aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
Diffstat (limited to 'net')
-rw-r--r--net/9p/client.c2
-rw-r--r--net/9p/trans_virtio.c5
-rw-r--r--net/batman-adv/bat_iv_ogm.c30
-rw-r--r--net/batman-adv/hard-interface.c22
-rw-r--r--net/batman-adv/originator.c36
-rw-r--r--net/batman-adv/originator.h4
-rw-r--r--net/batman-adv/routing.c4
-rw-r--r--net/batman-adv/send.c9
-rw-r--r--net/batman-adv/translation-table.c23
-rw-r--r--net/bluetooth/hidp/core.c16
-rw-r--r--net/bluetooth/hidp/hidp.h4
-rw-r--r--net/bridge/br_device.c54
-rw-r--r--net/bridge/br_fdb.c137
-rw-r--r--net/bridge/br_if.c6
-rw-r--r--net/bridge/br_input.c4
-rw-r--r--net/bridge/br_private.h13
-rw-r--r--net/bridge/br_stp_if.c2
-rw-r--r--net/bridge/br_vlan.c27
-rw-r--r--net/caif/caif_dev.c1
-rw-r--r--net/caif/cfsrvl.c1
-rw-r--r--net/can/af_can.c3
-rw-r--r--net/can/bcm.c4
-rw-r--r--net/can/raw.c27
-rw-r--r--net/ceph/messenger.c8
-rw-r--r--net/ceph/osd_client.c84
-rw-r--r--net/core/dev.c28
-rw-r--r--net/core/fib_rules.c7
-rw-r--r--net/core/flow_dissector.c20
-rw-r--r--net/core/neighbour.c8
-rw-r--r--net/core/netpoll.c4
-rw-r--r--net/core/rtnetlink.c21
-rw-r--r--net/core/skbuff.c3
-rw-r--r--net/core/sock.c6
-rw-r--r--net/dccp/ccids/lib/tfrc.c2
-rw-r--r--net/dccp/ccids/lib/tfrc.h1
-rw-r--r--net/decnet/af_decnet.c5
-rw-r--r--net/hsr/hsr_framereg.c2
-rw-r--r--net/ieee802154/6lowpan.c23
-rw-r--r--net/ipv4/af_inet.c7
-rw-r--r--net/ipv4/devinet.c3
-rw-r--r--net/ipv4/ip_forward.c71
-rw-r--r--net/ipv4/ip_output.c3
-rw-r--r--net/ipv4/ip_tunnel.c82
-rw-r--r--net/ipv4/ip_tunnel_core.c47
-rw-r--r--net/ipv4/ipconfig.c2
-rw-r--r--net/ipv4/netfilter/Kconfig5
-rw-r--r--net/ipv4/netfilter/Makefile1
-rw-r--r--net/ipv4/netfilter/nf_nat_h323.c5
-rw-r--r--net/ipv4/netfilter/nf_nat_snmp_basic.c4
-rw-r--r--net/ipv4/netfilter/nft_reject_ipv4.c75
-rw-r--r--net/ipv4/route.c13
-rw-r--r--net/ipv4/tcp.c10
-rw-r--r--net/ipv4/tcp_cong.c3
-rw-r--r--net/ipv4/tcp_input.c21
-rw-r--r--net/ipv4/tcp_output.c37
-rw-r--r--net/ipv4/udp_offload.c17
-rw-r--r--net/ipv6/Kconfig1
-rw-r--r--net/ipv6/addrconf.c2
-rw-r--r--net/ipv6/exthdrs_core.c2
-rw-r--r--net/ipv6/icmp.c2
-rw-r--r--net/ipv6/ip6_offload.c20
-rw-r--r--net/ipv6/ip6_output.c20
-rw-r--r--net/ipv6/netfilter/Kconfig5
-rw-r--r--net/ipv6/netfilter/Makefile1
-rw-r--r--net/ipv6/netfilter/nft_reject_ipv6.c76
-rw-r--r--net/ipv6/ping.c1
-rw-r--r--net/ipv6/sit.c19
-rw-r--r--net/ipv6/udp_offload.c2
-rw-r--r--net/ipx/af_ipx.c22
-rw-r--r--net/ipx/ipx_route.c4
-rw-r--r--net/mac80211/cfg.c44
-rw-r--r--net/mac80211/ht.c4
-rw-r--r--net/mac80211/ibss.c5
-rw-r--r--net/mac80211/ieee80211_i.h10
-rw-r--r--net/mac80211/iface.c33
-rw-r--r--net/mac80211/mlme.c24
-rw-r--r--net/mac80211/rx.c7
-rw-r--r--net/mac80211/sta_info.c66
-rw-r--r--net/mac80211/sta_info.h7
-rw-r--r--net/mac80211/tx.c17
-rw-r--r--net/mac80211/util.c48
-rw-r--r--net/mac80211/wme.c5
-rw-r--r--net/netfilter/Kconfig6
-rw-r--r--net/netfilter/Makefile1
-rw-r--r--net/netfilter/ipvs/ip_vs_conn.c8
-rw-r--r--net/netfilter/nf_conntrack_core.c55
-rw-r--r--net/netfilter/nf_conntrack_netlink.c35
-rw-r--r--net/netfilter/nf_nat_core.c56
-rw-r--r--net/netfilter/nf_synproxy_core.c5
-rw-r--r--net/netfilter/nf_tables_api.c82
-rw-r--r--net/netfilter/nf_tables_core.c6
-rw-r--r--net/netfilter/nft_ct.c16
-rw-r--r--net/netfilter/nft_log.c5
-rw-r--r--net/netfilter/nft_lookup.c1
-rw-r--r--net/netfilter/nft_meta.c4
-rw-r--r--net/netfilter/nft_payload.c3
-rw-r--r--net/netfilter/nft_queue.c4
-rw-r--r--net/netfilter/nft_rbtree.c16
-rw-r--r--net/netfilter/nft_reject.c89
-rw-r--r--net/netfilter/nft_reject_inet.c63
-rw-r--r--net/netfilter/xt_CT.c7
-rw-r--r--net/netlink/af_netlink.c4
-rw-r--r--net/nfc/nci/core.c2
-rw-r--r--net/openvswitch/datapath.c23
-rw-r--r--net/openvswitch/flow_table.c88
-rw-r--r--net/openvswitch/flow_table.h2
-rw-r--r--net/packet/af_packet.c26
-rw-r--r--net/sched/sch_pie.c21
-rw-r--r--net/sched/sch_tbf.c24
-rw-r--r--net/sctp/associola.c211
-rw-r--r--net/sctp/ipv6.c2
-rw-r--r--net/sctp/sm_sideeffect.c7
-rw-r--r--net/sctp/sm_statefuns.c9
-rw-r--r--net/sctp/socket.c47
-rw-r--r--net/sctp/sysctl.c18
-rw-r--r--net/sctp/ulpevent.c8
-rw-r--r--net/sunrpc/auth_gss/auth_gss.c19
-rw-r--r--net/sunrpc/backchannel_rqst.c6
-rw-r--r--net/sunrpc/svc_xprt.c6
-rw-r--r--net/sunrpc/xprtsock.c6
-rw-r--r--net/tipc/bearer.c7
-rw-r--r--net/tipc/config.c2
-rw-r--r--net/tipc/core.c109
-rw-r--r--net/tipc/core.h2
-rw-r--r--net/tipc/link.c7
-rw-r--r--net/tipc/name_table.c3
-rw-r--r--net/tipc/netlink.c8
-rw-r--r--net/tipc/ref.c3
-rw-r--r--net/tipc/server.c5
-rw-r--r--net/tipc/server.h2
-rw-r--r--net/tipc/socket.c8
-rw-r--r--net/wireless/core.c17
-rw-r--r--net/wireless/core.h4
-rw-r--r--net/wireless/nl80211.c32
-rw-r--r--net/wireless/nl80211.h8
-rw-r--r--net/wireless/reg.c12
-rw-r--r--net/wireless/scan.c40
-rw-r--r--net/wireless/sme.c2
-rw-r--r--net/xfrm/xfrm_policy.c2
-rw-r--r--net/xfrm/xfrm_state.c23
-rw-r--r--net/xfrm/xfrm_user.c5
141 files changed, 1749 insertions, 1052 deletions
diff --git a/net/9p/client.c b/net/9p/client.c
index a5e4d2dcb03e..9186550d77a6 100644
--- a/net/9p/client.c
+++ b/net/9p/client.c
@@ -204,7 +204,7 @@ free_and_return:
204 return ret; 204 return ret;
205} 205}
206 206
207struct p9_fcall *p9_fcall_alloc(int alloc_msize) 207static struct p9_fcall *p9_fcall_alloc(int alloc_msize)
208{ 208{
209 struct p9_fcall *fc; 209 struct p9_fcall *fc;
210 fc = kmalloc(sizeof(struct p9_fcall) + alloc_msize, GFP_NOFS); 210 fc = kmalloc(sizeof(struct p9_fcall) + alloc_msize, GFP_NOFS);
diff --git a/net/9p/trans_virtio.c b/net/9p/trans_virtio.c
index cd1e1ede73a4..ac2666c1d011 100644
--- a/net/9p/trans_virtio.c
+++ b/net/9p/trans_virtio.c
@@ -340,7 +340,10 @@ static int p9_get_mapped_pages(struct virtio_chan *chan,
340 int count = nr_pages; 340 int count = nr_pages;
341 while (nr_pages) { 341 while (nr_pages) {
342 s = rest_of_page(data); 342 s = rest_of_page(data);
343 pages[index++] = kmap_to_page(data); 343 if (is_vmalloc_addr(data))
344 pages[index++] = vmalloc_to_page(data);
345 else
346 pages[index++] = kmap_to_page(data);
344 data += s; 347 data += s;
345 nr_pages--; 348 nr_pages--;
346 } 349 }
diff --git a/net/batman-adv/bat_iv_ogm.c b/net/batman-adv/bat_iv_ogm.c
index 512159bf607f..8323bced8e5b 100644
--- a/net/batman-adv/bat_iv_ogm.c
+++ b/net/batman-adv/bat_iv_ogm.c
@@ -241,19 +241,19 @@ batadv_iv_ogm_orig_get(struct batadv_priv *bat_priv, const uint8_t *addr)
241 size = bat_priv->num_ifaces * sizeof(uint8_t); 241 size = bat_priv->num_ifaces * sizeof(uint8_t);
242 orig_node->bat_iv.bcast_own_sum = kzalloc(size, GFP_ATOMIC); 242 orig_node->bat_iv.bcast_own_sum = kzalloc(size, GFP_ATOMIC);
243 if (!orig_node->bat_iv.bcast_own_sum) 243 if (!orig_node->bat_iv.bcast_own_sum)
244 goto free_bcast_own; 244 goto free_orig_node;
245 245
246 hash_added = batadv_hash_add(bat_priv->orig_hash, batadv_compare_orig, 246 hash_added = batadv_hash_add(bat_priv->orig_hash, batadv_compare_orig,
247 batadv_choose_orig, orig_node, 247 batadv_choose_orig, orig_node,
248 &orig_node->hash_entry); 248 &orig_node->hash_entry);
249 if (hash_added != 0) 249 if (hash_added != 0)
250 goto free_bcast_own; 250 goto free_orig_node;
251 251
252 return orig_node; 252 return orig_node;
253 253
254free_bcast_own:
255 kfree(orig_node->bat_iv.bcast_own);
256free_orig_node: 254free_orig_node:
255 /* free twice, as batadv_orig_node_new sets refcount to 2 */
256 batadv_orig_node_free_ref(orig_node);
257 batadv_orig_node_free_ref(orig_node); 257 batadv_orig_node_free_ref(orig_node);
258 258
259 return NULL; 259 return NULL;
@@ -266,7 +266,7 @@ batadv_iv_ogm_neigh_new(struct batadv_hard_iface *hard_iface,
266 struct batadv_orig_node *orig_neigh) 266 struct batadv_orig_node *orig_neigh)
267{ 267{
268 struct batadv_priv *bat_priv = netdev_priv(hard_iface->soft_iface); 268 struct batadv_priv *bat_priv = netdev_priv(hard_iface->soft_iface);
269 struct batadv_neigh_node *neigh_node; 269 struct batadv_neigh_node *neigh_node, *tmp_neigh_node;
270 270
271 neigh_node = batadv_neigh_node_new(hard_iface, neigh_addr, orig_node); 271 neigh_node = batadv_neigh_node_new(hard_iface, neigh_addr, orig_node);
272 if (!neigh_node) 272 if (!neigh_node)
@@ -281,14 +281,24 @@ batadv_iv_ogm_neigh_new(struct batadv_hard_iface *hard_iface,
281 neigh_node->orig_node = orig_neigh; 281 neigh_node->orig_node = orig_neigh;
282 neigh_node->if_incoming = hard_iface; 282 neigh_node->if_incoming = hard_iface;
283 283
284 batadv_dbg(BATADV_DBG_BATMAN, bat_priv,
285 "Creating new neighbor %pM for orig_node %pM on interface %s\n",
286 neigh_addr, orig_node->orig, hard_iface->net_dev->name);
287
288 spin_lock_bh(&orig_node->neigh_list_lock); 284 spin_lock_bh(&orig_node->neigh_list_lock);
289 hlist_add_head_rcu(&neigh_node->list, &orig_node->neigh_list); 285 tmp_neigh_node = batadv_neigh_node_get(orig_node, hard_iface,
286 neigh_addr);
287 if (!tmp_neigh_node) {
288 hlist_add_head_rcu(&neigh_node->list, &orig_node->neigh_list);
289 } else {
290 kfree(neigh_node);
291 batadv_hardif_free_ref(hard_iface);
292 neigh_node = tmp_neigh_node;
293 }
290 spin_unlock_bh(&orig_node->neigh_list_lock); 294 spin_unlock_bh(&orig_node->neigh_list_lock);
291 295
296 if (!tmp_neigh_node)
297 batadv_dbg(BATADV_DBG_BATMAN, bat_priv,
298 "Creating new neighbor %pM for orig_node %pM on interface %s\n",
299 neigh_addr, orig_node->orig,
300 hard_iface->net_dev->name);
301
292out: 302out:
293 return neigh_node; 303 return neigh_node;
294} 304}
diff --git a/net/batman-adv/hard-interface.c b/net/batman-adv/hard-interface.c
index 3d417d3641c6..b851cc580853 100644
--- a/net/batman-adv/hard-interface.c
+++ b/net/batman-adv/hard-interface.c
@@ -241,7 +241,7 @@ int batadv_hardif_min_mtu(struct net_device *soft_iface)
241{ 241{
242 struct batadv_priv *bat_priv = netdev_priv(soft_iface); 242 struct batadv_priv *bat_priv = netdev_priv(soft_iface);
243 const struct batadv_hard_iface *hard_iface; 243 const struct batadv_hard_iface *hard_iface;
244 int min_mtu = ETH_DATA_LEN; 244 int min_mtu = INT_MAX;
245 245
246 rcu_read_lock(); 246 rcu_read_lock();
247 list_for_each_entry_rcu(hard_iface, &batadv_hardif_list, list) { 247 list_for_each_entry_rcu(hard_iface, &batadv_hardif_list, list) {
@@ -256,8 +256,6 @@ int batadv_hardif_min_mtu(struct net_device *soft_iface)
256 } 256 }
257 rcu_read_unlock(); 257 rcu_read_unlock();
258 258
259 atomic_set(&bat_priv->packet_size_max, min_mtu);
260
261 if (atomic_read(&bat_priv->fragmentation) == 0) 259 if (atomic_read(&bat_priv->fragmentation) == 0)
262 goto out; 260 goto out;
263 261
@@ -268,13 +266,21 @@ int batadv_hardif_min_mtu(struct net_device *soft_iface)
268 min_mtu = min_t(int, min_mtu, BATADV_FRAG_MAX_FRAG_SIZE); 266 min_mtu = min_t(int, min_mtu, BATADV_FRAG_MAX_FRAG_SIZE);
269 min_mtu -= sizeof(struct batadv_frag_packet); 267 min_mtu -= sizeof(struct batadv_frag_packet);
270 min_mtu *= BATADV_FRAG_MAX_FRAGMENTS; 268 min_mtu *= BATADV_FRAG_MAX_FRAGMENTS;
271 atomic_set(&bat_priv->packet_size_max, min_mtu);
272
273 /* with fragmentation enabled we can fragment external packets easily */
274 min_mtu = min_t(int, min_mtu, ETH_DATA_LEN);
275 269
276out: 270out:
277 return min_mtu - batadv_max_header_len(); 271 /* report to the other components the maximum amount of bytes that
272 * batman-adv can send over the wire (without considering the payload
273 * overhead). For example, this value is used by TT to compute the
274 * maximum local table table size
275 */
276 atomic_set(&bat_priv->packet_size_max, min_mtu);
277
278 /* the real soft-interface MTU is computed by removing the payload
279 * overhead from the maximum amount of bytes that was just computed.
280 *
281 * However batman-adv does not support MTUs bigger than ETH_DATA_LEN
282 */
283 return min_t(int, min_mtu - batadv_max_header_len(), ETH_DATA_LEN);
278} 284}
279 285
280/* adjusts the MTU if a new interface with a smaller MTU appeared. */ 286/* adjusts the MTU if a new interface with a smaller MTU appeared. */
diff --git a/net/batman-adv/originator.c b/net/batman-adv/originator.c
index 6df12a2e3605..853941629dc1 100644
--- a/net/batman-adv/originator.c
+++ b/net/batman-adv/originator.c
@@ -458,6 +458,42 @@ out:
458} 458}
459 459
460/** 460/**
461 * batadv_neigh_node_get - retrieve a neighbour from the list
462 * @orig_node: originator which the neighbour belongs to
463 * @hard_iface: the interface where this neighbour is connected to
464 * @addr: the address of the neighbour
465 *
466 * Looks for and possibly returns a neighbour belonging to this originator list
467 * which is connected through the provided hard interface.
468 * Returns NULL if the neighbour is not found.
469 */
470struct batadv_neigh_node *
471batadv_neigh_node_get(const struct batadv_orig_node *orig_node,
472 const struct batadv_hard_iface *hard_iface,
473 const uint8_t *addr)
474{
475 struct batadv_neigh_node *tmp_neigh_node, *res = NULL;
476
477 rcu_read_lock();
478 hlist_for_each_entry_rcu(tmp_neigh_node, &orig_node->neigh_list, list) {
479 if (!batadv_compare_eth(tmp_neigh_node->addr, addr))
480 continue;
481
482 if (tmp_neigh_node->if_incoming != hard_iface)
483 continue;
484
485 if (!atomic_inc_not_zero(&tmp_neigh_node->refcount))
486 continue;
487
488 res = tmp_neigh_node;
489 break;
490 }
491 rcu_read_unlock();
492
493 return res;
494}
495
496/**
461 * batadv_orig_ifinfo_free_rcu - free the orig_ifinfo object 497 * batadv_orig_ifinfo_free_rcu - free the orig_ifinfo object
462 * @rcu: rcu pointer of the orig_ifinfo object 498 * @rcu: rcu pointer of the orig_ifinfo object
463 */ 499 */
diff --git a/net/batman-adv/originator.h b/net/batman-adv/originator.h
index 37be290f63f6..db3a9ed734cb 100644
--- a/net/batman-adv/originator.h
+++ b/net/batman-adv/originator.h
@@ -29,6 +29,10 @@ void batadv_orig_node_free_ref_now(struct batadv_orig_node *orig_node);
29struct batadv_orig_node *batadv_orig_node_new(struct batadv_priv *bat_priv, 29struct batadv_orig_node *batadv_orig_node_new(struct batadv_priv *bat_priv,
30 const uint8_t *addr); 30 const uint8_t *addr);
31struct batadv_neigh_node * 31struct batadv_neigh_node *
32batadv_neigh_node_get(const struct batadv_orig_node *orig_node,
33 const struct batadv_hard_iface *hard_iface,
34 const uint8_t *addr);
35struct batadv_neigh_node *
32batadv_neigh_node_new(struct batadv_hard_iface *hard_iface, 36batadv_neigh_node_new(struct batadv_hard_iface *hard_iface,
33 const uint8_t *neigh_addr, 37 const uint8_t *neigh_addr,
34 struct batadv_orig_node *orig_node); 38 struct batadv_orig_node *orig_node);
diff --git a/net/batman-adv/routing.c b/net/batman-adv/routing.c
index 1ed9f7c9ecea..a953d5b196a3 100644
--- a/net/batman-adv/routing.c
+++ b/net/batman-adv/routing.c
@@ -688,7 +688,7 @@ static int batadv_check_unicast_ttvn(struct batadv_priv *bat_priv,
688 int is_old_ttvn; 688 int is_old_ttvn;
689 689
690 /* check if there is enough data before accessing it */ 690 /* check if there is enough data before accessing it */
691 if (pskb_may_pull(skb, hdr_len + ETH_HLEN) < 0) 691 if (!pskb_may_pull(skb, hdr_len + ETH_HLEN))
692 return 0; 692 return 0;
693 693
694 /* create a copy of the skb (in case of for re-routing) to modify it. */ 694 /* create a copy of the skb (in case of for re-routing) to modify it. */
@@ -918,6 +918,8 @@ int batadv_recv_unicast_tvlv(struct sk_buff *skb,
918 918
919 if (ret != NET_RX_SUCCESS) 919 if (ret != NET_RX_SUCCESS)
920 ret = batadv_route_unicast_packet(skb, recv_if); 920 ret = batadv_route_unicast_packet(skb, recv_if);
921 else
922 consume_skb(skb);
921 923
922 return ret; 924 return ret;
923} 925}
diff --git a/net/batman-adv/send.c b/net/batman-adv/send.c
index 579f5f00a385..843febd1e519 100644
--- a/net/batman-adv/send.c
+++ b/net/batman-adv/send.c
@@ -254,9 +254,9 @@ static int batadv_send_skb_unicast(struct batadv_priv *bat_priv,
254 struct batadv_orig_node *orig_node, 254 struct batadv_orig_node *orig_node,
255 unsigned short vid) 255 unsigned short vid)
256{ 256{
257 struct ethhdr *ethhdr = (struct ethhdr *)skb->data; 257 struct ethhdr *ethhdr;
258 struct batadv_unicast_packet *unicast_packet; 258 struct batadv_unicast_packet *unicast_packet;
259 int ret = NET_XMIT_DROP; 259 int ret = NET_XMIT_DROP, hdr_size;
260 260
261 if (!orig_node) 261 if (!orig_node)
262 goto out; 262 goto out;
@@ -265,12 +265,16 @@ static int batadv_send_skb_unicast(struct batadv_priv *bat_priv,
265 case BATADV_UNICAST: 265 case BATADV_UNICAST:
266 if (!batadv_send_skb_prepare_unicast(skb, orig_node)) 266 if (!batadv_send_skb_prepare_unicast(skb, orig_node))
267 goto out; 267 goto out;
268
269 hdr_size = sizeof(*unicast_packet);
268 break; 270 break;
269 case BATADV_UNICAST_4ADDR: 271 case BATADV_UNICAST_4ADDR:
270 if (!batadv_send_skb_prepare_unicast_4addr(bat_priv, skb, 272 if (!batadv_send_skb_prepare_unicast_4addr(bat_priv, skb,
271 orig_node, 273 orig_node,
272 packet_subtype)) 274 packet_subtype))
273 goto out; 275 goto out;
276
277 hdr_size = sizeof(struct batadv_unicast_4addr_packet);
274 break; 278 break;
275 default: 279 default:
276 /* this function supports UNICAST and UNICAST_4ADDR only. It 280 /* this function supports UNICAST and UNICAST_4ADDR only. It
@@ -279,6 +283,7 @@ static int batadv_send_skb_unicast(struct batadv_priv *bat_priv,
279 goto out; 283 goto out;
280 } 284 }
281 285
286 ethhdr = (struct ethhdr *)(skb->data + hdr_size);
282 unicast_packet = (struct batadv_unicast_packet *)skb->data; 287 unicast_packet = (struct batadv_unicast_packet *)skb->data;
283 288
284 /* inform the destination node that we are still missing a correct route 289 /* inform the destination node that we are still missing a correct route
diff --git a/net/batman-adv/translation-table.c b/net/batman-adv/translation-table.c
index b6071f675a3e..959dde721c46 100644
--- a/net/batman-adv/translation-table.c
+++ b/net/batman-adv/translation-table.c
@@ -1975,6 +1975,7 @@ static uint32_t batadv_tt_global_crc(struct batadv_priv *bat_priv,
1975 struct hlist_head *head; 1975 struct hlist_head *head;
1976 uint32_t i, crc_tmp, crc = 0; 1976 uint32_t i, crc_tmp, crc = 0;
1977 uint8_t flags; 1977 uint8_t flags;
1978 __be16 tmp_vid;
1978 1979
1979 for (i = 0; i < hash->size; i++) { 1980 for (i = 0; i < hash->size; i++) {
1980 head = &hash->table[i]; 1981 head = &hash->table[i];
@@ -2011,8 +2012,11 @@ static uint32_t batadv_tt_global_crc(struct batadv_priv *bat_priv,
2011 orig_node)) 2012 orig_node))
2012 continue; 2013 continue;
2013 2014
2014 crc_tmp = crc32c(0, &tt_common->vid, 2015 /* use network order to read the VID: this ensures that
2015 sizeof(tt_common->vid)); 2016 * every node reads the bytes in the same order.
2017 */
2018 tmp_vid = htons(tt_common->vid);
2019 crc_tmp = crc32c(0, &tmp_vid, sizeof(tmp_vid));
2016 2020
2017 /* compute the CRC on flags that have to be kept in sync 2021 /* compute the CRC on flags that have to be kept in sync
2018 * among nodes 2022 * among nodes
@@ -2046,6 +2050,7 @@ static uint32_t batadv_tt_local_crc(struct batadv_priv *bat_priv,
2046 struct hlist_head *head; 2050 struct hlist_head *head;
2047 uint32_t i, crc_tmp, crc = 0; 2051 uint32_t i, crc_tmp, crc = 0;
2048 uint8_t flags; 2052 uint8_t flags;
2053 __be16 tmp_vid;
2049 2054
2050 for (i = 0; i < hash->size; i++) { 2055 for (i = 0; i < hash->size; i++) {
2051 head = &hash->table[i]; 2056 head = &hash->table[i];
@@ -2064,8 +2069,11 @@ static uint32_t batadv_tt_local_crc(struct batadv_priv *bat_priv,
2064 if (tt_common->flags & BATADV_TT_CLIENT_NEW) 2069 if (tt_common->flags & BATADV_TT_CLIENT_NEW)
2065 continue; 2070 continue;
2066 2071
2067 crc_tmp = crc32c(0, &tt_common->vid, 2072 /* use network order to read the VID: this ensures that
2068 sizeof(tt_common->vid)); 2073 * every node reads the bytes in the same order.
2074 */
2075 tmp_vid = htons(tt_common->vid);
2076 crc_tmp = crc32c(0, &tmp_vid, sizeof(tmp_vid));
2069 2077
2070 /* compute the CRC on flags that have to be kept in sync 2078 /* compute the CRC on flags that have to be kept in sync
2071 * among nodes 2079 * among nodes
@@ -2262,6 +2270,7 @@ static bool batadv_tt_global_check_crc(struct batadv_orig_node *orig_node,
2262{ 2270{
2263 struct batadv_tvlv_tt_vlan_data *tt_vlan_tmp; 2271 struct batadv_tvlv_tt_vlan_data *tt_vlan_tmp;
2264 struct batadv_orig_node_vlan *vlan; 2272 struct batadv_orig_node_vlan *vlan;
2273 uint32_t crc;
2265 int i; 2274 int i;
2266 2275
2267 /* check if each received CRC matches the locally stored one */ 2276 /* check if each received CRC matches the locally stored one */
@@ -2281,7 +2290,10 @@ static bool batadv_tt_global_check_crc(struct batadv_orig_node *orig_node,
2281 if (!vlan) 2290 if (!vlan)
2282 return false; 2291 return false;
2283 2292
2284 if (vlan->tt.crc != ntohl(tt_vlan_tmp->crc)) 2293 crc = vlan->tt.crc;
2294 batadv_orig_node_vlan_free_ref(vlan);
2295
2296 if (crc != ntohl(tt_vlan_tmp->crc))
2285 return false; 2297 return false;
2286 } 2298 }
2287 2299
@@ -3218,7 +3230,6 @@ static void batadv_tt_update_orig(struct batadv_priv *bat_priv,
3218 3230
3219 spin_lock_bh(&orig_node->tt_lock); 3231 spin_lock_bh(&orig_node->tt_lock);
3220 3232
3221 tt_change = (struct batadv_tvlv_tt_change *)tt_buff;
3222 batadv_tt_update_changes(bat_priv, orig_node, tt_num_changes, 3233 batadv_tt_update_changes(bat_priv, orig_node, tt_num_changes,
3223 ttvn, tt_change); 3234 ttvn, tt_change);
3224 3235
diff --git a/net/bluetooth/hidp/core.c b/net/bluetooth/hidp/core.c
index 292e619db896..d9fb93451442 100644
--- a/net/bluetooth/hidp/core.c
+++ b/net/bluetooth/hidp/core.c
@@ -430,6 +430,16 @@ static void hidp_del_timer(struct hidp_session *session)
430 del_timer(&session->timer); 430 del_timer(&session->timer);
431} 431}
432 432
433static void hidp_process_report(struct hidp_session *session,
434 int type, const u8 *data, int len, int intr)
435{
436 if (len > HID_MAX_BUFFER_SIZE)
437 len = HID_MAX_BUFFER_SIZE;
438
439 memcpy(session->input_buf, data, len);
440 hid_input_report(session->hid, type, session->input_buf, len, intr);
441}
442
433static void hidp_process_handshake(struct hidp_session *session, 443static void hidp_process_handshake(struct hidp_session *session,
434 unsigned char param) 444 unsigned char param)
435{ 445{
@@ -502,7 +512,8 @@ static int hidp_process_data(struct hidp_session *session, struct sk_buff *skb,
502 hidp_input_report(session, skb); 512 hidp_input_report(session, skb);
503 513
504 if (session->hid) 514 if (session->hid)
505 hid_input_report(session->hid, HID_INPUT_REPORT, skb->data, skb->len, 0); 515 hidp_process_report(session, HID_INPUT_REPORT,
516 skb->data, skb->len, 0);
506 break; 517 break;
507 518
508 case HIDP_DATA_RTYPE_OTHER: 519 case HIDP_DATA_RTYPE_OTHER:
@@ -584,7 +595,8 @@ static void hidp_recv_intr_frame(struct hidp_session *session,
584 hidp_input_report(session, skb); 595 hidp_input_report(session, skb);
585 596
586 if (session->hid) { 597 if (session->hid) {
587 hid_input_report(session->hid, HID_INPUT_REPORT, skb->data, skb->len, 1); 598 hidp_process_report(session, HID_INPUT_REPORT,
599 skb->data, skb->len, 1);
588 BT_DBG("report len %d", skb->len); 600 BT_DBG("report len %d", skb->len);
589 } 601 }
590 } else { 602 } else {
diff --git a/net/bluetooth/hidp/hidp.h b/net/bluetooth/hidp/hidp.h
index ab5241400cf7..8798492a6e99 100644
--- a/net/bluetooth/hidp/hidp.h
+++ b/net/bluetooth/hidp/hidp.h
@@ -24,6 +24,7 @@
24#define __HIDP_H 24#define __HIDP_H
25 25
26#include <linux/types.h> 26#include <linux/types.h>
27#include <linux/hid.h>
27#include <linux/kref.h> 28#include <linux/kref.h>
28#include <net/bluetooth/bluetooth.h> 29#include <net/bluetooth/bluetooth.h>
29#include <net/bluetooth/l2cap.h> 30#include <net/bluetooth/l2cap.h>
@@ -179,6 +180,9 @@ struct hidp_session {
179 180
180 /* Used in hidp_output_raw_report() */ 181 /* Used in hidp_output_raw_report() */
181 int output_report_success; /* boolean */ 182 int output_report_success; /* boolean */
183
184 /* temporary input buffer */
185 u8 input_buf[HID_MAX_BUFFER_SIZE];
182}; 186};
183 187
184/* HIDP init defines */ 188/* HIDP init defines */
diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
index e4401a531afb..63f0455c0bc3 100644
--- a/net/bridge/br_device.c
+++ b/net/bridge/br_device.c
@@ -187,8 +187,7 @@ static int br_set_mac_address(struct net_device *dev, void *p)
187 187
188 spin_lock_bh(&br->lock); 188 spin_lock_bh(&br->lock);
189 if (!ether_addr_equal(dev->dev_addr, addr->sa_data)) { 189 if (!ether_addr_equal(dev->dev_addr, addr->sa_data)) {
190 memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN); 190 /* Mac address will be changed in br_stp_change_bridge_id(). */
191 br_fdb_change_mac_address(br, addr->sa_data);
192 br_stp_change_bridge_id(br, addr->sa_data); 191 br_stp_change_bridge_id(br, addr->sa_data);
193 } 192 }
194 spin_unlock_bh(&br->lock); 193 spin_unlock_bh(&br->lock);
@@ -226,6 +225,33 @@ static void br_netpoll_cleanup(struct net_device *dev)
226 br_netpoll_disable(p); 225 br_netpoll_disable(p);
227} 226}
228 227
228static int __br_netpoll_enable(struct net_bridge_port *p, gfp_t gfp)
229{
230 struct netpoll *np;
231 int err;
232
233 np = kzalloc(sizeof(*p->np), gfp);
234 if (!np)
235 return -ENOMEM;
236
237 err = __netpoll_setup(np, p->dev, gfp);
238 if (err) {
239 kfree(np);
240 return err;
241 }
242
243 p->np = np;
244 return err;
245}
246
247int br_netpoll_enable(struct net_bridge_port *p, gfp_t gfp)
248{
249 if (!p->br->dev->npinfo)
250 return 0;
251
252 return __br_netpoll_enable(p, gfp);
253}
254
229static int br_netpoll_setup(struct net_device *dev, struct netpoll_info *ni, 255static int br_netpoll_setup(struct net_device *dev, struct netpoll_info *ni,
230 gfp_t gfp) 256 gfp_t gfp)
231{ 257{
@@ -236,7 +262,7 @@ static int br_netpoll_setup(struct net_device *dev, struct netpoll_info *ni,
236 list_for_each_entry(p, &br->port_list, list) { 262 list_for_each_entry(p, &br->port_list, list) {
237 if (!p->dev) 263 if (!p->dev)
238 continue; 264 continue;
239 err = br_netpoll_enable(p, gfp); 265 err = __br_netpoll_enable(p, gfp);
240 if (err) 266 if (err)
241 goto fail; 267 goto fail;
242 } 268 }
@@ -249,28 +275,6 @@ fail:
249 goto out; 275 goto out;
250} 276}
251 277
252int br_netpoll_enable(struct net_bridge_port *p, gfp_t gfp)
253{
254 struct netpoll *np;
255 int err;
256
257 if (!p->br->dev->npinfo)
258 return 0;
259
260 np = kzalloc(sizeof(*p->np), gfp);
261 if (!np)
262 return -ENOMEM;
263
264 err = __netpoll_setup(np, p->dev, gfp);
265 if (err) {
266 kfree(np);
267 return err;
268 }
269
270 p->np = np;
271 return err;
272}
273
274void br_netpoll_disable(struct net_bridge_port *p) 278void br_netpoll_disable(struct net_bridge_port *p)
275{ 279{
276 struct netpoll *np = p->np; 280 struct netpoll *np = p->np;
diff --git a/net/bridge/br_fdb.c b/net/bridge/br_fdb.c
index c5f5a4a933f4..9203d5a1943f 100644
--- a/net/bridge/br_fdb.c
+++ b/net/bridge/br_fdb.c
@@ -27,6 +27,9 @@
27#include "br_private.h" 27#include "br_private.h"
28 28
29static struct kmem_cache *br_fdb_cache __read_mostly; 29static struct kmem_cache *br_fdb_cache __read_mostly;
30static struct net_bridge_fdb_entry *fdb_find(struct hlist_head *head,
31 const unsigned char *addr,
32 __u16 vid);
30static int fdb_insert(struct net_bridge *br, struct net_bridge_port *source, 33static int fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
31 const unsigned char *addr, u16 vid); 34 const unsigned char *addr, u16 vid);
32static void fdb_notify(struct net_bridge *br, 35static void fdb_notify(struct net_bridge *br,
@@ -89,11 +92,57 @@ static void fdb_delete(struct net_bridge *br, struct net_bridge_fdb_entry *f)
89 call_rcu(&f->rcu, fdb_rcu_free); 92 call_rcu(&f->rcu, fdb_rcu_free);
90} 93}
91 94
95/* Delete a local entry if no other port had the same address. */
96static void fdb_delete_local(struct net_bridge *br,
97 const struct net_bridge_port *p,
98 struct net_bridge_fdb_entry *f)
99{
100 const unsigned char *addr = f->addr.addr;
101 u16 vid = f->vlan_id;
102 struct net_bridge_port *op;
103
104 /* Maybe another port has same hw addr? */
105 list_for_each_entry(op, &br->port_list, list) {
106 if (op != p && ether_addr_equal(op->dev->dev_addr, addr) &&
107 (!vid || nbp_vlan_find(op, vid))) {
108 f->dst = op;
109 f->added_by_user = 0;
110 return;
111 }
112 }
113
114 /* Maybe bridge device has same hw addr? */
115 if (p && ether_addr_equal(br->dev->dev_addr, addr) &&
116 (!vid || br_vlan_find(br, vid))) {
117 f->dst = NULL;
118 f->added_by_user = 0;
119 return;
120 }
121
122 fdb_delete(br, f);
123}
124
125void br_fdb_find_delete_local(struct net_bridge *br,
126 const struct net_bridge_port *p,
127 const unsigned char *addr, u16 vid)
128{
129 struct hlist_head *head = &br->hash[br_mac_hash(addr, vid)];
130 struct net_bridge_fdb_entry *f;
131
132 spin_lock_bh(&br->hash_lock);
133 f = fdb_find(head, addr, vid);
134 if (f && f->is_local && !f->added_by_user && f->dst == p)
135 fdb_delete_local(br, p, f);
136 spin_unlock_bh(&br->hash_lock);
137}
138
92void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr) 139void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr)
93{ 140{
94 struct net_bridge *br = p->br; 141 struct net_bridge *br = p->br;
95 bool no_vlan = (nbp_get_vlan_info(p) == NULL) ? true : false; 142 struct net_port_vlans *pv = nbp_get_vlan_info(p);
143 bool no_vlan = !pv;
96 int i; 144 int i;
145 u16 vid;
97 146
98 spin_lock_bh(&br->hash_lock); 147 spin_lock_bh(&br->hash_lock);
99 148
@@ -104,38 +153,34 @@ void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr)
104 struct net_bridge_fdb_entry *f; 153 struct net_bridge_fdb_entry *f;
105 154
106 f = hlist_entry(h, struct net_bridge_fdb_entry, hlist); 155 f = hlist_entry(h, struct net_bridge_fdb_entry, hlist);
107 if (f->dst == p && f->is_local) { 156 if (f->dst == p && f->is_local && !f->added_by_user) {
108 /* maybe another port has same hw addr? */
109 struct net_bridge_port *op;
110 u16 vid = f->vlan_id;
111 list_for_each_entry(op, &br->port_list, list) {
112 if (op != p &&
113 ether_addr_equal(op->dev->dev_addr,
114 f->addr.addr) &&
115 nbp_vlan_find(op, vid)) {
116 f->dst = op;
117 goto insert;
118 }
119 }
120
121 /* delete old one */ 157 /* delete old one */
122 fdb_delete(br, f); 158 fdb_delete_local(br, p, f);
123insert:
124 /* insert new address, may fail if invalid
125 * address or dup.
126 */
127 fdb_insert(br, p, newaddr, vid);
128 159
129 /* if this port has no vlan information 160 /* if this port has no vlan information
130 * configured, we can safely be done at 161 * configured, we can safely be done at
131 * this point. 162 * this point.
132 */ 163 */
133 if (no_vlan) 164 if (no_vlan)
134 goto done; 165 goto insert;
135 } 166 }
136 } 167 }
137 } 168 }
138 169
170insert:
171 /* insert new address, may fail if invalid address or dup. */
172 fdb_insert(br, p, newaddr, 0);
173
174 if (no_vlan)
175 goto done;
176
177 /* Now add entries for every VLAN configured on the port.
178 * This function runs under RTNL so the bitmap will not change
179 * from under us.
