From 1de062e1e530408f5a06466742b26722c230c024 Mon Sep 17 00:00:00 2001 From: Dimitry Andric Date: Sat, 2 Jan 2021 22:13:33 +0100 Subject: [PATCH] Vendor import of llvm-project branch release/11.x llvmorg-11.0.1-rc2-0-g43ff75f2c3f (aka 11.0.1 rc2). --- clang/include/clang/AST/DeclTemplate.h | 0 clang/include/clang/Basic/Attr.td | 10 +- clang/include/clang/Basic/AttrDocs.td | 26 ++ clang/include/clang/Basic/Builtins.def | 27 ++- clang/include/clang/Basic/Builtins.h | 7 + .../clang/Basic/DiagnosticDriverKinds.td | 2 +- clang/include/clang/Basic/IdentifierTable.h | 12 - clang/include/clang/Format/Format.h | 0 clang/include/clang/Sema/Sema.h | 2 + .../Refactoring/Extract/SourceExtraction.h | 0 clang/lib/AST/Decl.cpp | 28 +-- clang/lib/AST/DeclTemplate.cpp | 0 clang/lib/CodeGen/CGBuiltin.cpp | 6 +- clang/lib/CodeGen/CGObjCGNU.cpp | 13 - clang/lib/CodeGen/TargetInfo.cpp | 16 +- clang/lib/Driver/ToolChains/Cuda.cpp | 115 +++++++-- clang/lib/Driver/ToolChains/Cuda.h | 3 - clang/lib/Format/FormatToken.h | 6 - clang/lib/Format/UnwrappedLineParser.cpp | 2 - clang/lib/Format/WhitespaceManager.cpp | 10 +- .../Headers/__clang_cuda_complex_builtins.h | 6 +- clang/lib/Headers/cuda_wrappers/new | 8 + clang/lib/Headers/intrin.h | 149 +++++------- clang/lib/Headers/openmp_wrappers/complex | 2 + clang/lib/Headers/openmp_wrappers/complex.h | 2 + clang/lib/Sema/SemaChecking.cpp | 5 - clang/lib/Sema/SemaConcept.cpp | 0 clang/lib/Sema/SemaDecl.cpp | 129 +++++----- clang/lib/Sema/SemaDeclAttr.cpp | 6 +- clang/lib/Sema/SemaExpr.cpp | 3 +- clang/lib/Sema/SemaLookup.cpp | 7 +- clang/lib/Sema/SemaTemplate.cpp | 3 +- .../lib/Sema/SemaTemplateInstantiateDecl.cpp | 13 +- clang/lib/Sema/UsedDeclVisitor.h | 11 +- clang/lib/Serialization/ASTReader.cpp | 12 +- clang/lib/Serialization/ASTWriter.cpp | 6 +- clang/lib/StaticAnalyzer/Checkers/Yaml.h | 0 .../sanitizer_common_interceptors_ioctl.inc | 0 lld/COFF/DebugTypes.cpp | 5 +- lld/ELF/LinkerScript.cpp | 7 +- lldb/include/lldb/Core/ModuleList.h | 19 +- lldb/include/lldb/Target/Platform.h | 9 +- lldb/source/Core/ModuleList.cpp | 39 +-- .../Process/minidump/ProcessMinidump.cpp | 152 ++++++++++-- .../Process/minidump/ProcessMinidump.h | 4 + lldb/source/Target/Platform.cpp | 43 ++-- lldb/source/Target/Target.cpp | 96 ++++++-- llvm/include/llvm/ADT/ImmutableMap.h | 2 +- llvm/include/llvm/BinaryFormat/ELF.h | 6 + .../Transforms/IPO/DeadArgumentElimination.h | 1 + .../DebugInfo/DWARF/DWARFDebugArangeSet.cpp | 13 +- llvm/lib/DebugInfo/DWARF/DWARFUnitIndex.cpp | 8 +- .../RuntimeDyld/RuntimeDyld.cpp | 19 +- llvm/lib/IR/ConstantFold.cpp | 2 +- llvm/lib/MC/MCObjectFileInfo.cpp | 2 + llvm/lib/MC/MCParser/ELFAsmParser.cpp | 7 +- llvm/lib/MCA/HardwareUnits/LSUnit.cpp | 2 + llvm/lib/Support/Host.cpp | 5 +- llvm/lib/Support/Windows/Path.inc | 21 +- llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp | 15 ++ .../AArch64/AArch64CallingConvention.cpp | 9 + .../Target/AArch64/AArch64ISelLowering.cpp | 17 +- .../AArch64/AsmParser/AArch64AsmParser.cpp | 30 +++ .../GISel/AArch64InstructionSelector.cpp | 134 ++++------- .../AArch64/GISel/AArch64LegalizerInfo.cpp | 38 +-- .../AArch64/GISel/AArch64RegisterBankInfo.cpp | 1 + .../MCTargetDesc/AArch64ELFStreamer.cpp | 8 + .../MCTargetDesc/AArch64TargetStreamer.h | 4 + llvm/lib/Target/AMDGPU/SIPreEmitPeephole.cpp | 24 +- llvm/lib/Target/BPF/BPFISelDAGToDAG.cpp | 4 +- llvm/lib/Target/Hexagon/HexagonAsmPrinter.h | 0 llvm/lib/Target/Sparc/LeonFeatures.td | 0 llvm/lib/Target/Sparc/LeonPasses.cpp | 0 llvm/lib/Target/Sparc/LeonPasses.h | 0 llvm/lib/Target/Sparc/SparcAsmPrinter.cpp | 2 +- llvm/lib/Target/Sparc/SparcSchedule.td | 0 .../WebAssembly/WebAssemblyFastISel.cpp | 17 +- .../WebAssembly/WebAssemblyInstrInfo.td | 4 +- .../WebAssembly/WebAssemblyRegisterInfo.cpp | 10 +- llvm/lib/Target/X86/X86EvexToVex.cpp | 0 llvm/lib/Target/X86/X86FrameLowering.cpp | 224 ++++++++++++++++-- llvm/lib/Target/X86/X86FrameLowering.h | 8 +- llvm/lib/Target/X86/X86ISelLowering.cpp | 29 ++- llvm/lib/Target/X86/X86SchedBroadwell.td | 0 llvm/lib/Target/X86/X86SchedSkylakeServer.td | 0 .../IPO/DeadArgumentElimination.cpp | 29 ++- .../InstCombine/InstCombineSelect.cpp | 20 +- .../Instrumentation/AddressSanitizer.cpp | 20 +- 88 files changed, 1155 insertions(+), 601 deletions(-) mode change 100644 => 100755 clang/include/clang/AST/DeclTemplate.h mode change 100644 => 100755 clang/include/clang/Format/Format.h mode change 100755 => 100644 clang/include/clang/Tooling/Refactoring/Extract/SourceExtraction.h mode change 100644 => 100755 clang/lib/AST/DeclTemplate.cpp mode change 100644 => 100755 clang/lib/Sema/SemaConcept.cpp mode change 100755 => 100644 clang/lib/StaticAnalyzer/Checkers/Yaml.h mode change 100755 => 100644 compiler-rt/lib/sanitizer_common/sanitizer_common_interceptors_ioctl.inc mode change 100755 => 100644 llvm/lib/Target/Hexagon/HexagonAsmPrinter.h mode change 100755 => 100644 llvm/lib/Target/Sparc/LeonFeatures.td mode change 100755 => 100644 llvm/lib/Target/Sparc/LeonPasses.cpp mode change 100755 => 100644 llvm/lib/Target/Sparc/LeonPasses.h mode change 100755 => 100644 llvm/lib/Target/Sparc/SparcSchedule.td mode change 100755 => 100644 llvm/lib/Target/X86/X86EvexToVex.cpp mode change 100755 => 100644 llvm/lib/Target/X86/X86SchedBroadwell.td mode change 100755 => 100644 llvm/lib/Target/X86/X86SchedSkylakeServer.td diff --git a/clang/include/clang/AST/DeclTemplate.h b/clang/include/clang/AST/DeclTemplate.h old mode 100644 new mode 100755 diff --git a/clang/include/clang/Basic/Attr.td b/clang/include/clang/Basic/Attr.td index bc4a380545af..60eaee7839e2 100644 --- a/clang/include/clang/Basic/Attr.td +++ b/clang/include/clang/Basic/Attr.td @@ -2115,7 +2115,7 @@ def InitPriority : InheritableAttr { let Spellings = [GCC<"init_priority", /*AllowInC*/0>]; let Args = [UnsignedArgument<"Priority">]; let Subjects = SubjectList<[Var], ErrorDiag>; - let Documentation = [Undocumented]; + let Documentation = [InitPriorityDocs]; } def Section : InheritableAttr { @@ -3444,3 +3444,11 @@ def ReleaseHandle : InheritableParamAttr { let Subjects = SubjectList<[ParmVar]>; let Documentation = [ReleaseHandleDocs]; } + +def Builtin : InheritableAttr { + let Spellings = []; + let Args = [UnsignedArgument<"ID">]; + let Subjects = SubjectList<[Function]>; + let SemaHandler = 0; + let Documentation = [Undocumented]; +} diff --git a/clang/include/clang/Basic/AttrDocs.td b/clang/include/clang/Basic/AttrDocs.td index 3cba3a3d96f9..833127ed44eb 100644 --- a/clang/include/clang/Basic/AttrDocs.td +++ b/clang/include/clang/Basic/AttrDocs.td @@ -57,6 +57,32 @@ global variable or function should be in after translation. let Heading = "section, __declspec(allocate)"; } +def InitPriorityDocs : Documentation { + let Category = DocCatVariable; + let Content = [{ +In C++, the order in which global variables are initialized across translation +units is unspecified, unlike the ordering within a single translation unit. The +``init_priority`` attribute allows you to specify a relative ordering for the +initialization of objects declared at namespace scope in C++. The priority is +given as an integer constant expression between 101 and 65535 (inclusive). +Priorities outside of that range are reserved for use by the implementation. A +lower value indicates a higher priority of initialization. Note that only the +relative ordering of values is important. For example: + +.. code-block:: c++ + + struct SomeType { SomeType(); }; + __attribute__((init_priority(200))) SomeType Obj1; + __attribute__((init_priority(101))) SomeType Obj2; + +``Obj1`` will be initialized *before* ``Obj2`` despite the usual order of +initialization being the opposite. + +This attribute is only supported for C++ and Objective-C++ and is ignored in +other language modes. + }]; +} + def InitSegDocs : Documentation { let Category = DocCatVariable; let Content = [{ diff --git a/clang/include/clang/Basic/Builtins.def b/clang/include/clang/Basic/Builtins.def index 1416a64543a4..5463b7dfc18c 100644 --- a/clang/include/clang/Basic/Builtins.def +++ b/clang/include/clang/Basic/Builtins.def @@ -75,6 +75,9 @@ // U -> pure // c -> const // t -> signature is meaningless, use custom typechecking +// T -> type is not important to semantic analysis and codegen; recognize as +// builtin even if type doesn't match signature, and don't warn if we +// can't be sure the type is right // F -> this is a libc/libm function with a '__builtin_' prefix added. // f -> this is a libc/libm function without the '__builtin_' prefix. It can // be followed by ':headername:' to state which header this function @@ -893,7 +896,7 @@ LANGBUILTIN(__va_start, "vc**.", "nt", ALL_MS_LANGUAGES) LANGBUILTIN(__fastfail, "vUi", "nr", ALL_MS_LANGUAGES) // Microsoft library builtins. -LIBBUILTIN(_setjmpex, "iJ", "fj", "setjmpex.h", ALL_MS_LANGUAGES) +LIBBUILTIN(_setjmpex, "iJ", "fjT", "setjmpex.h", ALL_MS_LANGUAGES) // C99 library functions // C99 stdarg.h @@ -987,8 +990,8 @@ LIBBUILTIN(wmemmove,"w*w*wC*z", "f", "wchar.h", ALL_LANGUAGES) // In some systems setjmp is a macro that expands to _setjmp. We undefine // it here to avoid having two identical LIBBUILTIN entries. #undef setjmp -LIBBUILTIN(setjmp, "iJ", "fj", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(longjmp, "vJi", "fr", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(setjmp, "iJ", "fjT", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(longjmp, "vJi", "frT", "setjmp.h", ALL_LANGUAGES) // Non-C library functions, active in GNU mode only. // Functions with (returns_twice) attribute (marked as "j") are still active in @@ -1015,20 +1018,22 @@ LIBBUILTIN(strcasecmp, "icC*cC*", "f", "strings.h", ALL_GNU_LANGUAGES) LIBBUILTIN(strncasecmp, "icC*cC*z", "f", "strings.h", ALL_GNU_LANGUAGES) // POSIX unistd.h LIBBUILTIN(_exit, "vi", "fr", "unistd.h", ALL_GNU_LANGUAGES) -LIBBUILTIN(vfork, "p", "fj", "unistd.h", ALL_LANGUAGES) +LIBBUILTIN(vfork, "p", "fjT", "unistd.h", ALL_LANGUAGES) // POSIX pthread.h +// FIXME: Should specify argument types. LIBBUILTIN(pthread_create, "", "fC<2,3>", "pthread.h", ALL_GNU_LANGUAGES) // POSIX setjmp.h -LIBBUILTIN(_setjmp, "iJ", "fj", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(__sigsetjmp, "iSJi", "fj", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(sigsetjmp, "iSJi", "fj", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(savectx, "iJ", "fj", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(getcontext, "iK*", "fj", "setjmp.h", ALL_LANGUAGES) +// FIXME: MinGW _setjmp has an additional void* parameter. +LIBBUILTIN(_setjmp, "iJ", "fjT", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(__sigsetjmp, "iSJi", "fjT", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(sigsetjmp, "iSJi", "fjT", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(savectx, "iJ", "fjT", "setjmp.h", ALL_LANGUAGES) +LIBBUILTIN(getcontext, "iK*", "fjT", "setjmp.h", ALL_LANGUAGES) -LIBBUILTIN(_longjmp, "vJi", "fr", "setjmp.h", ALL_GNU_LANGUAGES) -LIBBUILTIN(siglongjmp, "vSJi", "fr", "setjmp.h", ALL_GNU_LANGUAGES) +LIBBUILTIN(_longjmp, "vJi", "frT", "setjmp.h", ALL_GNU_LANGUAGES) +LIBBUILTIN(siglongjmp, "vSJi", "frT", "setjmp.h", ALL_GNU_LANGUAGES) // non-standard but very common LIBBUILTIN(strlcpy, "zc*cC*z", "f", "string.h", ALL_GNU_LANGUAGES) LIBBUILTIN(strlcat, "zc*cC*z", "f", "string.h", ALL_GNU_LANGUAGES) diff --git a/clang/include/clang/Basic/Builtins.h b/clang/include/clang/Basic/Builtins.h index e4ed482d9068..15bfcf797917 100644 --- a/clang/include/clang/Basic/Builtins.h +++ b/clang/include/clang/Basic/Builtins.h @@ -158,6 +158,13 @@ class Context { return strchr(getRecord(ID).Attributes, 't') != nullptr; } + /// Determines whether a declaration of this builtin should be recognized + /// even if the type doesn't match the specified signature. + bool allowTypeMismatch(unsigned ID) const { + return strchr(getRecord(ID).Attributes, 'T') != nullptr || + hasCustomTypechecking(ID); + } + /// Determines whether this builtin has a result or any arguments which /// are pointer types. bool hasPtrArgsOrResult(unsigned ID) const { diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td b/clang/include/clang/Basic/DiagnosticDriverKinds.td index 558639ecad6a..acdad15cdf6c 100644 --- a/clang/include/clang/Basic/DiagnosticDriverKinds.td +++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td @@ -69,7 +69,7 @@ def err_drv_cuda_version_unsupported : Error< "install, pass a different GPU arch with --cuda-gpu-arch, or pass " "--no-cuda-version-check.">; def warn_drv_unknown_cuda_version: Warning< - "Unknown CUDA version %0. Assuming the latest supported version %1">, + "Unknown CUDA version. %0 Assuming the latest supported version %1">, InGroup; def err_drv_cuda_host_arch : Error<"unsupported architecture '%0' for host compilation.">; def err_drv_mix_cuda_hip : Error<"Mixed Cuda and HIP compilation is not supported.">; diff --git a/clang/include/clang/Basic/IdentifierTable.h b/clang/include/clang/Basic/IdentifierTable.h index fc554a35e721..204a0f0cc0a5 100644 --- a/clang/include/clang/Basic/IdentifierTable.h +++ b/clang/include/clang/Basic/IdentifierTable.h @@ -225,18 +225,6 @@ class alignas(IdentifierInfoAlignment) IdentifierInfo { } void setObjCKeywordID(tok::ObjCKeywordKind ID) { ObjCOrBuiltinID = ID; } - /// True if setNotBuiltin() was called. - bool hasRevertedBuiltin() const { - return ObjCOrBuiltinID == tok::NUM_OBJC_KEYWORDS; - } - - /// Revert the identifier to a non-builtin identifier. We do this if - /// the name of a known builtin library function is used to declare that - /// function, but an unexpected type is specified. - void revertBuiltin() { - setBuiltinID(0); - } - /// Return a value indicating whether this is a builtin function. /// /// 0 is not-built-in. 1+ are specific builtin functions. diff --git a/clang/include/clang/Format/Format.h b/clang/include/clang/Format/Format.h old mode 100644 new mode 100755 diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h index 6f7ad8076718..16a7084f6b08 100644 --- a/clang/include/clang/Sema/Sema.h +++ b/clang/include/clang/Sema/Sema.h @@ -3957,6 +3957,8 @@ class Sema final { ObjCInterfaceDecl *getObjCInterfaceDecl(IdentifierInfo *&Id, SourceLocation IdLoc, bool TypoCorrection = false); + FunctionDecl *CreateBuiltin(IdentifierInfo *II, QualType Type, unsigned ID, + SourceLocation Loc); NamedDecl *LazilyCreateBuiltin(IdentifierInfo *II, unsigned ID, Scope *S, bool ForRedeclaration, SourceLocation Loc); diff --git a/clang/include/clang/Tooling/Refactoring/Extract/SourceExtraction.h b/clang/include/clang/Tooling/Refactoring/Extract/SourceExtraction.h old mode 100755 new mode 100644 diff --git a/clang/lib/AST/Decl.cpp b/clang/lib/AST/Decl.cpp index 5c0a98815dd7..67490da5c584 100644 --- a/clang/lib/AST/Decl.cpp +++ b/clang/lib/AST/Decl.cpp @@ -3162,44 +3162,24 @@ FunctionDecl *FunctionDecl::getCanonicalDecl() { return getFirstDecl(); } /// functions as their wrapped builtins. This shouldn't be done in general, but /// it's useful in Sema to diagnose calls to wrappers based on their semantics. unsigned FunctionDecl::getBuiltinID(bool ConsiderWrapperFunctions) const { - unsigned BuiltinID; + unsigned BuiltinID = 0; if (const auto *ABAA = getAttr()) { BuiltinID = ABAA->getBuiltinName()->getBuiltinID(); - } else { - if (!getIdentifier()) - return 0; - - BuiltinID = getIdentifier()->getBuiltinID(); + } else if (const auto *A = getAttr()) { + BuiltinID = A->getID(); } if (!BuiltinID) return 0; - ASTContext &Context = getASTContext(); - if (Context.getLangOpts().CPlusPlus) { - const auto *LinkageDecl = - dyn_cast(getFirstDecl()->getDeclContext()); - // In C++, the first declaration of a builtin is always inside an implicit - // extern "C". - // FIXME: A recognised library function may not be directly in an extern "C" - // declaration, for instance "extern "C" { namespace std { decl } }". - if (!LinkageDecl) { - if (BuiltinID == Builtin::BI__GetExceptionInfo && - Context.getTargetInfo().getCXXABI().isMicrosoft()) - return Builtin::BI__GetExceptionInfo; - return 0; - } - if (LinkageDecl->getLanguage() != LinkageSpecDecl::lang_c) - return 0; - } - // If the function is marked "overloadable", it has a different mangled name // and is not the C library function. if (!ConsiderWrapperFunctions && hasAttr() && !hasAttr()) return 0; + ASTContext &Context = getASTContext(); if (!Context.BuiltinInfo.isPredefinedLibFunction(BuiltinID)) return BuiltinID; diff --git a/clang/lib/AST/DeclTemplate.cpp b/clang/lib/AST/DeclTemplate.cpp old mode 100644 new mode 100755 diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp index 8994b939093e..6b93f1b60af5 100644 --- a/clang/lib/CodeGen/CGBuiltin.cpp +++ b/clang/lib/CodeGen/CGBuiltin.cpp @@ -3754,11 +3754,13 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID, case Builtin::BI_abnormal_termination: return RValue::get(EmitSEHAbnormalTermination()); case Builtin::BI_setjmpex: - if (getTarget().getTriple().isOSMSVCRT()) + if (getTarget().getTriple().isOSMSVCRT() && E->getNumArgs() == 1 && + E->getArg(0)->getType()->isPointerType()) return EmitMSVCRTSetJmp(*this, MSVCSetJmpKind::_setjmpex, E); break; case Builtin::BI_setjmp: - if (getTarget().getTriple().isOSMSVCRT()) { + if (getTarget().getTriple().isOSMSVCRT() && E->getNumArgs() == 1 && + E->getArg(0)->getType()->isPointerType()) { if (getTarget().getTriple().getArch() == llvm::Triple::x86) return EmitMSVCRTSetJmp(*this, MSVCSetJmpKind::_setjmp3, E); else if (getTarget().getTriple().getArch() == llvm::Triple::aarch64) diff --git a/clang/lib/CodeGen/CGObjCGNU.cpp b/clang/lib/CodeGen/CGObjCGNU.cpp index bb9c494ae68e..c64faf4c0af7 100644 --- a/clang/lib/CodeGen/CGObjCGNU.cpp +++ b/clang/lib/CodeGen/CGObjCGNU.cpp @@ -3511,19 +3511,6 @@ void CGObjCGNU::GenerateClass(const ObjCImplementationDecl *OID) { ClassMethods.insert(ClassMethods.begin(), OID->classmeth_begin(), OID->classmeth_end()); - // Collect the same information about