[Bug tree-optimization/105918] [12/13 Regression] Spurious Warray-bounds in std::to_chars since r12-2132-ga110855667782dac

["rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105918

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #1)
> Shorter testcase (still need to reduce it futher):
> #include <charconv>
> void toChars(int number) {
>     char temp[1];
>     std::to_chars(temp, temp + 1, number);
> }

So the failure mode is std::to_chars treating a negative 'number' specially,
in this case starting from the end.  We get

<bb 6> [local count: 536870912]:
# __first_9 = PHI <&temp(4), &MEM <char[1]> [(void *)&temp + 1B](5)>
# __unsigned_val_10 = PHI <__unsigned_val_5(4), __unsigned_val_8(5)>

and

<bb 21> [local count: 241591910]:
__num_47 = __val_28 * 2;
_48 = __num_47 + 1;
_49 = __digits[_48];
MEM[(char *)__first_9 + 1B] = _49;
_50 = __digits[__num_47];
*__first_9 = _50;

here the __first_9 + 1 store is _always_ out of bounds.  It might be
that this block is never reachable but we couldn't prove that.