180 */
181 for_each_set_bit(vid, pv->vlan_bitmap, VLAN_N_VID)
182 fdb_insert(br, p, newaddr, vid);
183
139done: 184done:
140 spin_unlock_bh(&br->hash_lock); 185 spin_unlock_bh(&br->hash_lock);
141} 186}
@@ -146,10 +191,12 @@ void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr)
146 struct net_port_vlans *pv; 191 struct net_port_vlans *pv;
147 u16 vid = 0; 192 u16 vid = 0;
148 193
194 spin_lock_bh(&br->hash_lock);
195
149 /* If old entry was unassociated with any port, then delete it. */ 196 /* If old entry was unassociated with any port, then delete it. */
150 f = __br_fdb_get(br, br->dev->dev_addr, 0); 197 f = __br_fdb_get(br, br->dev->dev_addr, 0);
151 if (f && f->is_local && !f->dst) 198 if (f && f->is_local && !f->dst)
152 fdb_delete(br, f); 199 fdb_delete_local(br, NULL, f);
153 200
154 fdb_insert(br, NULL, newaddr, 0); 201 fdb_insert(br, NULL, newaddr, 0);
155 202
@@ -159,14 +206,16 @@ void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr)
159 */ 206 */
160 pv = br_get_vlan_info(br); 207 pv = br_get_vlan_info(br);
161 if (!pv) 208 if (!pv)
162 return; 209 goto out;
163 210
164 for_each_set_bit_from(vid, pv->vlan_bitmap, VLAN_N_VID) { 211 for_each_set_bit_from(vid, pv->vlan_bitmap, VLAN_N_VID) {
165 f = __br_fdb_get(br, br->dev->dev_addr, vid); 212 f = __br_fdb_get(br, br->dev->dev_addr, vid);
166 if (f && f->is_local && !f->dst) 213 if (f && f->is_local && !f->dst)
167 fdb_delete(br, f); 214 fdb_delete_local(br, NULL, f);
168 fdb_insert(br, NULL, newaddr, vid); 215 fdb_insert(br, NULL, newaddr, vid);
169 } 216 }
217out:
218 spin_unlock_bh(&br->hash_lock);
170} 219}
171 220
172void br_fdb_cleanup(unsigned long _data) 221void br_fdb_cleanup(unsigned long _data)
@@ -235,25 +284,11 @@ void br_fdb_delete_by_port(struct net_bridge *br,
235 284
236 if (f->is_static && !do_all) 285 if (f->is_static && !do_all)
237 continue; 286 continue;
238 /*
239 * if multiple ports all have the same device address
240 * then when one port is deleted, assign
241 * the local entry to other port
242 */
243 if (f->is_local) {
244 struct net_bridge_port *op;
245 list_for_each_entry(op, &br->port_list, list) {
246 if (op != p &&
247 ether_addr_equal(op->dev->dev_addr,
248 f->addr.addr)) {
249 f->dst = op;
250 goto skip_delete;
251 }
252 }
253 }
254 287
255 fdb_delete(br, f); 288 if (f->is_local)
256 skip_delete: ; 289 fdb_delete_local(br, p, f);
290 else
291 fdb_delete(br, f);
257 } 292 }
258 } 293 }
259 spin_unlock_bh(&br->hash_lock); 294 spin_unlock_bh(&br->hash_lock);
@@ -397,6 +432,7 @@ static struct net_bridge_fdb_entry *fdb_create(struct hlist_head *head,
397 fdb->vlan_id = vid; 432 fdb->vlan_id = vid;
398 fdb->is_local = 0; 433 fdb->is_local = 0;
399 fdb->is_static = 0; 434 fdb->is_static = 0;
435 fdb->added_by_user = 0;
400 fdb->updated = fdb->used = jiffies; 436 fdb->updated = fdb->used = jiffies;
401 hlist_add_head_rcu(&fdb->hlist, head); 437 hlist_add_head_rcu(&fdb->hlist, head);
402 } 438 }
@@ -447,7 +483,7 @@ int br_fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
447} 483}
448 484
449void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source, 485void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source,
450 const unsigned char *addr, u16 vid) 486 const unsigned char *addr, u16 vid, bool added_by_user)
451{ 487{
452 struct hlist_head *head = &br->hash[br_mac_hash(addr, vid)]; 488 struct hlist_head *head = &br->hash[br_mac_hash(addr, vid)];
453 struct net_bridge_fdb_entry *fdb; 489 struct net_bridge_fdb_entry *fdb;
@@ -473,13 +509,18 @@ void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source,
473 /* fastpath: update of existing entry */ 509 /* fastpath: update of existing entry */
474 fdb->dst = source; 510 fdb->dst = source;
475 fdb->updated = jiffies; 511 fdb->updated = jiffies;
512 if (unlikely(added_by_user))
513 fdb->added_by_user = 1;
476 } 514 }
477 } else { 515 } else {
478 spin_lock(&br->hash_lock); 516 spin_lock(&br->hash_lock);
479 if (likely(!fdb_find(head, addr, vid))) { 517 if (likely(!fdb_find(head, addr, vid))) {
480 fdb = fdb_create(head, source, addr, vid); 518 fdb = fdb_create(head, source, addr, vid);
481 if (fdb) 519 if (fdb) {
520 if (unlikely(added_by_user))
521 fdb->added_by_user = 1;
482 fdb_notify(br, fdb, RTM_NEWNEIGH); 522 fdb_notify(br, fdb, RTM_NEWNEIGH);
523 }
483 } 524 }
484 /* else we lose race and someone else inserts 525 /* else we lose race and someone else inserts
485 * it first, don't bother updating 526 * it first, don't bother updating
@@ -647,6 +688,7 @@ static int fdb_add_entry(struct net_bridge_port *source, const __u8 *addr,
647 688
648 modified = true; 689 modified = true;
649 } 690 }
691 fdb->added_by_user = 1;
650 692
651 fdb->used = jiffies; 693 fdb->used = jiffies;
652 if (modified) { 694 if (modified) {
@@ -664,7 +706,7 @@ static int __br_fdb_add(struct ndmsg *ndm, struct net_bridge_port *p,
664 706
665 if (ndm->ndm_flags & NTF_USE) { 707 if (ndm->ndm_flags & NTF_USE) {
666 rcu_read_lock(); 708 rcu_read_lock();
667 br_fdb_update(p->br, p, addr, vid); 709 br_fdb_update(p->br, p, addr, vid, true);
668 rcu_read_unlock(); 710 rcu_read_unlock();
669 } else { 711 } else {
670 spin_lock_bh(&p->br->hash_lock); 712 spin_lock_bh(&p->br->hash_lock);
@@ -749,8 +791,7 @@ out:
749 return err; 791 return err;
750} 792}
751 793
752int fdb_delete_by_addr(struct net_bridge *br, const u8 *addr, 794static int fdb_delete_by_addr(struct net_bridge *br, const u8 *addr, u16 vlan)
753 u16 vlan)
754{ 795{
755 struct hlist_head *head = &br->hash[br_mac_hash(addr, vlan)]; 796 struct hlist_head *head = &br->hash[br_mac_hash(addr, vlan)];
756 struct net_bridge_fdb_entry *fdb; 797 struct net_bridge_fdb_entry *fdb;
diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c
index cffe1d666ba1..54d207d3a31c 100644
--- a/net/bridge/br_if.c
+++ b/net/bridge/br_if.c
@@ -389,6 +389,9 @@ int br_add_if(struct net_bridge *br, struct net_device *dev)
389 if (br->dev->needed_headroom < dev->needed_headroom) 389 if (br->dev->needed_headroom < dev->needed_headroom)
390 br->dev->needed_headroom = dev->needed_headroom; 390 br->dev->needed_headroom = dev->needed_headroom;
391 391
392 if (br_fdb_insert(br, p, dev->dev_addr, 0))
393 netdev_err(dev, "failed insert local address bridge forwarding table\n");
394
392 spin_lock_bh(&br->lock); 395 spin_lock_bh(&br->lock);
393 changed_addr = br_stp_recalculate_bridge_id(br); 396 changed_addr = br_stp_recalculate_bridge_id(br);
394 397
@@ -404,9 +407,6 @@ int br_add_if(struct net_bridge *br, struct net_device *dev)
404 407
405 dev_set_mtu(br->dev, br_min_mtu(br)); 408 dev_set_mtu(br->dev, br_min_mtu(br));
406 409
407 if (br_fdb_insert(br, p, dev->dev_addr, 0))
408 netdev_err(dev, "failed insert local address bridge forwarding table\n");
409
410 kobject_uevent(&p->kobj, KOBJ_ADD); 410 kobject_uevent(&p->kobj, KOBJ_ADD);
411 411
412 return 0; 412 return 0;
diff --git a/net/bridge/br_input.c b/net/bridge/br_input.c
index bf8dc7d308d6..28d544627422 100644
--- a/net/bridge/br_input.c
+++ b/net/bridge/br_input.c
@@ -77,7 +77,7 @@ int br_handle_frame_finish(struct sk_buff *skb)
77 /* insert into forwarding database after filtering to avoid spoofing */ 77 /* insert into forwarding database after filtering to avoid spoofing */
78 br = p->br; 78 br = p->br;
79 if (p->flags & BR_LEARNING) 79 if (p->flags & BR_LEARNING)
80 br_fdb_update(br, p, eth_hdr(skb)->h_source, vid); 80 br_fdb_update(br, p, eth_hdr(skb)->h_source, vid, false);
81 81
82 if (!is_broadcast_ether_addr(dest) && is_multicast_ether_addr(dest) && 82 if (!is_broadcast_ether_addr(dest) && is_multicast_ether_addr(dest) &&
83 br_multicast_rcv(br, p, skb, vid)) 83 br_multicast_rcv(br, p, skb, vid))
@@ -148,7 +148,7 @@ static int br_handle_local_finish(struct sk_buff *skb)
148 148
149 br_vlan_get_tag(skb, &vid); 149 br_vlan_get_tag(skb, &vid);
150 if (p->flags & BR_LEARNING) 150 if (p->flags & BR_LEARNING)
151 br_fdb_update(p->br, p, eth_hdr(skb)->h_source, vid); 151 br_fdb_update(p->br, p, eth_hdr(skb)->h_source, vid, false);
152 return 0; /* process further */ 152 return 0; /* process further */
153} 153}
154 154
diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index fcd12333c59b..3ba11bc99b65 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -104,6 +104,7 @@ struct net_bridge_fdb_entry
104 mac_addr addr; 104 mac_addr addr;
105 unsigned char is_local; 105 unsigned char is_local;
106 unsigned char is_static; 106 unsigned char is_static;
107 unsigned char added_by_user;
107 __u16 vlan_id; 108 __u16 vlan_id;
108}; 109};
109 110
@@ -370,6 +371,9 @@ static inline void br_netpoll_disable(struct net_bridge_port *p)
370int br_fdb_init(void); 371int br_fdb_init(void);
371void br_fdb_fini(void); 372void br_fdb_fini(void);
372void br_fdb_flush(struct net_bridge *br); 373void br_fdb_flush(struct net_bridge *br);
374void br_fdb_find_delete_local(struct net_bridge *br,
375 const struct net_bridge_port *p,
376 const unsigned char *addr, u16 vid);
373void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr); 377void br_fdb_changeaddr(struct net_bridge_port *p, const unsigned char *newaddr);
374void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr); 378void br_fdb_change_mac_address(struct net_bridge *br, const u8 *newaddr);
375void br_fdb_cleanup(unsigned long arg); 379void br_fdb_cleanup(unsigned long arg);
@@ -383,8 +387,7 @@ int br_fdb_fillbuf(struct net_bridge *br, void *buf, unsigned long count,
383int br_fdb_insert(struct net_bridge *br, struct net_bridge_port *source, 387int br_fdb_insert(struct net_bridge *br, struct net_bridge_port *source,
384 const unsigned char *addr, u16 vid); 388 const unsigned char *addr, u16 vid);
385void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source, 389void br_fdb_update(struct net_bridge *br, struct net_bridge_port *source,
386 const unsigned char *addr, u16 vid); 390 const unsigned char *addr, u16 vid, bool added_by_user);
387int fdb_delete_by_addr(struct net_bridge *br, const u8 *addr, u16 vid);
388 391
389int br_fdb_delete(struct ndmsg *ndm, struct nlattr *tb[], 392int br_fdb_delete(struct ndmsg *ndm, struct nlattr *tb[],
390 struct net_device *dev, const unsigned char *addr); 393 struct net_device *dev, const unsigned char *addr);
@@ -584,6 +587,7 @@ struct sk_buff *br_handle_vlan(struct net_bridge *br,
584int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags); 587int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags);
585int br_vlan_delete(struct net_bridge *br, u16 vid); 588int br_vlan_delete(struct net_bridge *br, u16 vid);
586void br_vlan_flush(struct net_bridge *br); 589void br_vlan_flush(struct net_bridge *br);
590bool br_vlan_find(struct net_bridge *br, u16 vid);
587int br_vlan_filter_toggle(struct net_bridge *br, unsigned long val); 591int br_vlan_filter_toggle(struct net_bridge *br, unsigned long val);
588int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags); 592int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags);
589int nbp_vlan_delete(struct net_bridge_port *port, u16 vid); 593int nbp_vlan_delete(struct net_bridge_port *port, u16 vid);
@@ -665,6 +669,11 @@ static inline void br_vlan_flush(struct net_bridge *br)
665{ 669{
666} 670}
667 671
672static inline bool br_vlan_find(struct net_bridge *br, u16 vid)
673{
674 return false;
675}
676
668static inline int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags) 677static inline int nbp_vlan_add(struct net_bridge_port *port, u16 vid, u16 flags)
669{ 678{
670 return -EOPNOTSUPP; 679 return -EOPNOTSUPP;
diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
index 656a6f3e40de..189ba1e7d851 100644
--- a/net/bridge/br_stp_if.c
+++ b/net/bridge/br_stp_if.c
@@ -194,6 +194,8 @@ void br_stp_change_bridge_id(struct net_bridge *br, const unsigned char *addr)
194 194
195 wasroot = br_is_root_bridge(br); 195 wasroot = br_is_root_bridge(br);
196 196
197 br_fdb_change_mac_address(br, addr);
198
197 memcpy(oldaddr, br->bridge_id.addr, ETH_ALEN); 199 memcpy(oldaddr, br->bridge_id.addr, ETH_ALEN);
198 memcpy(br->bridge_id.addr, addr, ETH_ALEN); 200 memcpy(br->bridge_id.addr, addr, ETH_ALEN);
199 memcpy(br->dev->dev_addr, addr, ETH_ALEN); 201 memcpy(br->dev->dev_addr, addr, ETH_ALEN);
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index 4ca4d0a0151c..8249ca764c79 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -275,9 +275,7 @@ int br_vlan_delete(struct net_bridge *br, u16 vid)
275 if (!pv) 275 if (!pv)
276 return -EINVAL; 276 return -EINVAL;
277 277
278 spin_lock_bh(&br->hash_lock); 278 br_fdb_find_delete_local(br, NULL, br->dev->dev_addr, vid);
279 fdb_delete_by_addr(br, br->dev->dev_addr, vid);
280 spin_unlock_bh(&br->hash_lock);
281 279
282 __vlan_del(pv, vid); 280 __vlan_del(pv, vid);
283 return 0; 281 return 0;
@@ -295,6 +293,25 @@ void br_vlan_flush(struct net_bridge *br)
295 __vlan_flush(pv); 293 __vlan_flush(pv);
296} 294}
297 295
296bool br_vlan_find(struct net_bridge *br, u16 vid)
297{
298 struct net_port_vlans *pv;
299 bool found = false;
300
301 rcu_read_lock();
302 pv = rcu_dereference(br->vlan_info);
303
304 if (!pv)
305 goto out;
306
307 if (test_bit(vid, pv->vlan_bitmap))
308 found = true;
309
310out:
311 rcu_read_unlock();
312 return found;
313}
314
298int br_vlan_filter_toggle(struct net_bridge *br, unsigned long val) 315int br_vlan_filter_toggle(struct net_bridge *br, unsigned long val)
299{ 316{
300 if (!rtnl_trylock()) 317 if (!rtnl_trylock())
@@ -359,9 +376,7 @@ int nbp_vlan_delete(struct net_bridge_port *port, u16 vid)
359 if (!pv) 376 if (!pv)
360 return -EINVAL; 377 return -EINVAL;
361 378
362 spin_lock_bh(&port->br->hash_lock); 379 br_fdb_find_delete_local(port->br, port, port->dev->dev_addr, vid);
363 fdb_delete_by_addr(port->br, port->dev->dev_addr, vid);
364 spin_unlock_bh(&port->br->hash_lock);
365 380
366 return __vlan_del(pv, vid); 381 return __vlan_del(pv, vid);
367} 382}
diff --git a/net/caif/caif_dev.c b/net/caif/caif_dev.c
index 4dca159435cf..edbca468fa73 100644
--- a/net/caif/caif_dev.c
+++ b/net/caif/caif_dev.c
@@ -22,6 +22,7 @@
22#include <net/pkt_sched.h> 22#include <net/pkt_sched.h>
23#include <net/caif/caif_device.h> 23#include <net/caif/caif_device.h>
24#include <net/caif/caif_layer.h> 24#include <net/caif/caif_layer.h>
25#include <net/caif/caif_dev.h>
25#include <net/caif/cfpkt.h> 26#include <net/caif/cfpkt.h>
26#include <net/caif/cfcnfg.h> 27#include <net/caif/cfcnfg.h>
27#include <net/caif/cfserl.h> 28#include <net/caif/cfserl.h>
diff --git a/net/caif/cfsrvl.c b/net/caif/cfsrvl.c
index 353f793d1b3b..a6e115463052 100644
--- a/net/caif/cfsrvl.c
+++ b/net/caif/cfsrvl.c
@@ -15,6 +15,7 @@
15#include <net/caif/caif_layer.h> 15#include <net/caif/caif_layer.h>
16#include <net/caif/cfsrvl.h> 16#include <net/caif/cfsrvl.h>
17#include <net/caif/cfpkt.h> 17#include <net/caif/cfpkt.h>
18#include <net/caif/caif_dev.h>
18 19
19#define SRVL_CTRL_PKT_SIZE 1 20#define SRVL_CTRL_PKT_SIZE 1
20#define SRVL_FLOW_OFF 0x81 21#define SRVL_FLOW_OFF 0x81
diff --git a/net/can/af_can.c b/net/can/af_can.c
index d249874a366d..a27f8aad9e99 100644
--- a/net/can/af_can.c
+++ b/net/can/af_can.c
@@ -57,6 +57,7 @@
57#include <linux/skbuff.h> 57#include <linux/skbuff.h>
58#include <linux/can.h> 58#include <linux/can.h>
59#include <linux/can/core.h> 59#include <linux/can/core.h>
60#include <linux/can/skb.h>
60#include <linux/ratelimit.h> 61#include <linux/ratelimit.h>
61#include <net/net_namespace.h> 62#include <net/net_namespace.h>
62#include <net/sock.h> 63#include <net/sock.h>
@@ -290,7 +291,7 @@ int can_send(struct sk_buff *skb, int loop)
290 return -ENOMEM; 291 return -ENOMEM;
291 } 292 }
292 293
293 newskb->sk = skb->sk; 294 can_skb_set_owner(newskb, skb->sk);
294 newskb->ip_summed = CHECKSUM_UNNECESSARY; 295 newskb->ip_summed = CHECKSUM_UNNECESSARY;
295 newskb->pkt_type = PACKET_BROADCAST; 296 newskb->pkt_type = PACKET_BROADCAST;
296 } 297 }
diff --git a/net/can/bcm.c b/net/can/bcm.c
index 3fc737b214c7..dcb75c0e66c1 100644
--- a/net/can/bcm.c
+++ b/net/can/bcm.c
@@ -268,7 +268,7 @@ static void bcm_can_tx(struct bcm_op *op)
268 268
269 /* send with loopback */ 269 /* send with loopback */
270 skb->dev = dev; 270 skb->dev = dev;
271 skb->sk = op->sk; 271 can_skb_set_owner(skb, op->sk);
272 can_send(skb, 1); 272 can_send(skb, 1);
273 273
274 /* update statistics */ 274 /* update statistics */
@@ -1223,7 +1223,7 @@ static int bcm_tx_send(struct msghdr *msg, int ifindex, struct sock *sk)
1223 1223
1224 can_skb_prv(skb)->ifindex = dev->ifindex; 1224 can_skb_prv(skb)->ifindex = dev->ifindex;
1225 skb->dev = dev; 1225 skb->dev = dev;
1226 skb->sk = sk; 1226 can_skb_set_owner(skb, sk);
1227 err = can_send(skb, 1); /* send with loopback */ 1227 err = can_send(skb, 1); /* send with loopback */
1228 dev_put(dev); 1228 dev_put(dev);
1229 1229
diff --git a/net/can/raw.c b/net/can/raw.c
index 07d72d852324..081e81fd017f 100644
--- a/net/can/raw.c
+++ b/net/can/raw.c
@@ -121,13 +121,9 @@ static void raw_rcv(struct sk_buff *oskb, void *data)
121 if (!ro->recv_own_msgs && oskb->sk == sk) 121 if (!ro->recv_own_msgs && oskb->sk == sk)
122 return; 122 return;
123 123
124 /* do not pass frames with DLC > 8 to a legacy socket */ 124 /* do not pass non-CAN2.0 frames to a legacy socket */
125 if (!ro->fd_frames) { 125 if (!ro->fd_frames && oskb->len != CAN_MTU)
126 struct canfd_frame *cfd = (struct canfd_frame *)oskb->data; 126 return;
127
128 if (unlikely(cfd->len > CAN_MAX_DLEN))
129 return;
130 }
131 127
132 /* clone the given skb to be able to enqueue it into the rcv queue */ 128 /* clone the given skb to be able to enqueue it into the rcv queue */
133 skb = skb_clone(oskb, GFP_ATOMIC); 129 skb = skb_clone(oskb, GFP_ATOMIC);
@@ -715,6 +711,7 @@ static int raw_sendmsg(struct kiocb *iocb, struct socket *sock,
715 711
716 skb->dev = dev; 712 skb->dev = dev;
717 skb->sk = sk; 713 skb->sk = sk;
714 skb->priority = sk->sk_priority;
718 715
719 err = can_send(skb, ro->loopback); 716 err = can_send(skb, ro->loopback);
720 717
@@ -737,9 +734,7 @@ static int raw_recvmsg(struct kiocb *iocb, struct socket *sock,
737 struct msghdr *msg, size_t size, int flags) 734 struct msghdr *msg, size_t size, int flags)
738{ 735{
739 struct sock *sk = sock->sk; 736 struct sock *sk = sock->sk;
740 struct raw_sock *ro = raw_sk(sk);
741 struct sk_buff *skb; 737 struct sk_buff *skb;
742 int rxmtu;
743 int err = 0; 738 int err = 0;
744 int noblock; 739 int noblock;
745 740
@@ -750,20 +745,10 @@ static int raw_recvmsg(struct kiocb *iocb, struct socket *sock,
750 if (!skb) 745 if (!skb)
751 return err; 746 return err;
752 747
753 /* 748 if (size < skb->len)
754 * when serving a legacy socket the DLC <= 8 is already checked inside
755 * raw_rcv(). Now check if we need to pass a canfd_frame to a legacy
756 * socket and cut the possible CANFD_MTU/CAN_MTU length to CAN_MTU
757 */
758 if (!ro->fd_frames)
759 rxmtu = CAN_MTU;
760 else
761 rxmtu = skb->len;
762
763 if (size < rxmtu)
764 msg->msg_flags |= MSG_TRUNC; 749 msg->msg_flags |= MSG_TRUNC;
765 else 750 else
766 size = rxmtu; 751 size = skb->len;
767 752
768 err = memcpy_toiovec(msg->msg_iov, skb->data, size); 753 err = memcpy_toiovec(msg->msg_iov, skb->data, size);
769 if (err < 0) { 754 if (err < 0) {
diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 0e478a0f4204..30efc5c18622 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -840,9 +840,13 @@ static bool ceph_msg_data_bio_advance(struct ceph_msg_data_cursor *cursor,
840 840
841 if (!cursor->bvec_iter.bi_size) { 841 if (!cursor->bvec_iter.bi_size) {
842 bio = bio->bi_next; 842 bio = bio->bi_next;
843 cursor->bvec_iter = bio->bi_iter; 843 cursor->bio = bio;
844 if (bio)
845 cursor->bvec_iter = bio->bi_iter;
846 else
847 memset(&cursor->bvec_iter, 0,
848 sizeof(cursor->bvec_iter));
844 } 849 }
845 cursor->bio = bio;
846 850
847 if (!cursor->last_piece) { 851 if (!cursor->last_piece) {
848 BUG_ON(!cursor->resid); 852 BUG_ON(!cursor->resid);
diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c
index 010ff3bd58ad..0676f2b199d6 100644
--- a/net/ceph/osd_client.c
+++ b/net/ceph/osd_client.c
@@ -1427,6 +1427,40 @@ static void __send_queued(struct ceph_osd_client *osdc)
1427} 1427}
1428 1428
1429/* 1429/*
1430 * Caller should hold map_sem for read and request_mutex.
1431 */
1432static int __ceph_osdc_start_request(struct ceph_osd_client *osdc,
1433 struct ceph_osd_request *req,
1434 bool nofail)
1435{
1436 int rc;
1437
1438 __register_request(osdc, req);
1439 req->r_sent = 0;
1440 req->r_got_reply = 0;
1441 rc = __map_request(osdc, req, 0);
1442 if (rc < 0) {
1443 if (nofail) {
1444 dout("osdc_start_request failed map, "
1445 " will retry %lld\n", req->r_tid);
1446 rc = 0;
1447 } else {
1448 __unregister_request(osdc, req);
1449 }
1450 return rc;
1451 }
1452
1453 if (req->r_osd == NULL) {
1454 dout("send_request %p no up osds in pg\n", req);
1455 ceph_monc_request_next_osdmap(&osdc->client->monc);
1456 } else {
1457 __send_queued(osdc);
1458 }
1459
1460 return 0;
1461}
1462
1463/*
1430 * Timeout callback, called every N seconds when 1 or more osd 1464 * Timeout callback, called every N seconds when 1 or more osd
1431 * requests has been active for more than N seconds. When this 1465 * requests has been active for more than N seconds. When this
1432 * happens, we ping all OSDs with requests who have timed out to 1466 * happens, we ping all OSDs with requests who have timed out to
@@ -1653,6 +1687,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1653 osdmap_epoch = ceph_decode_32(&p); 1687 osdmap_epoch = ceph_decode_32(&p);
1654 1688
1655 /* lookup */ 1689 /* lookup */
1690 down_read(&osdc->map_sem);
1656 mutex_lock(&osdc->request_mutex); 1691 mutex_lock(&osdc->request_mutex);
1657 req = __lookup_request(osdc, tid); 1692 req = __lookup_request(osdc, tid);
1658 if (req == NULL) { 1693 if (req == NULL) {
@@ -1709,7 +1744,6 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1709 dout("redirect pool %lld\n", redir.oloc.pool); 1744 dout("redirect pool %lld\n", redir.oloc.pool);
1710 1745
1711 __unregister_request(osdc, req); 1746 __unregister_request(osdc, req);
1712 mutex_unlock(&osdc->request_mutex);
1713 1747
1714 req->r_target_oloc = redir.oloc; /* struct */ 1748 req->r_target_oloc = redir.oloc; /* struct */
1715 1749
@@ -1721,10 +1755,10 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1721 * successfully. In the future we might want to follow 1755 * successfully. In the future we might want to follow
1722 * original request's nofail setting here. 1756 * original request's nofail setting here.
1723 */ 1757 */
1724 err = ceph_osdc_start_request(osdc, req, true); 1758 err = __ceph_osdc_start_request(osdc, req, true);
1725 BUG_ON(err); 1759 BUG_ON(err);
1726 1760
1727 goto done; 1761 goto out_unlock;
1728 } 1762 }
1729 1763
1730 already_completed = req->r_got_reply; 1764 already_completed = req->r_got_reply;
@@ -1742,8 +1776,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1742 req->r_got_reply = 1; 1776 req->r_got_reply = 1;
1743 } else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) { 1777 } else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) {
1744 dout("handle_reply tid %llu dup ack\n", tid); 1778 dout("handle_reply tid %llu dup ack\n", tid);
1745 mutex_unlock(&osdc->request_mutex); 1779 goto out_unlock;
1746 goto done;
1747 } 1780 }
1748 1781
1749 dout("handle_reply tid %llu flags %d\n", tid, flags); 1782 dout("handle_reply tid %llu flags %d\n", tid, flags);
@@ -1758,6 +1791,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1758 __unregister_request(osdc, req); 1791 __unregister_request(osdc, req);
1759 1792
1760 mutex_unlock(&osdc->request_mutex); 1793 mutex_unlock(&osdc->request_mutex);
1794 up_read(&osdc->map_sem);
1761 1795
1762 if (!already_completed) { 1796 if (!already_completed) {
1763 if (req->r_unsafe_callback && 1797 if (req->r_unsafe_callback &&
@@ -1775,10 +1809,14 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
1775 complete_request(req); 1809 complete_request(req);
1776 } 1810 }
1777 1811
1778done: 1812out:
1779 dout("req=%p req->r_linger=%d\n", req, req->r_linger); 1813 dout("req=%p req->r_linger=%d\n", req, req->r_linger);
1780 ceph_osdc_put_request(req); 1814 ceph_osdc_put_request(req);
1781 return; 1815 return;
1816out_unlock:
1817 mutex_unlock(&osdc->request_mutex);
1818 up_read(&osdc->map_sem);
1819 goto out;
1782 1820
1783bad_put: 1821bad_put:
1784 req->r_result = -EIO; 1822 req->r_result = -EIO;
@@ -1791,6 +1829,7 @@ bad_put:
1791 ceph_osdc_put_request(req); 1829 ceph_osdc_put_request(req);
1792bad_mutex: 1830bad_mutex:
1793 mutex_unlock(&osdc->request_mutex); 1831 mutex_unlock(&osdc->request_mutex);
1832 up_read(&osdc->map_sem);
1794bad: 1833bad:
1795 pr_err("corrupt osd_op_reply got %d %d\n", 1834 pr_err("corrupt osd_op_reply got %d %d\n",
1796 (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len)); 1835 (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len));
@@ -2351,34 +2390,16 @@ int ceph_osdc_start_request(struct ceph_osd_client *osdc,
2351 struct ceph_osd_request *req, 2390 struct ceph_osd_request *req,
2352 bool nofail) 2391 bool nofail)
2353{ 2392{
2354 int rc = 0; 2393 int rc;
2355 2394
2356 down_read(&osdc->map_sem); 2395 down_read(&osdc->map_sem);
2357 mutex_lock(&osdc->request_mutex); 2396 mutex_lock(&osdc->request_mutex);
2358 __register_request(osdc, req); 2397
2359 req->r_sent = 0; 2398 rc = __ceph_osdc_start_request(osdc, req, nofail);
2360 req->r_got_reply = 0; 2399
2361 rc = __map_request(osdc, req, 0);
2362 if (rc < 0) {
2363 if (nofail) {
2364 dout("osdc_start_request failed map, "
2365 " will retry %lld\n", req->r_tid);
2366 rc = 0;
2367 } else {
2368 __unregister_request(osdc, req);
2369 }
2370 goto out_unlock;
2371 }
2372 if (req->r_osd == NULL) {
2373 dout("send_request %p no up osds in pg\n", req);
2374 ceph_monc_request_next_osdmap(&osdc->client->monc);
2375 } else {
2376 __send_queued(osdc);
2377 }
2378 rc = 0;
2379out_unlock:
2380 mutex_unlock(&osdc->request_mutex); 2400 mutex_unlock(&osdc->request_mutex);
2381 up_read(&osdc->map_sem); 2401 up_read(&osdc->map_sem);
2402
2382 return rc; 2403 return rc;
2383} 2404}
2384EXPORT_SYMBOL(ceph_osdc_start_request); 2405EXPORT_SYMBOL(ceph_osdc_start_request);
@@ -2504,9 +2525,12 @@ int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
2504 err = -ENOMEM; 2525 err = -ENOMEM;
2505 osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify"); 2526 osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify");
2506 if (!osdc->notify_wq) 2527 if (!osdc->notify_wq)
2507 goto out_msgpool; 2528 goto out_msgpool_reply;
2529
2508 return 0; 2530 return 0;
2509 2531
2532out_msgpool_reply:
2533 ceph_msgpool_destroy(&osdc->msgpool_op_reply);
2510out_msgpool: 2534out_msgpool:
2511 ceph_msgpool_destroy(&osdc->msgpool_op); 2535 ceph_msgpool_destroy(&osdc->msgpool_op);
2512out_mempool: 2536out_mempool:
diff --git a/net/core/dev.c b/net/core/dev.c
index 3721db716350..b1b0c8d4d7df 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2420,7 +2420,7 @@ EXPORT_SYMBOL(netdev_rx_csum_fault);
2420 * 2. No high memory really exists on this machine. 2420 * 2. No high memory really exists on this machine.
2421 */ 2421 */
2422 2422
2423static int illegal_highdma(struct net_device *dev, struct sk_buff *skb) 2423static int illegal_highdma(const struct net_device *dev, struct sk_buff *skb)
2424{ 2424{
2425#ifdef CONFIG_HIGHMEM 2425#ifdef CONFIG_HIGHMEM
2426 int i; 2426 int i;
@@ -2495,34 +2495,36 @@ static int dev_gso_segment(struct sk_buff *skb, netdev_features_t features)
2495} 2495}
2496 2496
2497static netdev_features_t harmonize_features(struct sk_buff *skb, 2497static netdev_features_t harmonize_features(struct sk_buff *skb,
2498 netdev_features_t features) 2498 const struct net_device *dev,
2499 netdev_features_t features)
2499{ 2500{
2500 if (skb->ip_summed != CHECKSUM_NONE && 2501 if (skb->ip_summed != CHECKSUM_NONE &&
2501 !can_checksum_protocol(features, skb_network_protocol(skb))) { 2502 !can_checksum_protocol(features, skb_network_protocol(skb))) {
2502 features &= ~NETIF_F_ALL_CSUM; 2503 features &= ~NETIF_F_ALL_CSUM;
2503 } else if (illegal_highdma(skb->dev, skb)) { 2504 } else if (illegal_highdma(dev, skb)) {
2504 features &= ~NETIF_F_SG; 2505 features &= ~NETIF_F_SG;
2505 } 2506 }
2506 2507
2507 return features; 2508 return features;
2508} 2509}
2509 2510
2510netdev_features_t netif_skb_features(struct sk_buff *skb) 2511netdev_features_t netif_skb_dev_features(struct sk_buff *skb,
2512 const struct net_device *dev)
2511{ 2513{
2512 __be16 protocol = skb->protocol; 2514 __be16 protocol = skb->protocol;
2513 netdev_features_t features = skb->dev->features; 2515 netdev_features_t features = dev->features;
2514 2516
2515 if (skb_shinfo(skb)->gso_segs > skb->dev->gso_max_segs) 2517 if (skb_shinfo(skb)->gso_segs > dev->gso_max_segs)
2516 features &= ~NETIF_F_GSO_MASK; 2518 features &= ~NETIF_F_GSO_MASK;
2517 2519
2518 if (protocol == htons(ETH_P_8021Q) || protocol == htons(ETH_P_8021AD)) { 2520 if (protocol == htons(ETH_P_8021Q) || protocol == htons(ETH_P_8021AD)) {
2519 struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data; 2521 struct vlan_ethhdr *veh = (struct vlan_ethhdr *)skb->data;
2520 protocol = veh->h_vlan_encapsulated_proto; 2522 protocol = veh->h_vlan_encapsulated_proto;
2521 } else if (!vlan_tx_tag_present(skb)) { 2523 } else if (!vlan_tx_tag_present(skb)) {
2522 return harmonize_features(skb, features); 2524 return harmonize_features(skb, dev, features);
2523 } 2525 }
2524 2526
2525 features &= (skb->dev->vlan_features | NETIF_F_HW_VLAN_CTAG_TX | 2527 features &= (dev->vlan_features | NETIF_F_HW_VLAN_CTAG_TX |
2526 NETIF_F_HW_VLAN_STAG_TX); 2528 NETIF_F_HW_VLAN_STAG_TX);
2527 2529
2528 if (protocol == htons(ETH_P_8021Q) || protocol == htons(ETH_P_8021AD)) 2530 if (protocol == htons(ETH_P_8021Q) || protocol == htons(ETH_P_8021AD))
@@ -2530,9 +2532,9 @@ netdev_features_t netif_skb_features(struct sk_buff *skb)
2530 NETIF_F_GEN_CSUM | NETIF_F_HW_VLAN_CTAG_TX | 2532 NETIF_F_GEN_CSUM | NETIF_F_HW_VLAN_CTAG_TX |
2531 NETIF_F_HW_VLAN_STAG_TX; 2533 NETIF_F_HW_VLAN_STAG_TX;
2532 2534
2533 return harmonize_features(skb, features); 2535 return harmonize_features(skb, dev, features);
2534} 2536}
2535EXPORT_SYMBOL(netif_skb_features); 2537EXPORT_SYMBOL(netif_skb_dev_features);
2536 2538
2537int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev, 2539int dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
2538 struct netdev_queue *txq) 2540 struct netdev_queue *txq)
@@ -2803,7 +2805,7 @@ EXPORT_SYMBOL(dev_loopback_xmit);
2803 * the BH enable code must have IRQs enabled so that it will not deadlock. 2805 * the BH enable code must have IRQs enabled so that it will not deadlock.
2804 * --BLG 2806 * --BLG
2805 */ 2807 */
2806int __dev_queue_xmit(struct sk_buff *skb, void *accel_priv) 2808static int __dev_queue_xmit(struct sk_buff *skb, void *accel_priv)
2807{ 2809{
2808 struct net_device *dev = skb->dev; 2810 struct net_device *dev = skb->dev;
2809 struct netdev_queue *txq; 2811 struct netdev_queue *txq;
@@ -4637,7 +4639,7 @@ struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev)
4637} 4639}
4638EXPORT_SYMBOL(netdev_master_upper_dev_get_rcu); 4640EXPORT_SYMBOL(netdev_master_upper_dev_get_rcu);
4639 4641
4640int netdev_adjacent_sysfs_add(struct net_device *dev, 4642static int netdev_adjacent_sysfs_add(struct net_device *dev,
4641 struct net_device *adj_dev, 4643 struct net_device *adj_dev,
4642 struct list_head *dev_list) 4644 struct list_head *dev_list)
4643{ 4645{
@@ -4647,7 +4649,7 @@ int netdev_adjacent_sysfs_add(struct net_device *dev,
4647 return sysfs_create_link(&(dev->dev.kobj), &(adj_dev->dev.kobj), 4649 return sysfs_create_link(&(dev->dev.kobj), &(adj_dev->dev.kobj),
4648 linkname); 4650 linkname);
4649} 4651}
4650void netdev_adjacent_sysfs_del(struct net_device *dev, 4652static void netdev_adjacent_sysfs_del(struct net_device *dev,
4651 char *name, 4653 char *name,
4652 struct list_head *dev_list) 4654 struct list_head *dev_list)
4653{ 4655{
diff --git a/net/core/fib_rules.c b/net/core/fib_rules.c
index f409e0bd35c0..185c341fafbd 100644
--- a/net/core/fib_rules.c
+++ b/net/core/fib_rules.c
@@ -745,6 +745,13 @@ static int fib_rules_event(struct notifier_block *this, unsigned long event,
745 attach_rules(&ops->rules_list, dev); 745 attach_rules(&ops->rules_list, dev);
746 break; 746 break;
747 747
748 case NETDEV_CHANGENAME:
749 list_for_each_entry(ops, &net->rules_ops, list) {
750 detach_rules(&ops->rules_list, dev);
751 attach_rules(&ops->rules_list, dev);
752 }
753 break;
754
748 case NETDEV_UNREGISTER: 755 case NETDEV_UNREGISTER:
749 list_for_each_entry(ops, &net->rules_ops, list) 756 list_for_each_entry(ops, &net->rules_ops, list)
750 detach_rules(&ops->rules_list, dev); 757 detach_rules(&ops->rules_list, dev);
diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c
index 87577d447554..e29e810663d7 100644
--- a/net/core/flow_dissector.c
+++ b/net/core/flow_dissector.c
@@ -323,17 +323,6 @@ u32 __skb_get_poff(const struct sk_buff *skb)
323 return poff; 323 return poff;
324} 324}
325 325
326static inline u16 dev_cap_txqueue(struct net_device *dev, u16 queue_index)
327{
328 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
329 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
330 dev->name, queue_index,
331 dev->real_num_tx_queues);
332 return 0;
333 }
334 return queue_index;
335}
336
337static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb) 326static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
338{ 327{
339#ifdef CONFIG_XPS 328#ifdef CONFIG_XPS
@@ -372,7 +361,7 @@ static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
372#endif 361#endif
373} 362}
374 363
375u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb) 364static u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb)
376{ 365{
377 struct sock *sk = skb->sk; 366 struct sock *sk = skb->sk;
378 int queue_index = sk_tx_queue_get(sk); 367 int queue_index = sk_tx_queue_get(sk);
@@ -392,7 +381,6 @@ u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb)
392 381
393 return queue_index; 382 return queue_index;
394} 383}
395EXPORT_SYMBOL(__netdev_pick_tx);
396 384
397struct netdev_queue *netdev_pick_tx(struct net_device *dev, 385struct netdev_queue *netdev_pick_tx(struct net_device *dev,
398 struct sk_buff *skb, 386 struct sk_buff *skb,
@@ -403,13 +391,13 @@ struct netdev_queue *netdev_pick_tx(struct net_device *dev,
403 if (dev->real_num_tx_queues != 1) { 391 if (dev->real_num_tx_queues != 1) {
404 const struct net_device_ops *ops = dev->netdev_ops; 392 const struct net_device_ops *ops = dev->netdev_ops;
405 if (ops->ndo_select_queue) 393 if (ops->ndo_select_queue)
406 queue_index = ops->ndo_select_queue(dev, skb, 394 queue_index = ops->ndo_select_queue(dev, skb, accel_priv,
407 accel_priv); 395 __netdev_pick_tx);
408 else 396 else
409 queue_index = __netdev_pick_tx(dev, skb); 397 queue_index = __netdev_pick_tx(dev, skb);
410 398
411 if (!accel_priv) 399 if (!accel_priv)
412 queue_index = dev_cap_txqueue(dev, queue_index); 400 queue_index = netdev_cap_txqueue(dev, queue_index);
413 } 401 }
414 402
415 skb_set_queue_mapping(skb, queue_index); 403 skb_set_queue_mapping(skb, queue_index);
diff --git a/net/core/neighbour.c b/net/core/neighbour.c
index b9e9e0d38672..e16129019c66 100644
--- a/net/core/neighbour.c
+++ b/net/core/neighbour.c
@@ -766,9 +766,6 @@ static void neigh_periodic_work(struct work_struct *work)
766 nht = rcu_dereference_protected(tbl->nht, 766 nht = rcu_dereference_protected(tbl->nht,
767 lockdep_is_held(&tbl->lock)); 767 lockdep_is_held(&tbl->lock));
768 768
769 if (atomic_read(&tbl->entries) < tbl->gc_thresh1)
770 goto out;
771
772 /* 769 /*
773 * periodically recompute ReachableTime from random function 770 * periodically recompute ReachableTime from random function
774 */ 771 */
@@ -781,6 +778,9 @@ static void neigh_periodic_work(struct work_struct *work)
781 neigh_rand_reach_time(NEIGH_VAR(p, BASE_REACHABLE_TIME)); 778 neigh_rand_reach_time(NEIGH_VAR(p, BASE_REACHABLE_TIME));
782 } 779 }
783 780
781 if (atomic_read(&tbl->entries) < tbl->gc_thresh1)
782 goto out;
783
784 for (i = 0 ; i < (1 << nht->hash_shift); i++) { 784 for (i = 0 ; i < (1 << nht->hash_shift); i++) {
785 np = &nht->hash_buckets[i]; 785 np = &nht->hash_buckets[i];
786 786
@@ -3046,7 +3046,7 @@ int neigh_sysctl_register(struct net_device *dev, struct neigh_parms *p,
3046 if (!t) 3046 if (!t)
3047 goto err; 3047 goto err;
3048 3048
3049 for (i = 0; i < ARRAY_SIZE(t->neigh_vars); i++) { 3049 for (i = 0; i < NEIGH_VAR_GC_INTERVAL; i++) {
3050 t->neigh_vars[i].data += (long) p; 3050 t->neigh_vars[i].data += (long) p;
3051 t->neigh_vars[i].extra1 = dev; 3051 t->neigh_vars[i].extra1 = dev;
3052 t->neigh_vars[i].extra2 = p; 3052 t->neigh_vars[i].extra2 = p;
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index c03f3dec4763..a664f7829a6d 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -948,6 +948,7 @@ int netpoll_parse_options(struct netpoll *np, char *opt)
948{ 948{
949 char *cur=opt, *delim; 949 char *cur=opt, *delim;
950 int ipv6; 950 int ipv6;
951 bool ipversion_set = false;
951 952
952 if (*cur != '@') { 953 if (*cur != '@') {
953 if ((delim = strchr(cur, '@')) == NULL) 954 if ((delim = strchr(cur, '@')) == NULL)
@@ -960,6 +961,7 @@ int netpoll_parse_options(struct netpoll *np, char *opt)
960 cur++; 961 cur++;
961 962
962 if (*cur != '/') { 963 if (*cur != '/') {
964 ipversion_set = true;
963 if ((delim = strchr(cur, '/')) == NULL) 965 if ((delim = strchr(cur, '/')) == NULL)
964 goto parse_failed; 966 goto parse_failed;
965 *delim = 0; 967 *delim = 0;
@@ -1002,7 +1004,7 @@ int netpoll_parse_options(struct netpoll *np, char *opt)
1002 ipv6 = netpoll_parse_ip_addr(cur, &np->remote_ip); 1004 ipv6 = netpoll_parse_ip_addr(cur, &np->remote_ip);
1003 if (ipv6 < 0) 1005 if (ipv6 < 0)
1004 goto parse_failed; 1006 goto parse_failed;
1005 else if (np->ipv6 != (bool)ipv6) 1007 else if (ipversion_set && np->ipv6 != (bool)ipv6)
1006 goto parse_failed; 1008 goto parse_failed;
1007 else 1009 else
1008 np->ipv6 = (bool)ipv6; 1010 np->ipv6 = (bool)ipv6;
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index 393b1bc9a618..1a0dac2ef9ad 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -374,7 +374,7 @@ static size_t rtnl_link_get_slave_info_data_size(const struct net_device *dev)
374 if (!master_dev) 374 if (!master_dev)
375 return 0; 375 return 0;
376 ops = master_dev->rtnl_link_ops; 376 ops = master_dev->rtnl_link_ops;
377 if (!ops->get_slave_size) 377 if (!ops || !ops->get_slave_size)
378 return 0; 378 return 0;
379 /* IFLA_INFO_SLAVE_DATA + nested data */ 379 /* IFLA_INFO_SLAVE_DATA + nested data */
380 return nla_total_size(sizeof(struct nlattr)) + 380 return nla_total_size(sizeof(struct nlattr)) +
@@ -1963,16 +1963,21 @@ replay:
1963 1963
1964 dev->ifindex = ifm->ifi_index; 1964 dev->ifindex = ifm->ifi_index;
1965 1965
1966 if (ops->newlink) 1966 if (ops->newlink) {
1967 err = ops->newlink(net, dev, tb, data); 1967 err = ops->newlink(net, dev, tb, data);
1968 else 1968 /* Drivers should call free_netdev() in ->destructor
1969 * and unregister it on failure so that device could be
1970 * finally freed in rtnl_unlock.
