[Bug tree-optimization/127005] Wmaybe-uninitialized with optional>

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127005

Drea Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2026-08-22
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
           Keywords|                            |missed-optimization

--- Comment #1 from Drea Pinski <pinskia at gcc dot gnu.org> ---
So libstdc++ optional has this:
      constexpr void
      _M_destroy() noexcept
      {
        _M_engaged = false;
        _M_payload._M_value.~_Stored_type();


I wonder if we should set engaged before and after the call to ~_Stored_type.

That would allow to optimize away the check like this:
```
  c._M_engaged = 0;
  _13 = MEM[(struct lock_guard *)&c]._M_device;
  _14 = &MEM[(struct mutex *)_13].D.31492._M_mutex;
  pthread_mutex_unlock (_14);
  MEM[(struct lock_guard *)&c] ={v} {CLOBBER(eob)};
  _5 = MEM[(struct _Optional_payload_base *)&c]._M_engaged;
  if (_5 != 0)
```

I might take a look into this next week.