Merged main:2de3d00edc614cdce4b30b68be564a7a3123dbf9 into amd-gfx:96d9808c3cce

Local branch amd-gfx 96d9808 Revert "Reapply [APInt] Enable APInt ctor assertion by default (llvm#114539)"
Remote branch main 2de3d00 [mlir][tosa] Fix a crash in `PadOp::fold` (llvm#114921)

Author: SC llvm team

clang-tools-extra/clangd/TidyProvider.cpp

@@ -46,7 +46,7 @@ class DotClangTidyCache : private FileCache {
         [this](std::optional<llvm::StringRef> Data) {
           Value.reset();
           if (Data && !Data->empty()) {
-            tidy::DiagCallback Diagnostics = [](const llvm::SMDiagnostic &D) {
+            auto Diagnostics = [](const llvm::SMDiagnostic &D) {
               switch (D.getKind()) {
               case llvm::SourceMgr::DK_Error:
                 elog("tidy-config error at {0}:{1}:{2}: {3}", D.getFilename(),
@@ -149,7 +149,7 @@ static void mergeCheckList(std::optional<std::string> &Checks,
   *Checks = llvm::join_items(",", *Checks, List);
 }
 
-TidyProviderRef provideEnvironment() {
+TidyProvider provideEnvironment() {
   static const std::optional<std::string> User = [] {
     std::optional<std::string> Ret = llvm::sys::Process::GetEnv("USER");
 #ifdef _WIN32
@@ -167,7 +167,7 @@ TidyProviderRef provideEnvironment() {
   return [](tidy::ClangTidyOptions &, llvm::StringRef) {};
 }
 
-TidyProviderRef provideDefaultChecks() {
+TidyProvider provideDefaultChecks() {
   // These default checks are chosen for:
   //  - low false-positive rate
   //  - providing a lot of value
@@ -251,7 +251,7 @@ TidyProvider disableUnusableChecks(llvm::ArrayRef<std::string> ExtraBadChecks) {
   };
 }
 
-TidyProviderRef provideClangdConfig() {
+TidyProvider provideClangdConfig() {
   return [](tidy::ClangTidyOptions &Opts, llvm::StringRef) {
     const auto &CurTidyConfig = Config::current().Diagnostics.ClangTidy;
     if (!CurTidyConfig.Checks.empty())

clang-tools-extra/clangd/TidyProvider.h

@@ -30,11 +30,11 @@ using TidyProviderRef = llvm::function_ref<void(tidy::ClangTidyOptions &,
 TidyProvider combine(std::vector<TidyProvider> Providers);
 
 /// Provider that just sets the defaults.
-TidyProviderRef provideEnvironment();
+TidyProvider provideEnvironment();
 
 /// Provider that will enable a nice set of default checks if none are
 /// specified.
-TidyProviderRef provideDefaultChecks();
+TidyProvider provideDefaultChecks();
 
 /// Provider the enables a specific set of checks and warnings as errors.
 TidyProvider addTidyChecks(llvm::StringRef Checks,
@@ -51,7 +51,7 @@ disableUnusableChecks(llvm::ArrayRef<std::string> ExtraBadChecks = {});
 TidyProvider provideClangTidyFiles(ThreadsafeFS &);
 
 // Provider that uses clangd configuration files.
-TidyProviderRef provideClangdConfig();
+TidyProvider provideClangdConfig();
 
 tidy::ClangTidyOptions getTidyOptionsForFile(TidyProviderRef Provider,
                                              llvm::StringRef Filename);

clang/lib/AST/ByteCode/Compiler.cpp

@@ -6446,8 +6446,6 @@ bool Compiler<Emitter>::emitBuiltinBitCast(const CastExpr *E) {
   QualType ToType = E->getType();
   std::optional<PrimType> ToT = classify(ToType);
 
-  assert(!DiscardResult && "Implement DiscardResult mode for bitcasts.");
-
   if (ToType->isNullPtrType()) {
     if (!this->discard(SubExpr))
       return false;
@@ -6463,12 +6461,24 @@ bool Compiler<Emitter>::emitBuiltinBitCast(const CastExpr *E) {
   }
   assert(!ToType->isReferenceType());
 