1971 */
1972 if (err < 0)
1973 goto out;
1974 } else {
1969 err = register_netdevice(dev); 1975 err = register_netdevice(dev);
1970 1976 if (err < 0) {
1971 if (err < 0) { 1977 free_netdev(dev);
1972 free_netdev(dev); 1978 goto out;
1973 goto out; 1979 }
1974 } 1980 }
1975
1976 err = rtnl_configure_link(dev, ifm); 1981 err = rtnl_configure_link(dev, ifm);
1977 if (err < 0) 1982 if (err < 0)
1978 unregister_netdevice(dev); 1983 unregister_netdevice(dev);
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 5976ef0846bd..5d6236d9fdce 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -707,9 +707,6 @@ static void __copy_skb_header(struct sk_buff *new, const struct sk_buff *old)
707 new->mark = old->mark; 707 new->mark = old->mark;
708 new->skb_iif = old->skb_iif; 708 new->skb_iif = old->skb_iif;
709 __nf_copy(new, old); 709 __nf_copy(new, old);
710#if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
711 new->nf_trace = old->nf_trace;
712#endif
713#ifdef CONFIG_NET_SCHED 710#ifdef CONFIG_NET_SCHED
714 new->tc_index = old->tc_index; 711 new->tc_index = old->tc_index;
715#ifdef CONFIG_NET_CLS_ACT 712#ifdef CONFIG_NET_CLS_ACT
diff --git a/net/core/sock.c b/net/core/sock.c
index 0c127dcdf6a8..5b6a9431b017 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -1775,7 +1775,9 @@ struct sk_buff *sock_alloc_send_pskb(struct sock *sk, unsigned long header_len,
1775 while (order) { 1775 while (order) {
1776 if (npages >= 1 << order) { 1776 if (npages >= 1 << order) {
1777 page = alloc_pages(sk->sk_allocation | 1777 page = alloc_pages(sk->sk_allocation |
1778 __GFP_COMP | __GFP_NOWARN, 1778 __GFP_COMP |
1779 __GFP_NOWARN |
1780 __GFP_NORETRY,
1779 order); 1781 order);
1780 if (page) 1782 if (page)
1781 goto fill_page; 1783 goto fill_page;
@@ -1845,7 +1847,7 @@ bool skb_page_frag_refill(unsigned int sz, struct page_frag *pfrag, gfp_t prio)
1845 gfp_t gfp = prio; 1847 gfp_t gfp = prio;
1846 1848
1847 if (order) 1849 if (order)
1848 gfp |= __GFP_COMP | __GFP_NOWARN; 1850 gfp |= __GFP_COMP | __GFP_NOWARN | __GFP_NORETRY;
1849 pfrag->page = alloc_pages(gfp, order); 1851 pfrag->page = alloc_pages(gfp, order);
1850 if (likely(pfrag->page)) { 1852 if (likely(pfrag->page)) {
1851 pfrag->offset = 0; 1853 pfrag->offset = 0;
diff --git a/net/dccp/ccids/lib/tfrc.c b/net/dccp/ccids/lib/tfrc.c
index c073b81a1f3e..62b5828acde0 100644
--- a/net/dccp/ccids/lib/tfrc.c
+++ b/net/dccp/ccids/lib/tfrc.c
@@ -8,7 +8,7 @@
8#include "tfrc.h" 8#include "tfrc.h"
9 9
10#ifdef CONFIG_IP_DCCP_TFRC_DEBUG 10#ifdef CONFIG_IP_DCCP_TFRC_DEBUG
11static bool tfrc_debug; 11bool tfrc_debug;
12module_param(tfrc_debug, bool, 0644); 12module_param(tfrc_debug, bool, 0644);
13MODULE_PARM_DESC(tfrc_debug, "Enable TFRC debug messages"); 13MODULE_PARM_DESC(tfrc_debug, "Enable TFRC debug messages");
14#endif 14#endif
diff --git a/net/dccp/ccids/lib/tfrc.h b/net/dccp/ccids/lib/tfrc.h
index a3d8f7c76ae0..40ee7d62b652 100644
--- a/net/dccp/ccids/lib/tfrc.h
+++ b/net/dccp/ccids/lib/tfrc.h
@@ -21,6 +21,7 @@
21#include "packet_history.h" 21#include "packet_history.h"
22 22
23#ifdef CONFIG_IP_DCCP_TFRC_DEBUG 23#ifdef CONFIG_IP_DCCP_TFRC_DEBUG
24extern bool tfrc_debug;
24#define tfrc_pr_debug(format, a...) DCCP_PR_DEBUG(tfrc_debug, format, ##a) 25#define tfrc_pr_debug(format, a...) DCCP_PR_DEBUG(tfrc_debug, format, ##a)
25#else 26#else
26#define tfrc_pr_debug(format, a...) 27#define tfrc_pr_debug(format, a...)
diff --git a/net/decnet/af_decnet.c b/net/decnet/af_decnet.c
index 2954dcbca832..4c04848953bd 100644
--- a/net/decnet/af_decnet.c
+++ b/net/decnet/af_decnet.c
@@ -2104,8 +2104,6 @@ static struct notifier_block dn_dev_notifier = {
2104 .notifier_call = dn_device_event, 2104 .notifier_call = dn_device_event,
2105}; 2105};
2106 2106
2107extern int dn_route_rcv(struct sk_buff *, struct net_device *, struct packet_type *, struct net_device *);
2108
2109static struct packet_type dn_dix_packet_type __read_mostly = { 2107static struct packet_type dn_dix_packet_type __read_mostly = {
2110 .type = cpu_to_be16(ETH_P_DNA_RT), 2108 .type = cpu_to_be16(ETH_P_DNA_RT),
2111 .func = dn_route_rcv, 2109 .func = dn_route_rcv,
@@ -2353,9 +2351,6 @@ static const struct proto_ops dn_proto_ops = {
2353 .sendpage = sock_no_sendpage, 2351 .sendpage = sock_no_sendpage,
2354}; 2352};
2355 2353
2356void dn_register_sysctl(void);
2357void dn_unregister_sysctl(void);
2358
2359MODULE_DESCRIPTION("The Linux DECnet Network Protocol"); 2354MODULE_DESCRIPTION("The Linux DECnet Network Protocol");
2360MODULE_AUTHOR("Linux DECnet Project Team"); 2355MODULE_AUTHOR("Linux DECnet Project Team");
2361MODULE_LICENSE("GPL"); 2356MODULE_LICENSE("GPL");
diff --git a/net/hsr/hsr_framereg.c b/net/hsr/hsr_framereg.c
index 327060c6c874..7ae0d7f6dbd0 100644
--- a/net/hsr/hsr_framereg.c
+++ b/net/hsr/hsr_framereg.c
@@ -297,7 +297,7 @@ static bool seq_nr_after(u16 a, u16 b)
297 297
298void hsr_register_frame_in(struct node_entry *node, enum hsr_dev_idx dev_idx) 298void hsr_register_frame_in(struct node_entry *node, enum hsr_dev_idx dev_idx)
299{ 299{
300 if ((dev_idx < 0) || (dev_idx >= HSR_MAX_DEV)) { 300 if ((dev_idx < 0) || (dev_idx >= HSR_MAX_SLAVE)) {
301 WARN_ONCE(1, "%s: Invalid dev_idx (%d)\n", __func__, dev_idx); 301 WARN_ONCE(1, "%s: Invalid dev_idx (%d)\n", __func__, dev_idx);
302 return; 302 return;
303 } 303 }
diff --git a/net/ieee802154/6lowpan.c b/net/ieee802154/6lowpan.c
index 48b25c0af4d0..8edfea5da572 100644
--- a/net/ieee802154/6lowpan.c
+++ b/net/ieee802154/6lowpan.c
@@ -106,7 +106,6 @@ static int lowpan_header_create(struct sk_buff *skb,
106 unsigned short type, const void *_daddr, 106 unsigned short type, const void *_daddr,
107 const void *_saddr, unsigned int len) 107 const void *_saddr, unsigned int len)
108{ 108{
109 struct ipv6hdr *hdr;
110 const u8 *saddr = _saddr; 109 const u8 *saddr = _saddr;
111 const u8 *daddr = _daddr; 110 const u8 *daddr = _daddr;
112 struct ieee802154_addr sa, da; 111 struct ieee802154_addr sa, da;
@@ -117,8 +116,6 @@ static int lowpan_header_create(struct sk_buff *skb,
117 if (type != ETH_P_IPV6) 116 if (type != ETH_P_IPV6)
118 return 0; 117 return 0;
119 118
120 hdr = ipv6_hdr(skb);
121
122 if (!saddr) 119 if (!saddr)
123 saddr = dev->dev_addr; 120 saddr = dev->dev_addr;
124 121
@@ -533,7 +530,27 @@ static struct header_ops lowpan_header_ops = {
533 .create = lowpan_header_create, 530 .create = lowpan_header_create,
534}; 531};
535 532
533static struct lock_class_key lowpan_tx_busylock;
534static struct lock_class_key lowpan_netdev_xmit_lock_key;
535
536static void lowpan_set_lockdep_class_one(struct net_device *dev,
537 struct netdev_queue *txq,
538 void *_unused)
539{
540 lockdep_set_class(&txq->_xmit_lock,
541 &lowpan_netdev_xmit_lock_key);
542}
543
544
545static int lowpan_dev_init(struct net_device *dev)
546{
547 netdev_for_each_tx_queue(dev, lowpan_set_lockdep_class_one, NULL);
548 dev->qdisc_tx_busylock = &lowpan_tx_busylock;
549 return 0;
550}
551
536static const struct net_device_ops lowpan_netdev_ops = { 552static const struct net_device_ops lowpan_netdev_ops = {
553 .ndo_init = lowpan_dev_init,
537 .ndo_start_xmit = lowpan_xmit, 554 .ndo_start_xmit = lowpan_xmit,
538 .ndo_set_mac_address = lowpan_set_address, 555 .ndo_set_mac_address = lowpan_set_address,
539}; 556};
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index ecd2c3f245ce..19ab78aca547 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1296,8 +1296,11 @@ static struct sk_buff *inet_gso_segment(struct sk_buff *skb,
1296 1296
1297 segs = ERR_PTR(-EPROTONOSUPPORT); 1297 segs = ERR_PTR(-EPROTONOSUPPORT);
1298 1298
1299 /* Note : following gso_segment() might change skb->encapsulation */ 1299 if (skb->encapsulation &&
1300 udpfrag = !skb->encapsulation && proto == IPPROTO_UDP; 1300 skb_shinfo(skb)->gso_type & (SKB_GSO_SIT|SKB_GSO_IPIP))
1301 udpfrag = proto == IPPROTO_UDP && encap;
1302 else
1303 udpfrag = proto == IPPROTO_UDP && !skb->encapsulation;
1301 1304
1302 ops = rcu_dereference(inet_offloads[proto]); 1305 ops = rcu_dereference(inet_offloads[proto]);
1303 if (likely(ops && ops->callbacks.gso_segment)) 1306 if (likely(ops && ops->callbacks.gso_segment))
diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index ac2dff3c2c1c..bdbf68bb2e2d 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1443,7 +1443,8 @@ static size_t inet_nlmsg_size(void)
1443 + nla_total_size(4) /* IFA_LOCAL */ 1443 + nla_total_size(4) /* IFA_LOCAL */
1444 + nla_total_size(4) /* IFA_BROADCAST */ 1444 + nla_total_size(4) /* IFA_BROADCAST */
1445 + nla_total_size(IFNAMSIZ) /* IFA_LABEL */ 1445 + nla_total_size(IFNAMSIZ) /* IFA_LABEL */
1446 + nla_total_size(4); /* IFA_FLAGS */ 1446 + nla_total_size(4) /* IFA_FLAGS */
1447 + nla_total_size(sizeof(struct ifa_cacheinfo)); /* IFA_CACHEINFO */
1447} 1448}
1448 1449
1449static inline u32 cstamp_delta(unsigned long cstamp) 1450static inline u32 cstamp_delta(unsigned long cstamp)
diff --git a/net/ipv4/ip_forward.c b/net/ipv4/ip_forward.c
index e9f1217a8afd..f3869c186d97 100644
--- a/net/ipv4/ip_forward.c
+++ b/net/ipv4/ip_forward.c
@@ -39,6 +39,71 @@
39#include <net/route.h> 39#include <net/route.h>
40#include <net/xfrm.h> 40#include <net/xfrm.h>
41 41
42static bool ip_may_fragment(const struct sk_buff *skb)
43{
44 return unlikely((ip_hdr(skb)->frag_off & htons(IP_DF)) == 0) ||
45 !skb->local_df;
46}
47
48static bool ip_exceeds_mtu(const struct sk_buff *skb, unsigned int mtu)
49{
50 if (skb->len <= mtu || skb->local_df)
51 return false;
52
53 if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
54 return false;
55
56 return true;
57}
58
59static bool ip_gso_exceeds_dst_mtu(const struct sk_buff *skb)
60{
61 unsigned int mtu;
62
63 if (skb->local_df || !skb_is_gso(skb))
64 return false;
65
66 mtu = ip_dst_mtu_maybe_forward(skb_dst(skb), true);
67
68 /* if seglen > mtu, do software segmentation for IP fragmentation on
69 * output. DF bit cannot be set since ip_forward would have sent
70 * icmp error.
71 */
72 return skb_gso_network_seglen(skb) > mtu;
73}
74
75/* called if GSO skb needs to be fragmented on forward */
76static int ip_forward_finish_gso(struct sk_buff *skb)
77{
78 struct dst_entry *dst = skb_dst(skb);
79 netdev_features_t features;
80 struct sk_buff *segs;
81 int ret = 0;
82
83 features = netif_skb_dev_features(skb, dst->dev);
84 segs = skb_gso_segment(skb, features & ~NETIF_F_GSO_MASK);
85 if (IS_ERR(segs)) {
86 kfree_skb(skb);
87 return -ENOMEM;
88 }
89
90 consume_skb(skb);
91
92 do {
93 struct sk_buff *nskb = segs->next;
94 int err;
95
96 segs->next = NULL;
97 err = dst_output(segs);
98
99 if (err && ret == 0)
100 ret = err;
101 segs = nskb;
102 } while (segs);
103
104 return ret;
105}
106
42static int ip_forward_finish(struct sk_buff *skb) 107static int ip_forward_finish(struct sk_buff *skb)
43{ 108{
44 struct ip_options *opt = &(IPCB(skb)->opt); 109 struct ip_options *opt = &(IPCB(skb)->opt);
@@ -49,6 +114,9 @@ static int ip_forward_finish(struct sk_buff *skb)
49 if (unlikely(opt->optlen)) 114 if (unlikely(opt->optlen))
50 ip_forward_options(skb); 115 ip_forward_options(skb);
51 116
117 if (ip_gso_exceeds_dst_mtu(skb))
118 return ip_forward_finish_gso(skb);
119
52 return dst_output(skb); 120 return dst_output(skb);
53} 121}
54 122
@@ -91,8 +159,7 @@ int ip_forward(struct sk_buff *skb)
91 159
92 IPCB(skb)->flags |= IPSKB_FORWARDED; 160 IPCB(skb)->flags |= IPSKB_FORWARDED;
93 mtu = ip_dst_mtu_maybe_forward(&rt->dst, true); 161 mtu = ip_dst_mtu_maybe_forward(&rt->dst, true);
94 if (unlikely(skb->len > mtu && !skb_is_gso(skb) && 162 if (!ip_may_fragment(skb) && ip_exceeds_mtu(skb, mtu)) {
95 (ip_hdr(skb)->frag_off & htons(IP_DF))) && !skb->local_df) {
96 IP_INC_STATS(dev_net(rt->dst.dev), IPSTATS_MIB_FRAGFAILS); 163 IP_INC_STATS(dev_net(rt->dst.dev), IPSTATS_MIB_FRAGFAILS);
97 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, 164 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
98 htonl(mtu)); 165 htonl(mtu));
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index 8971780aec7c..73c6b63bba74 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -422,9 +422,6 @@ static void ip_copy_metadata(struct sk_buff *to, struct sk_buff *from)
422 to->tc_index = from->tc_index; 422 to->tc_index = from->tc_index;
423#endif 423#endif
424 nf_copy(to, from); 424 nf_copy(to, from);
425#if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
426 to->nf_trace = from->nf_trace;
427#endif
428#if defined(CONFIG_IP_VS) || defined(CONFIG_IP_VS_MODULE) 425#if defined(CONFIG_IP_VS) || defined(CONFIG_IP_VS_MODULE)
429 to->ipvs_property = from->ipvs_property; 426 to->ipvs_property = from->ipvs_property;
430#endif 427#endif
diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index bd28f386bd02..78a89e61925d 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c
@@ -93,83 +93,32 @@ static void tunnel_dst_reset(struct ip_tunnel *t)
93 tunnel_dst_set(t, NULL); 93 tunnel_dst_set(t, NULL);
94} 94}
95 95
96static void tunnel_dst_reset_all(struct ip_tunnel *t) 96void ip_tunnel_dst_reset_all(struct ip_tunnel *t)
97{ 97{
98 int i; 98 int i;
99 99
100 for_each_possible_cpu(i) 100 for_each_possible_cpu(i)
101 __tunnel_dst_set(per_cpu_ptr(t->dst_cache, i), NULL); 101 __tunnel_dst_set(per_cpu_ptr(t->dst_cache, i), NULL);
102} 102}
103EXPORT_SYMBOL(ip_tunnel_dst_reset_all);
103 104
104static struct dst_entry *tunnel_dst_get(struct ip_tunnel *t) 105static struct rtable *tunnel_rtable_get(struct ip_tunnel *t, u32 cookie)
105{ 106{
106 struct dst_entry *dst; 107 struct dst_entry *dst;
107 108
108 rcu_read_lock(); 109 rcu_read_lock();
109 dst = rcu_dereference(this_cpu_ptr(t->dst_cache)->dst); 110 dst = rcu_dereference(this_cpu_ptr(t->dst_cache)->dst);
110 if (dst) 111 if (dst) {
112 if (dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
113 rcu_read_unlock();
114 tunnel_dst_reset(t);
115 return NULL;
116 }
111 dst_hold(dst); 117 dst_hold(dst);
112 rcu_read_unlock();
113 return dst;
114}
115
116static struct dst_entry *tunnel_dst_check(struct ip_tunnel *t, u32 cookie)
117{
118 struct dst_entry *dst = tunnel_dst_get(t);
119
120 if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
121 tunnel_dst_reset(t);
122 return NULL;
123 }
124
125 return dst;
126}
127
128/* Often modified stats are per cpu, other are shared (netdev->stats) */
129struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
130 struct rtnl_link_stats64 *tot)
131{
132 int i;
133
134 for_each_possible_cpu(i) {
135 const struct pcpu_sw_netstats *tstats =
136 per_cpu_ptr(dev->tstats, i);
137 u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
138 unsigned int start;
139
140 do {
141 start = u64_stats_fetch_begin_bh(&tstats->syncp);
142 rx_packets = tstats->rx_packets;
143 tx_packets = tstats->tx_packets;
144 rx_bytes = tstats->rx_bytes;
145 tx_bytes = tstats->tx_bytes;
146 } while (u64_stats_fetch_retry_bh(&tstats->syncp, start));
147
148 tot->rx_packets += rx_packets;
149 tot->tx_packets += tx_packets;
150 tot->rx_bytes += rx_bytes;
151 tot->tx_bytes += tx_bytes;
152 } 118 }
153 119 rcu_read_unlock();
154 tot->multicast = dev->stats.multicast; 120 return (struct rtable *)dst;
155
156 tot->rx_crc_errors = dev->stats.rx_crc_errors;
157 tot->rx_fifo_errors = dev->stats.rx_fifo_errors;
158 tot->rx_length_errors = dev->stats.rx_length_errors;
159 tot->rx_frame_errors = dev->stats.rx_frame_errors;
160 tot->rx_errors = dev->stats.rx_errors;
161
162 tot->tx_fifo_errors = dev->stats.tx_fifo_errors;
163 tot->tx_carrier_errors = dev->stats.tx_carrier_errors;
164 tot->tx_dropped = dev->stats.tx_dropped;
165 tot->tx_aborted_errors = dev->stats.tx_aborted_errors;
166 tot->tx_errors = dev->stats.tx_errors;
167
168 tot->collisions = dev->stats.collisions;
169
170 return tot;
171} 121}
172EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64);
173 122
174static bool ip_tunnel_key_match(const struct ip_tunnel_parm *p, 123static bool ip_tunnel_key_match(const struct ip_tunnel_parm *p,
175 __be16 flags, __be32 key) 124 __be16 flags, __be32 key)
@@ -584,7 +533,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
584 struct flowi4 fl4; 533 struct flowi4 fl4;
585 u8 tos, ttl; 534 u8 tos, ttl;
586 __be16 df; 535 __be16 df;
587 struct rtable *rt = NULL; /* Route to the other host */ 536 struct rtable *rt; /* Route to the other host */
588 unsigned int max_headroom; /* The extra header space needed */ 537 unsigned int max_headroom; /* The extra header space needed */
589 __be32 dst; 538 __be32 dst;
590 int err; 539 int err;
@@ -657,8 +606,7 @@ void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
657 init_tunnel_flow(&fl4, protocol, dst, tnl_params->saddr, 606 init_tunnel_flow(&fl4, protocol, dst, tnl_params->saddr,
658 tunnel->parms.o_key, RT_TOS(tos), tunnel->parms.link); 607 tunnel->parms.o_key, RT_TOS(tos), tunnel->parms.link);
659 608
660 if (connected) 609 rt = connected ? tunnel_rtable_get(tunnel, 0) : NULL;
661 rt = (struct rtable *)tunnel_dst_check(tunnel, 0);
662 610
663 if (!rt) { 611 if (!rt) {
664 rt = ip_route_output_key(tunnel->net, &fl4); 612 rt = ip_route_output_key(tunnel->net, &fl4);
@@ -766,7 +714,7 @@ static void ip_tunnel_update(struct ip_tunnel_net *itn,
766 if (set_mtu) 714 if (set_mtu)
767 dev->mtu = mtu; 715 dev->mtu = mtu;
768 } 716 }
769 tunnel_dst_reset_all(t); 717 ip_tunnel_dst_reset_all(t);
770 netdev_state_change(dev); 718 netdev_state_change(dev);
771} 719}
772 720
@@ -1095,7 +1043,7 @@ void ip_tunnel_uninit(struct net_device *dev)
1095 if (itn->fb_tunnel_dev != dev) 1043 if (itn->fb_tunnel_dev != dev)
1096 ip_tunnel_del(netdev_priv(dev)); 1044 ip_tunnel_del(netdev_priv(dev));
1097 1045
1098 tunnel_dst_reset_all(tunnel); 1046 ip_tunnel_dst_reset_all(tunnel);
1099} 1047}
1100EXPORT_SYMBOL_GPL(ip_tunnel_uninit); 1048EXPORT_SYMBOL_GPL(ip_tunnel_uninit);
1101 1049
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index 6156f4ef5e91..6f847dd56dbc 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -108,7 +108,6 @@ int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto)
108 nf_reset(skb); 108 nf_reset(skb);
109 secpath_reset(skb); 109 secpath_reset(skb);
110 skb_clear_hash_if_not_l4(skb); 110 skb_clear_hash_if_not_l4(skb);
111 skb_dst_drop(skb);
112 skb->vlan_tci = 0; 111 skb->vlan_tci = 0;
113 skb_set_queue_mapping(skb, 0); 112 skb_set_queue_mapping(skb, 0);
114 skb->pkt_type = PACKET_HOST; 113 skb->pkt_type = PACKET_HOST;
@@ -148,3 +147,49 @@ error:
148 return ERR_PTR(err); 147 return ERR_PTR(err);
149} 148}
150EXPORT_SYMBOL_GPL(iptunnel_handle_offloads); 149EXPORT_SYMBOL_GPL(iptunnel_handle_offloads);
150
151/* Often modified stats are per cpu, other are shared (netdev->stats) */
152struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
153 struct rtnl_link_stats64 *tot)
154{
155 int i;
156
157 for_each_possible_cpu(i) {
158 const struct pcpu_sw_netstats *tstats =
159 per_cpu_ptr(dev->tstats, i);
160 u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
161 unsigned int start;
162
163 do {
164 start = u64_stats_fetch_begin_bh(&tstats->syncp);
165 rx_packets = tstats->rx_packets;
166 tx_packets = tstats->tx_packets;
167 rx_bytes = tstats->rx_bytes;
168 tx_bytes = tstats->tx_bytes;
169 } while (u64_stats_fetch_retry_bh(&tstats->syncp, start));
170
171 tot->rx_packets += rx_packets;
172 tot->tx_packets += tx_packets;
173 tot->rx_bytes += rx_bytes;
174 tot->tx_bytes += tx_bytes;
175 }
176
177 tot->multicast = dev->stats.multicast;
178
179 tot->rx_crc_errors = dev->stats.rx_crc_errors;
180 tot->rx_fifo_errors = dev->stats.rx_fifo_errors;
181 tot->rx_length_errors = dev->stats.rx_length_errors;
182 tot->rx_frame_errors = dev->stats.rx_frame_errors;
183 tot->rx_errors = dev->stats.rx_errors;
184
185 tot->tx_fifo_errors = dev->stats.tx_fifo_errors;
186 tot->tx_carrier_errors = dev->stats.tx_carrier_errors;
187 tot->tx_dropped = dev->stats.tx_dropped;
188 tot->tx_aborted_errors = dev->stats.tx_aborted_errors;
189 tot->tx_errors = dev->stats.tx_errors;
190
191 tot->collisions = dev->stats.collisions;
192
193 return tot;
194}
195EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64);
diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index efa1138fa523..b3e86ea7b71b 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -273,7 +273,7 @@ static int __init ic_open_devs(void)
273 273
274 msleep(1); 274 msleep(1);
275 275
276 if time_before(jiffies, next_msg) 276 if (time_before(jiffies, next_msg))
277 continue; 277 continue;
278 278
279 elapsed = jiffies_to_msecs(jiffies - start); 279 elapsed = jiffies_to_msecs(jiffies - start);
diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig
index 81c6910cfa92..a26ce035e3fa 100644
--- a/net/ipv4/netfilter/Kconfig
+++ b/net/ipv4/netfilter/Kconfig
@@ -61,6 +61,11 @@ config NFT_CHAIN_NAT_IPV4
61 packet transformations such as the source, destination address and 61 packet transformations such as the source, destination address and
62 source and destination ports. 62 source and destination ports.
63 63
64config NFT_REJECT_IPV4
65 depends on NF_TABLES_IPV4
66 default NFT_REJECT
67 tristate
68
64config NF_TABLES_ARP 69config NF_TABLES_ARP
65 depends on NF_TABLES 70 depends on NF_TABLES
66 tristate "ARP nf_tables support" 71 tristate "ARP nf_tables support"
diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile
index c16be9d58420..90b82405331e 100644
--- a/net/ipv4/netfilter/Makefile
+++ b/net/ipv4/netfilter/Makefile
@@ -30,6 +30,7 @@ obj-$(CONFIG_NF_NAT_PROTO_GRE) += nf_nat_proto_gre.o
30obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o 30obj-$(CONFIG_NF_TABLES_IPV4) += nf_tables_ipv4.o
31obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o 31obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV4) += nft_chain_route_ipv4.o
32obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o 32obj-$(CONFIG_NFT_CHAIN_NAT_IPV4) += nft_chain_nat_ipv4.o
33obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o
33obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o 34obj-$(CONFIG_NF_TABLES_ARP) += nf_tables_arp.o
34 35
35# generic IP tables 36# generic IP tables
diff --git a/net/ipv4/netfilter/nf_nat_h323.c b/net/ipv4/netfilter/nf_nat_h323.c
index 9eea059dd621..574f7ebba0b6 100644
--- a/net/ipv4/netfilter/nf_nat_h323.c
+++ b/net/ipv4/netfilter/nf_nat_h323.c
@@ -229,7 +229,10 @@ static int nat_rtp_rtcp(struct sk_buff *skb, struct nf_conn *ct,
229 ret = nf_ct_expect_related(rtcp_exp); 229 ret = nf_ct_expect_related(rtcp_exp);
230 if (ret == 0) 230 if (ret == 0)
231 break; 231 break;
232 else if (ret != -EBUSY) { 232 else if (ret == -EBUSY) {
233 nf_ct_unexpect_related(rtp_exp);
234 continue;
235 } else if (ret < 0) {
233 nf_ct_unexpect_related(rtp_exp); 236 nf_ct_unexpect_related(rtp_exp);
234 nated_port = 0; 237 nated_port = 0;
235 break; 238 break;
diff --git a/net/ipv4/netfilter/nf_nat_snmp_basic.c b/net/ipv4/netfilter/nf_nat_snmp_basic.c
index d551e31b416e..7c676671329d 100644
--- a/net/ipv4/netfilter/nf_nat_snmp_basic.c
+++ b/net/ipv4/netfilter/nf_nat_snmp_basic.c
@@ -1198,8 +1198,8 @@ static int snmp_translate(struct nf_conn *ct,
1198 map.to = NOCT1(&ct->tuplehash[!dir].tuple.dst.u3.ip); 1198 map.to = NOCT1(&ct->tuplehash[!dir].tuple.dst.u3.ip);
1199 } else { 1199 } else {
1200 /* DNAT replies */ 1200 /* DNAT replies */
1201 map.from = NOCT1(&ct->tuplehash[dir].tuple.src.u3.ip); 1201 map.from = NOCT1(&ct->tuplehash[!dir].tuple.src.u3.ip);
1202 map.to = NOCT1(&ct->tuplehash[!dir].tuple.dst.u3.ip); 1202 map.to = NOCT1(&ct->tuplehash[dir].tuple.dst.u3.ip);
1203 } 1203 }
1204 1204
1205 if (map.from == map.to) 1205 if (map.from == map.to)
diff --git a/net/ipv4/netfilter/nft_reject_ipv4.c b/net/ipv4/netfilter/nft_reject_ipv4.c
new file mode 100644
index 000000000000..e79718a382f2
--- /dev/null
+++ b/net/ipv4/netfilter/nft_reject_ipv4.c
@@ -0,0 +1,75 @@
1/*
2 * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
3 * Copyright (c) 2013 Eric Leblond <eric@regit.org>
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License version 2 as
7 * published by the Free Software Foundation.
8 *
9 * Development of this code funded by Astaro AG (http://www.astaro.com/)
10 */
11
12#include <linux/kernel.h>
13#include <linux/init.h>
14#include <linux/module.h>
15#include <linux/netlink.h>
16#include <linux/netfilter.h>
17#include <linux/netfilter/nf_tables.h>
18#include <net/netfilter/nf_tables.h>
19#include <net/icmp.h>
20#include <net/netfilter/ipv4/nf_reject.h>
21#include <net/netfilter/nft_reject.h>
22
23void nft_reject_ipv4_eval(const struct nft_expr *expr,
24 struct nft_data data[NFT_REG_MAX + 1],
25 const struct nft_pktinfo *pkt)
26{
27 struct nft_reject *priv = nft_expr_priv(expr);
28
29 switch (priv->type) {
30 case NFT_REJECT_ICMP_UNREACH:
31 nf_send_unreach(pkt->skb, priv->icmp_code);
32 break;
33 case NFT_REJECT_TCP_RST:
34 nf_send_reset(pkt->skb, pkt->ops->hooknum);
35 break;
36 }
37
38 data[NFT_REG_VERDICT].verdict = NF_DROP;
39}
40EXPORT_SYMBOL_GPL(nft_reject_ipv4_eval);
41
42static struct nft_expr_type nft_reject_ipv4_type;
43static const struct nft_expr_ops nft_reject_ipv4_ops = {
44 .type = &nft_reject_ipv4_type,
45 .size = NFT_EXPR_SIZE(sizeof(struct nft_reject)),
46 .eval = nft_reject_ipv4_eval,
47 .init = nft_reject_init,
48 .dump = nft_reject_dump,
49};
50
51static struct nft_expr_type nft_reject_ipv4_type __read_mostly = {
52 .family = NFPROTO_IPV4,
53 .name = "reject",
54 .ops = &nft_reject_ipv4_ops,
55 .policy = nft_reject_policy,
56 .maxattr = NFTA_REJECT_MAX,
57 .owner = THIS_MODULE,
58};
59
60static int __init nft_reject_ipv4_module_init(void)
61{
62 return nft_register_expr(&nft_reject_ipv4_type);
63}
64
65static void __exit nft_reject_ipv4_module_exit(void)
66{
67 nft_unregister_expr(&nft_reject_ipv4_type);
68}
69
70module_init(nft_reject_ipv4_module_init);
71module_exit(nft_reject_ipv4_module_exit);
72
73MODULE_LICENSE("GPL");
74MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
75MODULE_ALIAS_NFT_AF_EXPR(AF_INET, "reject");
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 25071b48921c..4c011ec69ed4 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1597,6 +1597,7 @@ static int __mkroute_input(struct sk_buff *skb,
1597 rth->rt_gateway = 0; 1597 rth->rt_gateway = 0;
1598 rth->rt_uses_gateway = 0; 1598 rth->rt_uses_gateway = 0;
1599 INIT_LIST_HEAD(&rth->rt_uncached); 1599 INIT_LIST_HEAD(&rth->rt_uncached);
1600 RT_CACHE_STAT_INC(in_slow_tot);
1600 1601
1601 rth->dst.input = ip_forward; 1602 rth->dst.input = ip_forward;
1602 rth->dst.output = ip_output; 1603 rth->dst.output = ip_output;
@@ -1695,10 +1696,11 @@ static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
1695 fl4.daddr = daddr; 1696 fl4.daddr = daddr;
1696 fl4.saddr = saddr; 1697 fl4.saddr = saddr;
1697 err = fib_lookup(net, &fl4, &res); 1698 err = fib_lookup(net, &fl4, &res);
1698 if (err != 0) 1699 if (err != 0) {
1700 if (!IN_DEV_FORWARD(in_dev))
1701 err = -EHOSTUNREACH;
1699 goto no_route; 1702 goto no_route;
1700 1703 }
1701 RT_CACHE_STAT_INC(in_slow_tot);
1702 1704
1703 if (res.type == RTN_BROADCAST) 1705 if (res.type == RTN_BROADCAST)
1704 goto brd_input; 1706 goto brd_input;
@@ -1712,8 +1714,10 @@ static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
1712 goto local_input; 1714 goto local_input;
1713 } 1715 }
1714 1716
1715 if (!IN_DEV_FORWARD(in_dev)) 1717 if (!IN_DEV_FORWARD(in_dev)) {
1718 err = -EHOSTUNREACH;
1716 goto no_route; 1719 goto no_route;
1720 }
1717 if (res.type != RTN_UNICAST) 1721 if (res.type != RTN_UNICAST)
1718 goto martian_destination; 1722 goto martian_destination;
1719 1723
@@ -1768,6 +1772,7 @@ local_input:
1768 rth->rt_gateway = 0; 1772 rth->rt_gateway = 0;
1769 rth->rt_uses_gateway = 0; 1773 rth->rt_uses_gateway = 0;
1770 INIT_LIST_HEAD(&rth->rt_uncached); 1774 INIT_LIST_HEAD(&rth->rt_uncached);
1775 RT_CACHE_STAT_INC(in_slow_tot);
1771 if (res.type == RTN_UNREACHABLE) { 1776 if (res.type == RTN_UNREACHABLE) {
1772 rth->dst.input= ip_error; 1777 rth->dst.input= ip_error;
1773 rth->dst.error= -err; 1778 rth->dst.error= -err;
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 4475b3bb494d..97c8f5620c43 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1044,7 +1044,8 @@ void tcp_free_fastopen_req(struct tcp_sock *tp)
1044 } 1044 }
1045} 1045}
1046 1046
1047static int tcp_sendmsg_fastopen(struct sock *sk, struct msghdr *msg, int *size) 1047static int tcp_sendmsg_fastopen(struct sock *sk, struct msghdr *msg,
1048 int *copied, size_t size)
1048{ 1049{
1049 struct tcp_sock *tp = tcp_sk(sk); 1050 struct tcp_sock *tp = tcp_sk(sk);
1050 int err, flags; 1051 int err, flags;
@@ -1059,11 +1060,12 @@ static int tcp_sendmsg_fastopen(struct sock *sk, struct msghdr *msg, int *size)
1059 if (unlikely(tp->fastopen_req == NULL)) 1060 if (unlikely(tp->fastopen_req == NULL))
1060 return -ENOBUFS; 1061 return -ENOBUFS;
1061 tp->fastopen_req->data = msg; 1062 tp->fastopen_req->data = msg;
1063 tp->fastopen_req->size = size;
1062 1064
1063 flags = (msg->msg_flags & MSG_DONTWAIT) ? O_NONBLOCK : 0; 1065 flags = (msg->msg_flags & MSG_DONTWAIT) ? O_NONBLOCK : 0;
1064 err = __inet_stream_connect(sk->sk_socket, msg->msg_name, 1066 err = __inet_stream_connect(sk->sk_socket, msg->msg_name,
1065 msg->msg_namelen, flags); 1067 msg->msg_namelen, flags);
1066 *size = tp->fastopen_req->copied; 1068 *copied = tp->fastopen_req->copied;
1067 tcp_free_fastopen_req(tp); 1069 tcp_free_fastopen_req(tp);
1068 return err; 1070 return err;
1069} 1071}
@@ -1083,7 +1085,7 @@ int tcp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
1083 1085
1084 flags = msg->msg_flags; 1086 flags = msg->msg_flags;
1085 if (flags & MSG_FASTOPEN) { 1087 if (flags & MSG_FASTOPEN) {
1086 err = tcp_sendmsg_fastopen(sk, msg, &copied_syn); 1088 err = tcp_sendmsg_fastopen(sk, msg, &copied_syn, size);
1087 if (err == -EINPROGRESS && copied_syn > 0) 1089 if (err == -EINPROGRESS && copied_syn > 0)
1088 goto out; 1090 goto out;
1089 else if (err) 1091 else if (err)
@@ -2229,7 +2231,7 @@ adjudge_to_death:
2229 /* This is a (useful) BSD violating of the RFC. There is a 2231 /* This is a (useful) BSD violating of the RFC. There is a
2230 * problem with TCP as specified in that the other end could 2232 * problem with TCP as specified in that the other end could
2231 * keep a socket open forever with no application left this end. 2233 * keep a socket open forever with no application left this end.
2232 * We use a 3 minute timeout (about the same as BSD) then kill 2234 * We use a 1 minute timeout (about the same as BSD) then kill
2233 * our end. If they send after that then tough - BUT: long enough 2235 * our end. If they send after that then tough - BUT: long enough
2234 * that we won't make the old 4*rto = almost no time - whoops 2236 * that we won't make the old 4*rto = almost no time - whoops
2235 * reset mistake. 2237 * reset mistake.
diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index ad37bf18ae4b..2388275adb9b 100644
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -290,8 +290,7 @@ bool tcp_is_cwnd_limited(const struct sock *sk, u32 in_flight)
290 left = tp->snd_cwnd - in_flight; 290 left = tp->snd_cwnd - in_flight;
291 if (sk_can_gso(sk) && 291 if (sk_can_gso(sk) &&
292 left * sysctl_tcp_tso_win_divisor < tp->snd_cwnd && 292 left * sysctl_tcp_tso_win_divisor < tp->snd_cwnd &&
293 left * tp->mss_cache < sk->sk_gso_max_size && 293 left < tp->xmit_size_goal_segs)
294 left < sk->sk_gso_max_segs)
295 return true; 294 return true;
296 return left <= tcp_max_tso_deferred_mss(tp); 295 return left <= tcp_max_tso_deferred_mss(tp);
297} 296}
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 65cf90e063d5..eeaac399420d 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -671,6 +671,7 @@ static void tcp_rtt_estimator(struct sock *sk, const __u32 mrtt)
671{ 671{
672 struct tcp_sock *tp = tcp_sk(sk); 672 struct tcp_sock *tp = tcp_sk(sk);
673 long m = mrtt; /* RTT */ 673 long m = mrtt; /* RTT */
674 u32 srtt = tp->srtt;
674 675
675 /* The following amusing code comes from Jacobson's 676 /* The following amusing code comes from Jacobson's
676 * article in SIGCOMM '88. Note that rtt and mdev 677 * article in SIGCOMM '88. Note that rtt and mdev
@@ -688,11 +689,9 @@ static void tcp_rtt_estimator(struct sock *sk, const __u32 mrtt)
688 * does not matter how to _calculate_ it. Seems, it was trap 689 * does not matter how to _calculate_ it. Seems, it was trap
689 * that VJ failed to avoid. 8) 690 * that VJ failed to avoid. 8)
690 */ 691 */
691 if (m == 0) 692 if (srtt != 0) {
692 m = 1; 693 m -= (srtt >> 3); /* m is now error in rtt est */
693 if (tp->srtt != 0) { 694 srtt += m; /* rtt = 7/8 rtt + 1/8 new */
694 m -= (tp->srtt >> 3); /* m is now error in rtt est */
695 tp->srtt += m; /* rtt = 7/8 rtt + 1/8 new */
696 if (m < 0) { 695 if (m < 0) {
697 m = -m; /* m is now abs(error) */ 696 m = -m; /* m is now abs(error) */
698 m -= (tp->mdev >> 2); /* similar update on mdev */ 697 m -= (tp->mdev >> 2); /* similar update on mdev */
@@ -723,11 +722,12 @@ static void tcp_rtt_estimator(struct sock *sk, const __u32 mrtt)
723 } 722 }
724 } else { 723 } else {
725 /* no previous measure. */ 724 /* no previous measure. */
726 tp->srtt = m << 3; /* take the measured time to be rtt */ 725 srtt = m << 3; /* take the measured time to be rtt */
727 tp->mdev = m << 1; /* make sure rto = 3*rtt */ 726 tp->mdev = m << 1; /* make sure rto = 3*rtt */
728 tp->mdev_max = tp->rttvar = max(tp->mdev, tcp_rto_min(sk)); 727 tp->mdev_max = tp->rttvar = max(tp->mdev, tcp_rto_min(sk));
729 tp->rtt_seq = tp->snd_nxt; 728 tp->rtt_seq = tp->snd_nxt;
730 } 729 }
730 tp->srtt = max(1U, srtt);
731} 731}
732 732
733/* Set the sk_pacing_rate to allow proper sizing of TSO packets. 733/* Set the sk_pacing_rate to allow proper sizing of TSO packets.
@@ -746,8 +746,10 @@ static void tcp_update_pacing_rate(struct sock *sk)
746 746
747 rate *= max(tp->snd_cwnd, tp->packets_out); 747 rate *= max(tp->snd_cwnd, tp->packets_out);
748 748
749 /* Correction for small srtt : minimum srtt being 8 (1 jiffy << 3), 749 /* Correction for small srtt and scheduling constraints.
750 * be conservative and assume srtt = 1 (125 us instead of 1.25 ms) 750 * For small rtt, consider noise is too high, and use
751 * the minimal value (srtt = 1 -> 125 us for HZ=1000)
752 *
751 * We probably need usec resolution in the future. 753 * We probably need usec resolution in the future.
752 * Note: This also takes care of possible srtt=0 case, 754 * Note: This also takes care of possible srtt=0 case,
753 * when tcp_rtt_estimator() was not yet called. 755 * when tcp_rtt_estimator() was not yet called.
@@ -1943,8 +1945,9 @@ void tcp_enter_loss(struct sock *sk, int how)
1943 if (skb == tcp_send_head(sk)) 1945 if (skb == tcp_send_head(sk))
1944 break; 1946 break;
1945 1947
1946 if (TCP_SKB_CB(skb)->sacked & TCPCB_RETRANS) 1948 if (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_RETRANS)
1947 tp->undo_marker = 0; 1949 tp->undo_marker = 0;
1950
1948 TCP_SKB_CB(skb)->sacked &= (~TCPCB_TAGBITS)|TCPCB_SACKED_ACKED; 1951 TCP_SKB_CB(skb)->sacked &= (~TCPCB_TAGBITS)|TCPCB_SACKED_ACKED;
1949 if (!(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED) || how) { 1952 if (!(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED) || how) {
1950 TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_ACKED; 1953 TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_ACKED;
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 03d26b85eab8..f0eb4e337ec8 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -698,7 +698,8 @@ static void tcp_tsq_handler(struct sock *sk)
698 if ((1 << sk->sk_state) & 698 if ((1 << sk->sk_state) &
699 (TCPF_ESTABLISHED | TCPF_FIN_WAIT1 | TCPF_CLOSING | 699 (TCPF_ESTABLISHED | TCPF_FIN_WAIT1 | TCPF_CLOSING |
700 TCPF_CLOSE_WAIT | TCPF_LAST_ACK)) 700 TCPF_CLOSE_WAIT | TCPF_LAST_ACK))
701 tcp_write_xmit(sk, tcp_current_mss(sk), 0, 0, GFP_ATOMIC); 701 tcp_write_xmit(sk, tcp_current_mss(sk), tcp_sk(sk)->nonagle,
702 0, GFP_ATOMIC);
702} 703}
703/* 704/*
704 * One tasklet per cpu tries to send more skbs. 705 * One tasklet per cpu tries to send more skbs.
