diff options
Diffstat (limited to 'llvm/include')
29 files changed, 505 insertions, 270 deletions
diff --git a/llvm/include/llvm-c/DebugInfo.h b/llvm/include/llvm-c/DebugInfo.h index 2ecd69a..70da3a6 100644 --- a/llvm/include/llvm-c/DebugInfo.h +++ b/llvm/include/llvm-c/DebugInfo.h @@ -204,6 +204,11 @@ enum { typedef unsigned LLVMMetadataKind; /** + * The kind of checksum to emit. + */ +typedef enum { CSK_MD5, CSK_SHA1, CSK_SHA256 } LLVMChecksumKind; + +/** * An LLVM DWARF type encoding. */ typedef unsigned LLVMDWARFTypeEncoding; @@ -327,6 +332,25 @@ LLVM_C_ABI LLVMMetadataRef LLVMDIBuilderCreateFile(LLVMDIBuilderRef Builder, size_t DirectoryLen); /** + * Create a file descriptor to hold debugging information for a file. + * \param Builder The \c DIBuilder. + * \param Filename File name. + * \param FilenameLen The length of the C string passed to \c Filename. + * \param Directory Directory. + * \param DirectoryLen The length of the C string passed to \c Directory. + * \param ChecksumKind The kind of checksum. eg MD5, SHA256 + * \param Checksum The checksum. + * \param ChecksumLen The length of the checksum. + * \param Souce The embedded source. + * \param SourceLen The length of the source. + */ +LLVM_C_ABI LLVMMetadataRef LLVMDIBuilderCreateFileWithChecksum( + LLVMDIBuilderRef Builder, const char *Filename, size_t FilenameLen, + const char *Directory, size_t DirectoryLen, LLVMChecksumKind ChecksumKind, + const char *Checksum, size_t ChecksumLen, const char *Source, + size_t SourceLen); + +/** * Creates a new descriptor for a module with the specified parent scope. * \param Builder The \c DIBuilder. * \param ParentScope The parent scope containing this module declaration. diff --git a/llvm/include/llvm/ADT/APFloat.h b/llvm/include/llvm/ADT/APFloat.h index a1bfce7..bccdb89 100644 --- a/llvm/include/llvm/ADT/APFloat.h +++ b/llvm/include/llvm/ADT/APFloat.h @@ -138,10 +138,16 @@ enum lostFraction { // Example of truncated bits: /// New operations: sqrt, IEEE remainder, C90 fmod, nexttoward. /// +namespace detail { +class IEEEFloat; +class DoubleAPFloat; +} // namespace detail + // This is the common type definitions shared by APFloat and its internal // implementation classes. This struct should not define any non-static data // members. -struct APFloatBase { +class APFloatBase { +public: typedef APInt::WordType integerPart; static constexpr unsigned integerPartWidth = APInt::APINT_BITS_PER_WORD; @@ -257,30 +263,64 @@ struct APFloatBase { LLVM_ABI static const llvm::fltSemantics &EnumToSemantics(Semantics S); LLVM_ABI static Semantics SemanticsToEnum(const llvm::fltSemantics &Sem); - LLVM_ABI static const fltSemantics &IEEEhalf() LLVM_READNONE; - LLVM_ABI static const fltSemantics &BFloat() LLVM_READNONE; - LLVM_ABI static const fltSemantics &IEEEsingle() LLVM_READNONE; - LLVM_ABI static const fltSemantics &IEEEdouble() LLVM_READNONE; - LLVM_ABI static const fltSemantics &IEEEquad() LLVM_READNONE; - LLVM_ABI static const fltSemantics &PPCDoubleDouble() LLVM_READNONE; - LLVM_ABI static const fltSemantics &PPCDoubleDoubleLegacy() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E5M2() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E5M2FNUZ() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E4M3() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E4M3FN() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E4M3FNUZ() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E4M3B11FNUZ() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E3M4() LLVM_READNONE; - LLVM_ABI static const fltSemantics &FloatTF32() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float8E8M0FNU() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float6E3M2FN() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float6E2M3FN() LLVM_READNONE; - LLVM_ABI static const fltSemantics &Float4E2M1FN() LLVM_READNONE; - LLVM_ABI static const fltSemantics &x87DoubleExtended() LLVM_READNONE; +private: + LLVM_ABI static const fltSemantics semIEEEhalf; + LLVM_ABI static const fltSemantics semBFloat; + LLVM_ABI static const fltSemantics semIEEEsingle; + LLVM_ABI static const fltSemantics semIEEEdouble; + LLVM_ABI static const fltSemantics semIEEEquad; + LLVM_ABI static const fltSemantics semFloat8E5M2; + LLVM_ABI static const fltSemantics semFloat8E5M2FNUZ; + LLVM_ABI static const fltSemantics semFloat8E4M3; + LLVM_ABI static const fltSemantics semFloat8E4M3FN; + LLVM_ABI static const fltSemantics semFloat8E4M3FNUZ; + LLVM_ABI static const fltSemantics semFloat8E4M3B11FNUZ; + LLVM_ABI static const fltSemantics semFloat8E3M4; + LLVM_ABI static const fltSemantics semFloatTF32; + LLVM_ABI static const fltSemantics semFloat8E8M0FNU; + LLVM_ABI static const fltSemantics semFloat6E3M2FN; + LLVM_ABI static const fltSemantics semFloat6E2M3FN; + LLVM_ABI static const fltSemantics semFloat4E2M1FN; + LLVM_ABI static const fltSemantics semX87DoubleExtended; + LLVM_ABI static const fltSemantics semBogus; + LLVM_ABI static const fltSemantics semPPCDoubleDouble; + LLVM_ABI static const fltSemantics semPPCDoubleDoubleLegacy; + + friend class detail::IEEEFloat; + friend class detail::DoubleAPFloat; + friend class APFloat; + +public: + static const fltSemantics &IEEEhalf() { return semIEEEhalf; } + static const fltSemantics &BFloat() { return semBFloat; } + static const fltSemantics &IEEEsingle() { return semIEEEsingle; } + static const fltSemantics &IEEEdouble() { return semIEEEdouble; } + static const fltSemantics &IEEEquad() { return semIEEEquad; } + static const fltSemantics &PPCDoubleDouble() { return semPPCDoubleDouble; } + static const fltSemantics &PPCDoubleDoubleLegacy() { + return semPPCDoubleDoubleLegacy; + } + static const fltSemantics &Float8E5M2() { return semFloat8E5M2; } + static const fltSemantics &Float8E5M2FNUZ() { return semFloat8E5M2FNUZ; } + static const fltSemantics &Float8E4M3() { return semFloat8E4M3; } + static const fltSemantics &Float8E4M3FN() { return semFloat8E4M3FN; } + static const fltSemantics &Float8E4M3FNUZ() { return semFloat8E4M3FNUZ; } + static const fltSemantics &Float8E4M3B11FNUZ() { + return semFloat8E4M3B11FNUZ; + } + static const fltSemantics &Float8E3M4() { return semFloat8E3M4; } + static const fltSemantics &FloatTF32() { return semFloatTF32; } + static const fltSemantics &Float8E8M0FNU() { return semFloat8E8M0FNU; } + static const fltSemantics &Float6E3M2FN() { return semFloat6E3M2FN; } + static const fltSemantics &Float6E2M3FN() { return semFloat6E2M3FN; } + static const fltSemantics &Float4E2M1FN() { return semFloat4E2M1FN; } + static const fltSemantics &x87DoubleExtended() { + return semX87DoubleExtended; + } /// A Pseudo fltsemantic used to construct APFloats that cannot conflict with /// anything real. - LLVM_ABI static const fltSemantics &Bogus() LLVM_READNONE; + static const fltSemantics &Bogus() { return semBogus; } // Returns true if any number described by this semantics can be precisely // represented by the specified semantics. Does not take into account @@ -927,69 +967,11 @@ class APFloat : public APFloatBase { llvm_unreachable("Unexpected semantics"); } - ~Storage() { - if (usesLayout<IEEEFloat>(*semantics)) { - IEEE.~IEEEFloat(); - return; - } - if (usesLayout<DoubleAPFloat>(*semantics)) { - Double.