by Clang. These
   /// loosely correspond to the options passed to '-march' or '-mtune' flags.
   enum CPUKind {
     CK_Generic,

lib/Basic/Targets/X86.cpp

for (const auto &Feature : Features) {
     } else if (Feature == "+movdir64b") {
       HasMOVDIR64B = true;
     } else if (Feature == "+pconfig") {
       HasPCONFIG = true;
     } else if (Feature == "+ptwrite") {
       HasPTWRITE = true;
     } else if (Feature == "+invpcid") {
       HasINVPCID = true;
+    } else if (Feature == "+enqcmd") {
+      HasENQCMD = true;
     }
 
     X86SSEEnum Level = llvm::StringSwitch<X86SSEEnum>(Feature)
                            .Case("+avx512f", AVX512F)
                            .Case("+avx2", AVX2)
                            .Case("+avx", AVX)
                            .Case("+sse4.2", SSE42)
                            .Case("+sse4.1", SSE41)
void X86TargetInfo::getTargetDefines(const LangOptions &Opts,
   if (HasMOVDIR64B)
     Builder.defineMacro("__MOVDIR64B__");
   if (HasPCONFIG)
     Builder.defineMacro("__PCONFIG__");
   if (HasPTWRITE)
     Builder.defineMacro("__PTWRITE__");
   if (HasINVPCID)
     Builder.defineMacro("__INVPCID__");
+  if (HasENQCMD)
+    Builder.defineMacro("__ENQCMD__");
 
   // Each case falls through to the previous one here.
   switch (SSELevel) {
   case AVX512F:
     Builder.defineMacro("__AVX512F__");
     LLVM_FALLTHROUGH;
   case AVX2:
     Builder.defineMacro("__AVX2__");
return llvm::StringSwitch<bool>(Name)
       .Case("avx512vp2intersect", true)
       .Case("bmi", true)
       .Case("bmi2", true)
       .Case("cldemote", true)
       .Case("clflushopt", true)
       .Case("clwb", true)
       .Case("clzero", true)
       .Case("cx16", true)
+      .Case("enqcmd", true)
       .Case("f16c", true)
       .Case("fma", true)
       .Case("fma4", true)
       .Case("fsgsbase", true)
       .Case("fxsr", true)
       .Case("gfni", true)
       .Case("invpcid", true)
       .Case("lwp", true)
return llvm::StringSwitch<bool>(Feature)
       .Case("bmi", HasBMI)
       .Case("bmi2", HasBMI2)
       .Case("cldemote", HasCLDEMOTE)
       .Case("clflushopt", HasCLFLUSHOPT)
       .Case("clwb", HasCLWB)
       .Case("clzero", HasCLZERO)
       .Case("cx8", HasCX8)
       .Case("cx16", HasCX16)
+      .Case("enqcmd", HasENQCMD)
       .Case("f16c", HasF16C)
       .Case("fma", HasFMA)
       .Case("fma4", XOPLevel >= FMA4)
       .Case("fsgsbase", HasFSGSBASE)
       .Case("fxsr", HasFXSR)
       .Case("gfni", HasGFNI)
       .Case("invpcid", HasINVPCID)
       .Case("lwp", HasLWP)

lib/Headers/CMakeLists.txt

set(files
   __clang_cuda_runtime_wrapper.h
   cetintrin.h
   cldemoteintrin.h
   clzerointrin.h
   cpuid.h
   clflushoptintrin.h
   clwbintrin.h
   emmintrin.h
+  enqcmdintrin.h
   f16cintrin.h
   float.h
   fma4intrin.h
   fmaintrin.h
   fxsrintrin.h
   gfniintrin.h
   htmintrin.h
   htmxlintrin.h

lib/Headers/cpuid.h

 #define bit_VPCLMULQDQ       0x00000400
 #define bit_AVX512VNNI       0x00000800
 #define bit_AVX512BITALG     0x00001000
 #define bit_AVX512VPOPCNTDQ  0x00004000
 #define bit_RDPID            0x00400000
 #define bit_CLDEMOTE         0x02000000
 #define bit_MOVDIRI          0x08000000
 #define bit_MOVDIR64B        0x10000000
+#define bit_ENQCMD           0x20000000
 
 /* Features in %edx for leaf 7 sub-leaf 0 */
 #define bit_AVX5124VNNIW  0x00000004
 #define bit_AVX5124FMAPS  0x00000008
 #define bit_PCONFIG       0x00040000
 #define bit_IBT           0x00100000
 