@@ -863,8 +864,8 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it,
863 864
864 if (unlikely(skb->fclone == SKB_FCLONE_ORIG && 865 if (unlikely(skb->fclone == SKB_FCLONE_ORIG &&
865 fclone->fclone == SKB_FCLONE_CLONE)) 866 fclone->fclone == SKB_FCLONE_CLONE))
866 NET_INC_STATS_BH(sock_net(sk), 867 NET_INC_STATS(sock_net(sk),
867 LINUX_MIB_TCPSPURIOUS_RTX_HOSTQUEUES); 868 LINUX_MIB_TCPSPURIOUS_RTX_HOSTQUEUES);
868 869
869 if (unlikely(skb_cloned(skb))) 870 if (unlikely(skb_cloned(skb)))
870 skb = pskb_copy(skb, gfp_mask); 871 skb = pskb_copy(skb, gfp_mask);
@@ -1904,7 +1905,15 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
1904 1905
1905 if (atomic_read(&sk->sk_wmem_alloc) > limit) { 1906 if (atomic_read(&sk->sk_wmem_alloc) > limit) {
1906 set_bit(TSQ_THROTTLED, &tp->tsq_flags); 1907 set_bit(TSQ_THROTTLED, &tp->tsq_flags);
1907 break; 1908 /* It is possible TX completion already happened
1909 * before we set TSQ_THROTTLED, so we must
1910 * test again the condition.
1911 * We abuse smp_mb__after_clear_bit() because
1912 * there is no smp_mb__after_set_bit() yet
1913 */
1914 smp_mb__after_clear_bit();
1915 if (atomic_read(&sk->sk_wmem_alloc) > limit)
1916 break;
1908 } 1917 }
1909 1918
1910 limit = mss_now; 1919 limit = mss_now;
@@ -1977,7 +1986,7 @@ bool tcp_schedule_loss_probe(struct sock *sk)
1977 /* Schedule a loss probe in 2*RTT for SACK capable connections 1986 /* Schedule a loss probe in 2*RTT for SACK capable connections
1978 * in Open state, that are either limited by cwnd or application. 1987 * in Open state, that are either limited by cwnd or application.
1979 */ 1988 */
1980 if (sysctl_tcp_early_retrans < 3 || !rtt || !tp->packets_out || 1989 if (sysctl_tcp_early_retrans < 3 || !tp->srtt || !tp->packets_out ||
1981 !tcp_is_sack(tp) || inet_csk(sk)->icsk_ca_state != TCP_CA_Open) 1990 !tcp_is_sack(tp) || inet_csk(sk)->icsk_ca_state != TCP_CA_Open)
1982 return false; 1991 return false;
1983 1992
@@ -2328,6 +2337,7 @@ int __tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
2328 struct tcp_sock *tp = tcp_sk(sk); 2337 struct tcp_sock *tp = tcp_sk(sk);
2329 struct inet_connection_sock *icsk = inet_csk(sk); 2338 struct inet_connection_sock *icsk = inet_csk(sk);
2330 unsigned int cur_mss; 2339 unsigned int cur_mss;
2340 int err;
2331 2341
2332 /* Inconslusive MTU probe */ 2342 /* Inconslusive MTU probe */
2333 if (icsk->icsk_mtup.probe_size) { 2343 if (icsk->icsk_mtup.probe_size) {
@@ -2391,11 +2401,15 @@ int __tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
2391 skb_headroom(skb) >= 0xFFFF)) { 2401 skb_headroom(skb) >= 0xFFFF)) {
2392 struct sk_buff *nskb = __pskb_copy(skb, MAX_TCP_HEADER, 2402 struct sk_buff *nskb = __pskb_copy(skb, MAX_TCP_HEADER,
2393 GFP_ATOMIC); 2403 GFP_ATOMIC);
2394 return nskb ? tcp_transmit_skb(sk, nskb, 0, GFP_ATOMIC) : 2404 err = nskb ? tcp_transmit_skb(sk, nskb, 0, GFP_ATOMIC) :
2395 -ENOBUFS; 2405 -ENOBUFS;
2396 } else { 2406 } else {
2397 return tcp_transmit_skb(sk, skb, 1, GFP_ATOMIC); 2407 err = tcp_transmit_skb(sk, skb, 1, GFP_ATOMIC);
2398 } 2408 }
2409
2410 if (likely(!err))
2411 TCP_SKB_CB(skb)->sacked |= TCPCB_EVER_RETRANS;
2412 return err;
2399} 2413}
2400 2414
2401int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb) 2415int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
@@ -2899,7 +2913,12 @@ static int tcp_send_syn_data(struct sock *sk, struct sk_buff *syn)
2899 space = __tcp_mtu_to_mss(sk, inet_csk(sk)->icsk_pmtu_cookie) - 2913 space = __tcp_mtu_to_mss(sk, inet_csk(sk)->icsk_pmtu_cookie) -
2900 MAX_TCP_OPTION_SPACE; 2914 MAX_TCP_OPTION_SPACE;
2901 2915
2902 syn_data = skb_copy_expand(syn, skb_headroom(syn), space, 2916 space = min_t(size_t, space, fo->size);
2917
2918 /* limit to order-0 allocations */
2919 space = min_t(size_t, space, SKB_MAX_HEAD(MAX_TCP_HEADER));
2920
2921 syn_data = skb_copy_expand(syn, MAX_TCP_HEADER, space,
2903 sk->sk_allocation); 2922 sk->sk_allocation);
2904 if (syn_data == NULL) 2923 if (syn_data == NULL)
2905 goto fallback; 2924 goto fallback;
diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
index 25f5cee3a08a..88b4023ecfcf 100644
--- a/net/ipv4/udp_offload.c
+++ b/net/ipv4/udp_offload.c
@@ -17,6 +17,8 @@
17static DEFINE_SPINLOCK(udp_offload_lock); 17static DEFINE_SPINLOCK(udp_offload_lock);
18static struct udp_offload_priv __rcu *udp_offload_base __read_mostly; 18static struct udp_offload_priv __rcu *udp_offload_base __read_mostly;
19 19
20#define udp_deref_protected(X) rcu_dereference_protected(X, lockdep_is_held(&udp_offload_lock))
21
20struct udp_offload_priv { 22struct udp_offload_priv {
21 struct udp_offload *offload; 23 struct udp_offload *offload;
22 struct rcu_head rcu; 24 struct rcu_head rcu;
@@ -100,8 +102,7 @@ out:
100 102
101int udp_add_offload(struct udp_offload *uo) 103int udp_add_offload(struct udp_offload *uo)
102{ 104{
103 struct udp_offload_priv __rcu **head = &udp_offload_base; 105 struct udp_offload_priv *new_offload = kzalloc(sizeof(*new_offload), GFP_ATOMIC);
104 struct udp_offload_priv *new_offload = kzalloc(sizeof(*new_offload), GFP_KERNEL);
105 106
106 if (!new_offload) 107 if (!new_offload)
107 return -ENOMEM; 108 return -ENOMEM;
@@ -109,8 +110,8 @@ int udp_add_offload(struct udp_offload *uo)
109 new_offload->offload = uo; 110 new_offload->offload = uo;
110 111
111 spin_lock(&udp_offload_lock); 112 spin_lock(&udp_offload_lock);
112 rcu_assign_pointer(new_offload->next, rcu_dereference(*head)); 113 new_offload->next = udp_offload_base;
113 rcu_assign_pointer(*head, new_offload); 114 rcu_assign_pointer(udp_offload_base, new_offload);
114 spin_unlock(&udp_offload_lock); 115 spin_unlock(&udp_offload_lock);
115 116
116 return 0; 117 return 0;
@@ -130,12 +131,12 @@ void udp_del_offload(struct udp_offload *uo)
130 131
131 spin_lock(&udp_offload_lock); 132 spin_lock(&udp_offload_lock);
132 133
133 uo_priv = rcu_dereference(*head); 134 uo_priv = udp_deref_protected(*head);
134 for (; uo_priv != NULL; 135 for (; uo_priv != NULL;
135 uo_priv = rcu_dereference(*head)) { 136 uo_priv = udp_deref_protected(*head)) {
136
137 if (uo_priv->offload == uo) { 137 if (uo_priv->offload == uo) {
138 rcu_assign_pointer(*head, rcu_dereference(uo_priv->next)); 138 rcu_assign_pointer(*head,
139 udp_deref_protected(uo_priv->next));
139 goto unlock; 140 goto unlock;
140 } 141 }
141 head = &uo_priv->next; 142 head = &uo_priv->next;
diff --git a/net/ipv6/Kconfig b/net/ipv6/Kconfig
index d92e5586783e..438a73aa777c 100644
--- a/net/ipv6/Kconfig
+++ b/net/ipv6/Kconfig
@@ -138,6 +138,7 @@ config INET6_XFRM_MODE_ROUTEOPTIMIZATION
138config IPV6_VTI 138config IPV6_VTI
139tristate "Virtual (secure) IPv6: tunneling" 139tristate "Virtual (secure) IPv6: tunneling"
140 select IPV6_TUNNEL 140 select IPV6_TUNNEL
141 select NET_IP_TUNNEL
141 depends on INET6_XFRM_MODE_TUNNEL 142 depends on INET6_XFRM_MODE_TUNNEL
142 ---help--- 143 ---help---
143 Tunneling means encapsulating data of one protocol type within 144 Tunneling means encapsulating data of one protocol type within
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index ad235690684c..fdbfeca36d63 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -2783,6 +2783,8 @@ static void addrconf_gre_config(struct net_device *dev)
2783 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0); 2783 ipv6_addr_set(&addr, htonl(0xFE800000), 0, 0, 0);
2784 if (!ipv6_generate_eui64(addr.s6_addr + 8, dev)) 2784 if (!ipv6_generate_eui64(addr.s6_addr + 8, dev))
2785 addrconf_add_linklocal(idev, &addr); 2785 addrconf_add_linklocal(idev, &addr);
2786 else
2787 addrconf_prefix_route(&addr, 64, dev, 0, 0);
2786} 2788}
2787#endif 2789#endif
2788 2790
diff --git a/net/ipv6/exthdrs_core.c b/net/ipv6/exthdrs_core.c
index 140748debc4a..8af3eb57f438 100644
--- a/net/ipv6/exthdrs_core.c
+++ b/net/ipv6/exthdrs_core.c
@@ -212,7 +212,7 @@ int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset,
212 found = (nexthdr == target); 212 found = (nexthdr == target);
213 213
214 if ((!ipv6_ext_hdr(nexthdr)) || nexthdr == NEXTHDR_NONE) { 214 if ((!ipv6_ext_hdr(nexthdr)) || nexthdr == NEXTHDR_NONE) {
215 if (target < 0) 215 if (target < 0 || found)
216 break; 216 break;
217 return -ENOENT; 217 return -ENOENT;
218 } 218 }
diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index f81f59686f21..f2610e157660 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -414,7 +414,7 @@ static void icmp6_send(struct sk_buff *skb, u8 type, u8 code, __u32 info)
414 addr_type = ipv6_addr_type(&hdr->daddr); 414 addr_type = ipv6_addr_type(&hdr->daddr);
415 415
416 if (ipv6_chk_addr(net, &hdr->daddr, skb->dev, 0) || 416 if (ipv6_chk_addr(net, &hdr->daddr, skb->dev, 0) ||
417 ipv6_anycast_destination(skb)) 417 ipv6_chk_acast_addr_src(net, skb->dev, &hdr->daddr))
418 saddr = &hdr->daddr; 418 saddr = &hdr->daddr;
419 419
420 /* 420 /*
diff --git a/net/ipv6/ip6_offload.c b/net/ipv6/ip6_offload.c
index 1e8683b135bb..59f95affceb0 100644
--- a/net/ipv6/ip6_offload.c
+++ b/net/ipv6/ip6_offload.c
@@ -89,7 +89,7 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb,
89 unsigned int unfrag_ip6hlen; 89 unsigned int unfrag_ip6hlen;
90 u8 *prevhdr; 90 u8 *prevhdr;
91 int offset = 0; 91 int offset = 0;
92 bool tunnel; 92 bool encap, udpfrag;
93 int nhoff; 93 int nhoff;
94 94
95 if (unlikely(skb_shinfo(skb)->gso_type & 95 if (unlikely(skb_shinfo(skb)->gso_type &
@@ -110,8 +110,8 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb,
110 if (unlikely(!pskb_may_pull(skb, sizeof(*ipv6h)))) 110 if (unlikely(!pskb_may_pull(skb, sizeof(*ipv6h))))
111 goto out; 111 goto out;
112 112
113 tunnel = SKB_GSO_CB(skb)->encap_level > 0; 113 encap = SKB_GSO_CB(skb)->encap_level > 0;
114 if (tunnel) 114 if (encap)
115 features = skb->dev->hw_enc_features & netif_skb_features(skb); 115 features = skb->dev->hw_enc_features & netif_skb_features(skb);
116 SKB_GSO_CB(skb)->encap_level += sizeof(*ipv6h); 116 SKB_GSO_CB(skb)->encap_level += sizeof(*ipv6h);
117 117
@@ -121,6 +121,12 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb,
121 121
122 proto = ipv6_gso_pull_exthdrs(skb, ipv6h->nexthdr); 122 proto = ipv6_gso_pull_exthdrs(skb, ipv6h->nexthdr);
123 123
124 if (skb->encapsulation &&
125 skb_shinfo(skb)->gso_type & (SKB_GSO_SIT|SKB_GSO_IPIP))
126 udpfrag = proto == IPPROTO_UDP && encap;
127 else
128 udpfrag = proto == IPPROTO_UDP && !skb->encapsulation;
129
124 ops = rcu_dereference(inet6_offloads[proto]); 130 ops = rcu_dereference(inet6_offloads[proto]);
125 if (likely(ops && ops->callbacks.gso_segment)) { 131 if (likely(ops && ops->callbacks.gso_segment)) {
126 skb_reset_transport_header(skb); 132 skb_reset_transport_header(skb);
@@ -133,13 +139,9 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb,
133 for (skb = segs; skb; skb = skb->next) { 139 for (skb = segs; skb; skb = skb->next) {
134 ipv6h = (struct ipv6hdr *)(skb_mac_header(skb) + nhoff); 140 ipv6h = (struct ipv6hdr *)(skb_mac_header(skb) + nhoff);
135 ipv6h->payload_len = htons(skb->len - nhoff - sizeof(*ipv6h)); 141 ipv6h->payload_len = htons(skb->len - nhoff - sizeof(*ipv6h));
136 if (tunnel) {
137 skb_reset_inner_headers(skb);
138 skb->encapsulation = 1;
139 }
140 skb->network_header = (u8 *)ipv6h - skb->head; 142 skb->network_header = (u8 *)ipv6h - skb->head;
141 143
142 if (!tunnel && proto == IPPROTO_UDP) { 144 if (udpfrag) {
143 unfrag_ip6hlen = ip6_find_1stfragopt(skb, &prevhdr); 145 unfrag_ip6hlen = ip6_find_1stfragopt(skb, &prevhdr);
144 fptr = (struct frag_hdr *)((u8 *)ipv6h + unfrag_ip6hlen); 146 fptr = (struct frag_hdr *)((u8 *)ipv6h + unfrag_ip6hlen);
145 fptr->frag_off = htons(offset); 147 fptr->frag_off = htons(offset);
@@ -148,6 +150,8 @@ static struct sk_buff *ipv6_gso_segment(struct sk_buff *skb,
148 offset += (ntohs(ipv6h->payload_len) - 150 offset += (ntohs(ipv6h->payload_len) -
149 sizeof(struct frag_hdr)); 151 sizeof(struct frag_hdr));
150 } 152 }
153 if (encap)
154 skb_reset_inner_headers(skb);
151 } 155 }
152 156
153out: 157out:
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index ef02b26ccf81..16f91a2e7888 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -342,6 +342,20 @@ static unsigned int ip6_dst_mtu_forward(const struct dst_entry *dst)
342 return mtu; 342 return mtu;
343} 343}
344 344
345static bool ip6_pkt_too_big(const struct sk_buff *skb, unsigned int mtu)
346{
347 if (skb->len <= mtu || skb->local_df)
348 return false;
349
350 if (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)
351 return true;
352
353 if (skb_is_gso(skb) && skb_gso_network_seglen(skb) <= mtu)
354 return false;
355
356 return true;
357}
358
345int ip6_forward(struct sk_buff *skb) 359int ip6_forward(struct sk_buff *skb)
346{ 360{
347 struct dst_entry *dst = skb_dst(skb); 361 struct dst_entry *dst = skb_dst(skb);
@@ -466,8 +480,7 @@ int ip6_forward(struct sk_buff *skb)
466 if (mtu < IPV6_MIN_MTU) 480 if (mtu < IPV6_MIN_MTU)
467 mtu = IPV6_MIN_MTU; 481 mtu = IPV6_MIN_MTU;
468 482
469 if ((!skb->local_df && skb->len > mtu && !skb_is_gso(skb)) || 483 if (ip6_pkt_too_big(skb, mtu)) {
470 (IP6CB(skb)->frag_max_size && IP6CB(skb)->frag_max_size > mtu)) {
471 /* Again, force OUTPUT device used as source address */ 484 /* Again, force OUTPUT device used as source address */
472 skb->dev = dst->dev; 485 skb->dev = dst->dev;
473 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu); 486 icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
@@ -517,9 +530,6 @@ static void ip6_copy_metadata(struct sk_buff *to, struct sk_buff *from)
517 to->tc_index = from->tc_index; 530 to->tc_index = from->tc_index;
518#endif 531#endif
519 nf_copy(to, from); 532 nf_copy(to, from);
520#if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
521 to->nf_trace = from->nf_trace;
522#endif
523 skb_copy_secmark(to, from); 533 skb_copy_secmark(to, from);
524} 534}
525 535
diff --git a/net/ipv6/netfilter/Kconfig b/net/ipv6/netfilter/Kconfig
index 35750df744dc..4bff1f297e39 100644
--- a/net/ipv6/netfilter/Kconfig
+++ b/net/ipv6/netfilter/Kconfig
@@ -50,6 +50,11 @@ config NFT_CHAIN_NAT_IPV6
50 packet transformations such as the source, destination address and 50 packet transformations such as the source, destination address and
51 source and destination ports. 51 source and destination ports.
52 52
53config NFT_REJECT_IPV6
54 depends on NF_TABLES_IPV6
55 default NFT_REJECT
56 tristate
57
53config IP6_NF_IPTABLES 58config IP6_NF_IPTABLES
54 tristate "IP6 tables support (required for filtering)" 59 tristate "IP6 tables support (required for filtering)"
55 depends on INET && IPV6 60 depends on INET && IPV6
diff --git a/net/ipv6/netfilter/Makefile b/net/ipv6/netfilter/Makefile
index d1b4928f34f7..70d3dd66f2cd 100644
--- a/net/ipv6/netfilter/Makefile
+++ b/net/ipv6/netfilter/Makefile
@@ -27,6 +27,7 @@ obj-$(CONFIG_NF_DEFRAG_IPV6) += nf_defrag_ipv6.o
27obj-$(CONFIG_NF_TABLES_IPV6) += nf_tables_ipv6.o 27obj-$(CONFIG_NF_TABLES_IPV6) += nf_tables_ipv6.o
28obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV6) += nft_chain_route_ipv6.o 28obj-$(CONFIG_NFT_CHAIN_ROUTE_IPV6) += nft_chain_route_ipv6.o
29obj-$(CONFIG_NFT_CHAIN_NAT_IPV6) += nft_chain_nat_ipv6.o 29obj-$(CONFIG_NFT_CHAIN_NAT_IPV6) += nft_chain_nat_ipv6.o
30obj-$(CONFIG_NFT_REJECT_IPV6) += nft_reject_ipv6.o
30 31
31# matches 32# matches
32obj-$(CONFIG_IP6_NF_MATCH_AH) += ip6t_ah.o 33obj-$(CONFIG_IP6_NF_MATCH_AH) += ip6t_ah.o
diff --git a/net/ipv6/netfilter/nft_reject_ipv6.c b/net/ipv6/netfilter/nft_reject_ipv6.c
new file mode 100644
index 000000000000..0bc19fa87821
--- /dev/null
+++ b/net/ipv6/netfilter/nft_reject_ipv6.c
@@ -0,0 +1,76 @@
1/*
2 * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
3 * Copyright (c) 2013 Eric Leblond <eric@regit.org>
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License version 2 as
7 * published by the Free Software Foundation.
8 *
9 * Development of this code funded by Astaro AG (http://www.astaro.com/)
10 */
11
12#include <linux/kernel.h>
13#include <linux/init.h>
14#include <linux/module.h>
15#include <linux/netlink.h>
16#include <linux/netfilter.h>
17#include <linux/netfilter/nf_tables.h>
18#include <net/netfilter/nf_tables.h>
19#include <net/netfilter/nft_reject.h>
20#include <net/netfilter/ipv6/nf_reject.h>
21
22void nft_reject_ipv6_eval(const struct nft_expr *expr,
23 struct nft_data data[NFT_REG_MAX + 1],
24 const struct nft_pktinfo *pkt)
25{
26 struct nft_reject *priv = nft_expr_priv(expr);
27 struct net *net = dev_net((pkt->in != NULL) ? pkt->in : pkt->out);
28
29 switch (priv->type) {
30 case NFT_REJECT_ICMP_UNREACH:
31 nf_send_unreach6(net, pkt->skb, priv->icmp_code,
32 pkt->ops->hooknum);
33 break;
34 case NFT_REJECT_TCP_RST:
35 nf_send_reset6(net, pkt->skb, pkt->ops->hooknum);
36 break;
37 }
38
39 data[NFT_REG_VERDICT].verdict = NF_DROP;
40}
41EXPORT_SYMBOL_GPL(nft_reject_ipv6_eval);
42
43static struct nft_expr_type nft_reject_ipv6_type;
44static const struct nft_expr_ops nft_reject_ipv6_ops = {
45 .type = &nft_reject_ipv6_type,
46 .size = NFT_EXPR_SIZE(sizeof(struct nft_reject)),
47 .eval = nft_reject_ipv6_eval,
48 .init = nft_reject_init,
49 .dump = nft_reject_dump,
50};
51
52static struct nft_expr_type nft_reject_ipv6_type __read_mostly = {
53 .family = NFPROTO_IPV6,
54 .name = "reject",
55 .ops = &nft_reject_ipv6_ops,
56 .policy = nft_reject_policy,
57 .maxattr = NFTA_REJECT_MAX,
58 .owner = THIS_MODULE,
59};
60
61static int __init nft_reject_ipv6_module_init(void)
62{
63 return nft_register_expr(&nft_reject_ipv6_type);
64}
65
66static void __exit nft_reject_ipv6_module_exit(void)
67{
68 nft_unregister_expr(&nft_reject_ipv6_type);
69}
70
71module_init(nft_reject_ipv6_module_init);
72module_exit(nft_reject_ipv6_module_exit);
73
74MODULE_LICENSE("GPL");
75MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
76MODULE_ALIAS_NFT_AF_EXPR(AF_INET6, "reject");
diff --git a/net/ipv6/ping.c b/net/ipv6/ping.c
index fb9beb78f00b..587bbdcb22b4 100644
--- a/net/ipv6/ping.c
+++ b/net/ipv6/ping.c
@@ -135,6 +135,7 @@ int ping_v6_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
135 fl6.flowi6_proto = IPPROTO_ICMPV6; 135 fl6.flowi6_proto = IPPROTO_ICMPV6;
136 fl6.saddr = np->saddr; 136 fl6.saddr = np->saddr;
137 fl6.daddr = *daddr; 137 fl6.daddr = *daddr;
138 fl6.flowi6_mark = sk->sk_mark;
138 fl6.fl6_icmp_type = user_icmph.icmp6_type; 139 fl6.fl6_icmp_type = user_icmph.icmp6_type;
139 fl6.fl6_icmp_code = user_icmph.icmp6_code; 140 fl6.fl6_icmp_code = user_icmph.icmp6_code;
140 security_sk_classify_flow(sk, flowi6_to_flowi(&fl6)); 141 security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 3dfbcf1dcb1c..b4d74c86586c 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -475,6 +475,7 @@ static void ipip6_tunnel_uninit(struct net_device *dev)
475 ipip6_tunnel_unlink(sitn, tunnel); 475 ipip6_tunnel_unlink(sitn, tunnel);
476 ipip6_tunnel_del_prl(tunnel, NULL); 476 ipip6_tunnel_del_prl(tunnel, NULL);
477 } 477 }
478 ip_tunnel_dst_reset_all(tunnel);
478 dev_put(dev); 479 dev_put(dev);
479} 480}
480 481
@@ -1082,6 +1083,7 @@ static void ipip6_tunnel_update(struct ip_tunnel *t, struct ip_tunnel_parm *p)
1082 t->parms.link = p->link; 1083 t->parms.link = p->link;
1083 ipip6_tunnel_bind_dev(t->dev); 1084 ipip6_tunnel_bind_dev(t->dev);
1084 } 1085 }
1086 ip_tunnel_dst_reset_all(t);
1085 netdev_state_change(t->dev); 1087 netdev_state_change(t->dev);
1086} 1088}
1087 1089
@@ -1112,6 +1114,7 @@ static int ipip6_tunnel_update_6rd(struct ip_tunnel *t,
1112 t->ip6rd.relay_prefix = relay_prefix; 1114 t->ip6rd.relay_prefix = relay_prefix;
1113 t->ip6rd.prefixlen = ip6rd->prefixlen; 1115 t->ip6rd.prefixlen = ip6rd->prefixlen;
1114 t->ip6rd.relay_prefixlen = ip6rd->relay_prefixlen; 1116 t->ip6rd.relay_prefixlen = ip6rd->relay_prefixlen;
1117 ip_tunnel_dst_reset_all(t);
1115 netdev_state_change(t->dev); 1118 netdev_state_change(t->dev);
1116 return 0; 1119 return 0;
1117} 1120}
@@ -1271,6 +1274,7 @@ ipip6_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
1271 err = ipip6_tunnel_add_prl(t, &prl, cmd == SIOCCHGPRL); 1274 err = ipip6_tunnel_add_prl(t, &prl, cmd == SIOCCHGPRL);
1272 break; 1275 break;
1273 } 1276 }
1277 ip_tunnel_dst_reset_all(t);
1274 netdev_state_change(dev); 1278 netdev_state_change(dev);
1275 break; 1279 break;
1276 1280
@@ -1326,6 +1330,9 @@ static const struct net_device_ops ipip6_netdev_ops = {
1326 1330
1327static void ipip6_dev_free(struct net_device *dev) 1331static void ipip6_dev_free(struct net_device *dev)
1328{ 1332{
1333 struct ip_tunnel *tunnel = netdev_priv(dev);
1334
1335 free_percpu(tunnel->dst_cache);
1329 free_percpu(dev->tstats); 1336 free_percpu(dev->tstats);
1330 free_netdev(dev); 1337 free_netdev(dev);
1331} 1338}
@@ -1375,6 +1382,12 @@ static int ipip6_tunnel_init(struct net_device *dev)
1375 u64_stats_init(&ipip6_tunnel_stats->syncp); 1382 u64_stats_init(&ipip6_tunnel_stats->syncp);
1376 } 1383 }
1377 1384
1385 tunnel->dst_cache = alloc_percpu(struct ip_tunnel_dst);
1386 if (!tunnel->dst_cache) {
1387 free_percpu(dev->tstats);
1388 return -ENOMEM;
1389 }
1390
1378 return 0; 1391 return 0;
1379} 1392}
1380 1393
@@ -1405,6 +1418,12 @@ static int __net_init ipip6_fb_tunnel_init(struct net_device *dev)
1405 u64_stats_init(&ipip6_fb_stats->syncp); 1418 u64_stats_init(&ipip6_fb_stats->syncp);
1406 } 1419 }
1407 1420
1421 tunnel->dst_cache = alloc_percpu(struct ip_tunnel_dst);
1422 if (!tunnel->dst_cache) {
1423 free_percpu(dev->tstats);
1424 return -ENOMEM;
1425 }
1426
1408 dev_hold(dev); 1427 dev_hold(dev);
1409 rcu_assign_pointer(sitn->tunnels_wc[0], tunnel); 1428 rcu_assign_pointer(sitn->tunnels_wc[0], tunnel);
1410 return 0; 1429 return 0;
diff --git a/net/ipv6/udp_offload.c b/net/ipv6/udp_offload.c
index e7359f9eaa8d..b261ee8b83fc 100644
--- a/net/ipv6/udp_offload.c
+++ b/net/ipv6/udp_offload.c
@@ -113,7 +113,7 @@ static struct sk_buff *udp6_ufo_fragment(struct sk_buff *skb,
113 fptr = (struct frag_hdr *)(skb_network_header(skb) + unfrag_ip6hlen); 113 fptr = (struct frag_hdr *)(skb_network_header(skb) + unfrag_ip6hlen);
114 fptr->nexthdr = nexthdr; 114 fptr->nexthdr = nexthdr;
115 fptr->reserved = 0; 115 fptr->reserved = 0;
116 ipv6_select_ident(fptr, (struct rt6_info *)skb_dst(skb)); 116 fptr->identification = skb_shinfo(skb)->ip6_frag_id;
117 117
118 /* Fragment the skb. ipv6 header and the remaining fields of the 118 /* Fragment the skb. ipv6 header and the remaining fields of the
119 * fragment header are updated in ipv6_gso_segment() 119 * fragment header are updated in ipv6_gso_segment()
diff --git a/net/ipx/af_ipx.c b/net/ipx/af_ipx.c
index 994e28bfb32e..00b2a6d1c009 100644
--- a/net/ipx/af_ipx.c
+++ b/net/ipx/af_ipx.c
@@ -52,18 +52,12 @@
52#include <net/p8022.h> 52#include <net/p8022.h>
53#include <net/psnap.h> 53#include <net/psnap.h>
54#include <net/sock.h> 54#include <net/sock.h>
55#include <net/datalink.h>
55#include <net/tcp_states.h> 56#include <net/tcp_states.h>
57#include <net/net_namespace.h>
56 58
57#include <asm/uaccess.h> 59#include <asm/uaccess.h>
58 60
59#ifdef CONFIG_SYSCTL
60extern void ipx_register_sysctl(void);
61extern void ipx_unregister_sysctl(void);
62#else
63#define ipx_register_sysctl()
64#define ipx_unregister_sysctl()
65#endif
66
67/* Configuration Variables */ 61/* Configuration Variables */
68static unsigned char ipxcfg_max_hops = 16; 62static unsigned char ipxcfg_max_hops = 16;
69static char ipxcfg_auto_select_primary; 63static char ipxcfg_auto_select_primary;
@@ -84,15 +78,6 @@ DEFINE_SPINLOCK(ipx_interfaces_lock);
84struct ipx_interface *ipx_primary_net; 78struct ipx_interface *ipx_primary_net;
85struct ipx_interface *ipx_internal_net; 79struct ipx_interface *ipx_internal_net;
86 80
87extern int ipxrtr_add_route(__be32 network, struct ipx_interface *intrfc,
88 unsigned char *node);
89extern void ipxrtr_del_routes(struct ipx_interface *intrfc);
90extern int ipxrtr_route_packet(struct sock *sk, struct sockaddr_ipx *usipx,
91 struct iovec *iov, size_t len, int noblock);
92extern int ipxrtr_route_skb(struct sk_buff *skb);
93extern struct ipx_route *ipxrtr_lookup(__be32 net);
94extern int ipxrtr_ioctl(unsigned int cmd, void __user *arg);
95
96struct ipx_interface *ipx_interfaces_head(void) 81struct ipx_interface *ipx_interfaces_head(void)
97{ 82{
98 struct ipx_interface *rc = NULL; 83 struct ipx_interface *rc = NULL;
@@ -1986,9 +1971,6 @@ static struct notifier_block ipx_dev_notifier = {
1986 .notifier_call = ipxitf_device_event, 1971 .notifier_call = ipxitf_device_event,
1987}; 1972};
1988 1973
1989extern struct datalink_proto *make_EII_client(void);
1990extern void destroy_EII_client(struct datalink_proto *);
1991
1992static const unsigned char ipx_8022_type = 0xE0; 1974static const unsigned char ipx_8022_type = 0xE0;
1993static const unsigned char ipx_snap_id[5] = { 0x0, 0x0, 0x0, 0x81, 0x37 }; 1975static const unsigned char ipx_snap_id[5] = { 0x0, 0x0, 0x0, 0x81, 0x37 };
1994static const char ipx_EII_err_msg[] __initconst = 1976static const char ipx_EII_err_msg[] __initconst =
diff --git a/net/ipx/ipx_route.c b/net/ipx/ipx_route.c
index 30f4519b092f..c1f03185c5e1 100644
--- a/net/ipx/ipx_route.c
+++ b/net/ipx/ipx_route.c
@@ -20,15 +20,11 @@ DEFINE_RWLOCK(ipx_routes_lock);
20 20
21extern struct ipx_interface *ipx_internal_net; 21extern struct ipx_interface *ipx_internal_net;
22 22
23extern __be16 ipx_cksum(struct ipxhdr *packet, int length);
24extern struct ipx_interface *ipxitf_find_using_net(__be32 net); 23extern struct ipx_interface *ipxitf_find_using_net(__be32 net);
25extern int ipxitf_demux_socket(struct ipx_interface *intrfc, 24extern int ipxitf_demux_socket(struct ipx_interface *intrfc,
26 struct sk_buff *skb, int copy); 25 struct sk_buff *skb, int copy);
27extern int ipxitf_demux_socket(struct ipx_interface *intrfc, 26extern int ipxitf_demux_socket(struct ipx_interface *intrfc,
28 struct sk_buff *skb, int copy); 27 struct sk_buff *skb, int copy);
29extern int ipxitf_send(struct ipx_interface *intrfc, struct sk_buff *skb,
30 char *node);
31extern struct ipx_interface *ipxitf_find_using_net(__be32 net);
32 28
33struct ipx_route *ipxrtr_lookup(__be32 net) 29struct ipx_route *ipxrtr_lookup(__be32 net)
34{ 30{
diff --git a/net/mac80211/cfg.c b/net/mac80211/cfg.c
index f9ae9b85d4c1..453e974287d1 100644
--- a/net/mac80211/cfg.c
+++ b/net/mac80211/cfg.c
@@ -1021,8 +1021,10 @@ static int ieee80211_start_ap(struct wiphy *wiphy, struct net_device *dev,
1021 IEEE80211_P2P_OPPPS_ENABLE_BIT; 1021 IEEE80211_P2P_OPPPS_ENABLE_BIT;
1022 1022
1023 err = ieee80211_assign_beacon(sdata, &params->beacon); 1023 err = ieee80211_assign_beacon(sdata, &params->beacon);
1024 if (err < 0) 1024 if (err < 0) {
1025 ieee80211_vif_release_channel(sdata);
1025 return err; 1026 return err;
1027 }
1026 changed |= err; 1028 changed |= err;
1027 1029
1028 err = drv_start_ap(sdata->local, sdata); 1030 err = drv_start_ap(sdata->local, sdata);
@@ -1032,6 +1034,7 @@ static int ieee80211_start_ap(struct wiphy *wiphy, struct net_device *dev,
1032 if (old) 1034 if (old)
1033 kfree_rcu(old, rcu_head); 1035 kfree_rcu(old, rcu_head);
1034 RCU_INIT_POINTER(sdata->u.ap.beacon, NULL); 1036 RCU_INIT_POINTER(sdata->u.ap.beacon, NULL);
1037 ieee80211_vif_release_channel(sdata);
1035 return err; 1038 return err;
1036 } 1039 }
1037 1040
@@ -1090,8 +1093,6 @@ static int ieee80211_stop_ap(struct wiphy *wiphy, struct net_device *dev)
1090 kfree(sdata->u.ap.next_beacon); 1093 kfree(sdata->u.ap.next_beacon);
1091 sdata->u.ap.next_beacon = NULL; 1094 sdata->u.ap.next_beacon = NULL;
1092 1095
1093 cancel_work_sync(&sdata->u.ap.request_smps_work);
1094
1095 /* turn off carrier for this interface and dependent VLANs */ 1096 /* turn off carrier for this interface and dependent VLANs */
1096 list_for_each_entry(vlan, &sdata->u.ap.vlans, u.vlan.list) 1097 list_for_each_entry(vlan, &sdata->u.ap.vlans, u.vlan.list)
1097 netif_carrier_off(vlan->dev); 1098 netif_carrier_off(vlan->dev);
@@ -1103,6 +1104,7 @@ static int ieee80211_stop_ap(struct wiphy *wiphy, struct net_device *dev)
1103 kfree_rcu(old_beacon, rcu_head); 1104 kfree_rcu(old_beacon, rcu_head);
1104 if (old_probe_resp) 1105 if (old_probe_resp)
1105 kfree_rcu(old_probe_resp, rcu_head); 1106 kfree_rcu(old_probe_resp, rcu_head);
1107 sdata->u.ap.driver_smps_mode = IEEE80211_SMPS_OFF;
1106 1108
1107 __sta_info_flush(sdata, true); 1109 __sta_info_flush(sdata, true);
1108 ieee80211_free_keys(sdata, true); 1110 ieee80211_free_keys(sdata, true);
@@ -2638,6 +2640,24 @@ static int ieee80211_start_roc_work(struct ieee80211_local *local,
2638 INIT_DELAYED_WORK(&roc->work, ieee80211_sw_roc_work); 2640 INIT_DELAYED_WORK(&roc->work, ieee80211_sw_roc_work);
2639 INIT_LIST_HEAD(&roc->dependents); 2641 INIT_LIST_HEAD(&roc->dependents);
2640 2642
2643 /*
2644 * cookie is either the roc cookie (for normal roc)
2645 * or the SKB (for mgmt TX)
2646 */
2647 if (!txskb) {
2648 /* local->mtx protects this */
2649 local->roc_cookie_counter++;
2650 roc->cookie = local->roc_cookie_counter;
2651 /* wow, you wrapped 64 bits ... more likely a bug */
2652 if (WARN_ON(roc->cookie == 0)) {
2653 roc->cookie = 1;
2654 local->roc_cookie_counter++;
2655 }
2656 *cookie = roc->cookie;
2657 } else {
2658 *cookie = (unsigned long)txskb;
2659 }
2660
2641 /* if there's one pending or we're scanning, queue this one */ 2661 /* if there's one pending or we're scanning, queue this one */
2642 if (!list_empty(&local->roc_list) || 2662 if (!list_empty(&local->roc_list) ||
2643 local->scanning || local->radar_detect_enabled) 2663 local->scanning || local->radar_detect_enabled)
@@ -2772,24 +2792,6 @@ static int ieee80211_start_roc_work(struct ieee80211_local *local,
2772 if (!queued) 2792 if (!queued)
2773 list_add_tail(&roc->list, &local->roc_list); 2793 list_add_tail(&roc->list, &local->roc_list);
2774 2794
2775 /*
2776 * cookie is either the roc cookie (for normal roc)
2777 * or the SKB (for mgmt TX)
2778 */
2779 if (!txskb) {
2780 /* local->mtx protects this */
2781 local->roc_cookie_counter++;
2782 roc->cookie = local->roc_cookie_counter;
2783 /* wow, you wrapped 64 bits ... more likely a bug */
2784 if (WARN_ON(roc->cookie == 0)) {
2785 roc->cookie = 1;
2786 local->roc_cookie_counter++;
2787 }
2788 *cookie = roc->cookie;
2789 } else {
2790 *cookie = (unsigned long)txskb;
2791 }
2792
2793 return 0; 2795 return 0;
2794} 2796}
2795 2797
diff --git a/net/mac80211/ht.c b/net/mac80211/ht.c
index fab7b91923e0..70dd013de836 100644
--- a/net/mac80211/ht.c
+++ b/net/mac80211/ht.c
@@ -466,7 +466,9 @@ void ieee80211_request_smps_ap_work(struct work_struct *work)
466 u.ap.request_smps_work); 466 u.ap.request_smps_work);
467 467
468 sdata_lock(sdata); 468 sdata_lock(sdata);
469 __ieee80211_request_smps_ap(sdata, sdata->u.ap.driver_smps_mode); 469 if (sdata_dereference(sdata->u.ap.beacon, sdata))
470 __ieee80211_request_smps_ap(sdata,
471 sdata->u.ap.driver_smps_mode);
470 sdata_unlock(sdata); 472 sdata_unlock(sdata);
471} 473}
472 474
diff --git a/net/mac80211/ibss.c b/net/mac80211/ibss.c
index 771080ec7212..2796a198728f 100644
--- a/net/mac80211/ibss.c
+++ b/net/mac80211/ibss.c
@@ -695,12 +695,9 @@ static void ieee80211_ibss_disconnect(struct ieee80211_sub_if_data *sdata)
695 struct cfg80211_bss *cbss; 695 struct cfg80211_bss *cbss;
696 struct beacon_data *presp; 696 struct beacon_data *presp;
697 struct sta_info *sta; 697 struct sta_info *sta;
698 int active_ibss;
699 u16 capability; 698 u16 capability;
700 699
701 active_ibss = ieee80211_sta_active_ibss(sdata); 700 if (!is_zero_ether_addr(ifibss->bssid)) {
702
703 if (!active_ibss && !is_zero_ether_addr(ifibss->bssid)) {
704 capability = WLAN_CAPABILITY_IBSS; 701 capability = WLAN_CAPABILITY_IBSS;
705 702
706 if (ifibss->privacy) 703 if (ifibss->privacy)
diff --git a/net/mac80211/ieee80211_i.h b/net/mac80211/ieee80211_i.h
index 3701930c6649..5e44e3179e02 100644
--- a/net/mac80211/ieee80211_i.h
+++ b/net/mac80211/ieee80211_i.h
@@ -1692,14 +1692,8 @@ void ieee80211_stop_queue_by_reason(struct ieee80211_hw *hw, int queue,
1692void ieee80211_propagate_queue_wake(struct ieee80211_local *local, int queue); 1692void ieee80211_propagate_queue_wake(struct ieee80211_local *local, int queue);
1693void ieee80211_add_pending_skb(struct ieee80211_local *local, 1693void ieee80211_add_pending_skb(struct ieee80211_local *local,
1694 struct sk_buff *skb); 1694 struct sk_buff *skb);
1695void ieee80211_add_pending_skbs_fn(struct ieee80211_local *local, 1695void ieee80211_add_pending_skbs(struct ieee80211_local *local,
1696 struct sk_buff_head *skbs, 1696 struct sk_buff_head *skbs);
1697 void (*fn)(void *data), void *data);
1698static inline void ieee80211_add_pending_skbs(struct ieee80211_local *local,
1699 struct sk_buff_head *skbs)
1700{
1701 ieee80211_add_pending_skbs_fn(local, skbs, NULL, NULL);
1702}
1703void ieee80211_flush_queues(struct ieee80211_local *local, 1697void ieee80211_flush_queues(struct ieee80211_local *local,
1704 struct ieee80211_sub_if_data *sdata); 1698 struct ieee80211_sub_if_data *sdata);
1705 1699
diff --git a/net/mac80211/iface.c b/net/mac80211/iface.c
index 3dfd20a453ab..ce1c44370610 100644
--- a/net/mac80211/iface.c
+++ b/net/mac80211/iface.c
@@ -418,20 +418,24 @@ int ieee80211_add_virtual_monitor(struct ieee80211_local *local)
418 return ret; 418 return ret;
419 } 419 }
420 420
421 mutex_lock(&local->iflist_mtx);
422 rcu_assign_pointer(local->monitor_sdata, sdata);
423 mutex_unlock(&local->iflist_mtx);
424
421 mutex_lock(&local->mtx); 425 mutex_lock(&local->mtx);
422 ret = ieee80211_vif_use_channel(sdata, &local->monitor_chandef, 426 ret = ieee80211_vif_use_channel(sdata, &local->monitor_chandef,
423 IEEE80211_CHANCTX_EXCLUSIVE); 427 IEEE80211_CHANCTX_EXCLUSIVE);
424 mutex_unlock(&local->mtx); 428 mutex_unlock(&local->mtx);
425 if (ret) { 429 if (ret) {
430 mutex_lock(&local->iflist_mtx);
431 rcu_assign_pointer(local->monitor_sdata, NULL);
432 mutex_unlock(&local->iflist_mtx);
433 synchronize_net();
426 drv_remove_interface(local, sdata); 434 drv_remove_interface(local, sdata);
427 kfree(sdata); 435 kfree(sdata);
428 return ret; 436 return ret;
429 } 437 }
430 438
431 mutex_lock(&local->iflist_mtx);
432 rcu_assign_pointer(local->monitor_sdata, sdata);
433 mutex_unlock(&local->iflist_mtx);
434
435 return 0; 439 return 0;
436} 440}
437 441
@@ -770,12 +774,19 @@ static void ieee80211_do_stop(struct ieee80211_sub_if_data *sdata,
770 774
771 ieee80211_roc_purge(local, sdata); 775 ieee80211_roc_purge(local, sdata);
772 776
773 if (sdata->vif.type == NL80211_IFTYPE_STATION) 777 switch (sdata->vif.type) {
778 case NL80211_IFTYPE_STATION:
774 ieee80211_mgd_stop(sdata); 779 ieee80211_mgd_stop(sdata);
775 780 break;
776 if (sdata->vif.type == NL80211_IFTYPE_ADHOC) 781 case NL80211_IFTYPE_ADHOC:
777 ieee80211_ibss_stop(sdata); 782 ieee80211_ibss_stop(sdata);
778 783 break;
784 case NL80211_IFTYPE_AP:
785 cancel_work_sync(&sdata->u.ap.request_smps_work);
786 break;
787 default:
788 break;
789 }
779 790
780 /* 791 /*
781 * Remove all stations associated with this interface. 792 * Remove all stations associated with this interface.
