r286129 - [OPENMP] Fixed codegen for __real/__imag expressions in atomic

[Alexey Bataev via cfe-commits]

Hi Dimitry,
I think so

Best regards,
Alexey Bataev

> 7 нояб. 2016 г., в 22:36, Dimitry Andric <dimitry at andric.com> написал(а):
> 
> Hi Alexey,
> 
> Does this look like a good candidate for 3.9.1?
> 
> -Dimitry
> 
>> On 07 Nov 2016, at 19:15, Alexey Bataev via cfe-commits <cfe-commits at lists.llvm.org> wrote:
>> 
>> Author: abataev
>> Date: Mon Nov  7 12:15:02 2016
>> New Revision: 286129
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=286129&view=rev
>> Log:
>> [OPENMP] Fixed codegen for __real/__imag expressions in atomic
>> constructs.
>> 
>> For __real/__imag unary expressions clang emits lvalue with the
>> associated type from the original complex expression, but not the
>> underlying builtin integer or float type. This causes crash in codegen
>> for atomic constructs, if __real/__imag expression are used in atomic
>> constructs.
>> 
>> Modified:
>>   cfe/trunk/lib/CodeGen/CGExpr.cpp
>>   cfe/trunk/test/OpenMP/atomic_write_codegen.c
>> 
>> Modified: cfe/trunk/lib/CodeGen/CGExpr.cpp
>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGExpr.cpp?rev=286129&r1=286128&r2=286129&view=diff
>> ==============================================================================
>> --- cfe/trunk/lib/CodeGen/CGExpr.cpp (original)
>> +++ cfe/trunk/lib/CodeGen/CGExpr.cpp Mon Nov  7 12:15:02 2016
>> @@ -2276,13 +2276,15 @@ LValue CodeGenFunction::EmitUnaryOpLValu
>>      return LV;
>>    }
>> 
>> -    assert(E->getSubExpr()->getType()->isAnyComplexType());
>> +    QualType T = ExprTy->castAs<ComplexType>()->getElementType();
>> 
>>    Address Component =
>>      (E->getOpcode() == UO_Real
>>         ? emitAddrOfRealComponent(LV.getAddress(), LV.getType())
>>         : emitAddrOfImagComponent(LV.getAddress(), LV.getType()));
>> -    return MakeAddrLValue(Component, ExprTy, LV.getAlignmentSource());
>> +    LValue ElemLV = MakeAddrLValue(Component, T, LV.getAlignmentSource());
>> +    ElemLV.getQuals().addQualifiers(LV.getQuals());
>> +    return ElemLV;
>>  }
>>  case UO_PreInc:
>>  case UO_PreDec: {
>> 
>> Modified: cfe/trunk/test/OpenMP/atomic_write_codegen.c
>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/atomic_write_codegen.c?rev=286129&r1=286128&r2=286129&view=diff
>> ==============================================================================
>> --- cfe/trunk/test/OpenMP/atomic_write_codegen.c (original)
>> +++ cfe/trunk/test/OpenMP/atomic_write_codegen.c Mon Nov  7 12:15:02 2016
>> @@ -78,6 +78,9 @@ float2 float2x;
>> register int rix __asm__("esp");
>> 
>> int main() {
>> +// CHECK: store atomic i32 1, i32* getelementptr inbounds ({ i32, i32 }, { i32, i32 }* @civ, i32 0, i32 1) monotonic,
>> +#pragma omp atomic write
>> + __imag(civ) = 1;
>> // CHECK: load i8, i8*
>> // CHECK: store atomic i8
>> #pragma omp atomic write
>> 
>> 
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
>