 /* Features in %eax for leaf 7 sub-leaf 1 */

lib/Headers/enqcmdintrin.h

+/*===------------------ enqcmdintrin.h - enqcmd intrinsics -----------------===
+ *
+ * Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+ * See https://llvm.org/LICENSE.txt for license information.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ *
+ *===-----------------------------------------------------------------------===
+ */
+
+#ifndef __IMMINTRIN_H
+#error "Never use <enqcmdintrin.h> directly; include <immintrin.h> instead."
+#endif
+
+#ifndef __ENQCMDINTRIN_H
+#define __ENQCMDINTRIN_H
+
+/* Define the default attributes for the functions in this file */
+#define _DEFAULT_FN_ATTRS \
+  __attribute__((__always_inline__, __nodebug__, __target__("enqcmd")))
+
+/// Reads 64-byte command pointed by \a __src, formats 64-byte enqueue store
+///    data, and performs 64-byte enqueue store to memory pointed by \a __dst.
+///    This intrinsics may only be used in User mode.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsics corresponds to the <c> ENQCMD </c> instruction.
+///
+/// \param __dst
+///    Pointer to the destination of the enqueue store.
+/// \param __src
+///    Pointer to 64-byte command data.
+/// \returns If the command data is successfully written to \a __dst then 0 is
+///    returned. Otherwise 1 is returned.
+static __inline__ int _DEFAULT_FN_ATTRS
+_enqcmd (void *__dst, const void *__src)
+{
+  return __builtin_ia32_enqcmd(__dst, __src);
+}
+
+/// Reads 64-byte command pointed by \a __src, formats 64-byte enqueue store
+///    data, and performs 64-byte enqueue store to memory pointed by \a __dst
+///    This intrinsic may only be used in Privileged mode.
+///
+/// \headerfile <x86intrin.h>
+///
+/// This intrinsics corresponds to the <c> ENQCMDS </c> instruction.
+///
+/// \param __dst
+///    Pointer to the destination of the enqueue store.
+/// \param __src
+///    Pointer to 64-byte command data.
+/// \returns If the command data is successfully written to \a __dst then 0 is
+///    returned. Otherwise 1 is returned.
+static __inline__ int _DEFAULT_FN_ATTRS
+_enqcmds (void *__dst, const void *__src)
+{
+  return __builtin_ia32_enqcmds(__dst, __src);
+}
+
+#undef _DEFAULT_FN_ATTRS
+
+#endif /* __ENQCMDINTRIN_H */

lib/Headers/immintrin.h

 #include <avx512vp2intersectintrin.h>
 #endif
 
 #if !defined(_MSC_VER) || __has_feature(modules) || \
   (defined(__AVX512VL__) && defined(__AVX512VP2INTERSECT__))
 #include <avx512vlvp2intersectintrin.h>
 #endif
 
+#if !defined(_MSC_VER) || __has_feature(modules) || defined(__ENQCMD__)
+#include <enqcmdintrin.h>
+#endif
+
 #if defined(_MSC_VER) && __has_extension(gnu_asm)
 /* Define the default attributes for these intrinsics */
 #define __DEFAULT_FN_ATTRS __attribute__((__always_inline__, __nodebug__))
 #ifdef __cplusplus
 extern "C" {
 #endif
 /*----------------------------------------------------------------------------*\
 |* Interlocked Exchange HLE

test/CodeGen/x86-enqcmd-builtins.c

+// RUN: %clang_cc1 %s -ffreestanding -triple i386-unknown-unknown -target-feature +enqcmd -emit-llvm -o - | FileCheck %s
+// RUN: %clang_cc1 %s -ffreestanding -triple x86_64-unknown-unknown -target-feature +enqcmd -emit-llvm -o - | FileCheck %s
+
+#include <immintrin.h>
+
+int test_enqcmd(void *dst, const void *src) {
+// CHECK-LABEL: @test_enqcmd
+// CHECK: %[[TMP0:.+]] = call i8 @llvm.x86.enqcmd(i8* %{{.+}}, i8* %{{.+}})
+// CHECK: %[[RET:.+]] = zext i8 %[[TMP0]] to i32
+// CHECK: ret i32 %[[RET]]
+    return _enqcmd(dst, src);
+}
+
+int test_enqcmds(void *dst, const void *src) {
+// CHECK-LABEL: @test_enqcmds
+// CHECK: %[[TMP0:.+]] = call i8 @llvm.x86.enqcmds(i8* %{{.+}}, i8* %{{.+}})
+// CHECK: %[[RET:.+]] = zext i8 %[[TMP0]] to i32
+// CHECK: ret i32 %[[RET]]
+    return _enqcmds(dst, src);
+}

test/Driver/x86-target-features.c

 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-invpcid %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-INVPCID %s
 // INVPCID: "-target-feature" "+invpcid"
 // NO-INVPCID: "-target-feature" "-invpcid"
 
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mavx512bf16 %s -### -o %t.o 2>&1 | FileCheck -check-prefix=AVX512BF16 %s
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-avx512bf16 %s -### -o %t.o 2>&1 | FileCheck -check-prefix=NO-AVX512BF16 %s
 // AVX512BF16: "-target-feature" "+avx512bf16"
 // NO-AVX512BF16: "-target-feature" "-avx512bf16"
+
+// RUN: %clang -target i386-unknown-linux-gnu -march=i386 -menqcmd %s -### -o %t.o 2>&1 | FileCheck --check-prefix=ENQCMD %s
+// RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-enqcmd %s -### -o %t.o 2>&1 | FileCheck --check-prefix=NO-ENQCMD %s
+// ENQCMD: "-target-feature" "+enqcmd"
+// NO-ENQCMD: "-target-feature" "-enqcmd"

test/Preprocessor/x86_target_features.c

 // VP2INTERSECT: #define __AVX512F__ 1
 // VP2INTERSECT: #define __AVX512VP2INTERSECT__ 1
 
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mno-avx512vp2intersect -x c -E -dM -o - %s | FileCheck  -check-prefix=NOVP2INTERSECT %s
 // RUN: %clang -target i386-unknown-linux-gnu -march=i386 -mavx512vp2intersect -mno-avx512f -x c -E -dM -o - %s | FileCheck  -check-prefix=NOVP2INTERSECT %s
 
 // NOVP2INTERSECT-NOT: #define __AVX512VP2INTERSECT__ 1
 
+// RUN: %clang -target i386-unknown-unknown -march=atom -menqcmd -x c -E -dM -o - %s | FileCheck -match-full-lines --check-prefix=ENQCMD %s
+
+// ENQCMD: #define __ENQCMD__ 1
+
+// RUN: %clang -target i386-unknown-unknown -march=atom -mno-enqcmd -x c -E -dM -o - %s | FileCheck -match-full-lines --check-prefix=NOENQCMD %s
+
+// NOENQCMD-NOT: #define __ENQCMD__ 1