+  // Prepare storage for the result in case we discard.
+  if (DiscardResult && !Initializing && !ToT) {
+    std::optional<unsigned> LocalIndex = allocateLocal(E);
+    if (!LocalIndex)
+      return false;
+    if (!this->emitGetPtrLocal(*LocalIndex, E))
+      return false;
+  }
+
   // Get a pointer to the value-to-cast on the stack.
   if (!this->visit(SubExpr))
     return false;
 
-  if (!ToT || ToT == PT_Ptr)
-    return this->emitBitCastPtr(E);
+  if (!ToT || ToT == PT_Ptr) {
+    if (!this->emitBitCastPtr(E))
+      return false;
+    return DiscardResult ? this->emitPopPtr(E) : true;
+  }
   assert(ToT);
 
   const llvm::fltSemantics *TargetSemantics = nullptr;

clang/lib/AST/ByteCode/InterpBuiltinBitCast.cpp

@@ -26,8 +26,8 @@ using namespace clang;
 using namespace clang::interp;
 
 /// Used to iterate over pointer fields.
-using DataFunc =
-    llvm::function_ref<bool(const Pointer &P, PrimType Ty, size_t BitOffset)>;
+using DataFunc = llvm::function_ref<bool(const Pointer &P, PrimType Ty,
+                                         size_t BitOffset, bool PackedBools)>;
 
 #define BITCAST_TYPE_SWITCH(Expr, B)                                           \
   do {                                                                         \
@@ -89,6 +89,7 @@ struct BitcastBuffer {
 
   std::byte *getBytes(unsigned BitOffset) const {
     assert(BitOffset % 8 == 0);
+    assert(BitOffset < SizeInBits);
     return const_cast<std::byte *>(data() + (BitOffset / 8));
   }
 
@@ -147,18 +148,20 @@ static bool enumerateData(const Pointer &P, const Context &Ctx, size_t Offset,
 
   // Primitives.
   if (FieldDesc->isPrimitive())
-    return F(P, FieldDesc->getPrimType(), Offset);
+    return F(P, FieldDesc->getPrimType(), Offset, false);
 
   // Primitive arrays.
   if (FieldDesc->isPrimitiveArray()) {
     bool BigEndianTarget = Ctx.getASTContext().getTargetInfo().isBigEndian();
     QualType ElemType = FieldDesc->getElemQualType();
     size_t ElemSizeInBits = Ctx.getASTContext().getTypeSize(ElemType);
     PrimType ElemT = *Ctx.classify(ElemType);
+    // Special case, since the bools here are packed.
+    bool PackedBools = FieldDesc->getType()->isExtVectorBoolType();
     bool Ok = true;
     for (unsigned I = 0; I != FieldDesc->getNumElems(); ++I) {
       unsigned Index = BigEndianTarget ? (FieldDesc->getNumElems() - 1 - I) : I;
-      Ok = Ok && F(P.atIndex(Index), ElemT, Offset);
+      Ok = Ok && F(P.atIndex(Index), ElemT, Offset, PackedBools);
       Offset += ElemSizeInBits;
     }
     return Ok;
@@ -302,7 +305,8 @@ static bool readPointerToBuffer(const Context &Ctx, const Pointer &FromPtr,
 
   return enumeratePointerFields(
       FromPtr, Ctx,
-      [&](const Pointer &P, PrimType T, size_t BitOffset) -> bool {
+      [&](const Pointer &P, PrimType T, size_t BitOffset,
+          bool PackedBools) -> bool {
         if (!P.isInitialized()) {
           assert(false && "Implement uninitialized value tracking");
           return ReturnOnUninit;
@@ -334,6 +338,8 @@ static bool readPointerToBuffer(const Context &Ctx, const Pointer &FromPtr,
         } else {
           if (const FieldDecl *FD = P.getField(); FD && FD->isBitField())
             BitWidth = FD->getBitWidthValue(ASTCtx);
+          else if (T == PT_Bool && PackedBools)
+            BitWidth = 1;
 