~DoubleAPFloat(); - return; - } - llvm_unreachable("Unexpected semantics"); - } - - Storage(const Storage &RHS) { - if (usesLayout<IEEEFloat>(*RHS.semantics)) { - new (this) IEEEFloat(RHS.IEEE); - return; - } - if (usesLayout<DoubleAPFloat>(*RHS.semantics)) { - new (this) DoubleAPFloat(RHS.Double); - return; - } - llvm_unreachable("Unexpected semantics"); - } - - Storage(Storage &&RHS) { - if (usesLayout<IEEEFloat>(*RHS.semantics)) { - new (this) IEEEFloat(std::move(RHS.IEEE)); - return; - } - if (usesLayout<DoubleAPFloat>(*RHS.semantics)) { - new (this) DoubleAPFloat(std::move(RHS.Double)); - return; - } - llvm_unreachable("Unexpected semantics"); - } - - Storage &operator=(const Storage &RHS) { - if (usesLayout<IEEEFloat>(*semantics) && - usesLayout<IEEEFloat>(*RHS.semantics)) { - IEEE = RHS.IEEE; - } else if (usesLayout<DoubleAPFloat>(*semantics) && - usesLayout<DoubleAPFloat>(*RHS.semantics)) { - Double = RHS.Double; - } else if (this != &RHS) { - this->~Storage(); - new (this) Storage(RHS); - } - return *this; - } - - Storage &operator=(Storage &&RHS) { - if (usesLayout<IEEEFloat>(*semantics) && - usesLayout<IEEEFloat>(*RHS.semantics)) { - IEEE = std::move(RHS.IEEE); - } else if (usesLayout<DoubleAPFloat>(*semantics) && - usesLayout<DoubleAPFloat>(*RHS.semantics)) { - Double = std::move(RHS.Double); - } else if (this != &RHS) { - this->~Storage(); - new (this) Storage(std::move(RHS)); - } - return *this; - } + LLVM_ABI ~Storage(); + LLVM_ABI Storage(const Storage &RHS); + LLVM_ABI Storage(Storage &&RHS); + LLVM_ABI Storage &operator=(const Storage &RHS); + LLVM_ABI Storage &operator=(Storage &&RHS); } U; template <typename T> static bool usesLayout(const fltSemantics &Semantics) { diff --git a/llvm/include/llvm/ADT/STLForwardCompat.h b/llvm/include/llvm/ADT/STLForwardCompat.h index da9d3ab0..273a5cf 100644 --- a/llvm/include/llvm/ADT/STLForwardCompat.h +++ b/llvm/include/llvm/ADT/STLForwardCompat.h @@ -26,6 +26,54 @@ namespace llvm { // Features from C++20 //===----------------------------------------------------------------------===// +namespace numbers { +// clang-format off +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T e_v = T(0x1.5bf0a8b145769P+1); // (2.7182818284590452354) https://oeis.org/A001113 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T egamma_v = T(0x1.2788cfc6fb619P-1); // (.57721566490153286061) https://oeis.org/A001620 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T ln2_v = T(0x1.62e42fefa39efP-1); // (.69314718055994530942) https://oeis.org/A002162 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T ln10_v = T(0x1.26bb1bbb55516P+1); // (2.3025850929940456840) https://oeis.org/A002392 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T log2e_v = T(0x1.71547652b82feP+0); // (1.4426950408889634074) +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T log10e_v = T(0x1.bcb7b1526e50eP-2); // (.43429448190325182765) +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T pi_v = T(0x1.921fb54442d18P+1); // (3.1415926535897932385) https://oeis.org/A000796 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T inv_pi_v = T(0x1.45f306dc9c883P-2); // (.31830988618379067154) https://oeis.org/A049541 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T inv_sqrtpi_v = T(0x1.20dd750429b6dP-1); // (.56418958354775628695) https://oeis.org/A087197 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T sqrt2_v = T(0x1.6a09e667f3bcdP+0); // (1.4142135623730950488) https://oeis.org/A00219 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T inv_sqrt2_v = T(0x1.6a09e667f3bcdP-1); // (.70710678118654752440) +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T sqrt3_v = T(0x1.bb67ae8584caaP+0); // (1.7320508075688772935) https://oeis.org/A002194 +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T inv_sqrt3_v = T(0x1.279a74590331cP-1); // (.57735026918962576451) +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T phi_v = T(0x1.9e3779b97f4a8P+0); // (1.6180339887498948482) https://oeis.org/A001622 + +inline constexpr double e = e_v<double>; +inline constexpr double egamma = egamma_v<double>; +inline constexpr double ln2 = ln2_v<double>; +inline constexpr double ln10 = ln10_v<double>; +inline constexpr double log2e = log2e_v<double>; +inline constexpr double log10e = log10e_v<double>; +inline constexpr double pi = pi_v<double>; +inline constexpr double inv_pi = inv_pi_v<double>; +inline constexpr double inv_sqrtpi = inv_sqrtpi_v<double>; +inline constexpr double sqrt2 = sqrt2_v<double>; +inline constexpr double inv_sqrt2 = inv_sqrt2_v<double>; +inline constexpr double sqrt3 = sqrt3_v<double>; +inline constexpr double inv_sqrt3 = inv_sqrt3_v<double>; +inline constexpr double phi = phi_v<double>; +// clang-format on +} // namespace numbers + template <typename T> struct remove_cvref // NOLINT(readability-identifier-naming) { diff --git a/llvm/include/llvm/ADT/bit.h b/llvm/include/llvm/ADT/bit.h index 66c4f94..8c68d0a 100644 --- a/llvm/include/llvm/ADT/bit.h +++ b/llvm/include/llvm/ADT/bit.h @@ -336,33 +336,28 @@ template <typename T> [[nodiscard]] T bit_ceil(T Value) { return T(1) << llvm::bit_width<T>(Value - 1u); } -// Forward-declare rotr so that rotl can use it. -template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>> -[[nodiscard]] constexpr T rotr(T V, int R); - template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>> [[nodiscard]] constexpr T rotl(T V, int R) { - unsigned N = std::numeric_limits<T>::digits; + constexpr unsigned N = std::numeric_limits<T>::digits; - R = R % N; - if (!R) - return V; + static_assert(has_single_bit(N), "& (N - 1) is only valid for powers of two"); + R = R & (N - 1); - if (R < 0) - return llvm::rotr(V, -R); + if (R == 0) + return V; return (V << R) | (V >> (N - R)); } -template <typename T, typename> [[nodiscard]] constexpr T rotr(T V, int R) { - unsigned N = std::numeric_limits<T>::digits; +template <typename T, typename = std::enable_if_t<std::is_unsigned_v<T>>> +[[nodiscard]] constexpr T rotr(T V, int R) { + constexpr unsigned N = std::numeric_limits<T>::digits; - R = R % N; - if (!R) - return V; + static_assert(has_single_bit(N), "& (N - 1) is only valid for powers of two"); + R = R & (N - 1); - if (R < 0) - return llvm::rotl(V, -R); + if (R == 0) + return V; return (V >> R) | (V << (N - R)); } diff --git a/llvm/include/llvm/Analysis/DXILResource.h b/llvm/include/llvm/Analysis/DXILResource.h index 88ac0a1..c7aff16 100644 --- a/llvm/include/llvm/Analysis/DXILResource.h +++ b/llvm/include/llvm/Analysis/DXILResource.h @@ -243,6 +243,25 @@ public: } }; +/// The dx.Padding target extension type +/// +/// `target("dx.Padding", NumBytes)` +class PaddingExtType : public TargetExtType { +public: + PaddingExtType() = delete; + PaddingExtType(const PaddingExtType &) = delete; + PaddingExtType &operator=(const PaddingExtType &) = delete; + + unsigned getNumBytes() const { return getIntParameter(0); } + + static bool classof(const TargetExtType *T) { + return T->getName() == "dx.Padding"; + } + static bool classof(const Type *T) { + return isa<TargetExtType>(T) && classof(cast<TargetExtType>(T)); + } +}; + //===----------------------------------------------------------------------===// class ResourceTypeInfo { diff --git a/llvm/include/llvm/Analysis/LoopAnalysisManager.h b/llvm/include/llvm/Analysis/LoopAnalysisManager.h index fc69cb0..1755257 100644 --- a/llvm/include/llvm/Analysis/LoopAnalysisManager.h +++ b/llvm/include/llvm/Analysis/LoopAnalysisManager.h @@ -36,7 +36,6 @@ namespace llvm { class AAResults; class AssumptionCache; -class BlockFrequencyInfo; class DominatorTree; class Function; class Loop; @@ -58,7 +57,6 @@ struct LoopStandardAnalysisResults { ScalarEvolution &SE; TargetLibraryInfo &TLI; TargetTransformInfo &TTI; - BlockFrequencyInfo *BFI; MemorySSA *MSSA; }; diff --git a/llvm/include/llvm/Analysis/LoopInfo.h b/llvm/include/llvm/Analysis/LoopInfo.h index a7a6a27..0ecb114 100644 --- a/llvm/include/llvm/Analysis/LoopInfo.h +++ b/llvm/include/llvm/Analysis/LoopInfo.h @@ -617,7 +617,7 @@ public: }; /// Function to print a loop's contents as LLVM's text IR assembly. -LLVM_ABI void printLoop(Loop &L, raw_ostream &OS, +LLVM_ABI void printLoop(const Loop &L, raw_ostream &OS, const std::string &Banner = ""); /// Find and return the loop attribute node for the attribute @p Name in diff --git a/llvm/include/llvm/Analysis/ScalarEvolution.h b/llvm/include/llvm/Analysis/ScalarEvolution.h index e5a6c8c..3d3ec14 100644 --- a/llvm/include/llvm/Analysis/ScalarEvolution.h +++ b/llvm/include/llvm/Analysis/ScalarEvolution.h @@ -1345,6 +1345,7 @@ public: class LoopGuards { DenseMap<const SCEV *, const SCEV *> RewriteMap; + SmallDenseSet<std::pair<const SCEV *, const SCEV *>> NotEqual; bool PreserveNUW = false; bool PreserveNSW = false; ScalarEvolution &SE; diff --git a/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h b/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h index 871028d..9354eef 100644 --- a/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h +++ b/llvm/include/llvm/Analysis/ScalarEvolutionPatternMatch.h @@ -95,6 +95,10 @@ inline bind_ty<const SCEVAddExpr> m_scev_Add(const SCEVAddExpr *&V) { return V; } +inline bind_ty<const SCEVMulExpr> m_scev_Mul(const SCEVMulExpr *&V) { + return V; +} + /// Match a specified const SCEV *. struct specificscev_ty { const SCEV *Expr; @@ -252,6 +256,18 @@ m_scev_UDiv(const Op0_t &Op0, const Op1_t &Op1) { return m_scev_Binary<SCEVUDivExpr>(Op0, Op1); } +template <typename Op0_t, typename Op1_t> +inline SCEVBinaryExpr_match<SCEVSMaxExpr, Op0_t, Op1_t> +m_scev_SMax(const Op0_t &Op0, const Op1_t &Op1) { + return m_scev_Binary<SCEVSMaxExpr>(Op0, Op1); +} + +template <typename Op0_t, typename Op1_t> +inline SCEVBinaryExpr_match<SCEVMinMaxExpr, Op0_t, Op1_t> +m_scev_MinMax(const Op0_t &Op0, const Op1_t &Op1) { + return m_scev_Binary<SCEVMinMaxExpr>(Op0, Op1); +} + /// Match unsigned remainder pattern. /// Matches patterns generated by getURemExpr. template <typename Op0_t, typename Op1_t> struct SCEVURem_match { @@ -284,14 +300,10 @@ template <typename Op0_t, typename Op1_t> struct SCEVURem_match { << SE.getTypeSizeInBits(TruncTy)); return Op0.match(LHS) && Op1.match(RHS); } - const auto *Add = dyn_cast<SCEVAddExpr>(Expr); - if (Add == nullptr || Add->getNumOperands() != 2) - return false; - - const SCEV *A = Add->getOperand(1); - const auto *Mul = dyn_cast<SCEVMulExpr>(Add->getOperand(0)); - if (Mul == nullptr) + const SCEV *A; + const SCEVMulExpr *Mul; + if (!SCEVPatternMatch::match(Expr, m_scev_Add(m_scev_Mul(Mul), m_SCEV(A)))) return false; const auto MatchURemWithDivisor = [&](const SCEV *B) { diff --git a/llvm/include/llvm/BinaryFormat/ELF.h b/llvm/include/llvm/BinaryFormat/ELF.h index e619b18..8d0dc64 100644 --- a/llvm/include/llvm/BinaryFormat/ELF.h +++ b/llvm/include/llvm/BinaryFormat/ELF.h @@ -647,6 +647,7 @@ enum { EF_HEXAGON_ISA_V85 = 0x00000085, // Hexagon V85 ISA EF_HEXAGON_ISA_V87 = 0x00000087, // Hexagon V87 ISA EF_HEXAGON_ISA_V89 = 0x00000089, // Hexagon V89 ISA + EF_HEXAGON_ISA_V91 = 0x00000091, // Hexagon V91 ISA EF_HEXAGON_ISA = 0x000003ff, // Hexagon V.. ISA // Tiny core flag, bit[15] @@ -680,6 +681,7 @@ enum { EF_HEXAGON_MACH_V85 = EF_HEXAGON_ISA_V85, // Hexagon V85 EF_HEXAGON_MACH_V87 = EF_HEXAGON_ISA_V87, // Hexagon V87 EF_HEXAGON_MACH_V89 = EF_HEXAGON_ISA_V89, // Hexagon V89 + EF_HEXAGON_MACH_V91 = EF_HEXAGON_ISA_V91, // Hexagon V91 EF_HEXAGON_MACH = 0x0000ffff, // Hexagon V.. }; diff --git a/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericJITLinkMemoryManager.h b/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericJITLinkMemoryManager.h index f9070af..eb71e9a 100644 --- a/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericJITLinkMemoryManager.h +++ b/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericJITLinkMemoryManager.h @@ -32,8 +32,9 @@ public: struct SymbolAddrs { ExecutorAddr Allocator; ExecutorAddr Reserve; - ExecutorAddr Finalize; - ExecutorAddr Deallocate; + ExecutorAddr Initialize; + ExecutorAddr Deinitialize; + ExecutorAddr Release; }; /// Create an EPCGenericJITLinkMemoryManager instance from a given set of diff --git a/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericRTDyldMemoryManager.h b/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericRTDyldMemoryManager.h index faec25d..fa48480 100644 --- a/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericRTDyldMemoryManager.h +++ b/llvm/include/llvm/ExecutionEngine/Orc/EPCGenericRTDyldMemoryManager.h @@ -31,8 +31,8 @@ public: struct SymbolAddrs { ExecutorAddr Instance; ExecutorAddr Reserve; - ExecutorAddr Finalize; - ExecutorAddr Deallocate; + ExecutorAddr Initialize; + ExecutorAddr Release; ExecutorAddr RegisterEHFrame; ExecutorAddr DeregisterEHFrame; }; diff --git a/llvm/include/llvm/ExecutionEngine/Orc/Shared/OrcRTBridge.h b/llvm/include/llvm/ExecutionEngine/Orc/Shared/OrcRTBridge.h index 99ba456..d68a689 100644 --- a/llvm/include/llvm/ExecutionEngine/Orc/Shared/OrcRTBridge.h +++ b/llvm/include/llvm/ExecutionEngine/Orc/Shared/OrcRTBridge.h @@ -29,8 +29,9 @@ LLVM_ABI extern const char *SimpleExecutorDylibManagerResolveWrapperName; LLVM_ABI extern const char *SimpleExecutorMemoryManagerInstanceName; LLVM_ABI extern const char *SimpleExecutorMemoryManagerReserveWrapperName; -LLVM_ABI extern const char *SimpleExecutorMemoryManagerFinalizeWrapperName; -LLVM_ABI extern const char *SimpleExecutorMemoryManagerDeallocateWrapperName; +LLVM_ABI extern const char *SimpleExecutorMemoryManagerInitializeWrapperName; +LLVM_ABI extern const char *SimpleExecutorMemoryManagerDeinitializeWrapperName; +LLVM_ABI extern const char *SimpleExecutorMemoryManagerReleaseWrapperName; LLVM_ABI extern const char *ExecutorSharedMemoryMapperServiceInstanceName; LLVM_ABI extern const char *ExecutorSharedMemoryMapperServiceReserveWrapperName; @@ -73,9 +74,12 @@ using SPSSimpleExecutorDylibManagerResolveSignature = shared::SPSExpected< using SPSSimpleExecutorMemoryManagerReserveSignature = shared::SPSExpected<shared::SPSExecutorAddr>(shared::SPSExecutorAddr, uint64_t); -using SPSSimpleExecutorMemoryManagerFinalizeSignature = - shared::SPSError(shared::SPSExecutorAddr, shared::SPSFinalizeRequest); -using SPSSimpleExecutorMemoryManagerDeallocateSignature = shared::SPSError( +using SPSSimpleExecutorMemoryManagerInitializeSignature = + shared::SPSExpected<shared::SPSExecutorAddr>(shared::SPSExecutorAddr, + shared::SPSFinalizeRequest); +using SPSSimpleExecutorMemoryManagerDeinitializeSignature = shared::SPSError( + shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSExecutorAddr>); +using SPSSimpleExecutorMemoryManagerReleaseSignature = shared::SPSError( shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSExecutorAddr>); // ExecutorSharedMemoryMapperService @@ -93,6 +97,18 @@ using SPSExecutorSharedMemoryMapperServiceDeinitializeSignature = using SPSExecutorSharedMemoryMapperServiceReleaseSignature = shared::SPSError( shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSExecutorAddr>); +// SimpleNativeMemoryMap APIs. +using SPSSimpleRemoteMemoryMapReserveSignature = + shared::SPSExpected<shared::SPSExecutorAddr>(shared::SPSExecutorAddr, + uint64_t); +using SPSSimpleRemoteMemoryMapInitializeSignature = + shared::SPSExpected<shared::SPSExecutorAddr>(shared::SPSExecutorAddr, + shared::SPSFinalizeRequest); +using SPSSimpleRemoteMemoryMapDeinitializeSignature = shared::SPSError( + shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSExecutorAddr>); +using SPSSimpleRemoteMemoryMapReleaseSignature = shared::SPSError( + shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSExecutorAddr>); + using SPSRunAsMainSignature = int64_t(shared::SPSExecutorAddr, shared::SPSSequence<shared::SPSString>); using SPSRunAsVoidFunctionSignature = int32_t(shared::SPSExecutorAddr); diff --git a/llvm/include/llvm/ExecutionEngine/Orc/SimpleRemoteMemoryMapper.h b/llvm/include/llvm/ExecutionEngine/Orc/SimpleRemoteMemoryMapper.h new file mode 100644 index 0000000..644c4f61 --- /dev/null +++ b/llvm/include/llvm/ExecutionEngine/Orc/SimpleRemoteMemoryMapper.h @@ -0,0 +1,87 @@ +//===- SimpleRemoteMemoryMapper.h - Remote memory mapper --------*- C++ -*-===// +// +// 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 +// +//===----------------------------------------------------------------------===// +// +// A simple memory mapper that uses EPC calls to implement reserve, initialize, +// deinitialize, and release. +// +//===----------------------------------------------------------------------===// + +#ifndef LLVM_EXECUTIONENGINE_ORC_SIMPLEREMOTEMEMORYMAPPER_H +#define LLVM_EXECUTIONENGINE_ORC_SIMPLEREMOTEMEMORYMAPPER_H + +#include "llvm/ExecutionEngine/Orc/MemoryMapper.h" + +namespace llvm::orc { + +/// Manages remote memory by making SPS-based EPC calls. +class LLVM_ABI SimpleRemoteMemoryMapper