https://github.com/hokein updated 
https://github.com/llvm/llvm-project/pull/86512

>From 011d6bbb434bdb46efe92891fe356885f82c4445 Mon Sep 17 00:00:00 2001
From: Haojian Wu <hokein...@gmail.com>
Date: Mon, 25 Mar 2024 15:10:51 +0100
Subject: [PATCH] ination of 2 commits.

[clang] Implement a bitwise_copyable builtin type trait.

This patch implements a `__is_bitwise_copyable` builtin in clang.

The bitwise copyable types act as the trivially copyable types, but
they support a wider range of types (e.g. classes with virtual methods) --
their underlying types can be safely copied by `memcopy` or `memmove`,
the clang compiler guarantees that both source and destination objects
have the same *object* representations after the copy operation, and the
lifetime of the destination object implicitly starts.

A particular use case of this builtin is to clone an object via memcopy
(without running the constructor):

```
Message* clone(const Message* src, char* buffer, int size) {
  if constexpr __is_bitwise_copyable(Message) {
    // bitwise copy to buffer
    __builtin_memcpy(buffer, src, size);
    // using __builtin_launder to prevent miscompile for 
-fstrict-vtable-pointers.
    return __builtin_launder(reinterpret_cast<Message*>(buffer));
  }
  // Fallback the operator new, which calls the constructor to start the 
lifetime.
  return new(buffer) Message(src);
}
```

Note that the definition of bitwise copyable is not tied to the Rule Of
Five, so users of this builtin must guarantee that program semantic constraints
are satisfied, e.g. no double resource deallocations.

Context: https://discourse.llvm.org/t/extension-for-creating-objects-via-memcpy
---
 clang/docs/LanguageExtensions.rst             | 28 +++++++++++++++
 clang/docs/ReleaseNotes.rst                   |  3 ++
 clang/include/clang/AST/Type.h                | 16 +++++++++
 clang/include/clang/Basic/TokenKinds.def      |  2 ++
 clang/lib/AST/Type.cpp                        | 35 +++++++++++++++++++
 clang/lib/Sema/SemaExprCXX.cpp                |  3 ++
 ...builtin-is-bitwise-cloneable-fsanitize.cpp | 34 ++++++++++++++++++
 .../SemaCXX/builtin-is-bitwise-cloneable.cpp  |  9 +++++
 clang/test/SemaObjCXX/arc-type-traits.mm      |  9 +++++
 9 files changed, 139 insertions(+)
 create mode 100644 
clang/test/SemaCXX/builtin-is-bitwise-cloneable-fsanitize.cpp
 create mode 100644 clang/test/SemaCXX/builtin-is-bitwise-cloneable.cpp

diff --git a/clang/docs/LanguageExtensions.rst 
b/clang/docs/LanguageExtensions.rst
index 46f99d0bbdd06..da569b9577e75 100644
--- a/clang/docs/LanguageExtensions.rst
+++ b/clang/docs/LanguageExtensions.rst
@@ -4016,6 +4016,34 @@ Note that the `size` argument must be a compile time 
constant.
 
 Note that this intrinsic cannot yet be called in a ``constexpr`` context.
 
+``__is_bitwise_cloneable``
+--------------------------
+
+A type trait is used to check whether a type can be safely copied by memcpy.
+
+**Syntax**:
+
+.. code-block:: c++
+
+  bool __is_bitwise_cloneable(Type)
+
+**Description**:
+
+This trait is similar to `std::is_trivially_copyable`, but additionally allows
+to have user-defined constructors, virtual functions and virtual bases. It is 
up
+to the user code to guarantee that a bitwise copy results in non-broken object.
+
+Objects of bitwise cloneable types can be bitwise copied by memcpy/memmove. The
+Clang compiler warrants that this behavior is well defined, and won't be
+broken by compiler optimizations.
+
+For implicit-lifetime types, the lifetime of the new object is implicitly
+started after the copy. For other types (e.g., classes with virtual methods),
+the lifetime isn't started, and using the object results in undefined behavior
+according to the C++ Standard.
+
+This builtin can be used in constant expressions.
+
 Atomic Min/Max builtins with memory ordering
 --------------------------------------------
 
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 58e407a724a9c..2143611560278 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -318,6 +318,9 @@ Non-comprehensive list of changes in this release
 - Builtins ``__builtin_shufflevector()`` and ``__builtin_convertvector()`` may
   now be used within constant expressions.
 
