On Thu, Apr 21, 2005 at 03:52:43PM -0700, Tom Duffy wrote:
> On Thu, 2005-04-21 at 15:36 -0700, Roland Dreier wrote:
> > Is this really the only change required? It seems that the socket
> > allocation function changed too -- 2.6.11 has
> >
> > extern struct sock * sk_alloc(int family, int priority, int
> > zero_it,
> > kmem_cache_t *slab);
> >
> > while my up-to-date Linus tree has
> >
> > extern struct sock *sk_alloc(int family, int priority,
> > struct proto *prot, int
> > zero_it);
> >
> > so I think sdp_conn.c at least needs some fixing up.
>
> Oh, you are right, I missed the compile warning, but I see it now.
>
> Why does the SDP code pass in sizeof(struct inet_sock) for the zero_it
> bool?
Is this a trick question? :) Because it's not a bool but an integer,
which use to be a bool in the 2.4 kernel days. Here's the relevant
code snip from net/core/sock.c:
if (zero_it) {
memset(sk, 0,
zero_it == 1 ? sizeof(struct sock) : zero_it);
sk->sk_family = family;
sock_lock_init(sk);
}
-Libor
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
