tipc: handle collisions of 32-bit node address hash values

When a 32-bit node address is generated from a 128-bit identifier,
there is a risk of collisions which must be discovered and handled.

We do this as follows:
- We don't apply the generated address immediately to the node, but do
instead initiate a 1 sec trial period to allow other cluster members
to discover and handle such collisions.

- During the trial period the node periodically sends out a new type
of message, DSC_TRIAL_MSG, using broadcast or emulated broadcast,
to all the other nodes in the cluster.

- When a node is receiving such a message, it must check that the
presented 32-bit identifier either is unused, or was used by the very
same peer in a previous session. In both cases it accepts the request
by not responding to it.

- If it finds that the same node has been up before using a different
address, it responds with a DSC_TRIAL_FAIL_MSG containing that
address.

- If it finds that the address has already been taken by some other
node, it generates a new, unused address and returns it to the
requester.

- During the trial period the requesting node must always be prepared
to accept a failure message, i.e., a message where a peer suggests a
different (or equal) address to the one tried. In those cases it
must apply the suggested value as trial address and restart the trial
period.

This algorithm ensures that in the vast majority of cases a node will
have the same address before and after a reboot. If a legacy user
configures the address explicitly, there will be no trial period and
messages, so this protocol addition is completely backwards compatible.

Acked-by: Ying Xue <[e-mail]>
Signed-off-by: Jon Maloy <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

net/tipc/addr.c

···
 
 	memcpy(tn->node_id, id, NODE_ID_LEN);
 	tipc_nodeid2string(tn->node_id_string, id);
-	tn->node_addr = tmp[0] ^ tmp[1] ^ tmp[2] ^ tmp[3];
+	tn->trial_addr = tmp[0] ^ tmp[1] ^ tmp[2] ^ tmp[3];
 	pr_info("Own node identity %s, cluster identity %u\n",
 		tipc_own_id_string(net), tn->net_id);
 }
···
 		sprintf(node_id, "%x", addr);
 		tipc_set_node_id(net, node_id);
 	}
+	tn->trial_addr = addr;
 	pr_info("32-bit node address hash set to %x\n", addr);
 }
 

net/tipc/bearer.c

···
 {
 	struct tipc_net *tn = tipc_net(net);
 	struct tipc_bearer_names b_names;
-	u32 self = tipc_own_addr(net);
 	int with_this_prio = 1;
 	struct tipc_bearer *b;
 	struct tipc_media *m;
···
 	int res = -EINVAL;
 	char *errstr = "";
 
-	if (!self) {
+	if (!tipc_own_id(net)) {
 		errstr = "not supported in standalone mode";
 		res = -ENOPROTOOPT;
 		goto rejected;

net/tipc/core.c

···
 
 	tn->net_id = 4711;
 	tn->node_addr = 0;
+	tn->trial_addr = 0;
+	tn->addr_trial_end = 0;
 	memset(tn->node_id, 0, sizeof(tn->node_id));
 	memset(tn->node_id_string, 0, sizeof(tn->node_id_string));
 	tn->mon_threshold = TIPC_DEF_MON_THRESHOLD;

net/tipc/core.h

···
 struct tipc_net {
 	u8  node_id[NODE_ID_LEN];
 	u32 node_addr;
+	u32 trial_addr;
+	unsigned long addr_trial_end;
 	char node_id_string[NODE_ID_STR_LEN];
 	int net_id;
 	int random;

net/tipc/discover.c

···
 /*
  * net/tipc/discover.c
  *
- * Copyright (c) 2003-2006, 2014-2015, Ericsson AB
+ * Copyright (c) 2003-2006, 2014-2018, Ericsson AB
  * Copyright (c) 2005-2006, 2010-2011, Wind River Systems
  * All rights reserved.
  *
···
  * @b: ptr to bearer issuing message
  */
 static void tipc_disc_init_msg(struct net *net, struct sk_buff *skb,
-			       u32 mtyp, struct tipc_bearer *b)
+			       u32 mtyp,  struct tipc_bearer *b)
 {
 	struct tipc_net *tn = tipc_net(net);
-	u32 self = tipc_own_addr(net);
 	u32 dest_domain = b->domain;
 	struct tipc_msg *hdr;
 
