[Bug libstdc++/106772] atomic<T>::wait shouldn't touch waiter pool if used platform wait

["rodgertq at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106772

--- Comment #6 from Thomas Rodgers <rodgertq at gcc dot gnu.org> ---
(In reply to Mkkt Bkkt from comment #5)
> Single reason why I want to use atomic::wait/notify is cross platform api
> for futex, wait/wake on address, ulock, etc
> Not because I need YOU decide instead of me how many spins, or other
> optimization I need.
> 
> boost::atomic already do this.
> 
> Why do you cannot make same?

(In reply to Mkkt Bkkt from comment #4)
> Nice, thanks for benchmarks without code.
> 
> Also, why do I need call notify when don't have wait? 
> 
> Common usage from my point of view looks like:
> 
> atomic.wait(value, mo):
> 
> while (atomic.load(mo) == value) {
>   futex_wait();
> }
> 
> Notify want to looks like:
> 
> atomic.store(1, mo)
> atomic.notify_one();
> 
> See:
> 
> https://github.com/lewissbaker/cppcoro/blob/master/lib/
> lightweight_manual_reset_event.cpp
> 

I have every confidence that Lewis knows how to bring a paper for a
'lightweight manual reset event' to SG1, I suspect it will be well received
when he does.

> https://github.com/YACLib/YACLib/blob/main/src/util/atomic_event.cpp
> 
> and others
> 
> So your optimization is pretty useless.
> 
> Also if for some reason possible few notification, or notify before wait,
> then possible to have code can looks like:
> 
> if (atomic.exchange(non_wait_value, mo) == wait_value) {
>   atomic.notify_one();
> }

I