We got a bug report from a customer pointing out that calling
promise::set_value on a moved-from promise crashes instead of throwing
an exception with error code future_errc::no_state.
This fixes it, by moving the _S_check calls to *before* we deference
the pointer that the calls check!
This passes all tests, including the more comprehensive ones I've
added as part of this commit, but I think it can wait for stage 1
anyway. We've been shipping this bug for a couple of releases already.
PR libstdc++/80316
* include/std/future (_State_baseV2::_Setter::operator()): Remove
_S_check calls that are done after the pointer to the shared state
is
already dereferenced.
(_State_baseV2::_Setter<_Res, void>): Define specialization for void
as partial specialization so it can be defined within the definition
of _State_baseV2.
(_State_baseV2::__setter): Call _S_check.
(_State_baseV2::__setter(promise<void>*)): Add overload for use by
promise<void>::set_value and
promise<void>::set_value_at_thread_exit.
(promise<T>, promise<T&>, promise<void>): Make _State a friend.
(_State_baseV2::_Setter<void, void>): Remove explicit
specialization.
(promise<void>::set_value, promise<void>::set_value_at_thread_exit):
Use new __setter overload.
* testsuite/30_threads/promise/members/at_thread_exit2.cc: New test.
* testsuite/30_threads/promise/members/set_exception.cc: Test
promise<T&> and promise<void> specializations.
* testsuite/30_threads/promise/members/set_exception2.cc: Likewise.
Test for no_state error condition.
* testsuite/30_threads/promise/members/set_value2.cc: Likewise.