https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121177
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|normal |enhancement
Last reconfirmed| |2025-07-19
Component|c++ |tree-optimization
Status|UNCONFIRMED |NEW
Ever confirmed|0 |1
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
```
_24 = (long unsigned int) n_2(D);
_25 = operator new [] (_24);
g (_25, n_2(D));
...
_34 = operator new (_24);
__builtin_memcpy (_34, _25, _24);
operator delete [] (_25);
_44 = operator new (_24);
_4(D)->D.85381._M_impl.D.84649._M_start = _44;
_31 = _44 + _24;
_4(D)->D.85381._M_impl.D.84649._M_end_of_storage = _31;
__builtin_memcpy (_44, _34, _24);
_4(D)->D.85381._M_impl.D.84649._M_finish = _31;
operator delete (_34, _24);
```
At least one copy (_34) could be removed I think.
GCC does remove the "copies" when the size is 0 but that is via jump threading.
