https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86701
--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #5)
> I think this is a valid testcase showing that we can't optimize it:
>
> #include<string>
> #include<cstdio>
> #include<cstring>
> int
> main()
> {
> std::string str="";
> str.push_back (0);
> printf ("%i %i\n", str.length(), strlen (str.c_str()));
> return 0;
> }
That is some other issue.
The strlen pass should have figured that out:
MEM <vector(2) char> [(char_type &)&str + 16] = { 0, 0 };
_1 = strlen (&str.D.25443._M_local_buf);
Reduced testcase for that:
```
struct f
{
char s[2];
};
f s;
int f(void)
{
char &t = s.s[0];
char &t1 = s.s[1];
t = 0;
t1 = 0;
int t0 = __builtin_strlen(s.s);
return t0;
}
```
