1762e01cca
Fix codegen of consteval functions returning an empty class, and related issues If a class is empty, don't store it to memory: the store might overwrite useful data. Similarly, if a class has tail padding that might overlap other fields, don't store the tail padding to memory. The problem here turned out a bit more general than I initially thought: basically all uses of EmitAggregateStore were broken. Call lowering had a method that did mostly the right thing, though: CreateCoercedStore. Adapt CreateCoercedStore so it always does the conservatively right thing, and use it for both calls and ConstantExpr. Also, along the way, fix the "overlap" bit in AggValueSlot: the bit was set incorrectly for empty classes in some cases. Fixes #93040.
55 lines
1.3 KiB
C++
55 lines
1.3 KiB
C++
// RUN: %clang_cc1 %s -triple=amdgcn-amd-amdhsa -emit-llvm -o - | FileCheck %s
|
|
|
|
template<typename T, unsigned int n> struct my_vector_base;
|
|
|
|
template<typename T>
|
|
struct my_vector_base<T, 1> {
|
|
typedef T Native_vec_ __attribute__((ext_vector_type(1)));
|
|
|
|
union {
|
|
Native_vec_ data;
|
|
struct {
|
|
T x;
|
|
};
|
|
};
|
|
};
|
|
|
|
template<typename T, unsigned int rank>
|
|
struct my_vector_type : public my_vector_base<T, rank> {
|
|
using my_vector_base<T, rank>::data;
|
|
using typename my_vector_base<T, rank>::Native_vec_;
|
|
|
|
template< typename U>
|
|
my_vector_type(U x) noexcept
|
|
{
|
|
for (auto i = 0u; i != rank; ++i) data[i] = x;
|
|
}
|
|
my_vector_type& operator+=(const my_vector_type& x) noexcept
|
|
{
|
|
data += x.data;
|
|
return *this;
|
|
}
|
|
};
|
|
|
|
template<typename T, unsigned int n>
|
|
inline
|
|
my_vector_type<T, n> operator+(
|
|
const my_vector_type<T, n>& x, const my_vector_type<T, n>& y) noexcept
|
|
{
|
|
return my_vector_type<T, n>{x} += y;
|
|
}
|
|
|
|
using char1 = my_vector_type<char, 1>;
|
|
|
|
int mane() {
|
|
|
|
char1 f1{1};
|
|
char1 f2{1};
|
|
|
|
// CHECK: [[CALL:%.*]] = call i16
|
|
// CHECK: [[TRUNC:%.*]] = trunc i16 [[CALL]] to i8
|
|
// CHECK: store i8 [[TRUNC]]
|
|
|
|
char1 f3 = f1 + f2;
|
|
}
|