Merge branch 'tipc-next'

Jon Maloy says:

====================
tipc: separate link aggregation from link layer

We continue the work on separating the roles of the link aggregation and
link layers, as well as making code cleanups in general.

This second commit batch focuses on moving the orchestration of link
failover and synchronization to the node level, as well as preparing the
node lock structure for further future impovements. We also make some
changes to message delivery between link and socket layer, in order to
make this mechanism safer and less obscure.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

net/tipc/bearer.c

@@ -343,7 +343,7 @@ static int tipc_enable_bearer(struct net *net, const char *name,
 static int tipc_reset_bearer(struct net *net, struct tipc_bearer *b_ptr)
 {
 	pr_info("Resetting bearer <%s>\n", b_ptr->name);
-	tipc_link_delete_list(net, b_ptr->identity);
+	tipc_node_delete_links(net, b_ptr->identity);
 	tipc_disc_reset(net, b_ptr);
 	return 0;
 }
@@ -361,7 +361,7 @@ static void bearer_disable(struct net *net, struct tipc_bearer *b_ptr)
 	pr_info("Disabling bearer <%s>\n", b_ptr->name);
 	b_ptr->media->disable_media(b_ptr);
 
-	tipc_link_delete_list(net, b_ptr->identity);
+	tipc_node_delete_links(net, b_ptr->identity);
 	if (b_ptr->link_req)
 		tipc_disc_delete(b_ptr->link_req);
 

net/tipc/core.h

@@ -109,6 +109,11 @@ struct tipc_net {
 	atomic_t subscription_count;
 };
 
+static inline struct tipc_net *tipc_net(struct net *net)
+{
+	return net_generic(net, tipc_net_id);
+}
+
 static inline u16 mod(u16 x)
 {
 	return x & 0xffffu;

net/tipc/discover.c

@@ -120,29 +120,24 @@ static void disc_dupl_alert(struct tipc_bearer *b_ptr, u32 node_addr,
  * @buf: buffer containing message
  * @bearer: bearer that message arrived on
  */
-void tipc_disc_rcv(struct net *net, struct sk_buff *buf,
+void tipc_disc_rcv(struct net *net, struct sk_buff *skb,
 		   struct tipc_bearer *bearer)
 {
 	struct tipc_net *tn = net_generic(net, tipc_net_id);
-	struct tipc_node *node;
 	struct tipc_media_addr maddr;
-	struct sk_buff *rbuf;
-	struct tipc_msg *msg = buf_msg(buf);
-	u32 ddom = msg_dest_domain(msg);
-	u32 onode = msg_prevnode(msg);
-	u32 net_id = msg_bc_netid(msg);
-	u32 mtyp = msg_type(msg);
-	u32 signature = msg_node_sig(msg);
-	u16 caps = msg_node_capabilities(msg);
-	bool addr_match = false;
-	bool sign_match = false;
-	bool link_up = false;
-	bool accept_addr = false;
-	bool accept_sign = false;
+	struct sk_buff *rskb;
+	struct tipc_msg *hdr = buf_msg(skb);
+	u32 ddom = msg_dest_domain(hdr);
+	u32 onode = msg_prevnode(hdr);
+	u32 net_id = msg_bc_netid(hdr);
+	u32 mtyp = msg_type(hdr);
+	u32 signature = msg_node_sig(hdr);
+	u16 caps = msg_node_capabilities(hdr);
 	bool respond = false;
+	bool dupl_addr = false;
 
-	bearer->media->msg2addr(bearer, &maddr, msg_media_addr(msg));
-	kfree_skb(buf);
+	bearer->media->msg2addr(bearer, &maddr, msg_media_addr(hdr));
+	kfree_skb(skb);
 
 	/* Ensure message from node is valid and communication is permitted */
 	if (net_id != tn->net_id)
@@ -164,91 +159,20 @@ void tipc_disc_rcv(struct net *net, struct sk_buff *buf,
 	if (!tipc_in_scope(bearer->domain, onode))
 		return;
 
