mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-25 13:46:07 +00:00
7ff4f48adb
Fixes a compiler assert on passing a compile time integer to atomic builtins. Assert introduced in D61522 Function changed from ->bool to ->Optional in D76646 Simplifies call sites to getIntegerConstantExpr to elide the now-redundant isValueDependent checks. Reviewed By: aaron.ballman Differential Revision: https://reviews.llvm.org/D112159
28 lines
593 B
C++
28 lines
593 B
C++
// RUN: %clang_cc1 -fsyntax-only -verify %s
|
|
// expected-no-diagnostics
|
|
|
|
// PR8345
|
|
template<typename T> T f(T* value) {
|
|
return __sync_add_and_fetch(value, 1);
|
|
}
|
|
int g(long long* x) { return f(x); }
|
|
int g(int* x) { return f(x); }
|
|
|
|
namespace PR11320 {
|
|
template<typename T>
|
|
void g(unsigned *x) {
|
|
__sync_bool_compare_and_swap(x, 1, 4);
|
|
}
|
|
void h() { g<int>(0); }
|
|
}
|
|
|
|
// Can pass value dependent integer to atomic builtin
|
|
template <int Order>
|
|
void fetchAdd(int *A, int V) {
|
|
__atomic_fetch_add(A, V, Order);
|
|
}
|
|
|
|
void fetchAddUse(int *A, int V) {
|
|
fetchAdd<__ATOMIC_ACQ_REL>(A, V);
|
|
}
|