final : public MemoryMapper { +public: + struct SymbolAddrs { + ExecutorAddr Instance; + ExecutorAddr Reserve; + ExecutorAddr Initialize; + ExecutorAddr Deinitialize; + ExecutorAddr Release; + }; + + SimpleRemoteMemoryMapper(ExecutorProcessControl &EPC, SymbolAddrs SAs); + + static Expected<std::unique_ptr<SimpleRemoteMemoryMapper>> + Create(ExecutorProcessControl &EPC, SymbolAddrs SAs) { + return std::make_unique<SimpleRemoteMemoryMapper>(EPC, SAs); + } + + unsigned int getPageSize() override { return EPC.getPageSize(); } + + /// Reserves memory in the remote process by calling a remote + /// SPS-wrapper-function with signature + /// + /// SPSExpected<SPSExecutorAddr>(uint64_t Size). + /// + /// On success, returns the base address of the reserved range. + void reserve(size_t NumBytes, OnReservedFunction OnReserved) override; + + char *prepare(jitlink::LinkGraph &G, ExecutorAddr Addr, + size_t ContentSize) override; + + /// Initializes memory within a previously reserved region (applying + /// protections and running any finalization actions) by calling a remote + /// SPS-wrapper-function with signature + /// + /// SPSExpected<SPSExecutorAddr>(SPSFinalizeRequest) + /// + /// On success, returns a key that can be used to deinitialize the region. + void initialize(AllocInfo &AI, OnInitializedFunction OnInitialized) override; + + /// Given a series of keys from previous initialize calls, deinitialize + /// previously initialized memory regions (running dealloc actions, resetting + /// permissions and decommitting if possible) by calling a remote + /// SPS-wrapper-function with signature + /// + /// SPSError(SPSSequence<SPSExecutorAddr> Keys) + /// + void deinitialize(ArrayRef<ExecutorAddr> Allocations, + OnDeinitializedFunction OnDeInitialized) override; + + /// Given a sequence of base addresses from previous reserve calls, release + /// the underlying ranges (deinitializing any remaining regions within them) + /// by calling a remote SPS-wrapper-function with signature + /// + /// SPSError(SPSSequence<SPSExecutorAddr> Bases) + /// + void release(ArrayRef<ExecutorAddr> Reservations, + OnReleasedFunction OnRelease) override; + +private: + ExecutorProcessControl &EPC; + SymbolAddrs SAs; +}; + +} // namespace llvm::orc + +#endif // LLVM_EXECUTIONENGINE_ORC_SIMPLEREMOTEMEMORYMAPPER_H diff --git a/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/SimpleExecutorMemoryManager.h b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/SimpleExecutorMemoryManager.h index 741f203..6224e92 100644 --- a/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/SimpleExecutorMemoryManager.h +++ b/llvm/include/llvm/ExecutionEngine/Orc/TargetProcess/SimpleExecutorMemoryManager.h @@ -34,34 +34,65 @@ class LLVM_ABI SimpleExecutorMemoryManager : public ExecutorBootstrapService { public: virtual ~SimpleExecutorMemoryManager(); - Expected<ExecutorAddr> allocate(uint64_t Size); - Error finalize(tpctypes::FinalizeRequest &FR); - Error deallocate(const std::vector<ExecutorAddr> &Bases); + Expected<ExecutorAddr> reserve(uint64_t Size); + Expected<ExecutorAddr> initialize(tpctypes::FinalizeRequest &FR); + Error deinitialize(const std::vector<ExecutorAddr> &InitKeys); + Error release(const std::vector<ExecutorAddr> &Bases); Error shutdown() override; void addBootstrapSymbols(StringMap<ExecutorAddr> &M) override; private: - struct Allocation { + struct RegionInfo { size_t Size = 0; - std::vector<shared::WrapperFunctionCall> DeallocationActions; + std::vector<shared::WrapperFunctionCall> DeallocActions; }; - using AllocationsMap = DenseMap<void *, Allocation>; + struct SlabInfo { + using RegionMap = std::map<ExecutorAddr, RegionInfo>; + size_t Size = 0; + RegionMap Regions; + }; + + using SlabMap = std::map<void *, SlabInfo>; + + /// Get a reference to the slab information for the slab containing the given + /// address. + Expected<SlabInfo &> getSlabInfo(ExecutorAddr A, StringRef Context); + + /// Get a reference to the slab information for the slab *covering* the given + /// range. The given range must be a subrange of e(possibly equal to) the + /// range of the slab itself. + Expected<SlabInfo &> getSlabInfo(ExecutorAddrRange R, StringRef Context); - Error deallocateImpl(void *Base, Allocation &A); + /// Create a RegionInfo for the given range, which must not overlap any + /// existing region. + Expected<RegionInfo &> createRegionInfo(ExecutorAddrRange R, + StringRef Context); + + /// Get a reference to the region information for the given address. This + /// address must represent the start of an existing initialized region. + Expected<RegionInfo &> getRegionInfo(SlabInfo &Slab, ExecutorAddr A, + StringRef Context); + + /// Get a reference to the region information for the given address. This + /// address must represent the start of an existing initialized region. + Expected<RegionInfo &> getRegionInfo(ExecutorAddr A, StringRef Context); static llvm::orc::shared::CWrapperFunctionResult reserveWrapper(const char *ArgData, size_t ArgSize); static llvm::orc::shared::CWrapperFunctionResult - finalizeWrapper(const char *ArgData, size_t ArgSize); + initializeWrapper(const char *ArgData, size_t ArgSize); + + static llvm::orc::shared::CWrapperFunctionResult + deinitializeWrapper(const char *ArgData, size_t ArgSize); static llvm::orc::shared::CWrapperFunctionResult - deallocateWrapper(const char *ArgData, size_t ArgSize); + releaseWrapper(const char *ArgData, size_t ArgSize); std::mutex M; - AllocationsMap Allocations; + SlabMap Slabs; }; } // end namespace rt_bootstrap diff --git a/llvm/include/llvm/Frontend/OpenMP/ClauseT.h b/llvm/include/llvm/Frontend/OpenMP/ClauseT.h index db781b58..1a01fa6 100644 --- a/llvm/include/llvm/Frontend/OpenMP/ClauseT.h +++ b/llvm/include/llvm/Frontend/OpenMP/ClauseT.h @@ -571,7 +571,9 @@ struct DoacrossT { // V5.2: [8.2.1] `requirement` clauses template <typename T, typename I, typename E> // struct DynamicAllocatorsT { - using EmptyTrait = std::true_type; + using Requires = E; + using WrapperTrait = std::true_type; + OPT(Requires) v; }; template <typename T, typename I, typename E> // @@ -802,6 +804,7 @@ template <typename T, typename I, typename E> // struct MapT { using LocatorList = ObjectListT<I, E>; ENUM(MapType, To, From, Tofrom, Storage); + ENUM(AttachModifier, Always, Auto, Never); ENUM(MapTypeModifier, Always, Close, Delete, Present, Self, OmpxHold); ENUM(RefModifier, RefPtee, RefPtr, RefPtrPtee); // See note at the definition of the MapperT type. @@ -810,8 +813,8 @@ struct MapT { using MapTypeModifiers = ListT<MapTypeModifier>; // Not a spec name using TupleTrait = std::true_type; - std::tuple<OPT(MapType), OPT(MapTypeModifiers), OPT(RefModifier), - OPT(Mappers), OPT(Iterator), LocatorList> + std::tuple<OPT(MapType), OPT(MapTypeModifiers), OPT(AttachModifier), + OPT(RefModifier), OPT(Mappers), OPT(Iterator), LocatorList> t; }; @@ -1055,7 +1058,9 @@ struct ReplayableT { // V5.2: [8.2.1] `requirement` clauses template <typename T, typename I, typename E> // struct ReverseOffloadT { - using EmptyTrait = std::true_type; + using Requires = E; + using WrapperTrait = std::true_type; + OPT(Requires) v; }; // V5.2: [10.4.2] `safelen` clause @@ -1077,6 +1082,14 @@ struct ScheduleT { std::tuple<Kind, OPT(OrderingModifier), OPT(ChunkModifier), OPT(ChunkSize)> t; }; +// [6.0:361] +template <typename T, typename I, typename E> // +struct SelfMapsT { + using Requires = E; + using WrapperTrait = std::true_type; + OPT(Requires) v; +}; + // V5.2: [15.8.1] Memory-order clauses template <typename T, typename I, typename E> // struct SeqCstT { @@ -1168,18 +1181,17 @@ struct TransparentT { // V5.2: [8.2.1] `requirement` clauses template <typename T, typename I, typename E> // struct UnifiedAddressT { - using EmptyTrait = std::true_type; + using Requires = E; + using WrapperTrait = std::true_type; + OPT(Requires) v; }; // V5.2: [8.2.1] `requirement` clauses template <typename T, typename I, typename E> // struct UnifiedSharedMemoryT { - using EmptyTrait = std::true_type; -}; - -template <typename T, typename I, typename E> // -struct SelfMapsT { - using EmptyTrait = std::true_type; + using Requires = E; + using WrapperTrait = std::true_type; + OPT(Requires) v; }; // V5.2: [5.10] `uniform` clause @@ -1287,14 +1299,12 @@ using ExtensionClausesT = template <typename T, typename I, typename E> using EmptyClausesT = std::variant< AcqRelT<T, I, E>, AcquireT<T, I, E>, CaptureT<T, I, E>, CompareT<T, I, E>, - DynamicAllocatorsT<T, I, E>, FullT<T, I, E>, InbranchT<T, I, E>, - MergeableT<T, I, E>, NogroupT<T, I, E>, NoOpenmpRoutinesT<T, I, E>, + FullT<T, I, E>, InbranchT<T, I, E>, MergeableT<T, I, E>, NogroupT<T, I, E>, + NoOpenmpConstructsT<T, I, E>, NoOpenmpRoutinesT<T, I, E>, NoOpenmpT<T, I, E>, NoParallelismT<T, I, E>, NotinbranchT<T, I, E>, NowaitT<T, I, E>, ReadT<T, I, E>, RelaxedT<T, I, E>, ReleaseT<T, I, E>, - ReverseOffloadT<T, I, E>, SeqCstT<T, I, E>, SimdT<T, I, E>, - ThreadsT<T, I, E>, UnifiedAddressT<T, I, E>, UnifiedSharedMemoryT<T, I, E>, - UnknownT<T, I, E>, UntiedT<T, I, E>, UseT<T, I, E>, WeakT<T, I, E>, - WriteT<T, I, E>, NoOpenmpConstructsT<T, I, E>, SelfMapsT<T, I, E>>; + SeqCstT<T, I, E>, SimdT<T, I, E>, ThreadsT<T, I, E>, UnknownT<T, I, E>, + UntiedT<T, I, E>, UseT<T, I, E>, WeakT<T, I, E>, WriteT<T, I, E>>; template <typename T, typename I, typename E> using IncompleteClausesT = @@ -1322,18 +1332,20 @@ using WrapperClausesT = std::variant< AtomicDefaultMemOrderT<T, I, E>, AtT<T, I, E>, BindT<T, I, E>, CollapseT<T, I, E>, ContainsT<T, I, E>, CopyinT<T, I, E>, CopyprivateT<T, I, E>, DefaultT<T, I, E>, DestroyT<T, I, E>, - DetachT<T, I, E>, DeviceTypeT<T, I, E>, EnterT<T, I, E>, - ExclusiveT<T, I, E>, FailT<T, I, E>, FilterT<T, I, E>, FinalT<T, I, E>, - FirstprivateT<T, I, E>, HasDeviceAddrT<T, I, E>, HintT<T, I, E>, - HoldsT<T, I, E>, InclusiveT<T, I, E>, IndirectT<T, I, E>, + DetachT<T, I, E>, DeviceTypeT<T, I, E>, DynamicAllocatorsT<T, I, E>, + EnterT<T, I, E>, ExclusiveT<T, I, E>, FailT<T, I, E>, FilterT<T, I, E>, + FinalT<T, I, E>, FirstprivateT<T, I, E>, HasDeviceAddrT<T, I, E>, + HintT<T, I, E>, HoldsT<T, I, E>, InclusiveT<T, I, E>, IndirectT<T, I, E>, InitializerT<T, I, E>, IsDevicePtrT<T, I, E>, LinkT<T, I, E>, MessageT<T, I, E>, NocontextT<T, I, E>, NontemporalT<T, I, E>, NovariantsT<T, I, E>, NumTeamsT<T, I, E>, NumThreadsT<T, I, E>, OrderedT<T, I, E>, PartialT<T, I, E>, PriorityT<T, I, E>, PrivateT<T, I, E>, - ProcBindT<T, I, E>, SafelenT<T, I, E>, SeverityT<T, I, E>, SharedT<T, I, E>, - SimdlenT<T, I, E>, SizesT<T, I, E>, PermutationT<T, I, E>, - ThreadLimitT<T, I, E>, UniformT<T, I, E>, UpdateT<T, I, E>, - UseDeviceAddrT<T, I, E>, UseDevicePtrT<T, I, E>, UsesAllocatorsT<T, I, E>>; + ProcBindT<T, I, E>, ReverseOffloadT<T, I, E>, SafelenT<T, I, E>, + SelfMapsT<T, I, E>, SeverityT<T, I, E>, SharedT<T, I, E>, SimdlenT<T, I, E>, + SizesT<T, I, E>, PermutationT<T, I, E>, ThreadLimitT<T, I, E>, + UnifiedAddressT<T, I, E>, UnifiedSharedMemoryT<T, I, E>, UniformT<T, I, E>, + UpdateT<T, I, E>, UseDeviceAddrT<T, I, E>, UseDevicePtrT<T, I, E>, + UsesAllocatorsT<T, I, E>>; template <typename T, typename I, typename E> using UnionOfAllClausesT = typename type::Union< // diff --git a/llvm/include/llvm/Frontend/OpenMP/ConstructDecompositionT.h b/llvm/include/llvm/Frontend/OpenMP/ConstructDecompositionT.h index 047baa3..6d6eb5cd 100644 --- a/llvm/include/llvm/Frontend/OpenMP/ConstructDecompositionT.h +++ b/llvm/include/llvm/Frontend/OpenMP/ConstructDecompositionT.h @@ -708,6 +708,7 @@ bool ConstructDecompositionT<C, H>::applyClause( tomp::clause::MapT<TypeTy, IdTy, ExprTy>{ {/*MapType=*/MapType::Tofrom, /*MapTypeModifier=*/std::nullopt, + /*AttachModifier=*/std::nullopt, /*RefModifier=*/std::nullopt, /*Mapper=*/std::nullopt, /*Iterator=*/std::nullopt, /*LocatorList=*/std::move(tofrom)}}); @@ -970,8 +971,9 @@ bool ConstructDecompositionT<C, H>::applyClause( llvm::omp::Clause::OMPC_map, tomp::clause::MapT<TypeTy, IdTy, ExprTy>{ {/*MapType=*/MapType::Tofrom, /*MapTypeModifier=*/std::nullopt, - /*RefModifier=*/std::nullopt, /*Mapper=*/std::nullopt, - /*Iterator=*/std::nullopt, /*LocatorList=*/std::move(tofrom)}}); + /*AttachModifier=*/std::nullopt, /*RefModifier=*/std::nullopt, + /*Mapper=*/std::nullopt, /*Iterator=*/std::nullopt, + /*LocatorList=*/std::move(tofrom)}}); dirTarget->clauses.push_back(map); applied = true; diff --git a/llvm/include/llvm/Frontend/OpenMP/OMP.td b/llvm/include/llvm/Frontend/OpenMP/OMP.td index 86a9e24..edcf7a9 100644 --- a/llvm/include/llvm/Frontend/OpenMP/OMP.td +++ b/llvm/include/llvm/Frontend/OpenMP/OMP.td @@ -177,6 +177,8 @@ def OMPC_Doacross : Clause<[Spelling<"doacross">]> { } def OMPC_DynamicAllocators : Clause<[Spelling<"dynamic_allocators">]> { let clangClass = "OMPDynamicAllocatorsClause"; + let flangClass = "OmpDynamicAllocatorsClause"; + let isValueOptional = true; } def OMPC_DynGroupprivate : Clause<[Spelling<"dyn_groupprivate">]> { let flangClass = "OmpDynGroupprivateClause"; @@ -467,6 +469,8 @@ def OMPC_Replayable : Clause<[Spelling<"replayable">]> { } def OMPC_ReverseOffload : Clause<[Spelling<"reverse_offload">]> { let clangClass = "OMPReverseOffloadClause"; + let flangClass = "OmpReverseOffloadClause"; + let isValueOptional = true; } def OMPC_SafeLen : Clause<[Spelling<"safelen">]> { let clangClass = "OMPSafelenClause"; @@ -541,12 +545,18 @@ def OMPC_Transparent : Clause<[Spelling<"transparent">]> { } def OMPC_UnifiedAddress : Clause<[Spelling<"unified_address">]> { let clangClass = "OMPUnifiedAddressClause"; + let flangClass = "OmpUnifiedAddressClause"; + let isValueOptional = true; } def OMPC_UnifiedSharedMemory : Clause<[Spelling<"unified_shared_memory">]> { let clangClass = "OMPUnifiedSharedMemoryClause"; + let flangClass = "OmpUnifiedSharedMemoryClause"; + let isValueOptional = true; } def OMPC_SelfMaps : Clause<[Spelling<"self_maps">]> { let clangClass = "OMPSelfMapsClause"; + let flangClass = "OmpSelfMapsClause"; + let isValueOptional = true; } def OMPC_Uniform : Clause<[Spelling<"uniform">]> { let flangClass = "Name"; diff --git a/llvm/include/llvm/IR/DebugProgramInstruction.h b/llvm/include/llvm/IR/DebugProgramInstruction.h index e0292c2..457c60e3b 100644 --- a/llvm/include/llvm/IR/DebugProgramInstruction.h +++ b/llvm/include/llvm/IR/DebugProgramInstruction.h @@ -14,7 +14,7 @@ // dbg.value(metadata i32 %foo, ...) // %bar = void call @ext(%foo); // -// and all information is stored in the Value / Metadata hierachy defined +// and all information is stored in the Value / Metadata hierarchy defined // elsewhere in LLVM. In the "DbgRecord" design, each instruction /may/ have a // connection with a DbgMarker, which identifies a position immediately before // the instruction, and each DbgMarker /may/ then have connections to DbgRecords @@ -37,7 +37,7 @@ // // This structure separates the two concerns of the position of the debug-info // in the function, and the Value that it refers to. It also creates a new -// "place" in-between the Value / Metadata hierachy where we can customise +// "place" in-between the Value / Metadata hierarchy where we can customise // storage and allocation techniques to better suite debug-info workloads. // NB: as of the initial prototype, none of that has actually been attempted // yet. @@ -162,7 +162,7 @@ public: LLVM_ABI bool isIdenticalToWhenDefined(const DbgRecord &R) const; /// Convert this DbgRecord back into an appropriate llvm.dbg.* intrinsic. /// \p InsertBefore Optional position to insert this intrinsic. - /// \returns A new llvm.dbg.* intrinsic representiung this DbgRecord. + /// \returns A new llvm.dbg.