[NETLINK]: Add properly module refcounting for kernel netlink sockets.

- Remove bogus code for compiling netlink as module
- Add module refcounting support for modules implementing a netlink
  protocol
- Add support for autoloading modules that implement a netlink protocol
  as soon as someone opens a socket for that protocol

Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 files changed, 107 insertions, 31 deletions

diff --git a/net/bridge/netfilter/ebt_ulog.c b/net/bridge/netfilter/ebt_ulog.c
index 01af4fcef26d..561d75c8ed5a 100644
--- a/net/bridge/netfilter/ebt_ulog.c
+++ b/net/bridge/netfilter/ebt_ulog.c
@@ -258,7 +258,7 @@ static int __init init(void)
                 spin_lock_init(&ulog_buffers[i].lock);
         }
 
-        ebtulognl = netlink_kernel_create(NETLINK_NFLOG, NULL);
+        ebtulognl = netlink_kernel_create(NETLINK_NFLOG, NULL, THIS_MODULE);
         if (!ebtulognl)
                 ret = -ENOMEM;
         else if ((ret = ebt_register_watcher(&ulog)))
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 4b1bb30e6381..9b3c61f1a37d 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -708,7 +708,7 @@ void __init rtnetlink_init(void)
         if (!rta_buf)
                 panic("rtnetlink_init: cannot allocate rta_buf\n");
 
-        rtnl = netlink_kernel_create(NETLINK_ROUTE, rtnetlink_rcv);
+        rtnl = netlink_kernel_create(NETLINK_ROUTE, rtnetlink_rcv, THIS_MODULE);
         if (rtnl == NULL)
                 panic("rtnetlink_init: cannot initialize rtnetlink\n");
         netlink_set_nonroot(NETLINK_ROUTE, NL_NONROOT_RECV);
diff --git a/net/decnet/netfilter/dn_rtmsg.c b/net/decnet/netfilter/dn_rtmsg.c
index 284a9998e53d..3068fddb2da3 100644
--- a/net/decnet/netfilter/dn_rtmsg.c
+++ b/net/decnet/netfilter/dn_rtmsg.c
@@ -138,7 +138,8 @@ static int __init init(void)
 {
         int rv = 0;
 
-        dnrmg = netlink_kernel_create(NETLINK_DNRTMSG, dnrmg_receive_user_sk);
+        dnrmg = netlink_kernel_create(NETLINK_DNRTMSG, dnrmg_receive_user_sk,
+                                      THIS_MODULE);
         if (dnrmg == NULL) {
                 printk(KERN_ERR "dn_rtmsg: Cannot create netlink socket");
                 return -ENOMEM;
@@ -162,6 +163,7 @@ static void __exit fini(void)
 MODULE_DESCRIPTION("DECnet Routing Message Grabulator");
 MODULE_AUTHOR("Steven Whitehouse <steve@chygwyn.com>");
 MODULE_LICENSE("GPL");
+MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_DNRTMSG);
 
 module_init(init);
 module_exit(fini);
diff --git a/net/ipv4/fib_frontend.c b/net/ipv4/fib_frontend.c
index e5722084239b..b5e2f1550c91 100644
--- a/net/ipv4/fib_frontend.c
+++ b/net/ipv4/fib_frontend.c
@@ -567,7 +567,7 @@ static void nl_fib_input(struct sock *sk, int len)
 
 static void nl_fib_lookup_init(void)
 {
-      netlink_kernel_create(NETLINK_FIB_LOOKUP, nl_fib_input);
+      netlink_kernel_create(NETLINK_FIB_LOOKUP, nl_fib_input, THIS_MODULE);
 }
 
 static void fib_disable_ip(struct net_device *dev, int force)
diff --git a/net/ipv4/netfilter/ip_queue.c b/net/ipv4/netfilter/ip_queue.c
index ae975ac59c6a..b237f7fcad92 100644
--- a/net/ipv4/netfilter/ip_queue.c
+++ b/net/ipv4/netfilter/ip_queue.c
@@ -692,7 +692,8 @@ init_or_cleanup(int init)
                 goto cleanup;
 