           BITCAST_TYPE_SWITCH(T, {
             T Val = P.deref<T>();
@@ -401,7 +407,7 @@ bool clang::interp::DoBitCastPtr(InterpState &S, CodePtr OpPC,
   size_t BitOffset = 0;
   bool Success = enumeratePointerFields(
       ToPtr, S.getContext(),
-      [&](const Pointer &P, PrimType T, size_t _) -> bool {
+      [&](const Pointer &P, PrimType T, size_t _, bool PackedBools) -> bool {
         if (T == PT_Float) {
           CharUnits ObjectReprChars = ASTCtx.getTypeSizeInChars(P.getType());
           const auto &Semantics = ASTCtx.getFloatTypeSemantics(P.getType());

clang/lib/Sema/SemaFunctionEffects.cpp

@@ -971,6 +971,7 @@ class Analyzer {
     PendingFunctionAnalysis &CurrentFunction;
     CallableInfo &CurrentCaller;
     ViolationSite VSite;
+    const Expr *TrailingRequiresClause = nullptr;
 
     FunctionBodyASTVisitor(Analyzer &Outer,
                            PendingFunctionAnalysis &CurrentFunction,
@@ -985,6 +986,9 @@ class Analyzer {
       if (auto *Dtor = dyn_cast<CXXDestructorDecl>(CurrentCaller.CDecl))
         followDestructor(dyn_cast<CXXRecordDecl>(Dtor->getParent()), Dtor);
 
+      if (auto *FD = dyn_cast<FunctionDecl>(CurrentCaller.CDecl))
+        TrailingRequiresClause = FD->getTrailingRequiresClause();
+
       // Do an AST traversal of the function/block body
       TraverseDecl(const_cast<Decl *>(CurrentCaller.CDecl));
     }
@@ -1259,6 +1263,17 @@ class Analyzer {
       return true;
     }
 
+    bool TraverseStmt(Stmt *Statement) {
+      // If this statement is a `requires` clause from the top-level function
+      // being traversed, ignore it, since it's not generating runtime code.
+      // We skip the traversal of lambdas (beyond their captures, see
+      // TraverseLambdaExpr below), so just caching this from our constructor
+      // should suffice.
+      if (Statement != TrailingRequiresClause)
+        return Base::TraverseStmt(Statement);
+      return true;
+    }
+
     bool TraverseConstructorInitializer(CXXCtorInitializer *Init) {
       ViolationSite PrevVS = VSite;
       if (Init->isAnyMemberInitializer())
@@ -1297,6 +1312,7 @@ class Analyzer {
     }
 
     bool TraverseBlockExpr(BlockExpr * /*unused*/) {
+      // As with lambdas, don't traverse the block's body.
       // TODO: are the capture expressions (ctor call?) safe?
       return true;
     }

clang/test/AST/ByteCode/builtin-bit-cast.cpp

@@ -38,6 +38,13 @@ constexpr Init round_trip(const Init &init) {
   return bit_cast<Init>(bit_cast<Intermediate>(init));
 }
 
+
+namespace Discarding {
+  struct S { int a; };
+  constexpr int f = (__builtin_bit_cast(int, 2), 0);
+  constexpr int f2 = (__builtin_bit_cast(S, 2), 0);
+}
+
 namespace std {
 enum byte : unsigned char {};
 } // namespace std
@@ -468,8 +475,52 @@ struct ref_mem {
 // both-note@+1 {{bit_cast from a type with a reference member is not allowed in a constant expression}}
 constexpr intptr_t run_ref_mem = __builtin_bit_cast(intptr_t, ref_mem{global_int});
 
+namespace test_vector {
+
+typedef unsigned uint2 __attribute__((vector_size(2 * sizeof(unsigned))));
+typedef char byte8 __attribute__((vector_size(sizeof(unsigned long long))));
+
+constexpr uint2 test_vector = { 0x0C05FEFE, 0xCAFEBABE };
+
+static_assert(bit_cast<unsigned long long>(test_vector) == (LITTLE_END
+                                                                ? 0xCAFEBABE0C05FEFE
+                                                                : 0x0C05FEFECAFEBABE), "");
+static_assert(check_round_trip<uint2>(0xCAFEBABE0C05FEFEULL), "");
+static_assert(check_round_trip<byte8>(0xCAFEBABE0C05FEFEULL), "");
+
+typedef bool bool8 __attribute__((ext_vector_type(8)));
+typedef bool bool9 __attribute__((ext_vector_type(9)));
+typedef bool bool16 __attribute__((ext_vector_type(16)));
+typedef bool bool17 __attribute__((ext_vector_type(17)));
+typedef bool bool32 __attribute__((ext_vector_type(32)));
+typedef bool bool128 __attribute__((ext_vector_type(128)));
 
