[Atomic] Merge alignment information from Decl and from Type when emit atomic expression.

[wmi]

Summary

This is to fix PR34347.

EmitAtomicExpr will only use alignment information from Type, instead of Decl, so when the declaration of an atomic variable is marked to have the alignment equal as its size, EmitAtomicExpr don't know about it and will generate libcall instead of atomic op.

The patch merge the alignment information from Decl and from Type.

[wmi]

Ping

[rjmccall]

lib/CodeGen/CGAtomic.cpp

680 ↗

(On Diff #113298)

Just use EmitPointerWithAlignment instead of EmitScalarExpr to emit the pointer operand.

[wmi]

Address John's comment.

[rjmccall]

LGTM, thanks!

[twoh]

Hello, I observed a case where atomic builtin generates libcall when the corresponding sync builtin generates an atomic instruction (https://bugs.llvm.org/show_bug.cgi?id=38846). It seems that the alignment checking for __atomic builtins (line 759 of this patch) results the difference, and wonder if the check is actually necessary. Could anyone please shed some light on understanding this? Thanks!

cfe/trunk/lib/CodeGen/CGAtomic.cpp

AddDirectArgument(CodeGenFunction &CGF, CallArgList &Args,
   }
 }
 
 RValue CodeGenFunction::EmitAtomicExpr(AtomicExpr *E) {
   QualType AtomicTy = E->getPtr()->getType()->getPointeeType();
   QualType MemTy = AtomicTy;
   if (const AtomicType *AT = AtomicTy->getAs<AtomicType>())
     MemTy = AT->getValueType();
-  CharUnits sizeChars, alignChars;
-  std::tie(sizeChars, alignChars) = getContext().getTypeInfoInChars(AtomicTy);
-  uint64_t Size = sizeChars.getQuantity();
-  unsigned MaxInlineWidthInBits = getTarget().getMaxAtomicInlineWidth();
-  bool UseLibcall = (sizeChars != alignChars ||
-                     getContext().toBits(sizeChars) > MaxInlineWidthInBits);
-
   llvm::Value *IsWeak = nullptr, *OrderFail = nullptr;
 
   Address Val1 = Address::invalid();
   Address Val2 = Address::invalid();
   Address Dest = Address::invalid();
-  Address Ptr(EmitScalarExpr(E->getPtr()), alignChars);
+  Address Ptr = EmitPointerWithAlignment(E->getPtr());
+
+  CharUnits sizeChars, alignChars;
+  std::tie(sizeChars, alignChars) = getContext().getTypeInfoInChars(AtomicTy);
+  uint64_t Size = sizeChars.getQuantity();
+  unsigned MaxInlineWidthInBits = getTarget().getMaxAtomicInlineWidth();
+  bool UseLibcall = (sizeChars != Ptr.getAlignment() ||
+                     getContext().toBits(sizeChars) > MaxInlineWidthInBits);
 
   if (E->getOp() == AtomicExpr::AO__c11_atomic_init ||
       E->getOp() == AtomicExpr::AO__opencl_atomic_init) {
     LValue lvalue = MakeAddrLValue(Ptr, AtomicTy);
     EmitAtomicInit(E->getVal1(), lvalue);
     return RValue::get(nullptr);
   }
 

cfe/trunk/test/CodeGenCXX/atomic-align.cpp

+// RUN: %clang_cc1 %s -std=c++11 -emit-llvm -o - -triple=x86_64-linux-gnu | FileCheck %s
+
+struct AM {
+  int f1, f2;
+};
+alignas(8) AM m;
+AM load1() {
+  AM am;
+  // m is declared to align to 8bytes, so generate load atomic instead
+  // of libcall.
+  // CHECK-LABEL: @_Z5load1v
+  // CHECK: load atomic {{.*}} monotonic
+  __atomic_load(&m, &am, 0);
+  return am;
+}
+
+struct BM {
+  int f1;
+  alignas(8) AM f2;
+};
+BM bm;
+AM load2() {
+  AM am;
+  // BM::f2 is declared to align to 8bytes, so generate load atomic instead
+  // of libcall.
+  // CHECK-LABEL: @_Z5load2v
+  // CHECK: load atomic {{.*}} monotonic
+  __atomic_load(&bm.f2, &am, 0);
+  return am;
+}