synthesized properties, which don't - // show up in the instance method lists. - for (auto *propertyImpl : OID->property_impls()) - if (propertyImpl->getPropertyImplementation() == - ObjCPropertyImplDecl::Synthesize) { - auto addPropertyMethod = [&](const ObjCMethodDecl *accessor) { - if (accessor) - InstanceMethods.push_back(accessor); - }; - addPropertyMethod(propertyImpl->getGetterMethodDecl()); - addPropertyMethod(propertyImpl->getSetterMethodDecl()); - } - llvm::Constant *Properties = GeneratePropertyList(OID, ClassDecl); // Collect the names of referenced protocols diff --git a/clang/lib/CodeGen/TargetInfo.cpp b/clang/lib/CodeGen/TargetInfo.cpp index 9cd63ebe29ee..a061651d8b21 100644 --- a/clang/lib/CodeGen/TargetInfo.cpp +++ b/clang/lib/CodeGen/TargetInfo.cpp @@ -5471,6 +5471,11 @@ class AArch64ABIInfo : public SwiftABIInfo { Address EmitVAArg(CodeGenFunction &CGF, Address VAListAddr, QualType Ty) const override { + llvm::Type *BaseTy = CGF.ConvertType(Ty); + if (isa(BaseTy)) + llvm::report_fatal_error("Passing SVE types to variadic functions is " + "currently not supported"); + return Kind == Win64 ? EmitMSVAArg(CGF, VAListAddr, Ty) : isDarwinPCS() ? EmitDarwinVAArg(VAListAddr, Ty, CGF) : EmitAAPCSVAArg(VAListAddr, Ty, CGF); @@ -10390,7 +10395,6 @@ bool RISCVABIInfo::detectFPCCEligibleStructHelper(QualType Ty, CharUnits CurOff, return false; Field1Ty = CGT.ConvertType(EltTy); Field1Off = CurOff; - assert(CurOff.isZero() && "Unexpected offset for first field"); Field2Ty = Field1Ty; Field2Off = Field1Off + getContext().getTypeSizeInChars(EltTy); return true; @@ -10485,7 +10489,7 @@ bool RISCVABIInfo::detectFPCCEligibleStruct(QualType Ty, llvm::Type *&Field1Ty, NeededArgFPRs++; else if (Field2Ty) NeededArgGPRs++; - return IsCandidate; + return true; } // Call getCoerceAndExpand for the two-element flattened struct described by @@ -10511,15 +10515,15 @@ ABIArgInfo RISCVABIInfo::coerceAndExpandFPCCEligibleStruct( CharUnits Field2Align = CharUnits::fromQuantity(getDataLayout().getABITypeAlignment(Field2Ty)); - CharUnits Field1Size = + CharUnits Field1End = Field1Off + CharUnits::fromQuantity(getDataLayout().getTypeStoreSize(Field1Ty)); - CharUnits Field2OffNoPadNoPack = Field1Size.alignTo(Field2Align); + CharUnits Field2OffNoPadNoPack = Field1End.alignTo(Field2Align); CharUnits Padding = CharUnits::Zero(); if (Field2Off > Field2OffNoPadNoPack) Padding = Field2Off - Field2OffNoPadNoPack; - else if (Field2Off != Field2Align && Field2Off > Field1Size) - Padding = Field2Off - Field1Size; + else if (Field2Off != Field2Align && Field2Off > Field1End) + Padding = Field2Off - Field1End; bool IsPacked = !Field2Off.isMultipleOf(Field2Align); diff --git a/clang/lib/Driver/ToolChains/Cuda.cpp b/clang/lib/Driver/ToolChains/Cuda.cpp index 110a0bca9bc1..ffc606dd554b 100644 --- a/clang/lib/Driver/ToolChains/Cuda.cpp +++ b/clang/lib/Driver/ToolChains/Cuda.cpp @@ -16,6 +16,7 @@ #include "clang/Driver/Driver.h" #include "clang/Driver/DriverDiagnostic.h" #include "clang/Driver/Options.h" +#include "llvm/ADT/Optional.h" #include "llvm/Option/ArgList.h" #include "llvm/Support/FileSystem.h" #include "llvm/Support/Host.h" @@ -32,30 +33,81 @@ using namespace clang::driver::tools; using namespace clang; using namespace llvm::opt; +namespace { +struct CudaVersionInfo { + std::string DetectedVersion; + CudaVersion Version; +}; // Parses the contents of version.txt in an CUDA installation. It should // contain one line of the from e.g. "CUDA Version 7.5.2". -void CudaInstallationDetector::ParseCudaVersionFile(llvm::StringRef V) { - Version = CudaVersion::UNKNOWN; +CudaVersionInfo parseCudaVersionFile(llvm::StringRef V) { + V = V.trim(); if (!V.startswith("CUDA Version ")) - return; + return {V.str(), CudaVersion::UNKNOWN}; V = V.substr(strlen("CUDA Version ")); SmallVector VersionParts; V.split(VersionParts, '.'); - if (VersionParts.size() < 2) - return; - DetectedVersion = join_items(".", VersionParts[0], VersionParts[1]); - Version = CudaStringToVersion(DetectedVersion); - if (Version != CudaVersion::UNKNOWN) { - // TODO(tra): remove the warning once we have all features of 10.2 and 11.0 - // implemented. - DetectedVersionIsNotSupported = Version > CudaVersion::LATEST_SUPPORTED; - return; - } - - Version = CudaVersion::LATEST_SUPPORTED; - DetectedVersionIsNotSupported = true; + return {"version.txt: " + V.str() + ".", + VersionParts.size() < 2 + ? CudaVersion::UNKNOWN + : CudaStringToVersion( + join_items(".", VersionParts[0], VersionParts[1]))}; } +CudaVersion getCudaVersion(uint32_t raw_version) { + if (raw_version < 7050) + return CudaVersion::CUDA_70; + if (raw_version < 8000) + return CudaVersion::CUDA_75; + if (raw_version < 9000) + return CudaVersion::CUDA_80; + if (raw_version < 9010) + return CudaVersion::CUDA_90; + if (raw_version < 9020) + return CudaVersion::CUDA_91; + if (raw_version < 10000) + return CudaVersion::CUDA_92; + if (raw_version < 10010) + return CudaVersion::CUDA_100; + if (raw_version < 10020) + return CudaVersion::CUDA_101; + if (raw_version < 11000) + return CudaVersion::CUDA_102; + if (raw_version < 11010) + return CudaVersion::CUDA_110; + return CudaVersion::LATEST; +} + +CudaVersionInfo parseCudaHFile(llvm::StringRef Input) { + // Helper lambda which skips the words if the line starts with them or returns + // None otherwise. + auto StartsWithWords = + [](llvm::StringRef Line, + const SmallVector words) -> llvm::Optional { + for (StringRef word : words) { + if (!Line.consume_front(word)) + return {}; + Line = Line.ltrim(); + } + return Line; + }; + + Input = Input.ltrim(); + while (!Input.empty()) { + if (auto Line = + StartsWithWords(Input.ltrim(), {"#", "define", "CUDA_VERSION"})) { + uint32_t RawVersion; + Line->consumeInteger(10, RawVersion); + return {"cuda.h: CUDA_VERSION=" + Twine(RawVersion).str() + ".", + getCudaVersion(RawVersion)}; + } + // Find next non-empty line. + Input = Input.drop_front(Input.find_first_of("\n\r")).ltrim(); + } + return {"cuda.h: CUDA_VERSION not found.", CudaVersion::UNKNOWN}; +} +} // namespace + void CudaInstallationDetector::WarnIfUnsupportedVersion() { if (DetectedVersionIsNotSupported) D.Diag(diag::warn_drv_unknown_cuda_version) @@ -152,16 +204,31 @@ CudaInstallationDetector::CudaInstallationDetector( else continue; - llvm::ErrorOr> VersionFile = - FS.getBufferForFile(InstallPath + "/version.txt"); - if (!VersionFile) { - // CUDA 7.0 doesn't have a version.txt, so guess that's our version if - // version.txt isn't present. - Version = CudaVersion::CUDA_70; - } else { - ParseCudaVersionFile((*VersionFile)->getBuffer()); + CudaVersionInfo VersionInfo = {"", CudaVersion::UNKNOWN}; + if (auto VersionFile = FS.getBufferForFile(InstallPath + "/version.txt")) + VersionInfo = parseCudaVersionFile((*VersionFile)->getBuffer()); + // If version file didn't give us the version, try to find it in cuda.h + if (VersionInfo.Version == CudaVersion::UNKNOWN) + if (auto CudaHFile = FS.getBufferForFile(InstallPath + "/include/cuda.h")) + VersionInfo = parseCudaHFile((*CudaHFile)->getBuffer()); + // As the last resort, make an educated guess between CUDA-7.0, (which had + // no version.txt file and had old-style libdevice bitcode ) and an unknown + // recent CUDA version (no version.txt, new style bitcode). + if (VersionInfo.Version == CudaVersion::UNKNOWN) { + VersionInfo.Version = (FS.exists(LibDevicePath + "/libdevice.10.bc")) + ? Version = CudaVersion::LATEST + : Version = CudaVersion::CUDA_70; + VersionInfo.DetectedVersion = + "No version found in version.txt or cuda.h."; } + Version = VersionInfo.Version; + DetectedVersion = VersionInfo.DetectedVersion; + + // TODO(tra): remove the warning once we have all features of 10.2 + // and 11.0 implemented. + DetectedVersionIsNotSupported = Version > CudaVersion::LATEST_SUPPORTED; + if (Version >= CudaVersion::CUDA_90) { // CUDA-9+ uses single libdevice file for all GPU variants. std::string FilePath = LibDevicePath + "/libdevice.10.bc"; diff --git a/clang/lib/Driver/ToolChains/Cuda.h b/clang/lib/Driver/ToolChains/Cuda.h index 873eb7338a30..bbf272c468a5 100644 --- a/clang/lib/Driver/ToolChains/Cuda.h +++ b/clang/lib/Driver/ToolChains/Cuda.h @@ -78,9 +78,6 @@ class CudaInstallationDetector { return LibDeviceMap.lookup(Gpu); } void WarnIfUnsupportedVersion(); - -private: - void ParseCudaVersionFile(llvm::StringRef V); }; namespace tools { diff --git a/clang/lib/Format/FormatToken.h b/clang/lib/Format/FormatToken.h index d4287f53fde3..b132a3e84da5 100644 --- a/clang/lib/Format/FormatToken.h +++ b/clang/lib/Format/FormatToken.h @@ -183,12 +183,6 @@ struct FormatToken { /// before the token. bool MustBreakBefore = false; - /// Whether to not align across this token - /// - /// This happens for example when a preprocessor directive ended directly - /// before the token, but very rarely otherwise. - bool MustBreakAlignBefore = false; - /// The raw text of the token. /// /// Contains the raw token text without leading whitespace and without leading diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp index a37386425aae..ea8a41cfba82 100644 --- a/clang/lib/Format/UnwrappedLineParser.cpp +++ b/clang/lib/Format/UnwrappedLineParser.cpp @@ -3037,7 +3037,6 @@ void UnwrappedLineParser::readToken(int LevelDifference) { } FormatTok = Tokens->getNextToken(); FormatTok->MustBreakBefore = true; - FormatTok->MustBreakAlignBefore = true; } if (!PPStack.empty() && (PPStack.back().Kind == PP_Unreachable) && @@ -3062,7 +3061,6 @@ void UnwrappedLineParser::pushToken(FormatToken *Tok) { Line->Tokens.push_back(UnwrappedLineNode(Tok)); if (MustBreakBeforeNextToken) { Line->Tokens.back().Tok->MustBreakBefore = true; - Line->Tokens.back().Tok->MustBreakAlignBefore = true; MustBreakBeforeNextToken = false; } } diff --git a/clang/lib/Format/WhitespaceManager.cpp b/clang/lib/Format/WhitespaceManager.cpp index 32e0b685ea0f..3a265bd09168 100644 --- a/clang/lib/Format/WhitespaceManager.cpp +++ b/clang/lib/Format/WhitespaceManager.cpp @@ -411,11 +411,9 @@ static unsigned AlignTokens(const FormatStyle &Style, F &&Matches, if (Changes[i].NewlinesBefore != 0) { CommasBeforeMatch = 0; EndOfSequence = i; - // If there is a blank line, there is a forced-align-break (eg, - // preprocessor), or if the last line didn't contain any matching token, - // the sequence ends here. - if (Changes[i].NewlinesBefore > 1 || - Changes[i].Tok->MustBreakAlignBefore || !FoundMatchOnLine) + // If there is a blank line, or if the last line didn't contain any + // matching token, the sequence ends here. + if (Changes[i].NewlinesBefore > 1 || !FoundMatchOnLine) AlignCurrentSequence(); FoundMatchOnLine = false; @@ -726,8 +724,6 @@ void WhitespaceManager::alignTrailingComments() { if (Changes[i].StartOfBlockComment) continue; Newlines += Changes[i].NewlinesBefore; - if (Changes[i].Tok->MustBreakAlignBefore) - BreakBeforeNext = true; if (!Changes[i].IsTrailingComment) continue; diff --git a/clang/lib/Headers/__clang_cuda_complex_builtins.h b/clang/lib/Headers/__clang_cuda_complex_builtins.h index 8c10ff6b461f..d924487ab285 100644 --- a/clang/lib/Headers/__clang_cuda_complex_builtins.h +++ b/clang/lib/Headers/__clang_cuda_complex_builtins.h @@ -16,7 +16,7 @@ // to work with CUDA and OpenMP target offloading [in C and C++ mode].) #pragma push_macro("__DEVICE__") -#ifdef _OPENMP +#ifdef __OPENMP_NVPTX__ #pragma omp declare target #define __DEVICE__ __attribute__((noinline, nothrow, cold, weak)) #else @@ -26,7 +26,7 @@ // To make the algorithms available for C and C++ in CUDA and OpenMP we select // different but equivalent function versions. TODO: For OpenMP we currently // select the native builtins as the overload support for templates is lacking. -#if !defined(_OPENMP) +#if !defined(__OPENMP_NVPTX__) #define _ISNANd std::isnan #define _ISNANf std::isnan #define _ISINFd std::isinf @@ -250,7 +250,7 @@ __DEVICE__ float _Complex __divsc3(float __a, float __b, float __c, float __d) { #undef _LOGBd #undef _LOGBf -#ifdef _OPENMP +#ifdef __OPENMP_NVPTX__ #pragma omp end declare target #endif diff --git a/clang/lib/Headers/cuda_wrappers/new b/clang/lib/Headers/cuda_wrappers/new index f49811c5a57c..7f255314056a 100644 --- a/clang/lib/Headers/cuda_wrappers/new +++ b/clang/lib/Headers/cuda_wrappers/new @@ -26,6 +26,13 @@ #include_next +#if !defined(__device__) +// The header has been included too early from the standard C++ library +// and CUDA-specific macros are not available yet. +// Undo the include guard and try again later. +#undef __CLANG_CUDA_WRAPPERS_NEW +#else + #pragma push_macro("CUDA_NOEXCEPT") #if __cplusplus >= 201103L #define CUDA_NOEXCEPT noexcept @@ -95,4 +102,5 @@ __device__ inline void operator delete[](void *, void *) CUDA_NOEXCEPT {} #pragma pop_macro("CUDA_NOEXCEPT") +#endif // __device__ #endif // include guard diff --git a/clang/lib/Headers/intrin.h b/clang/lib/Headers/intrin.h index 871b47ca8267..e7b76a3bb2ed 100644 --- a/clang/lib/Headers/intrin.h +++ b/clang/lib/Headers/intrin.h @@ -57,16 +57,11 @@ void __addfsbyte(unsigned long, unsigned char); void __addfsdword(unsigned long, unsigned long); void __addfsword(unsigned long, unsigned short); void __code_seg(const char *); -static __inline__ void __cpuid(int[4], int); -static __inline__ void __cpuidex(int[4], int, int); -static __inline__ __int64 __emul(int, int); -static __inline__ unsigned __int64 __emulu(unsigned int, unsigned int); unsigned int __getcallerseflags(void); -static __inline__ void __halt(void); unsigned char __inbyte(unsigned short); void __inbytestring(unsigned short, unsigned char *, unsigned long); @@ -82,13 +77,9 @@ void __inwordstring(unsigned short, unsigned short *, unsigned long); void __lidt(void *); unsigned __int64 __ll_lshift(unsigned __int64, int); __int64 __ll_rshift(__int64, int); -static __inline__ void __movsb(unsigned char *, unsigned char const *, size_t); -static __inline__ void __movsd(unsigned long *, unsigned long const *, size_t); -static __inline__ void __movsw(unsigned short *, unsigned short const *, size_t); -static __inline__ void __nop(void); void __nvreg_restore_fence(void); void __nvreg_save_fence(void); @@ -105,23 +96,16 @@ unsigned long __readcr4(void); unsigned long __readcr8(void); unsigned int __readdr(unsigned int); #ifdef __i386__ -static __inline__ unsigned char __readfsbyte(unsigned long); -static __inline__ unsigned __int64 __readfsqword(unsigned long); -static __inline__ unsigned short __readfsword(unsigned long); #endif -static __inline__ unsigned __int64 __readmsr(unsigned long); unsigned __int64 __readpmc(unsigned long); unsigned long __segmentlimit(unsigned long); void __sidt(void *); -static __inline__ void __stosb(unsigned char *, unsigned char, size_t); -static __inline__ void __stosd(unsigned long *, unsigned long, size_t); -static __inline__ void __stosw(unsigned short *, unsigned short, size_t); void __svm_clgi(void); void __svm_invlpga(void *, int); @@ -136,7 +120,6 @@ void __vmx_off(void); void __vmx_vmptrst(unsigned __int64 *); void __wbinvd(void); void __writecr0(unsigned int); -static __inline__ void __writecr3(unsigned __INTPTR_TYPE__); void __writecr4(unsigned int); void __writecr8(unsigned int); @@ -146,11 +129,8 @@ void __writefsdword(unsigned long, unsigned long); void __writefsqword(unsigned long, unsigned __int64); void __writefsword(unsigned long, unsigned short); void __writemsr(unsigned long, unsigned __int64); -static __inline__ void *_AddressOfReturnAddress(void); -static __inline__ unsigned char _BitScanForward(unsigned long *_Index, unsigned long _Mask); -static __inline__ unsigned char _BitScanReverse(unsigned long *_Index, unsigned long _Mask); unsigned char _bittest(long const *, long); unsigned char _bittestandcomplement(long *, long); @@ -169,12 +149,10 @@ long _InterlockedExchangeAdd_HLEAcquire(long volatile *, long); long _InterlockedExchangeAdd_HLERelease(long volatile *, long); __int64 _InterlockedExchangeAdd64_HLEAcquire(__int64 volatile *, __int64); __int64 _InterlockedExchangeAdd64_HLERelease(__int64 volatile *, __int64); -static __inline__ void -__attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) -_ReadBarrier(void); -static __inline__ void -__attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) -_ReadWriteBarrier(void); +void __attribute__((__deprecated__( + "use other intrinsics or C++11 atomics instead"))) _ReadBarrier(void); +void __attribute__((__deprecated__( + "use other intrinsics or C++11 atomics instead"))) _ReadWriteBarrier(void); unsigned int _rorx_u32(unsigned int, const unsigned int); int _sarx_i32(int, unsigned int); #if __STDC_HOSTED__ @@ -185,9 +163,8 @@ unsigned int _shrx_u32(unsigned int, unsigned int); void _Store_HLERelease(long volatile *, long); void _Store64_HLERelease(__int64 volatile *, __int64); void _StorePointer_HLERelease(void *volatile *, void *); -static __inline__ void -__attribute__((__deprecated__("use other intrinsics or C++11 atomics instead"))) -_WriteBarrier(void); +void __attribute__((__deprecated__( + "use other intrinsics or C++11 atomics instead"))) _WriteBarrier(void); unsigned __int32 xbegin(void); void _xend(void); @@ -197,19 +174,14 @@ void __addgsbyte(unsigned long, unsigned char); void __addgsdword(unsigned long, unsigned long); void __addgsqword(unsigned long, unsigned __int64); void __addgsword(unsigned long, unsigned short); -static __inline__ void __faststorefence(void); void __incgsbyte(unsigned long); void __incgsdword(unsigned long); void __incgsqword(unsigned long); void __incgsword(unsigned long); -static __inline__ void __movsq(unsigned long long *, unsigned long long const *, size_t); -static __inline__ unsigned char __readgsbyte(unsigned long); -static __inline__ unsigned long __readgsdword(unsigned long); -static __inline__ unsigned __int64 __readgsqword(unsigned long); unsigned short __readgsword(unsigned long); unsigned __int64 __shiftleft128(unsigned __int64 _LowPart, @@ -218,7 +190,6 @@ unsigned __int64 __shiftleft128(unsigned __int64 _LowPart, unsigned __int64 __shiftright128(unsigned __int64 _LowPart, unsigned __int64 _HighPart, unsigned char _Shift); -static __inline__ void __stosq(unsigned __int64 *, unsigned __int64, size_t); unsigned char __vmx_on(unsigned __int64 *); unsigned char __vmx_vmclear(unsigned __int64 *); @@ -269,13 +240,9 @@ unsigned __int64 _rorx_u64(unsigned __int64, const unsigned int); __int64 _sarx_i64(__int64, unsigned int); unsigned __int64 _shlx_u64(unsigned __int64, unsigned int); unsigned __int64 _shrx_u64(unsigned __int64, unsigned int); -static __inline__ __int64 __mulh(__int64, __int64); -static __inline__ unsigned __int64 __umulh(unsigned __int64, unsigned __int64); -static __inline__ __int64 _mul128(__int64, __int64, __int64*); -static __inline__ unsigned __int64 _umul128(unsigned __int64, unsigned __int64, unsigned __int64*); @@ -284,29 +251,19 @@ unsigned __int64 _umul128(unsigned __int64, #if defined(__x86_64__) || defined(__arm__) || defined(__aarch64__) -static __inline__ unsigned char _BitScanForward64(unsigned long *_Index, unsigned __int64 _Mask); -static __inline__ unsigned char _BitScanReverse64(unsigned long *_Index, unsigned __int64 _Mask); #endif #if defined(__i386__) || defined(__x86_64__) || defined(__arm__) || defined(__aarch64__) -static __inline__ __int64 _InterlockedDecrement64(__int64 volatile *_Addend); -static __inline__ __int64 _InterlockedExchange64(__int64 volatile *_Target, __int64 _Value); -static __inline__ __int64 _InterlockedExchangeAdd64(__int64 volatile *_Addend, __int64 _Value); -static __inline__ __int64 _InterlockedExchangeSub64(__int64 volatile *_Subend, __int64 _Value); -static __inline__ __int64 _InterlockedIncrement64(__int64 volatile *_Addend); -static __inline__ __int64 _InterlockedOr64(__int64 volatile *_Value, __int64 _Mask); -static __inline__ __int64 _InterlockedXor64(__int64 volatile *_Value, __int64 _Mask); -static __inline__ __int64 _InterlockedAnd64(__int64 volatile *_Value, __int64 _Mask); #endif @@ -475,40 +432,56 @@ __int64 _InterlockedCompareExchange64_rel(__int64 volatile *_Destination, |* movs, stos \*----------------------------------------------------------------------------*/ #if defined(__i386__) || defined(__x86_64__) -static __inline__ void __DEFAULT_FN_ATTRS -__movsb(unsigned char *__dst, unsigned char const *__src, size_t __n) { +static __inline__ void __DEFAULT_FN_ATTRS __movsb(unsigned char *__dst, + unsigned char const *__src, + size_t __n) { __asm__ __volatile__("rep movsb" : "+D"(__dst), "+S"(__src), "+c"(__n) : : "memory"); } -static __inline__ void __DEFAULT_FN_ATTRS -__movsd(unsigned long *__dst, unsigned long const *__src, size_t __n) { - __asm__ __volatile__("rep movsl" : "+D"(__dst), "+S"(__src), "+c"(__n) - : : "memory"); -} -static __inline__ void __DEFAULT_FN_ATTRS -__movsw(unsigned short *__dst, unsigned short const *__src, size_t __n) { - __asm__ __volatile__("rep movsw" : "+D"(__dst), "+S"(__src), "+c"(__n) - : : "memory"); -} -static __inline__ void __DEFAULT_FN_ATTRS -__stosd(unsigned long *__dst, unsigned long __x, size_t __n) { - __asm__ __volatile__("rep stosl" : "+D"(__dst), "+c"(__n) : "a"(__x) +static __inline__ void __DEFAULT_FN_ATTRS __movsd(unsigned long *__dst, + unsigned long const *__src, + size_t __n) { + __asm__ __volatile__("rep movsl" + : "+D"(__dst), "+S"(__src), "+c"(__n) + : : "memory"); } -static __inline__ void __DEFAULT_FN_ATTRS -__stosw(unsigned short *__dst, unsigned short __x, size_t __n) { - __asm__ __volatile__("rep stosw" : "+D"(__dst), "+c"(__n) : "a"(__x) +static __inline__ void __DEFAULT_FN_ATTRS __movsw(unsigned short *__dst, + unsigned short const *__src, + size_t __n) { + __asm__ __volatile__("rep movsw" + : "+D"(__dst), "+S"(__src), "+c"(__n) + : + : "memory"); +} +static __inline__ void __DEFAULT_FN_ATTRS __stosd(unsigned long *__dst, + unsigned long __x, + size_t __n) { + __asm__ __volatile__("rep stosl" + : "+D"(__dst), "+c"(__n) + : "a"(__x) + : "memory"); +} +static __inline__ void __DEFAULT_FN_ATTRS __stosw(unsigned short *__dst, + unsigned short __x, + size_t __n) { + __asm__ __volatile__("rep stosw" + : "+D"(__dst), "+c"(__n) + : "a"(__x) : "memory"); } #endif #ifdef __x86_64__ -static __inline__ void __DEFAULT_FN_ATTRS -__movsq(unsigned long long *__dst, unsigned long long const *__src, size_t __n) { - __asm__ __volatile__("rep movsq" : "+D"(__dst), "+S"(__src), "+c"(__n) - : : "memory"); +static __inline__ void __DEFAULT_FN_ATTRS __movsq( + unsigned long long *__dst, unsigned long long const *__src, size_t __n) { + __asm__ __volatile__("rep movsq" + : "+D"(__dst), "+S"(__src), "+c"(__n) + : + : "memory"); } -static __inline__ void __DEFAULT_FN_ATTRS -__stosq(unsigned __int64 *__dst, unsigned __int64 __x, size_t __n) { +static __inline__ void __DEFAULT_FN_ATTRS __stosq(unsigned __int64 *__dst, + unsigned __int64 __x, + size_t __n) { __asm__ __volatile__("rep stosq" : "+D"(__dst), "+c"(__n) : "a"(__x) : "memory"); } @@ -518,26 +491,25 @@ __stosq(unsigned __int64 *__dst, unsigned __int64 __x, size_t __n) { |* Misc \*----------------------------------------------------------------------------*/ #if defined(__i386__) || defined(__x86_64__) -static __inline__ void __DEFAULT_FN_ATTRS -__cpuid(int __info[4], int __level) { - __asm__ ("cpuid" : "=a"(__info[0]), "=b" (__info[1]), "=c"(__info[2]), "=d"(__info[3]) - : "a"(__level), "c"(0)); +static __inline__ void __DEFAULT_FN_ATTRS __cpuid(int __info[4], int __level) { + __asm__("cpuid" + : "=a"(__info[0]), "=b"(__info[1]), "=c"(__info[2]), "=d"(__info[3]) + : "a"(__level), "c"(0)); } -static __inline__ void __DEFAULT_FN_ATTRS -__cpuidex(int __info[4], int __level, int __ecx) { - __asm__ ("cpuid" : "=a"(__info[0]), "=b" (__info[1]), "=c"(__info[2]), "=d"(__info[3]) - : "a"(__level), "c"(__ecx)); +static __inline__ void __DEFAULT_FN_ATTRS __cpuidex(int __info[4], int __level, + int __ecx) { + __asm__("cpuid" + : "=a"(__info[0]), "=b"(__info[1]), "=c"(__info[2]), "=d"(__info[3]) + : "a"(__level), "c"(__ecx)); } -static __inline__ void __DEFAULT_FN_ATTRS -__halt(void) { - __asm__ volatile ("hlt"); +static __inline__ void __DEFAULT_FN_ATTRS __halt(void) { + __asm__ volatile("hlt"); } #endif #if defined(__i386__) || defined(__x86_64__) || defined(__aarch64__) -static __inline__ void __DEFAULT_FN_ATTRS -__nop(void) { - __asm__ volatile ("nop"); +static __inline__ void __DEFAULT_FN_ATTRS __nop(void) { + __asm__ volatile("nop"); } #endif @@ -574,8 +546,7 @@ __readmsr(unsigned long __register) { } #endif -static __inline__ unsigned __LPTRINT_TYPE__ __DEFAULT_FN_ATTRS -__readcr3(void) { +static __inline__ unsigned __LPTRINT_TYPE__ __DEFAULT_FN_ATTRS __readcr3(void) { unsigned __LPTRINT_TYPE__ __cr3_val; __asm__ __volatile__ ("mov %%cr3, %0" : "=r"(__cr3_val) : : "memory"); return __cr3_val; diff --git a/clang/lib/Headers/openmp_wrappers/complex b/clang/lib/Headers/openmp_wrappers/complex index 1ed0b14879ef..d8dcd41670ee 100644 --- a/clang/lib/Headers/openmp_wrappers/complex +++ b/clang/lib/Headers/openmp_wrappers/complex @@ -18,7 +18,9 @@ #include #define __CUDA__ +#define __OPENMP_NVPTX__ #include <__clang_cuda_complex_builtins.h> +#undef __OPENMP_NVPTX__ #endif // Grab the host header too. diff --git a/clang/lib/Headers/openmp_wrappers/complex.h b/clang/lib/Headers/openmp_wrappers/complex.h index 829c7a785725..00d278548f82 100644 --- a/clang/lib/Headers/openmp_wrappers/complex.h +++ b/clang/lib/Headers/openmp_wrappers/complex.h @@ -18,7 +18,9 @@ #include #define __CUDA__ +#define __OPENMP_NVPTX__ #include <__clang_cuda_complex_builtins.h> +#undef __OPENMP_NVPTX__ #endif // Grab the host header too. diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp index b00d2ff5f1d5..1bf04d9cb4f2 100644 --- a/clang/lib/Sema/SemaChecking.cpp +++ b/clang/lib/Sema/SemaChecking.cpp @@ -1573,11 +1573,6 @@ Sema::CheckBuiltinFunctionCall(FunctionDecl *FDecl, unsigned BuiltinID, if (SemaBuiltinSetjmp(TheCall)) return ExprError(); break; - case Builtin::BI_setjmp: - case Builtin::BI_setjmpex: - if (checkArgCount(*this, TheCall, 1)) - return true; - break; case Builtin::BI__builtin_classify_type: if (checkArgCount(*this, TheCall, 1)) return true; TheCall->setType(Context.IntTy); diff --git a/clang/lib/Sema/SemaConcept.cpp b/clang/lib/Sema/SemaConcept.cpp old mode 100644 new mode 100755 diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp index 3e2b61ae8cdf..5b0417fa8859 100644 --- a/clang/lib/Sema/SemaDecl.cpp +++ b/clang/lib/Sema/SemaDecl.cpp @@ -2054,6 +2054,42 @@ static StringRef getHeaderName(Builtin::Context &BuiltinInfo, unsigned ID, llvm_unreachable("unhandled error kind"); } +FunctionDecl *Sema::CreateBuiltin(IdentifierInfo *II, QualType Type, + unsigned ID, SourceLocation Loc) { + DeclContext *Parent = Context.getTranslationUnitDecl(); + + if (getLangOpts().CPlusPlus) { + LinkageSpecDecl *CLinkageDecl = LinkageSpecDecl::Create( + Context, Parent, Loc, Loc, LinkageSpecDecl::lang_c, false); + CLinkageDecl->setImplicit(); + Parent->addDecl(CLinkageDecl); + Parent = CLinkageDecl; + } + + FunctionDecl *New = FunctionDecl::Create(Context, Parent, Loc, Loc, II, Type, + /*TInfo=*/nullptr, SC_Extern, false, + Type->isFunctionProtoType()); + New->setImplicit(); + New->addAttr(BuiltinAttr::CreateImplicit(Context, ID)); + + // Create Decl objects for each parameter, adding them to the + // FunctionDecl. + if (const FunctionProtoType *FT = dyn_cast(Type)) { + SmallVector Params; + for (unsigned i = 0, e = FT->getNumParams(); i != e; ++i) { + ParmVarDecl *parm = ParmVarDecl::Create( + Context, New, SourceLocation(), SourceLocation(), nullptr, + FT->getParamType(i), /*TInfo=*/nullptr, SC_None, nullptr); + parm->setScopeInfo(0, i); + Params.push_back(parm); + } + New->setParams(Params); + } + + AddKnownFunctionAttributes(New); + return New; +} + /// LazilyCreateBuiltin - The specified Builtin-ID was first used at /// file scope. lazily create a decl for it. ForRedeclaration is true /// if we're creating this built-in in anticipation of redeclaring the @@ -2071,7 +2107,8 @@ NamedDecl *Sema::LazilyCreateBuiltin(IdentifierInfo *II, unsigned ID, // If we have a builtin without an associated type we should not emit a // warning when we were not able to find a type for it. - if (Error == ASTContext::GE_Missing_type) + if (Error == ASTContext::GE_Missing_type || + Context.BuiltinInfo.allowTypeMismatch(ID)) return nullptr; // If we could not find a type for setjmp it is because the jmp_buf type was @@ -2095,50 +2132,15 @@ NamedDecl *Sema::LazilyCreateBuiltin(IdentifierInfo *II, unsigned ID, Context.BuiltinInfo.isHeaderDependentFunction(ID))) { Diag(Loc, diag::ext_implicit_lib_function_decl) << Context.BuiltinInfo.getName(ID) << R; - if (Context.BuiltinInfo.getHeaderName(ID) && - !Diags.isIgnored(diag::ext_implicit_lib_function_decl, Loc)) + if (const char *Header = Context.BuiltinInfo.getHeaderName(ID)) Diag(Loc, diag::note_include_header_or_declare) - << Context.BuiltinInfo.getHeaderName(ID) - << Context.BuiltinInfo.getName(ID); + << Header << Context.BuiltinInfo.getName(ID); } if (R.isNull()) return nullptr; - DeclContext *Parent = Context.getTranslationUnitDecl(); - if (getLangOpts().CPlusPlus) { - LinkageSpecDecl *CLinkageDecl = - LinkageSpecDecl::Create(Context, Parent, Loc, Loc, - LinkageSpecDecl::lang_c, false); - CLinkageDecl->setImplicit(); - Parent->addDecl(CLinkageDecl); - Parent = CLinkageDecl; - } - - FunctionDecl *New = FunctionDecl::Create(Context, - Parent, - Loc, Loc, II, R, /*TInfo=*/nullptr, - SC_Extern, - false, - R->isFunctionProtoType()); - New->setImplicit(); - - // Create Decl objects for each parameter, adding them to the - // FunctionDecl. - if (const FunctionProtoType *FT = dyn_cast(R)) { - SmallVector Params; - for (unsigned i = 0, e = FT->getNumParams(); i != e; ++i) { - ParmVarDecl *parm = - ParmVarDecl::Create(Context, New, SourceLocation(), SourceLocation(), - nullptr, FT->getParamType(i), /*TInfo=*/nullptr, - SC_None, nullptr); - parm->setScopeInfo(0, i); - Params.push_back(parm); - } - New->setParams(Params); - } - - AddKnownFunctionAttributes(New); + FunctionDecl *New = CreateBuiltin(II, R, ID, Loc); RegisterLocallyScopedExternCDecl(New, S); // TUScope is the translation-unit scope to insert this function into. @@ -2146,7 +2148,7 @@ NamedDecl *Sema::LazilyCreateBuiltin(IdentifierInfo *II, unsigned ID, // relate Scopes to DeclContexts, and probably eliminate CurContext // entirely, but we're not there yet. DeclContext *SavedContext = CurContext; - CurContext = Parent; + CurContext = New->getDeclContext(); PushOnScopeChains(New, TUScope); CurContext = SavedContext; return New; @@ -3348,7 +3350,10 @@ bool Sema::MergeFunctionDecl(FunctionDecl *New, NamedDecl *&OldD, // there but not here. NewTypeInfo = NewTypeInfo.withCallingConv(OldTypeInfo.getCC()); RequiresAdjustment = true; - } else if (New->getBuiltinID()) { + } else if (Old->getBuiltinID()) { + // Builtin attribute isn't propagated to the new one yet at this point, + // so we check if the old one is a builtin. + // Calling Conventions on a Builtin aren't really useful and setting a // default calling convention and cdecl'ing some builtin redeclarations is // common, so warn and ignore the calling convention on the redeclaration. @@ -3781,18 +3786,6 @@ bool Sema::MergeFunctionDecl(FunctionDecl *New, NamedDecl *&OldD, Diag(New->getLocation(), diag::warn_redecl_library_builtin) << New; Diag(OldLocation, diag::note_previous_builtin_declaration) << Old << Old->getType(); - - // If this is a global redeclaration, just forget hereafter - // about the "builtin-ness" of the function. - // - // Doing this for local extern declarations is problematic. If - // the builtin declaration remains visible, a second invalid - // local declaration will produce a hard error; if it doesn't - // remain visible, a single bogus local redeclaration (which is - // actually only a warning) could break all the downstream code. - if (!New->getLexicalDeclContext()->isFunctionOrMethod()) - New->getIdentifier()->revertBuiltin(); - return false; } @@ -9636,6 +9629,36 @@ Sema::ActOnFunctionDeclarator(Scope *S, Declarator &D, DeclContext *DC, } } + // If this is the first declaration of a library builtin function, add + // attributes as appropriate. + if (!D.isRedeclaration() && + NewFD->getDeclContext()->getRedeclContext()->isFileContext()) { + if (IdentifierInfo *II = Previous.getLookupName().getAsIdentifierInfo()) { + if (unsigned BuiltinID = II->getBuiltinID()) { + if (NewFD->getLanguageLinkage() == CLanguageLinkage) { + // Validate the type matches unless this builtin is specified as + // matching regardless of its declared type. + if (Context.BuiltinInfo.allowTypeMismatch(BuiltinID)) { + NewFD->addAttr(BuiltinAttr::CreateImplicit(Context, BuiltinID)); + } else { + ASTContext::GetBuiltinTypeError Error; + LookupPredefedObjCSuperType(*this, S, NewFD->getIdentifier()); + QualType BuiltinType = Context.GetBuiltinType(BuiltinID, Error); + + if (!Error && !BuiltinType.isNull() && + Context.hasSameFunctionTypeIgnoringExceptionSpec( + NewFD->getType(), BuiltinType)) + NewFD->addAttr(BuiltinAttr::CreateImplicit(Context, BuiltinID)); + } + } else if (BuiltinID == Builtin::BI__GetExceptionInfo && + Context.getTargetInfo().getCXXABI().isMicrosoft()) { + // FIXME: We should consider this a builtin only in the std namespace. + NewFD->addAttr(BuiltinAttr::CreateImplicit(Context, BuiltinID)); + } + } + } + } + ProcessPragmaWeak(S, NewFD); checkAttributesAfterMerging(*this, *NewFD); diff --git a/clang/lib/Sema/SemaDeclAttr.cpp b/clang/lib/Sema/SemaDeclAttr.cpp index 1a0594512a60..a9a2a19b4797 100644 --- a/clang/lib/Sema/SemaDeclAttr.cpp +++ b/clang/lib/Sema/SemaDeclAttr.cpp @@ -3301,7 +3301,11 @@ static void handleInitPriorityAttr(Sema &S, Decl *D, const ParsedAttr &AL) { return; } - if (prioritynum < 101 || prioritynum > 65535) { + // Only perform the priority check if the attribute is outside of a system + // header. Values <= 100 are reserved for the implementation, and libc++ + // benefits from being able to specify values in that range. + if ((prioritynum < 101 || prioritynum > 65535) && + !S.getSourceManager().isInSystemHeader(AL.getLoc())) { S.Diag(AL.getLoc(), diag::err_attribute_argument_out_of_range) << E->getSourceRange() << AL << 101 << 65535; AL.setInvalid(); diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp index ccae79636f32..d301e6c732ab 100644 --- a/clang/lib/Sema/SemaExpr.cpp +++ b/clang/lib/Sema/SemaExpr.cpp @@ -6158,6 +6158,7 @@ static FunctionDecl *rewriteBuiltinFunctionDecl(Sema *Sema, ASTContext &Context, Params.push_back(Parm); } OverloadDecl->setParams(Params); + Sema->mergeDeclAttributes(OverloadDecl, FDecl); return OverloadDecl; } @@ -7401,7 +7402,7 @@ Sema::ActOnCastExpr(Scope *S, SourceLocation LParenLoc, } if (PE || PLE->getNumExprs() == 1) { Expr *E = (PE ? PE->getSubExpr() : PLE->getExpr(0)); - if (!E->getType()->isVectorType()) + if (!E->isTypeDependent() && !E->getType()->isVectorType()) isVectorLiteral = true; } else diff --git a/clang/lib/Sema/SemaLookup.cpp b/clang/lib/Sema/SemaLookup.cpp index 5757eaf3fac0..bcbecd545398 100644 --- a/clang/lib/Sema/SemaLookup.cpp +++ b/clang/lib/Sema/SemaLookup.cpp @@ -894,10 +894,9 @@ bool Sema::LookupBuiltin(LookupResult &R) { Context.BuiltinInfo.isPredefinedLibFunction(BuiltinID)) return false; - if (NamedDecl *D = LazilyCreateBuiltin((IdentifierInfo *)II, - BuiltinID, TUScope, - R.isForRedeclaration(), - R.getNameLoc())) { + if (NamedDecl *D = + LazilyCreateBuiltin(II, BuiltinID, TUScope, + R.isForRedeclaration(), R.getNameLoc())) { R.addDecl(D); return true; } diff --git a/clang/lib/Sema/SemaTemplate.cpp b/clang/lib/Sema/SemaTemplate.cpp index c05ed0b14e3e..f788cf103503 100644 --- a/clang/lib/Sema/SemaTemplate.cpp +++ b/clang/lib/Sema/SemaTemplate.cpp @@ -1963,8 +1963,7 @@ class ExtractTypeForDeductionGuide TypeLocBuilder InnerTLB; QualType