-	node = tipc_node_create(net, onode);
-	if (!node)
-		return;
-	tipc_node_lock(node);
-	node->capabilities = caps;
-
-	/* Prepare to validate requesting node's signature and media address */
-	sign_match = (signature == node->signature);
-	tipc_node_check_dest(node, bearer, &link_up, &addr_match, &maddr);
-
-	/* These three flags give us eight permutations: */
-
-	if (sign_match && addr_match && link_up) {
-		/* All is fine. Do nothing. */
-	} else if (sign_match && addr_match && !link_up) {
-		/* Respond. The link will come up in due time */
-		respond = true;
-	} else if (sign_match && !addr_match && link_up) {
-		/* Peer has changed i/f address without rebooting.
-		 * If so, the link will reset soon, and the next
-		 * discovery will be accepted. So we can ignore it.
-		 * It may also be an cloned or malicious peer having
-		 * chosen the same node address and signature as an
-		 * existing one.
-		 * Ignore requests until the link goes down, if ever.
-		 */
-		disc_dupl_alert(bearer, onode, &maddr);
-	} else if (sign_match && !addr_match && !link_up) {
-		/* Peer link has changed i/f address without rebooting.
-		 * It may also be a cloned or malicious peer; we can't
-		 * distinguish between the two.
-		 * The signature is correct, so we must accept.
-		 */
-		accept_addr = true;
-		respond = true;
-	} else if (!sign_match && addr_match && link_up) {
-		/* Peer node rebooted. Two possibilities:
-		 *  - Delayed re-discovery; this link endpoint has already
-		 *    reset and re-established contact with the peer, before
-		 *    receiving a discovery message from that node.
-		 *    (The peer happened to receive one from this node first).
-		 *  - The peer came back so fast that our side has not
-		 *    discovered it yet. Probing from this side will soon
-		 *    reset the link, since there can be no working link
-		 *    endpoint at the peer end, and the link will re-establish.
-		 *  Accept the signature, since it comes from a known peer.
-		 */
-		accept_sign = true;
-	} else if (!sign_match && addr_match && !link_up) {
-		/*  The peer node has rebooted.
-		 *  Accept signature, since it is a known peer.
-		 */
-		accept_sign = true;
-		respond = true;
-	} else if (!sign_match && !addr_match && link_up) {
-		/* Peer rebooted with new address, or a new/duplicate peer.
-		 * Ignore until the link goes down, if ever.
-		 */
+	tipc_node_check_dest(net, onode, bearer, caps, signature,
+			     &maddr, &respond, &dupl_addr);
+	if (dupl_addr)
 		disc_dupl_alert(bearer, onode, &maddr);
-	} else if (!sign_match && !addr_match && !link_up) {
-		/* Peer rebooted with new address, or it is a new peer.
-		 * Accept signature and address.
-		*/
-		accept_sign = true;
-		accept_addr = true;
-		respond = true;
-	}
-
-	if (accept_sign)
-		node->signature = signature;
-
-	if (accept_addr && !tipc_node_update_dest(node, bearer, &maddr))
-		respond = false;
 
 	/* Send response, if necessary */
 	if (respond && (mtyp == DSC_REQ_MSG)) {
-		rbuf = tipc_buf_acquire(MAX_H_SIZE);
-		if (rbuf) {
-			tipc_disc_init_msg(net, rbuf, DSC_RESP_MSG, bearer);
-			tipc_bearer_send(net, bearer->identity, rbuf, &maddr);
-			kfree_skb(rbuf);
+		rskb = tipc_buf_acquire(MAX_H_SIZE);
+		if (rskb) {
+			tipc_disc_init_msg(net, rskb, DSC_RESP_MSG, bearer);
+			tipc_bearer_send(net, bearer->identity, rskb, &maddr);
+			kfree_skb(rskb);
 		}
 	}
-	tipc_node_unlock(node);
-	tipc_node_put(node);
 }
 