+- Added ``__buitlin_bitwise_clonable`` which is used to check whether a type
+  can be safely copied by memcpy/memmove.
+
 New Compiler Flags
 ------------------
 - ``-fsanitize=implicit-bitfield-conversion`` checks implicit truncation and
diff --git a/clang/include/clang/AST/Type.h b/clang/include/clang/AST/Type.h
index 263b632df23ce..0adada068bbfe 100644
--- a/clang/include/clang/AST/Type.h
+++ b/clang/include/clang/AST/Type.h
@@ -1120,6 +1120,22 @@ class QualType {
   /// Return true if this is a trivially copyable type (C++0x [basic.types]p9)
   bool isTriviallyCopyableType(const ASTContext &Context) const;
 
+  /// Return true if the type is safe to bitwise copy using memcpy/memmove.
+  ///
+  /// This is an extension in clang: bitwise clonable types act as trivially
+  /// copyable types, meaning their underlying bytes can be safely copied by
+  /// memcpy or memmove. After the copy, the destination object has the same
+  /// object representation.
+  ///
+  /// However, there are cases where it is not safe to copy:
+  ///  - When sanitizers, such as AddressSanitizer, add padding with poison,
+  ///    which can cause issues if those poisoned padding bits are accessed.
+  ///  - Types with Objective-C lifetimes, where specific runtime
+  ///    semantics may not be preserved during a bitwise copy.
+  ///
+  // FIXME: each call will trigger a full computation, cache the result.
+  bool isBitwiseCloneableType(const ASTContext &Context) const;
+
   /// Return true if this is a trivially copyable type
   bool isTriviallyCopyConstructibleType(const ASTContext &Context) const;
 
diff --git a/clang/include/clang/Basic/TokenKinds.def 
b/clang/include/clang/Basic/TokenKinds.def
index b5a0e9df9f7ae..9c4b17465e18a 100644
--- a/clang/include/clang/Basic/TokenKinds.def
+++ b/clang/include/clang/Basic/TokenKinds.def
@@ -542,6 +542,8 @@ TYPE_TRAIT_2(__reference_converts_from_temporary, 
ReferenceConvertsFromTemporary
 // is not exposed to users.
 TYPE_TRAIT_2(/*EmptySpellingName*/, IsDeducible, KEYCXX)
 
+TYPE_TRAIT_1(__is_bitwise_cloneable, IsBitwiseCloneable, KEYALL)
+
 // Embarcadero Expression Traits
 EXPRESSION_TRAIT(__is_lvalue_expr, IsLValueExpr, KEYCXX)
 EXPRESSION_TRAIT(__is_rvalue_expr, IsRValueExpr, KEYCXX)
diff --git a/clang/lib/AST/Type.cpp b/clang/lib/AST/Type.cpp
index 04f105c128872..a35bf93bb7d54 100644
--- a/clang/lib/AST/Type.cpp
+++ b/clang/lib/AST/Type.cpp
@@ -2749,6 +2749,41 @@ bool QualType::isTriviallyCopyableType(const ASTContext 
&Context) const {
                                      /*IsCopyConstructible=*/false);
 }
 
