Merge remote-tracking branch 'upstream/main' into make-cxx-dr-status-utf8

Author: Endilll

.ci/metrics/metrics.py

@@ -43,7 +43,7 @@
 # This means we essentially have a list of workflows sorted by creation date,
 # and that's all we can deduce from it. So for each iteration, we'll blindly
 # process the last N workflows.
-GITHUB_WORKFLOWS_MAX_PROCESS_COUNT = 1000
+GITHUB_WORKFLOWS_MAX_PROCESS_COUNT = 2000
 # Second reason for the cut: reaching a workflow older than X.
 # This means we will miss long-tails (exceptional jobs running for more than
 # X hours), but that's also the case with the count cutoff above.

bolt/lib/Target/AArch64/AArch64MCSymbolizer.cpp

@@ -125,15 +125,39 @@ AArch64MCSymbolizer::adjustRelocation(const Relocation &Rel,
     // instruction pairs and will perform necessary adjustments.
     ErrorOr<uint64_t> SymbolValue = BC.getSymbolValue(*Rel.Symbol);
     assert(SymbolValue && "Symbol value should be set");
-    (void)SymbolValue;
-
-    AdjustedRel.Symbol = BC.registerNameAtAddress("__BOLT_got_zero", 0, 0, 0);
-    AdjustedRel.Addend = Rel.Value;
+    const uint64_t SymbolPageAddr = *SymbolValue & ~0xfffULL;
+
+    // Check if defined symbol and GOT are on the same page. If they are not,
+    // disambiguate the operand.
+    if (BC.MIB->isADRP(Inst) && Rel.Addend == 0 &&
+        SymbolPageAddr == Rel.Value &&
+        !isPageAddressValidForGOT(SymbolPageAddr)) {
+      AdjustedRel.Type = ELF::R_AARCH64_ADR_PREL_PG_HI21;
+    } else {
+      AdjustedRel.Symbol = BC.registerNameAtAddress("__BOLT_got_zero", 0, 0, 0);
+      AdjustedRel.Addend = Rel.Value;
+    }
   }
 
   return AdjustedRel;
 }
 
+bool AArch64MCSymbolizer::isPageAddressValidForGOT(uint64_t PageAddress) const {
+  assert(!(PageAddress & 0xfffULL) && "Page address not aligned at 4KB");
+
+  ErrorOr<BinarySection &> GOT =
+      Function.getBinaryContext().getUniqueSectionByName(".got");
+  if (!GOT || !GOT->getSize())
+    return false;
+
+  const uint64_t GOTFirstPageAddress = GOT->getAddress() & ~0xfffULL;
+  const uint64_t GOTLastPageAddress =
+      (GOT->getAddress() + GOT->getSize() - 1) & ~0xfffULL;
+
+  return PageAddress >= GOTFirstPageAddress &&
+         PageAddress <= GOTLastPageAddress;
+}
+
 void AArch64MCSymbolizer::tryAddingPcLoadReferenceComment(raw_ostream &CStream,
                                                           int64_t Value,
                                                           uint64_t Address) {}

bolt/lib/Target/AArch64/AArch64MCSymbolizer.h

@@ -28,6 +28,9 @@ class AArch64MCSymbolizer : public MCSymbolizer {
   std::optional<Relocation> adjustRelocation(const Relocation &Rel,
                                              const MCInst &Inst) const;
 
+  /// Return true if \p PageAddress is a valid page address for .got section.
+  bool isPageAddressValidForGOT(uint64_t PageAddress) const;
+
 public:
   AArch64MCSymbolizer(BinaryFunction &Function, bool CreateNewSymbols = true)
       : MCSymbolizer(*Function.getBinaryContext().Ctx.get(), nullptr),