 /**

net/tipc/link.h

@@ -49,13 +49,17 @@
  */
 #define INVALID_LINK_SEQ 0x10000
 
-
-/* Link endpoint receive states
+/* Link FSM events:
  */
 enum {
-	TIPC_LINK_OPEN,
-	TIPC_LINK_BLOCKED,
-	TIPC_LINK_TUNNEL
+	LINK_ESTABLISH_EVT       = 0xec1ab1e,
+	LINK_PEER_RESET_EVT      = 0x9eed0e,
+	LINK_FAILURE_EVT         = 0xfa110e,
+	LINK_RESET_EVT           = 0x10ca1d0e,
+	LINK_FAILOVER_BEGIN_EVT  = 0xfa110bee,
+	LINK_FAILOVER_END_EVT    = 0xfa110ede,
+	LINK_SYNCH_BEGIN_EVT     = 0xc1ccbee,
+	LINK_SYNCH_END_EVT       = 0xc1ccede
 };
 
 /* Events returned from link at packet reception or at timeout
@@ -120,7 +124,6 @@ struct tipc_stats {
  * @pmsg: convenience pointer to "proto_msg" field
  * @priority: current link priority
  * @net_plane: current link network plane ('A' through 'H')
- * @exec_mode: transmit/receive mode for link endpoint instance
  * @backlog_limit: backlog queue congestion thresholds (indexed by importance)
  * @exp_msg_count: # of tunnelled messages expected during link changeover
  * @reset_rcv_checkpt: seq # of last acknowledged message at time of link reset
@@ -145,7 +148,7 @@ struct tipc_stats {
 struct tipc_link {
 	u32 addr;
 	char name[TIPC_MAX_LINK_NAME];
-	struct tipc_media_addr media_addr;
+	struct tipc_media_addr *media_addr;
 	struct tipc_node *owner;
 
 	/* Management and link supervision data */
@@ -155,7 +158,7 @@ struct tipc_link {
 	u32 tolerance;
 	unsigned long keepalive_intv;
 	u32 abort_limit;
-	int state;
+	u32 state;
 	u32 silent_intv_cnt;
 	struct {
 		unchar hdr[INT_H_SIZE];
@@ -164,13 +167,10 @@ struct tipc_link {
 	struct tipc_msg *pmsg;
 	u32 priority;
 	char net_plane;
-	u8 exec_mode;
-	u16 synch_point;
 
-	/* Failover */
-	u16 failover_pkts;
-	u16 failover_checkpt;
-	struct sk_buff *failover_skb;
+	/* Failover/synch */
+	u16 drop_point;
+	struct sk_buff *failover_reasm_skb;
 
 	/* Max packet negotiation */
 	u16 mtu;
@@ -205,25 +205,25 @@ struct tipc_link {
 	struct tipc_stats stats;
 };
 