+bool QualType::isBitwiseCloneableType(const ASTContext & Context) const {
+  auto CanonicalType = getCanonicalType();
+  if (CanonicalType.hasNonTrivialObjCLifetime())
+    return false;
+  if (CanonicalType->isArrayType())
+    return Context.getBaseElementType(CanonicalType)
+        .isBitwiseCloneableType(Context);
+
+  if (CanonicalType->isIncompleteType())
+    return false;
+
+  if (const auto *RD = CanonicalType->getAsRecordDecl()) { // 
struct/union/class
+    // Never allow memcpy when we're adding poisoned padding bits to the 
struct.
+    // Accessing these posioned bits will trigger false alarms on
+    // SanitizeAddressFieldPadding etc.
+    if (RD->mayInsertExtraPadding())
+      return false;
+
+    for (auto *const Field : RD->fields()) {
+      if (!Field->getType().isBitwiseCloneableType(Context))
+        return false;
+    }
+
+    if (const auto *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
+      for (auto Base : CXXRD->bases())
+        if (!Base.getType().isBitwiseCloneableType(Context))
+          return false;
+      for (auto VBase : CXXRD->vbases())
+        if (!VBase.getType().isBitwiseCloneableType(Context))
+          return false;
+    }
+  }
+  return true;
+}
+
 bool QualType::isTriviallyCopyConstructibleType(
     const ASTContext &Context) const {
   return isTriviallyCopyableTypeImpl(*this, Context,
diff --git a/clang/lib/Sema/SemaExprCXX.cpp b/clang/lib/Sema/SemaExprCXX.cpp
index d3e9dcb4f4399..1206e460c2428 100644
--- a/clang/lib/Sema/SemaExprCXX.cpp
+++ b/clang/lib/Sema/SemaExprCXX.cpp
@@ -5125,6 +5125,7 @@ static bool CheckUnaryTypeTraitTypeCompleteness(Sema &S, 
TypeTrait UTT,
   case UTT_IsStandardLayout:
   case UTT_IsPOD:
   case UTT_IsLiteral:
+  case UTT_IsBitwiseCloneable:
   // By analogy, is_trivially_relocatable and is_trivially_equality_comparable
   // impose the same constraints.
   case UTT_IsTriviallyRelocatable:
@@ -5618,6 +5619,8 @@ static bool EvaluateUnaryTypeTrait(Sema &Self, TypeTrait 
UTT,
     return C.hasUniqueObjectRepresentations(T);
   case UTT_IsTriviallyRelocatable:
     return T.isTriviallyRelocatableType(C);
+  case UTT_IsBitwiseCloneable:
+    return T.isBitwiseCloneableType(C);
   case UTT_IsReferenceable:
     return T.isReferenceable();
   case UTT_CanPassInRegs:
diff --git a/clang/test/SemaCXX/builtin-is-bitwise-cloneable-fsanitize.cpp 
b/clang/test/SemaCXX/builtin-is-bitwise-cloneable-fsanitize.cpp
new file mode 100644
index 0000000000000..d47a39a0754c5
--- /dev/null
+++ b/clang/test/SemaCXX/builtin-is-bitwise-cloneable-fsanitize.cpp
@@ -0,0 +1,34 @@
+// RUN: %clang_cc1 -triple x86_64-unknown-linux -DSANITIZER_ENABLED 
-fsanitize=address -fsanitize-address-field-padding=1 %s
+// RUN: %clang_cc1 -triple x86_64-unknown-linux %s
+
+struct S {
+  ~S() {}
+  virtual void foo() {}
+
+  int buffer[1];
+  int other_field = 0;
+};
+
+union U {
+  S s;
+};
+
+struct Derived : S {};
+
+static_assert(!__is_trivially_copyable(S));
+#ifdef SANITIZER_ENABLED
+// Don't allow memcpy when the struct has poisoned padding bits.
+// The sanitizer adds posion padding bits to struct S.
+static_assert(sizeof(S) > 16);
+static_assert(!__is_bitwise_cloneable(S));
+static_assert(sizeof(U) == sizeof(S)); // no padding bit for U.
+static_assert(!__is_bitwise_cloneable(U));
+static_assert(!__is_bitwise_cloneable(S[2]));
+static_assert(!__is_bitwise_cloneable(Derived));
+#else
+static_assert(sizeof(S) == 16);
+static_assert(__is_bitwise_cloneable(S));
+static_assert(__is_bitwise_cloneable(U));
+static_assert(__is_bitwise_cloneable(S[2]));
+static_assert(__is_bitwise_cloneable(Derived));
+#endif
diff --git a/clang/test/SemaCXX/builtin-is-bitwise-cloneable.cpp 
b/clang/test/SemaCXX/builtin-is-bitwise-cloneable.cpp
new file mode 100644
index 0000000000000..706a0521a5882
--- /dev/null
+++ b/clang/test/SemaCXX/builtin-is-bitwise-cloneable.cpp
@@ -0,0 +1,9 @@
+// RUN: %clang_cc1 -fsyntax-only -verify %s
+// expected-no-diagnostics
+//
+struct DynamicClass { virtual int Foo(); };
+static_assert(!__is_trivially_copyable(DynamicClass));
+static_assert(__is_bitwise_cloneable(DynamicClass));
+
+struct InComplete;
+static_assert(!__is_bitwise_cloneable(InComplete));
diff --git a/clang/test/SemaObjCXX/arc-type-traits.mm 
b/clang/test/SemaObjCXX/arc-type-traits.mm
index 2d30ae450f3b0..25bc8b362140a 100644
--- a/clang/test/SemaObjCXX/arc-type-traits.mm
+++ b/clang/test/SemaObjCXX/arc-type-traits.mm
@@ -221,3 +221,12 @@
 TRAIT_IS_TRUE(__is_trivially_relocatable, HasStrong);
 TRAIT_IS_FALSE(__is_trivially_relocatable, HasWeak);
 TRAIT_IS_TRUE(__is_trivially_relocatable, HasUnsafeUnretained);
+
+// __is_bitwise_cloneable
+TRAIT_IS_FALSE(__is_bitwise_cloneable, __strong id);
+TRAIT_IS_FALSE(__is_bitwise_cloneable, __weak id);
+TRAIT_IS_FALSE(__is_bitwise_cloneable, __autoreleasing id);
+TRAIT_IS_TRUE(__is_trivial, __unsafe_unretained id);
+TRAIT_IS_FALSE(__is_bitwise_cloneable, HasStrong);
+TRAIT_IS_FALSE(__is_bitwise_cloneable, HasWeak);
+TRAIT_IS_TRUE(__is_bitwise_cloneable, HasUnsafeUnretained);

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to