Re: [PATCH 01/24] man/man7/unix.7: ERRORS: EINVAL: The address field is sun_family, not sun_type

["Günther Noack" <gnoack3000@gmail.com>]

Hello Alejandro!

On Sun, Oct 18, 2026 at 12:46:17PM +0200, Alejandro Colomar wrote:
> > Date: 2026-10-16 19:31:16+0200
> > From: Günther Noack <gnoack3000@gmail.com>
> >
> > struct sockaddr_un has no sun_type field; the address family is in
> > sun_family, as described in "Address format" above.  The check is done
> > by unix_validate_addr() in linux.git net/unix/af_unix.c.
> > 
> > Assisted-by: LLM
> > Signed-off-by: Günther Noack <gnoack3000@gmail.com>
> > ---
> 
> I'd like to understand the reason why the bogus documentation existed at
> all.  For that, a 'Fixes:' tag would be necessary.  It'd also be
> interesting to know whether .sun_type was a human hallucination, or it
> existed at some point.
> 
> In this specific case, it seems to be:
> 
> 	Fixes: 9d1fd1157306 (1999-06-15, 2022-12-19; "man-pages 1.24")

Thanks, added the "Fixes:" tag.


> > Notes:
> >     Fixes an obvious typo
> 
> Most likely; still worth having a Fixes:.  Since I don't see any
> occurences of sun_type in glibc's or Linux's source code, let's assume
> it is a typo.

I also don't think sun_type ever existed.  I've read a bunch of old
documentation on BSD sockets, some of it dating back to the 90s, as
part of wrapping my head around Landlock's networking features, and
I've only come across a "family" field.

The best references to this are sockaddr(3type) and
sockaddr_un(3type).  The first and only common field of each of these
structs is the "family" field, which always has a different prefix
depending on the socket's address family (sun_family, sin_family,
etc.).  struct sockaddr_un only has a sun_family and sun_path field.

Another indicator that "family" is right is that it also talks about
AF_UNIX as the proper value for the field.  This is the value that
sun_family needs to have for it to be recognized as an AF_UNIX
address.  The AF stands for "address family".

–Günther