@@ -1046,7 +1057,8 @@ static void ieee80211_uninit(struct net_device *dev)
1046 1057
1047static u16 ieee80211_netdev_select_queue(struct net_device *dev, 1058static u16 ieee80211_netdev_select_queue(struct net_device *dev,
1048 struct sk_buff *skb, 1059 struct sk_buff *skb,
1049 void *accel_priv) 1060 void *accel_priv,
1061 select_queue_fallback_t fallback)
1050{ 1062{
1051 return ieee80211_select_queue(IEEE80211_DEV_TO_SUB_IF(dev), skb); 1063 return ieee80211_select_queue(IEEE80211_DEV_TO_SUB_IF(dev), skb);
1052} 1064}
@@ -1064,7 +1076,8 @@ static const struct net_device_ops ieee80211_dataif_ops = {
1064 1076
1065static u16 ieee80211_monitor_select_queue(struct net_device *dev, 1077static u16 ieee80211_monitor_select_queue(struct net_device *dev,
1066 struct sk_buff *skb, 1078 struct sk_buff *skb,
1067 void *accel_priv) 1079 void *accel_priv,
1080 select_queue_fallback_t fallback)
1068{ 1081{
1069 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev); 1082 struct ieee80211_sub_if_data *sdata = IEEE80211_DEV_TO_SUB_IF(dev);
1070 struct ieee80211_local *local = sdata->local; 1083 struct ieee80211_local *local = sdata->local;
diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
index fc1d82465b3c..245dce969b31 100644
--- a/net/mac80211/mlme.c
+++ b/net/mac80211/mlme.c
@@ -222,6 +222,7 @@ ieee80211_determine_chantype(struct ieee80211_sub_if_data *sdata,
222 switch (vht_oper->chan_width) { 222 switch (vht_oper->chan_width) {
223 case IEEE80211_VHT_CHANWIDTH_USE_HT: 223 case IEEE80211_VHT_CHANWIDTH_USE_HT:
224 vht_chandef.width = chandef->width; 224 vht_chandef.width = chandef->width;
225 vht_chandef.center_freq1 = chandef->center_freq1;
225 break; 226 break;
226 case IEEE80211_VHT_CHANWIDTH_80MHZ: 227 case IEEE80211_VHT_CHANWIDTH_80MHZ:
227 vht_chandef.width = NL80211_CHAN_WIDTH_80; 228 vht_chandef.width = NL80211_CHAN_WIDTH_80;
@@ -271,6 +272,28 @@ ieee80211_determine_chantype(struct ieee80211_sub_if_data *sdata,
271 ret = 0; 272 ret = 0;
272 273
273out: 274out:
275 /*
276 * When tracking the current AP, don't do any further checks if the
277 * new chandef is identical to the one we're currently using for the
278 * connection. This keeps us from playing ping-pong with regulatory,
279 * without it the following can happen (for example):
280 * - connect to an AP with 80 MHz, world regdom allows 80 MHz
281 * - AP advertises regdom US
282 * - CRDA loads regdom US with 80 MHz prohibited (old database)
283 * - the code below detects an unsupported channel, downgrades, and
284 * we disconnect from the AP in the caller
285 * - disconnect causes CRDA to reload world regdomain and the game
286 * starts anew.
287 * (see https://bugzilla.kernel.org/show_bug.cgi?id=70881)
288 *
289 * It seems possible that there are still scenarios with CSA or real
290 * bandwidth changes where a this could happen, but those cases are
291 * less common and wouldn't completely prevent using the AP.
292 */
293 if (tracking &&
294 cfg80211_chandef_identical(chandef, &sdata->vif.bss_conf.chandef))
295 return ret;
296
274 /* don't print the message below for VHT mismatch if VHT is disabled */ 297 /* don't print the message below for VHT mismatch if VHT is disabled */
275 if (ret & IEEE80211_STA_DISABLE_VHT) 298 if (ret & IEEE80211_STA_DISABLE_VHT)
276 vht_chandef = *chandef; 299 vht_chandef = *chandef;
@@ -3753,6 +3776,7 @@ static int ieee80211_prep_connection(struct ieee80211_sub_if_data *sdata,
3753 chanctx_conf = rcu_dereference(sdata->vif.chanctx_conf); 3776 chanctx_conf = rcu_dereference(sdata->vif.chanctx_conf);
3754 if (WARN_ON(!chanctx_conf)) { 3777 if (WARN_ON(!chanctx_conf)) {
3755 rcu_read_unlock(); 3778 rcu_read_unlock();
3779 sta_info_free(local, new_sta);
3756 return -EINVAL; 3780 return -EINVAL;
3757 } 3781 }
3758 rate_flags = ieee80211_chandef_rate_flags(&chanctx_conf->def); 3782 rate_flags = ieee80211_chandef_rate_flags(&chanctx_conf->def);
diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c
index c24ca0d0f469..3e57f96c9666 100644
--- a/net/mac80211/rx.c
+++ b/net/mac80211/rx.c
@@ -1128,6 +1128,13 @@ static void sta_ps_end(struct sta_info *sta)
1128 sta->sta.addr, sta->sta.aid); 1128 sta->sta.addr, sta->sta.aid);
1129 1129
1130 if (test_sta_flag(sta, WLAN_STA_PS_DRIVER)) { 1130 if (test_sta_flag(sta, WLAN_STA_PS_DRIVER)) {
1131 /*
1132 * Clear the flag only if the other one is still set
1133 * so that the TX path won't start TX'ing new frames
1134 * directly ... In the case that the driver flag isn't
1135 * set ieee80211_sta_ps_deliver_wakeup() will clear it.
1136 */
1137 clear_sta_flag(sta, WLAN_STA_PS_STA);
1131 ps_dbg(sta->sdata, "STA %pM aid %d driver-ps-blocked\n", 1138 ps_dbg(sta->sdata, "STA %pM aid %d driver-ps-blocked\n",
1132 sta->sta.addr, sta->sta.aid); 1139 sta->sta.addr, sta->sta.aid);
1133 return; 1140 return;
diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c
index decd30c1e290..a023b432143b 100644
--- a/net/mac80211/sta_info.c
+++ b/net/mac80211/sta_info.c
@@ -91,7 +91,7 @@ static int sta_info_hash_del(struct ieee80211_local *local,
91 return -ENOENT; 91 return -ENOENT;
92} 92}
93 93
94static void cleanup_single_sta(struct sta_info *sta) 94static void __cleanup_single_sta(struct sta_info *sta)
95{ 95{
96 int ac, i; 96 int ac, i;
97 struct tid_ampdu_tx *tid_tx; 97 struct tid_ampdu_tx *tid_tx;
@@ -99,7 +99,8 @@ static void cleanup_single_sta(struct sta_info *sta)
99 struct ieee80211_local *local = sdata->local; 99 struct ieee80211_local *local = sdata->local;
100 struct ps_data *ps; 100 struct ps_data *ps;
101 101
102 if (test_sta_flag(sta, WLAN_STA_PS_STA)) { 102 if (test_sta_flag(sta, WLAN_STA_PS_STA) ||
103 test_sta_flag(sta, WLAN_STA_PS_DRIVER)) {
103 if (sta->sdata->vif.type == NL80211_IFTYPE_AP || 104 if (sta->sdata->vif.type == NL80211_IFTYPE_AP ||
104 sta->sdata->vif.type == NL80211_IFTYPE_AP_VLAN) 105 sta->sdata->vif.type == NL80211_IFTYPE_AP_VLAN)
105 ps = &sdata->bss->ps; 106 ps = &sdata->bss->ps;
@@ -109,6 +110,7 @@ static void cleanup_single_sta(struct sta_info *sta)
109 return; 110 return;
110 111
111 clear_sta_flag(sta, WLAN_STA_PS_STA); 112 clear_sta_flag(sta, WLAN_STA_PS_STA);
113 clear_sta_flag(sta, WLAN_STA_PS_DRIVER);
112 114
113 atomic_dec(&ps->num_sta_ps); 115 atomic_dec(&ps->num_sta_ps);
114 sta_info_recalc_tim(sta); 116 sta_info_recalc_tim(sta);
@@ -139,7 +141,14 @@ static void cleanup_single_sta(struct sta_info *sta)
139 ieee80211_purge_tx_queue(&local->hw, &tid_tx->pending); 141 ieee80211_purge_tx_queue(&local->hw, &tid_tx->pending);
140 kfree(tid_tx); 142 kfree(tid_tx);
141 } 143 }
144}
142 145
146static void cleanup_single_sta(struct sta_info *sta)
147{
148 struct ieee80211_sub_if_data *sdata = sta->sdata;
149 struct ieee80211_local *local = sdata->local;
150
151 __cleanup_single_sta(sta);
143 sta_info_free(local, sta); 152 sta_info_free(local, sta);
144} 153}
145 154
@@ -330,6 +339,7 @@ struct sta_info *sta_info_alloc(struct ieee80211_sub_if_data *sdata,
330 rcu_read_unlock(); 339 rcu_read_unlock();
331 340
332 spin_lock_init(&sta->lock); 341 spin_lock_init(&sta->lock);
342 spin_lock_init(&sta->ps_lock);
333 INIT_WORK(&sta->drv_unblock_wk, sta_unblock); 343 INIT_WORK(&sta->drv_unblock_wk, sta_unblock);
334 INIT_WORK(&sta->ampdu_mlme.work, ieee80211_ba_session_work); 344 INIT_WORK(&sta->ampdu_mlme.work, ieee80211_ba_session_work);
335 mutex_init(&sta->ampdu_mlme.mtx); 345 mutex_init(&sta->ampdu_mlme.mtx);
@@ -487,21 +497,26 @@ static int sta_info_insert_finish(struct sta_info *sta) __acquires(RCU)
487 goto out_err; 497 goto out_err;
488 } 498 }
489 499
490 /* notify driver */
491 err = sta_info_insert_drv_state(local, sdata, sta);
492 if (err)
493 goto out_err;
494
495 local->num_sta++; 500 local->num_sta++;
496 local->sta_generation++; 501 local->sta_generation++;
497 smp_mb(); 502 smp_mb();
498 503
504 /* simplify things and don't accept BA sessions yet */
505 set_sta_flag(sta, WLAN_STA_BLOCK_BA);
506
499 /* make the station visible */ 507 /* make the station visible */
500 sta_info_hash_add(local, sta); 508 sta_info_hash_add(local, sta);
501 509
502 list_add_rcu(&sta->list, &local->sta_list); 510 list_add_rcu(&sta->list, &local->sta_list);
503 511
512 /* notify driver */
513 err = sta_info_insert_drv_state(local, sdata, sta);
514 if (err)
515 goto out_remove;
516
504 set_sta_flag(sta, WLAN_STA_INSERTED); 517 set_sta_flag(sta, WLAN_STA_INSERTED);
518 /* accept BA sessions now */
519 clear_sta_flag(sta, WLAN_STA_BLOCK_BA);
505 520
506 ieee80211_recalc_min_chandef(sdata); 521 ieee80211_recalc_min_chandef(sdata);
507 ieee80211_sta_debugfs_add(sta); 522 ieee80211_sta_debugfs_add(sta);
@@ -522,6 +537,12 @@ static int sta_info_insert_finish(struct sta_info *sta) __acquires(RCU)
522 mesh_accept_plinks_update(sdata); 537 mesh_accept_plinks_update(sdata);
523 538
524 return 0; 539 return 0;
540 out_remove:
541 sta_info_hash_del(local, sta);
542 list_del_rcu(&sta->list);
543 local->num_sta--;
544 synchronize_net();
545 __cleanup_single_sta(sta);
525 out_err: 546 out_err:
526 mutex_unlock(&local->sta_mtx); 547 mutex_unlock(&local->sta_mtx);
527 rcu_read_lock(); 548 rcu_read_lock();
@@ -1071,10 +1092,14 @@ struct ieee80211_sta *ieee80211_find_sta(struct ieee80211_vif *vif,
1071} 1092}
1072EXPORT_SYMBOL(ieee80211_find_sta); 1093EXPORT_SYMBOL(ieee80211_find_sta);
1073 1094
1074static void clear_sta_ps_flags(void *_sta) 1095/* powersave support code */
1096void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta)
1075{ 1097{
1076 struct sta_info *sta = _sta;
1077 struct ieee80211_sub_if_data *sdata = sta->sdata; 1098 struct ieee80211_sub_if_data *sdata = sta->sdata;
1099 struct ieee80211_local *local = sdata->local;
1100 struct sk_buff_head pending;
1101 int filtered = 0, buffered = 0, ac;
1102 unsigned long flags;
1078 struct ps_data *ps; 1103 struct ps_data *ps;
1079 1104
1080 if (sdata->vif.type == NL80211_IFTYPE_AP || 1105 if (sdata->vif.type == NL80211_IFTYPE_AP ||
@@ -1085,20 +1110,6 @@ static void clear_sta_ps_flags(void *_sta)
1085 else 1110 else
1086 return; 1111 return;
1087 1112
1088 clear_sta_flag(sta, WLAN_STA_PS_DRIVER);
1089 if (test_and_clear_sta_flag(sta, WLAN_STA_PS_STA))
1090 atomic_dec(&ps->num_sta_ps);
1091}
1092
1093/* powersave support code */
1094void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta)
1095{
1096 struct ieee80211_sub_if_data *sdata = sta->sdata;
1097 struct ieee80211_local *local = sdata->local;
1098 struct sk_buff_head pending;
1099 int filtered = 0, buffered = 0, ac;
1100 unsigned long flags;
1101
1102 clear_sta_flag(sta, WLAN_STA_SP); 1113 clear_sta_flag(sta, WLAN_STA_SP);
1103 1114
1104 BUILD_BUG_ON(BITS_TO_LONGS(IEEE80211_NUM_TIDS) > 1); 1115 BUILD_BUG_ON(BITS_TO_LONGS(IEEE80211_NUM_TIDS) > 1);
@@ -1109,6 +1120,8 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta)
1109 1120
1110 skb_queue_head_init(&pending); 1121 skb_queue_head_init(&pending);
1111 1122
1123 /* sync with ieee80211_tx_h_unicast_ps_buf */
1124 spin_lock(&sta->ps_lock);
1112 /* Send all buffered frames to the station */ 1125 /* Send all buffered frames to the station */
1113 for (ac = 0; ac < IEEE80211_NUM_ACS; ac++) { 1126 for (ac = 0; ac < IEEE80211_NUM_ACS; ac++) {
1114 int count = skb_queue_len(&pending), tmp; 1127 int count = skb_queue_len(&pending), tmp;
@@ -1127,7 +1140,12 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta)
1127 buffered += tmp - count; 1140 buffered += tmp - count;
1128 } 1141 }
1129 1142
1130 ieee80211_add_pending_skbs_fn(local, &pending, clear_sta_ps_flags, sta); 1143 ieee80211_add_pending_skbs(local, &pending);
1144 clear_sta_flag(sta, WLAN_STA_PS_DRIVER);
1145 clear_sta_flag(sta, WLAN_STA_PS_STA);
1146 spin_unlock(&sta->ps_lock);
1147
1148 atomic_dec(&ps->num_sta_ps);
1131 1149
1132 /* This station just woke up and isn't aware of our SMPS state */ 1150 /* This station just woke up and isn't aware of our SMPS state */
1133 if (!ieee80211_smps_is_restrictive(sta->known_smps_mode, 1151 if (!ieee80211_smps_is_restrictive(sta->known_smps_mode,
diff --git a/net/mac80211/sta_info.h b/net/mac80211/sta_info.h
index d77ff7090630..d3a6d8208f2f 100644
--- a/net/mac80211/sta_info.h
+++ b/net/mac80211/sta_info.h
@@ -267,6 +267,7 @@ struct ieee80211_tx_latency_stat {
267 * @drv_unblock_wk: used for driver PS unblocking 267 * @drv_unblock_wk: used for driver PS unblocking
268 * @listen_interval: listen interval of this station, when we're acting as AP 268 * @listen_interval: listen interval of this station, when we're acting as AP
269 * @_flags: STA flags, see &enum ieee80211_sta_info_flags, do not use directly 269 * @_flags: STA flags, see &enum ieee80211_sta_info_flags, do not use directly
270 * @ps_lock: used for powersave (when mac80211 is the AP) related locking
270 * @ps_tx_buf: buffers (per AC) of frames to transmit to this station 271 * @ps_tx_buf: buffers (per AC) of frames to transmit to this station
271 * when it leaves power saving state or polls 272 * when it leaves power saving state or polls
272 * @tx_filtered: buffers (per AC) of frames we already tried to 273 * @tx_filtered: buffers (per AC) of frames we already tried to
@@ -356,10 +357,8 @@ struct sta_info {
356 /* use the accessors defined below */ 357 /* use the accessors defined below */
357 unsigned long _flags; 358 unsigned long _flags;
358 359
359 /* 360 /* STA powersave lock and frame queues */
360 * STA powersave frame queues, no more than the internal 361 spinlock_t ps_lock;
361 * locking required.
362 */
363 struct sk_buff_head ps_tx_buf[IEEE80211_NUM_ACS]; 362 struct sk_buff_head ps_tx_buf[IEEE80211_NUM_ACS];
364 struct sk_buff_head tx_filtered[IEEE80211_NUM_ACS]; 363 struct sk_buff_head tx_filtered[IEEE80211_NUM_ACS];
365 unsigned long driver_buffered_tids; 364 unsigned long driver_buffered_tids;
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c
index 27c990bf2320..4080c615636f 100644
--- a/net/mac80211/tx.c
+++ b/net/mac80211/tx.c
@@ -478,6 +478,20 @@ ieee80211_tx_h_unicast_ps_buf(struct ieee80211_tx_data *tx)
478 sta->sta.addr, sta->sta.aid, ac); 478 sta->sta.addr, sta->sta.aid, ac);
479 if (tx->local->total_ps_buffered >= TOTAL_MAX_TX_BUFFER) 479 if (tx->local->total_ps_buffered >= TOTAL_MAX_TX_BUFFER)
480 purge_old_ps_buffers(tx->local); 480 purge_old_ps_buffers(tx->local);
481
482 /* sync with ieee80211_sta_ps_deliver_wakeup */
483 spin_lock(&sta->ps_lock);
484 /*
485 * STA woke up the meantime and all the frames on ps_tx_buf have
486 * been queued to pending queue. No reordering can happen, go
487 * ahead and Tx the packet.
488 */
489 if (!test_sta_flag(sta, WLAN_STA_PS_STA) &&
490 !test_sta_flag(sta, WLAN_STA_PS_DRIVER)) {
491 spin_unlock(&sta->ps_lock);
492 return TX_CONTINUE;
493 }
494
481 if (skb_queue_len(&sta->ps_tx_buf[ac]) >= STA_MAX_TX_BUFFER) { 495 if (skb_queue_len(&sta->ps_tx_buf[ac]) >= STA_MAX_TX_BUFFER) {
482 struct sk_buff *old = skb_dequeue(&sta->ps_tx_buf[ac]); 496 struct sk_buff *old = skb_dequeue(&sta->ps_tx_buf[ac]);
483 ps_dbg(tx->sdata, 497 ps_dbg(tx->sdata,
@@ -492,6 +506,7 @@ ieee80211_tx_h_unicast_ps_buf(struct ieee80211_tx_data *tx)
492 info->flags |= IEEE80211_TX_INTFL_NEED_TXPROCESSING; 506 info->flags |= IEEE80211_TX_INTFL_NEED_TXPROCESSING;
493 info->flags &= ~IEEE80211_TX_TEMPORARY_FLAGS; 507 info->flags &= ~IEEE80211_TX_TEMPORARY_FLAGS;
494 skb_queue_tail(&sta->ps_tx_buf[ac], tx->skb); 508 skb_queue_tail(&sta->ps_tx_buf[ac], tx->skb);
509 spin_unlock(&sta->ps_lock);
495 510
496 if (!timer_pending(&local->sta_cleanup)) 511 if (!timer_pending(&local->sta_cleanup))
497 mod_timer(&local->sta_cleanup, 512 mod_timer(&local->sta_cleanup,
@@ -878,7 +893,7 @@ static int ieee80211_fragment(struct ieee80211_tx_data *tx,
878 } 893 }
879 894
880 /* adjust first fragment's length */ 895 /* adjust first fragment's length */
881 skb->len = hdrlen + per_fragm; 896 skb_trim(skb, hdrlen + per_fragm);
882 return 0; 897 return 0;
883} 898}
884 899
diff --git a/net/mac80211/util.c b/net/mac80211/util.c
index 676dc0967f37..b8700d417a9c 100644
--- a/net/mac80211/util.c
+++ b/net/mac80211/util.c
@@ -435,9 +435,8 @@ void ieee80211_add_pending_skb(struct ieee80211_local *local,
435 spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags); 435 spin_unlock_irqrestore(&local->queue_stop_reason_lock, flags);
436} 436}
437 437
438void ieee80211_add_pending_skbs_fn(struct ieee80211_local *local, 438void ieee80211_add_pending_skbs(struct ieee80211_local *local,
439 struct sk_buff_head *skbs, 439 struct sk_buff_head *skbs)
440 void (*fn)(void *data), void *data)
441{ 440{
442 struct ieee80211_hw *hw = &local->hw; 441 struct ieee80211_hw *hw = &local->hw;
443 struct sk_buff *skb; 442 struct sk_buff *skb;
@@ -461,9 +460,6 @@ void ieee80211_add_pending_skbs_fn(struct ieee80211_local *local,
461 __skb_queue_tail(&local->pending[queue], skb); 460 __skb_queue_tail(&local->pending[queue], skb);
462 } 461 }
463 462
464 if (fn)
465 fn(data);
466
467 for (i = 0; i < hw->queues; i++) 463 for (i = 0; i < hw->queues; i++)
468 __ieee80211_wake_queue(hw, i, 464 __ieee80211_wake_queue(hw, i,
469 IEEE80211_QUEUE_STOP_REASON_SKB_ADD); 465 IEEE80211_QUEUE_STOP_REASON_SKB_ADD);
@@ -1741,6 +1737,26 @@ int ieee80211_reconfig(struct ieee80211_local *local)
1741 IEEE80211_QUEUE_STOP_REASON_SUSPEND); 1737 IEEE80211_QUEUE_STOP_REASON_SUSPEND);
1742 1738
1743 /* 1739 /*
1740 * Reconfigure sched scan if it was interrupted by FW restart or
1741 * suspend.
1742 */
1743 mutex_lock(&local->mtx);
1744 sched_scan_sdata = rcu_dereference_protected(local->sched_scan_sdata,
1745 lockdep_is_held(&local->mtx));
1746 if (sched_scan_sdata && local->sched_scan_req)
1747 /*
1748 * Sched scan stopped, but we don't want to report it. Instead,
1749 * we're trying to reschedule.
1750 */
1751 if (__ieee80211_request_sched_scan_start(sched_scan_sdata,
1752 local->sched_scan_req))
1753 sched_scan_stopped = true;
1754 mutex_unlock(&local->mtx);
1755
1756 if (sched_scan_stopped)
1757 cfg80211_sched_scan_stopped(local->hw.wiphy);
1758
1759 /*
1744 * If this is for hw restart things are still running. 1760 * If this is for hw restart things are still running.
1745 * We may want to change that later, however. 1761 * We may want to change that later, however.
1746 */ 1762 */
@@ -1768,26 +1784,6 @@ int ieee80211_reconfig(struct ieee80211_local *local)
1768 WARN_ON(1); 1784 WARN_ON(1);
1769#endif 1785#endif
1770 1786
1771 /*
1772 * Reconfigure sched scan if it was interrupted by FW restart or
1773 * suspend.
1774 */
1775 mutex_lock(&local->mtx);
1776 sched_scan_sdata = rcu_dereference_protected(local->sched_scan_sdata,
1777 lockdep_is_held(&local->mtx));
1778 if (sched_scan_sdata && local->sched_scan_req)
1779 /*
1780 * Sched scan stopped, but we don't want to report it. Instead,
1781 * we're trying to reschedule.
1782 */
1783 if (__ieee80211_request_sched_scan_start(sched_scan_sdata,
1784 local->sched_scan_req))
1785 sched_scan_stopped = true;
1786 mutex_unlock(&local->mtx);
1787
1788 if (sched_scan_stopped)
1789 cfg80211_sched_scan_stopped(local->hw.wiphy);
1790
1791 return 0; 1787 return 0;
1792} 1788}
1793 1789
diff --git a/net/mac80211/wme.c b/net/mac80211/wme.c
index 21211c60ca98..d51422c778de 100644
--- a/net/mac80211/wme.c
+++ b/net/mac80211/wme.c
@@ -154,6 +154,11 @@ u16 ieee80211_select_queue(struct ieee80211_sub_if_data *sdata,
154 return IEEE80211_AC_BE; 154 return IEEE80211_AC_BE;
155 } 155 }
156 156
157 if (skb->protocol == sdata->control_port_protocol) {
158 skb->priority = 7;
159 return ieee80211_downgrade_queue(sdata, skb);
160 }
161
157 /* use the data classifier to determine what 802.1d tag the 162 /* use the data classifier to determine what 802.1d tag the
158 * data frame has */ 163 * data frame has */
159 rcu_read_lock(); 164 rcu_read_lock();
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
index c37467562fd0..e9410d17619d 100644
--- a/net/netfilter/Kconfig
+++ b/net/netfilter/Kconfig
@@ -513,7 +513,6 @@ config NFT_QUEUE
513 513
514config NFT_REJECT 514config NFT_REJECT
515 depends on NF_TABLES 515 depends on NF_TABLES
516 depends on NF_TABLES_IPV6 || !NF_TABLES_IPV6
517 default m if NETFILTER_ADVANCED=n 516 default m if NETFILTER_ADVANCED=n
518 tristate "Netfilter nf_tables reject support" 517 tristate "Netfilter nf_tables reject support"
519 help 518 help
@@ -521,6 +520,11 @@ config NFT_REJECT
521 explicitly deny and notify via TCP reset/ICMP informational errors 520 explicitly deny and notify via TCP reset/ICMP informational errors
522 unallowed traffic. 521 unallowed traffic.
523 522
523config NFT_REJECT_INET
524 depends on NF_TABLES_INET
525 default NFT_REJECT
526 tristate
527
524config NFT_COMPAT 528config NFT_COMPAT
525 depends on NF_TABLES 529 depends on NF_TABLES
526 depends on NETFILTER_XTABLES 530 depends on NETFILTER_XTABLES
diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
index ee9c4de5f8ed..bffdad774da7 100644
--- a/net/netfilter/Makefile
+++ b/net/netfilter/Makefile
@@ -79,6 +79,7 @@ obj-$(CONFIG_NFT_LIMIT) += nft_limit.o
79obj-$(CONFIG_NFT_NAT) += nft_nat.o 79obj-$(CONFIG_NFT_NAT) += nft_nat.o
80obj-$(CONFIG_NFT_QUEUE) += nft_queue.o 80obj-$(CONFIG_NFT_QUEUE) += nft_queue.o
81obj-$(CONFIG_NFT_REJECT) += nft_reject.o 81obj-$(CONFIG_NFT_REJECT) += nft_reject.o
82obj-$(CONFIG_NFT_REJECT_INET) += nft_reject_inet.o
82obj-$(CONFIG_NFT_RBTREE) += nft_rbtree.o 83obj-$(CONFIG_NFT_RBTREE) += nft_rbtree.o
83obj-$(CONFIG_NFT_HASH) += nft_hash.o 84obj-$(CONFIG_NFT_HASH) += nft_hash.o
84obj-$(CONFIG_NFT_COUNTER) += nft_counter.o 85obj-$(CONFIG_NFT_COUNTER) += nft_counter.o
diff --git a/net/netfilter/ipvs/ip_vs_conn.c b/net/netfilter/ipvs/ip_vs_conn.c
index 59a1a85bcb3e..a8eb0a89326a 100644
--- a/net/netfilter/ipvs/ip_vs_conn.c
+++ b/net/netfilter/ipvs/ip_vs_conn.c
@@ -871,11 +871,11 @@ ip_vs_conn_new(const struct ip_vs_conn_param *p,
871 cp->protocol = p->protocol; 871 cp->protocol = p->protocol;
872 ip_vs_addr_set(p->af, &cp->caddr, p->caddr); 872 ip_vs_addr_set(p->af, &cp->caddr, p->caddr);
873 cp->cport = p->cport; 873 cp->cport = p->cport;
874 ip_vs_addr_set(p->af, &cp->vaddr, p->vaddr); 874 /* proto should only be IPPROTO_IP if p->vaddr is a fwmark */
875 cp->vport = p->vport;
876 /* proto should only be IPPROTO_IP if d_addr is a fwmark */
877 ip_vs_addr_set(p->protocol == IPPROTO_IP ? AF_UNSPEC : p->af, 875 ip_vs_addr_set(p->protocol == IPPROTO_IP ? AF_UNSPEC : p->af,
878 &cp->daddr, daddr); 876 &cp->vaddr, p->vaddr);
877 cp->vport = p->vport;
878 ip_vs_addr_set(p->af, &cp->daddr, daddr);
879 cp->dport = dport; 879 cp->dport = dport;
880 cp->flags = flags; 880 cp->flags = flags;
881 cp->fwmark = fwmark; 881 cp->fwmark = fwmark;
diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 8824ed0ccc9c..356bef519fe5 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -312,6 +312,21 @@ static void death_by_timeout(unsigned long ul_conntrack)
312 nf_ct_delete((struct nf_conn *)ul_conntrack, 0, 0); 312 nf_ct_delete((struct nf_conn *)ul_conntrack, 0, 0);
313} 313}
314 314
315static inline bool
316nf_ct_key_equal(struct nf_conntrack_tuple_hash *h,
317 const struct nf_conntrack_tuple *tuple,
318 u16 zone)
319{
320 struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h);
321
322 /* A conntrack can be recreated with the equal tuple,
323 * so we need to check that the conntrack is confirmed
324 */
325 return nf_ct_tuple_equal(tuple, &h->tuple) &&
326 nf_ct_zone(ct) == zone &&
327 nf_ct_is_confirmed(ct);
328}
329
315/* 330/*
316 * Warning : 331 * Warning :
317 * - Caller must take a reference on returned object 332 * - Caller must take a reference on returned object
@@ -333,8 +348,7 @@ ____nf_conntrack_find(struct net *net, u16 zone,
333 local_bh_disable(); 348 local_bh_disable();
334begin: 349begin:
335 hlist_nulls_for_each_entry_rcu(h, n, &net->ct.hash[bucket], hnnode) { 350 hlist_nulls_for_each_entry_rcu(h, n, &net->ct.hash[bucket], hnnode) {
336 if (nf_ct_tuple_equal(tuple, &h->tuple) && 351 if (nf_ct_key_equal(h, tuple, zone)) {
337 nf_ct_zone(nf_ct_tuplehash_to_ctrack(h)) == zone) {
338 NF_CT_STAT_INC(net, found); 352 NF_CT_STAT_INC(net, found);
339 local_bh_enable(); 353 local_bh_enable();
340 return h; 354 return h;
@@ -372,8 +386,7 @@ begin:
372 !atomic_inc_not_zero(&ct->ct_general.use))) 386 !atomic_inc_not_zero(&ct->ct_general.use)))
373 h = NULL; 387 h = NULL;
374 else { 388 else {
375 if (unlikely(!nf_ct_tuple_equal(tuple, &h->tuple) || 389 if (unlikely(!nf_ct_key_equal(h, tuple, zone))) {
376 nf_ct_zone(ct) != zone)) {
377 nf_ct_put(ct); 390 nf_ct_put(ct);
378 goto begin; 391 goto begin;
379 } 392 }
@@ -435,7 +448,9 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)
435 goto out; 448 goto out;
436 449
437 add_timer(&ct->timeout); 450 add_timer(&ct->timeout);
438 nf_conntrack_get(&ct->ct_general); 451 smp_wmb();
452 /* The caller holds a reference to this object */
453 atomic_set(&ct->ct_general.use, 2);
439 __nf_conntrack_hash_insert(ct, hash, repl_hash); 454 __nf_conntrack_hash_insert(ct, hash, repl_hash);
440 NF_CT_STAT_INC(net, insert); 455 NF_CT_STAT_INC(net, insert);
441 spin_unlock_bh(&nf_conntrack_lock); 456 spin_unlock_bh(&nf_conntrack_lock);
@@ -449,6 +464,21 @@ out:
449} 464}
450EXPORT_SYMBOL_GPL(nf_conntrack_hash_check_insert); 465EXPORT_SYMBOL_GPL(nf_conntrack_hash_check_insert);
451 466
467/* deletion from this larval template list happens via nf_ct_put() */
468void nf_conntrack_tmpl_insert(struct net *net, struct nf_conn *tmpl)
469{
470 __set_bit(IPS_TEMPLATE_BIT, &tmpl->status);
471 __set_bit(IPS_CONFIRMED_BIT, &tmpl->status);
472 nf_conntrack_get(&tmpl->ct_general);
473
474 spin_lock_bh(&nf_conntrack_lock);
475 /* Overload tuple linked list to put us in template list. */
476 hlist_nulls_add_head_rcu(&tmpl->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
477 &net->ct.tmpl);
478 spin_unlock_bh(&nf_conntrack_lock);
479}
480EXPORT_SYMBOL_GPL(nf_conntrack_tmpl_insert);
481
452/* Confirm a connection given skb; places it in hash table */ 482/* Confirm a connection given skb; places it in hash table */
453int 483int
454__nf_conntrack_confirm(struct sk_buff *skb) 484__nf_conntrack_confirm(struct sk_buff *skb)
@@ -720,11 +750,10 @@ __nf_conntrack_alloc(struct net *net, u16 zone,
720 nf_ct_zone->id = zone; 750 nf_ct_zone->id = zone;
721 } 751 }
722#endif 752#endif
723 /* 753 /* Because we use RCU lookups, we set ct_general.use to zero before
724 * changes to lookup keys must be done before setting refcnt to 1 754 * this is inserted in any list.
725 */ 755 */
726 smp_wmb(); 756 atomic_set(&ct->ct_general.use, 0);
727 atomic_set(&ct->ct_general.use, 1);
728 return ct; 757 return ct;
729 758
730#ifdef CONFIG_NF_CONNTRACK_ZONES 759#ifdef CONFIG_NF_CONNTRACK_ZONES
@@ -748,6 +777,11 @@ void nf_conntrack_free(struct nf_conn *ct)
748{ 777{
749 struct net *net = nf_ct_net(ct); 778 struct net *net = nf_ct_net(ct);
750 779
780 /* A freed object has refcnt == 0, that's
781 * the golden rule for SLAB_DESTROY_BY_RCU
782 */
783 NF_CT_ASSERT(atomic_read(&ct->ct_general.use) == 0);
784
751 nf_ct_ext_destroy(ct); 785 nf_ct_ext_destroy(ct);
752 nf_ct_ext_free(ct); 786 nf_ct_ext_free(ct);
753 kmem_cache_free(net->ct.nf_conntrack_cachep, ct); 787 kmem_cache_free(net->ct.nf_conntrack_cachep, ct);
@@ -843,6 +877,9 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,
843 NF_CT_STAT_INC(net, new); 877 NF_CT_STAT_INC(net, new);
844 } 878 }
845 879
880 /* Now it is inserted into the unconfirmed list, bump refcount */
881 nf_conntrack_get(&ct->ct_general);
882
846 /* Overload tuple linked list to put us in unconfirmed list. */ 883 /* Overload tuple linked list to put us in unconfirmed list. */
847 hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode, 884 hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
848 &net->ct.unconfirmed); 885 &net->ct.unconfirmed);
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index bb322d0beb48..b9f0e0374322 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -1310,27 +1310,22 @@ ctnetlink_change_status(struct nf_conn *ct, const struct nlattr * const cda[])
1310} 1310}
1311 1311
1312static int 1312static int
1313ctnetlink_change_nat(struct nf_conn *ct, const struct nlattr * const cda[]) 1313ctnetlink_setup_nat(struct nf_conn *ct, const struct nlattr * const cda[])
1314{ 1314{
1315#ifdef CONFIG_NF_NAT_NEEDED 1315#ifdef CONFIG_NF_NAT_NEEDED
1316 int ret; 1316 int ret;
1317 1317
1318 if (cda[CTA_NAT_DST]) { 1318 ret = ctnetlink_parse_nat_setup(ct, NF_NAT_MANIP_DST,
1319 ret = ctnetlink_parse_nat_setup(ct, 1319 cda[CTA_NAT_DST]);
1320 NF_NAT_MANIP_DST, 1320 if (ret < 0)
1321 cda[CTA_NAT_DST]); 1321 return ret;
1322 if (ret < 0) 1322
1323 return ret; 1323 ret = ctnetlink_parse_nat_setup(ct, NF_NAT_MANIP_SRC,
1324 } 1324 cda[CTA_NAT_SRC]);
1325 if (cda[CTA_NAT_SRC]) { 1325 return ret;
1326 ret = ctnetlink_parse_nat_setup(ct,
1327 NF_NAT_MANIP_SRC,
1328 cda[CTA_NAT_SRC]);
1329 if (ret < 0)
1330 return ret;
1331 }
1332 return 0;
1333#else 1326#else
1327 if (!cda[CTA_NAT_DST] && !cda[CTA_NAT_SRC])
1328 return 0;
1334 return -EOPNOTSUPP; 1329 return -EOPNOTSUPP;
1335#endif 1330#endif
1336} 1331}
@@ -1659,11 +1654,9 @@ ctnetlink_create_conntrack(struct net *net, u16 zone,
1659 goto err2; 1654 goto err2;
1660 } 1655 }
1661 1656
1662 if (cda[CTA_NAT_SRC] || cda[CTA_NAT_DST]) { 1657 err = ctnetlink_setup_nat(ct, cda);
1663 err = ctnetlink_change_nat(ct, cda); 1658 if (err < 0)
1664 if (err < 0) 1659 goto err2;
1665 goto err2;
1666 }
1667 1660
1668 nf_ct_acct_ext_add(ct, GFP_ATOMIC); 1661 nf_ct_acct_ext_add(ct, GFP_ATOMIC);
1669 nf_ct_tstamp_ext_add(ct, GFP_ATOMIC); 1662 nf_ct_tstamp_ext_add(ct, GFP_ATOMIC);
diff --git a/net/netfilter/nf_nat_core.c b/net/netfilter/nf_nat_core.c
index d3f5cd6dd962..52ca952b802c 100644
--- a/net/netfilter/nf_nat_core.c
+++ b/net/netfilter/nf_nat_core.c
@@ -432,15 +432,15 @@ nf_nat_setup_info(struct nf_conn *ct,
432} 432}
433EXPORT_SYMBOL(nf_nat_setup_info); 433EXPORT_SYMBOL(nf_nat_setup_info);
434 434
435unsigned int 435static unsigned int
436nf_nat_alloc_null_binding(struct nf_conn *ct, unsigned int hooknum) 436__nf_nat_alloc_null_binding(struct nf_conn *ct, enum nf_nat_manip_type manip)
437{ 437{
438 /* Force range to this IP; let proto decide mapping for 438 /* Force range to this IP; let proto decide mapping for
439 * per-proto parts (hence not IP_NAT_RANGE_PROTO_SPECIFIED). 439 * per-proto parts (hence not IP_NAT_RANGE_PROTO_SPECIFIED).
440 * Use reply in case it's already been mangled (eg local packet). 440 * Use reply in case it's already been mangled (eg local packet).
441 */ 441 */
442 union nf_inet_addr ip = 442 union nf_inet_addr ip =
443 (HOOK2MANIP(hooknum) == NF_NAT_MANIP_SRC ? 443 (manip == NF_NAT_MANIP_SRC ?
444 ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3 : 444 ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3 :
445 ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3); 445 ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3);
446 struct nf_nat_range range = { 446 struct nf_nat_range range = {
@@ -448,7 +448,13 @@ nf_nat_alloc_null_binding(struct nf_conn *ct, unsigned int hooknum)
448 .min_addr = ip, 448 .min_addr = ip,
449 .max_addr = ip, 449 .max_addr = ip,
450 }; 450 };
451 return nf_nat_setup_info(ct, &range, HOOK2MANIP(hooknum)); 451 return nf_nat_setup_info(ct, &range, manip);
452}
453
454unsigned int
455nf_nat_alloc_null_binding(struct nf_conn *ct, unsigned int hooknum)
456{
457 return __nf_nat_alloc_null_binding(ct, HOOK2MANIP(hooknum));
452} 458}
453EXPORT_SYMBOL_GPL(nf_nat_alloc_null_binding); 459EXPORT_SYMBOL_GPL(nf_nat_alloc_null_binding);
454 460
@@ -702,9 +708,9 @@ static const struct nla_policy nat_nla_policy[CTA_NAT_MAX+1] = {
702 708
703static int 709static int
704nfnetlink_parse_nat(const struct nlattr *nat, 710nfnetlink_parse_nat(const struct nlattr *nat,
705 const struct nf_conn *ct, struct nf_nat_range *range) 711 const struct nf_conn *ct, struct nf_nat_range *range,
712 const struct nf_nat_l3proto *l3proto)
706{ 713{
707 const struct nf_nat_l3proto *l3proto;
708 struct nlattr *tb[CTA_NAT_MAX+1]; 714 struct nlattr *tb[CTA_NAT_MAX+1];
709 int err; 715 int err;
710 716
@@ -714,38 +720,46 @@ nfnetlink_parse_nat(const struct nlattr *nat,
714 if (err < 0) 720 if (err < 0)
715 return err; 721 return err;
716 722
717 rcu_read_lock();
718 l3proto = __nf_nat_l3proto_find(nf_ct_l3num(ct));
719 if (l3proto == NULL) {
720 err = -EAGAIN;
721 goto out;
722 }
723 err = l3proto->nlattr_to_range(tb, range); 723 err = l3proto->nlattr_to_range(tb, range);
724 if (err < 0) 724 if (err < 0)
725 goto out; 725 return err;
726 726
727 if (!tb[CTA_NAT_PROTO]) 727 if (!tb[CTA_NAT_PROTO])
728 goto out; 728 return 0;
729 729
730 err = nfnetlink_parse_nat_proto(tb[CTA_NAT_PROTO], ct, range); 730 return nfnetlink_parse_nat_proto(tb[CTA_NAT_PROTO], ct, range);
731out:
732 rcu_read_unlock();
733 return err;
734} 731}
735 732
733/* This function is called under rcu_read_lock() */
736static int 734static int
737nfnetlink_parse_nat_setup(struct nf_conn *ct, 735nfnetlink_parse_nat_setup(struct nf_conn *ct,
738 enum nf_nat_manip_type manip, 736 enum nf_nat_manip_type manip,
739 const struct nlattr *attr) 737 const struct nlattr *attr)
740{ 738{
741 struct nf_nat_range range; 739 struct nf_nat_range range;
740 const struct nf_nat_l3proto *l3proto;
742 int err; 741 int err;
743 742
744 err = nfnetlink_parse_nat(attr, ct, &range); 743 /* Should not happen, restricted to creating new conntracks
744 * via ctnetlink.
745 */
746 if (WARN_ON_ONCE(nf_nat_initialized(ct, manip)))
747 return -EEXIST;
748
749 /* Make sure that L3 NAT is there by when we call nf_nat_setup_info to
750 * attach the null binding, otherwise this may oops.