         netlink_register_notifier(&ipq_nl_notifier);
-        ipqnl = netlink_kernel_create(NETLINK_FIREWALL, ipq_rcv_sk);
+        ipqnl = netlink_kernel_create(NETLINK_FIREWALL, ipq_rcv_sk,
+                                      THIS_MODULE);
         if (ipqnl == NULL) {
                 printk(KERN_ERR "ip_queue: failed to create netlink socket\n");
                 goto cleanup_netlink_notifier;
diff --git a/net/ipv4/netfilter/ipt_ULOG.c b/net/ipv4/netfilter/ipt_ULOG.c
index 52a0076302a7..4ea8371ab270 100644
--- a/net/ipv4/netfilter/ipt_ULOG.c
+++ b/net/ipv4/netfilter/ipt_ULOG.c
@@ -62,6 +62,7 @@
 MODULE_LICENSE("GPL");
 MODULE_AUTHOR("Harald Welte <laforge@gnumonks.org>");
 MODULE_DESCRIPTION("iptables userspace logging module");
+MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_NFLOG);
 
 #define ULOG_NL_EVENT           111             /* Harald's favorite number */
 #define ULOG_MAXNLGROUPS        32              /* numer of nlgroups */
@@ -372,7 +373,7 @@ static int __init init(void)
                 ulog_buffers[i].timer.data = i;
         }
 
-        nflognl = netlink_kernel_create(NETLINK_NFLOG, NULL);
+        nflognl = netlink_kernel_create(NETLINK_NFLOG, NULL, THIS_MODULE);
         if (!nflognl)
                 return -ENOMEM;
 
diff --git a/net/ipv4/tcp_diag.c b/net/ipv4/tcp_diag.c
index f66945cb158f..f79bd11a4701 100644
--- a/net/ipv4/tcp_diag.c
+++ b/net/ipv4/tcp_diag.c
@@ -774,7 +774,8 @@ static void tcpdiag_rcv(struct sock *sk, int len)
 
 static int __init tcpdiag_init(void)
 {
-        tcpnl = netlink_kernel_create(NETLINK_TCPDIAG, tcpdiag_rcv);
+        tcpnl = netlink_kernel_create(NETLINK_TCPDIAG, tcpdiag_rcv,
+                                      THIS_MODULE);
         if (tcpnl == NULL)
                 return -ENOMEM;
         return 0;
diff --git a/net/ipv6/netfilter/ip6_queue.c b/net/ipv6/netfilter/ip6_queue.c
index 7130603a32c5..1c3d247a22cc 100644
--- a/net/ipv6/netfilter/ip6_queue.c
+++ b/net/ipv6/netfilter/ip6_queue.c
@@ -685,7 +685,7 @@ init_or_cleanup(int init)
                 goto cleanup;
 
         netlink_register_notifier(&ipq_nl_notifier);
-        ipqnl = netlink_kernel_create(NETLINK_IP6_FW, ipq_rcv_sk);
+        ipqnl = netlink_kernel_create(NETLINK_IP6_FW, ipq_rcv_sk, THIS_MODULE);
         if (ipqnl == NULL) {
                 printk(KERN_ERR "ip6_queue: failed to create netlink socket\n");
                 goto cleanup_netlink_notifier;
diff --git a/net/netfilter/nfnetlink.c b/net/netfilter/nfnetlink.c
index b0ed57981847..6210ca42166c 100644
--- a/net/netfilter/nfnetlink.c
+++ b/net/netfilter/nfnetlink.c
@@ -38,6 +38,8 @@
 #include <linux/netfilter/nfnetlink.h>
 
 MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Harald Welte <laforge@netfilter.org>");
+MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_NETFILTER);
 
 static char __initdata nfversion[] = "0.30";
 
