[OpenWrt-Devel] Q: musl: ether_aton() behaves differently than in glibc

Zefir Kurtisi zefir.kurtisi at neratec.com
Sun Jul 1 05:47:45 EDT 2018


On 06/30/2018 08:56 PM, Rosen Penev wrote:
> On Wed, Jun 27, 2018 at 7:55 AM Zefir Kurtisi <zefir.kurtisi at neratec.com> wrote:
>>
>> Hi,
>>
>> I found a proprietary package stopped working after moving from glibc to musl and
>> ended up identifying a difference in processing of ether_aton().
>>
>> In musl, the ether_addr string is expected to be NULL terminated (see
>> https://git.musl-libc.org/cgit/musl/tree/src/network/ether.c#n20), while other
>> libc implementations are less strict (i.e. only evaluate the leading ether_addr
>> and ignoring subsequent chars).
>>
>> Tried to search the net for a reliable spec on whether it must be NULL terminated
>> or not, but found nothing specific.
>>
>>
>> What do you guys think, is musl just more strict here and therefore correct, or is
>> the less strict behaviour of the other implementations the way to go?
> Probably the musl behavior is correct.

Was my impression, too. Means need to double check whichever package uses that
function and ensure it provides a NULL terminated string.

Cheers

_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel



More information about the openwrt-devel mailing list