751 */
752 l3proto = __nf_nat_l3proto_find(nf_ct_l3num(ct));
753 if (l3proto == NULL)
754 return -EAGAIN;
755
756 /* No NAT information has been passed, allocate the null-binding */
757 if (attr == NULL)
758 return __nf_nat_alloc_null_binding(ct, manip);
759
760 err = nfnetlink_parse_nat(attr, ct, &range, l3proto);
745 if (err < 0) 761 if (err < 0)
746 return err; 762 return err;
747 if (nf_nat_initialized(ct, manip))
748 return -EEXIST;
749 763
750 return nf_nat_setup_info(ct, &range, manip); 764 return nf_nat_setup_info(ct, &range, manip);
751} 765}
diff --git a/net/netfilter/nf_synproxy_core.c b/net/netfilter/nf_synproxy_core.c
index 9858e3e51a3a..52e20c9a46a5 100644
--- a/net/netfilter/nf_synproxy_core.c
+++ b/net/netfilter/nf_synproxy_core.c
@@ -363,9 +363,8 @@ static int __net_init synproxy_net_init(struct net *net)
363 goto err2; 363 goto err2;
364 if (!nfct_synproxy_ext_add(ct)) 364 if (!nfct_synproxy_ext_add(ct))
365 goto err2; 365 goto err2;
366 __set_bit(IPS_TEMPLATE_BIT, &ct->status);
367 __set_bit(IPS_CONFIRMED_BIT, &ct->status);
368 366
367 nf_conntrack_tmpl_insert(net, ct);
369 snet->tmpl = ct; 368 snet->tmpl = ct;
370 369
371 snet->stats = alloc_percpu(struct synproxy_stats); 370 snet->stats = alloc_percpu(struct synproxy_stats);
@@ -390,7 +389,7 @@ static void __net_exit synproxy_net_exit(struct net *net)
390{ 389{
391 struct synproxy_net *snet = synproxy_pernet(net); 390 struct synproxy_net *snet = synproxy_pernet(net);
392 391
393 nf_conntrack_free(snet->tmpl); 392 nf_ct_put(snet->tmpl);
394 synproxy_proc_exit(net); 393 synproxy_proc_exit(net);
395 free_percpu(snet->stats); 394 free_percpu(snet->stats);
396} 395}
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 117bbaaddde6..adce01e8bb57 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1008,10 +1008,8 @@ notify:
1008 return 0; 1008 return 0;
1009} 1009}
1010 1010
1011static void nf_tables_rcu_chain_destroy(struct rcu_head *head) 1011static void nf_tables_chain_destroy(struct nft_chain *chain)
1012{ 1012{
1013 struct nft_chain *chain = container_of(head, struct nft_chain, rcu_head);
1014
1015 BUG_ON(chain->use > 0); 1013 BUG_ON(chain->use > 0);
1016 1014
1017 if (chain->flags & NFT_BASE_CHAIN) { 1015 if (chain->flags & NFT_BASE_CHAIN) {
@@ -1045,7 +1043,7 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb,
1045 if (IS_ERR(chain)) 1043 if (IS_ERR(chain))
1046 return PTR_ERR(chain); 1044 return PTR_ERR(chain);
1047 1045
1048 if (!list_empty(&chain->rules)) 1046 if (!list_empty(&chain->rules) || chain->use > 0)
1049 return -EBUSY; 1047 return -EBUSY;
1050 1048
1051 list_del(&chain->list); 1049 list_del(&chain->list);
@@ -1059,7 +1057,9 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb,
1059 family); 1057 family);
1060 1058
1061 /* Make sure all rule references are gone before this is released */ 1059 /* Make sure all rule references are gone before this is released */
1062 call_rcu(&chain->rcu_head, nf_tables_rcu_chain_destroy); 1060 synchronize_rcu();
1061
1062 nf_tables_chain_destroy(chain);
1063 return 0; 1063 return 0;
1064} 1064}
1065 1065
@@ -1114,35 +1114,45 @@ void nft_unregister_expr(struct nft_expr_type *type)
1114} 1114}
1115EXPORT_SYMBOL_GPL(nft_unregister_expr); 1115EXPORT_SYMBOL_GPL(nft_unregister_expr);
1116 1116
1117static const struct nft_expr_type *__nft_expr_type_get(struct nlattr *nla) 1117static const struct nft_expr_type *__nft_expr_type_get(u8 family,
1118 struct nlattr *nla)
1118{ 1119{
1119 const struct nft_expr_type *type; 1120 const struct nft_expr_type *type;
1120 1121
1121 list_for_each_entry(type, &nf_tables_expressions, list) { 1122 list_for_each_entry(type, &nf_tables_expressions, list) {
1122 if (!nla_strcmp(nla, type->name)) 1123 if (!nla_strcmp(nla, type->name) &&
1124 (!type->family || type->family == family))
1123 return type; 1125 return type;
1124 } 1126 }
1125 return NULL; 1127 return NULL;
1126} 1128}
1127 1129
1128static const struct nft_expr_type *nft_expr_type_get(struct nlattr *nla) 1130static const struct nft_expr_type *nft_expr_type_get(u8 family,
1131 struct nlattr *nla)
1129{ 1132{
1130 const struct nft_expr_type *type; 1133 const struct nft_expr_type *type;
1131 1134
1132 if (nla == NULL) 1135 if (nla == NULL)
1133 return ERR_PTR(-EINVAL); 1136 return ERR_PTR(-EINVAL);
1134 1137
1135 type = __nft_expr_type_get(nla); 1138 type = __nft_expr_type_get(family, nla);
1136 if (type != NULL && try_module_get(type->owner)) 1139 if (type != NULL && try_module_get(type->owner))
1137 return type; 1140 return type;
1138 1141
1139#ifdef CONFIG_MODULES 1142#ifdef CONFIG_MODULES
1140 if (type == NULL) { 1143 if (type == NULL) {
1141 nfnl_unlock(NFNL_SUBSYS_NFTABLES); 1144 nfnl_unlock(NFNL_SUBSYS_NFTABLES);
1145 request_module("nft-expr-%u-%.*s", family,
1146 nla_len(nla), (char *)nla_data(nla));
1147 nfnl_lock(NFNL_SUBSYS_NFTABLES);
1148 if (__nft_expr_type_get(family, nla))
1149 return ERR_PTR(-EAGAIN);
1150
1151 nfnl_unlock(NFNL_SUBSYS_NFTABLES);
1142 request_module("nft-expr-%.*s", 1152 request_module("nft-expr-%.*s",
1143 nla_len(nla), (char *)nla_data(nla)); 1153 nla_len(nla), (char *)nla_data(nla));
1144 nfnl_lock(NFNL_SUBSYS_NFTABLES); 1154 nfnl_lock(NFNL_SUBSYS_NFTABLES);
1145 if (__nft_expr_type_get(nla)) 1155 if (__nft_expr_type_get(family, nla))
1146 return ERR_PTR(-EAGAIN); 1156 return ERR_PTR(-EAGAIN);
1147 } 1157 }
1148#endif 1158#endif
@@ -1193,7 +1203,7 @@ static int nf_tables_expr_parse(const struct nft_ctx *ctx,
1193 if (err < 0) 1203 if (err < 0)
1194 return err; 1204 return err;
1195 1205
1196 type = nft_expr_type_get(tb[NFTA_EXPR_NAME]); 1206 type = nft_expr_type_get(ctx->afi->family, tb[NFTA_EXPR_NAME]);
1197 if (IS_ERR(type)) 1207 if (IS_ERR(type))
1198 return PTR_ERR(type); 1208 return PTR_ERR(type);
1199 1209
@@ -1521,9 +1531,8 @@ err:
1521 return err; 1531 return err;
1522} 1532}
1523 1533
1524static void nf_tables_rcu_rule_destroy(struct rcu_head *head) 1534static void nf_tables_rule_destroy(struct nft_rule *rule)
1525{ 1535{
1526 struct nft_rule *rule = container_of(head, struct nft_rule, rcu_head);
1527 struct nft_expr *expr; 1536 struct nft_expr *expr;
1528 1537
1529 /* 1538 /*
@@ -1538,11 +1547,6 @@ static void nf_tables_rcu_rule_destroy(struct rcu_head *head)
1538 kfree(rule); 1547 kfree(rule);
1539} 1548}
1540 1549
1541static void nf_tables_rule_destroy(struct nft_rule *rule)
1542{
1543 call_rcu(&rule->rcu_head, nf_tables_rcu_rule_destroy);
1544}
1545
1546#define NFT_RULE_MAXEXPRS 128 1550#define NFT_RULE_MAXEXPRS 128
1547 1551
1548static struct nft_expr_info *info; 1552static struct nft_expr_info *info;
@@ -1809,9 +1813,6 @@ static int nf_tables_commit(struct sk_buff *skb)
1809 synchronize_rcu(); 1813 synchronize_rcu();
1810 1814
1811 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { 1815 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) {
1812 /* Delete this rule from the dirty list */
1813 list_del(&rupd->list);
1814
1815 /* This rule was inactive in the past and just became active. 1816 /* This rule was inactive in the past and just became active.
1816 * Clear the next bit of the genmask since its meaning has 1817 * Clear the next bit of the genmask since its meaning has
1817 * changed, now it is the future. 1818 * changed, now it is the future.
@@ -1822,6 +1823,7 @@ static int nf_tables_commit(struct sk_buff *skb)
1822 rupd->chain, rupd->rule, 1823 rupd->chain, rupd->rule,
1823 NFT_MSG_NEWRULE, 0, 1824 NFT_MSG_NEWRULE, 0,
1824 rupd->family); 1825 rupd->family);
1826 list_del(&rupd->list);
1825 kfree(rupd); 1827 kfree(rupd);
1826 continue; 1828 continue;
1827 } 1829 }
@@ -1831,7 +1833,15 @@ static int nf_tables_commit(struct sk_buff *skb)
1831 nf_tables_rule_notify(skb, rupd->nlh, rupd->table, rupd->chain, 1833 nf_tables_rule_notify(skb, rupd->nlh, rupd->table, rupd->chain,
1832 rupd->rule, NFT_MSG_DELRULE, 0, 1834 rupd->rule, NFT_MSG_DELRULE, 0,
1833 rupd->family); 1835 rupd->family);
1836 }
1837
1838 /* Make sure we don't see any packet traversing old rules */
1839 synchronize_rcu();
1840
1841 /* Now we can safely release unused old rules */
1842 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) {
1834 nf_tables_rule_destroy(rupd->rule); 1843 nf_tables_rule_destroy(rupd->rule);
1844 list_del(&rupd->list);
1835 kfree(rupd); 1845 kfree(rupd);
1836 } 1846 }
1837 1847
@@ -1844,20 +1854,26 @@ static int nf_tables_abort(struct sk_buff *skb)
1844 struct nft_rule_trans *rupd, *tmp; 1854 struct nft_rule_trans *rupd, *tmp;
1845 1855
1846 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { 1856 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) {
1847 /* Delete all rules from the dirty list */
1848 list_del(&rupd->list);
1849
1850 if (!nft_rule_is_active_next(net, rupd->rule)) { 1857 if (!nft_rule_is_active_next(net, rupd->rule)) {
1851 nft_rule_clear(net, rupd->rule); 1858 nft_rule_clear(net, rupd->rule);
1859 list_del(&rupd->list);
1852 kfree(rupd); 1860 kfree(rupd);
1853 continue; 1861 continue;
1854 } 1862 }
1855 1863
1856 /* This rule is inactive, get rid of it */ 1864 /* This rule is inactive, get rid of it */
1857 list_del_rcu(&rupd->rule->list); 1865 list_del_rcu(&rupd->rule->list);
1866 }
1867
1868 /* Make sure we don't see any packet accessing aborted rules */
1869 synchronize_rcu();
1870
1871 list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) {
1858 nf_tables_rule_destroy(rupd->rule); 1872 nf_tables_rule_destroy(rupd->rule);
1873 list_del(&rupd->list);
1859 kfree(rupd); 1874 kfree(rupd);
1860 } 1875 }
1876
1861 return 0; 1877 return 0;
1862} 1878}
1863 1879
@@ -1943,6 +1959,9 @@ static int nft_ctx_init_from_setattr(struct nft_ctx *ctx,
1943 } 1959 }
1944 1960
1945 if (nla[NFTA_SET_TABLE] != NULL) { 1961 if (nla[NFTA_SET_TABLE] != NULL) {
1962 if (afi == NULL)
1963 return -EAFNOSUPPORT;
1964
1946 table = nf_tables_table_lookup(afi, nla[NFTA_SET_TABLE]); 1965 table = nf_tables_table_lookup(afi, nla[NFTA_SET_TABLE]);
1947 if (IS_ERR(table)) 1966 if (IS_ERR(table))
1948 return PTR_ERR(table); 1967 return PTR_ERR(table);
@@ -1989,13 +2008,13 @@ static int nf_tables_set_alloc_name(struct nft_ctx *ctx, struct nft_set *set,
1989 2008
1990 if (!sscanf(i->name, name, &tmp)) 2009 if (!sscanf(i->name, name, &tmp))
1991 continue; 2010 continue;
1992 if (tmp < 0 || tmp > BITS_PER_LONG * PAGE_SIZE) 2011 if (tmp < 0 || tmp >= BITS_PER_BYTE * PAGE_SIZE)
1993 continue; 2012 continue;
1994 2013
1995 set_bit(tmp, inuse); 2014 set_bit(tmp, inuse);
1996 } 2015 }
1997 2016
1998 n = find_first_zero_bit(inuse, BITS_PER_LONG * PAGE_SIZE); 2017 n = find_first_zero_bit(inuse, BITS_PER_BYTE * PAGE_SIZE);
1999 free_page((unsigned long)inuse); 2018 free_page((unsigned long)inuse);
2000 } 2019 }
2001 2020
@@ -2428,6 +2447,8 @@ static int nf_tables_delset(struct sock *nlsk, struct sk_buff *skb,
2428 struct nft_ctx ctx; 2447 struct nft_ctx ctx;
2429 int err; 2448 int err;
2430 2449
2450 if (nfmsg->nfgen_family == NFPROTO_UNSPEC)
2451 return -EAFNOSUPPORT;
2431 if (nla[NFTA_SET_TABLE] == NULL) 2452 if (nla[NFTA_SET_TABLE] == NULL)
2432 return -EINVAL; 2453 return -EINVAL;
2433 2454
@@ -2435,9 +2456,6 @@ static int nf_tables_delset(struct sock *nlsk, struct sk_buff *skb,
2435 if (err < 0) 2456 if (err < 0)
2436 return err; 2457 return err;
2437 2458
2438 if (nfmsg->nfgen_family == NFPROTO_UNSPEC)
2439 return -EAFNOSUPPORT;
2440
2441 set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_NAME]); 2459 set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_NAME]);
2442 if (IS_ERR(set)) 2460 if (IS_ERR(set))
2443 return PTR_ERR(set); 2461 return PTR_ERR(set);
@@ -2723,6 +2741,9 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set,
2723 if (nla[NFTA_SET_ELEM_DATA] == NULL && 2741 if (nla[NFTA_SET_ELEM_DATA] == NULL &&
2724 !(elem.flags & NFT_SET_ELEM_INTERVAL_END)) 2742 !(elem.flags & NFT_SET_ELEM_INTERVAL_END))
2725 return -EINVAL; 2743 return -EINVAL;
2744 if (nla[NFTA_SET_ELEM_DATA] != NULL &&
2745 elem.flags & NFT_SET_ELEM_INTERVAL_END)
2746 return -EINVAL;
2726 } else { 2747 } else {
2727 if (nla[NFTA_SET_ELEM_DATA] != NULL) 2748 if (nla[NFTA_SET_ELEM_DATA] != NULL)
2728 return -EINVAL; 2749 return -EINVAL;
@@ -2977,6 +2998,9 @@ static int nf_tables_loop_check_setelem(const struct nft_ctx *ctx,
2977 const struct nft_set_iter *iter, 2998 const struct nft_set_iter *iter,
2978 const struct nft_set_elem *elem) 2999 const struct nft_set_elem *elem)
2979{ 3000{
3001 if (elem->flags & NFT_SET_ELEM_INTERVAL_END)
3002 return 0;
3003
2980 switch (elem->data.verdict) { 3004 switch (elem->data.verdict) {
2981 case NFT_JUMP: 3005 case NFT_JUMP:
2982 case NFT_GOTO: 3006 case NFT_GOTO:
diff --git a/net/netfilter/nf_tables_core.c b/net/netfilter/nf_tables_core.c
index 0d879fcb8763..90998a6ff8b9 100644
--- a/net/netfilter/nf_tables_core.c
+++ b/net/netfilter/nf_tables_core.c
@@ -103,9 +103,9 @@ static struct nf_loginfo trace_loginfo = {
103 }, 103 },
104}; 104};
105 105
106static inline void nft_trace_packet(const struct nft_pktinfo *pkt, 106static void nft_trace_packet(const struct nft_pktinfo *pkt,
107 const struct nft_chain *chain, 107 const struct nft_chain *chain,
108 int rulenum, enum nft_trace type) 108 int rulenum, enum nft_trace type)
109{ 109{
110 struct net *net = dev_net(pkt->in ? pkt->in : pkt->out); 110 struct net *net = dev_net(pkt->in ? pkt->in : pkt->out);
111 111
diff --git a/net/netfilter/nft_ct.c b/net/netfilter/nft_ct.c
index 917052e20602..46e275403838 100644
--- a/net/netfilter/nft_ct.c
+++ b/net/netfilter/nft_ct.c
@@ -226,6 +226,7 @@ static int nft_ct_init_validate_get(const struct nft_expr *expr,
226 if (tb[NFTA_CT_DIRECTION] != NULL) 226 if (tb[NFTA_CT_DIRECTION] != NULL)
227 return -EINVAL; 227 return -EINVAL;
228 break; 228 break;
229 case NFT_CT_L3PROTOCOL:
229 case NFT_CT_PROTOCOL: 230 case NFT_CT_PROTOCOL:
230 case NFT_CT_SRC: 231 case NFT_CT_SRC:
231 case NFT_CT_DST: 232 case NFT_CT_DST:
@@ -311,8 +312,19 @@ static int nft_ct_get_dump(struct sk_buff *skb, const struct nft_expr *expr)
311 goto nla_put_failure; 312 goto nla_put_failure;
312 if (nla_put_be32(skb, NFTA_CT_KEY, htonl(priv->key))) 313 if (nla_put_be32(skb, NFTA_CT_KEY, htonl(priv->key)))
313 goto nla_put_failure; 314 goto nla_put_failure;
314 if (nla_put_u8(skb, NFTA_CT_DIRECTION, priv->dir)) 315
315 goto nla_put_failure; 316 switch (priv->key) {
317 case NFT_CT_PROTOCOL:
318 case NFT_CT_SRC:
319 case NFT_CT_DST:
320 case NFT_CT_PROTO_SRC:
321 case NFT_CT_PROTO_DST:
322 if (nla_put_u8(skb, NFTA_CT_DIRECTION, priv->dir))
323 goto nla_put_failure;
324 default:
325 break;
326 }
327
316 return 0; 328 return 0;
317 329
318nla_put_failure: 330nla_put_failure:
diff --git a/net/netfilter/nft_log.c b/net/netfilter/nft_log.c
index 5af790123ad8..26c5154e05f3 100644
--- a/net/netfilter/nft_log.c
+++ b/net/netfilter/nft_log.c
@@ -23,7 +23,6 @@ static const char *nft_log_null_prefix = "";
23struct nft_log { 23struct nft_log {
24 struct nf_loginfo loginfo; 24 struct nf_loginfo loginfo;
25 char *prefix; 25 char *prefix;
26 int family;
27}; 26};
28 27
29static void nft_log_eval(const struct nft_expr *expr, 28static void nft_log_eval(const struct nft_expr *expr,
@@ -33,7 +32,7 @@ static void nft_log_eval(const struct nft_expr *expr,
33 const struct nft_log *priv = nft_expr_priv(expr); 32 const struct nft_log *priv = nft_expr_priv(expr);
34 struct net *net = dev_net(pkt->in ? pkt->in : pkt->out); 33 struct net *net = dev_net(pkt->in ? pkt->in : pkt->out);
35 34
36 nf_log_packet(net, priv->family, pkt->ops->hooknum, pkt->skb, pkt->in, 35 nf_log_packet(net, pkt->ops->pf, pkt->ops->hooknum, pkt->skb, pkt->in,
37 pkt->out, &priv->loginfo, "%s", priv->prefix); 36 pkt->out, &priv->loginfo, "%s", priv->prefix);
38} 37}
39 38
@@ -52,8 +51,6 @@ static int nft_log_init(const struct nft_ctx *ctx,
52 struct nf_loginfo *li = &priv->loginfo; 51 struct nf_loginfo *li = &priv->loginfo;
53 const struct nlattr *nla; 52 const struct nlattr *nla;
54 53
55 priv->family = ctx->afi->family;
56
57 nla = tb[NFTA_LOG_PREFIX]; 54 nla = tb[NFTA_LOG_PREFIX];
58 if (nla != NULL) { 55 if (nla != NULL) {
59 priv->prefix = kmalloc(nla_len(nla) + 1, GFP_KERNEL); 56 priv->prefix = kmalloc(nla_len(nla) + 1, GFP_KERNEL);
diff --git a/net/netfilter/nft_lookup.c b/net/netfilter/nft_lookup.c
index 8a6116b75b5a..bb4ef4cccb6e 100644
--- a/net/netfilter/nft_lookup.c
+++ b/net/netfilter/nft_lookup.c
@@ -16,6 +16,7 @@
16#include <linux/netfilter.h> 16#include <linux/netfilter.h>
17#include <linux/netfilter/nf_tables.h> 17#include <linux/netfilter/nf_tables.h>
18#include <net/netfilter/nf_tables.h> 18#include <net/netfilter/nf_tables.h>
19#include <net/netfilter/nf_tables_core.h>
19 20
20struct nft_lookup { 21struct nft_lookup {
21 struct nft_set *set; 22 struct nft_set *set;
diff --git a/net/netfilter/nft_meta.c b/net/netfilter/nft_meta.c
index e8254ad2e5a9..425cf39af890 100644
--- a/net/netfilter/nft_meta.c
+++ b/net/netfilter/nft_meta.c
@@ -116,7 +116,7 @@ static void nft_meta_get_eval(const struct nft_expr *expr,
116 skb->sk->sk_socket->file->f_cred->fsgid); 116 skb->sk->sk_socket->file->f_cred->fsgid);
117 read_unlock_bh(&skb->sk->sk_callback_lock); 117 read_unlock_bh(&skb->sk->sk_callback_lock);
118 break; 118 break;
119#ifdef CONFIG_NET_CLS_ROUTE 119#ifdef CONFIG_IP_ROUTE_CLASSID
120 case NFT_META_RTCLASSID: { 120 case NFT_META_RTCLASSID: {
121 const struct dst_entry *dst = skb_dst(skb); 121 const struct dst_entry *dst = skb_dst(skb);
122 122
@@ -199,7 +199,7 @@ static int nft_meta_init_validate_get(uint32_t key)
199 case NFT_META_OIFTYPE: 199 case NFT_META_OIFTYPE:
200 case NFT_META_SKUID: 200 case NFT_META_SKUID:
201 case NFT_META_SKGID: 201 case NFT_META_SKGID:
202#ifdef CONFIG_NET_CLS_ROUTE 202#ifdef CONFIG_IP_ROUTE_CLASSID
203 case NFT_META_RTCLASSID: 203 case NFT_META_RTCLASSID:
204#endif 204#endif
205#ifdef CONFIG_NETWORK_SECMARK 205#ifdef CONFIG_NETWORK_SECMARK
diff --git a/net/netfilter/nft_payload.c b/net/netfilter/nft_payload.c
index a2aeb318678f..85daa84bfdfe 100644
--- a/net/netfilter/nft_payload.c
+++ b/net/netfilter/nft_payload.c
@@ -135,7 +135,8 @@ nft_payload_select_ops(const struct nft_ctx *ctx,
135 if (len == 0 || len > FIELD_SIZEOF(struct nft_data, data)) 135 if (len == 0 || len > FIELD_SIZEOF(struct nft_data, data))
136 return ERR_PTR(-EINVAL); 136 return ERR_PTR(-EINVAL);
137 137
138 if (len <= 4 && IS_ALIGNED(offset, len) && base != NFT_PAYLOAD_LL_HEADER) 138 if (len <= 4 && is_power_of_2(len) && IS_ALIGNED(offset, len) &&
139 base != NFT_PAYLOAD_LL_HEADER)
139 return &nft_payload_fast_ops; 140 return &nft_payload_fast_ops;
140 else 141 else
141 return &nft_payload_ops; 142 return &nft_payload_ops;
diff --git a/net/netfilter/nft_queue.c b/net/netfilter/nft_queue.c
index cbea473d69e9..e8ae2f6bf232 100644
--- a/net/netfilter/nft_queue.c
+++ b/net/netfilter/nft_queue.c
@@ -25,7 +25,6 @@ struct nft_queue {
25 u16 queuenum; 25 u16 queuenum;
26 u16 queues_total; 26 u16 queues_total;
27 u16 flags; 27 u16 flags;
28 u8 family;
29}; 28};
30 29
31static void nft_queue_eval(const struct nft_expr *expr, 30static void nft_queue_eval(const struct nft_expr *expr,
@@ -43,7 +42,7 @@ static void nft_queue_eval(const struct nft_expr *expr,
43 queue = priv->queuenum + cpu % priv->queues_total; 42 queue = priv->queuenum + cpu % priv->queues_total;
44 } else { 43 } else {
45 queue = nfqueue_hash(pkt->skb, queue, 44 queue = nfqueue_hash(pkt->skb, queue,
46 priv->queues_total, priv->family, 45 priv->queues_total, pkt->ops->pf,
47 jhash_initval); 46 jhash_initval);
48 } 47 }
49 } 48 }
@@ -71,7 +70,6 @@ static int nft_queue_init(const struct nft_ctx *ctx,
71 return -EINVAL; 70 return -EINVAL;
72 71
73 init_hashrandom(&jhash_initval); 72 init_hashrandom(&jhash_initval);
74 priv->family = ctx->afi->family;
75 priv->queuenum = ntohs(nla_get_be16(tb[NFTA_QUEUE_NUM])); 73 priv->queuenum = ntohs(nla_get_be16(tb[NFTA_QUEUE_NUM]));
76 74
77 if (tb[NFTA_QUEUE_TOTAL] != NULL) 75 if (tb[NFTA_QUEUE_TOTAL] != NULL)
diff --git a/net/netfilter/nft_rbtree.c b/net/netfilter/nft_rbtree.c
index ca0c1b231bfe..e21d69d13506 100644
--- a/net/netfilter/nft_rbtree.c
+++ b/net/netfilter/nft_rbtree.c
@@ -69,8 +69,10 @@ static void nft_rbtree_elem_destroy(const struct nft_set *set,
69 struct nft_rbtree_elem *rbe) 69 struct nft_rbtree_elem *rbe)
70{ 70{
71 nft_data_uninit(&rbe->key, NFT_DATA_VALUE); 71 nft_data_uninit(&rbe->key, NFT_DATA_VALUE);
72 if (set->flags & NFT_SET_MAP) 72 if (set->flags & NFT_SET_MAP &&
73 !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
73 nft_data_uninit(rbe->data, set->dtype); 74 nft_data_uninit(rbe->data, set->dtype);
75
74 kfree(rbe); 76 kfree(rbe);
75} 77}
76 78
@@ -108,7 +110,8 @@ static int nft_rbtree_insert(const struct nft_set *set,
108 int err; 110 int err;
109 111
110 size = sizeof(*rbe); 112 size = sizeof(*rbe);
111 if (set->flags & NFT_SET_MAP) 113 if (set->flags & NFT_SET_MAP &&
114 !(elem->flags & NFT_SET_ELEM_INTERVAL_END))
112 size += sizeof(rbe->data[0]); 115 size += sizeof(rbe->data[0]);
113 116
114 rbe = kzalloc(size, GFP_KERNEL); 117 rbe = kzalloc(size, GFP_KERNEL);
@@ -117,7 +120,8 @@ static int nft_rbtree_insert(const struct nft_set *set,
117 120
118 rbe->flags = elem->flags; 121 rbe->flags = elem->flags;
119 nft_data_copy(&rbe->key, &elem->key); 122 nft_data_copy(&rbe->key, &elem->key);
120 if (set->flags & NFT_SET_MAP) 123 if (set->flags & NFT_SET_MAP &&
124 !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
121 nft_data_copy(rbe->data, &elem->data); 125 nft_data_copy(rbe->data, &elem->data);
122 126
123 err = __nft_rbtree_insert(set, rbe); 127 err = __nft_rbtree_insert(set, rbe);
@@ -153,7 +157,8 @@ static int nft_rbtree_get(const struct nft_set *set, struct nft_set_elem *elem)
153 parent = parent->rb_right; 157 parent = parent->rb_right;
154 else { 158 else {
155 elem->cookie = rbe; 159 elem->cookie = rbe;
156 if (set->flags & NFT_SET_MAP) 160 if (set->flags & NFT_SET_MAP &&
161 !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
157 nft_data_copy(&elem->data, rbe->data); 162 nft_data_copy(&elem->data, rbe->data);
158 elem->flags = rbe->flags; 163 elem->flags = rbe->flags;
159 return 0; 164 return 0;
@@ -177,7 +182,8 @@ static void nft_rbtree_walk(const struct nft_ctx *ctx,
177 182
178 rbe = rb_entry(node, struct nft_rbtree_elem, node); 183 rbe = rb_entry(node, struct nft_rbtree_elem, node);
179 nft_data_copy(&elem.key, &rbe->key); 184 nft_data_copy(&elem.key, &rbe->key);
180 if (set->flags & NFT_SET_MAP) 185 if (set->flags & NFT_SET_MAP &&
186 !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
181 nft_data_copy(&elem.data, rbe->data); 187 nft_data_copy(&elem.data, rbe->data);
182 elem.flags = rbe->flags; 188 elem.flags = rbe->flags;
183 189
diff --git a/net/netfilter/nft_reject.c b/net/netfilter/nft_reject.c
index 5e204711d704..f3448c296446 100644
--- a/net/netfilter/nft_reject.c
+++ b/net/netfilter/nft_reject.c
@@ -16,65 +16,23 @@
16#include <linux/netfilter.h> 16#include <linux/netfilter.h>
17#include <linux/netfilter/nf_tables.h> 17#include <linux/netfilter/nf_tables.h>
18#include <net/netfilter/nf_tables.h> 18#include <net/netfilter/nf_tables.h>
19#include <net/icmp.h> 19#include <net/netfilter/nft_reject.h>
20#include <net/netfilter/ipv4/nf_reject.h>
21 20
22#if IS_ENABLED(CONFIG_NF_TABLES_IPV6) 21const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
23#include <net/netfilter/ipv6/nf_reject.h>
24#endif
25
26struct nft_reject {
27 enum nft_reject_types type:8;
28 u8 icmp_code;
29 u8 family;
30};
31
32static void nft_reject_eval(const struct nft_expr *expr,
33 struct nft_data data[NFT_REG_MAX + 1],
34 const struct nft_pktinfo *pkt)
35{
36 struct nft_reject *priv = nft_expr_priv(expr);
37#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
38 struct net *net = dev_net((pkt->in != NULL) ? pkt->in : pkt->out);
39#endif
40 switch (priv->type) {
41 case NFT_REJECT_ICMP_UNREACH:
42 if (priv->family == NFPROTO_IPV4)
43 nf_send_unreach(pkt->skb, priv->icmp_code);
44#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
45 else if (priv->family == NFPROTO_IPV6)
46 nf_send_unreach6(net, pkt->skb, priv->icmp_code,
47 pkt->ops->hooknum);
48#endif
49 break;
50 case NFT_REJECT_TCP_RST:
51 if (priv->family == NFPROTO_IPV4)
52 nf_send_reset(pkt->skb, pkt->ops->hooknum);
53#if IS_ENABLED(CONFIG_NF_TABLES_IPV6)
54 else if (priv->family == NFPROTO_IPV6)
55 nf_send_reset6(net, pkt->skb, pkt->ops->hooknum);
56#endif
57 break;
58 }
59
60 data[NFT_REG_VERDICT].verdict = NF_DROP;
61}
62
63static const struct nla_policy nft_reject_policy[NFTA_REJECT_MAX + 1] = {
64 [NFTA_REJECT_TYPE] = { .type = NLA_U32 }, 22 [NFTA_REJECT_TYPE] = { .type = NLA_U32 },
65 [NFTA_REJECT_ICMP_CODE] = { .type = NLA_U8 }, 23 [NFTA_REJECT_ICMP_CODE] = { .type = NLA_U8 },
66}; 24};
25EXPORT_SYMBOL_GPL(nft_reject_policy);
67 26
68static int nft_reject_init(const struct nft_ctx *ctx, 27int nft_reject_init(const struct nft_ctx *ctx,
69 const struct nft_expr *expr, 28 const struct nft_expr *expr,
70 const struct nlattr * const tb[]) 29 const struct nlattr * const tb[])
71{ 30{
72 struct nft_reject *priv = nft_expr_priv(expr); 31 struct nft_reject *priv = nft_expr_priv(expr);
73 32
74 if (tb[NFTA_REJECT_TYPE] == NULL) 33 if (tb[NFTA_REJECT_TYPE] == NULL)
75 return -EINVAL; 34 return -EINVAL;
76 35
77 priv->family = ctx->afi->family;
78 priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE])); 36 priv->type = ntohl(nla_get_be32(tb[NFTA_REJECT_TYPE]));
79 switch (priv->type) { 37 switch (priv->type) {
80 case NFT_REJECT_ICMP_UNREACH: 38 case NFT_REJECT_ICMP_UNREACH:
@@ -89,8 +47,9 @@ static int nft_reject_init(const struct nft_ctx *ctx,
89 47
90 return 0; 48 return 0;
91} 49}
50EXPORT_SYMBOL_GPL(nft_reject_init);
92 51
93static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr) 52int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
94{ 53{
95 const struct nft_reject *priv = nft_expr_priv(expr); 54 const struct nft_reject *priv = nft_expr_priv(expr);
96 55
@@ -109,37 +68,7 @@ static int nft_reject_dump(struct sk_buff *skb, const struct nft_expr *expr)
109nla_put_failure: 68nla_put_failure:
110 return -1; 69 return -1;
111} 70}
112 71EXPORT_SYMBOL_GPL(nft_reject_dump);
113static struct nft_expr_type nft_reject_type;
114static const struct nft_expr_ops nft_reject_ops = {
115 .type = &nft_reject_type,
116 .size = NFT_EXPR_SIZE(sizeof(struct nft_reject)),
117 .eval = nft_reject_eval,
118 .init = nft_reject_init,
119 .dump = nft_reject_dump,
120};
121
122static struct nft_expr_type nft_reject_type __read_mostly = {
123 .name = "reject",
124 .ops = &nft_reject_ops,
125 .policy = nft_reject_policy,
126 .maxattr = NFTA_REJECT_MAX,
127 .owner = THIS_MODULE,
128};
129
130static int __init nft_reject_module_init(void)
131{
132 return nft_register_expr(&nft_reject_type);
133}
134
135static void __exit nft_reject_module_exit(void)
136{
137 nft_unregister_expr(&nft_reject_type);
138}
139
140module_init(nft_reject_module_init);
141module_exit(nft_reject_module_exit);
142 72
143MODULE_LICENSE("GPL"); 73MODULE_LICENSE("GPL");
144MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>"); 74MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
145MODULE_ALIAS_NFT_EXPR("reject");
diff --git a/net/netfilter/nft_reject_inet.c b/net/netfilter/nft_reject_inet.c
new file mode 100644
index 000000000000..b718a52a4654
--- /dev/null
+++ b/net/netfilter/nft_reject_inet.c
@@ -0,0 +1,63 @@
1/*
2 * Copyright (c) 2014 Patrick McHardy <kaber@trash.net>
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License version 2 as
6 * published by the Free Software Foundation.
7 */
8
9#include <linux/kernel.h>
10#include <linux/init.h>
11#include <linux/module.h>
12#include <linux/netlink.h>
13#include <linux/netfilter.h>
14#include <linux/netfilter/nf_tables.h>
15#include <net/netfilter/nf_tables.h>
16#include <net/netfilter/nft_reject.h>
17
18static void nft_reject_inet_eval(const struct nft_expr *expr,
19 struct nft_data data[NFT_REG_MAX + 1],
20 const struct nft_pktinfo *pkt)
21{
22 switch (pkt->ops->pf) {
23 case NFPROTO_IPV4:
24 return nft_reject_ipv4_eval(expr, data, pkt);
25 case NFPROTO_IPV6:
26 return nft_reject_ipv6_eval(expr, data, pkt);
27 }
28}
29
30static struct nft_expr_type nft_reject_inet_type;
31static const struct nft_expr_ops nft_reject_inet_ops = {
32 .type = &nft_reject_inet_type,
33 .size = NFT_EXPR_SIZE(sizeof(struct nft_reject)),
34 .eval = nft_reject_inet_eval,
35 .init = nft_reject_init,
36 .dump = nft_reject_dump,
37};
38
39static struct nft_expr_type nft_reject_inet_type __read_mostly = {
40 .family = NFPROTO_INET,
41 .name = "reject",
42 .ops = &nft_reject_inet_ops,
43 .policy = nft_reject_policy,
44 .maxattr = NFTA_REJECT_MAX,
45 .owner = THIS_MODULE,
46};
47
48static int __init nft_reject_inet_module_init(void)
49{
50 return nft_register_expr(&nft_reject_inet_type);
51}
52
53static void __exit nft_reject_inet_module_exit(void)
54{
55 nft_unregister_expr(&nft_reject_inet_type);
56}
57
58module_init(nft_reject_inet_module_init);
59module_exit(nft_reject_inet_module_exit);
60
61MODULE_LICENSE("GPL");
62MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
63MODULE_ALIAS_NFT_AF_EXPR(1, "reject");
diff --git a/net/netfilter/xt_CT.c b/net/netfilter/xt_CT.c
index 5929be622c5c..75747aecdebe 100644
--- a/net/netfilter/xt_CT.c
+++ b/net/netfilter/xt_CT.c
@@ -228,12 +228,7 @@ static int xt_ct_tg_check(const struct xt_tgchk_param *par,
228 goto err3; 228 goto err3;
229 } 229 }
230 230
231 __set_bit(IPS_TEMPLATE_BIT, &ct->status); 231 nf_conntrack_tmpl_insert(par->net, ct);
232 __set_bit(IPS_CONFIRMED_BIT, &ct->status);
233
234 /* Overload tuple linked list to put us in template list. */
235 hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
236 &par->net->ct.tmpl);
237out: 232out:
238 info->ct = ct; 233 info->ct = ct;
239 return 0; 234 return 0;
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index fdf51353cf78..04748ab649c2 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -1489,8 +1489,8 @@ static int netlink_connect(struct socket *sock, struct sockaddr *addr,
1489 if (addr->sa_family != AF_NETLINK) 1489 if (addr->sa_family != AF_NETLINK)
1490 return -EINVAL; 1490 return -EINVAL;
1491 1491
1492 /* Only superuser is allowed to send multicasts */ 1492 if ((nladdr->nl_groups || nladdr->nl_pid) &&
1493 if (nladdr->nl_groups && !netlink_capable(sock, NL_CFG_F_NONROOT_SEND)) 1493 !netlink_capable(sock, NL_CFG_F_NONROOT_SEND))
1494 return -EPERM; 1494 return -EPERM;
1495 1495
1496 if (!nlk->portid) 1496 if (!nlk->portid)
diff --git a/net/nfc/nci/core.c b/net/nfc/nci/core.c
index 46bda010bf11..56db888b1cd5 100644
--- a/net/nfc/nci/core.c
+++ b/net/nfc/nci/core.c
@@ -301,7 +301,7 @@ static int nci_open_device(struct nci_dev *ndev)
301 rc = __nci_request(ndev, nci_reset_req, 0, 301 rc = __nci_request(ndev, nci_reset_req, 0,
302 msecs_to_jiffies(NCI_RESET_TIMEOUT)); 302 msecs_to_jiffies(NCI_RESET_TIMEOUT));
303 303
304 if (ndev->ops->setup(ndev)) 304 if (ndev->ops->setup)
305 ndev->ops->setup(ndev); 305 ndev->ops->setup(ndev);
306 306
307 if (!rc) { 307 if (!rc) {
diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
index df4692826ead..e9a48baf8551 100644
--- a/net/openvswitch/datapath.c
+++ b/net/openvswitch/datapath.c
@@ -55,6 +55,7 @@
55 55
56#include "datapath.h" 56#include "datapath.h"
57#include "flow.h" 57#include "flow.h"
58#include "flow_table.h"
58#include "flow_netlink.h" 59#include "flow_netlink.h"
59#include "vport-internal_dev.h" 60#include "vport-internal_dev.h"
60#include "vport-netdev.h" 61#include "vport-netdev.h"
@@ -160,7 +161,6 @@ static void destroy_dp_rcu(struct rcu_head *rcu)
160{ 161{
161 struct datapath *dp = container_of(rcu, struct datapath, rcu); 162 struct datapath *dp = container_of(rcu, struct datapath, rcu);
162 163
163 ovs_flow_tbl_destroy(&dp->table);
164 free_percpu(dp->stats_percpu); 164 free_percpu(dp->stats_percpu);
165 release_net(ovs_dp_get_net(dp)); 165 release_net(ovs_dp_get_net(dp));
166 kfree(dp->ports); 166 kfree(dp->ports);
@@ -466,6 +466,14 @@ static int queue_userspace_packet(struct datapath *dp, struct sk_buff *skb,
466 466
467 skb_zerocopy(user_skb, skb, skb->len, hlen); 467 skb_zerocopy(user_skb, skb, skb->len, hlen);
468 468
469 /* Pad OVS_PACKET_ATTR_PACKET if linear copy was performed */
470 if (!(dp->user_features & OVS_DP_F_UNALIGNED)) {
471 size_t plen = NLA_ALIGN(user_skb->len) - user_skb->len;
472
473 if (plen > 0)
474 memset(skb_put(user_skb, plen), 0, plen);
475 }
476
469 ((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len; 477 ((struct nlmsghdr *) user_skb->data)->nlmsg_len = user_skb->len;
470 478
471 err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid); 479 err = genlmsg_unicast(ovs_dp_get_net(dp), user_skb, upcall_info->portid);
@@ -852,11 +860,8 @@ static int ovs_flow_cmd_new_or_set(struct sk_buff *skb, struct genl_info *info)
852 goto err_unlock_ovs; 860 goto err_unlock_ovs;
853 861
854 /* The unmasked key has to be the same for flow updates. */ 862 /* The unmasked key has to be the same for flow updates. */
855 error = -EINVAL; 863 if (!ovs_flow_cmp_unmasked_key(flow, &match))
856 if (!ovs_flow_cmp_unmasked_key(flow, &match)) {
857 OVS_NLERR("Flow modification message rejected, unmasked key does not match.\n");
858 goto err_unlock_ovs; 864 goto err_unlock_ovs;
859 }
860 865
861 /* Update actions. */ 866 /* Update actions. */
862 old_acts = ovsl_dereference(flow->sf_acts); 867 old_acts = ovsl_dereference(flow->sf_acts);
@@ -1079,6 +1084,7 @@ static size_t ovs_dp_cmd_msg_size(void)
1079 msgsize += nla_total_size(IFNAMSIZ); 1084 msgsize += nla_total_size(IFNAMSIZ);
1080 msgsize += nla_total_size(sizeof(struct ovs_dp_stats)); 1085 msgsize += nla_total_size(sizeof(struct ovs_dp_stats));
1081 msgsize += nla_total_size(sizeof(struct ovs_dp_megaflow_stats)); 1086 msgsize += nla_total_size(sizeof(struct ovs_dp_megaflow_stats));
1087 msgsize += nla_total_size(sizeof(u32)); /* OVS_DP_ATTR_USER_FEATURES */
1082 1088
1083 return msgsize; 1089 return msgsize;
1084} 1090}
@@ -1279,7 +1285,7 @@ err_destroy_ports_array:
1279err_destroy_percpu: 1285err_destroy_percpu:
1280 free_percpu(dp->stats_percpu); 1286 free_percpu(dp->stats_percpu);
1281err_destroy_table: 1287err_destroy_table:
1282 ovs_flow_tbl_destroy(&dp->table); 1288 ovs_flow_tbl_destroy(&dp->table, false);
1283err_free_dp: 1289err_free_dp:
1284 release_net(ovs_dp_get_net(dp)); 1290 release_net(ovs_dp_get_net(dp));
1285 kfree(dp); 1291 kfree(dp);
@@ -1306,10 +1312,13 @@ static void __dp_destroy(struct datapath *dp)
1306 list_del_rcu(&dp->list_node); 1312 list_del_rcu(&dp->list_node);
1307 1313
1308 /* OVSP_LOCAL is datapath internal port. We need to make sure that 1314 /* OVSP_LOCAL is datapath internal port. We need to make sure that
1309 * all port in datapath are destroyed first before freeing datapath. 1315 * all ports in datapath are destroyed first before freeing datapath.