bolt/test/AArch64/got-load-symbolization.s

@@ -0,0 +1,100 @@
+## Check that BOLT symbolizer properly handles loads from GOT, including
+## instruction sequences changed/relaxed by the linker.
+
+# RUN: split-file %s %t
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %t/main.s \
+# RUN:   -o %t/main.o
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %t/near.s \
+# RUN:   -o %t/near.o
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %t/far.s \
+# RUN:   -o %t/far.o
+# RUN: %clang %cflags %t/main.o %t/near.o %t/far.o -o %t/main.exe -Wl,-q -static
+# RUN: llvm-bolt %t/main.exe -o %t/main.bolt --keep-nops --print-disasm \
+# RUN:   --print-only=_start | FileCheck %s
+
+#--- main.s
+
+	.text
+	.globl _start
+	.p2align        2
+	.type _start, @function
+# CHECK-LABEL: _start
+_start:
+
+## Function address load relaxable into nop+adr.
+# CHECK: 			nop
+# CHECK-NEXT: adr x0, near
+	adrp    x0, :got:near
+	ldr     x0, [x0, :got_lo12:near]
+
+## Function address load relaxable into adrp+add.
+# CHECK-NEXT: adrp x1, far
+# CHECK-NEXT: add  x1, x1, :lo12:far
+	adrp    x1, :got:far
+	ldr     x1, [x1, :got_lo12:far]
+
+## Non-relaxable due to the instruction in-between.
+# CHECK-NEXT: adrp x2, __BOLT_got_zero
+# CHECK-NEXT: nop
+# CHECK-NEXT: ldr  x2, [x2, :lo12:__BOLT_got_zero{{.*}}]
+	adrp    x2, :got:near
+	nop
+	ldr     x2, [x2, :got_lo12:near]
+
+## Load data object with local visibility. Relaxable into adrp+add.
+# CHECK-NEXT: adrp x3, "local_far_data/1"
+# CHECK-NEXT: add  x3, x3, :lo12:"local_far_data/1"
+  adrp    x3, :got:local_far_data
+  ldr     x3, [x3, :got_lo12:local_far_data]
+
+## Global data reference relaxable into adrp+add.
+# CHECK-NEXT: adrp x4, far_data
+# CHECK-NEXT: add  x4, x4, :lo12:far_data
+  adrp    x4, :got:far_data
+  ldr     x4, [x4, :got_lo12:far_data]
+
+	ret
+	.size _start, .-_start
+
+.weak near
+.weak far
+.weak far_data
+
+## Data object separated by more than 1MB from _start.
+  .data
+  .type local_far_data, @object
+local_far_data:
+  .xword 0
+.size   local_far_data, .-local_far_data
+
+#--- near.s
+
+	.text
+	.globl near
+	.type near, @function
+near:
+  ret
+.size   near, .-near
+
+#--- far.s
+
+  .text
+
+## Insert 1MB of empty space to make objects after it unreachable by adr
+## instructions in _start.
+  .space 0x100000
+
+	.globl far
+	.type far, @function
+far:
+  ret
+.size   far, .-far
+
+  .data
+  .globl far_data
+  .type far_data, @object
+far_data:
+  .xword 0
+.size   far_data, .-far_data
+

clang/docs/LanguageExtensions.rst

@@ -1912,6 +1912,40 @@ A simplistic usage example as might be seen in standard C++ headers follows:
   // Emulate type trait for compatibility with other compilers.
   #endif
 
+
+.. _builtin_structured_binding_size-doc:
+
+__builtin_structured_binding_size (C++)
+---------------------------------------
+
+The ``__builtin_structured_binding_size(T)`` type trait returns
+the *structured binding size* ([dcl.struct.bind]) of type ``T``
+
+This is equivalent to the size of the pack ``p`` in ``auto&& [...p] = declval<T&>();``.
+If the argument cannot be decomposed, ``__builtin_structured_binding_size(T)``
+is not a valid expression (``__builtin_structured_binding_size`` is SFINAE-friendly).
+
+builtin arrays, builtin SIMD vectors,
+builtin complex types, *tuple-like* types, and decomposable class types
+are decomposable types.
+
+A type is considered a valid *tuple-like* if ``std::tuple_size_v<T>`` is a valid expression,
+even if there is no valid ``std::tuple_element`` specialization or suitable
+``get`` function for that type.
+
+.. code-block:: c++
+
+  template<std::size_t Idx, typename T>
+  requires (Idx < __builtin_structured_binding_size(T))
+  decltype(auto) constexpr get_binding(T&& obj) {
+      auto && [...p] = std::forward<T>(obj);
+      return p...[Idx];
+  }
+  struct S { int a = 0, b = 42; };
+  static_assert(__builtin_structured_binding_size(S) == 2);
+  static_assert(get_binding<1>(S{}) == 42);
+
+
 Blocks
 ======
 

clang/docs/ReleaseNotes.rst

@@ -74,6 +74,9 @@ What's New in Clang |release|?
 C++ Language Changes
 --------------------
 
