Re: [PATCH 1/3] kstrtox: introduce a safer version of memparse()

[Qu Wenruo]

On 2023/12/27 23:56, David Disseldorp wrote:
> On Sat, 23 Dec 2023 20:28:05 +1030, Qu Wenruo wrote:
>
> > +	s = _parse_integer_fixup_radix(s, &base);
> > +	rv = _parse_integer(s, base, &value);
> > +	if (rv & KSTRTOX_OVERFLOW)
> > +		return -ERANGE;
> > +	if (rv == 0)
> > +		return -EINVAL;
>
> I was playing around with your unit tests and noticed that "0xG" didn't
> reach the expected rv == 0 -> -EINVAL above. It seems that
> _parse_integer_fixup_radix() should handle 0x<non hex> differently, or
> at least step past any autodetected '0' octal prefix.

Yes, that's also the problem I hit, but I'm not 100% sure if changing
_parse_integer_fixup_radix() is safe for other call sites thus I didn't
put such test case here.

My initial failure cases includes things like "0x" which would still
return 0 is already a warning sign.

Thanks,
Qu
> Cheers, David