1310 */ 1316 */
1311 ovs_dp_detach_port(ovs_vport_ovsl(dp, OVSP_LOCAL)); 1317 ovs_dp_detach_port(ovs_vport_ovsl(dp, OVSP_LOCAL));
1312 1318
1319 /* RCU destroy the flow table */
1320 ovs_flow_tbl_destroy(&dp->table, true);
1321
1313 call_rcu(&dp->rcu, destroy_dp_rcu); 1322 call_rcu(&dp->rcu, destroy_dp_rcu);
1314} 1323}
1315 1324
diff --git a/net/openvswitch/flow_table.c b/net/openvswitch/flow_table.c
index c58a0fe3c889..3c268b3d71c3 100644
--- a/net/openvswitch/flow_table.c
+++ b/net/openvswitch/flow_table.c
@@ -153,29 +153,29 @@ static void rcu_free_flow_callback(struct rcu_head *rcu)
153 flow_free(flow); 153 flow_free(flow);
154} 154}
155 155
156static void flow_mask_del_ref(struct sw_flow_mask *mask, bool deferred)
157{
158 if (!mask)
159 return;
160
161 BUG_ON(!mask->ref_count);
162 mask->ref_count--;
163
164 if (!mask->ref_count) {
165 list_del_rcu(&mask->list);
166 if (deferred)
167 kfree_rcu(mask, rcu);
168 else
169 kfree(mask);
170 }
171}
172
173void ovs_flow_free(struct sw_flow *flow, bool deferred) 156void ovs_flow_free(struct sw_flow *flow, bool deferred)
174{ 157{
175 if (!flow) 158 if (!flow)
176 return; 159 return;
177 160
178 flow_mask_del_ref(flow->mask, deferred); 161 if (flow->mask) {
162 struct sw_flow_mask *mask = flow->mask;
163
164 /* ovs-lock is required to protect mask-refcount and
165 * mask list.
166 */
167 ASSERT_OVSL();
168 BUG_ON(!mask->ref_count);
169 mask->ref_count--;
170
171 if (!mask->ref_count) {
172 list_del_rcu(&mask->list);
173 if (deferred)
174 kfree_rcu(mask, rcu);
175 else
176 kfree(mask);
177 }
178 }
179 179
180 if (deferred) 180 if (deferred)
181 call_rcu(&flow->rcu, rcu_free_flow_callback); 181 call_rcu(&flow->rcu, rcu_free_flow_callback);
@@ -188,26 +188,9 @@ static void free_buckets(struct flex_array *buckets)
188 flex_array_free(buckets); 188 flex_array_free(buckets);
189} 189}
190 190
191
191static void __table_instance_destroy(struct table_instance *ti) 192static void __table_instance_destroy(struct table_instance *ti)
192{ 193{
193 int i;
194
195 if (ti->keep_flows)
196 goto skip_flows;
197
198 for (i = 0; i < ti->n_buckets; i++) {
199 struct sw_flow *flow;
200 struct hlist_head *head = flex_array_get(ti->buckets, i);
201 struct hlist_node *n;
202 int ver = ti->node_ver;
203
204 hlist_for_each_entry_safe(flow, n, head, hash_node[ver]) {
205 hlist_del(&flow->hash_node[ver]);
206 ovs_flow_free(flow, false);
207 }
208 }
209
210skip_flows:
211 free_buckets(ti->buckets); 194 free_buckets(ti->buckets);
212 kfree(ti); 195 kfree(ti);
213} 196}
@@ -258,20 +241,38 @@ static void flow_tbl_destroy_rcu_cb(struct rcu_head *rcu)
258 241
259static void table_instance_destroy(struct table_instance *ti, bool deferred) 242static void table_instance_destroy(struct table_instance *ti, bool deferred)
260{ 243{
244 int i;
245
261 if (!ti) 246 if (!ti)
262 return; 247 return;
263 248
249 if (ti->keep_flows)
250 goto skip_flows;
251
252 for (i = 0; i < ti->n_buckets; i++) {
253 struct sw_flow *flow;
254 struct hlist_head *head = flex_array_get(ti->buckets, i);
255 struct hlist_node *n;
256 int ver = ti->node_ver;
257
258 hlist_for_each_entry_safe(flow, n, head, hash_node[ver]) {
259 hlist_del_rcu(&flow->hash_node[ver]);
260 ovs_flow_free(flow, deferred);
261 }
262 }
263
264skip_flows:
264 if (deferred) 265 if (deferred)
265 call_rcu(&ti->rcu, flow_tbl_destroy_rcu_cb); 266 call_rcu(&ti->rcu, flow_tbl_destroy_rcu_cb);
266 else 267 else
267 __table_instance_destroy(ti); 268 __table_instance_destroy(ti);
268} 269}
269 270
270void ovs_flow_tbl_destroy(struct flow_table *table) 271void ovs_flow_tbl_destroy(struct flow_table *table, bool deferred)
271{ 272{
272 struct table_instance *ti = ovsl_dereference(table->ti); 273 struct table_instance *ti = ovsl_dereference(table->ti);
273 274
274 table_instance_destroy(ti, false); 275 table_instance_destroy(ti, deferred);
275} 276}
276 277
277struct sw_flow *ovs_flow_tbl_dump_next(struct table_instance *ti, 278struct sw_flow *ovs_flow_tbl_dump_next(struct table_instance *ti,
@@ -504,16 +505,11 @@ static struct sw_flow_mask *mask_alloc(void)
504 505
505 mask = kmalloc(sizeof(*mask), GFP_KERNEL); 506 mask = kmalloc(sizeof(*mask), GFP_KERNEL);
506 if (mask) 507 if (mask)
507 mask->ref_count = 0; 508 mask->ref_count = 1;
508 509
509 return mask; 510 return mask;
510} 511}
511 512
512static void mask_add_ref(struct sw_flow_mask *mask)
513{
514 mask->ref_count++;
515}
516
517static bool mask_equal(const struct sw_flow_mask *a, 513static bool mask_equal(const struct sw_flow_mask *a,
518 const struct sw_flow_mask *b) 514 const struct sw_flow_mask *b)
519{ 515{
@@ -554,9 +550,11 @@ static int flow_mask_insert(struct flow_table *tbl, struct sw_flow *flow,
554 mask->key = new->key; 550 mask->key = new->key;
555 mask->range = new->range; 551 mask->range = new->range;
556 list_add_rcu(&mask->list, &tbl->mask_list); 552 list_add_rcu(&mask->list, &tbl->mask_list);
553 } else {
554 BUG_ON(!mask->ref_count);
555 mask->ref_count++;
557 } 556 }
558 557
559 mask_add_ref(mask);
560 flow->mask = mask; 558 flow->mask = mask;
561 return 0; 559 return 0;
562} 560}
diff --git a/net/openvswitch/flow_table.h b/net/openvswitch/flow_table.h
index 1996e34c0fd8..baaeb101924d 100644
--- a/net/openvswitch/flow_table.h
+++ b/net/openvswitch/flow_table.h
@@ -60,7 +60,7 @@ void ovs_flow_free(struct sw_flow *, bool deferred);
60 60
61int ovs_flow_tbl_init(struct flow_table *); 61int ovs_flow_tbl_init(struct flow_table *);
62int ovs_flow_tbl_count(struct flow_table *table); 62int ovs_flow_tbl_count(struct flow_table *table);
63void ovs_flow_tbl_destroy(struct flow_table *table); 63void ovs_flow_tbl_destroy(struct flow_table *table, bool deferred);
64int ovs_flow_tbl_flush(struct flow_table *flow_table); 64int ovs_flow_tbl_flush(struct flow_table *flow_table);
65 65
66int ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow, 66int ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow,
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 6a2bb37506c5..48a6a93db296 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -308,11 +308,27 @@ static bool packet_use_direct_xmit(const struct packet_sock *po)
308 return po->xmit == packet_direct_xmit; 308 return po->xmit == packet_direct_xmit;
309} 309}
310 310
311static u16 packet_pick_tx_queue(struct net_device *dev) 311static u16 __packet_pick_tx_queue(struct net_device *dev, struct sk_buff *skb)
312{ 312{
313 return (u16) raw_smp_processor_id() % dev->real_num_tx_queues; 313 return (u16) raw_smp_processor_id() % dev->real_num_tx_queues;
314} 314}
315 315
316static void packet_pick_tx_queue(struct net_device *dev, struct sk_buff *skb)
317{
318 const struct net_device_ops *ops = dev->netdev_ops;
319 u16 queue_index;
320
321 if (ops->ndo_select_queue) {
322 queue_index = ops->ndo_select_queue(dev, skb, NULL,
323 __packet_pick_tx_queue);
324 queue_index = netdev_cap_txqueue(dev, queue_index);
325 } else {
326 queue_index = __packet_pick_tx_queue(dev, skb);
327 }
328
329 skb_set_queue_mapping(skb, queue_index);
330}
331
316/* register_prot_hook must be invoked with the po->bind_lock held, 332/* register_prot_hook must be invoked with the po->bind_lock held,
317 * or from a context in which asynchronous accesses to the packet 333 * or from a context in which asynchronous accesses to the packet
318 * socket is not possible (packet_create()). 334 * socket is not possible (packet_create()).
@@ -2285,7 +2301,8 @@ static int tpacket_snd(struct packet_sock *po, struct msghdr *msg)
2285 } 2301 }
2286 } 2302 }
2287 2303
2288 skb_set_queue_mapping(skb, packet_pick_tx_queue(dev)); 2304 packet_pick_tx_queue(dev, skb);
2305
2289 skb->destructor = tpacket_destruct_skb; 2306 skb->destructor = tpacket_destruct_skb;
2290 __packet_set_status(po, ph, TP_STATUS_SENDING); 2307 __packet_set_status(po, ph, TP_STATUS_SENDING);
2291 packet_inc_pending(&po->tx_ring); 2308 packet_inc_pending(&po->tx_ring);
@@ -2499,7 +2516,8 @@ static int packet_snd(struct socket *sock, struct msghdr *msg, size_t len)
2499 skb->dev = dev; 2516 skb->dev = dev;
2500 skb->priority = sk->sk_priority; 2517 skb->priority = sk->sk_priority;
2501 skb->mark = sk->sk_mark; 2518 skb->mark = sk->sk_mark;
2502 skb_set_queue_mapping(skb, packet_pick_tx_queue(dev)); 2519
2520 packet_pick_tx_queue(dev, skb);
2503 2521
2504 if (po->has_vnet_hdr) { 2522 if (po->has_vnet_hdr) {
2505 if (vnet_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) { 2523 if (vnet_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
@@ -3786,7 +3804,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
3786 */ 3804 */
3787 if (!tx_ring) 3805 if (!tx_ring)
3788 init_prb_bdqc(po, rb, pg_vec, req_u, tx_ring); 3806 init_prb_bdqc(po, rb, pg_vec, req_u, tx_ring);
3789 break; 3807 break;
3790 default: 3808 default:
3791 break; 3809 break;
3792 } 3810 }
diff --git a/net/sched/sch_pie.c b/net/sched/sch_pie.c
index a255d0200a59..fefeeb73f15f 100644
--- a/net/sched/sch_pie.c
+++ b/net/sched/sch_pie.c
@@ -15,6 +15,11 @@
15 * 15 *
16 * ECN support is added by Naeem Khademi <naeemk@ifi.uio.no> 16 * ECN support is added by Naeem Khademi <naeemk@ifi.uio.no>
17 * University of Oslo, Norway. 17 * University of Oslo, Norway.
18 *
19 * References:
20 * IETF draft submission: http://tools.ietf.org/html/draft-pan-aqm-pie-00
21 * IEEE Conference on High Performance Switching and Routing 2013 :
22 * "PIE: A * Lightweight Control Scheme to Address the Bufferbloat Problem"
18 */ 23 */
19 24
20#include <linux/module.h> 25#include <linux/module.h>
@@ -36,7 +41,7 @@ struct pie_params {
36 psched_time_t target; /* user specified target delay in pschedtime */ 41 psched_time_t target; /* user specified target delay in pschedtime */
37 u32 tupdate; /* timer frequency (in jiffies) */ 42 u32 tupdate; /* timer frequency (in jiffies) */
38 u32 limit; /* number of packets that can be enqueued */ 43 u32 limit; /* number of packets that can be enqueued */
39 u32 alpha; /* alpha and beta are between -4 and 4 */ 44 u32 alpha; /* alpha and beta are between 0 and 32 */
40 u32 beta; /* and are used for shift relative to 1 */ 45 u32 beta; /* and are used for shift relative to 1 */
41 bool ecn; /* true if ecn is enabled */ 46 bool ecn; /* true if ecn is enabled */
42 bool bytemode; /* to scale drop early prob based on pkt size */ 47 bool bytemode; /* to scale drop early prob based on pkt size */
@@ -326,10 +331,16 @@ static void calculate_probability(struct Qdisc *sch)
326 if (qdelay == 0 && qlen != 0) 331 if (qdelay == 0 && qlen != 0)
327 update_prob = false; 332 update_prob = false;
328 333
329 /* Add ranges for alpha and beta, more aggressive for high dropping 334 /* In the algorithm, alpha and beta are between 0 and 2 with typical
330 * mode and gentle steps for light dropping mode 335 * value for alpha as 0.125. In this implementation, we use values 0-32
331 * In light dropping mode, take gentle steps; in medium dropping mode, 336 * passed from user space to represent this. Also, alpha and beta have
332 * take medium steps; in high dropping mode, take big steps. 337 * unit of HZ and need to be scaled before they can used to update
338 * probability. alpha/beta are updated locally below by 1) scaling them
339 * appropriately 2) scaling down by 16 to come to 0-2 range.
340 * Please see paper for details.
341 *
342 * We scale alpha and beta differently depending on whether we are in
343 * light, medium or high dropping mode.
333 */ 344 */
334 if (q->vars.prob < MAX_PROB / 100) { 345 if (q->vars.prob < MAX_PROB / 100) {
335 alpha = 346 alpha =
diff --git a/net/sched/sch_tbf.c b/net/sched/sch_tbf.c
index 1cb413fead89..4f505a006896 100644
--- a/net/sched/sch_tbf.c
+++ b/net/sched/sch_tbf.c
@@ -334,18 +334,6 @@ static int tbf_change(struct Qdisc *sch, struct nlattr *opt)
334 qdisc_put_rtab(qdisc_get_rtab(&qopt->peakrate, 334 qdisc_put_rtab(qdisc_get_rtab(&qopt->peakrate,
335 tb[TCA_TBF_PTAB])); 335 tb[TCA_TBF_PTAB]));
336 336
337 if (q->qdisc != &noop_qdisc) {
338 err = fifo_set_limit(q->qdisc, qopt->limit);
339 if (err)
340 goto done;
341 } else if (qopt->limit > 0) {
342 child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit);
343 if (IS_ERR(child)) {
344 err = PTR_ERR(child);
345 goto done;
346 }
347 }
348
349 buffer = min_t(u64, PSCHED_TICKS2NS(qopt->buffer), ~0U); 337 buffer = min_t(u64, PSCHED_TICKS2NS(qopt->buffer), ~0U);
350 mtu = min_t(u64, PSCHED_TICKS2NS(qopt->mtu), ~0U); 338 mtu = min_t(u64, PSCHED_TICKS2NS(qopt->mtu), ~0U);
351 339
@@ -390,6 +378,18 @@ static int tbf_change(struct Qdisc *sch, struct nlattr *opt)
390 goto done; 378 goto done;
391 } 379 }
392 380
381 if (q->qdisc != &noop_qdisc) {
382 err = fifo_set_limit(q->qdisc, qopt->limit);
383 if (err)
384 goto done;
385 } else if (qopt->limit > 0) {
386 child = fifo_create_dflt(sch, &bfifo_qdisc_ops, qopt->limit);
387 if (IS_ERR(child)) {
388 err = PTR_ERR(child);
389 goto done;
390 }
391 }
392
393 sch_tree_lock(sch); 393 sch_tree_lock(sch);
394 if (child) { 394 if (child) {
395 qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen); 395 qdisc_tree_decrease_qlen(q->qdisc, q->qdisc->q.qlen);
diff --git a/net/sctp/associola.c b/net/sctp/associola.c
index 5ae609200674..ee13d28d39d1 100644
--- a/net/sctp/associola.c
+++ b/net/sctp/associola.c
@@ -1239,78 +1239,107 @@ void sctp_assoc_update(struct sctp_association *asoc,
1239} 1239}
1240 1240
1241/* Update the retran path for sending a retransmitted packet. 1241/* Update the retran path for sending a retransmitted packet.
1242 * Round-robin through the active transports, else round-robin 1242 * See also RFC4960, 6.4. Multi-Homed SCTP Endpoints:
1243 * through the inactive transports as this is the next best thing 1243 *
1244 * we can try. 1244 * When there is outbound data to send and the primary path
1245 * becomes inactive (e.g., due to failures), or where the
1246 * SCTP user explicitly requests to send data to an
1247 * inactive destination transport address, before reporting
1248 * an error to its ULP, the SCTP endpoint should try to send
1249 * the data to an alternate active destination transport
1250 * address if one exists.
1251 *
1252 * When retransmitting data that timed out, if the endpoint
1253 * is multihomed, it should consider each source-destination
1254 * address pair in its retransmission selection policy.
1255 * When retransmitting timed-out data, the endpoint should
1256 * attempt to pick the most divergent source-destination
1257 * pair from the original source-destination pair to which
1258 * the packet was transmitted.
1259 *
1260 * Note: Rules for picking the most divergent source-destination
1261 * pair are an implementation decision and are not specified
1262 * within this document.
1263 *
1264 * Our basic strategy is to round-robin transports in priorities
1265 * according to sctp_state_prio_map[] e.g., if no such
1266 * transport with state SCTP_ACTIVE exists, round-robin through
1267 * SCTP_UNKNOWN, etc. You get the picture.
1245 */ 1268 */
1246void sctp_assoc_update_retran_path(struct sctp_association *asoc) 1269static const u8 sctp_trans_state_to_prio_map[] = {
1270 [SCTP_ACTIVE] = 3, /* best case */
1271 [SCTP_UNKNOWN] = 2,
1272 [SCTP_PF] = 1,
1273 [SCTP_INACTIVE] = 0, /* worst case */
1274};
1275
1276static u8 sctp_trans_score(const struct sctp_transport *trans)
1247{ 1277{
1248 struct sctp_transport *t, *next; 1278 return sctp_trans_state_to_prio_map[trans->state];
1249 struct list_head *head = &asoc->peer.transport_addr_list; 1279}
1250 struct list_head *pos;
1251 1280
1252 if (asoc->peer.transport_count == 1) 1281static struct sctp_transport *sctp_trans_elect_best(struct sctp_transport *curr,
1253 return; 1282 struct sctp_transport *best)
1283{
1284 if (best == NULL)
1285 return curr;
1254 1286
1255 /* Find the next transport in a round-robin fashion. */ 1287 return sctp_trans_score(curr) > sctp_trans_score(best) ? curr : best;
1256 t = asoc->peer.retran_path; 1288}
1257 pos = &t->transports;
1258 next = NULL;
1259 1289
1260 while (1) { 1290void sctp_assoc_update_retran_path(struct sctp_association *asoc)
1261 /* Skip the head. */ 1291{
1262 if (pos->next == head) 1292 struct sctp_transport *trans = asoc->peer.retran_path;
1263 pos = head->next; 1293 struct sctp_transport *trans_next = NULL;
1264 else
1265 pos = pos->next;
1266 1294
1267 t = list_entry(pos, struct sctp_transport, transports); 1295 /* We're done as we only have the one and only path. */
1296 if (asoc->peer.transport_count == 1)
1297 return;
1298 /* If active_path and retran_path are the same and active,
1299 * then this is the only active path. Use it.
1300 */
1301 if (asoc->peer.active_path == asoc->peer.retran_path &&
1302 asoc->peer.active_path->state == SCTP_ACTIVE)
1303 return;
1268 1304
1269 /* We have exhausted the list, but didn't find any 1305 /* Iterate from retran_path's successor back to retran_path. */
1270 * other active transports. If so, use the next 1306 for (trans = list_next_entry(trans, transports); 1;
1271 * transport. 1307 trans = list_next_entry(trans, transports)) {
1272 */ 1308 /* Manually skip the head element. */
1273 if (t == asoc->peer.retran_path) { 1309 if (&trans->transports == &asoc->peer.transport_addr_list)
1274 t = next; 1310 continue;
1311 if (trans->state == SCTP_UNCONFIRMED)
1312 continue;
1313 trans_next = sctp_trans_elect_best(trans, trans_next);
1314 /* Active is good enough for immediate return. */
1315 if (trans_next->state == SCTP_ACTIVE)
1275 break; 1316 break;
1276 } 1317 /* We've reached the end, time to update path. */
1277 1318 if (trans == asoc->peer.retran_path)
1278 /* Try to find an active transport. */
1279
1280 if ((t->state == SCTP_ACTIVE) ||
1281 (t->state == SCTP_UNKNOWN)) {
1282 break; 1319 break;
1283 } else {
1284 /* Keep track of the next transport in case
1285 * we don't find any active transport.
1286 */
1287 if (t->state != SCTP_UNCONFIRMED && !next)
1288 next = t;
1289 }
1290 } 1320 }
1291 1321
1292 if (t) 1322 if (trans_next != NULL)
1293 asoc->peer.retran_path = t; 1323 asoc->peer.retran_path = trans_next;
1294 else
1295 t = asoc->peer.retran_path;
1296 1324
1297 pr_debug("%s: association:%p addr:%pISpc\n", __func__, asoc, 1325 pr_debug("%s: association:%p updated new path to addr:%pISpc\n",
1298 &t->ipaddr.sa); 1326 __func__, asoc, &asoc->peer.retran_path->ipaddr.sa);
1299} 1327}
1300 1328
1301/* Choose the transport for sending retransmit packet. */ 1329struct sctp_transport *
1302struct sctp_transport *sctp_assoc_choose_alter_transport( 1330sctp_assoc_choose_alter_transport(struct sctp_association *asoc,
1303 struct sctp_association *asoc, struct sctp_transport *last_sent_to) 1331 struct sctp_transport *last_sent_to)
1304{ 1332{
1305 /* If this is the first time packet is sent, use the active path, 1333 /* If this is the first time packet is sent, use the active path,
1306 * else use the retran path. If the last packet was sent over the 1334 * else use the retran path. If the last packet was sent over the
1307 * retran path, update the retran path and use it. 1335 * retran path, update the retran path and use it.
1308 */ 1336 */
1309 if (!last_sent_to) 1337 if (last_sent_to == NULL) {
1310 return asoc->peer.active_path; 1338 return asoc->peer.active_path;
1311 else { 1339 } else {
1312 if (last_sent_to == asoc->peer.retran_path) 1340 if (last_sent_to == asoc->peer.retran_path)
1313 sctp_assoc_update_retran_path(asoc); 1341 sctp_assoc_update_retran_path(asoc);
1342
1314 return asoc->peer.retran_path; 1343 return asoc->peer.retran_path;
1315 } 1344 }
1316} 1345}
@@ -1367,44 +1396,35 @@ static inline bool sctp_peer_needs_update(struct sctp_association *asoc)
1367 return false; 1396 return false;
1368} 1397}
1369 1398
1370/* Increase asoc's rwnd by len and send any window update SACK if needed. */ 1399/* Update asoc's rwnd for the approximated state in the buffer,
1371void sctp_assoc_rwnd_increase(struct sctp_association *asoc, unsigned int len) 1400 * and check whether SACK needs to be sent.
1401 */
1402void sctp_assoc_rwnd_update(struct sctp_association *asoc, bool update_peer)
1372{ 1403{
1404 int rx_count;
1373 struct sctp_chunk *sack; 1405 struct sctp_chunk *sack;
1374 struct timer_list *timer; 1406 struct timer_list *timer;
1375 1407
1376 if (asoc->rwnd_over) { 1408 if (asoc->ep->rcvbuf_policy)
1377 if (asoc->rwnd_over >= len) { 1409 rx_count = atomic_read(&asoc->rmem_alloc);
1378 asoc->rwnd_over -= len; 1410 else
1379 } else { 1411 rx_count = atomic_read(&asoc->base.sk->sk_rmem_alloc);
1380 asoc->rwnd += (len - asoc->rwnd_over);
1381 asoc->rwnd_over = 0;
1382 }
1383 } else {
1384 asoc->rwnd += len;
1385 }
1386 1412
1387 /* If we had window pressure, start recovering it 1413 if ((asoc->base.sk->sk_rcvbuf - rx_count) > 0)
1388 * once our rwnd had reached the accumulated pressure 1414 asoc->rwnd = (asoc->base.sk->sk_rcvbuf - rx_count) >> 1;
1389 * threshold. The idea is to recover slowly, but up 1415 else
1390 * to the initial advertised window. 1416 asoc->rwnd = 0;
1391 */
1392 if (asoc->rwnd_press && asoc->rwnd >= asoc->rwnd_press) {
1393 int change = min(asoc->pathmtu, asoc->rwnd_press);
1394 asoc->rwnd += change;
1395 asoc->rwnd_press -= change;
1396 }
1397 1417
1398 pr_debug("%s: asoc:%p rwnd increased by %d to (%u, %u) - %u\n", 1418 pr_debug("%s: asoc:%p rwnd=%u, rx_count=%d, sk_rcvbuf=%d\n",
1399 __func__, asoc, len, asoc->rwnd, asoc->rwnd_over, 1419 __func__, asoc, asoc->rwnd, rx_count,
1400 asoc->a_rwnd); 1420 asoc->base.sk->sk_rcvbuf);
1401 1421
1402 /* Send a window update SACK if the rwnd has increased by at least the 1422 /* Send a window update SACK if the rwnd has increased by at least the
1403 * minimum of the association's PMTU and half of the receive buffer. 1423 * minimum of the association's PMTU and half of the receive buffer.
1404 * The algorithm used is similar to the one described in 1424 * The algorithm used is similar to the one described in
1405 * Section 4.2.3.3 of RFC 1122. 1425 * Section 4.2.3.3 of RFC 1122.
1406 */ 1426 */
1407 if (sctp_peer_needs_update(asoc)) { 1427 if (update_peer && sctp_peer_needs_update(asoc)) {
1408 asoc->a_rwnd = asoc->rwnd; 1428 asoc->a_rwnd = asoc->rwnd;
1409 1429
1410 pr_debug("%s: sending window update SACK- asoc:%p rwnd:%u " 1430 pr_debug("%s: sending window update SACK- asoc:%p rwnd:%u "
@@ -1426,45 +1446,6 @@ void sctp_assoc_rwnd_increase(struct sctp_association *asoc, unsigned int len)
1426 } 1446 }
1427} 1447}
1428 1448
1429/* Decrease asoc's rwnd by len. */
1430void sctp_assoc_rwnd_decrease(struct sctp_association *asoc, unsigned int len)
1431{
1432 int rx_count;
1433 int over = 0;
1434
1435 if (unlikely(!asoc->rwnd || asoc->rwnd_over))
1436 pr_debug("%s: association:%p has asoc->rwnd:%u, "
1437 "asoc->rwnd_over:%u!\n", __func__, asoc,
1438 asoc->rwnd, asoc->rwnd_over);
1439
1440 if (asoc->ep->rcvbuf_policy)
1441 rx_count = atomic_read(&asoc->rmem_alloc);
1442 else
1443 rx_count = atomic_read(&asoc->base.sk->sk_rmem_alloc);
1444
1445 /* If we've reached or overflowed our receive buffer, announce
1446 * a 0 rwnd if rwnd would still be positive. Store the
1447 * the potential pressure overflow so that the window can be restored
1448 * back to original value.
1449 */
1450 if (rx_count >= asoc->base.sk->sk_rcvbuf)
1451 over = 1;
1452
1453 if (asoc->rwnd >= len) {
1454 asoc->rwnd -= len;
1455 if (over) {
1456 asoc->rwnd_press += asoc->rwnd;
1457 asoc->rwnd = 0;
1458 }
1459 } else {
1460 asoc->rwnd_over = len - asoc->rwnd;
1461 asoc->rwnd = 0;
1462 }
1463
1464 pr_debug("%s: asoc:%p rwnd decreased by %d to (%u, %u, %u)\n",
1465 __func__, asoc, len, asoc->rwnd, asoc->rwnd_over,
1466 asoc->rwnd_press);
1467}
1468 1449
1469/* Build the bind address list for the association based on info from the 1450/* Build the bind address list for the association based on info from the
1470 * local endpoint and the remote peer. 1451 * local endpoint and the remote peer.
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index 0f6259a6a932..2b1738ef9394 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -662,6 +662,8 @@ static struct sock *sctp_v6_create_accept_sk(struct sock *sk,
662 */ 662 */
663 sctp_v6_to_sk_daddr(&asoc->peer.primary_addr, newsk); 663 sctp_v6_to_sk_daddr(&asoc->peer.primary_addr, newsk);
664 664
665 newsk->sk_v6_rcv_saddr = sk->sk_v6_rcv_saddr;
666
665 sk_refcnt_debug_inc(newsk); 667 sk_refcnt_debug_inc(newsk);
666 668
667 if (newsk->sk_prot->init(newsk)) { 669 if (newsk->sk_prot->init(newsk)) {
diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
index bd859154000e..5d6883ff00c3 100644
--- a/net/sctp/sm_sideeffect.c
+++ b/net/sctp/sm_sideeffect.c
@@ -495,11 +495,12 @@ static void sctp_do_8_2_transport_strike(sctp_cmd_seq_t *commands,
495 } 495 }
496 496
497 /* If the transport error count is greater than the pf_retrans 497 /* If the transport error count is greater than the pf_retrans
498 * threshold, and less than pathmaxrtx, then mark this transport 498 * threshold, and less than pathmaxrtx, and if the current state
499 * as Partially Failed, ee SCTP Quick Failover Draft, secon 5.1, 499 * is not SCTP_UNCONFIRMED, then mark this transport as Partially
500 * point 1 500 * Failed, see SCTP Quick Failover Draft, section 5.1
501 */ 501 */
502 if ((transport->state != SCTP_PF) && 502 if ((transport->state != SCTP_PF) &&
503 (transport->state != SCTP_UNCONFIRMED) &&
503 (asoc->pf_retrans < transport->pathmaxrxt) && 504 (asoc->pf_retrans < transport->pathmaxrxt) &&
504 (transport->error_count > asoc->pf_retrans)) { 505 (transport->error_count > asoc->pf_retrans)) {
505 506
diff --git a/net/sctp/sm_statefuns.c b/net/sctp/sm_statefuns.c
index 483dcd71b3c5..ae65b6b5973a 100644
--- a/net/sctp/sm_statefuns.c
+++ b/net/sctp/sm_statefuns.c
@@ -758,6 +758,13 @@ sctp_disposition_t sctp_sf_do_5_1D_ce(struct net *net,
758 struct sctp_chunk auth; 758 struct sctp_chunk auth;
759 sctp_ierror_t ret; 759 sctp_ierror_t ret;
760 760
761 /* Make sure that we and the peer are AUTH capable */
762 if (!net->sctp.auth_enable || !new_asoc->peer.auth_capable) {
763 kfree_skb(chunk->auth_chunk);
764 sctp_association_free(new_asoc);
765 return sctp_sf_pdiscard(net, ep, asoc, type, arg, commands);
766 }
767
761 /* set-up our fake chunk so that we can process it */ 768 /* set-up our fake chunk so that we can process it */
762 auth.skb = chunk->auth_chunk; 769 auth.skb = chunk->auth_chunk;
763 auth.asoc = chunk->asoc; 770 auth.asoc = chunk->asoc;
@@ -6176,7 +6183,7 @@ static int sctp_eat_data(const struct sctp_association *asoc,
6176 * PMTU. In cases, such as loopback, this might be a rather 6183 * PMTU. In cases, such as loopback, this might be a rather
6177 * large spill over. 6184 * large spill over.
6178 */ 6185 */
6179 if ((!chunk->data_accepted) && (!asoc->rwnd || asoc->rwnd_over || 6186 if ((!chunk->data_accepted) && (!asoc->rwnd ||
6180 (datalen > asoc->rwnd + asoc->frag_point))) { 6187 (datalen > asoc->rwnd + asoc->frag_point))) {
6181 6188
6182 /* If this is the next TSN, consider reneging to make 6189 /* If this is the next TSN, consider reneging to make
diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index 9e91d6e5df63..981aaf8b6ace 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -64,6 +64,7 @@
64#include <linux/crypto.h> 64#include <linux/crypto.h>
65#include <linux/slab.h> 65#include <linux/slab.h>
66#include <linux/file.h> 66#include <linux/file.h>
67#include <linux/compat.h>
67 68
68#include <net/ip.h> 69#include <net/ip.h>
69#include <net/icmp.h> 70#include <net/icmp.h>
@@ -1368,11 +1369,19 @@ static int sctp_setsockopt_connectx(struct sock *sk,
1368/* 1369/*
1369 * New (hopefully final) interface for the API. 1370 * New (hopefully final) interface for the API.
1370 * We use the sctp_getaddrs_old structure so that use-space library 1371 * We use the sctp_getaddrs_old structure so that use-space library
1371 * can avoid any unnecessary allocations. The only defferent part 1372 * can avoid any unnecessary allocations. The only different part
1372 * is that we store the actual length of the address buffer into the 1373 * is that we store the actual length of the address buffer into the
1373 * addrs_num structure member. That way we can re-use the existing 1374 * addrs_num structure member. That way we can re-use the existing
1374 * code. 1375 * code.
1375 */ 1376 */
1377#ifdef CONFIG_COMPAT
1378struct compat_sctp_getaddrs_old {
1379 sctp_assoc_t assoc_id;
1380 s32 addr_num;
1381 compat_uptr_t addrs; /* struct sockaddr * */
1382};
1383#endif
1384
1376static int sctp_getsockopt_connectx3(struct sock *sk, int len, 1385static int sctp_getsockopt_connectx3(struct sock *sk, int len,
1377 char __user *optval, 1386 char __user *optval,
1378 int __user *optlen) 1387 int __user *optlen)
@@ -1381,16 +1390,30 @@ static int sctp_getsockopt_connectx3(struct sock *sk, int len,
1381 sctp_assoc_t assoc_id = 0; 1390 sctp_assoc_t assoc_id = 0;
1382 int err = 0; 1391 int err = 0;
1383 1392
1384 if (len < sizeof(param)) 1393#ifdef CONFIG_COMPAT
1385 return -EINVAL; 1394 if (is_compat_task()) {
1395 struct compat_sctp_getaddrs_old param32;
1386 1396
1387 if (copy_from_user(&param, optval, sizeof(param))) 1397 if (len < sizeof(param32))
1388 return -EFAULT; 1398 return -EINVAL;
1399 if (copy_from_user(&param32, optval, sizeof(param32)))
1400 return -EFAULT;
1389 1401
1390 err = __sctp_setsockopt_connectx(sk, 1402 param.assoc_id = param32.assoc_id;
1391 (struct sockaddr __user *)param.addrs, 1403 param.addr_num = param32.addr_num;
1392 param.addr_num, &assoc_id); 1404 param.addrs = compat_ptr(param32.addrs);
1405 } else
1406#endif
1407 {
1408 if (len < sizeof(param))
1409 return -EINVAL;
1410 if (copy_from_user(&param, optval, sizeof(param)))
1411 return -EFAULT;
1412 }
1393 1413
1414 err = __sctp_setsockopt_connectx(sk, (struct sockaddr __user *)
1415 param.addrs, param.addr_num,
1416 &assoc_id);
1394 if (err == 0 || err == -EINPROGRESS) { 1417 if (err == 0 || err == -EINPROGRESS) {
1395 if (copy_to_user(optval, &assoc_id, sizeof(assoc_id))) 1418 if (copy_to_user(optval, &assoc_id, sizeof(assoc_id)))
1396 return -EFAULT; 1419 return -EFAULT;
@@ -2092,12 +2115,6 @@ static int sctp_recvmsg(struct kiocb *iocb, struct sock *sk,
2092 sctp_skb_pull(skb, copied); 2115 sctp_skb_pull(skb, copied);
2093 skb_queue_head(&sk->sk_receive_queue, skb); 2116 skb_queue_head(&sk->sk_receive_queue, skb);
2094 2117
2095 /* When only partial message is copied to the user, increase
2096 * rwnd by that amount. If all the data in the skb is read,
2097 * rwnd is updated when the event is freed.
