On 02/03/2017 02:51 AM, Eric Dumazet wrote:
On Fri, 2017-02-03 at 10:18 +0900, Lorenzo Colitti wrote:
On Fri, Feb 3, 2017 at 9:31 AM, Eric Dumazet <eric.duma...@gmail.com> wrote:
It should be safe to call sock_net_uid on any type of socket
(including NULL). sk_uid was added to struct sock in 86741ec25462
("net: core: Add a UID field to struct sock.")
But a request socket or a timewait socket do not have this field.
Daniel point is valid.
My bad. Yes.
It would definitely be useful to have the UID available in request
sockets, and perhaps timewait sockets as well. That could be done by
moving the UID to sock_common, or with something along the lines of:
static inline kuid_t sock_net_uid(const struct net *net, const struct sock
*sk)
{
+ if (sk && sk->sk_state == TCP_NEW_SYN_RECV)
+ sk = sk->__sk_common.skc_listener;
+ else if (sk && !sk_fullsock(sk))
+ sk = NULL;
+
return sk ? sk->sk_uid : make_kuid(net->user_ns, 0);
}
Any thoughts on which is better?
You could use
if (sk) {
sk = sk_to_full_sk(sk);
if (sk_fullsock(sk))
return sk->sk_uid;
}
Yeah, if that moves into the sock_net_uid() helper, then you could
remove the sk && sk_fullsock(sk) ? sk : NULL tests from the current
sock_net_uid() call sites such as in tcp code. Maybe then also make
the sock_net_uid() as __always_inline, so that most of the callers
with sock_net_uid(net, NULL) are guaranteed to optimize away their
sk checks at compile time?
