https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103483
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
See Also| |https://gcc.gnu.org/bugzill
| |a/show_bug.cgi?id=103332
Ever confirmed|0 |1
Last reconfirmed| |2021-11-30
--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #1)
> I think the warning is correct really.
It's wrong.
> as if
> d is a+1000 it is undefined. Without inlining there is no way to know if d
> is any value.
It's able to inline the size of the buffer all the way down to the
char_traits::copy call, but isn't able to inline the length argument as well.
If it can't use the compile-time constant length that is present in the code,
it shouldn't bother doing these checks. The middle-end should not assume all
code is wrong by default. These bogus stringop warnings in std::string are
getting out of hand.
See also PR103332 and others.