+- Added a :ref:`__builtin_structured_binding_size <builtin_structured_binding_size-doc>` (T)
+  builtin that returns the number of structured bindings that would be produced by destructuring ``T``.
+
 - Similarly to GCC, Clang now supports constant expressions in
   the strings of a GNU ``asm`` statement.
 

clang/include/clang/AST/ExprCXX.h

@@ -51,6 +51,7 @@
 #include <cstdint>
 #include <memory>
 #include <optional>
+#include <variant>
 
 namespace clang {
 
@@ -2765,27 +2766,27 @@ class CXXPseudoDestructorExpr : public Expr {
 /// \endcode
 class TypeTraitExpr final
     : public Expr,
-      private llvm::TrailingObjects<TypeTraitExpr, TypeSourceInfo *> {
+      private llvm::TrailingObjects<TypeTraitExpr, APValue, TypeSourceInfo *> {
   /// The location of the type trait keyword.
   SourceLocation Loc;
 
   ///  The location of the closing parenthesis.
   SourceLocation RParenLoc;
 
-  // Note: The TypeSourceInfos for the arguments are allocated after the
-  // TypeTraitExpr.
-
   TypeTraitExpr(QualType T, SourceLocation Loc, TypeTrait Kind,
-                ArrayRef<TypeSourceInfo *> Args,
-                SourceLocation RParenLoc,
-                bool Value);
+                ArrayRef<TypeSourceInfo *> Args, SourceLocation RParenLoc,
+                std::variant<bool, APValue> Value);
 
   TypeTraitExpr(EmptyShell Empty) : Expr(TypeTraitExprClass, Empty) {}
 
   size_t numTrailingObjects(OverloadToken<TypeSourceInfo *>) const {
     return getNumArgs();
   }
 
+  size_t numTrailingObjects(OverloadToken<APValue>) const {
+    return TypeTraitExprBits.IsBooleanTypeTrait ? 0 : 1;
+  }
+
 public:
   friend class ASTStmtReader;
   friend class ASTStmtWriter;
@@ -2798,19 +2799,34 @@ class TypeTraitExpr final
                                SourceLocation RParenLoc,
                                bool Value);
 
+  static TypeTraitExpr *Create(const ASTContext &C, QualType T,
+                               SourceLocation Loc, TypeTrait Kind,
+                               ArrayRef<TypeSourceInfo *> Args,
+                               SourceLocation RParenLoc, APValue Value);
+
   static TypeTraitExpr *CreateDeserialized(const ASTContext &C,
+                                           bool IsStoredAsBool,
                                            unsigned NumArgs);
 
   /// Determine which type trait this expression uses.
   TypeTrait getTrait() const {
     return static_cast<TypeTrait>(TypeTraitExprBits.Kind);
   }
 
-  bool getValue() const {
-    assert(!isValueDependent());
+  bool isStoredAsBoolean() const {
+    return TypeTraitExprBits.IsBooleanTypeTrait;
+  }
+
+  bool getBoolValue() const {
+    assert(!isValueDependent() && TypeTraitExprBits.IsBooleanTypeTrait);
     return TypeTraitExprBits.Value;
   }
 
+  const APValue &getAPValue() const {
+    assert(!isValueDependent() && !TypeTraitExprBits.IsBooleanTypeTrait);
+    return *getTrailingObjects<APValue>();
+  }
+
   /// Determine the number of arguments to this type trait.
   unsigned getNumArgs() const { return TypeTraitExprBits.NumArgs; }
 

clang/include/clang/AST/Stmt.h

@@ -954,11 +954,13 @@ class alignas(void *) Stmt {
     LLVM_PREFERRED_TYPE(TypeTrait)
     unsigned Kind : 8;
 
-    /// If this expression is not value-dependent, this indicates whether
-    /// the trait evaluated true or false.
     LLVM_PREFERRED_TYPE(bool)
-    unsigned Value : 1;
+    unsigned IsBooleanTypeTrait : 1;
 