Transformed = TransformType(InnerTLB, OrigDecl->getTypeSourceInfo()->getTypeLoc()); - TypeSourceInfo *TSI = - TransformType(InnerTLB.getTypeSourceInfo(Context, Transformed)); + TypeSourceInfo *TSI = InnerTLB.getTypeSourceInfo(Context, Transformed); TypedefNameDecl *Decl = nullptr; diff --git a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp index baec13ba627c..7e6efe6105bf 100644 --- a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp +++ b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp @@ -3629,8 +3629,11 @@ Decl *TemplateDeclInstantiator::VisitVarTemplateSpecializationDecl( SemaRef.Context, Owner, D->getInnerLocStart(), D->getLocation(), VarTemplate, DI->getType(), DI, D->getStorageClass(), Converted); Var->setTemplateArgsInfo(TemplateArgsInfo); - if (InsertPos) + if (!PrevDecl) { + void *InsertPos = nullptr; + VarTemplate->findSpecialization(Converted, InsertPos); VarTemplate->AddSpecialization(Var, InsertPos); + } if (SemaRef.getLangOpts().OpenCL) SemaRef.deduceOpenCLAddressSpace(Var); @@ -5311,7 +5314,7 @@ void Sema::InstantiateVariableDefinition(SourceLocation PointOfInstantiation, TemplateArgs); Var = cast_or_null(Instantiator.VisitVarTemplateSpecializationDecl( VarSpec->getSpecializedTemplate(), Def, nullptr, - VarSpec->getTemplateArgsInfo(), VarSpec->getTemplateArgs().asArray())); + VarSpec->getTemplateArgsInfo(), VarSpec->getTemplateArgs().asArray(), VarSpec)); if (Var) { llvm::PointerUnion PatternPtr = @@ -5321,12 +5324,6 @@ void Sema::InstantiateVariableDefinition(SourceLocation PointOfInstantiation, cast(Var)->setInstantiationOf( Partial, &VarSpec->getTemplateInstantiationArgs()); - // Merge the definition with the declaration. - LookupResult R(*this, Var->getDeclName(), Var->getLocation(), - LookupOrdinaryName, forRedeclarationInCurContext()); - R.addDecl(OldVar); - MergeVarDecl(Var, R); - // Attach the initializer. InstantiateVariableInitializer(Var, Def, TemplateArgs); } diff --git a/clang/lib/Sema/UsedDeclVisitor.h b/clang/lib/Sema/UsedDeclVisitor.h index d207e07f451a..c33d30478e2a 100644 --- a/clang/lib/Sema/UsedDeclVisitor.h +++ b/clang/lib/Sema/UsedDeclVisitor.h @@ -67,10 +67,13 @@ class UsedDeclVisitor : public EvaluatedExprVisitor { void VisitCXXDeleteExpr(CXXDeleteExpr *E) { if (E->getOperatorDelete()) asImpl().visitUsedDecl(E->getBeginLoc(), E->getOperatorDelete()); - QualType Destroyed = S.Context.getBaseElementType(E->getDestroyedType()); - if (const RecordType *DestroyedRec = Destroyed->getAs()) { - CXXRecordDecl *Record = cast(DestroyedRec->getDecl()); - asImpl().visitUsedDecl(E->getBeginLoc(), S.LookupDestructor(Record)); + QualType DestroyedOrNull = E->getDestroyedType(); + if (!DestroyedOrNull.isNull()) { + QualType Destroyed = S.Context.getBaseElementType(DestroyedOrNull); + if (const RecordType *DestroyedRec = Destroyed->getAs()) { + CXXRecordDecl *Record = cast(DestroyedRec->getDecl()); + asImpl().visitUsedDecl(E->getBeginLoc(), S.LookupDestructor(Record)); + } } Inherited::VisitCXXDeleteExpr(E); diff --git a/clang/lib/Serialization/ASTReader.cpp b/clang/lib/Serialization/ASTReader.cpp index 4a1a995204e5..4d439691ef83 100644 --- a/clang/lib/Serialization/ASTReader.cpp +++ b/clang/lib/Serialization/ASTReader.cpp @@ -908,9 +908,8 @@ ASTIdentifierLookupTraitBase::ReadKey(const unsigned char* d, unsigned n) { /// Whether the given identifier is "interesting". static bool isInterestingIdentifier(ASTReader &Reader, IdentifierInfo &II, bool IsModule) { - return II.hadMacroDefinition() || - II.isPoisoned() || - (IsModule ? II.hasRevertedBuiltin() : II.getObjCOrBuiltinID()) || + return II.hadMacroDefinition() || II.isPoisoned() || + (!IsModule && II.getObjCOrBuiltinID()) || II.hasRevertedTokenIDToIdentifier() || (!(IsModule && Reader.getPreprocessor().getLangOpts().CPlusPlus) && II.getFETokenInfo()); @@ -970,7 +969,6 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k, unsigned Bits = endian::readNext(d); bool CPlusPlusOperatorKeyword = readBit(Bits); bool HasRevertedTokenIDToIdentifier = readBit(Bits); - bool HasRevertedBuiltin = readBit(Bits); bool Poisoned = readBit(Bits); bool ExtensionToken = readBit(Bits); bool HadMacroDefinition = readBit(Bits); @@ -984,12 +982,6 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k, II->revertTokenIDToIdentifier(); if (!F.isModule()) II->setObjCOrBuiltinID(ObjCOrBuiltinID); - else if (HasRevertedBuiltin && II->getBuiltinID()) { - II->revertBuiltin(); - assert((II->hasRevertedBuiltin() || - II->getObjCOrBuiltinID() == ObjCOrBuiltinID) && - "Incorrect ObjC keyword or builtin ID"); - } assert(II->isExtensionToken() == ExtensionToken && "Incorrect extension token flag"); (void)ExtensionToken; diff --git a/clang/lib/Serialization/ASTWriter.cpp b/clang/lib/Serialization/ASTWriter.cpp index 2345a12caeb2..16e363fdd42c 100644 --- a/clang/lib/Serialization/ASTWriter.cpp +++ b/clang/lib/Serialization/ASTWriter.cpp @@ -3276,9 +3276,8 @@ class ASTIdentifierTableTrait { /// doesn't check whether the name has macros defined; use PublicMacroIterator /// to check that. bool isInterestingIdentifier(const IdentifierInfo *II, uint64_t MacroOffset) { - if (MacroOffset || - II->isPoisoned() || - (IsModule ? II->hasRevertedBuiltin() : II->getObjCOrBuiltinID()) || + if (MacroOffset || II->isPoisoned() || + (!IsModule && II->getObjCOrBuiltinID()) || II->hasRevertedTokenIDToIdentifier() || (NeedDecls && II->getFETokenInfo())) return true; @@ -3385,7 +3384,6 @@ class ASTIdentifierTableTrait { Bits = (Bits << 1) | unsigned(HadMacroDefinition); Bits = (Bits << 1) | unsigned(II->isExtensionToken()); Bits = (Bits << 1) | unsigned(II->isPoisoned()); - Bits = (Bits << 1) | unsigned(II->hasRevertedBuiltin()); Bits = (Bits << 1) | unsigned(II->hasRevertedTokenIDToIdentifier()); Bits = (Bits << 1) | unsigned(II->isCPlusPlusOperatorKeyword()); LE.write(Bits); diff --git a/clang/lib/StaticAnalyzer/Checkers/Yaml.h b/clang/lib/StaticAnalyzer/Checkers/Yaml.h old mode 100755 new mode 100644 diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_common_interceptors_ioctl.inc b/compiler-rt/lib/sanitizer_common/sanitizer_common_interceptors_ioctl.inc old mode 100755 new mode 100644 diff --git a/lld/COFF/DebugTypes.cpp b/lld/COFF/DebugTypes.cpp index 4790b0166799..abe3bb9eef5b 100644 --- a/lld/COFF/DebugTypes.cpp +++ b/lld/COFF/DebugTypes.cpp @@ -202,6 +202,9 @@ Expected TpiSource::mergeDebugT(TypeMerger *m, BinaryStreamReader reader(file->debugTypes, support::little); cantFail(reader.readArray(types, reader.getLength())); + // When dealing with PCH.OBJ, some indices were already merged. + unsigned nbHeadIndices = indexMap->tpiMap.size(); + if (config->debugGHashes) { ArrayRef hashes; std::vector ownedHashes; @@ -232,7 +235,7 @@ Expected TpiSource::mergeDebugT(TypeMerger *m, // collecting statistics. m->tpiCounts.resize(m->getTypeTable().size()); m->ipiCounts.resize(m->getIDTable().size()); - uint32_t srcIdx = 0; + uint32_t srcIdx = nbHeadIndices; for (CVType &ty : types) { TypeIndex dstIdx = indexMap->tpiMap[srcIdx++]; // Type merging may fail, so a complex source type may become the simple diff --git a/lld/ELF/LinkerScript.cpp b/lld/ELF/LinkerScript.cpp index 6de2cd65b973..7314b27659bb 100644 --- a/lld/ELF/LinkerScript.cpp +++ b/lld/ELF/LinkerScript.cpp @@ -679,8 +679,11 @@ addInputSec(StringMap> &map, auto *firstIsec = cast( cast(sec->sectionCommands[0]) ->sectionBases[0]); - if (firstIsec->getLinkOrderDep()->getOutputSection() != - isec->getLinkOrderDep()->getOutputSection()) + OutputSection *firstIsecOut = + firstIsec->flags & SHF_LINK_ORDER + ? firstIsec->getLinkOrderDep()->getOutputSection() + : nullptr; + if (firstIsecOut != isec->getLinkOrderDep()->getOutputSection()) continue; } diff --git a/lldb/include/lldb/Core/ModuleList.h b/lldb/include/lldb/Core/ModuleList.h index ae1f6fdb20a2..d90b27e474ac 100644 --- a/lldb/include/lldb/Core/ModuleList.h +++ b/lldb/include/lldb/Core/ModuleList.h @@ -139,7 +139,13 @@ class ModuleList { /// /// \param[in] module_sp /// A shared pointer to a module to replace in this collection. - void ReplaceEquivalent(const lldb::ModuleSP &module_sp); + /// + /// \param[in] old_modules + /// Optional pointer to a vector which, if provided, will have shared + /// pointers to the replaced module(s) appended to it. + void ReplaceEquivalent( + const lldb::ModuleSP &module_sp, + llvm::SmallVectorImpl *old_modules = nullptr); /// Append a module to the module list, if it is not already there. /// @@ -443,12 +449,11 @@ class ModuleList { static bool ModuleIsInCache(const Module *module_ptr); - static Status GetSharedModule(const ModuleSpec &module_spec, - lldb::ModuleSP &module_sp, - const FileSpecList *module_search_paths_ptr, - lldb::ModuleSP *old_module_sp_ptr, - bool *did_create_ptr, - bool always_create = false); + static Status + GetSharedModule(const ModuleSpec &module_spec, lldb::ModuleSP &module_sp, + const FileSpecList *module_search_paths_ptr, + llvm::SmallVectorImpl *old_modules, + bool *did_create_ptr, bool always_create = false); static bool RemoveSharedModule(lldb::ModuleSP &module_sp); diff --git a/lldb/include/lldb/Target/Platform.h b/lldb/include/lldb/Target/Platform.h index 6234b8244b3f..277fcf68cb0c 100644 --- a/lldb/include/lldb/Target/Platform.h +++ b/lldb/include/lldb/Target/Platform.h @@ -301,11 +301,10 @@ class Platform : public PluginInterface { LocateExecutableScriptingResources(Target *target, Module &module, Stream *feedback_stream); - virtual Status GetSharedModule(const ModuleSpec &module_spec, - Process *process, lldb::ModuleSP &module_sp, - const FileSpecList *module_search_paths_ptr, - lldb::ModuleSP *old_module_sp_ptr, - bool *did_create_ptr); + virtual Status GetSharedModule( + const ModuleSpec &module_spec, Process *process, + lldb::ModuleSP &module_sp, const FileSpecList *module_search_paths_ptr, + llvm::SmallVectorImpl *old_modules, bool *did_create_ptr); virtual bool GetModuleSpec(const FileSpec &module_file_spec, const ArchSpec &arch, ModuleSpec &module_spec); diff --git a/lldb/source/Core/ModuleList.cpp b/lldb/source/Core/ModuleList.cpp index 0345678ddaff..1701cb56338e 100644 --- a/lldb/source/Core/ModuleList.cpp +++ b/lldb/source/Core/ModuleList.cpp @@ -171,7 +171,9 @@ void ModuleList::Append(const ModuleSP &module_sp, bool notify) { AppendImpl(module_sp, notify); } -void ModuleList::ReplaceEquivalent(const ModuleSP &module_sp) { +void ModuleList::ReplaceEquivalent( + const ModuleSP &module_sp, + llvm::SmallVectorImpl *old_modules) { if (module_sp) { std::lock_guard guard(m_modules_mutex); @@ -184,11 +186,14 @@ void ModuleList::ReplaceEquivalent(const ModuleSP &module_sp) { size_t idx = 0; while (idx < m_modules.size()) { - ModuleSP module_sp(m_modules[idx]); - if (module_sp->MatchesModuleSpec(equivalent_module_spec)) + ModuleSP test_module_sp(m_modules[idx]); + if (test_module_sp->MatchesModuleSpec(equivalent_module_spec)) { + if (old_modules) + old_modules->push_back(test_module_sp); RemoveImpl(m_modules.begin() + idx); - else + } else { ++idx; + } } // Now add the new module to the list Append(module_sp); @@ -731,11 +736,11 @@ size_t ModuleList::RemoveOrphanSharedModules(bool mandatory) { return GetSharedModuleList().RemoveOrphans(mandatory); } -Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, - ModuleSP &module_sp, - const FileSpecList *module_search_paths_ptr, - ModuleSP *old_module_sp_ptr, - bool *did_create_ptr, bool always_create) { +Status +ModuleList::GetSharedModule(const ModuleSpec &module_spec, ModuleSP &module_sp, + const FileSpecList *module_search_paths_ptr, + llvm::SmallVectorImpl *old_modules, + bool *did_create_ptr, bool always_create) { ModuleList &shared_module_list = GetSharedModuleList(); std::lock_guard guard( shared_module_list.m_modules_mutex); @@ -747,8 +752,6 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, if (did_create_ptr) *did_create_ptr = false; - if (old_module_sp_ptr) - old_module_sp_ptr->reset(); const UUID *uuid_ptr = module_spec.GetUUIDPtr(); const FileSpec &module_file_spec = module_spec.GetFileSpec(); @@ -769,8 +772,8 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, // Make sure the file for the module hasn't been modified if (module_sp->FileHasChanged()) { - if (old_module_sp_ptr && !*old_module_sp_ptr) - *old_module_sp_ptr = module_sp; + if (old_modules) + old_modules->push_back(module_sp); Log *log(lldb_private::GetLogIfAnyCategoriesSet(LIBLLDB_LOG_MODULES)); if (log != nullptr) @@ -812,7 +815,7 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, *did_create_ptr = true; } - shared_module_list.ReplaceEquivalent(module_sp); + shared_module_list.ReplaceEquivalent(module_sp, old_modules); return error; } } @@ -849,7 +852,7 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, if (did_create_ptr) *did_create_ptr = true; - shared_module_list.ReplaceEquivalent(module_sp); + shared_module_list.ReplaceEquivalent(module_sp, old_modules); return Status(); } } @@ -924,8 +927,8 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, located_binary_modulespec.GetFileSpec()); if (file_spec_mod_time != llvm::sys::TimePoint<>()) { if (file_spec_mod_time != module_sp->GetModificationTime()) { - if (old_module_sp_ptr) - *old_module_sp_ptr = module_sp; + if (old_modules) + old_modules->push_back(module_sp); shared_module_list.Remove(module_sp); module_sp.reset(); } @@ -947,7 +950,7 @@ Status ModuleList::GetSharedModule(const ModuleSpec &module_spec, if (did_create_ptr) *did_create_ptr = true; - shared_module_list.ReplaceEquivalent(module_sp); + shared_module_list.ReplaceEquivalent(module_sp, old_modules); } } else { located_binary_modulespec.GetFileSpec().GetPath(path, sizeof(path)); diff --git a/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp b/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp index fc8ee346f449..1041f63aa2e2 100644 --- a/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp +++ b/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp @@ -121,6 +121,72 @@ class PlaceholderObjectFile : public ObjectFile { lldb::addr_t m_base; lldb::addr_t m_size; }; + +/// Duplicate the HashElfTextSection() from the breakpad sources. +/// +/// Breakpad, a Google crash log reporting tool suite, creates minidump files +/// for many different architectures. When using Breakpad to create ELF +/// minidumps, it will check for a GNU build ID when creating a minidump file +/// and if one doesn't exist in the file, it will say the UUID of the file is a +/// checksum of up to the first 4096 bytes of the .text section. Facebook also +/// uses breakpad and modified this hash to avoid collisions so we can +/// calculate and check for this as well. +/// +/// The breakpad code might end up hashing up to 15 bytes that immediately +/// follow the .text section in the file, so this code must do exactly what it +/// does so we can get an exact match for the UUID. +/// +/// \param[in] module_sp The module to grab the .text section from. +/// +/// \param[in/out] breakpad_uuid A vector that will receive the calculated +/// breakpad .text hash. +/// +/// \param[in/out] facebook_uuid A vector that will receive the calculated +/// facebook .text hash. +/// +void HashElfTextSection(ModuleSP module_sp, std::vector &breakpad_uuid, + std::vector &facebook_uuid) { + SectionList *sect_list = module_sp->GetSectionList(); + if (sect_list == nullptr) + return; + SectionSP sect_sp = sect_list->FindSectionByName(ConstString(".text")); + if (!sect_sp) + return; + constexpr size_t kMDGUIDSize = 16; + constexpr size_t kBreakpadPageSize = 4096; + // The breakpad code has a bug where it might access beyond the end of a + // .text section by up to 15 bytes, so we must ensure we round up to the + // next kMDGUIDSize byte boundary. + DataExtractor data; + const size_t text_size = sect_sp->GetFileSize(); + const size_t read_size = std::min( + llvm::alignTo(text_size, kMDGUIDSize), kBreakpadPageSize); + sect_sp->GetObjectFile()->GetData(sect_sp->GetFileOffset(), read_size, data); + + breakpad_uuid.assign(kMDGUIDSize, 0); + facebook_uuid.assign(kMDGUIDSize, 0); + + // The only difference between the breakpad hash and the facebook hash is the + // hashing of the text section size into the hash prior to hashing the .text + // contents. + for (size_t i = 0; i < kMDGUIDSize; i++) + facebook_uuid[i] ^= text_size % 255; + + // This code carefully duplicates how the hash was created in Breakpad + // sources, including the error where it might has an extra 15 bytes past the + // end of the .text section if the .text section is less than a page size in + // length. + const uint8_t *ptr = data.GetDataStart(); + const uint8_t *ptr_end = data.GetDataEnd(); + while (ptr < ptr_end) { + for (unsigned i = 0; i < kMDGUIDSize; i++) { + breakpad_uuid[i] ^= ptr[i]; + facebook_uuid[i] ^= ptr[i]; + } + ptr += kMDGUIDSize; + } +} + } // namespace ConstString ProcessMinidump::GetPluginNameStatic() { @@ -444,6 +510,53 @@ bool ProcessMinidump::UpdateThreadList(ThreadList &old_thread_list, return new_thread_list.GetSize(false) > 0; } +ModuleSP ProcessMinidump::GetOrCreateModule(UUID minidump_uuid, + llvm::StringRef name, + ModuleSpec module_spec) { + Log *log(GetLogIfAllCategoriesSet(LIBLLDB_LOG_DYNAMIC_LOADER)); + Status error; + + ModuleSP module_sp = + GetTarget().GetOrCreateModule(module_spec, true /* notify */, &error); + if (!module_sp) + return module_sp; + // We consider the module to be a match if the minidump UUID is a + // prefix of the actual UUID, or if either of the UUIDs are empty. + const auto dmp_bytes = minidump_uuid.GetBytes(); + const auto mod_bytes = module_sp->GetUUID().GetBytes(); + const bool match = dmp_bytes.empty() || mod_bytes.empty() || + mod_bytes.take_front(dmp_bytes.size()) == dmp_bytes; + if (match) { + LLDB_LOG(log, "Partial uuid match for {0}.", name); + return module_sp; + } + + // Breakpad generates minindump files, and if there is no GNU build + // ID in the binary, it will calculate a UUID by hashing