@@ -324,7 +326,8 @@ int __init nfnetlink_init(void)
 {
         printk("Netfilter messages via NETLINK v%s.\n", nfversion);
 
-        nfnl = netlink_kernel_create(NETLINK_NETFILTER, nfnetlink_rcv);
+        nfnl = netlink_kernel_create(NETLINK_NETFILTER, nfnetlink_rcv,
+                                     THIS_MODULE);
         if (!nfnl) {
                 printk(KERN_ERR "cannot initialize nfnetlink!\n");
                 return -1;
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index ff774a06c89d..5d487cd69c8c 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -13,7 +13,12 @@
  *                               added netlink_proto_exit
  * Tue Jan 22 18:32:44 BRST 2002 Arnaldo C. de Melo <acme@conectiva.com.br>
  *                               use nlk_sk, as sk->protinfo is on a diet 8)
- *
+ * Fri Jul 22 19:51:12 MEST 2005 Harald Welte <laforge@gnumonks.org>
+ *                               - inc module use count of module that owns
+ *                                 the kernel socket in case userspace opens
+ *                                 socket of same protocol
+ *                               - remove all module support, since netlink is
+ *                                 mandatory if CONFIG_NET=y these days
  */
 
 #include <linux/config.h>
@@ -92,6 +97,7 @@ struct netlink_table {
         struct nl_pid_hash hash;
         struct hlist_head mc_list;
         unsigned int nl_nonroot;
+        struct proto_ops *p_ops;
 };
 
 static struct netlink_table *nl_table;
@@ -341,7 +347,21 @@ static int netlink_create(struct socket *sock, int protocol)
         if (protocol<0 || protocol >= MAX_LINKS)
                 return -EPROTONOSUPPORT;
 
-        sock->ops = &netlink_ops;
+        netlink_table_grab();
+        if (!nl_table[protocol].hash.entries) {
+#ifdef CONFIG_KMOD
+                /* We do 'best effort'.  If we find a matching module,
+                 * it is loaded.  If not, we don't return an error to
+                 * allow pure userspace<->userspace communication. -HW
+                 */
+                netlink_table_ungrab();
+                request_module("net-pf-%d-proto-%d", PF_NETLINK, protocol);
+                netlink_table_grab();
+#endif
+        }
+        netlink_table_ungrab();
+
+        sock->ops = nl_table[protocol].p_ops;
 
         sk = sk_alloc(PF_NETLINK, GFP_KERNEL, &netlink_proto, 1);
         if (!sk)
@@ -394,6 +414,22 @@ static int netlink_release(struct socket *sock)
                                           };
                 notifier_call_chain(&netlink_chain, NETLINK_URELEASE, &n);
         }       
+
+        /* When this is a kernel socket, we need to remove the owner pointer,
+         * since we don't know whether the module will be dying at any given
+         * point - HW
+         */
+        if (!nlk->pid) {
+                struct proto_ops *p_tmp;
+
+                netlink_table_grab();
+                p_tmp = nl_table[sk->sk_protocol].p_ops;
+                if (p_tmp != &netlink_ops) {
+                        nl_table[sk->sk_protocol].p_ops = &netlink_ops;
+                        kfree(p_tmp);
+                }
+                netlink_table_ungrab();
+        }
         
         sock_put(sk);
         return 0;
@@ -1023,8 +1059,9 @@ static void netlink_data_ready(struct sock *sk, int len)
  */
 
 struct sock *
-netlink_kernel_create(int unit, void (*input)(struct sock *sk, int len))
+netlink_kernel_create(int unit, void (*input)(struct sock *sk, int len), struct module *module)
 {
+        struct proto_ops *p_ops;
         struct socket *sock;
         struct sock *sk;
 
@@ -1034,22 +1071,63 @@ netlink_kernel_create(int unit, void (*input)(struct sock *sk, int len))
         if (unit<0 || unit>=MAX_LINKS)
                 return NULL;
 