-struct tipc_port;
-
-struct tipc_link *tipc_link_create(struct tipc_node *n,
-				   struct tipc_bearer *b,
-				   const struct tipc_media_addr *maddr,
-				   struct sk_buff_head *inputq,
-				   struct sk_buff_head *namedq);
-void tipc_link_delete(struct tipc_link *link);
-void tipc_link_delete_list(struct net *net, unsigned int bearer_id);
-void tipc_link_failover_send_queue(struct tipc_link *l_ptr);
-void tipc_link_dup_queue_xmit(struct tipc_link *l_ptr, struct tipc_link *dest);
+bool tipc_link_create(struct tipc_node *n, struct tipc_bearer *b, u32 session,
+		      u32 ownnode, u32 peer, struct tipc_media_addr *maddr,
+		      struct sk_buff_head *inputq, struct sk_buff_head *namedq,
+		      struct tipc_link **link);
+void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
+			   int mtyp, struct sk_buff_head *xmitq);
+void tipc_link_build_bcast_sync_msg(struct tipc_link *l,
+				    struct sk_buff_head *xmitq);
+int tipc_link_fsm_evt(struct tipc_link *l, int evt);
 void tipc_link_reset_fragments(struct tipc_link *l_ptr);
-int tipc_link_is_up(struct tipc_link *l_ptr);
+bool tipc_link_is_up(struct tipc_link *l);
+bool tipc_link_is_reset(struct tipc_link *l);
+bool tipc_link_is_synching(struct tipc_link *l);
+bool tipc_link_is_failingover(struct tipc_link *l);
+bool tipc_link_is_blocked(struct tipc_link *l);
 int tipc_link_is_active(struct tipc_link *l_ptr);
 void tipc_link_purge_queues(struct tipc_link *l_ptr);
 void tipc_link_purge_backlog(struct tipc_link *l);
-void tipc_link_reset_all(struct tipc_node *node);
 void tipc_link_reset(struct tipc_link *l_ptr);
-void tipc_link_activate(struct tipc_link *link);
 int __tipc_link_xmit(struct net *net, struct tipc_link *link,
 		     struct sk_buff_head *list);
 int tipc_link_xmit(struct tipc_link *link,	struct sk_buff_head *list,
@@ -243,13 +243,8 @@ int tipc_nl_link_get(struct sk_buff *skb, struct genl_info *info);
 int tipc_nl_link_set(struct sk_buff *skb, struct genl_info *info);
 int tipc_nl_link_reset_stats(struct sk_buff *skb, struct genl_info *info);
 int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[]);
-void link_prepare_wakeup(struct tipc_link *l);
 int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq);
 int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
 		  struct sk_buff_head *xmitq);
-static inline u32 link_own_addr(struct tipc_link *l)
-{
-	return msg_prevnode(l->pmsg);
-}
 
 #endif

net/tipc/msg.h

@@ -110,7 +110,6 @@ struct tipc_skb_cb {
 	struct sk_buff *tail;
 	bool validated;
 	bool wakeup_pending;
-	bool bundling;
 	u16 chain_sz;
 	u16 chain_imp;
 };
@@ -559,15 +558,6 @@ static inline void msg_set_node_capabilities(struct tipc_msg *m, u32 n)
 	msg_set_bits(m, 1, 15, 0x1fff, n);
 }
 
-static inline bool msg_dup(struct tipc_msg *m)
-{
-	if (likely(msg_user(m) != TUNNEL_PROTOCOL))
-		return false;
-	if (msg_type(m) != SYNCH_MSG)
-		return false;
-	return true;
-}
-
 /*
  * Word 2
  */
@@ -621,12 +611,12 @@ static inline void msg_set_fragm_no(struct tipc_msg *m, u32 n)
 }
 
 
-static inline u32 msg_next_sent(struct tipc_msg *m)
+static inline u16 msg_next_sent(struct tipc_msg *m)
 {
 	return msg_bits(m, 4, 0, 0xffff);
 }
 
-static inline void msg_set_next_sent(struct tipc_msg *m, u32 n)
+static inline void msg_set_next_sent(struct tipc_msg *m, u16 n)
 {
 	msg_set_bits(m, 4, 0, 0xffff, n);
 }
@@ -727,12 +717,12 @@ static inline char *msg_media_addr(struct tipc_msg *m)
 /*
  * Word 9
  */
-static inline u32 msg_msgcnt(struct tipc_msg *m)
+static inline u16 msg_msgcnt(struct tipc_msg *m)
 {
 	return msg_bits(m, 9, 16, 0xffff);
 }
 
-static inline void msg_set_msgcnt(struct tipc_msg *m, u32 n)
+static inline void msg_set_msgcnt(struct tipc_msg *m, u16 n)
 {
 	msg_set_bits(m, 9, 16, 0xffff, n);
 }
@@ -767,19 +757,19 @@ static inline void msg_set_link_tolerance(struct tipc_msg *m, u32 n)
 	msg_set_bits(m, 9, 0, 0xffff, n);
 }
 
