XFAIL align.pass.pass.cpp on aix regarding atomic alignment

[jasonliu]

Summary

AIX currently fails because it does not support 16-byte MaxAtomicPromoteWidth while the type is implied to be lock-free.

[hubert.reinterpretcast]

libcxx/test/libcxx/atomics/atomics.align/align.pass.pass.cpp

18

Just trying to understand... this is a "known bug" (properly tracked) that is now being marked XFAIL to make it easier to spot other problems in regular regression testing? Might it be better to apply #ifndef _AIX to a subset of the calls?

35

So, this is returning the wrong value?

[hubert.reinterpretcast]

libcxx/test/libcxx/atomics/atomics.align/align.pass.pass.cpp

37

This is a statically-determined fact, but the storage for the object could be aligned even if the type's guaranteed alignment is less-than-natural. We now have is_always_lock_free.

[cebowleratibm]

libcxx/test/libcxx/atomics/atomics.align/align.pass.pass.cpp

37

Are you saying the test is invalid because some of the types below may have less than natural alignment and is_lock_free() is permitted to return true based on the actual alignment of the particular object?

[hubert.reinterpretcast]

libcxx/test/libcxx/atomics/atomics.align/align.pass.pass.cpp

37

I recall reading this before... the test is valid:
In any given program execution, the result of the lock-free query is the same for all atomic objects of the same type.

[jasonliu]

is_lock_free actually returns false for this structure:
struct LLIArr2 { long long int i[2]; };

So I'm seeing this test case passing on pwr8 and pwr9.
I don't think we need this change for AIX.

libcxx/test/libcxx/atomics/atomics.align/align.pass.pass.cpp

 //===----------------------------------------------------------------------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // UNSUPPORTED: libcpp-has-no-threads, c++03
 // REQUIRES: is-lockfree-runtime-function
 
 // GCC currently fails because it needs -fabi-version=6 to fix mangling of
 // std::atomic when used with __attribute__((vector(X))).
 // XFAIL: gcc
 
+// AIX currently fails becase it does not support 16-byte MaxAtomicPromoteWidth
+// while the type is implied to be lock-free.
+// XFAIL: aix
+
 // <atomic>
 
 // Verify that the content of atomic<T> is properly aligned if the type is
 // lock-free. This can't be observed through the atomic<T> API. It is
 // nonetheless required for correctness of the implementation: lock-free implies
 // that ISA instructions are used, and these instructions assume "suitable
 // alignment". Supported architectures all require natural alignment for
 // lock-freedom (e.g. load-linked / store-conditional, or cmpxchg).
 
 #include <atomic>
 #include <cassert>
 
 template <typename T>
 struct atomic_test : public std::__atomic_base<T> {
   atomic_test() {
     if (this->is_lock_free()) {
       using AtomicImpl = decltype(this->__a_);
       assert(alignof(AtomicImpl) >= sizeof(AtomicImpl) &&
              "expected natural alignment for lock-free type");
     }
   }
 };
 
 int main(int, char**) {
 
 // structs and unions can't be defined in the template invocation.