+        /* Do a quick check, to make us not go down to netlink_insert()
+         * if protocol already has kernel socket.
+         */
+        sk = netlink_lookup(unit, 0);
+        if (unlikely(sk)) {
+                sock_put(sk);
+                return NULL;
+        }
+
         if (sock_create_lite(PF_NETLINK, SOCK_DGRAM, unit, &sock))
                 return NULL;
 
+        sk = NULL;
+        if (module) {
+                /* Every registering protocol implemented in a module needs
+                 * it's own p_ops, since the socket code cannot deal with
+                 * module refcounting otherwise.  -HW
+                 */
+                p_ops = kmalloc(sizeof(*p_ops), GFP_KERNEL);
+                if (!p_ops)
+                        goto out_sock_release;
+
+                memcpy(p_ops, &netlink_ops, sizeof(*p_ops));
+                p_ops->owner = module;
+        } else
+                p_ops = &netlink_ops;
+
+        netlink_table_grab();
+        nl_table[unit].p_ops = p_ops;
+        netlink_table_ungrab();
+
         if (netlink_create(sock, unit) < 0) {
-                sock_release(sock);
-                return NULL;
+                sk = NULL;
+                goto out_kfree_p_ops;
         }
+
         sk = sock->sk;
         sk->sk_data_ready = netlink_data_ready;
         if (input)
                 nlk_sk(sk)->data_ready = input;
 
         if (netlink_insert(sk, 0)) {
-                sock_release(sock);
-                return NULL;
+                sk = NULL;
+                goto out_kfree_p_ops;
+        }
+
+        return sk;
+
+out_kfree_p_ops:
+        netlink_table_grab();
+        if (nl_table[unit].p_ops != &netlink_ops) {
+                kfree(nl_table[unit].p_ops);
+                nl_table[unit].p_ops = &netlink_ops;
         }
+        netlink_table_ungrab();
+out_sock_release:
+        sock_release(sock);
         return sk;
 }
 
@@ -1413,6 +1491,8 @@ enomem:
         for (i = 0; i < MAX_LINKS; i++) {
                 struct nl_pid_hash *hash = &nl_table[i].hash;
 
+                nl_table[i].p_ops = &netlink_ops;
+
                 hash->table = nl_pid_hash_alloc(1 * sizeof(*hash->table));
                 if (!hash->table) {
                         while (i-- > 0)
@@ -1438,21 +1518,7 @@ out:
         return err;
 }
 
-static void __exit netlink_proto_exit(void)
-{
-        sock_unregister(PF_NETLINK);
-        proc_net_remove("netlink");
-        kfree(nl_table);
-        nl_table = NULL;
-        proto_unregister(&netlink_proto);
-}
-
 core_initcall(netlink_proto_init);
-module_exit(netlink_proto_exit);
-
-MODULE_LICENSE("GPL");
-
-MODULE_ALIAS_NETPROTO(PF_NETLINK);
 
 EXPORT_SYMBOL(netlink_ack);
 EXPORT_SYMBOL(netlink_broadcast);
diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
index 8da3e25b2c4c..33ceeea783b1 100644
--- a/net/xfrm/xfrm_user.c
+++ b/net/xfrm/xfrm_user.c
@@ -1519,7 +1519,8 @@ static int __init xfrm_user_init(void)
 {
         printk(KERN_INFO "Initializing IPsec netlink socket\n");
 
-        xfrm_nl = netlink_kernel_create(NETLINK_XFRM, xfrm_netlink_rcv);
+        xfrm_nl = netlink_kernel_create(NETLINK_XFRM, xfrm_netlink_rcv,
+                                        THIS_MODULE);
         if (xfrm_nl == NULL)
                 return -ENOMEM;
 
@@ -1537,3 +1538,4 @@ static void __exit xfrm_user_exit(void)
 module_init(xfrm_user_init);
 module_exit(xfrm_user_exit);
 MODULE_LICENSE("GPL");
+MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_XFRM);