+static_assert(bit_cast<unsigned char>(bool8{1,0,1,0,1,0,1,0}) == (LITTLE_END ? 0x55 : 0xAA), "");
+constexpr bool8 b8 = __builtin_bit_cast(bool8, 0x55); // both-error {{__builtin_bit_cast source size does not equal destination size (4 vs 1)}}
+#if 0
+static_assert(check_round_trip<bool8>(static_cast<unsigned char>(0)), "");
+static_assert(check_round_trip<bool8>(static_cast<unsigned char>(1)), "");
+static_assert(check_round_trip<bool8>(static_cast<unsigned char>(0x55)), "");
+
+static_assert(bit_cast<unsigned short>(bool16{1,1,1,1,1,0,0,0, 1,1,1,1,0,1,0,0}) == (LITTLE_END ? 0x2F1F : 0xF8F4), "");
+
+static_assert(check_round_trip<bool16>(static_cast<short>(0xCAFE)), "");
+static_assert(check_round_trip<bool32>(static_cast<int>(0xCAFEBABE)), "");
+static_assert(check_round_trip<bool128>(static_cast<__int128_t>(0xCAFEBABE0C05FEFEULL)), "");
+#endif
 
+#if 0
+// expected-error@+2 {{constexpr variable 'bad_bool9_to_short' must be initialized by a constant expression}}
+// expected-note@+1 {{bit_cast involving type 'bool __attribute__((ext_vector_type(9)))' (vector of 9 'bool' values) is not allowed in a constant expression; element size 1 * element count 9 is not a multiple of the byte size 8}}
+constexpr unsigned short bad_bool9_to_short = __builtin_bit_cast(unsigned short, bool9{1,1,0,1,0,1,0,1,0});
+// expected-error@+2 {{constexpr variable 'bad_short_to_bool9' must be initialized by a constant expression}}
+// expected-note@+1 {{bit_cast involving type 'bool __attribute__((ext_vector_type(9)))' (vector of 9 'bool' values) is not allowed in a constant expression; element size 1 * element count 9 is not a multiple of the byte size 8}}
+constexpr bool9 bad_short_to_bool9 = __builtin_bit_cast(bool9, static_cast<unsigned short>(0));
+// expected-error@+2 {{constexpr variable 'bad_int_to_bool17' must be initialized by a constant expression}}
+// expected-note@+1 {{bit_cast involving type 'bool __attribute__((ext_vector_type(17)))' (vector of 17 'bool' values) is not allowed in a constant expression; element size 1 * element count 17 is not a multiple of the byte size 8}}
+constexpr bool17 bad_int_to_bool17 = __builtin_bit_cast(bool17, 0x0001CAFEU);
+#endif
+}
 
 namespace test_complex {
   constexpr _Complex unsigned test_int_complex = { 0x0C05FEFE, 0xCAFEBABE };

clang/test/Sema/attr-nonblocking-constraints.cpp

@@ -388,6 +388,51 @@ void nb26() [[clang::nonblocking]] {
 	abort_wrapper(); // no diagnostic
 }
 
+// --- Make sure we don't traverse a requires clause. ---
+
+// Apparently some requires clauses are able to be collapsed into a constant before the nonblocking
+// analysis sees any function calls. This example (extracted from a real-world case where
+// `operator&&` in <valarray>, preceding the inclusion of <expected>) is sufficiently complex
+// to look like it contains function calls. There may be simpler examples.
+
+namespace ExpectedTest {
+
+template <class _Tp>
+inline constexpr bool is_copy_constructible_v = __is_constructible(_Tp, _Tp&);
+
+template <bool, class _Tp = void>
+struct enable_if {};
+template <class _Tp>
+struct enable_if<true, _Tp> {
+  typedef _Tp type;
+};
+
+template <bool _Bp, class _Tp = void>
+using enable_if_t = typename enable_if<_Bp, _Tp>::type;
+
+// Doesn't seem to matter whether the enable_if is true or false.
+template <class E1, class E2, enable_if_t<is_copy_constructible_v<E1>> = 0>
+inline bool operator&&(const E1& x, const E2& y);
+
+template <class _Tp, class _Err>
+class expected {
+public:
+  constexpr expected()
+    {}
+
+  constexpr expected(const expected&)
+    requires(is_copy_constructible_v<_Tp> && is_copy_constructible_v<_Err>)
+  = default;
+};
+
+void test() [[clang::nonblocking]]
+{
+	expected<int, int> a;
+	auto b = a;
+}
+
+} // namespace ExpectedTest
+
 // --- nonblocking implies noexcept ---
 #pragma clang diagnostic warning "-Wperf-constraint-implies-noexcept"
 

