[PATCH] inet: make ether_aton() reject garbage after the address

[Florian Weimer]

* Adhemerval Zanella Netto:

> On 30/07/24 18:03, Miklós Máté wrote:
>> On 29/07/2024 14:11, Andreas Schwab wrote:
>>> On Jul 29 2024, Florian Weimer wrote:
>>>
>>>> However, inet_aton does not, where is is coming from.  See
>>>> resolv/tst-inet_aton_exact.c.
>>>>
>>>> We should move off this behavior because it is really surprising.  I
>>>> think for ether_aton, we can do it without introducing a compatibility
>>>> symbol.
>>>>
>>>> Andreas, what do you think?
>>> BSD does not reject trailing garbage.
>>>
>> I just checked the FreeBSD implementation [0], and it's even worse than the one in glibc (and BTW their struct ether_addr is incompatible). Their sscanf accepts "abcdef:fedcba:1234567890:0987654321:deadbeef:00000000000Mxyzptlk" as a valid MAC address.
>> 
>> [0] https://github.com/freebsd/freebsd-src/blob/6f25b46721a18cf4f036d041e7e5d275800a00b3/lib/libc/net/ether_addr.c#L86
>
> Andrea, Florian, do we have a consensus about this change? I think Florian
> is ok with this, but I am not sure about Andreas position.

I think we can make the change and probably should, but Andreas seems to
disagree?

Thanks,
Florian