Re: bind EADDRNOTAVAIL at wrong postition

[Philipp Takacs <philipp@bureaucracy.de>]

Hi Alex

[2024-11-01 13:48] Alejandro Colomar <alx@kernel.org>
> Hi Philipp,
>
> On Sun, Oct 06, 2024 at 03:17:48AM +0200, Philipp Takacs wrote:
> > Hi
> > 
> > In bind(2) there are general errors and errors specific for AF_UNIX. The
> > EADDRNOTAVAIL error is sorted at the AF_UNIX specific errors. But in
> > posix[0] EADDRNOTAVAIL is a normal error.
>
> Do you know any system under which it happens with a socket other than
> AF_UNIX?  It would be interesting to know if that's just a POSIX thing,
> or if actual systems do have that.

I found this while debugging a bug[0] on in the OpenSMTPD package of
Debian. The problem was that the interface was not fully up but already
had IP(v6) addresses configured. So OpenSMTPD get the addresses from
the Interface and the bind failed with EADDRNOTAVAIL.

There are also some other places where the linux kernel returns
EADDRNOTAVAIL on non AF_UNIX sockets, for example in sctp[1].

Philipp

[0] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1059700
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/net/sctp/socket.c#n399

>
> Have a lovely day!
> Alex
>
> > 
> > Philipp
> > 
> > [0] https://pubs.opengroup.org/onlinepubs/9699919799/functions/bind.html
>
> -- 
> <https://www.alejandro-colomar.es/>
> part 2     application/pgp-signatur   833