llvm/docs/Contributing.rst

@@ -181,8 +181,8 @@ of LLVM's high-level design, as well as its internals:
 .. _irc.oftc.net: irc://irc.oftc.net/llvm
 .. _good first issue: https://github.com/llvm/llvm-project/issues?q=is%3Aopen+is%3Aissue+label%3A%22good+first+issue%22
 .. _bug tracker: https://github.com/llvm/llvm-project/issues
-.. _clang-format-diff.py: https://reviews.llvm.org/source/llvm-github/browse/main/clang/tools/clang-format/clang-format-diff.py
-.. _git-clang-format: https://reviews.llvm.org/source/llvm-github/browse/main/clang/tools/clang-format/git-clang-format
+.. _clang-format-diff.py: https://github.com/llvm/llvm-project/blob/main/clang/tools/clang-format/clang-format-diff.py
+.. _git-clang-format: https://github.com/llvm/llvm-project/blob/main/clang/tools/clang-format/git-clang-format
 .. _LLVM's GitHub: https://github.com/llvm/llvm-project
 .. _LLVM's Phabricator (read-only): https://reviews.llvm.org/
 .. _LLVM's Open Projects page: https://llvm.org/OpenProjects.html#what

llvm/docs/LangRef.rst

@@ -4386,7 +4386,7 @@ is defined inline with other types (e.g. ``[2 x {i32, i32}]``) whereas
 identified types are always defined at the top level with a name.
 Literal types are uniqued by their contents and can never be recursive
 or opaque since there is no way to write one. Identified types can be
-recursive, can be opaqued, and are never uniqued.
+opaqued and are never uniqued. Identified types must not be recursive.
 
 :Syntax:
 

llvm/docs/ReleaseNotes.md

@@ -56,6 +56,8 @@ Makes programs 10x faster by doing Special New Thing.
 Changes to the LLVM IR
 ----------------------
 
+* Types are no longer allowed to be recursive.
+
 * The `x86_mmx` IR type has been removed. It will be translated to
   the standard vector type `<1 x i64>` in bitcode upgrade.
 * Renamed `llvm.experimental.stepvector` intrinsic to `llvm.stepvector`.

llvm/include/llvm/Config/llvm-config.h.cmake

@@ -16,7 +16,7 @@
 
 /* Indicate that this is LLVM compiled from the amd-gfx branch. */
 #define LLVM_HAVE_BRANCH_AMD_GFX
-#define LLVM_MAIN_REVISION 517092
+#define LLVM_MAIN_REVISION 517104
 
 /* Define if LLVM_ENABLE_DUMP is enabled */
 #cmakedefine LLVM_ENABLE_DUMP

llvm/include/llvm/IR/DerivedTypes.h

@@ -33,6 +33,7 @@ class Value;
 class APInt;
 class LLVMContext;
 template <typename T> class Expected;
+class Error;
 
 /// Class to represent integer types. Note that this class is also used to
 /// represent the built-in integer types: Int1Ty, Int8Ty, Int16Ty, Int32Ty and
@@ -317,9 +318,18 @@ class StructType : public Type {
   /// suffix if there is a collision. Do not call this on an literal type.
   void setName(StringRef Name);
 
-  /// Specify a body for an opaque identified type.
+  /// Specify a body for an opaque identified type, which must not make the type
+  /// recursive.
   void setBody(ArrayRef<Type*> Elements, bool isPacked = false);
 
+  /// Specify a body for an opaque identified type or return an error if it
+  /// would make the type recursive.
+  Error setBodyOrError(ArrayRef<Type *> Elements, bool isPacked = false);
+
+  /// Return an error if the body for an opaque identified type would make it
+  /// recursive.
+  Error checkBody(ArrayRef<Type *> Elements);
+
   template <typename... Tys>
   std::enable_if_t<are_base_of<Type, Tys...>::value, void>
   setBody(Type *elt1, Tys *... elts) {