first 4096 + // bytes of the .text section and using that as the UUID for a module + // in the minidump. Facebook uses a modified breakpad client that + // uses a slightly modified this hash to avoid collisions. Check for + // UUIDs from the minindump that match these cases and accept the + // module we find if they do match. + std::vector breakpad_uuid; + std::vector facebook_uuid; + HashElfTextSection(module_sp, breakpad_uuid, facebook_uuid); + if (dmp_bytes == llvm::ArrayRef(breakpad_uuid)) { + LLDB_LOG(log, "Breakpad .text hash match for {0}.", name); + return module_sp; + } + if (dmp_bytes == llvm::ArrayRef(facebook_uuid)) { + LLDB_LOG(log, "Facebook .text hash match for {0}.", name); + return module_sp; + } + // The UUID wasn't a partial match and didn't match the .text hash + // so remove the module from the target, we will need to create a + // placeholder object file. + GetTarget().GetImages().Remove(module_sp); + module_sp.reset(); + return module_sp; +} + void ProcessMinidump::ReadModuleList() { std::vector filtered_modules = m_minidump_parser->GetFilteredModuleList(); @@ -473,30 +586,21 @@ void ProcessMinidump::ReadModuleList() { // add the module to the target if it finds one. lldb::ModuleSP module_sp = GetTarget().GetOrCreateModule(module_spec, true /* notify */, &error); - if (!module_sp) { - // Try and find a module without specifying the UUID and only looking for - // the file given a basename. We then will look for a partial UUID match - // if we find any matches. This function will add the module to the - // target if it finds one, so we need to remove the module from the target - // if the UUID doesn't match during our manual UUID verification. This - // allows the "target.exec-search-paths" setting to specify one or more - // directories that contain executables that can be searched for matches. - ModuleSpec basename_module_spec(module_spec); - basename_module_spec.GetUUID().Clear(); - basename_module_spec.GetFileSpec().GetDirectory().Clear(); - module_sp = GetTarget().GetOrCreateModule(basename_module_spec, - true /* notify */, &error); - if (module_sp) { - // We consider the module to be a match if the minidump UUID is a - // prefix of the actual UUID, or if either of the UUIDs are empty. - const auto dmp_bytes = uuid.GetBytes(); - const auto mod_bytes = module_sp->GetUUID().GetBytes(); - const bool match = dmp_bytes.empty() || mod_bytes.empty() || - mod_bytes.take_front(dmp_bytes.size()) == dmp_bytes; - if (!match) { - GetTarget().GetImages().Remove(module_sp); - module_sp.reset(); - } + if (module_sp) { + LLDB_LOG(log, "Full uuid match for {0}.", name); + } else { + // We couldn't find a module with an exactly-matching UUID. Sometimes + // a minidump UUID is only a partial match or is a hash. So try again + // without specifying the UUID, then again without specifying the + // directory if that fails. This will allow us to find modules with + // partial matches or hash UUIDs in user-provided sysroots or search + // directories (target.exec-search-paths). + ModuleSpec partial_module_spec = module_spec; + partial_module_spec.GetUUID().Clear(); + module_sp = GetOrCreateModule(uuid, name, partial_module_spec); + if (!module_sp) { + partial_module_spec.GetFileSpec().GetDirectory().Clear(); + module_sp = GetOrCreateModule(uuid, name, partial_module_spec); } } if (module_sp) { diff --git a/lldb/source/Plugins/Process/minidump/ProcessMinidump.h b/lldb/source/Plugins/Process/minidump/ProcessMinidump.h index 839b0e7563f7..bfdace7ea33e 100644 --- a/lldb/source/Plugins/Process/minidump/ProcessMinidump.h +++ b/lldb/source/Plugins/Process/minidump/ProcessMinidump.h @@ -102,6 +102,10 @@ class ProcessMinidump : public Process { void ReadModuleList(); + lldb::ModuleSP GetOrCreateModule(lldb_private::UUID minidump_uuid, + llvm::StringRef name, + lldb_private::ModuleSpec module_spec); + JITLoaderList &GetJITLoaders() override; private: diff --git a/lldb/source/Target/Platform.cpp b/lldb/source/Target/Platform.cpp index 16787141bee0..e5afb4c7b8d7 100644 --- a/lldb/source/Target/Platform.cpp +++ b/lldb/source/Target/Platform.cpp @@ -218,15 +218,14 @@ Platform::LocateExecutableScriptingResources(Target *target, Module &module, // return PlatformSP(); //} -Status Platform::GetSharedModule(const ModuleSpec &module_spec, - Process *process, ModuleSP &module_sp, - const FileSpecList *module_search_paths_ptr, - ModuleSP *old_module_sp_ptr, - bool *did_create_ptr) { +Status Platform::GetSharedModule( + const ModuleSpec &module_spec, Process *process, ModuleSP &module_sp, + const FileSpecList *module_search_paths_ptr, + llvm::SmallVectorImpl *old_modules, bool *did_create_ptr) { if (IsHost()) - return ModuleList::GetSharedModule( - module_spec, module_sp, module_search_paths_ptr, old_module_sp_ptr, - did_create_ptr, false); + return ModuleList::GetSharedModule(module_spec, module_sp, + module_search_paths_ptr, old_modules, + did_create_ptr, false); // Module resolver lambda. auto resolver = [&](const ModuleSpec &spec) { @@ -239,17 +238,17 @@ Status Platform::GetSharedModule(const ModuleSpec &module_spec, resolved_spec.GetFileSpec().PrependPathComponent( m_sdk_sysroot.GetStringRef()); // Try to get shared module with resolved spec. - error = ModuleList::GetSharedModule( - resolved_spec, module_sp, module_search_paths_ptr, old_module_sp_ptr, - did_create_ptr, false); + error = ModuleList::GetSharedModule(resolved_spec, module_sp, + module_search_paths_ptr, old_modules, + did_create_ptr, false); } // If we don't have sysroot or it didn't work then // try original module spec. if (!error.Success()) { resolved_spec = spec; - error = ModuleList::GetSharedModule( - resolved_spec, module_sp, module_search_paths_ptr, old_module_sp_ptr, - did_create_ptr, false); + error = ModuleList::GetSharedModule(resolved_spec, module_sp, + module_search_paths_ptr, old_modules, + did_create_ptr, false); } if (error.Success() && module_sp) module_sp->SetPlatformFileSpec(resolved_spec.GetFileSpec()); @@ -1564,21 +1563,29 @@ Status Platform::GetRemoteSharedModule(const ModuleSpec &module_spec, if (error.Success() && module_sp) break; } - if (module_sp) + if (module_sp) { + resolved_module_spec = arch_module_spec; got_module_spec = true; + } } if (!got_module_spec) { // Get module information from a target. - if (!GetModuleSpec(module_spec.GetFileSpec(), module_spec.GetArchitecture(), - resolved_module_spec)) { + if (GetModuleSpec(module_spec.GetFileSpec(), module_spec.GetArchitecture(), + resolved_module_spec)) { if (!module_spec.GetUUID().IsValid() || module_spec.GetUUID() == resolved_module_spec.GetUUID()) { - return module_resolver(module_spec); + got_module_spec = true; } } } + if (!got_module_spec) { + // Fall back to the given module resolver, which may have its own + // search logic. + return module_resolver(module_spec); + } + // If we are looking for a specific UUID, make sure resolved_module_spec has // the same one before we search. if (module_spec.GetUUID().IsValid()) { diff --git a/lldb/source/Target/Target.cpp b/lldb/source/Target/Target.cpp index 707344f99fcb..19d0c3d477eb 100644 --- a/lldb/source/Target/Target.cpp +++ b/lldb/source/Target/Target.cpp @@ -1965,8 +1965,9 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, module_sp = m_images.FindFirstModule(module_spec); if (!module_sp) { - ModuleSP old_module_sp; // This will get filled in if we have a new version - // of the library + llvm::SmallVector + old_modules; // This will get filled in if we have a new version + // of the library bool did_create_module = false; FileSpecList search_paths = GetExecutableSearchPaths(); // If there are image search path entries, try to use them first to acquire @@ -1979,7 +1980,7 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, transformed_spec.GetFileSpec().GetFilename() = module_spec.GetFileSpec().GetFilename(); error = ModuleList::GetSharedModule(transformed_spec, module_sp, - &search_paths, &old_module_sp, + &search_paths, &old_modules, &did_create_module); } } @@ -1997,7 +1998,7 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, // We have a UUID, it is OK to check the global module list... error = ModuleList::GetSharedModule(module_spec, module_sp, &search_paths, - &old_module_sp, &did_create_module); + &old_modules, &did_create_module); } if (!module_sp) { @@ -2006,7 +2007,7 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, if (m_platform_sp) { error = m_platform_sp->GetSharedModule( module_spec, m_process_sp.get(), module_sp, &search_paths, - &old_module_sp, &did_create_module); + &old_modules, &did_create_module); } else { error.SetErrorString("no platform is currently set"); } @@ -2057,18 +2058,18 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, // this target. So let's remove the UUID from the module list, and look // in the target's module list. Only do this if there is SOMETHING else // in the module spec... - if (!old_module_sp) { - if (module_spec.GetUUID().IsValid() && - !module_spec.GetFileSpec().GetFilename().IsEmpty() && - !module_spec.GetFileSpec().GetDirectory().IsEmpty()) { - ModuleSpec module_spec_copy(module_spec.GetFileSpec()); - module_spec_copy.GetUUID().Clear(); + if (module_spec.GetUUID().IsValid() && + !module_spec.GetFileSpec().GetFilename().IsEmpty() && + !module_spec.GetFileSpec().GetDirectory().IsEmpty()) { + ModuleSpec module_spec_copy(module_spec.GetFileSpec()); + module_spec_copy.GetUUID().Clear(); - ModuleList found_modules; - m_images.FindModules(module_spec_copy, found_modules); - if (found_modules.GetSize() == 1) - old_module_sp = found_modules.GetModuleAtIndex(0); - } + ModuleList found_modules; + m_images.FindModules(module_spec_copy, found_modules); + found_modules.ForEach([&](const ModuleSP &found_module) -> bool { + old_modules.push_back(found_module); + return true; + }); } // Preload symbols outside of any lock, so hopefully we can do this for @@ -2076,14 +2077,67 @@ ModuleSP Target::GetOrCreateModule(const ModuleSpec &module_spec, bool notify, if (GetPreloadSymbols()) module_sp->PreloadSymbols(); - if (old_module_sp && m_images.GetIndexForModule(old_module_sp.get()) != - LLDB_INVALID_INDEX32) { - m_images.ReplaceModule(old_module_sp, module_sp); + llvm::SmallVector replaced_modules; + for (ModuleSP &old_module_sp : old_modules) { + if (m_images.GetIndexForModule(old_module_sp.get()) != + LLDB_INVALID_INDEX32) { + if (replaced_modules.empty()) + m_images.ReplaceModule(old_module_sp, module_sp); + else + m_images.Remove(old_module_sp); + + replaced_modules.push_back(std::move(old_module_sp)); + } + } + + if (replaced_modules.size() > 1) { + // The same new module replaced multiple old modules + // simultaneously. It's not clear this should ever + // happen (if we always replace old modules as we add + // new ones, presumably we should never have more than + // one old one). If there are legitimate cases where + // this happens, then the ModuleList::Notifier interface + // may need to be adjusted to allow reporting this. + // In the meantime, just log that this has happened; just + // above we called ReplaceModule on the first one, and Remove + // on the rest. + if (Log *log = GetLogIfAnyCategoriesSet(LIBLLDB_LOG_TARGET | + LIBLLDB_LOG_MODULES)) { + StreamString message; + auto dump = [&message](Module &dump_module) -> void { + UUID dump_uuid = dump_module.GetUUID(); + + message << '['; + dump_module.GetDescription(message.AsRawOstream()); + message << " (uuid "; + + if (dump_uuid.IsValid()) + dump_uuid.Dump(&message); + else + message << "not specified"; + + message << ")]"; + }; + + message << "New module "; + dump(*module_sp); + message.AsRawOstream() + << llvm::formatv(" simultaneously replaced {0} old modules: ", + replaced_modules.size()); + for (ModuleSP &replaced_module_sp : replaced_modules) + dump(*replaced_module_sp); + + log->PutString(message.GetString()); + } + } + + if (replaced_modules.empty()) + m_images.Append(module_sp, notify); + + for (ModuleSP &old_module_sp : replaced_modules) { Module *old_module_ptr = old_module_sp.get(); old_module_sp.reset(); ModuleList::RemoveSharedModuleIfOrphaned(old_module_ptr); - } else { - m_images.Append(module_sp, notify); } } else module_sp.reset(); diff --git a/llvm/include/llvm/ADT/ImmutableMap.h b/llvm/include/llvm/ADT/ImmutableMap.h index 30689d2274a8..81b21a7319a7 100644 --- a/llvm/include/llvm/ADT/ImmutableMap.h +++ b/llvm/include/llvm/ADT/ImmutableMap.h @@ -355,7 +355,7 @@ class ImmutableMapRef { unsigned getHeight() const { return Root ? Root->getHeight() : 0; } static inline void Profile(FoldingSetNodeID &ID, const ImmutableMapRef &M) { - ID.AddPointer(M.Root); + ID.AddPointer(M.Root.get()); } inline void Profile(FoldingSetNodeID &ID) const { return Profile(ID, *this); } diff --git a/llvm/include/llvm/BinaryFormat/ELF.h b/llvm/include/llvm/BinaryFormat/ELF.h index bdcf10fd1640..21a5c26883cd 100644 --- a/llvm/include/llvm/BinaryFormat/ELF.h +++ b/llvm/include/llvm/BinaryFormat/ELF.h @@ -405,6 +405,12 @@ enum { #include "ELFRelocs/AArch64.def" }; +// Special values for the st_other field in the symbol table entry for AArch64. +enum { + // Symbol may follow different calling convention than base PCS. + STO_AARCH64_VARIANT_PCS = 0x80 +}; + // ARM Specific e_flags enum : unsigned { EF_ARM_SOFT_FLOAT = 0x00000200U, // Legacy pre EABI_VER5 diff --git a/llvm/include/llvm/Transforms/IPO/DeadArgumentElimination.h b/llvm/include/llvm/Transforms/IPO/DeadArgumentElimination.h index 73797bc10017..496ceea12bc9 100644 --- a/llvm/include/llvm/Transforms/IPO/DeadArgumentElimination.h +++ b/llvm/include/llvm/Transforms/IPO/DeadArgumentElimination.h @@ -128,6 +128,7 @@ class DeadArgumentEliminationPass Liveness SurveyUses(const Value *V, UseVector &MaybeLiveUses); void SurveyFunction(const Function &F); + bool IsLive(const RetOrArg &RA); void MarkValue(const RetOrArg &RA, Liveness L, const UseVector &MaybeLiveUses); void MarkLive(const RetOrArg &RA); diff --git a/llvm/lib/DebugInfo/DWARF/DWARFDebugArangeSet.cpp b/llvm/lib/DebugInfo/DWARF/DWARFDebugArangeSet.cpp index 608fc0388af0..c3b039b05f30 100644 --- a/llvm/lib/DebugInfo/DWARF/DWARFDebugArangeSet.cpp +++ b/llvm/lib/DebugInfo/DWARF/DWARFDebugArangeSet.cpp @@ -132,19 +132,20 @@ Error DWARFDebugArangeSet::extract(DWARFDataExtractor data, uint64_t end_offset = Offset + full_length; while (*offset_ptr < end_offset) { + uint64_t EntryOffset = *offset_ptr; arangeDescriptor.Address = data.getUnsigned(offset_ptr, HeaderData.AddrSize); arangeDescriptor.Length = data.getUnsigned(offset_ptr, HeaderData.AddrSize); - if (arangeDescriptor.Length == 0) { - // Each set of tuples is terminated by a 0 for the address and 0 - // for the length. - if (arangeDescriptor.Address == 0 && *offset_ptr == end_offset) + // Each set of tuples is terminated by a 0 for the address and 0 + // for the length. + if (arangeDescriptor.Length == 0 && arangeDescriptor.Address == 0) { + if (*offset_ptr == end_offset) return ErrorSuccess(); return createStringError( errc::invalid_argument, "address range table at offset 0x%" PRIx64 - " has an invalid tuple (length = 0) at offset 0x%" PRIx64, - Offset, *offset_ptr - tuple_size); + " has a premature terminator entry at offset 0x%" PRIx64, + Offset, EntryOffset); } ArangeDescriptors.push_back(arangeDescriptor); diff --git a/llvm/lib/DebugInfo/DWARF/DWARFUnitIndex.cpp b/llvm/lib/DebugInfo/DWARF/DWARFUnitIndex.cpp index 3d4cecce27db..d27fd08db14e 100644 --- a/llvm/lib/DebugInfo/DWARF/DWARFUnitIndex.cpp +++ b/llvm/lib/DebugInfo/DWARF/DWARFUnitIndex.cpp @@ -286,10 +286,14 @@ const DWARFUnitIndex::Entry *DWARFUnitIndex::getFromHash(uint64_t S) const { auto H = S & Mask; auto HP = ((S >> 32) & Mask) | 1; - while (Rows[H].getSignature() != S && Rows[H].getSignature() != 0) + // The spec says "while 0 is a valid hash value, the row index in a used slot + // will always be non-zero". Loop until we find a match or an empty slot. + while (Rows[H].getSignature() != S && Rows[H].Index != nullptr) H = (H + HP) & Mask; - if (Rows[H].getSignature() != S) + // If the slot is empty, we don't care whether the signature matches (it could + // be zero and still match the zeros in the empty slot). + if (Rows[H].Index == nullptr) return nullptr; return &Rows[H]; diff --git a/llvm/lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp b/llvm/lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp index 7e9b0690ccea..04f541b59557 100644 --- a/llvm/lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp +++ b/llvm/lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp @@ -308,7 +308,9 @@ RuntimeDyldImpl::loadObjectImpl(const object::ObjectFile &Obj) { << " SID: " << SectionID << " Offset: " << format("%p", (uintptr_t)Addr) << " flags: " << *FlagsOrErr << "\n"); - GlobalSymbolTable[Name] = SymbolTableEntry(SectionID, Addr, *JITSymFlags); + if (!Name.empty()) // Skip absolute symbol relocations. + GlobalSymbolTable[Name] = + SymbolTableEntry(SectionID, Addr, *JITSymFlags); } else if (SymType == object::SymbolRef::ST_Function || SymType == object::SymbolRef::ST_Data || SymType == object::SymbolRef::ST_Unknown || @@ -340,8 +342,9 @@ RuntimeDyldImpl::loadObjectImpl(const object::ObjectFile &Obj) { << " SID: " << SectionID << " Offset: " << format("%p", (uintptr_t)SectOffset) << " flags: " << *FlagsOrErr << "\n"); - GlobalSymbolTable[Name] = - SymbolTableEntry(SectionID, SectOffset, *JITSymFlags); + if (!Name.empty()) // Skip absolute symbol relocations + GlobalSymbolTable[Name] = + SymbolTableEntry(SectionID, SectOffset, *JITSymFlags); } } @@ -769,8 +772,9 @@ Error RuntimeDyldImpl::emitCommonSymbols(const ObjectFile &Obj, LLVM_DEBUG(dbgs() << "Allocating common symbol " << Name << " address " << format("%p", Addr) << "\n"); - GlobalSymbolTable[Name] = - SymbolTableEntry(SectionID, Offset, std::move(*JITSymFlags)); + if (!Name.empty()) // Skip absolute symbol relocations. + GlobalSymbolTable[Name] = + SymbolTableEntry(SectionID, Offset, std::move(*JITSymFlags)); Offset += Size; Addr += Size; } @@ -930,6 +934,8 @@ void RuntimeDyldImpl::addRelocationForSymbol(const RelocationEntry &RE, if (Loc == GlobalSymbolTable.end()) { ExternalSymbolRelocations[SymbolName].push_back(RE); } else { + assert(!SymbolName.empty() && + "Empty symbol should not be in GlobalSymbolTable"); // Copy the RE since we want to modify its addend. RelocationEntry RECopy = RE; const auto &SymInfo = Loc->second; @@ -1234,7 +1240,8 @@ void RuntimeDyldImpl::finalizeAsync( for (auto &RelocKV : SharedThis->ExternalSymbolRelocations) { StringRef Name = RelocKV.first(); - assert(!Name.empty() && "Symbol has no name?"); + if (Name.empty()) // Skip absolute symbol relocations. + continue; assert(!SharedThis->GlobalSymbolTable.count(Name) && "Name already processed. RuntimeDyld instances can not be re-used " "when finalizing with finalizeAsync."); diff --git a/llvm/lib/IR/ConstantFold.cpp b/llvm/lib/IR/ConstantFold.cpp index f3c3e9ad9f69..c20d0955f3d8 100644 --- a/llvm/lib/IR/ConstantFold.cpp +++ b/llvm/lib/IR/ConstantFold.cpp @@ -1589,7 +1589,7 @@ static FCmpInst::Predicate evaluateFCmpRelation(Constant *V1, Constant *V2) { static ICmpInst::Predicate areGlobalsPotentiallyEqual(const GlobalValue *GV1, const GlobalValue *GV2) { auto isGlobalUnsafeForEquality = [](const GlobalValue *GV) { - if (GV->hasExternalWeakLinkage() || GV->hasWeakAnyLinkage()) + if (GV->isInterposable() || GV->hasGlobalUnnamedAddr()) return true; if (const auto *GVar = dyn_cast(GV)) { Type *Ty = GVar->getValueType(); diff --git a/llvm/lib/MC/MCObjectFileInfo.cpp b/llvm/lib/MC/MCObjectFileInfo.cpp index b77a9635f64c..b9b4416fde21 100644 --- a/llvm/lib/MC/MCObjectFileInfo.cpp +++ b/llvm/lib/MC/MCObjectFileInfo.cpp @@ -317,6 +317,8 @@ void MCObjectFileInfo::initELFMCObjectFileInfo(const Triple &T, bool Large) { break; case Triple::ppc64: case Triple::ppc64le: + case Triple::aarch64: + case Triple::aarch64_be: case Triple::x86_64: FDECFIEncoding = dwarf::DW_EH_PE_pcrel | (Large ? dwarf::DW_EH_PE_sdata8 : dwarf::DW_EH_PE_sdata4); diff --git a/llvm/lib/MC/MCParser/ELFAsmParser.cpp b/llvm/lib/MC/MCParser/ELFAsmParser.cpp index e5ab13bc719d..fb8215ef2281 100644 --- a/llvm/lib/MC/MCParser/ELFAsmParser.cpp +++ b/llvm/lib/MC/MCParser/ELFAsmParser.cpp @@ -644,10 +644,13 @@ bool ELFAsmParser::ParseSectionArguments(bool IsPush, SMLoc loc) { !