Re: [patch 7/8] fdmap v2 - implement sys_socket2

[Ulrich Drepper <drepper@redhat.com>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David Miller wrote:
> Since the valid range of "domain" values is quite small,
> we could avoid the new system call by cribbing some of the
> upper bits of the 'domain' argument.
> 
> Valid existing programs pass in valid 'domain' values and
> thus will not set any of the new flags.

I can see several problems with that:

- - experimental implementers might choose domain values which definitely
won't collide with others

- - the flags parameter ideally allows using the same values used for
open's mode argument.  The lowest value I can see making sense is
O_NONBLOCK (04000).

- - how to recognize kernels without the support?  -EAFNOSUPPORT can also
with new kernels mean it's actually the domain which is wrong

- - there might be new flags we want to use over time

I would strongly argue that any change we're doing in this area at
userlevel would involve a new interface.  Programs also need new
definitions from headers files.  This means a recent enough glibc will
be needed in any case.  Unless programs use their own definitions in
which case they might as well use the syscall() function.

- --
➧ Ulrich Drepper ➧ Red Hat, Inc. ➧ 444 Castro St ➧ Mountain View, CA ❖
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)

iD8DBQFGZzvl2ijCOnn/RHQRAgV0AKDBDhqSQ/cs4qGYLKGL4dwzpFZ2zgCgl/qO
oFKnQ2eRuiziRu/N5vwWCeM=
=tttP
-----END PGP SIGNATURE-----