* intrinsic representing this DbgRecord. LLVM_ABI DbgInfoIntrinsic * createDebugIntrinsic(Module *M, Instruction *InsertBefore) const; ///@} @@ -530,7 +530,7 @@ public: LLVM_ABI void setKillAddress(); /// Check whether this kills the address component. This doesn't take into /// account the position of the intrinsic, therefore a returned value of false - /// does not guarentee the address is a valid location for the variable at the + /// does not guarantee the address is a valid location for the variable at the /// intrinsic's position in IR. LLVM_ABI bool isKillAddress() const; @@ -539,7 +539,7 @@ public: LLVM_ABI DbgVariableRecord *clone() const; /// Convert this DbgVariableRecord back into a dbg.value intrinsic. /// \p InsertBefore Optional position to insert this intrinsic. - /// \returns A new dbg.value intrinsic representiung this DbgVariableRecord. + /// \returns A new dbg.value intrinsic representing this DbgVariableRecord. LLVM_ABI DbgVariableIntrinsic * createDebugIntrinsic(Module *M, Instruction *InsertBefore) const; diff --git a/llvm/include/llvm/IR/RuntimeLibcalls.td b/llvm/include/llvm/IR/RuntimeLibcalls.td index 6183a7e..a8b647c 100644 --- a/llvm/include/llvm/IR/RuntimeLibcalls.td +++ b/llvm/include/llvm/IR/RuntimeLibcalls.td @@ -405,17 +405,19 @@ def MIPS16_RET_DF : RuntimeLibcall; def MIPS16_RET_SC : RuntimeLibcall; def MIPS16_RET_SF : RuntimeLibcall; -multiclass LibmLongDoubleLibCall<string libcall_basename = !toupper(NAME), - string rtbasename = NAME> { +multiclass LibmLongDoubleLibCall<string libcall_basename = !toupper(!substr(NAME, 0, !sub(!size(NAME), 1))), + string rtname = NAME> { + + def NAME#"_f128" : RuntimeLibcallImpl<!cast<RuntimeLibcall>(libcall_basename#"_F128"), - !strconcat(rtbasename, "l")>; + rtname>; def NAME#"_ppcf128" : RuntimeLibcallImpl<!cast<RuntimeLibcall>(libcall_basename#"_PPCF128"), - !strconcat(rtbasename, "l")>; + rtname>; def NAME#"_f80" : RuntimeLibcallImpl<!cast<RuntimeLibcall>(libcall_basename#"_F80"), - !strconcat(rtbasename, "l")>; + rtname>; } // AArch64 calls @@ -765,19 +767,19 @@ def fmodl_ppc128 : RuntimeLibcallImpl<REM_PPCF128, "fmodl">; def fmaf : RuntimeLibcallImpl<FMA_F32>; def fma : RuntimeLibcallImpl<FMA_F64>; -defm fma : LibmLongDoubleLibCall; +defm fmal : LibmLongDoubleLibCall; def sqrtf : RuntimeLibcallImpl<SQRT_F32>; def sqrt : RuntimeLibcallImpl<SQRT_F64>; -defm sqrt : LibmLongDoubleLibCall; +defm sqrtl : LibmLongDoubleLibCall; def cbrtf : RuntimeLibcallImpl<CBRT_F32>; def cbrt : RuntimeLibcallImpl<CBRT_F64>; -defm cbrt : LibmLongDoubleLibCall; +defm cbrtl : LibmLongDoubleLibCall; def logf : RuntimeLibcallImpl<LOG_F32>; def log : RuntimeLibcallImpl<LOG_F64>; -defm log : LibmLongDoubleLibCall; +defm logl : LibmLongDoubleLibCall; def __logf_finite : RuntimeLibcallImpl<LOG_FINITE_F32>; def __log_finite : RuntimeLibcallImpl<LOG_FINITE_F64>; @@ -787,7 +789,7 @@ def __logl_finite_ppcf128 : RuntimeLibcallImpl<LOG_FINITE_PPCF128, "__logl_finit def log2f : RuntimeLibcallImpl<LOG2_F32>; def log2 : RuntimeLibcallImpl<LOG2_F64>; -defm log2 : LibmLongDoubleLibCall; +defm log2l : LibmLongDoubleLibCall; def __log2f_finite : RuntimeLibcallImpl<LOG2_FINITE_F32>; def __log2_finite : RuntimeLibcallImpl<LOG2_FINITE_F64>; @@ -797,7 +799,7 @@ def __log2l_finite_ppcf128 : RuntimeLibcallImpl<LOG2_FINITE_PPCF128, "__log2l_fi def log10f : RuntimeLibcallImpl<LOG10_F32>; def log10 : RuntimeLibcallImpl<LOG10_F64>; -defm log10 : LibmLongDoubleLibCall; +defm log10l : LibmLongDoubleLibCall; def __log10f_finite : RuntimeLibcallImpl<LOG10_FINITE_F32>; def __log10_finite : RuntimeLibcallImpl<LOG10_FINITE_F64>; @@ -807,7 +809,7 @@ def __log10l_finite_ppcf128 : RuntimeLibcallImpl<LOG10_FINITE_PPCF128, "__log10l def expf : RuntimeLibcallImpl<EXP_F32>; def exp : RuntimeLibcallImpl<EXP_F64>; -defm exp : LibmLongDoubleLibCall<"EXP", "exp">; +defm expl : LibmLongDoubleLibCall<"EXP">; def __expf_finite : RuntimeLibcallImpl<EXP_FINITE_F32>; def __exp_finite : RuntimeLibcallImpl<EXP_FINITE_F64>; @@ -817,7 +819,7 @@ def __expl_finite_ppcf128 : RuntimeLibcallImpl<EXP_FINITE_PPCF128, "__expl_finit def exp2f : RuntimeLibcallImpl<EXP2_F32>; def exp2 : RuntimeLibcallImpl<EXP2_F64>; -defm exp2 : LibmLongDoubleLibCall<"EXP2", "exp2">; +defm exp2l : LibmLongDoubleLibCall<"EXP2">; def __exp2f_finite : RuntimeLibcallImpl<EXP2_FINITE_F32>; def __exp2_finite : RuntimeLibcallImpl<EXP2_FINITE_F64>; @@ -827,47 +829,47 @@ def __exp2l_finite_ppcf128 : RuntimeLibcallImpl<EXP2_FINITE_PPCF128, "__exp2l_fi def sinf : RuntimeLibcallImpl<SIN_F32>; def sin : RuntimeLibcallImpl<SIN_F64>; -defm sin : LibmLongDoubleLibCall; +defm sinl : LibmLongDoubleLibCall; def cosf : RuntimeLibcallImpl<COS_F32>; def cos : RuntimeLibcallImpl<COS_F64>; -defm cos : LibmLongDoubleLibCall; +defm cosl : LibmLongDoubleLibCall; def tanf : RuntimeLibcallImpl<TAN_F32>; def tan : RuntimeLibcallImpl<TAN_F64>; -defm tan : LibmLongDoubleLibCall; +defm tanl : LibmLongDoubleLibCall; def sinhf : RuntimeLibcallImpl<SINH_F32>; def sinh : RuntimeLibcallImpl<SINH_F64>; -defm sinh : LibmLongDoubleLibCall; +defm sinhl : LibmLongDoubleLibCall; def coshf : RuntimeLibcallImpl<COSH_F32>; def cosh : RuntimeLibcallImpl<COSH_F64>; -defm cosh : LibmLongDoubleLibCall; +defm coshl : LibmLongDoubleLibCall; def tanhf : RuntimeLibcallImpl<TANH_F32>; def tanh : RuntimeLibcallImpl<TANH_F64>; -defm tanh : LibmLongDoubleLibCall; +defm tanhl : LibmLongDoubleLibCall; def asinf : RuntimeLibcallImpl<ASIN_F32>; def asin : RuntimeLibcallImpl<ASIN_F64>; -defm asin : LibmLongDoubleLibCall; +defm asinl : LibmLongDoubleLibCall; def acosf : RuntimeLibcallImpl<ACOS_F32>; def acos : RuntimeLibcallImpl<ACOS_F64>; -defm acos : LibmLongDoubleLibCall; +defm acosl : LibmLongDoubleLibCall; def atanf : RuntimeLibcallImpl<ATAN_F32>; def atan : RuntimeLibcallImpl<ATAN_F64>; -defm atan : LibmLongDoubleLibCall; +defm atanl : LibmLongDoubleLibCall; def atan2f : RuntimeLibcallImpl<ATAN2_F32>; def atan2 : RuntimeLibcallImpl<ATAN2_F64>; -defm atan2 : LibmLongDoubleLibCall; +defm atan2l : LibmLongDoubleLibCall; def powf : RuntimeLibcallImpl<POW_F32>; def pow : RuntimeLibcallImpl<POW_F64>; -defm pow : LibmLongDoubleLibCall; +defm powl : LibmLongDoubleLibCall; def __powf_finite : RuntimeLibcallImpl<POW_FINITE_F32>; def __pow_finite : RuntimeLibcallImpl<POW_FINITE_F64>; @@ -877,91 +879,91 @@ def __powl_finite_ppcf128 : RuntimeLibcallImpl<POW_FINITE_PPCF128, "__powl_finit def ceilf : RuntimeLibcallImpl<CEIL_F32>; def ceil : RuntimeLibcallImpl<CEIL_F64>; -defm ceil : LibmLongDoubleLibCall; +defm ceill : LibmLongDoubleLibCall; def truncf : RuntimeLibcallImpl<TRUNC_F32>; def trunc : RuntimeLibcallImpl<TRUNC_F64>; -defm trunc : LibmLongDoubleLibCall; +defm truncl : LibmLongDoubleLibCall; def rintf : RuntimeLibcallImpl<RINT_F32>; def rint : RuntimeLibcallImpl<RINT_F64>; -defm rint : LibmLongDoubleLibCall; +defm rintl : LibmLongDoubleLibCall; def nearbyintf : RuntimeLibcallImpl<NEARBYINT_F32>; def nearbyint : RuntimeLibcallImpl<NEARBYINT_F64>; -defm nearbyint : LibmLongDoubleLibCall; +defm nearbyintl : LibmLongDoubleLibCall; def roundf : RuntimeLibcallImpl<ROUND_F32>; def round : RuntimeLibcallImpl<ROUND_F64>; -defm round : LibmLongDoubleLibCall; +defm roundl : LibmLongDoubleLibCall; def roundevenf : RuntimeLibcallImpl<ROUNDEVEN_F32>; def roundeven : RuntimeLibcallImpl<ROUNDEVEN_F64>; -defm roundeven : LibmLongDoubleLibCall; +defm roundevenl : LibmLongDoubleLibCall; def floorf : RuntimeLibcallImpl<FLOOR_F32>; def floor : RuntimeLibcallImpl<FLOOR_F64>; -defm floor : LibmLongDoubleLibCall; +defm floorl : LibmLongDoubleLibCall; def copysignf : RuntimeLibcallImpl<COPYSIGN_F32>; def copysign : RuntimeLibcallImpl<COPYSIGN_F64>; -defm copysign : LibmLongDoubleLibCall; +defm copysignl : LibmLongDoubleLibCall; def fminf : RuntimeLibcallImpl<FMIN_F32>; def fmin : RuntimeLibcallImpl<FMIN_F64>; -defm fmin : LibmLongDoubleLibCall; +defm fminl : LibmLongDoubleLibCall; def fmaxf : RuntimeLibcallImpl<FMAX_F32>; def fmax : RuntimeLibcallImpl<FMAX_F64>; -defm fmax : LibmLongDoubleLibCall; +defm fmaxl : LibmLongDoubleLibCall; def fminimumf : RuntimeLibcallImpl<FMINIMUM_F32>; def fminimum : RuntimeLibcallImpl<FMINIMUM_F64>; -defm fminimum : LibmLongDoubleLibCall; +defm fminimuml : LibmLongDoubleLibCall; def fmaximumf : RuntimeLibcallImpl<FMAXIMUM_F32>; def fmaximum : RuntimeLibcallImpl<FMAXIMUM_F64>; -defm fmaximum : LibmLongDoubleLibCall; +defm fmaximuml : LibmLongDoubleLibCall; def fminimum_numf : RuntimeLibcallImpl<FMINIMUM_NUM_F32>; def