(SectionName == ".eh_frame" && Type == ELF::SHT_PROGBITS)) Error(loc, "changed section type for " + SectionName + ", expected: 0x" + utohexstr(Section->getType())); - if (Section->getFlags() != Flags) + // Check that flags are used consistently. However, the GNU assembler permits + // to leave out in subsequent uses of the same sections; for compatibility, + // do likewise. + if ((Flags || Size || !TypeName.empty()) && Section->getFlags() != Flags) Error(loc, "changed section flags for " + SectionName + ", expected: 0x" + utohexstr(Section->getFlags())); - if (Section->getEntrySize() != Size) + if ((Flags || Size || !TypeName.empty()) && Section->getEntrySize() != Size) Error(loc, "changed section entsize for " + SectionName + ", expected: " + Twine(Section->getEntrySize())); diff --git a/llvm/lib/MCA/HardwareUnits/LSUnit.cpp b/llvm/lib/MCA/HardwareUnits/LSUnit.cpp index e945e8cecce9..4594368fc0e9 100644 --- a/llvm/lib/MCA/HardwareUnits/LSUnit.cpp +++ b/llvm/lib/MCA/HardwareUnits/LSUnit.cpp @@ -243,6 +243,8 @@ void LSUnit::onInstructionExecuted(const InstRef &IR) { CurrentStoreGroupID = 0; if (GroupID == CurrentLoadBarrierGroupID) CurrentLoadBarrierGroupID = 0; + if (GroupID == CurrentStoreBarrierGroupID) + CurrentStoreBarrierGroupID = 0; } } diff --git a/llvm/lib/Support/Host.cpp b/llvm/lib/Support/Host.cpp index 658c1ee74cfe..36cecf9b2a16 100644 --- a/llvm/lib/Support/Host.cpp +++ b/llvm/lib/Support/Host.cpp @@ -760,14 +760,15 @@ getIntelProcessorTypeAndSubtype(unsigned Family, unsigned Model, *Type = X86::INTEL_GOLDMONT_PLUS; break; case 0x86: + CPU = "tremont"; *Type = X86::INTEL_TREMONT; break; + // Xeon Phi (Knights Landing + Knights Mill): case 0x57: - CPU = "tremont"; + CPU = "knl"; *Type = X86::INTEL_KNL; break; - case 0x85: CPU = "knm"; *Type = X86::INTEL_KNM; diff --git a/llvm/lib/Support/Windows/Path.inc b/llvm/lib/Support/Windows/Path.inc index e352beb77616..a4ffc0ec4313 100644 --- a/llvm/lib/Support/Windows/Path.inc +++ b/llvm/lib/Support/Windows/Path.inc @@ -19,7 +19,6 @@ #include "llvm/Support/ConvertUTF.h" #include "llvm/Support/WindowsError.h" #include -#include #include #include @@ -352,13 +351,13 @@ std::error_code is_local(const Twine &path, bool &result) { static std::error_code realPathFromHandle(HANDLE H, SmallVectorImpl &Buffer) { DWORD CountChars = ::GetFinalPathNameByHandleW( - H, Buffer.begin(), Buffer.capacity() - 1, FILE_NAME_NORMALIZED); - if (CountChars > Buffer.capacity()) { + H, Buffer.begin(), Buffer.capacity(), FILE_NAME_NORMALIZED); + if (CountChars && CountChars >= Buffer.capacity()) { // The buffer wasn't big enough, try again. In this case the return value // *does* indicate the size of the null terminator. Buffer.reserve(CountChars); CountChars = ::GetFinalPathNameByHandleW( - H, Buffer.data(), Buffer.capacity() - 1, FILE_NAME_NORMALIZED); + H, Buffer.begin(), Buffer.capacity(), FILE_NAME_NORMALIZED); } if (CountChars == 0) return mapWindowsError(GetLastError()); @@ -403,6 +402,20 @@ std::error_code is_local(int FD, bool &Result) { } static std::error_code setDeleteDisposition(HANDLE Handle, bool Delete) { + // First, check if the file is on a network (non-local) drive. If so, don't + // set DeleteFile to true, since it prevents opening the file for writes. + SmallVector FinalPath; + if (std::error_code EC = realPathFromHandle(Handle, FinalPath)) + return EC; + + bool IsLocal; + if (std::error_code EC = is_local_internal(FinalPath, IsLocal)) + return EC; + + if (!IsLocal) + return std::error_code(); + + // The file is on a local drive, set the DeleteFile to true. FILE_DISPOSITION_INFO Disposition; Disposition.DeleteFile = Delete; if (!SetFileInformationByHandle(Handle, FileDispositionInfo, &Disposition, diff --git a/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp b/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp index 3a94820dac8d..7ec7ffe309f7 100644 --- a/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp +++ b/llvm/lib/Target/AArch64/AArch64AsmPrinter.cpp @@ -89,6 +89,8 @@ class AArch64AsmPrinter : public AsmPrinter { void emitJumpTableEntry(const MachineJumpTableInfo *MJTI, const MachineBasicBlock *MBB, unsigned JTI); + void emitFunctionEntryLabel() override; + void LowerJumpTableDestSmall(MCStreamer &OutStreamer, const MachineInstr &MI); void LowerSTACKMAP(MCStreamer &OutStreamer, StackMaps &SM, @@ -822,6 +824,19 @@ void AArch64AsmPrinter::emitJumpTableEntry(const MachineJumpTableInfo *MJTI, OutStreamer->emitValue(Value, Size); } +void AArch64AsmPrinter::emitFunctionEntryLabel() { + if (MF->getFunction().getCallingConv() == CallingConv::AArch64_VectorCall || + MF->getFunction().getCallingConv() == + CallingConv::AArch64_SVE_VectorCall || + STI->getRegisterInfo()->hasSVEArgsOrReturn(MF)) { + auto *TS = + static_cast(OutStreamer->getTargetStreamer()); + TS->emitDirectiveVariantPCS(CurrentFnSym); + } + + return AsmPrinter::emitFunctionEntryLabel(); +} + /// Small jump tables contain an unsigned byte or half, representing the offset /// from the lowest-addressed possible destination to the desired basic /// block. Since all instructions are 4-byte aligned, this is further compressed diff --git a/llvm/lib/Target/AArch64/AArch64CallingConvention.cpp b/llvm/lib/Target/AArch64/AArch64CallingConvention.cpp index 84ec5afcc9c1..9ae2b465e247 100644 --- a/llvm/lib/Target/AArch64/AArch64CallingConvention.cpp +++ b/llvm/lib/Target/AArch64/AArch64CallingConvention.cpp @@ -35,6 +35,9 @@ static const MCPhysReg DRegList[] = {AArch64::D0, AArch64::D1, AArch64::D2, static const MCPhysReg QRegList[] = {AArch64::Q0, AArch64::Q1, AArch64::Q2, AArch64::Q3, AArch64::Q4, AArch64::Q5, AArch64::Q6, AArch64::Q7}; +static const MCPhysReg ZRegList[] = {AArch64::Z0, AArch64::Z1, AArch64::Z2, + AArch64::Z3, AArch64::Z4, AArch64::Z5, + AArch64::Z6, AArch64::Z7}; static bool finishStackBlock(SmallVectorImpl &PendingMembers, MVT LocVT, ISD::ArgFlagsTy &ArgFlags, @@ -97,6 +100,8 @@ static bool CC_AArch64_Custom_Block(unsigned &ValNo, MVT &ValVT, MVT &LocVT, RegList = DRegList; else if (LocVT.SimpleTy == MVT::f128 || LocVT.is128BitVector()) RegList = QRegList; + else if (LocVT.isScalableVector()) + RegList = ZRegList; else { // Not an array we want to split up after all. return false; @@ -141,6 +146,10 @@ static bool CC_AArch64_Custom_Block(unsigned &ValNo, MVT &ValVT, MVT &LocVT, return true; } + if (LocVT.isScalableVector()) + report_fatal_error( + "Passing consecutive scalable vector registers unsupported"); + // Mark all regs in the class as unavailable for (auto Reg : RegList) State.AllocateReg(Reg); diff --git a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp index 45bfa85bdc07..48ca9039b1bd 100644 --- a/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp +++ b/llvm/lib/Target/AArch64/AArch64ISelLowering.cpp @@ -4366,6 +4366,10 @@ AArch64TargetLowering::LowerCall(CallLoweringInfo &CLI, for (unsigned i = 0; i != NumArgs; ++i) { MVT ArgVT = Outs[i].VT; + if (!Outs[i].IsFixed && ArgVT.isScalableVector()) + report_fatal_error("Passing SVE types to variadic functions is " + "currently not supported"); + ISD::ArgFlagsTy ArgFlags = Outs[i].Flags; CCAssignFn *AssignFn = CCAssignFnForCall(CallConv, /*IsVarArg=*/ !Outs[i].IsFixed); @@ -6168,6 +6172,10 @@ SDValue AArch64TargetLowering::LowerVAARG(SDValue Op, SelectionDAG &DAG) const { Chain = VAList.getValue(1); VAList = DAG.getZExtOrTrunc(VAList, DL, PtrVT); + if (VT.isScalableVector()) + report_fatal_error("Passing SVE types to variadic functions is " + "currently not supported"); + if (Align && *Align > MinSlotSize) { VAList = DAG.getNode(ISD::ADD, DL, PtrVT, VAList, DAG.getConstant(Align->value() - 1, DL, PtrVT)); @@ -14702,7 +14710,14 @@ Value *AArch64TargetLowering::emitStoreConditional(IRBuilder<> &Builder, bool AArch64TargetLowering::functionArgumentNeedsConsecutiveRegisters( Type *Ty, CallingConv::ID CallConv, bool isVarArg) const { - return Ty->isArrayTy(); + if (Ty->isArrayTy()) + return true; + + const TypeSize &TySize = Ty->getPrimitiveSizeInBits(); + if (TySize.isScalable() && TySize.getKnownMinSize() > 128) + return true; + + return false; } bool AArch64TargetLowering::shouldNormalizeToSelectSequence(LLVMContext &, diff --git a/llvm/lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp b/llvm/lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp index 0ac09c4f96f0..e72ae0e62cb7 100644 --- a/llvm/lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp +++ b/llvm/lib/Target/AArch64/AsmParser/AArch64AsmParser.cpp @@ -179,6 +179,8 @@ class AArch64AsmParser : public MCTargetAsmParser { bool parseDirectiveCFINegateRAState(); bool parseDirectiveCFIBKeyFrame(); + bool parseDirectiveVariantPCS(SMLoc L); + bool validateInstruction(MCInst &Inst, SMLoc &IDLoc, SmallVectorImpl &Loc); bool MatchAndEmitInstruction(SMLoc IDLoc, unsigned &Opcode, @@ -5077,6 +5079,8 @@ bool AArch64AsmParser::ParseDirective(AsmToken DirectiveID) { parseDirectiveCFIBKeyFrame(); else if (IDVal == ".arch_extension") parseDirectiveArchExtension(Loc); + else if (IDVal == ".variant_pcs") + parseDirectiveVariantPCS(Loc); else if (IsMachO) { if (IDVal == MCLOHDirectiveName()) parseDirectiveLOH(IDVal, Loc); @@ -5507,6 +5511,32 @@ bool AArch64AsmParser::parseDirectiveCFIBKeyFrame() { return false; } +/// parseDirectiveVariantPCS +/// ::= .variant_pcs symbolname +bool AArch64AsmParser::parseDirectiveVariantPCS(SMLoc L) { + MCAsmParser &Parser = getParser(); + + const AsmToken &Tok = Parser.getTok(); + if (Tok.isNot(AsmToken::Identifier)) + return TokError("expected symbol name"); + + StringRef SymbolName = Tok.getIdentifier(); + + MCSymbol *Sym = getContext().lookupSymbol(SymbolName); + if (!Sym) + return TokError("unknown symbol in '.variant_pcs' directive"); + + Parser.Lex(); // Eat the symbol + + // Shouldn't be any more tokens + if (parseToken(AsmToken::EndOfStatement)) + return addErrorSuffix(" in '.variant_pcs' directive"); + + getTargetStreamer().emitDirectiveVariantPCS(Sym); + + return false; +} + bool AArch64AsmParser::classifySymbolRef(const MCExpr *Expr, AArch64MCExpr::VariantKind &ELFRefKind, diff --git a/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp b/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp index 408f0cb77e73..7733fe7f7b24 100644 --- a/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp +++ b/llvm/lib/Target/AArch64/GISel/AArch64InstructionSelector.cpp @@ -289,14 +289,15 @@ class AArch64InstructionSelector : public InstructionSelector { getExtendTypeForInst(MachineInstr &MI, MachineRegisterInfo &MRI, bool IsLoadStore = false) const; - /// Instructions that accept extend modifiers like UXTW expect the register - /// being extended to be a GPR32. Narrow ExtReg to a 32-bit register using a - /// subregister copy if necessary. Return either ExtReg, or the result of the - /// new copy. - Register narrowExtendRegIfNeeded(Register ExtReg, - MachineIRBuilder &MIB) const; - Register widenGPRBankRegIfNeeded(Register Reg, unsigned Size, - MachineIRBuilder &MIB) const; + /// Move \p Reg to \p RC if \p Reg is not already on \p RC. + /// + /// \returns Either \p Reg if no change was necessary, or the new register + /// created by moving \p Reg. + /// + /// Note: This uses emitCopy right now. + Register moveScalarRegClass(Register Reg, const TargetRegisterClass &RC, + MachineIRBuilder &MIB) const; + ComplexRendererFns selectArithExtendedRegister(MachineOperand &Root) const; void renderTruncImm(MachineInstrBuilder &MIB, const MachineInstr &MI, @@ -1195,10 +1196,10 @@ MachineInstr *AArch64InstructionSelector::emitTestBit( // TBNZW work. bool UseWReg = Bit < 32; unsigned NecessarySize = UseWReg ? 32 : 64; - if (Size < NecessarySize) - TestReg = widenGPRBankRegIfNeeded(TestReg, NecessarySize, MIB); - else if (Size > NecessarySize) - TestReg = narrowExtendRegIfNeeded(TestReg, MIB); + if (Size != NecessarySize) + TestReg = moveScalarRegClass( + TestReg, UseWReg ? AArch64::GPR32RegClass : AArch64::GPR64RegClass, + MIB); static const unsigned OpcTable[2][2] = {{AArch64::TBZX, AArch64::TBNZX}, {AArch64::TBZW, AArch64::TBNZW}}; @@ -4904,9 +4905,19 @@ AArch64InstructionSelector::selectExtendedSHL( return None; unsigned OffsetOpc = OffsetInst->getOpcode(); - if (OffsetOpc != TargetOpcode::G_SHL && OffsetOpc != TargetOpcode::G_MUL) - return None; + bool LookedThroughZExt = false; + if (OffsetOpc != TargetOpcode::G_SHL && OffsetOpc != TargetOpcode::G_MUL) { + // Try to look through a ZEXT. + if (OffsetOpc != TargetOpcode::G_ZEXT || !WantsExt) + return None; + OffsetInst = MRI.getVRegDef(OffsetInst->getOperand(1).getReg()); + OffsetOpc = OffsetInst->getOpcode(); + LookedThroughZExt = true; + + if (OffsetOpc != TargetOpcode::G_SHL && OffsetOpc != TargetOpcode::G_MUL) + return None; + } // Make sure that the memory op is a valid size. int64_t LegalShiftVal = Log2_32(SizeInBytes); if (LegalShiftVal == 0) @@ -4957,21 +4968,24 @@ AArch64InstructionSelector::selectExtendedSHL( unsigned SignExtend = 0; if (WantsExt) { - // Check if the offset is defined by an extend. - MachineInstr *ExtInst = getDefIgnoringCopies(OffsetReg, MRI); - auto Ext = getExtendTypeForInst(*ExtInst, MRI, true); - if (Ext == AArch64_AM::InvalidShiftExtend) - return None; + // Check if the offset is defined by an extend, unless we looked through a + // G_ZEXT earlier. + if (!LookedThroughZExt) { + MachineInstr *ExtInst = getDefIgnoringCopies(OffsetReg, MRI); + auto Ext = getExtendTypeForInst(*ExtInst, MRI, true); + if (Ext == AArch64_AM::InvalidShiftExtend) + return None; - SignExtend = isSignExtendShiftType(Ext) ? 1 : 0; - // We only support SXTW for signed extension here. - if (SignExtend && Ext != AArch64_AM::SXTW) - return None; + SignExtend = isSignExtendShiftType(Ext) ? 