-static inline bool msg_is_traffic(struct tipc_msg *m)
+static inline bool msg_peer_link_is_up(struct tipc_msg *m)
 {
 	if (likely(msg_user(m) != LINK_PROTOCOL))
 		return true;
-	if ((msg_type(m) == RESET_MSG) || (msg_type(m) == ACTIVATE_MSG))
-		return false;
+	if (msg_type(m) == STATE_MSG)
 		return true;
+	return false;
 }
 
-static inline bool msg_peer_is_up(struct tipc_msg *m)
+static inline bool msg_peer_node_is_up(struct tipc_msg *m)
 {
-	if (likely(msg_is_traffic(m)))
-		return false;
+	if (msg_peer_link_is_up(m))
+		return true;
 	return msg_redundant_link(m);
 }
 
@@ -872,28 +862,6 @@ static inline struct sk_buff *tipc_skb_dequeue(struct sk_buff_head *list,
 	return skb;
 }
 
-/* tipc_skb_queue_tail(): add buffer to tail of list;
- * @list: list to be appended to
- * @skb: buffer to append. Always appended
- * @dport: the destination port of the buffer
- * returns true if dport differs from previous destination
- */
-static inline bool tipc_skb_queue_tail(struct sk_buff_head *list,
-				       struct sk_buff *skb, u32 dport)
-{
-	struct sk_buff *_skb = NULL;
-	bool rv = false;
-
-	spin_lock_bh(&list->lock);
-	_skb = skb_peek_tail(list);
-	if (!_skb || (msg_destport(buf_msg(_skb)) != dport) ||
-	    (skb_queue_len(list) > 32))
-		rv = true;
-	__skb_queue_tail(list, skb);
-	spin_unlock_bh(&list->lock);
-	return rv;
-}
-
 /* tipc_skb_queue_sorted(); sort pkt into list according to sequence number
  * @list: list to be appended to
  * @skb: buffer to add
@@ -926,4 +894,33 @@ static inline bool __tipc_skb_queue_sorted(struct sk_buff_head *list,
 	return false;
 }
 
+/* tipc_skb_queue_splice_tail - append an skb list to lock protected list
+ * @list: the new list to append. Not lock protected
+ * @head: target list. Lock protected.
+ */
+static inline void tipc_skb_queue_splice_tail(struct sk_buff_head *list,
+					      struct sk_buff_head *head)
+{
+	spin_lock_bh(&head->lock);
+	skb_queue_splice_tail(list, head);
+	spin_unlock_bh(&head->lock);
+}
+
+/* tipc_skb_queue_splice_tail_init - merge two lock protected skb lists
+ * @list: the new list to add. Lock protected. Will be reinitialized
+ * @head: target list. Lock protected.
+ */
+static inline void tipc_skb_queue_splice_tail_init(struct sk_buff_head *list,
+						   struct sk_buff_head *head)
+{
+	struct sk_buff_head tmp;
+
+	__skb_queue_head_init(&tmp);
+
+	spin_lock_bh(&list->lock);
+	skb_queue_splice_tail_init(list, &tmp);
+	spin_unlock_bh(&list->lock);
+	tipc_skb_queue_splice_tail(&tmp, head);
+}
+
 #endif

