Tested x86_64-linux, pushed to trunk. The test was only failing for me with -m32 (and not -m64), so I didn't notice until now. That probably means we should make the test fail more reliably if the padding isn't being cleared.
-- >8 -- This test was written assuming that std::atomic_ref clears its target's padding on construction, but that could introduce data races. Change the test to store a value after construction and check that its padding is cleared by the store. libstdc++-v3/ChangeLog: * testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc: Store value with non-zero padding bits after construction. --- .../29_atomics/atomic_ref/compare_exchange_padding.cc | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/libstdc++-v3/testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc b/libstdc++-v3/testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc index 1b1a12dddda..e9f8a4bdf2a 100644 --- a/libstdc++-v3/testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc +++ b/libstdc++-v3/testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc @@ -20,14 +20,15 @@ int main () { S s; - fill_struct(s); - s.c = 'a'; - s.s = 42; - S ss{ s }; + fill_struct(ss); + ss.c = 'a'; + ss.s = 42; + std::atomic_ref<S> as{ s }; + as.store(ss); auto ts = as.load(); - VERIFY( !compare_struct(ss, ts) ); // padding cleared on construction + VERIFY( !compare_struct(ss, ts) ); // padding cleared on store as.exchange(ss); auto es = as.load(); VERIFY( compare_struct(ts, es) ); // padding cleared on exchange -- 2.37.3