1 : 0; + // We only support SXTW for signed extension here. + if (SignExtend && Ext != AArch64_AM::SXTW) + return None; + OffsetReg = ExtInst->getOperand(1).getReg(); + } // Need a 32-bit wide register here. MachineIRBuilder MIB(*MRI.getVRegDef(Root.getReg())); - OffsetReg = ExtInst->getOperand(1).getReg(); - OffsetReg = narrowExtendRegIfNeeded(OffsetReg, MIB); + OffsetReg = moveScalarRegClass(OffsetReg, AArch64::GPR32RegClass, MIB); } // We can use the LHS of the GEP as the base, and the LHS of the shift as an @@ -5143,8 +5157,8 @@ AArch64InstructionSelector::selectAddrModeWRO(MachineOperand &Root, // Need a 32-bit wide register. MachineIRBuilder MIB(*PtrAdd); - Register ExtReg = - narrowExtendRegIfNeeded(OffsetInst->getOperand(1).getReg(), MIB); + Register ExtReg = moveScalarRegClass(OffsetInst->getOperand(1).getReg(), + AArch64::GPR32RegClass, MIB); unsigned SignExtend = Ext == AArch64_AM::SXTW; // Base is LHS, offset is ExtReg. @@ -5418,67 +5432,21 @@ AArch64_AM::ShiftExtendType AArch64InstructionSelector::getExtendTypeForInst( } } -Register AArch64InstructionSelector::narrowExtendRegIfNeeded( - Register ExtReg, MachineIRBuilder &MIB) const { +Register AArch64InstructionSelector::moveScalarRegClass( + Register Reg, const TargetRegisterClass &RC, MachineIRBuilder &MIB) const { MachineRegisterInfo &MRI = *MIB.getMRI(); - if (MRI.getType(ExtReg).getSizeInBits() == 32) - return ExtReg; + auto Ty = MRI.getType(Reg); + assert(!Ty.isVector() && "Expected scalars only!"); + if (Ty.getSizeInBits() == TRI.getRegSizeInBits(RC)) + return Reg; - // Insert a copy to move ExtReg to GPR32. - Register NarrowReg = MRI.createVirtualRegister(&AArch64::GPR32RegClass); - auto Copy = MIB.buildCopy({NarrowReg}, {ExtReg}); - - // Select the copy into a subregister copy. + // Create a copy and immediately select it. + // FIXME: We should have an emitCopy function? + auto Copy = MIB.buildCopy({&RC}, {Reg}); selectCopy(*Copy, TII, MRI, TRI, RBI); return Copy.getReg(0); } -Register AArch64InstructionSelector::widenGPRBankRegIfNeeded( - Register Reg, unsigned WideSize, MachineIRBuilder &MIB) const { - assert(WideSize >= 8 && "WideSize is smaller than all possible registers?"); - MachineRegisterInfo &MRI = *MIB.getMRI(); - unsigned NarrowSize = MRI.getType(Reg).getSizeInBits(); - assert(WideSize >= NarrowSize && - "WideSize cannot be smaller than NarrowSize!"); - - // If the sizes match, just return the register. - // - // If NarrowSize is an s1, then we can select it to any size, so we'll treat - // it as a don't care. - if (NarrowSize == WideSize || NarrowSize == 1) - return Reg; - - // Now check the register classes. - const RegisterBank *RB = RBI.getRegBank(Reg, MRI, TRI); - const TargetRegisterClass *OrigRC = getMinClassForRegBank(*RB, NarrowSize); - const TargetRegisterClass *WideRC = getMinClassForRegBank(*RB, WideSize); - assert(OrigRC && "Could not determine narrow RC?"); - assert(WideRC && "Could not determine wide RC?"); - - // If the sizes differ, but the register classes are the same, there is no - // need to insert a SUBREG_TO_REG. - // - // For example, an s8 that's supposed to be a GPR will be selected to either - // a GPR32 or a GPR64 register. Note that this assumes that the s8 will - // always end up on a GPR32. - if (OrigRC == WideRC) - return Reg; - - // We have two different register classes. Insert a SUBREG_TO_REG. - unsigned SubReg = 0; - getSubRegForClass(OrigRC, TRI, SubReg); - assert(SubReg && "Couldn't determine subregister?"); - - // Build the SUBREG_TO_REG and return the new, widened register. - auto SubRegToReg = - MIB.buildInstr(AArch64::SUBREG_TO_REG, {WideRC}, {}) - .addImm(0) - .addUse(Reg) - .addImm(SubReg); - constrainSelectedInstRegOperands(*SubRegToReg, TII, TRI, RBI); - return SubRegToReg.getReg(0); -} - /// Select an "extended register" operand. This operand folds in an extend /// followed by an optional left shift. InstructionSelector::ComplexRendererFns @@ -5539,7 +5507,7 @@ AArch64InstructionSelector::selectArithExtendedRegister( // We require a GPR32 here. Narrow the ExtReg if needed using a subregister // copy. MachineIRBuilder MIB(*RootDef); - ExtReg = narrowExtendRegIfNeeded(ExtReg, MIB); + ExtReg = moveScalarRegClass(ExtReg, AArch64::GPR32RegClass, MIB); return {{[=](MachineInstrBuilder &MIB) { MIB.addUse(ExtReg); }, [=](MachineInstrBuilder &MIB) { diff --git a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp index 2eaec0b970fa..4ffde2a7e3c4 100644 --- a/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp +++ b/llvm/lib/Target/AArch64/GISel/AArch64LegalizerInfo.cpp @@ -97,15 +97,25 @@ AArch64LegalizerInfo::AArch64LegalizerInfo(const AArch64Subtarget &ST) .moreElementsToNextPow2(0); getActionDefinitionsBuilder(G_SHL) - .legalFor({{s32, s32}, {s64, s64}, - {v2s32, v2s32}, {v4s32, v4s32}, {v2s64, v2s64}}) - .clampScalar(1, s32, s64) - .clampScalar(0, s32, s64) - .widenScalarToNextPow2(0) - .clampNumElements(0, v2s32, v4s32) - .clampNumElements(0, v2s64, v2s64) - .moreElementsToNextPow2(0) - .minScalarSameAs(1, 0); + .customIf([=](const LegalityQuery &Query) { + const auto &SrcTy = Query.Types[0]; + const auto &AmtTy = Query.Types[1]; + return !SrcTy.isVector() && SrcTy.getSizeInBits() == 32 && + AmtTy.getSizeInBits() == 32; + }) + .legalFor({{s32, s32}, + {s64, s64}, + {s32, s64}, + {v2s32, v2s32}, + {v4s32, v4s32}, + {v2s64, v2s64}}) + .clampScalar(1, s32, s64) + .clampScalar(0, s32, s64) + .widenScalarToNextPow2(0) + .clampNumElements(0, v2s32, v4s32) + .clampNumElements(0, v2s64, v2s64) + .moreElementsToNextPow2(0) + .minScalarSameAs(1, 0); getActionDefinitionsBuilder(G_PTR_ADD) .legalFor({{p0, s64}, {v2p0, v2s64}}) @@ -710,16 +720,14 @@ bool AArch64LegalizerInfo::legalizeShlAshrLshr( // If the shift amount is a G_CONSTANT, promote it to a 64 bit type so the // imported patterns can select it later. Either way, it will be legal. Register AmtReg = MI.getOperand(2).getReg(); - auto *CstMI = MRI.getVRegDef(AmtReg); - assert(CstMI && "expected to find a vreg def"); - if (CstMI->getOpcode() != TargetOpcode::G_CONSTANT) + auto VRegAndVal = getConstantVRegValWithLookThrough(AmtReg, MRI); + if (!VRegAndVal) return true; // Check the shift amount is in range for an immediate form. - unsigned Amount = CstMI->getOperand(1).getCImm()->getZExtValue(); + int64_t Amount = VRegAndVal->Value; if (Amount > 31) return true; // This will have to remain a register variant. - assert(MRI.getType(AmtReg).getSizeInBits() == 32); - auto ExtCst = MIRBuilder.buildZExt(LLT::scalar(64), AmtReg); + auto ExtCst = MIRBuilder.buildConstant(LLT::scalar(64), Amount); MI.getOperand(2).setReg(ExtCst.getReg(0)); return true; } diff --git a/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp b/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp index 7e3ff1948dad..93213f5977e5 100644 --- a/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp +++ b/llvm/lib/Target/AArch64/GISel/AArch64RegisterBankInfo.cpp @@ -261,6 +261,7 @@ AArch64RegisterBankInfo::getRegBankFromRegClass(const TargetRegisterClass &RC, case AArch64::GPR64common_and_GPR64noipRegClassID: case AArch64::GPR64noip_and_tcGPR64RegClassID: case AArch64::tcGPR64RegClassID: + case AArch64::rtcGPR64RegClassID: case AArch64::WSeqPairsClassRegClassID: case AArch64::XSeqPairsClassRegClassID: return getRegBank(AArch64::GPRRegBankID); diff --git a/llvm/lib/Target/AArch64/MCTargetDesc/AArch64ELFStreamer.cpp b/llvm/lib/Target/AArch64/MCTargetDesc/AArch64ELFStreamer.cpp index fe4c34be1519..6dfda8217628 100644 --- a/llvm/lib/Target/AArch64/MCTargetDesc/AArch64ELFStreamer.cpp +++ b/llvm/lib/Target/AArch64/MCTargetDesc/AArch64ELFStreamer.cpp @@ -47,6 +47,10 @@ class AArch64TargetAsmStreamer : public AArch64TargetStreamer { void emitInst(uint32_t Inst) override; + void emitDirectiveVariantPCS(MCSymbol *Symbol) override { + OS << "\t.variant_pcs " << Symbol->getName() << "\n"; + } + public: AArch64TargetAsmStreamer(MCStreamer &S, formatted_raw_ostream &OS); }; @@ -194,6 +198,10 @@ void AArch64TargetELFStreamer::emitInst(uint32_t Inst) { getStreamer().emitInst(Inst); } +void AArch64TargetELFStreamer::emitDirectiveVariantPCS(MCSymbol *Symbol) { + cast(Symbol)->setOther(ELF::STO_AARCH64_VARIANT_PCS); +} + MCTargetStreamer *createAArch64AsmTargetStreamer(MCStreamer &S, formatted_raw_ostream &OS, MCInstPrinter *InstPrint, diff --git a/llvm/lib/Target/AArch64/MCTargetDesc/AArch64TargetStreamer.h b/llvm/lib/Target/AArch64/MCTargetDesc/AArch64TargetStreamer.h index 3a0c5d8318dd..1af978a806d1 100644 --- a/llvm/lib/Target/AArch64/MCTargetDesc/AArch64TargetStreamer.h +++ b/llvm/lib/Target/AArch64/MCTargetDesc/AArch64TargetStreamer.h @@ -36,6 +36,9 @@ class AArch64TargetStreamer : public MCTargetStreamer { /// Callback used to implement the .inst directive. virtual void emitInst(uint32_t Inst); + /// Callback used to implement the .variant_pcs directive. + virtual void emitDirectiveVariantPCS(MCSymbol *Symbol) {}; + virtual void EmitARM64WinCFIAllocStack(unsigned Size) {} virtual void EmitARM64WinCFISaveFPLR(int Offset) {} virtual void EmitARM64WinCFISaveFPLRX(int Offset) {} @@ -63,6 +66,7 @@ class AArch64TargetELFStreamer : public AArch64TargetStreamer { AArch64ELFStreamer &getStreamer(); void emitInst(uint32_t Inst) override; + void emitDirectiveVariantPCS(MCSymbol *Symbol) override; public: AArch64TargetELFStreamer(MCStreamer &S) : AArch64TargetStreamer(S) {} diff --git a/llvm/lib/Target/AMDGPU/SIPreEmitPeephole.cpp b/llvm/lib/Target/AMDGPU/SIPreEmitPeephole.cpp index f31c722db1b2..442be886a8ac 100644 --- a/llvm/lib/Target/AMDGPU/SIPreEmitPeephole.cpp +++ b/llvm/lib/Target/AMDGPU/SIPreEmitPeephole.cpp @@ -254,16 +254,24 @@ bool SIPreEmitPeephole::runOnMachineFunction(MachineFunction &MF) { for (MachineBasicBlock &MBB : MF) { MachineBasicBlock::iterator MBBE = MBB.getFirstTerminator(); - if (MBBE != MBB.end()) { - MachineInstr &MI = *MBBE; + MachineBasicBlock::iterator TermI = MBBE; + // Check first terminator for VCC branches to optimize + if (TermI != MBB.end()) { + MachineInstr &MI = *TermI; switch (MI.getOpcode()) { case AMDGPU::S_CBRANCH_VCCZ: case AMDGPU::S_CBRANCH_VCCNZ: Changed |= optimizeVccBranch(MI); continue; - case AMDGPU::SI_RETURN_TO_EPILOG: - // FIXME: This is not an optimization and should be - // moved somewhere else. + default: + break; + } + } + // Check all terminators for SI_RETURN_TO_EPILOG + // FIXME: This is not an optimization and should be moved somewhere else. + while (TermI != MBB.end()) { + MachineInstr &MI = *TermI; + if (MI.getOpcode() == AMDGPU::SI_RETURN_TO_EPILOG) { assert(!MF.getInfo()->returnsVoid()); // Graphics shaders returning non-void shouldn't contain S_ENDPGM, @@ -281,11 +289,11 @@ bool SIPreEmitPeephole::runOnMachineFunction(MachineFunction &MF) { .addMBB(EmptyMBBAtEnd); MI.eraseFromParent(); MBBE = MBB.getFirstTerminator(); + TermI = MBBE; + continue; } - break; - default: - break; } + TermI++; } if (!ST.hasVGPRIndexMode()) diff --git a/llvm/lib/Target/BPF/BPFISelDAGToDAG.cpp b/llvm/lib/Target/BPF/BPFISelDAGToDAG.cpp index d407edfbd966..77f565fb5957 100644 --- a/llvm/lib/Target/BPF/BPFISelDAGToDAG.cpp +++ b/llvm/lib/Target/BPF/BPFISelDAGToDAG.cpp @@ -254,7 +254,7 @@ void BPFDAGToDAGISel::PreprocessLoad(SDNode *Node, const LoadSDNode *LD = cast(Node); uint64_t size = LD->getMemOperand()->getSize(); - if (!size || size > 8 || (size & (size - 1))) + if (!size || size > 8 || (size & (size - 1)) || !LD->isSimple()) return; SDNode *LDAddrNode = LD->getOperand(1).getNode(); @@ -342,7 +342,7 @@ bool BPFDAGToDAGISel::getConstantFieldValue(const GlobalAddressSDNode *Node, unsigned char *ByteSeq) { const GlobalVariable *V = dyn_cast(Node->getGlobal()); - if (!V || !V->hasInitializer()) + if (!V || !V->hasInitializer() || !V->isConstant()) return false; const Constant *Init = V->getInitializer(); diff --git a/llvm/lib/Target/Hexagon/HexagonAsmPrinter.h b/llvm/lib/Target/Hexagon/HexagonAsmPrinter.h old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/Sparc/LeonFeatures.td b/llvm/lib/Target/Sparc/LeonFeatures.td old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/Sparc/LeonPasses.cpp b/llvm/lib/Target/Sparc/LeonPasses.cpp old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/Sparc/LeonPasses.h b/llvm/lib/Target/Sparc/LeonPasses.h old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/Sparc/SparcAsmPrinter.cpp b/llvm/lib/Target/Sparc/SparcAsmPrinter.cpp index 069e43c6f544..7845a18b14c1 100644 --- a/llvm/lib/Target/Sparc/SparcAsmPrinter.cpp +++ b/llvm/lib/Target/Sparc/SparcAsmPrinter.cpp @@ -351,7 +351,7 @@ void SparcAsmPrinter::printOperand(const MachineInstr *MI, int opNum, break; case MachineOperand::MO_Immediate: - O << (int)MO.getImm(); + O << MO.getImm(); break; case MachineOperand::MO_MachineBasicBlock: MO.getMBB()->getSymbol()->print(O, MAI); diff --git a/llvm/lib/Target/Sparc/SparcSchedule.td b/llvm/lib/Target/Sparc/SparcSchedule.td old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyFastISel.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyFastISel.cpp index 8a0092a3f298..c2a0d3e01740 100644 --- a/llvm/lib/Target/WebAssembly/WebAssemblyFastISel.cpp +++ b/llvm/lib/Target/WebAssembly/WebAssemblyFastISel.cpp @@ -58,6 +58,9 @@ class WebAssemblyFastISel final : public FastISel { int FI; } Base; + // Whether the base has been determined yet + bool IsBaseSet = false; + int64_t Offset = 0; const GlobalValue *GV = nullptr; @@ -74,8 +77,9 @@ class WebAssemblyFastISel final : public FastISel { bool isFIBase() const { return Kind == FrameIndexBase; } void setReg(unsigned Reg) { assert(isRegBase() && "Invalid base register access!"); - assert(Base.Reg == 0 && "Overwriting non-zero register"); + assert(!IsBaseSet && "Base cannot be reset"); Base.Reg = Reg; + IsBaseSet = true; } unsigned getReg() const { assert(isRegBase() && "Invalid base register access!"); @@ -83,8 +87,9 @@ class WebAssemblyFastISel final : public FastISel { } void setFI(unsigned FI) { assert(isFIBase() && "Invalid base frame index access!"); - assert(Base.FI == 0 && "Overwriting non-zero frame index"); + assert(!IsBaseSet && "Base cannot be reset"); Base.FI = FI; + IsBaseSet = true; } unsigned getFI() const { assert(isFIBase() && "Invalid base frame index access!"); @@ -98,13 +103,7 @@ class WebAssemblyFastISel final : public FastISel { int64_t getOffset() const { return Offset; } void setGlobalValue(const GlobalValue *G) { GV = G; } const GlobalValue *getGlobalValue() const { return GV; } - bool isSet() const { - if (isRegBase()) { - return Base.Reg != 0; - } else { - return Base.FI != 0; - } - } + bool isSet() const { return IsBaseSet; } }; /// Keep a pointer to the WebAssemblySubtarget around so that we can make the diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyInstrInfo.td b/llvm/lib/Target/WebAssembly/WebAssemblyInstrInfo.td index 5ff0d73534a6..085910f01ee6 100644 --- a/llvm/lib/Target/WebAssembly/WebAssemblyInstrInfo.td +++ b/llvm/lib/Target/WebAssembly/WebAssemblyInstrInfo.td @@ -328,7 +328,9 @@ defm CONST_F64 : I<(outs F64:$res), (ins f64imm_op:$imm), } // isMoveImm = 1, isAsCheapAsAMove = 1, isReMaterializable = 1 def : Pat<(i32 (WebAssemblywrapper tglobaladdr:$addr)), - (CONST_I32 tglobaladdr:$addr)>, Requires<[IsNotPIC]>; + (CONST_I32 tglobaladdr:$addr)>, Requires<[IsNotPIC, HasAddr32]>; +def : Pat<(i64 (WebAssemblywrapper tglobaladdr:$addr)), + (CONST_I64 tglobaladdr:$addr)>, Requires<[IsNotPIC, HasAddr64]>; def : Pat<(i32 (WebAssemblywrapper tglobaladdr:$addr)), (GLOBAL_GET_I32 tglobaladdr:$addr)>, Requires<[IsPIC]>; diff --git a/llvm/lib/Target/WebAssembly/WebAssemblyRegisterInfo.cpp b/llvm/lib/Target/WebAssembly/WebAssemblyRegisterInfo.cpp index 130589c9df8c..6b6394a58339 100644 --- a/llvm/lib/Target/WebAssembly/WebAssemblyRegisterInfo.cpp +++ b/llvm/lib/Target/WebAssembly/WebAssemblyRegisterInfo.cpp @@ -101,10 +101,12 @@ void WebAssemblyRegisterInfo::eliminateFrameIndex( WebAssemblyFrameLowering::getOpcConst(MF) && MRI.hasOneNonDBGUse(Def->getOperand(0).getReg())) { MachineOperand &ImmMO = Def->getOperand(1); - ImmMO.setImm(ImmMO.getImm() + uint32_t(FrameOffset)); - MI.getOperand(FIOperandNum) - .ChangeToRegister(FrameRegister, /*isDef=*/false); - return; + if (ImmMO.isImm()) { + ImmMO.setImm(ImmMO.getImm() + uint32_t(FrameOffset)); + MI.getOperand(FIOperandNum) + .ChangeToRegister(FrameRegister, /*isDef=*/false); + return; + } } } } diff --git a/llvm/lib/Target/X86/X86EvexToVex.cpp b/llvm/lib/Target/X86/X86EvexToVex.cpp old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/X86/X86FrameLowering.cpp b/llvm/lib/Target/X86/X86FrameLowering.cpp index c7ca6fb2a4fc..db6b68659493 100644 --- a/llvm/lib/Target/X86/X86FrameLowering.cpp +++ b/llvm/lib/Target/X86/X86FrameLowering.cpp @@ -586,29 +586,55 @@ void X86FrameLowering::emitStackProbeInlineGeneric( const uint64_t StackProbeSize = TLI.getStackProbeSize(MF); uint64_t ProbeChunk = StackProbeSize * 8; + uint64_t MaxAlign = + TRI->needsStackRealignment(MF) ? calculateMaxStackAlign(MF) : 0; + // Synthesize a loop or unroll it, depending on the number of iterations. + // BuildStackAlignAND ensures that only MaxAlign % StackProbeSize bits left + // between the unaligned rsp and current rsp. if (Offset > ProbeChunk) { - emitStackProbeInlineGenericLoop(MF, MBB, MBBI, DL, Offset); + emitStackProbeInlineGenericLoop(MF, MBB, MBBI, DL, Offset, + MaxAlign % StackProbeSize); } else { - emitStackProbeInlineGenericBlock(MF, MBB, MBBI, DL, Offset); + emitStackProbeInlineGenericBlock(MF, MBB, MBBI, DL, Offset, + MaxAlign % StackProbeSize); } } void X86FrameLowering::emitStackProbeInlineGenericBlock( MachineFunction &MF, MachineBasicBlock &MBB, - MachineBasicBlock::iterator MBBI, const DebugLoc &DL, - uint64_t Offset) const { + MachineBasicBlock::iterator MBBI, const DebugLoc &DL, uint64_t Offset, + uint64_t AlignOffset) const { const X86Subtarget &STI = MF.getSubtarget(); const X86TargetLowering &TLI = *STI.getTargetLowering(); const unsigned Opc = getSUBriOpcode(Uses64BitFramePtr, Offset); const unsigned MovMIOpc = Is64Bit ? X86::MOV64mi32 : X86::MOV32mi; const uint64_t StackProbeSize = TLI.getStackProbeSize(MF); - uint64_t CurrentOffset = 0; - // 0 Thanks to return address being saved on the stack - uint64_t CurrentProbeOffset = 0; - // For the first N - 1 pages, just probe. I tried to take advantage of + uint64_t CurrentOffset = 0; + + assert(AlignOffset < StackProbeSize); + + // If the offset is so small it fits within a page, there's nothing to do. + if (StackProbeSize < Offset + AlignOffset) { + + MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(Opc), StackPtr) + .addReg(StackPtr) + .addImm(StackProbeSize - AlignOffset) + .setMIFlag(MachineInstr::FrameSetup); + MI->getOperand(3).setIsDead(); // The EFLAGS implicit def is dead. + + addRegOffset(BuildMI(MBB, MBBI, DL, TII.get(MovMIOpc)) + .setMIFlag(MachineInstr::FrameSetup), + StackPtr, false, 0) + .addImm(0) + .setMIFlag(MachineInstr::FrameSetup); + NumFrameExtraProbe++; + CurrentOffset = StackProbeSize - AlignOffset; + } + + // For the next N - 1 pages, just probe. I tried to take advantage of // natural probes but it implies much more logic and there was very few // interesting natural probes to interleave. while (CurrentOffset + StackProbeSize < Offset) { @@ -626,9 +652,9 @@ void X86FrameLowering::emitStackProbeInlineGenericBlock( .setMIFlag(MachineInstr::FrameSetup); NumFrameExtraProbe++; CurrentOffset += StackProbeSize; - CurrentProbeOffset += StackProbeSize; } + // No need to probe the tail, it is smaller than a Page. uint64_t ChunkSize = Offset - CurrentOffset; MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(Opc), StackPtr) .addReg(StackPtr) @@ -639,8 +665,8 @@ void X86FrameLowering::emitStackProbeInlineGenericBlock( void