fminimum_num : RuntimeLibcallImpl<FMINIMUM_NUM_F64>; -defm fminimum_num : LibmLongDoubleLibCall; +defm fminimum_numl : LibmLongDoubleLibCall; def fmaximum_numf : RuntimeLibcallImpl<FMAXIMUM_NUM_F32>; def fmaximum_num : RuntimeLibcallImpl<FMAXIMUM_NUM_F64>; -defm fmaximum_num : LibmLongDoubleLibCall; +defm fmaximum_numl : LibmLongDoubleLibCall; def lroundf : RuntimeLibcallImpl<LROUND_F32>; def lround : RuntimeLibcallImpl<LROUND_F64>; -defm lround : LibmLongDoubleLibCall; +defm lroundl : LibmLongDoubleLibCall; def llroundf : RuntimeLibcallImpl<LLROUND_F32>; def llround : RuntimeLibcallImpl<LLROUND_F64>; -defm llround : LibmLongDoubleLibCall; +defm llroundl : LibmLongDoubleLibCall; def lrintf : RuntimeLibcallImpl<LRINT_F32>; def lrint : RuntimeLibcallImpl<LRINT_F64>; -defm lrint : LibmLongDoubleLibCall; +defm lrintl : LibmLongDoubleLibCall; def llrintf : RuntimeLibcallImpl<LLRINT_F32>; def llrint : RuntimeLibcallImpl<LLRINT_F64>; -defm llrint : LibmLongDoubleLibCall; +defm llrintl : LibmLongDoubleLibCall; def ldexpf : RuntimeLibcallImpl<LDEXP_F32>; def ldexp : RuntimeLibcallImpl<LDEXP_F64>; -defm ldexp : LibmLongDoubleLibCall; +defm ldexpl : LibmLongDoubleLibCall; def frexpf : RuntimeLibcallImpl<FREXP_F32>; def frexp : RuntimeLibcallImpl<FREXP_F64>; -defm frexp : LibmLongDoubleLibCall; +defm frexpl : LibmLongDoubleLibCall; def sincospif : RuntimeLibcallImpl<SINCOSPI_F32>; def sincospi : RuntimeLibcallImpl<SINCOSPI_F64>; -defm sincospi : LibmLongDoubleLibCall; +defm sincospil : LibmLongDoubleLibCall; def modff : RuntimeLibcallImpl<MODF_F32>; def modf : RuntimeLibcallImpl<MODF_F64>; -defm modf : LibmLongDoubleLibCall; +defm modfl : LibmLongDoubleLibCall; // Floating point environment def fegetenv : RuntimeLibcallImpl<FEGETENV>; @@ -1033,7 +1035,7 @@ def __sincos_stret : RuntimeLibcallImpl<SINCOS_STRET_F64>; def sincosf : RuntimeLibcallImpl<SINCOS_F32>; def sincos : RuntimeLibcallImpl<SINCOS_F64>; -defm sincos : LibmLongDoubleLibCall; +defm sincosl : LibmLongDoubleLibCall; def bzero : RuntimeLibcallImpl<BZERO>; def __bzero : RuntimeLibcallImpl<BZERO>; @@ -1198,9 +1200,9 @@ defvar SecurityCheckCookieIfWinMSVC = defvar LibmHasSinCosF32 = LibcallImpls<(add sincosf), hasSinCos>; defvar LibmHasSinCosF64 = LibcallImpls<(add sincos), hasSinCos>; -defvar LibmHasSinCosF80 = LibcallImpls<(add sincos_f80), hasSinCos>; -defvar LibmHasSinCosF128 = LibcallImpls<(add sincos_f128), hasSinCos>; -defvar LibmHasSinCosPPCF128 = LibcallImpls<(add sincos_ppcf128), hasSinCos>; +defvar LibmHasSinCosF80 = LibcallImpls<(add sincosl_f80), hasSinCos>; +defvar LibmHasSinCosF128 = LibcallImpls<(add sincosl_f128), hasSinCos>; +defvar LibmHasSinCosPPCF128 = LibcallImpls<(add sincosl_ppcf128), hasSinCos>; defvar LibmHasExp10F32 = LibcallImpls<(add exp10f), hasExp10>; defvar LibmHasExp10F64 = LibcallImpls<(add exp10), hasExp10>; @@ -1214,8 +1216,8 @@ defvar DefaultLibmExp10 = [ defvar WindowsMathRemovals = [ - ldexpf, ldexp_f80, ldexp_f128, ldexp_ppcf128, - frexpf, frexp_f80, frexp_f128, frexp_ppcf128 + ldexpf, ldexpl_f80, ldexpl_f128, ldexpl_ppcf128, + frexpf, frexpl_f80, frexpl_f128, frexpl_ppcf128 ]; defvar MostPowI = !listremove(PowiLibcallImpls, [__powitf2_f128, __powitf2_ppc128]); @@ -1233,11 +1235,11 @@ defvar WinDefaultLibcallImpls = (add WinDefaultLibcallImplsBaseList, defvar LibmHasFrexpF32 = LibcallImpls<(add frexpf), isNotOSWindowsOrIsCygwinMinGW>; defvar LibmHasLdexpF32 = LibcallImpls<(add ldexpf), isNotOSWindowsOrIsCygwinMinGW>; -defvar LibmHasFrexpF80 = LibcallImpls<(add frexp_f80), isNotOSWindowsOrIsCygwinMinGW>; -defvar LibmHasLdexpF80 = LibcallImpls<(add ldexp_f80), isNotOSWindowsOrIsCygwinMinGW>; +defvar LibmHasFrexpF80 = LibcallImpls<(add frexpl_f80), isNotOSWindowsOrIsCygwinMinGW>; +defvar LibmHasLdexpF80 = LibcallImpls<(add ldexpl_f80), isNotOSWindowsOrIsCygwinMinGW>; -defvar LibmHasFrexpF128 = LibcallImpls<(add frexp_f128), isNotOSWindowsOrIsCygwinMinGW>; -defvar LibmHasLdexpF128 = LibcallImpls<(add ldexp_f128), isNotOSWindowsOrIsCygwinMinGW>; +defvar LibmHasFrexpF128 = LibcallImpls<(add frexpl_f128), isNotOSWindowsOrIsCygwinMinGW>; +defvar LibmHasLdexpF128 = LibcallImpls<(add ldexpl_f128), isNotOSWindowsOrIsCygwinMinGW>; defvar has__stack_chk_fail = LibcallImpls<(add __stack_chk_fail), isNotOSOpenBSD>; defvar has__stack_chk_guard = @@ -2459,7 +2461,7 @@ defvar X86CommonLibcalls = LibcallImpls<(add __bzero), darwinHas__bzero>, LibmHasFrexpF32, LibmHasLdexpF32, LibmHasFrexpF80, LibmHasLdexpF80, - LibcallImpls<(add frexp_f128, ldexp_f128, exp10l_f128), hasExpFrexplLdexplF128>, + LibcallImpls<(add frexpl_f128, ldexpl_f128, exp10l_f128), hasExpFrexplLdexplF128>, DefaultRuntimeLibcallImpls_f80, LibmHasExp10F32, LibmHasExp10F64, LibmHasExp10F80, LibcallImpls<(add MostPowI), isNotOSMSVCRT>, diff --git a/llvm/include/llvm/IR/Value.h b/llvm/include/llvm/IR/Value.h index 04d0391..58822a0 100644 --- a/llvm/include/llvm/IR/Value.h +++ b/llvm/include/llvm/IR/Value.h @@ -484,8 +484,8 @@ public: /// Remove every uses that can safely be removed. /// /// This will remove for example uses in llvm.assume. - /// This should be used when performing want to perform a tranformation but - /// some Droppable uses pervent it. + /// This should be used when performing want to perform a transformation but + /// some Droppable uses prevent it. /// This function optionally takes a filter to only remove some droppable /// uses. LLVM_ABI void diff --git a/llvm/include/llvm/LTO/LTO.h b/llvm/include/llvm/LTO/LTO.h index 3a9a7f7..000472f 100644 --- a/llvm/include/llvm/LTO/LTO.h +++ b/llvm/include/llvm/LTO/LTO.h @@ -105,12 +105,6 @@ setupStatsFile(StringRef StatsFilename); /// ordered indices to elements in the input array. LLVM_ABI std::vector<int> generateModulesOrdering(ArrayRef<BitcodeModule *> R); -/// Updates MemProf attributes (and metadata) based on whether the index -/// has recorded that we are linking with allocation libraries containing -/// the necessary APIs for downstream transformations. -LLVM_ABI void updateMemProfAttributes(Module &Mod, - const ModuleSummaryIndex &Index); - class LTO; struct SymbolResolution; diff --git a/llvm/include/llvm/Support/DOTGraphTraits.h b/llvm/include/llvm/Support/DOTGraphTraits.h index ffa9abe..bf30aa4 100644 --- a/llvm/include/llvm/Support/DOTGraphTraits.h +++ b/llvm/include/llvm/Support/DOTGraphTraits.h @@ -164,7 +164,7 @@ public: /// template <typename Ty> struct DOTGraphTraits : public DefaultDOTGraphTraits { - DOTGraphTraits (bool simple=false) : DefaultDOTGraphTraits (simple) {} + using DefaultDOTGraphTraits::DefaultDOTGraphTraits; }; } // End llvm namespace diff --git a/llvm/include/llvm/Support/MathExtras.h b/llvm/include/llvm/Support/MathExtras.h index c2716a9..41232335 100644 --- a/llvm/include/llvm/Support/MathExtras.h +++ b/llvm/include/llvm/Support/MathExtras.h @@ -13,6 +13,7 @@ #ifndef LLVM_SUPPORT_MATHEXTRAS_H #define LLVM_SUPPORT_MATHEXTRAS_H +#include "llvm/ADT/STLForwardCompat.h" #include "llvm/ADT/bit.h" #include "llvm/Support/Compiler.h" #include <cassert> @@ -42,38 +43,28 @@ using common_sint = /// Mathematical constants. namespace numbers { -// TODO: Track C++20 std::numbers. // clang-format off -constexpr double e = 0x1.5bf0a8b145769P+1, // (2.7182818284590452354) https://oeis.org/A001113 - egamma = 0x1.2788cfc6fb619P-1, // (.57721566490153286061) https://oeis.org/A001620 - ln2 = 0x1.62e42fefa39efP-1, // (.69314718055994530942) https://oeis.org/A002162 - ln10 = 0x1.26bb1bbb55516P+1, // (2.3025850929940456840) https://oeis.org/A002392 - log2e = 0x1.71547652b82feP+0, // (1.4426950408889634074) - log10e = 0x1.bcb7b1526e50eP-2, // (.43429448190325182765) - pi = 0x1.921fb54442d18P+1, // (3.1415926535897932385) https://oeis.org/A000796 - inv_pi = 0x1.45f306dc9c883P-2, // (.31830988618379067154) https://oeis.org/A049541 - sqrtpi = 0x1.c5bf891b4ef6bP+0, // (1.7724538509055160273) https://oeis.org/A002161 - inv_sqrtpi = 0x1.20dd750429b6dP-1, // (.56418958354775628695) https://oeis.org/A087197 - sqrt2 = 0x1.6a09e667f3bcdP+0, // (1.4142135623730950488) https://oeis.org/A00219 - inv_sqrt2 = 0x1.6a09e667f3bcdP-1, // (.70710678118654752440) - sqrt3 = 0x1.bb67ae8584caaP+0, // (1.7320508075688772935) https://oeis.org/A002194 - inv_sqrt3 = 0x1.279a74590331cP-1, // (.57735026918962576451) - phi = 0x1.9e3779b97f4a8P+0; // (1.6180339887498948482) https://oeis.org/A001622 -constexpr float ef = 0x1.5bf0a8P+1F, // (2.71828183) https://oeis.org/A001113 - egammaf = 0x1.2788d0P-1F, // (.577215665) https://oeis.org/A001620 - ln2f = 0x1.62e430P-1F, // (.693147181) https://oeis.org/A002162 - ln10f = 0x1.26bb1cP+1F, // (2.30258509) https://oeis.org/A002392 - log2ef = 0x1.715476P+0F, // (1.44269504) - log10ef = 0x1.bcb7b2P-2F, // (.434294482) - pif = 0x1.921fb6P+1F, // (3.14159265) https://oeis.org/A000796 - inv_pif = 0x1.45f306P-2F, // (.318309886) https://oeis.org/A049541 - sqrtpif = 0x1.c5bf8aP+0F, // (1.77245385) https://oeis.org/A002161 - inv_sqrtpif = 0x1.20dd76P-1F, // (.564189584) https://oeis.org/A087197 - sqrt2f = 0x1.6a09e6P+0F, // (1.41421356) https://oeis.org/A002193 - inv_sqrt2f = 0x1.6a09e6P-1F, // (.707106781) - sqrt3f = 0x1.bb67aeP+0F, // (1.73205081) https://oeis.org/A002194 - inv_sqrt3f = 0x1.279a74P-1F, // (.577350269) - phif = 0x1.9e377aP+0F; // (1.61803399) https://oeis.org/A001622 +inline constexpr float ef = e_v<float>; +inline constexpr float egammaf = egamma_v<float>; +inline constexpr float ln2f = ln2_v<float>; +inline constexpr float ln10f = ln10_v<float>; +inline constexpr float log2ef = log2e_v<float>; +inline constexpr float log10ef = log10e_v<float>; +inline constexpr float pif = pi_v<float>; +inline constexpr float inv_pif = inv_pi_v<float>; +inline constexpr float inv_sqrtpif = inv_sqrtpi_v<float>; +inline constexpr float sqrt2f = sqrt2_v<float>; +inline constexpr float inv_sqrt2f = inv_sqrt2_v<float>; +inline constexpr float sqrt3f = sqrt3_v<float>; +inline constexpr float inv_sqrt3f = inv_sqrt3_v<float>; +inline constexpr float phif = phi_v<float>; + +// sqrtpi is not in C++20 std::numbers. +template <typename T, typename = std::enable_if_t<std::is_floating_point_v<T>>> +inline constexpr T sqrtpi_v = T(0x1.c5bf891b4ef6bP+0); // (1.7724538509055160273) https://oeis.org/A002161 +inline constexpr double sqrtpi = sqrtpi_v<double>; +inline constexpr float sqrtpif = sqrtpi_v<float>; + // These string literals are taken from below: // https://github.com/bminor/glibc/blob/8543577b04ded6d979ffcc5a818930e4d74d0645/math/math.h#L1215-L1229 constexpr const char *pis = "3.141592653589793238462643383279502884", diff --git a/llvm/include/llvm/Support/X86DisassemblerDecoderCommon.h b/llvm/include/llvm/Support/X86DisassemblerDecoderCommon.h index faaff4a..4aa6c01 100644 --- a/llvm/include/llvm/Support/X86DisassemblerDecoderCommon.h +++ b/llvm/include/llvm/Support/X86DisassemblerDecoderCommon.h @@ -121,6 +121,7 @@ enum attributeBits { "The Dynamic Duo! Prefer over all else because this changes " \ "most operands' meaning") \ ENUM_ENTRY(IC_64BIT_REX2, 2, "requires a REX2 prefix") \ + ENUM_ENTRY(IC_64BIT_REX2_REXW, 3, "requires a REX2 and the W prefix") \ ENUM_ENTRY(IC_VEX, 1, "requires a VEX prefix") \ ENUM_ENTRY(IC_VEX_XS, 2, "requires VEX and the XS prefix") \ ENUM_ENTRY(IC_VEX_XD, 2, "requires VEX and the XD prefix") \ diff --git a/llvm/include/llvm/TargetParser/X86TargetParser.def b/llvm/include/llvm/TargetParser/X86TargetParser.def index 254587b..a94eab1 100644 --- a/llvm/include/llvm/TargetParser/X86TargetParser.def +++ b/llvm/include/llvm/TargetParser/X86TargetParser.def @@ -108,6 +108,7 @@ X86_CPU_SUBTYPE(INTEL_COREI7_ARROWLAKE_S, "arrowlake-s") X86_CPU_SUBTYPE(INTEL_COREI7_PANTHERLAKE, "pantherlake") X86_CPU_SUBTYPE(AMDFAM1AH_ZNVER5, "znver5") X86_CPU_SUBTYPE(INTEL_COREI7_DIAMONDRAPIDS, "diamondrapids") +X86_CPU_SUBTYPE(INTEL_COREI7_NOVALAKE, "novalake") // Alternate names supported by __builtin_cpu_is and target multiversioning. X86_CPU_SUBTYPE_ALIAS(INTEL_COREI7_ALDERLAKE, "raptorlake") diff --git a/llvm/include/llvm/TargetParser/X86TargetParser.h b/llvm/include/llvm/TargetParser/X86TargetParser.h index e4c43cd..80f3d35 100644 --- a/llvm/include/llvm/TargetParser/X86TargetParser.h +++ b/llvm/include/llvm/TargetParser/X86TargetParser.h @@ -117,6 +117,7 @@ enum CPUKind { CK_Lunarlake, CK_Pantherlake, CK_Wildcatlake, + CK_Novalake, CK_Sierraforest, CK_Grandridge, CK_Graniterapids, diff --git a/llvm/include/llvm/Transforms/IPO/MemProfContextDisambiguation.h b/llvm/include/llvm/Transforms/IPO/MemProfContextDisambiguation.h index f2de083..576f1eb 100644 --- a/llvm/include/llvm/Transforms/IPO/MemProfContextDisambiguation.h +++ b/llvm/include/llvm/Transforms/IPO/MemProfContextDisambiguation.h @@ -95,6 +95,16 @@ public: function_ref<bool(GlobalValue::GUID, const GlobalValueSummary *)> isPrevailing); }; + +/// Strips MemProf attributes and metadata. Can be invoked by the pass pipeline +/// when we don't have an index that has recorded that we are linking with +/// allocation libraries containing the necessary APIs for downstream +/// transformations. +class MemProfRemoveInfo : public PassInfoMixin<MemProfRemoveInfo> { +public: + PreservedAnalyses run(Module &M, ModuleAnalysisManager &AM); +}; + } // end namespace llvm #endif // LLVM_TRANSFORMS_IPO_MEMPROF_CONTEXT_DISAMBIGUATION_H diff --git a/llvm/include/llvm/Transforms/Scalar/LoopPassManager.h b/llvm/include/llvm/Transforms/Scalar/LoopPassManager.h index 750f954..1842d2d 100644 --- a/llvm/include/llvm/Transforms/Scalar/LoopPassManager.h +++ b/llvm/include/llvm/Transforms/Scalar/LoopPassManager.h @@ -404,10 +404,8 @@ public: explicit FunctionToLoopPassAdaptor(std::unique_ptr<PassConceptT> Pass, bool UseMemorySSA = false, - bool UseBlockFrequencyInfo = false, bool LoopNestMode = false) : Pass(std::move(Pass)), UseMemorySSA(UseMemorySSA), - UseBlockFrequencyInfo(UseBlockFrequencyInfo), LoopNestMode(LoopNestMode) { LoopCanonicalizationFPM.addPass(LoopSimplifyPass()); LoopCanonicalizationFPM.addPass(LCSSAPass()); @@ -429,7 +427,6 @@ private: FunctionPassManager LoopCanonicalizationFPM; bool UseMemorySSA = false; - bool UseBlockFrequencyInfo = false; const bool LoopNestMode; }; @@ -442,8 +439,7 @@ private: /// \c LoopPassManager and the returned adaptor will be in loop-nest mode. template <typename LoopPassT> inline FunctionToLoopPassAdaptor -createFunctionToLoopPassAdaptor(LoopPassT &&Pass, bool UseMemorySSA = false, - bool UseBlockFrequencyInfo = false) { +createFunctionToLoopPassAdaptor(LoopPassT &&Pass, bool UseMemorySSA = false) { if constexpr (is_detected<HasRunOnLoopT, LoopPassT>::value) { using PassModelT = detail::PassModel<Loop, LoopPassT, LoopAnalysisManager, @@ -453,7 +449,7 @@ createFunctionToLoopPassAdaptor(LoopPassT &&Pass, bool UseMemorySSA = false, return FunctionToLoopPassAdaptor( std::unique_ptr<FunctionToLoopPassAdaptor::PassConceptT>( new PassModelT(std::forward<LoopPassT>(Pass))), - UseMemorySSA, UseBlockFrequencyInfo, false); + UseMemorySSA, false); } else { LoopPassManager LPM; LPM.addPass(std::forward<LoopPassT>(Pass)); @@ -465,7 +461,7 @@ createFunctionToLoopPassAdaptor(LoopPassT &&Pass, bool UseMemorySSA = false, return FunctionToLoopPassAdaptor( std::unique_ptr<FunctionToLoopPassAdaptor::PassConceptT>( new PassModelT(std::move(LPM))), - UseMemorySSA, UseBlockFrequencyInfo, true); + UseMemorySSA, true); } } @@ -474,8 +470,7 @@ createFunctionToLoopPassAdaptor(LoopPassT &&Pass, bool UseMemorySSA = false, template <> inline FunctionToLoopPassAdaptor createFunctionToLoopPassAdaptor<LoopPassManager>(LoopPassManager &&LPM, - bool UseMemorySSA, - bool UseBlockFrequencyInfo) { + bool UseMemorySSA) { // Check if LPM contains any loop pass and if it does not, returns an adaptor // in loop-nest mode. using PassModelT = @@ -487,7 +482,7 @@ createFunctionToLoopPassAdaptor<LoopPassManager>(LoopPassManager &&LPM, return FunctionToLoopPassAdaptor( std::unique_ptr<FunctionToLoopPassAdaptor::PassConceptT>( new PassModelT(std::move(LPM))), - UseMemorySSA, UseBlockFrequencyInfo, LoopNestMode); + UseMemorySSA, LoopNestMode); } /// Pass for printing a loop's contents as textual IR. |