net: Move IP_ROUTER_ALERT out of lock_sock(sk)
authorKirill Tkhai <ktkhai@virtuozzo.com>
Thu, 22 Mar 2018 09:45:12 +0000 (12:45 +0300)
committerDavid S. Miller <davem@davemloft.net>
Thu, 22 Mar 2018 19:12:55 +0000 (15:12 -0400)
ip_ra_control() does not need sk_lock. Who are the another
users of ip_ra_chain? ip_mroute_setsockopt() doesn't take
sk_lock, while parallel IP_ROUTER_ALERT syscalls are
synchronized by ip_ra_lock. So, we may move this command
out of sk_lock.

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/ip_sockglue.c

index be7c3b71914de1e83f6a854e25c04b951aa1aed2..dcbf6afe27e7266788ea6351f7c607c19c60788f 100644 (file)
@@ -647,6 +647,8 @@ static int do_ip_setsockopt(struct sock *sk, int level,
 
        /* If optlen==0, it is equivalent to val == 0 */
 
+       if (optname == IP_ROUTER_ALERT)
+               return ip_ra_control(sk, val ? 1 : 0, NULL);
        if (ip_mroute_opt(optname))
                return ip_mroute_setsockopt(sk, optname, optval, optlen);
 
@@ -1157,9 +1159,6 @@ mc_msf_out:
                        goto e_inval;
                inet->mc_all = val;
                break;
-       case IP_ROUTER_ALERT:
-               err = ip_ra_control(sk, val ? 1 : 0, NULL);
-               break;
 
        case IP_FREEBIND:
                if (optlen < 1)