X86FrameLowering::emitStackProbeInlineGenericLoop( MachineFunction &MF, MachineBasicBlock &MBB, - MachineBasicBlock::iterator MBBI, const DebugLoc &DL, - uint64_t Offset) const { + MachineBasicBlock::iterator MBBI, const DebugLoc &DL, uint64_t Offset, + uint64_t AlignOffset) const { assert(Offset && "null offset"); const X86Subtarget &STI = MF.getSubtarget(); @@ -648,6 +674,26 @@ void X86FrameLowering::emitStackProbeInlineGenericLoop( const unsigned MovMIOpc = Is64Bit ? X86::MOV64mi32 : X86::MOV32mi; const uint64_t StackProbeSize = TLI.getStackProbeSize(MF); + if (AlignOffset) { + if (AlignOffset < StackProbeSize) { + // Perform a first smaller allocation followed by a probe. + const unsigned SUBOpc = getSUBriOpcode(Uses64BitFramePtr, AlignOffset); + MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(SUBOpc), StackPtr) + .addReg(StackPtr) + .addImm(AlignOffset) + .setMIFlag(MachineInstr::FrameSetup); + MI->getOperand(3).setIsDead(); // The EFLAGS implicit def is dead. + + addRegOffset(BuildMI(MBB, MBBI, DL, TII.get(MovMIOpc)) + .setMIFlag(MachineInstr::FrameSetup), + StackPtr, false, 0) + .addImm(0) + .setMIFlag(MachineInstr::FrameSetup); + NumFrameExtraProbe++; + Offset -= AlignOffset; + } + } + // Synthesize a loop NumFrameLoopProbe++; const BasicBlock *LLVM_BB = MBB.getBasicBlock(); @@ -666,8 +712,8 @@ void X86FrameLowering::emitStackProbeInlineGenericLoop( // save loop bound { - const unsigned Opc = getSUBriOpcode(Uses64BitFramePtr, Offset); - BuildMI(MBB, MBBI, DL, TII.get(Opc), FinalStackProbed) + const unsigned SUBOpc = getSUBriOpcode(Uses64BitFramePtr, Offset); + BuildMI(MBB, MBBI, DL, TII.get(SUBOpc), FinalStackProbed) .addReg(FinalStackProbed) .addImm(Offset / StackProbeSize * StackProbeSize) .setMIFlag(MachineInstr::FrameSetup); @@ -675,8 +721,8 @@ void X86FrameLowering::emitStackProbeInlineGenericLoop( // allocate a page { - const unsigned Opc = getSUBriOpcode(Uses64BitFramePtr, StackProbeSize); - BuildMI(testMBB, DL, TII.get(Opc), StackPtr) + const unsigned SUBOpc = getSUBriOpcode(Uses64BitFramePtr, StackProbeSize); + BuildMI(testMBB, DL, TII.get(SUBOpc), StackPtr) .addReg(StackPtr) .addImm(StackProbeSize) .setMIFlag(MachineInstr::FrameSetup); @@ -1052,13 +1098,149 @@ void X86FrameLowering::BuildStackAlignAND(MachineBasicBlock &MBB, uint64_t MaxAlign) const { uint64_t Val = -MaxAlign; unsigned AndOp = getANDriOpcode(Uses64BitFramePtr, Val); - MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(AndOp), Reg) - .addReg(Reg) - .addImm(Val) - .setMIFlag(MachineInstr::FrameSetup); - // The EFLAGS implicit def is dead. - MI->getOperand(3).setIsDead(); + MachineFunction &MF = *MBB.getParent(); + const X86Subtarget &STI = MF.getSubtarget(); + const X86TargetLowering &TLI = *STI.getTargetLowering(); + const uint64_t StackProbeSize = TLI.getStackProbeSize(MF); + const bool EmitInlineStackProbe = TLI.hasInlineStackProbe(MF); + + // We want to make sure that (in worst case) less than StackProbeSize bytes + // are not probed after the AND. This assumption is used in + // emitStackProbeInlineGeneric. + if (Reg == StackPtr && EmitInlineStackProbe && MaxAlign >= StackProbeSize) { + { + NumFrameLoopProbe++; + MachineBasicBlock *entryMBB = + MF.CreateMachineBasicBlock(MBB.getBasicBlock()); + MachineBasicBlock *headMBB = + MF.CreateMachineBasicBlock(MBB.getBasicBlock()); + MachineBasicBlock *bodyMBB = + MF.CreateMachineBasicBlock(MBB.getBasicBlock()); + MachineBasicBlock *footMBB = + MF.CreateMachineBasicBlock(MBB.getBasicBlock()); + + MachineFunction::iterator MBBIter = MBB.getIterator(); + MF.insert(MBBIter, entryMBB); + MF.insert(MBBIter, headMBB); + MF.insert(MBBIter, bodyMBB); + MF.insert(MBBIter, footMBB); + const unsigned MovMIOpc = Is64Bit ? X86::MOV64mi32 : X86::MOV32mi; + Register FinalStackProbed = Uses64BitFramePtr ? X86::R11 : X86::R11D; + + // Setup entry block + { + + entryMBB->splice(entryMBB->end(), &MBB, MBB.begin(), MBBI); + BuildMI(entryMBB, DL, TII.get(TargetOpcode::COPY), FinalStackProbed) + .addReg(StackPtr) + .setMIFlag(MachineInstr::FrameSetup); + MachineInstr *MI = + BuildMI(entryMBB, DL, TII.get(AndOp), FinalStackProbed) + .addReg(FinalStackProbed) + .addImm(Val) + .setMIFlag(MachineInstr::FrameSetup); + + // The EFLAGS implicit def is dead. + MI->getOperand(3).setIsDead(); + + BuildMI(entryMBB, DL, + TII.get(Uses64BitFramePtr ? X86::CMP64rr : X86::CMP32rr)) + .addReg(FinalStackProbed) + .addReg(StackPtr) + .setMIFlag(MachineInstr::FrameSetup); + BuildMI(entryMBB, DL, TII.get(X86::JCC_1)) + .addMBB(&MBB) + .addImm(X86::COND_E) + .setMIFlag(MachineInstr::FrameSetup); + entryMBB->addSuccessor(headMBB); + entryMBB->addSuccessor(&MBB); + } + + // Loop entry block + + { + const unsigned SUBOpc = + getSUBriOpcode(Uses64BitFramePtr, StackProbeSize); + BuildMI(headMBB, DL, TII.get(SUBOpc), StackPtr) + .addReg(StackPtr) + .addImm(StackProbeSize) + .setMIFlag(MachineInstr::FrameSetup); + + BuildMI(headMBB, DL, + TII.get(Uses64BitFramePtr ? X86::CMP64rr : X86::CMP32rr)) + .addReg(FinalStackProbed) + .addReg(StackPtr) + .setMIFlag(MachineInstr::FrameSetup); + + // jump + BuildMI(headMBB, DL, TII.get(X86::JCC_1)) + .addMBB(footMBB) + .addImm(X86::COND_B) + .setMIFlag(MachineInstr::FrameSetup); + + headMBB->addSuccessor(bodyMBB); + headMBB->addSuccessor(footMBB); + } + + // setup loop body + { + addRegOffset(BuildMI(bodyMBB, DL, TII.get(MovMIOpc)) + .setMIFlag(MachineInstr::FrameSetup), + StackPtr, false, 0) + .addImm(0) + .setMIFlag(MachineInstr::FrameSetup); + + const unsigned SUBOpc = + getSUBriOpcode(Uses64BitFramePtr, StackProbeSize); + BuildMI(bodyMBB, DL, TII.get(SUBOpc), StackPtr) + .addReg(StackPtr) + .addImm(StackProbeSize) + .setMIFlag(MachineInstr::FrameSetup); + + // cmp with stack pointer bound + BuildMI(bodyMBB, DL, + TII.get(Uses64BitFramePtr ? X86::CMP64rr : X86::CMP32rr)) + .addReg(FinalStackProbed) + .addReg(StackPtr) + .setMIFlag(MachineInstr::FrameSetup); + + // jump + BuildMI(bodyMBB, DL, TII.get(X86::JCC_1)) + .addMBB(bodyMBB) + .addImm(X86::COND_B) + .setMIFlag(MachineInstr::FrameSetup); + bodyMBB->addSuccessor(bodyMBB); + bodyMBB->addSuccessor(footMBB); + } + + // setup loop footer + { + BuildMI(footMBB, DL, TII.get(TargetOpcode::COPY), StackPtr) + .addReg(FinalStackProbed) + .setMIFlag(MachineInstr::FrameSetup); + addRegOffset(BuildMI(footMBB, DL, TII.get(MovMIOpc)) + .setMIFlag(MachineInstr::FrameSetup), + StackPtr, false, 0) + .addImm(0) + .setMIFlag(MachineInstr::FrameSetup); + footMBB->addSuccessor(&MBB); + } + + recomputeLiveIns(*headMBB); + recomputeLiveIns(*bodyMBB); + recomputeLiveIns(*footMBB); + recomputeLiveIns(MBB); + } + } else { + MachineInstr *MI = BuildMI(MBB, MBBI, DL, TII.get(AndOp), Reg) + .addReg(Reg) + .addImm(Val) + .setMIFlag(MachineInstr::FrameSetup); + + // The EFLAGS implicit def is dead. + MI->getOperand(3).setIsDead(); + } } bool X86FrameLowering::has128ByteRedZone(const MachineFunction& MF) const { diff --git a/llvm/lib/Target/X86/X86FrameLowering.h b/llvm/lib/Target/X86/X86FrameLowering.h index c0b4be95f88d..bb2e83205e71 100644 --- a/llvm/lib/Target/X86/X86FrameLowering.h +++ b/llvm/lib/Target/X86/X86FrameLowering.h @@ -213,14 +213,14 @@ class X86FrameLowering : public TargetFrameLowering { void emitStackProbeInlineGenericBlock(MachineFunction &MF, MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, - const DebugLoc &DL, - uint64_t Offset) const; + const DebugLoc &DL, uint64_t Offset, + uint64_t Align) const; void emitStackProbeInlineGenericLoop(MachineFunction &MF, MachineBasicBlock &MBB, MachineBasicBlock::iterator MBBI, - const DebugLoc &DL, - uint64_t Offset) const; + const DebugLoc &DL, uint64_t Offset, + uint64_t Align) const; /// Emit a stub to later inline the target stack probe. MachineInstr *emitStackProbeInlineStub(MachineFunction &MF, diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp index 1671917157f4..56690c3c555b 100644 --- a/llvm/lib/Target/X86/X86ISelLowering.cpp +++ b/llvm/lib/Target/X86/X86ISelLowering.cpp @@ -30285,6 +30285,13 @@ void X86TargetLowering::ReplaceNodeResults(SDNode *N, Results.push_back(V); return; } + case ISD::BITREVERSE: + assert(N->getValueType(0) == MVT::i64 && "Unexpected VT!"); + assert(Subtarget.hasXOP() && "Expected XOP"); + // We can use VPPERM by copying to a vector register and back. We'll need + // to move the scalar in two i32 pieces. + Results.push_back(LowerBITREVERSE(SDValue(N, 0), Subtarget, DAG)); + return; } } @@ -31876,7 +31883,7 @@ X86TargetLowering::EmitLoweredProbedAlloca(MachineInstr &MI, BuildMI(testMBB, DL, TII->get(X86::JCC_1)) .addMBB(tailMBB) - .addImm(X86::COND_L); + .addImm(X86::COND_GE); testMBB->addSuccessor(blockMBB); testMBB->addSuccessor(tailMBB); @@ -31892,9 +31899,9 @@ X86TargetLowering::EmitLoweredProbedAlloca(MachineInstr &MI, // // The property we want to enforce is to never have more than [page alloc] between two probes. - const unsigned MovMIOpc = - TFI.Uses64BitFramePtr ? X86::MOV64mi32 : X86::MOV32mi; - addRegOffset(BuildMI(blockMBB, DL, TII->get(MovMIOpc)), physSPReg, false, 0) + const unsigned XORMIOpc = + TFI.Uses64BitFramePtr ? X86::XOR64mi8 : X86::XOR32mi8; + addRegOffset(BuildMI(blockMBB, DL, TII->get(XORMIOpc)), physSPReg, false, 0) .addImm(0); BuildMI(blockMBB, DL, @@ -36018,8 +36025,10 @@ static SDValue combineTargetShuffle(SDValue N, SelectionDAG &DAG, return DAG.getNode(X86ISD::VBROADCAST, DL, VT, Src.getOperand(0)); // Share broadcast with the longest vector and extract low subvector (free). + // Ensure the same SDValue from the SDNode use is being used. for (SDNode *User : Src->uses()) if (User != N.getNode() && User->getOpcode() == X86ISD::VBROADCAST && + Src == User->getOperand(0) && User->getValueSizeInBits(0) > VT.getSizeInBits()) { return extractSubVector(SDValue(User, 0), 0, DAG, DL, VT.getSizeInBits()); @@ -39588,10 +39597,14 @@ combineVSelectWithAllOnesOrZeros(SDNode *N, SelectionDAG &DAG, // vselect Cond, 000..., X -> andn Cond, X if (TValIsAllZeros) { - MVT AndNVT = MVT::getVectorVT(MVT::i64, CondVT.getSizeInBits() / 64); - SDValue CastCond = DAG.getBitcast(AndNVT, Cond); - SDValue CastRHS = DAG.getBitcast(AndNVT, RHS); - SDValue AndN = DAG.getNode(X86ISD::ANDNP, DL, AndNVT, CastCond, CastRHS); + SDValue CastRHS = DAG.getBitcast(CondVT, RHS); + SDValue AndN; + // The canonical form differs for i1 vectors - x86andnp is not used + if (CondVT.getScalarType() == MVT::i1) + AndN = DAG.getNode(ISD::AND, DL, CondVT, DAG.getNOT(DL, Cond, CondVT), + CastRHS); + else + AndN = DAG.getNode(X86ISD::ANDNP, DL, CondVT, Cond, CastRHS); return DAG.getBitcast(VT, AndN); } diff --git a/llvm/lib/Target/X86/X86SchedBroadwell.td b/llvm/lib/Target/X86/X86SchedBroadwell.td old mode 100755 new mode 100644 diff --git a/llvm/lib/Target/X86/X86SchedSkylakeServer.td b/llvm/lib/Target/X86/X86SchedSkylakeServer.td old mode 100755 new mode 100644 diff --git a/llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp b/llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp index 54c51b6e7161..f2588938d964 100644 --- a/llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp +++ b/llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp @@ -357,7 +357,7 @@ DeadArgumentEliminationPass::Liveness DeadArgumentEliminationPass::MarkIfNotLive(RetOrArg Use, UseVector &MaybeLiveUses) { // We're live if our use or its Function is already marked as live. - if (LiveFunctions.count(Use.F) || LiveValues.count(Use)) + if (IsLive(Use)) return Live; // We're maybe live otherwise, but remember that we must become live if @@ -657,10 +657,18 @@ void DeadArgumentEliminationPass::MarkValue(const RetOrArg &RA, Liveness L, MarkLive(RA); break; case MaybeLive: - // Note any uses of this value, so this return value can be - // marked live whenever one of the uses becomes live. - for (const auto &MaybeLiveUse : MaybeLiveUses) - Uses.insert(std::make_pair(MaybeLiveUse, RA)); + assert(!IsLive(RA) && "Use is already live!"); + for (const auto &MaybeLiveUse : MaybeLiveUses) { + if (IsLive(MaybeLiveUse)) { + // A use is live, so this value is live. + MarkLive(RA); + break; + } else { + // Note any uses of this value, so this value can be + // marked live whenever one of the uses becomes live. + Uses.insert(std::make_pair(MaybeLiveUse, RA)); + } + } break; } } @@ -686,17 +694,20 @@ void DeadArgumentEliminationPass::MarkLive(const Function &F) { /// mark any values that are used by this value (according to Uses) live as /// well. void DeadArgumentEliminationPass::MarkLive(const RetOrArg &RA) { - if (LiveFunctions.count(RA.F)) - return; // Function was already marked Live. + if (IsLive(RA)) + return; // Already marked Live. - if (!LiveValues.insert(RA).second) - return; // We were already marked Live. + LiveValues.insert(RA); LLVM_DEBUG(dbgs() << "DeadArgumentEliminationPass - Marking " << RA.getDescription() << " live\n"); PropagateLiveness(RA); } +bool DeadArgumentEliminationPass::IsLive(const RetOrArg &RA) { + return LiveFunctions.count(RA.F) || LiveValues.count(RA); +} + /// PropagateLiveness - Given that RA is a live value, propagate it's liveness /// to any other values it uses (according to Uses). void DeadArgumentEliminationPass::PropagateLiveness(const RetOrArg &RA) { diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp index fa695c39cd1e..1e43014e7d32 100644 --- a/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp +++ b/llvm/lib/Transforms/InstCombine/InstCombineSelect.cpp @@ -782,25 +782,24 @@ static Value *canonicalizeSaturatedAdd(ICmpInst *Cmp, Value *TVal, Value *FVal, // Match unsigned saturated add of 2 variables with an unnecessary 'not'. // There are 8 commuted variants. - // Canonicalize -1 (saturated result) to true value of the select. Just - // swapping the compare operands is legal, because the selected value is the - // same in case of equality, so we can interchange u< and u<=. + // Canonicalize -1 (saturated result) to true value of the select. if (match(FVal, m_AllOnes())) { std::swap(TVal, FVal); - std::swap(Cmp0, Cmp1); + Pred = CmpInst::getInversePredicate(Pred); } if (!match(TVal, m_AllOnes())) return nullptr; - // Canonicalize predicate to 'ULT'. - if (Pred == ICmpInst::ICMP_UGT) { - Pred = ICmpInst::ICMP_ULT; + // Canonicalize predicate to less-than or less-or-equal-than. + if (Pred == ICmpInst::ICMP_UGT || Pred == ICmpInst::ICMP_UGE) { std::swap(Cmp0, Cmp1); + Pred = CmpInst::getSwappedPredicate(Pred); } - if (Pred != ICmpInst::ICMP_ULT) + if (Pred != ICmpInst::ICMP_ULT && Pred != ICmpInst::ICMP_ULE) return nullptr; // Match unsigned saturated add of 2 variables with an unnecessary 'not'. + // Strictness of the comparison is irrelevant. Value *Y; if (match(Cmp0, m_Not(m_Value(X))) && match(FVal, m_c_Add(m_Specific(X), m_Value(Y))) && Y == Cmp1) { @@ -809,6 +808,7 @@ static Value *canonicalizeSaturatedAdd(ICmpInst *Cmp, Value *TVal, Value *FVal, return Builder.CreateBinaryIntrinsic(Intrinsic::uadd_sat, X, Y); } // The 'not' op may be included in the sum but not the compare. + // Strictness of the comparison is irrelevant. X = Cmp0; Y = Cmp1; if (match(FVal, m_c_Add(m_Not(m_Specific(X)), m_Specific(Y)))) { @@ -819,7 +819,9 @@ static Value *canonicalizeSaturatedAdd(ICmpInst *Cmp, Value *TVal, Value *FVal, Intrinsic::uadd_sat, BO->getOperand(0), BO->getOperand(1)); } // The overflow may be detected via the add wrapping round. - if (match(Cmp0, m_c_Add(m_Specific(Cmp1), m_Value(Y))) && + // This is only valid for strict comparison! + if (Pred == ICmpInst::ICMP_ULT && + match(Cmp0, m_c_Add(m_Specific(Cmp1), m_Value(Y))) && match(FVal, m_c_Add(m_Specific(Cmp1), m_Specific(Y)))) { // ((X + Y) u< X) ? -1 : (X + Y) --> uadd.sat(X, Y) // ((X + Y) u< Y) ? -1 : (X + Y) --> uadd.sat(X, Y) diff --git a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp index ee09a4d9db7e..1557fad4d372 100644 --- a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp +++ b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp @@ -792,7 +792,7 @@ class ModuleAddressSanitizer { StringRef InternalSuffix); Instruction *CreateAsanModuleDtor(Module &M); - bool canInstrumentAliasedGlobal(const GlobalAlias &GA) const; + const GlobalVariable *getExcludedAliasedGlobal(const GlobalAlias &GA) const; bool shouldInstrumentGlobal(GlobalVariable *G) const; bool ShouldUseMachOGlobalsSection() const; StringRef getGlobalMetadataSection() const; @@ -1784,20 +1784,22 @@ void ModuleAddressSanitizer::createInitializerPoisonCalls( } } -bool ModuleAddressSanitizer::canInstrumentAliasedGlobal( - const GlobalAlias &GA) const { +const GlobalVariable * +ModuleAddressSanitizer::getExcludedAliasedGlobal(const GlobalAlias &GA) const { // In case this function should be expanded to include rules that do not just // apply when CompileKernel is true, either guard all existing rules with an // 'if (CompileKernel) { ... }' or be absolutely sure that all these rules // should also apply to user space. assert(CompileKernel && "Only expecting to be called when compiling kernel"); + const Constant *C = GA.getAliasee(); + // When compiling the kernel, globals that are aliased by symbols prefixed // by "__" are special and cannot be padded with a redzone. if (GA.getName().startswith("__")) - return false; + return dyn_cast(C->stripPointerCastsAndAliases()); - return true; + return nullptr; } bool ModuleAddressSanitizer::shouldInstrumentGlobal(GlobalVariable *G) const { @@ -2256,14 +2258,12 @@ bool ModuleAddressSanitizer::InstrumentGlobals(IRBuilder<> &IRB, Module &M, *CtorComdat = false; // Build set of globals that are aliased by some GA, where - // canInstrumentAliasedGlobal(GA) returns false. + // getExcludedAliasedGlobal(GA) returns the relevant GlobalVariable. SmallPtrSet AliasedGlobalExclusions; if (CompileKernel) { for (auto &GA : M.aliases()) { - if (const auto *GV = dyn_cast(GA.getAliasee())) { - if (!canInstrumentAliasedGlobal(GA)) - AliasedGlobalExclusions.insert(GV); - } + if (const GlobalVariable *GV = getExcludedAliasedGlobal(GA)) + AliasedGlobalExclusions.insert(GV); } }