Sebastien Roy writes:
> > I don't think you need it for SOCK_RAW, because you can set the source
> > using IP_HDRINCL.
> 
> You can't set the output interface, nor receive the input interface with
> that.  If it could do everything that's proposed here, then I really
> don't see why we're doing this at all instead of just doing IP_HDRINCL
> for UDP sockets...

The original problem being solved is about source address on UDP
output, and getting it right per RFC 1123.  You can already do that
with raw IP.

Interfaces are a red herring.  If it supports them as well, that's
interesting, but supporting them has nothing to do with solving the
original problem.

Plus, IP_RECVIF and IP_XMIT_IF should work for raw, and shouldn't
require any changes (other than perhaps a man page entry).  Right?

> Plus, when the ancillary data and socket options are generalized across
> transports (instead of the copy-n-paste mess we have today), it's going
> to be more work to disallow this option for raw sockets than it is to
> simply let it work.

Possibly so.  I see nothing wrong with having it work for raw sockets.
Just that it's not really needed, and thus it doesn't matter if it
does or does not work there.  It's only connectionless transport
protocols above IP that need this, and (today) that's just UDP.

-- 
James Carlson, KISS Network                    <[EMAIL PROTECTED]>
Sun Microsystems / 1 Network Drive         71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to