https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119990
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Is this reduced from a program or were you trying things and ran into this
issue?
The reason why I ask is that GCC knows that GCC is called only once so it
inlines less into it if not inside a loop.
The IR:
```
std::__cxx11::basic_string<char>::basic_string<> (&obj.one, "", &D.46982);
std::__cxx11::basic_string<char>::basic_string<> (&obj.d.b, "hi", &D.47023);
<bb 3> [local count: 1073741824]:
std::__cxx11::basic_string<char>::basic_string<> (&obj.c, "hi2", &D.47025);
<bb 4> [local count: 1073741824]:
D.47025 ={v} {CLOBBER(eos)};
D.47023 ={v} {CLOBBER(eos)};
D.46982 ={v} {CLOBBER(eos)};
_17 = MEM[(const struct basic_string *)&obj]._M_dataplus._M_p;
__builtin_puts (_17);
```
I think it is warning on the _17 here being passed to puts here. Which is
almost definitely an incorrect warning.
MEM[(const struct basic_string *)&obj]._M_dataplus._M_p is basically
&obj.one._M_dataplus._M_p but somehow GCC got lost. and not noticing MEM[(const
struct basic_string *)&obj] and obj.one are the same.