net/tipc/node.h

@@ -47,39 +47,17 @@
 
 #define INVALID_BEARER_ID -1
 
-/* Node FSM states and events:
- */
-enum {
-	SELF_DOWN_PEER_DOWN    = 0xdd,
-	SELF_UP_PEER_UP        = 0xaa,
-	SELF_DOWN_PEER_LEAVING = 0xd1,
-	SELF_UP_PEER_COMING    = 0xac,
-	SELF_COMING_PEER_UP    = 0xca,
-	SELF_LEAVING_PEER_DOWN = 0x1d,
-};
-
-enum {
-	SELF_ESTABL_CONTACT_EVT = 0xec,
-	SELF_LOST_CONTACT_EVT   = 0x1c,
-	PEER_ESTABL_CONTACT_EVT = 0xfec,
-	PEER_LOST_CONTACT_EVT   = 0xf1c
-};
-
 /* Flags used to take different actions according to flag type
- * TIPC_WAIT_PEER_LINKS_DOWN: wait to see that peer's links are down
- * TIPC_WAIT_OWN_LINKS_DOWN: wait until peer node is declared down
  * TIPC_NOTIFY_NODE_DOWN: notify node is down
  * TIPC_NOTIFY_NODE_UP: notify node is up
  * TIPC_DISTRIBUTE_NAME: publish or withdraw link state name type
  */
 enum {
-	TIPC_MSG_EVT                    = 1,
 	TIPC_NOTIFY_NODE_DOWN		= (1 << 3),
 	TIPC_NOTIFY_NODE_UP		= (1 << 4),
 	TIPC_WAKEUP_BCAST_USERS		= (1 << 5),
 	TIPC_NOTIFY_LINK_UP		= (1 << 6),
 	TIPC_NOTIFY_LINK_DOWN		= (1 << 7),
-	TIPC_NAMED_MSG_EVT		= (1 << 8),
 	TIPC_BCAST_MSG_EVT		= (1 << 9),
 	TIPC_BCAST_RESET		= (1 << 10)
 };
@@ -127,6 +105,8 @@ struct tipc_link_entry {
  * @links: array containing references to all links to node
  * @action_flags: bit mask of different types of node actions
  * @bclink: broadcast-related info
+ * @state: connectivity state vs peer node
+ * @sync_point: sequence number where synch/failover is finished
  * @list: links to adjacent nodes in sorted list of cluster's nodes
  * @working_links: number of working links to node (both active and standby)
  * @link_cnt: number of links to node
@@ -142,14 +122,13 @@ struct tipc_node {
 	spinlock_t lock;
 	struct net *net;
 	struct hlist_node hash;
-	struct sk_buff_head *inputq;
-	struct sk_buff_head *namedq;
 	int active_links[2];
 	struct tipc_link_entry links[MAX_BEARERS];
 	int action_flags;
 	struct tipc_node_bclink bclink;
 	struct list_head list;
 	int state;
+	u16 sync_point;
 	int link_cnt;
 	u16 working_links;
 	u16 capabilities;
@@ -164,17 +143,15 @@ struct tipc_node {
 
 struct tipc_node *tipc_node_find(struct net *net, u32 addr);
 void tipc_node_put(struct tipc_node *node);
-struct tipc_node *tipc_node_create(struct net *net, u32 addr);
 void tipc_node_stop(struct net *net);
-void tipc_node_check_dest(struct tipc_node *n, struct tipc_bearer *bearer,
-			  bool *link_up, bool *addr_match,
-			  struct tipc_media_addr *maddr);
-bool tipc_node_update_dest(struct tipc_node *n, struct tipc_bearer *bearer,
-			   struct tipc_media_addr *maddr);
+void tipc_node_check_dest(struct net *net, u32 onode,
+			  struct tipc_bearer *bearer,
+			  u16 capabilities, u32 signature,
+			  struct tipc_media_addr *maddr,
+			  bool *respond, bool *dupl_addr);
+void tipc_node_delete_links(struct net *net, int bearer_id);
 void tipc_node_attach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr);
 void tipc_node_detach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr);
-void tipc_node_link_down(struct tipc_node *n_ptr, int bearer_id);
-void tipc_node_link_up(struct tipc_node *n_ptr, int bearer_id);
 bool tipc_node_is_up(struct tipc_node *n);
 int tipc_node_get_linkname(struct net *net, u32 bearer_id, u32 node,
 			   char *linkname, size_t len);