+    /// If this expression is a non value-dependent boolean trait,
+    /// this indicates whether the trait evaluated true or false.
+    LLVM_PREFERRED_TYPE(bool)
+    unsigned Value : 1;
     /// The number of arguments to this type trait. According to [implimits]
     /// 8 bits would be enough, but we require (and test for) at least 16 bits
     /// to mirror FunctionType.

clang/include/clang/Analysis/ProgramPoint.h

@@ -39,6 +39,9 @@ class ProgramPointTag {
 public:
   ProgramPointTag(void *tagKind = nullptr) : TagKind(tagKind) {}
   virtual ~ProgramPointTag();
+
+  /// The description of this program point which will be displayed when the
+  /// ExplodedGraph is dumped in DOT format for debugging.
   virtual StringRef getTagDescription() const = 0;
 
   /// Used to implement 'isKind' in subclasses.

clang/include/clang/Basic/DiagnosticSemaKinds.td

@@ -591,6 +591,8 @@ def err_decomp_decl_std_tuple_size_not_constant : Error<
   "is not a valid integral constant expression">;
 def note_in_binding_decl_init : Note<
   "in implicit initialization of binding declaration %0">;
+def err_arg_is_not_destructurable : Error<
+  "type %0 cannot be decomposed">;
 
 def err_std_type_trait_not_class_template : Error<
   "unsupported standard library implementation: "
@@ -7639,9 +7641,13 @@ def warn_arith_conv_mixed_enum_types_cxx20 : Warning<
   "%sub{select_arith_conv_kind}0 "
   "different enumeration types%diff{ ($ and $)|}1,2 is deprecated">,
   InGroup<DeprecatedEnumEnumConversion>;
-def err_conv_mixed_enum_types_cxx26 : Error<
+
+def err_conv_mixed_enum_types: Error <
   "invalid %sub{select_arith_conv_kind}0 "
   "different enumeration types%diff{ ($ and $)|}1,2">;
+def warn_conv_mixed_enum_types_cxx26 : Warning <
+  err_conv_mixed_enum_types.Summary>,
+  InGroup<EnumEnumConversion>, DefaultError;
 
 def warn_arith_conv_mixed_anon_enum_types : Warning<
   warn_arith_conv_mixed_enum_types.Summary>,
@@ -12519,16 +12525,31 @@ def err_builtin_is_within_lifetime_invalid_arg : Error<
   "%select{non-|function }0pointer argument to '__builtin_is_within_lifetime' "
   "is not allowed">;
 
-def err_builtin_invalid_arg_type: Error <
-  "%ordinal0 argument must be "
-  "%select{a vector, integer or floating point type|a matrix|"
-  "a pointer to a valid matrix element type|"
-  "a signed integer or floating point type|a vector type|"
-  "a floating point type|"
-  "a vector of integers|"
-  "an unsigned integer|"
-  "an 'int'|"
-  "a vector of floating points}1 (was %2)">;
+// A multi-component builtin type diagnostic. The first component broadly
+// selects a scalar or container type (scalar, vector or matrix). The second
+// component selects integer types and the third component selects
+// floating-point types. Any component can be left empty.
+def err_builtin_invalid_arg_type: Error<
+  "%ordinal0 argument must be a "
+  // First component: scalar or container types
+  "%select{|scalar|vector|matrix|vector of|scalar or vector of}1"
+  // A comma after generic vector/matrix types if there are non-empty second
+  // and third components, to initiate a list.
+  "%plural{[2,3]:%plural{0:|:%plural{0:|:,}2}3|:}1"
+  // A space after a non-empty first component
+  "%plural{0:|: }1"
+  // Second component: integer-like types
+  "%select{|integer|signed integer|unsigned integer|'int'|"
+  "pointer to a valid matrix element}2"
+  // A space after a non-empty second component
+  "%plural{0:|: }2"
+  // An 'or' if non-empty second and third components are combined
+  "%plural{0:|:%plural{0:|:or }2}3"
+  // Third component: floating-point types
+  "%select{|floating-point}3"
+  // A space after a non-empty third component
+  "%plural{0:|: }3"
+  "%plural{[0,3]:type|:types}1 (was %4)">;
 
 def err_builtin_matrix_disabled: Error<
   "matrix types extension is disabled. Pass -fenable-matrix to enable it">;