2098 */
2099 if (!sctp_ulpevent_is_notification(event))
2100 sctp_assoc_rwnd_increase(event->asoc, copied);
2101 goto out; 2118 goto out;
2102 } else if ((event->msg_flags & MSG_NOTIFICATION) || 2119 } else if ((event->msg_flags & MSG_NOTIFICATION) ||
2103 (event->msg_flags & MSG_EOR)) 2120 (event->msg_flags & MSG_EOR))
diff --git a/net/sctp/sysctl.c b/net/sctp/sysctl.c
index 7135e617ab0f..35c8923b5554 100644
--- a/net/sctp/sysctl.c
+++ b/net/sctp/sysctl.c
@@ -151,6 +151,7 @@ static struct ctl_table sctp_net_table[] = {
151 }, 151 },
152 { 152 {
153 .procname = "cookie_hmac_alg", 153 .procname = "cookie_hmac_alg",
154 .data = &init_net.sctp.sctp_hmac_alg,
154 .maxlen = 8, 155 .maxlen = 8,
155 .mode = 0644, 156 .mode = 0644,
156 .proc_handler = proc_sctp_do_hmac_alg, 157 .proc_handler = proc_sctp_do_hmac_alg,
@@ -401,15 +402,18 @@ static int proc_sctp_do_rto_max(struct ctl_table *ctl, int write,
401 402
402int sctp_sysctl_net_register(struct net *net) 403int sctp_sysctl_net_register(struct net *net)
403{ 404{
404 struct ctl_table *table; 405 struct ctl_table *table = sctp_net_table;
405 int i; 406
407 if (!net_eq(net, &init_net)) {
408 int i;
406 409
407 table = kmemdup(sctp_net_table, sizeof(sctp_net_table), GFP_KERNEL); 410 table = kmemdup(sctp_net_table, sizeof(sctp_net_table), GFP_KERNEL);
408 if (!table) 411 if (!table)
409 return -ENOMEM; 412 return -ENOMEM;
410 413
411 for (i = 0; table[i].data; i++) 414 for (i = 0; table[i].data; i++)
412 table[i].data += (char *)(&net->sctp) - (char *)&init_net.sctp; 415 table[i].data += (char *)(&net->sctp) - (char *)&init_net.sctp;
416 }
413 417
414 net->sctp.sysctl_header = register_net_sysctl(net, "net/sctp", table); 418 net->sctp.sysctl_header = register_net_sysctl(net, "net/sctp", table);
415 return 0; 419 return 0;
diff --git a/net/sctp/ulpevent.c b/net/sctp/ulpevent.c
index 85c64658bd0b..8d198ae03606 100644
--- a/net/sctp/ulpevent.c
+++ b/net/sctp/ulpevent.c
@@ -989,7 +989,7 @@ static void sctp_ulpevent_receive_data(struct sctp_ulpevent *event,
989 skb = sctp_event2skb(event); 989 skb = sctp_event2skb(event);
990 /* Set the owner and charge rwnd for bytes received. */ 990 /* Set the owner and charge rwnd for bytes received. */
991 sctp_ulpevent_set_owner(event, asoc); 991 sctp_ulpevent_set_owner(event, asoc);
992 sctp_assoc_rwnd_decrease(asoc, skb_headlen(skb)); 992 sctp_assoc_rwnd_update(asoc, false);
993 993
994 if (!skb->data_len) 994 if (!skb->data_len)
995 return; 995 return;
@@ -1011,6 +1011,7 @@ static void sctp_ulpevent_release_data(struct sctp_ulpevent *event)
1011{ 1011{
1012 struct sk_buff *skb, *frag; 1012 struct sk_buff *skb, *frag;
1013 unsigned int len; 1013 unsigned int len;
1014 struct sctp_association *asoc;
1014 1015
1015 /* Current stack structures assume that the rcv buffer is 1016 /* Current stack structures assume that the rcv buffer is
1016 * per socket. For UDP style sockets this is not true as 1017 * per socket. For UDP style sockets this is not true as
@@ -1035,8 +1036,11 @@ static void sctp_ulpevent_release_data(struct sctp_ulpevent *event)
1035 } 1036 }
1036 1037
1037done: 1038done:
1038 sctp_assoc_rwnd_increase(event->asoc, len); 1039 asoc = event->asoc;
1040 sctp_association_hold(asoc);
1039 sctp_ulpevent_release_owner(event); 1041 sctp_ulpevent_release_owner(event);
1042 sctp_assoc_rwnd_update(asoc, true);
1043 sctp_association_put(asoc);
1040} 1044}
1041 1045
1042static void sctp_ulpevent_release_frag_data(struct sctp_ulpevent *event) 1046static void sctp_ulpevent_release_frag_data(struct sctp_ulpevent *event)
diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index 6c0513a7f992..36e431ee1c90 100644
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -108,6 +108,7 @@ struct gss_auth {
108static DEFINE_SPINLOCK(pipe_version_lock); 108static DEFINE_SPINLOCK(pipe_version_lock);
109static struct rpc_wait_queue pipe_version_rpc_waitqueue; 109static struct rpc_wait_queue pipe_version_rpc_waitqueue;
110static DECLARE_WAIT_QUEUE_HEAD(pipe_version_waitqueue); 110static DECLARE_WAIT_QUEUE_HEAD(pipe_version_waitqueue);
111static void gss_put_auth(struct gss_auth *gss_auth);
111 112
112static void gss_free_ctx(struct gss_cl_ctx *); 113static void gss_free_ctx(struct gss_cl_ctx *);
113static const struct rpc_pipe_ops gss_upcall_ops_v0; 114static const struct rpc_pipe_ops gss_upcall_ops_v0;
@@ -320,6 +321,7 @@ gss_release_msg(struct gss_upcall_msg *gss_msg)
320 if (gss_msg->ctx != NULL) 321 if (gss_msg->ctx != NULL)
321 gss_put_ctx(gss_msg->ctx); 322 gss_put_ctx(gss_msg->ctx);
322 rpc_destroy_wait_queue(&gss_msg->rpc_waitqueue); 323 rpc_destroy_wait_queue(&gss_msg->rpc_waitqueue);
324 gss_put_auth(gss_msg->auth);
323 kfree(gss_msg); 325 kfree(gss_msg);
324} 326}
325 327
@@ -498,9 +500,12 @@ gss_alloc_msg(struct gss_auth *gss_auth,
498 default: 500 default:
499 err = gss_encode_v1_msg(gss_msg, service_name, gss_auth->target_name); 501 err = gss_encode_v1_msg(gss_msg, service_name, gss_auth->target_name);
500 if (err) 502 if (err)
501 goto err_free_msg; 503 goto err_put_pipe_version;
502 }; 504 };
505 kref_get(&gss_auth->kref);
503 return gss_msg; 506 return gss_msg;
507err_put_pipe_version:
508 put_pipe_version(gss_auth->net);
504err_free_msg: 509err_free_msg:
505 kfree(gss_msg); 510 kfree(gss_msg);
506err: 511err:
@@ -991,6 +996,8 @@ gss_create_new(struct rpc_auth_create_args *args, struct rpc_clnt *clnt)
991 gss_auth->service = gss_pseudoflavor_to_service(gss_auth->mech, flavor); 996 gss_auth->service = gss_pseudoflavor_to_service(gss_auth->mech, flavor);
992 if (gss_auth->service == 0) 997 if (gss_auth->service == 0)
993 goto err_put_mech; 998 goto err_put_mech;
999 if (!gssd_running(gss_auth->net))
1000 goto err_put_mech;
994 auth = &gss_auth->rpc_auth; 1001 auth = &gss_auth->rpc_auth;
995 auth->au_cslack = GSS_CRED_SLACK >> 2; 1002 auth->au_cslack = GSS_CRED_SLACK >> 2;
996 auth->au_rslack = GSS_VERF_SLACK >> 2; 1003 auth->au_rslack = GSS_VERF_SLACK >> 2;
@@ -1062,6 +1069,12 @@ gss_free_callback(struct kref *kref)
1062} 1069}
1063 1070
1064static void 1071static void
1072gss_put_auth(struct gss_auth *gss_auth)
1073{
1074 kref_put(&gss_auth->kref, gss_free_callback);
1075}
1076
1077static void
1065gss_destroy(struct rpc_auth *auth) 1078gss_destroy(struct rpc_auth *auth)
1066{ 1079{
1067 struct gss_auth *gss_auth = container_of(auth, 1080 struct gss_auth *gss_auth = container_of(auth,
@@ -1082,7 +1095,7 @@ gss_destroy(struct rpc_auth *auth)
1082 gss_auth->gss_pipe[1] = NULL; 1095 gss_auth->gss_pipe[1] = NULL;
1083 rpcauth_destroy_credcache(auth); 1096 rpcauth_destroy_credcache(auth);
1084 1097
1085 kref_put(&gss_auth->kref, gss_free_callback); 1098 gss_put_auth(gss_auth);
1086} 1099}
1087 1100
1088/* 1101/*
@@ -1253,7 +1266,7 @@ gss_destroy_nullcred(struct rpc_cred *cred)
1253 call_rcu(&cred->cr_rcu, gss_free_cred_callback); 1266 call_rcu(&cred->cr_rcu, gss_free_cred_callback);
1254 if (ctx) 1267 if (ctx)
1255 gss_put_ctx(ctx); 1268 gss_put_ctx(ctx);
1256 kref_put(&gss_auth->kref, gss_free_callback); 1269 gss_put_auth(gss_auth);
1257} 1270}
1258 1271
1259static void 1272static void
diff --git a/net/sunrpc/backchannel_rqst.c b/net/sunrpc/backchannel_rqst.c
index 890a29912d5a..e860d4f7ed2a 100644
--- a/net/sunrpc/backchannel_rqst.c
+++ b/net/sunrpc/backchannel_rqst.c
@@ -64,7 +64,6 @@ static void xprt_free_allocation(struct rpc_rqst *req)
64 free_page((unsigned long)xbufp->head[0].iov_base); 64 free_page((unsigned long)xbufp->head[0].iov_base);
65 xbufp = &req->rq_snd_buf; 65 xbufp = &req->rq_snd_buf;
66 free_page((unsigned long)xbufp->head[0].iov_base); 66 free_page((unsigned long)xbufp->head[0].iov_base);
67 list_del(&req->rq_bc_pa_list);
68 kfree(req); 67 kfree(req);
69} 68}
70 69
@@ -168,8 +167,10 @@ out_free:
168 /* 167 /*
169 * Memory allocation failed, free the temporary list 168 * Memory allocation failed, free the temporary list
170 */ 169 */
171 list_for_each_entry_safe(req, tmp, &tmp_list, rq_bc_pa_list) 170 list_for_each_entry_safe(req, tmp, &tmp_list, rq_bc_pa_list) {
171 list_del(&req->rq_bc_pa_list);
172 xprt_free_allocation(req); 172 xprt_free_allocation(req);
173 }
173 174
174 dprintk("RPC: setup backchannel transport failed\n"); 175 dprintk("RPC: setup backchannel transport failed\n");
175 return -ENOMEM; 176 return -ENOMEM;
@@ -198,6 +199,7 @@ void xprt_destroy_backchannel(struct rpc_xprt *xprt, unsigned int max_reqs)
198 xprt_dec_alloc_count(xprt, max_reqs); 199 xprt_dec_alloc_count(xprt, max_reqs);
199 list_for_each_entry_safe(req, tmp, &xprt->bc_pa_list, rq_bc_pa_list) { 200 list_for_each_entry_safe(req, tmp, &xprt->bc_pa_list, rq_bc_pa_list) {
200 dprintk("RPC: req=%p\n", req); 201 dprintk("RPC: req=%p\n", req);
202 list_del(&req->rq_bc_pa_list);
201 xprt_free_allocation(req); 203 xprt_free_allocation(req);
202 if (--max_reqs == 0) 204 if (--max_reqs == 0)
203 break; 205 break;
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 80a6640f329b..06c6ff0cb911 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -571,7 +571,7 @@ static void svc_check_conn_limits(struct svc_serv *serv)
571 } 571 }
572} 572}
573 573
574int svc_alloc_arg(struct svc_rqst *rqstp) 574static int svc_alloc_arg(struct svc_rqst *rqstp)
575{ 575{
576 struct svc_serv *serv = rqstp->rq_server; 576 struct svc_serv *serv = rqstp->rq_server;
577 struct xdr_buf *arg; 577 struct xdr_buf *arg;
@@ -612,7 +612,7 @@ int svc_alloc_arg(struct svc_rqst *rqstp)
612 return 0; 612 return 0;
613} 613}
614 614
615struct svc_xprt *svc_get_next_xprt(struct svc_rqst *rqstp, long timeout) 615static struct svc_xprt *svc_get_next_xprt(struct svc_rqst *rqstp, long timeout)
616{ 616{
617 struct svc_xprt *xprt; 617 struct svc_xprt *xprt;
618 struct svc_pool *pool = rqstp->rq_pool; 618 struct svc_pool *pool = rqstp->rq_pool;
@@ -691,7 +691,7 @@ struct svc_xprt *svc_get_next_xprt(struct svc_rqst *rqstp, long timeout)
691 return xprt; 691 return xprt;
692} 692}
693 693
694void svc_add_new_temp_xprt(struct svc_serv *serv, struct svc_xprt *newxpt) 694static void svc_add_new_temp_xprt(struct svc_serv *serv, struct svc_xprt *newxpt)
695{ 695{
696 spin_lock_bh(&serv->sv_lock); 696 spin_lock_bh(&serv->sv_lock);
697 set_bit(XPT_TEMP, &newxpt->xpt_flags); 697 set_bit(XPT_TEMP, &newxpt->xpt_flags);
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 817a1e523969..0addefca8e77 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -510,6 +510,7 @@ static int xs_nospace(struct rpc_task *task)
510 struct rpc_rqst *req = task->tk_rqstp; 510 struct rpc_rqst *req = task->tk_rqstp;
511 struct rpc_xprt *xprt = req->rq_xprt; 511 struct rpc_xprt *xprt = req->rq_xprt;
512 struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt); 512 struct sock_xprt *transport = container_of(xprt, struct sock_xprt, xprt);
513 struct sock *sk = transport->inet;
513 int ret = -EAGAIN; 514 int ret = -EAGAIN;
514 515
515 dprintk("RPC: %5u xmit incomplete (%u left of %u)\n", 516 dprintk("RPC: %5u xmit incomplete (%u left of %u)\n",
@@ -527,7 +528,7 @@ static int xs_nospace(struct rpc_task *task)
527 * window size 528 * window size
528 */ 529 */
529 set_bit(SOCK_NOSPACE, &transport->sock->flags); 530 set_bit(SOCK_NOSPACE, &transport->sock->flags);
530 transport->inet->sk_write_pending++; 531 sk->sk_write_pending++;
531 /* ...and wait for more buffer space */ 532 /* ...and wait for more buffer space */
532 xprt_wait_for_buffer_space(task, xs_nospace_callback); 533 xprt_wait_for_buffer_space(task, xs_nospace_callback);
533 } 534 }
@@ -537,6 +538,9 @@ static int xs_nospace(struct rpc_task *task)
537 } 538 }
538 539
539 spin_unlock_bh(&xprt->transport_lock); 540 spin_unlock_bh(&xprt->transport_lock);
541
542 /* Race breaker in case memory is freed before above code is called */
543 sk->sk_write_space(sk);
540 return ret; 544 return ret;
541} 545}
542 546
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index a38c89969c68..574b86193b15 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -610,8 +610,13 @@ static struct notifier_block notifier = {
610 610
611int tipc_bearer_setup(void) 611int tipc_bearer_setup(void)
612{ 612{
613 int err;
614
615 err = register_netdevice_notifier(&notifier);
616 if (err)
617 return err;
613 dev_add_pack(&tipc_packet_type); 618 dev_add_pack(&tipc_packet_type);
614 return register_netdevice_notifier(&notifier); 619 return 0;
615} 620}
616 621
617void tipc_bearer_cleanup(void) 622void tipc_bearer_cleanup(void)
diff --git a/net/tipc/config.c b/net/tipc/config.c
index c301a9a592d8..e74eef2e7490 100644
--- a/net/tipc/config.c
+++ b/net/tipc/config.c
@@ -181,7 +181,7 @@ static struct sk_buff *cfg_set_own_addr(void)
181 if (tipc_own_addr) 181 if (tipc_own_addr)
182 return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED 182 return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED
183 " (cannot change node address once assigned)"); 183 " (cannot change node address once assigned)");
184 tipc_core_start_net(addr); 184 tipc_net_start(addr);
185 return tipc_cfg_reply_none(); 185 return tipc_cfg_reply_none();
186} 186}
187 187
diff --git a/net/tipc/core.c b/net/tipc/core.c
index f9e88d8b04ca..80c20647b3d2 100644
--- a/net/tipc/core.c
+++ b/net/tipc/core.c
@@ -77,37 +77,13 @@ struct sk_buff *tipc_buf_acquire(u32 size)
77} 77}
78 78
79/** 79/**
80 * tipc_core_stop_net - shut down TIPC networking sub-systems
81 */
82static void tipc_core_stop_net(void)
83{
84 tipc_net_stop();
85 tipc_bearer_cleanup();
86}
87
88/**
89 * start_net - start TIPC networking sub-systems
90 */
91int tipc_core_start_net(unsigned long addr)
92{
93 int res;
94
95 tipc_net_start(addr);
96 res = tipc_bearer_setup();
97 if (res < 0)
98 goto err;
99 return res;
100
101err:
102 tipc_core_stop_net();
103 return res;
104}
105
106/**
107 * tipc_core_stop - switch TIPC from SINGLE NODE to NOT RUNNING mode 80 * tipc_core_stop - switch TIPC from SINGLE NODE to NOT RUNNING mode
108 */ 81 */
109static void tipc_core_stop(void) 82static void tipc_core_stop(void)
110{ 83{
84 tipc_handler_stop();
85 tipc_net_stop();
86 tipc_bearer_cleanup();
111 tipc_netlink_stop(); 87 tipc_netlink_stop();
112 tipc_cfg_stop(); 88 tipc_cfg_stop();
113 tipc_subscr_stop(); 89 tipc_subscr_stop();
@@ -122,30 +98,65 @@ static void tipc_core_stop(void)
122 */ 98 */
123static int tipc_core_start(void) 99static int tipc_core_start(void)
124{ 100{
125 int res; 101 int err;
126 102
127 get_random_bytes(&tipc_random, sizeof(tipc_random)); 103 get_random_bytes(&tipc_random, sizeof(tipc_random));
128 104
129 res = tipc_handler_start(); 105 err = tipc_handler_start();
130 if (!res) 106 if (err)
131 res = tipc_ref_table_init(tipc_max_ports, tipc_random); 107 goto out_handler;
132 if (!res) 108
133 res = tipc_nametbl_init(); 109 err = tipc_ref_table_init(tipc_max_ports, tipc_random);
134 if (!res) 110 if (err)
135 res = tipc_netlink_start(); 111 goto out_reftbl;
136 if (!res) 112
137 res = tipc_socket_init(); 113 err = tipc_nametbl_init();
138 if (!res) 114 if (err)
139 res = tipc_register_sysctl(); 115 goto out_nametbl;
140 if (!res) 116
141 res = tipc_subscr_start(); 117 err = tipc_netlink_start();
142 if (!res) 118 if (err)
143 res = tipc_cfg_init(); 119 goto out_netlink;
144 if (res) { 120
145 tipc_handler_stop(); 121 err = tipc_socket_init();
146 tipc_core_stop(); 122 if (err)
147 } 123 goto out_socket;
148 return res; 124
125 err = tipc_register_sysctl();
126 if (err)
127 goto out_sysctl;
128
129 err = tipc_subscr_start();
130 if (err)
131 goto out_subscr;
132
133 err = tipc_cfg_init();
134 if (err)
135 goto out_cfg;
136
137 err = tipc_bearer_setup();
138 if (err)
139 goto out_bearer;
140
141 return 0;
142out_bearer:
143 tipc_cfg_stop();
144out_cfg:
145 tipc_subscr_stop();
146out_subscr:
147 tipc_unregister_sysctl();
148out_sysctl:
149 tipc_socket_stop();
150out_socket:
151 tipc_netlink_stop();
152out_netlink:
153 tipc_nametbl_stop();
154out_nametbl:
155 tipc_ref_table_stop();
156out_reftbl:
157 tipc_handler_stop();
158out_handler:
159 return err;
149} 160}
150 161
151static int __init tipc_init(void) 162static int __init tipc_init(void)
@@ -174,8 +185,6 @@ static int __init tipc_init(void)
174 185
175static void __exit tipc_exit(void) 186static void __exit tipc_exit(void)
176{ 187{
177 tipc_handler_stop();
178 tipc_core_stop_net();
179 tipc_core_stop(); 188 tipc_core_stop();
180 pr_info("Deactivated\n"); 189 pr_info("Deactivated\n");
181} 190}
diff --git a/net/tipc/core.h b/net/tipc/core.h
index 1ff477b0450d..4dfe137587bb 100644
--- a/net/tipc/core.h
+++ b/net/tipc/core.h
@@ -90,7 +90,6 @@ extern int tipc_random __read_mostly;
90/* 90/*
91 * Routines available to privileged subsystems 91 * Routines available to privileged subsystems
92 */ 92 */
93int tipc_core_start_net(unsigned long);
94int tipc_handler_start(void); 93int tipc_handler_start(void);
95void tipc_handler_stop(void); 94void tipc_handler_stop(void);
96int tipc_netlink_start(void); 95int tipc_netlink_start(void);
@@ -192,6 +191,7 @@ static inline void k_term_timer(struct timer_list *timer)
192 191
193struct tipc_skb_cb { 192struct tipc_skb_cb {
194 void *handle; 193 void *handle;
194 bool deferred;
195}; 195};
196 196
197#define TIPC_SKB_CB(__skb) ((struct tipc_skb_cb *)&((__skb)->cb[0])) 197#define TIPC_SKB_CB(__skb) ((struct tipc_skb_cb *)&((__skb)->cb[0]))
diff --git a/net/tipc/link.c b/net/tipc/link.c
index d4b5de41b682..da6018beb6eb 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -1391,6 +1391,12 @@ static int link_recv_buf_validate(struct sk_buff *buf)
1391 u32 hdr_size; 1391 u32 hdr_size;
1392 u32 min_hdr_size; 1392 u32 min_hdr_size;
1393 1393
1394 /* If this packet comes from the defer queue, the skb has already
1395 * been validated
1396 */
1397 if (unlikely(TIPC_SKB_CB(buf)->deferred))
1398 return 1;
1399
1394 if (unlikely(buf->len < MIN_H_SIZE)) 1400 if (unlikely(buf->len < MIN_H_SIZE))
1395 return 0; 1401 return 0;
1396 1402
@@ -1703,6 +1709,7 @@ static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
1703 &l_ptr->newest_deferred_in, buf)) { 1709 &l_ptr->newest_deferred_in, buf)) {
1704 l_ptr->deferred_inqueue_sz++; 1710 l_ptr->deferred_inqueue_sz++;
1705 l_ptr->stats.deferred_recv++; 1711 l_ptr->stats.deferred_recv++;
1712 TIPC_SKB_CB(buf)->deferred = true;
1706 if ((l_ptr->deferred_inqueue_sz % 16) == 1) 1713 if ((l_ptr->deferred_inqueue_sz % 16) == 1)
1707 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0); 1714 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1708 } else 1715 } else
diff --git a/net/tipc/name_table.c b/net/tipc/name_table.c
index 92a1533af4e0..48302be175ce 100644
--- a/net/tipc/name_table.c
+++ b/net/tipc/name_table.c
@@ -945,9 +945,6 @@ void tipc_nametbl_stop(void)
945{ 945{
946 u32 i; 946 u32 i;
947 947
948 if (!table.types)
949 return;
950
951 /* Verify name table is empty, then release it */ 948 /* Verify name table is empty, then release it */
952 write_lock_bh(&tipc_nametbl_lock); 949 write_lock_bh(&tipc_nametbl_lock);
953 for (i = 0; i < TIPC_NAMETBL_SIZE; i++) { 950 for (i = 0; i < TIPC_NAMETBL_SIZE; i++) {
diff --git a/net/tipc/netlink.c b/net/tipc/netlink.c
index 9f72a6376362..3aaf73de9e2d 100644
--- a/net/tipc/netlink.c
+++ b/net/tipc/netlink.c
@@ -83,8 +83,6 @@ static struct genl_ops tipc_genl_ops[] = {
83 }, 83 },
84}; 84};
85 85
86static int tipc_genl_family_registered;
87
88int tipc_netlink_start(void) 86int tipc_netlink_start(void)
89{ 87{
90 int res; 88 int res;
@@ -94,16 +92,10 @@ int tipc_netlink_start(void)
94 pr_err("Failed to register netlink interface\n"); 92 pr_err("Failed to register netlink interface\n");
95 return res; 93 return res;
96 } 94 }
97
98 tipc_genl_family_registered = 1;
99 return 0; 95 return 0;
100} 96}
101 97
102void tipc_netlink_stop(void) 98void tipc_netlink_stop(void)
103{ 99{
104 if (!tipc_genl_family_registered)
105 return;
106
107 genl_unregister_family(&tipc_genl_family); 100 genl_unregister_family(&tipc_genl_family);
108 tipc_genl_family_registered = 0;
109} 101}
diff --git a/net/tipc/ref.c b/net/tipc/ref.c
index 2a2a938dc22c..de3d593e2fee 100644
--- a/net/tipc/ref.c
+++ b/net/tipc/ref.c
@@ -126,9 +126,6 @@ int tipc_ref_table_init(u32 requested_size, u32 start)
126 */ 126 */
127void tipc_ref_table_stop(void) 127void tipc_ref_table_stop(void)
128{ 128{
129 if (!tipc_ref_table.entries)
130 return;
131
132 vfree(tipc_ref_table.entries); 129 vfree(tipc_ref_table.entries);
133 tipc_ref_table.entries = NULL; 130 tipc_ref_table.entries = NULL;
134} 131}
diff --git a/net/tipc/server.c b/net/tipc/server.c
index b635ca347a87..373979789a73 100644
--- a/net/tipc/server.c
+++ b/net/tipc/server.c
@@ -573,7 +573,6 @@ int tipc_server_start(struct tipc_server *s)
573 kmem_cache_destroy(s->rcvbuf_cache); 573 kmem_cache_destroy(s->rcvbuf_cache);
574 return ret; 574 return ret;
575 } 575 }
576 s->enabled = 1;
577 return ret; 576 return ret;
578} 577}
579 578
@@ -583,10 +582,6 @@ void tipc_server_stop(struct tipc_server *s)
583 int total = 0; 582 int total = 0;
584 int id; 583 int id;
585 584
586 if (!s->enabled)
587 return;
588
589 s->enabled = 0;
590 spin_lock_bh(&s->idr_lock); 585 spin_lock_bh(&s->idr_lock);
591 for (id = 0; total < s->idr_in_use; id++) { 586 for (id = 0; total < s->idr_in_use; id++) {
592 con = idr_find(&s->conn_idr, id); 587 con = idr_find(&s->conn_idr, id);
diff --git a/net/tipc/server.h b/net/tipc/server.h
index 98b23f20bc0f..be817b0b547e 100644
--- a/net/tipc/server.h
+++ b/net/tipc/server.h
@@ -56,7 +56,6 @@
56 * @name: server name 56 * @name: server name
57 * @imp: message importance 57 * @imp: message importance
58 * @type: socket type 58 * @type: socket type
59 * @enabled: identify whether server is launched or not
60 */ 59 */
61struct tipc_server { 60struct tipc_server {
62 struct idr conn_idr; 61 struct idr conn_idr;
@@ -74,7 +73,6 @@ struct tipc_server {
74 const char name[TIPC_SERVER_NAME_LEN]; 73 const char name[TIPC_SERVER_NAME_LEN];
75 int imp; 74 int imp;
76 int type; 75 int type;
77 int enabled;
78}; 76};
79 77
80int tipc_conn_sendmsg(struct tipc_server *s, int conid, 78int tipc_conn_sendmsg(struct tipc_server *s, int conid,
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index aab4948f0aff..a4cf274455aa 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -70,8 +70,6 @@ static const struct proto_ops msg_ops;
70static struct proto tipc_proto; 70static struct proto tipc_proto;
71static struct proto tipc_proto_kern; 71static struct proto tipc_proto_kern;
72 72
73static int sockets_enabled;
74
75/* 73/*
76 * Revised TIPC socket locking policy: 74 * Revised TIPC socket locking policy:
77 * 75 *
@@ -2027,8 +2025,6 @@ int tipc_socket_init(void)
2027 proto_unregister(&tipc_proto); 2025 proto_unregister(&tipc_proto);
2028 goto out; 2026 goto out;
2029 } 2027 }
2030
2031 sockets_enabled = 1;
2032 out: 2028 out:
2033 return res; 2029 return res;
2034} 2030}
@@ -2038,10 +2034,6 @@ int tipc_socket_init(void)
2038 */ 2034 */
2039void tipc_socket_stop(void) 2035void tipc_socket_stop(void)
2040{ 2036{
2041 if (!sockets_enabled)
2042 return;
2043
2044 sockets_enabled = 0;
2045 sock_unregister(tipc_family_ops.family); 2037 sock_unregister(tipc_family_ops.family);
2046 proto_unregister(&tipc_proto); 2038 proto_unregister(&tipc_proto);
2047} 2039}
diff --git a/net/wireless/core.c b/net/wireless/core.c
index d89dee2259b5..010892b81a06 100644
--- a/net/wireless/core.c
+++ b/net/wireless/core.c
@@ -203,8 +203,11 @@ void cfg80211_stop_p2p_device(struct cfg80211_registered_device *rdev,
203 203
204 rdev->opencount--; 204 rdev->opencount--;
205 205
206 WARN_ON(rdev->scan_req && rdev->scan_req->wdev == wdev && 206 if (rdev->scan_req && rdev->scan_req->wdev == wdev) {
207 !rdev->scan_req->notified); 207 if (WARN_ON(!rdev->scan_req->notified))
208 rdev->scan_req->aborted = true;
209 ___cfg80211_scan_done(rdev, false);
210 }
208} 211}
209 212
210static int cfg80211_rfkill_set_block(void *data, bool blocked) 213static int cfg80211_rfkill_set_block(void *data, bool blocked)
@@ -440,9 +443,6 @@ int wiphy_register(struct wiphy *wiphy)
440 int i; 443 int i;
441 u16 ifmodes = wiphy->interface_modes; 444 u16 ifmodes = wiphy->interface_modes;
442 445
443 /* support for 5/10 MHz is broken due to nl80211 API mess - disable */
444 wiphy->flags &= ~WIPHY_FLAG_SUPPORTS_5_10_MHZ;
445
446 /* 446 /*
447 * There are major locking problems in nl80211/mac80211 for CSA, 447 * There are major locking problems in nl80211/mac80211 for CSA,
448 * disable for all drivers until this has been reworked. 448 * disable for all drivers until this has been reworked.
@@ -859,8 +859,11 @@ static int cfg80211_netdev_notifier_call(struct notifier_block *nb,
859 break; 859 break;
860 case NETDEV_DOWN: 860 case NETDEV_DOWN:
861 cfg80211_update_iface_num(rdev, wdev->iftype, -1); 861 cfg80211_update_iface_num(rdev, wdev->iftype, -1);
862 WARN_ON(rdev->scan_req && rdev->scan_req->wdev == wdev && 862 if (rdev->scan_req && rdev->scan_req->wdev == wdev) {
863 !rdev->scan_req->notified); 863 if (WARN_ON(!rdev->scan_req->notified))
864 rdev->scan_req->aborted = true;
865 ___cfg80211_scan_done(rdev, false);
866 }
864 867
865 if (WARN_ON(rdev->sched_scan_req && 868 if (WARN_ON(rdev->sched_scan_req &&
866 rdev->sched_scan_req->dev == wdev->netdev)) { 869 rdev->sched_scan_req->dev == wdev->netdev)) {
diff --git a/net/wireless/core.h b/net/wireless/core.h
index 37ec16d7bb1a..f1d193b557b6 100644
--- a/net/wireless/core.h
+++ b/net/wireless/core.h
@@ -62,6 +62,7 @@ struct cfg80211_registered_device {
62 struct rb_root bss_tree; 62 struct rb_root bss_tree;
63 u32 bss_generation; 63 u32 bss_generation;
64 struct cfg80211_scan_request *scan_req; /* protected by RTNL */ 64 struct cfg80211_scan_request *scan_req; /* protected by RTNL */
65 struct sk_buff *scan_msg;
65 struct cfg80211_sched_scan_request *sched_scan_req; 66 struct cfg80211_sched_scan_request *sched_scan_req;
66 unsigned long suspend_at; 67 unsigned long suspend_at;
67 struct work_struct scan_done_wk; 68 struct work_struct scan_done_wk;
@@ -361,7 +362,8 @@ int cfg80211_validate_key_settings(struct cfg80211_registered_device *rdev,
361 struct key_params *params, int key_idx, 362 struct key_params *params, int key_idx,
362 bool pairwise, const u8 *mac_addr); 363 bool pairwise, const u8 *mac_addr);
363void __cfg80211_scan_done(struct work_struct *wk); 364void __cfg80211_scan_done(struct work_struct *wk);
364void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev); 365void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev,
366 bool send_message);
365void __cfg80211_sched_scan_results(struct work_struct *wk); 367void __cfg80211_sched_scan_results(struct work_struct *wk);
366int __cfg80211_stop_sched_scan(struct cfg80211_registered_device *rdev, 368int __cfg80211_stop_sched_scan(struct cfg80211_registered_device *rdev,
367 bool driver_initiated); 369 bool driver_initiated);
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 7a742594916e..4fe2e6e2bc76 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -1719,9 +1719,10 @@ static int nl80211_dump_wiphy(struct sk_buff *skb, struct netlink_callback *cb)
1719 * We can then retry with the larger buffer. 1719 * We can then retry with the larger buffer.
1720 */ 1720 */
1721 if ((ret == -ENOBUFS || ret == -EMSGSIZE) && 1721 if ((ret == -ENOBUFS || ret == -EMSGSIZE) &&
1722 !skb->len && 1722 !skb->len && !state->split &&
1723 cb->min_dump_alloc < 4096) { 1723 cb->min_dump_alloc < 4096) {
1724 cb->min_dump_alloc = 4096; 1724 cb->min_dump_alloc = 4096;
1725 state->split_start = 0;
1725 rtnl_unlock(); 1726 rtnl_unlock();
1726 return 1; 1727 return 1;
1727 } 1728 }
@@ -5244,7 +5245,7 @@ static int nl80211_trigger_scan(struct sk_buff *skb, struct genl_info *info)
5244 if (!rdev->ops->scan) 5245 if (!rdev->ops->scan)
5245 return -EOPNOTSUPP; 5246 return -EOPNOTSUPP;
5246 5247
5247 if (rdev->scan_req) { 5248 if (rdev->scan_req || rdev->scan_msg) {
5248 err = -EBUSY; 5249 err = -EBUSY;
5249 goto unlock; 5250 goto unlock;
5250 } 5251 }
@@ -10011,40 +10012,31 @@ void nl80211_send_scan_start(struct cfg80211_registered_device *rdev,
10011 NL80211_MCGRP_SCAN, GFP_KERNEL); 10012 NL80211_MCGRP_SCAN, GFP_KERNEL);
10012} 10013}
10013 10014
10014void nl80211_send_scan_done(struct cfg80211_registered_device *rdev, 10015struct sk_buff *nl80211_build_scan_msg(struct cfg80211_registered_device *rdev,
10015 struct wireless_dev *wdev) 10016 struct wireless_dev *wdev, bool aborted)
10016{ 10017{
10017 struct sk_buff *msg; 10018 struct sk_buff *msg;
10018 10019
10019 msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL); 10020 msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
10020 if (!msg) 10021 if (!msg)
10021 return; 10022 return NULL;
10022 10023
10023 if (nl80211_send_scan_msg(msg, rdev, wdev, 0, 0, 0, 10024 if (nl80211_send_scan_msg(msg, rdev, wdev, 0, 0, 0,
10024 NL80211_CMD_NEW_SCAN_RESULTS) < 0) { 10025 aborted ? NL80211_CMD_SCAN_ABORTED :
10026 NL80211_CMD_NEW_SCAN_RESULTS) < 0) {
10025 nlmsg_free(msg); 10027 nlmsg_free(msg);
10026 return; 10028 return NULL;
10027 } 10029 }
10028 10030
10029 genlmsg_multicast_netns(&nl80211_fam, wiphy_net(&rdev->wiphy), msg, 0, 10031 return msg;
10030 NL80211_MCGRP_SCAN, GFP_KERNEL);
10031} 10032}
10032 10033
10033void nl80211_send_scan_aborted(struct cfg80211_registered_device *rdev, 10034void nl80211_send_scan_result(struct cfg80211_registered_device *rdev,
10034 struct wireless_dev *wdev) 10035 struct sk_buff *msg)
10035{ 10036{
10036 struct sk_buff *msg;
10037
10038 msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
10039 if (!msg) 10037 if (!msg)
10040 return; 10038 return;
10041 10039
10042 if (nl80211_send_scan_msg(msg, rdev, wdev, 0, 0, 0,
10043 NL80211_CMD_SCAN_ABORTED) < 0) {
10044 nlmsg_free(msg);
10045 return;
10046 }
10047
10048 genlmsg_multicast_netns(&nl80211_fam, wiphy_net(&rdev->wiphy), msg, 0, 10040 genlmsg_multicast_netns(&nl80211_fam, wiphy_net(&rdev->wiphy), msg, 0,
10049 NL80211_MCGRP_SCAN, GFP_KERNEL); 10041 NL80211_MCGRP_SCAN, GFP_KERNEL);
10050} 10042}
diff --git a/net/wireless/nl80211.h b/net/wireless/nl80211.h
index b1b231324e10..75799746d845 100644
--- a/net/wireless/nl80211.h
+++ b/net/wireless/nl80211.h
@@ -8,10 +8,10 @@ void nl80211_exit(void);
8void nl80211_notify_dev_rename(struct cfg80211_registered_device *rdev); 8void nl80211_notify_dev_rename(struct cfg80211_registered_device *rdev);
9void nl80211_send_scan_start(struct cfg80211_registered_device *rdev, 9void nl80211_send_scan_start(struct cfg80211_registered_device *rdev,
10 struct wireless_dev *wdev); 10 struct wireless_dev *wdev);
11void nl80211_send_scan_done(struct cfg80211_registered_device *rdev, 11struct sk_buff *nl80211_build_scan_msg(struct cfg80211_registered_device *rdev,
12 struct wireless_dev *wdev); 12 struct wireless_dev *wdev, bool aborted);
13void nl80211_send_scan_aborted(struct cfg80211_registered_device *rdev, 13void nl80211_send_scan_result(struct cfg80211_registered_device *rdev,
14 struct wireless_dev *wdev); 14 struct sk_buff *msg);
15void nl80211_send_sched_scan(struct cfg80211_registered_device *rdev, 15void nl80211_send_sched_scan(struct cfg80211_registered_device *rdev,
16 struct net_device *netdev, u32 cmd); 16 struct net_device *netdev, u32 cmd);
17void nl80211_send_sched_scan_results(struct cfg80211_registered_device *rdev, 17void nl80211_send_sched_scan_results(struct cfg80211_registered_device *rdev,
diff --git a/net/wireless/reg.c b/net/wireless/reg.c
index 9b897fca7487..f0541370e68e 100644
--- a/net/wireless/reg.c
+++ b/net/wireless/reg.c
@@ -1700,7 +1700,7 @@ static void reg_process_hint(struct regulatory_request *reg_request)
1700 return; 1700 return;
1701 case NL80211_REGDOM_SET_BY_USER: 1701 case NL80211_REGDOM_SET_BY_USER:
1702 treatment = reg_process_hint_user(reg_request); 1702 treatment = reg_process_hint_user(reg_request);
1703 if (treatment == REG_REQ_OK || 1703 if (treatment == REG_REQ_IGNORE ||
1704 treatment == REG_REQ_ALREADY_SET) 1704 treatment == REG_REQ_ALREADY_SET)
1705 return; 1705 return;
1706 schedule_delayed_work(&reg_timeout, msecs_to_jiffies(3142)); 1706 schedule_delayed_work(&reg_timeout, msecs_to_jiffies(3142));
@@ -2373,6 +2373,7 @@ static int reg_set_rd_country_ie(const struct ieee80211_regdomain *rd,
2373int set_regdom(const struct ieee80211_regdomain *rd) 2373int set_regdom(const struct ieee80211_regdomain *rd)
2374{ 2374{
2375 struct regulatory_request *lr; 2375 struct regulatory_request *lr;
2376 bool user_reset = false;
2376 int r; 2377 int r;
2377 2378
2378 if (!reg_is_valid_request(rd->alpha2)) { 2379 if (!reg_is_valid_request(rd->alpha2)) {
@@ -2389,6 +2390,7 @@ int set_regdom(const struct ieee80211_regdomain *rd)
2389 break; 2390 break;
2390 case NL80211_REGDOM_SET_BY_USER: 2391 case NL80211_REGDOM_SET_BY_USER:
2391 r = reg_set_rd_user(rd, lr); 2392 r = reg_set_rd_user(rd, lr);
2393 user_reset = true;
2392 break; 2394 break;
2393 case NL80211_REGDOM_SET_BY_DRIVER: 2395 case NL80211_REGDOM_SET_BY_DRIVER:
2394 r = reg_set_rd_driver(rd, lr); 2396 r = reg_set_rd_driver(rd, lr);
@@ -2402,8 +2404,14 @@ int set_regdom(const struct ieee80211_regdomain *rd)
2402 } 2404 }
2403 2405
2404 if (r) { 2406 if (r) {
2405 if (r == -EALREADY) 2407 switch (r) {
2408 case -EALREADY:
2406 reg_set_request_processed(); 2409 reg_set_request_processed();
2410 break;
2411 default:
2412 /* Back to world regulatory in case of errors */
2413 restore_regulatory_settings(user_reset);
2414 }
2407 2415
2408 kfree(rd); 2416 kfree(rd);
2409 return r; 2417 return r;
diff --git a/net/wireless/scan.c b/net/wireless/scan.c
index b528e31da2cf..d1ed4aebbbb7 100644
--- a/net/wireless/scan.c
+++ b/net/wireless/scan.c
@@ -161,18 +161,25 @@ static void __cfg80211_bss_expire(struct cfg80211_registered_device *dev,
161 dev->bss_generation++; 161 dev->bss_generation++;
162} 162}
163 163
164void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev) 164void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev,
165 bool send_message)
165{ 166{
166 struct cfg80211_scan_request *request; 167 struct cfg80211_scan_request *request;
167 struct wireless_dev *wdev; 168 struct wireless_dev *wdev;
169 struct sk_buff *msg;
168#ifdef CONFIG_CFG80211_WEXT 170#ifdef CONFIG_CFG80211_WEXT
169 union iwreq_data wrqu; 171 union iwreq_data wrqu;
170#endif 172#endif
171 173
172 ASSERT_RTNL(); 174 ASSERT_RTNL();
173 175
174 request = rdev->scan_req; 176 if (rdev->scan_msg) {
177 nl80211_send_scan_result(rdev, rdev->scan_msg);
178 rdev->scan_msg = NULL;
179 return;
180 }
175 181
182 request = rdev->scan_req;
176 if (!request) 183 if (!request)
177 return; 184 return;
178 185
@@ -186,18 +193,16 @@ void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev)
186 if (wdev->netdev) 193 if (wdev->netdev)
187 cfg80211_sme_scan_done(wdev->netdev); 194 cfg80211_sme_scan_done(wdev->netdev);
188 195
189 if (request->aborted) { 196 if (!request->aborted &&
190 nl80211_send_scan_aborted(rdev, wdev); 197 request->flags & NL80211_SCAN_FLAG_FLUSH) {
191 } else { 198 /* flush entries from previous scans */
192 if (request->flags & NL80211_SCAN_FLAG_FLUSH) { 199 spin_lock_bh(&rdev->bss_lock);
193 /* flush entries from previous scans */ 200 __cfg80211_bss_expire(rdev, request->scan_start);
194 spin_lock_bh(&rdev->bss_lock); 201 spin_unlock_bh(&rdev->bss_lock);
195 __cfg80211_bss_expire(rdev, request->scan_start);
196 spin_unlock_bh(&rdev->bss_lock);
197 }
198 nl80211_send_scan_done(rdev, wdev);
199 } 202 }
200 203
204 msg = nl80211_build_scan_msg(rdev, wdev, request->aborted);
205
201#ifdef CONFIG_CFG80211_WEXT 206#ifdef CONFIG_CFG80211_WEXT
202 if (wdev->netdev && !request->aborted) { 207 if (wdev->netdev && !request->aborted) {
203 memset(&wrqu, 0, sizeof(wrqu)); 208 memset(&wrqu, 0, sizeof(wrqu));
@@ -211,6 +216,11 @@ void ___cfg80211_scan_done(struct cfg80211_registered_device *rdev)
211 216
212 rdev->scan_req = NULL; 217 rdev->scan_req = NULL;
213 kfree(request); 218 kfree(request);
219
220 if (!send_message)
221 rdev->scan_msg = msg;
222 else
223 nl80211_send_scan_result(rdev, msg);
214} 224}
215 225
216void __cfg80211_scan_done(struct work_struct *wk) 226void __cfg80211_scan_done(struct work_struct *wk)
@@ -221,7 +231,7 @@ void __cfg80211_scan_done(struct work_struct *wk)
221 scan_done_wk); 231 scan_done_wk);
222 232
223 rtnl_lock(); 233 rtnl_lock();
224 ___cfg80211_scan_done(rdev); 234 ___cfg80211_scan_done(rdev, true);
225 rtnl_unlock(); 235 rtnl_unlock();
226} 236}
227 237
@@ -1079,7 +1089,7 @@ int cfg80211_wext_siwscan(struct net_device *dev,
1079 if (IS_ERR(rdev)) 1089 if (IS_ERR(rdev))
1080 return PTR_ERR(rdev); 1090 return PTR_ERR(rdev);
1081 1091
1082 if (rdev->scan_req) { 1092 if (rdev->scan_req || rdev->scan_msg) {
1083 err = -EBUSY; 1093 err = -EBUSY;
1084 goto out; 1094 goto out;
1085 } 1095 }
@@ -1481,7 +1491,7 @@ int cfg80211_wext_giwscan(struct net_device *dev,
1481 if (IS_ERR(rdev)) 1491 if (IS_ERR(rdev))
1482 return PTR_ERR(rdev); 1492 return PTR_ERR(rdev);
1483 1493
1484 if (rdev->scan_req) 1494 if (rdev->scan_req || rdev->scan_msg)
1485 return -EAGAIN; 1495 return -EAGAIN;
1486 1496
1487 res = ieee80211_scan_results(rdev, info, extra, data->length); 1497 res = ieee80211_scan_results(rdev, info, extra, data->length);
diff --git a/net/wireless/sme.c b/net/wireless/sme.c
index a63509118508..f04d4c32e96e 100644
--- a/net/wireless/sme.c
+++ b/net/wireless/sme.c
@@ -67,7 +67,7 @@ static int cfg80211_conn_scan(struct wireless_dev *wdev)
67 ASSERT_RDEV_LOCK(rdev); 67 ASSERT_RDEV_LOCK(rdev);
68 ASSERT_WDEV_LOCK(wdev); 68 ASSERT_WDEV_LOCK(wdev);
69 69
70 if (rdev->scan_req) 70 if (rdev->scan_req || rdev->scan_msg)
71 return -EBUSY; 71 return -EBUSY;
72 72
73 if (wdev->conn->params.channel) 73 if (wdev->conn->params.channel)
diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
index 4b98b25793c5..1d5c7bf29938 100644
--- a/net/xfrm/xfrm_policy.c
+++ b/net/xfrm/xfrm_policy.c
@@ -1158,7 +1158,7 @@ static struct xfrm_policy *__xfrm_policy_unlink(struct xfrm_policy *pol,
1158 if (hlist_unhashed(&pol->bydst)) 1158 if (hlist_unhashed(&pol->bydst))
1159 return NULL; 1159 return NULL;
1160 1160
1161 hlist_del(&pol->bydst); 1161 hlist_del_init(&pol->bydst);
1162 hlist_del(&pol->byidx); 1162 hlist_del(&pol->byidx);
1163 list_del(&pol->walk.all); 1163 list_del(&pol->walk.all);
1164 net->xfrm.policy_count[dir]--; 1164 net->xfrm.policy_count[dir]--;
diff --git a/net/xfrm/xfrm_state.c b/net/xfrm/xfrm_state.c
index a26b7aa79475..40f1b3e92e78 100644
--- a/net/xfrm/xfrm_state.c
+++ b/net/xfrm/xfrm_state.c
@@ -1159,6 +1159,11 @@ static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
1159 } 1159 }
1160 x->props.aalgo = orig->props.aalgo; 1160 x->props.aalgo = orig->props.aalgo;
1161 1161
1162 if (orig->aead) {
1163 x->aead = xfrm_algo_aead_clone(orig->aead);
1164 if (!x->aead)
1165 goto error;
1166 }
1162 if (orig->ealg) { 1167 if (orig->ealg) {
1163 x->ealg = xfrm_algo_clone(orig->ealg); 1168 x->ealg = xfrm_algo_clone(orig->ealg);
1164 if (!x->ealg) 1169 if (!x->ealg)
@@ -1201,6 +1206,9 @@ static struct xfrm_state *xfrm_state_clone(struct xfrm_state *orig, int *errp)
1201 x->props.flags = orig->props.flags; 1206 x->props.flags = orig->props.flags;
1202 x->props.extra_flags = orig->props.extra_flags; 1207 x->props.extra_flags = orig->props.extra_flags;
1203 1208
1209 x->tfcpad = orig->tfcpad;
1210 x->replay_maxdiff = orig->replay_maxdiff;
1211 x->replay_maxage = orig->replay_maxage;
1204 x->curlft.add_time = orig->curlft.add_time; 1212 x->curlft.add_time = orig->curlft.add_time;
1205 x->km.state = orig->km.state; 1213 x->km.state = orig->km.state;
1206 x->km.seq = orig->km.seq; 1214 x->km.seq = orig->km.seq;
@@ -1215,11 +1223,12 @@ out:
1215 return NULL; 1223 return NULL;
1216} 1224}
1217 1225
1218/* net->xfrm.xfrm_state_lock is held */
1219struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *net) 1226struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *net)
1220{ 1227{
1221 unsigned int h; 1228 unsigned int h;
1222 struct xfrm_state *x; 1229 struct xfrm_state *x = NULL;
1230
1231 spin_lock_bh(&net->xfrm.xfrm_state_lock);
1223 1232
1224 if (m->reqid) { 1233 if (m->reqid) {
1225 h = xfrm_dst_hash(net, &m->old_daddr, &m->old_saddr, 1234 h = xfrm_dst_hash(net, &m->old_daddr, &m->old_saddr,
@@ -1236,7 +1245,7 @@ struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *n
1236 m->old_family)) 1245 m->old_family))
1237 continue; 1246 continue;
1238 xfrm_state_hold(x); 1247 xfrm_state_hold(x);
1239 return x; 1248 break;
1240 } 1249 }
1241 } else { 1250 } else {
1242 h = xfrm_src_hash(net, &m->old_daddr, &m->old_saddr, 1251 h = xfrm_src_hash(net, &m->old_daddr, &m->old_saddr,
@@ -1251,11 +1260,13 @@ struct xfrm_state *xfrm_migrate_state_find(struct xfrm_migrate *m, struct net *n
1251 m->old_family)) 1260 m->old_family))
1252 continue; 1261 continue;
1253 xfrm_state_hold(x); 1262 xfrm_state_hold(x);
1254 return x; 1263 break;
1255 } 1264 }
1256 } 1265 }
1257 1266
1258 return NULL; 1267 spin_unlock_bh(&net->xfrm.xfrm_state_lock);
1268
1269 return x;
1259} 1270}
1260EXPORT_SYMBOL(xfrm_migrate_state_find); 1271EXPORT_SYMBOL(xfrm_migrate_state_find);
1261 1272
@@ -1451,7 +1462,7 @@ xfrm_state_sort(struct xfrm_state **dst, struct xfrm_state **src, int n,
1451{ 1462{
1452 int err = 0; 1463 int err = 0;
1453 struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family); 1464 struct xfrm_state_afinfo *afinfo = xfrm_state_get_afinfo(family);
1454 struct net *net = xs_net(*dst); 1465 struct net *net = xs_net(*src);
1455 1466
1456 if (!afinfo) 1467 if (!afinfo)
1457 return -EAFNOSUPPORT; 1468 return -EAFNOSUPPORT;
diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
index 1ae3ec7c18b0..c274179d60a2 100644
--- a/net/xfrm/xfrm_user.c
+++ b/net/xfrm/xfrm_user.c
@@ -32,11 +32,6 @@
32#include <linux/in6.h> 32#include <linux/in6.h>
33#endif 33#endif
34 34
35static inline int aead_len(struct xfrm_algo_aead *alg)
36{
37 return sizeof(*alg) + ((alg->alg_key_len + 7) / 8);
38}
39
40static int verify_one_alg(struct nlattr **attrs, enum xfrm_attr_type_t type) 35static int verify_one_alg(struct nlattr **attrs, enum xfrm_attr_type_t type)
41{ 36{
42 struct nlattr *rt = attrs[type]; 37 struct nlattr *rt = attrs[type];