 	hdr = buf_msg(skb);
-	tipc_msg_init(self, hdr, LINK_CONFIG, mtyp,
+	tipc_msg_init(tn->trial_addr, hdr, LINK_CONFIG, mtyp,
 		      MAX_H_SIZE, dest_domain);
+	msg_set_size(hdr, MAX_H_SIZE + NODE_ID_LEN);
 	msg_set_non_seq(hdr, 1);
 	msg_set_node_sig(hdr, tn->random);
 	msg_set_node_capabilities(hdr, TIPC_NODE_CAPABILITIES);
 	msg_set_dest_domain(hdr, dest_domain);
 	msg_set_bc_netid(hdr, tn->net_id);
 	b->media->addr2msg(msg_media_addr(hdr), &b->addr);
+	msg_set_node_id(hdr, tipc_own_id(net));
 }
 
-static void tipc_disc_msg_xmit(struct net *net, u32 mtyp, u32 dst, u32 src,
+static void tipc_disc_msg_xmit(struct net *net, u32 mtyp, u32 dst,
+			       u32 src, u32 sugg_addr,
 			       struct tipc_media_addr *maddr,
 			       struct tipc_bearer *b)
 {
+	struct tipc_msg *hdr;
 	struct sk_buff *skb;
 
-	skb = tipc_buf_acquire(MAX_H_SIZE, GFP_ATOMIC);
+	skb = tipc_buf_acquire(MAX_H_SIZE + NODE_ID_LEN, GFP_ATOMIC);
 	if (!skb)
 		return;
+	hdr = buf_msg(skb);
 	tipc_disc_init_msg(net, skb, mtyp, b);
+	msg_set_sugg_node_addr(hdr, sugg_addr);
+	msg_set_dest_domain(hdr, dst);
 	tipc_bearer_xmit_skb(net, b->identity, skb, maddr);
 }
 
···
 		media_addr_str, b->name);
 }
 
+/* tipc_disc_addr_trial(): - handle an address uniqueness trial from peer
+ */
+bool tipc_disc_addr_trial_msg(struct tipc_discoverer *d,
+			      struct tipc_media_addr *maddr,
+			      struct tipc_bearer *b,
+			      u32 dst, u32 src,
+			      u32 sugg_addr,
+			      u8 *peer_id,
+			      int mtyp)
+{
+	struct net *net = d->net;
+	struct tipc_net *tn = tipc_net(net);
+	bool trial = time_before(jiffies, tn->addr_trial_end);
+	u32 self = tipc_own_addr(net);
+
+	if (mtyp == DSC_TRIAL_FAIL_MSG) {
+		if (!trial)
+			return true;
+
+		/* Ignore if somebody else already gave new suggestion */
+		if (dst != tn->trial_addr)
+			return true;
+
+		/* Otherwise update trial address and restart trial period */
+		tn->trial_addr = sugg_addr;
+		msg_set_prevnode(buf_msg(d->skb), sugg_addr);
+		tn->addr_trial_end = jiffies + msecs_to_jiffies(1000);
+		return true;
+	}
+
+	/* Apply trial address if we just left trial period */
+	if (!trial && !self) {
+		tipc_net_finalize(net, tn->trial_addr);
+		msg_set_type(buf_msg(d->skb), DSC_REQ_MSG);
+	}
+
+	if (mtyp != DSC_TRIAL_MSG)
+		return false;
+
+	sugg_addr = tipc_node_try_addr(net, peer_id, src);
+	if (sugg_addr)
+		tipc_disc_msg_xmit(net, DSC_TRIAL_FAIL_MSG, src,
+				   self, sugg_addr, maddr, b);
+	return true;
+}
+
 /**
  * tipc_disc_rcv - handle incoming discovery message (request or response)
  * @net: applicable net namespace
···
 	struct tipc_msg *hdr = buf_msg(skb);
 	u16 caps = msg_node_capabilities(hdr);
 	bool legacy = tn->legacy_addr_format;
+	u32 sugg = msg_sugg_node_addr(hdr);
 	u32 signature = msg_node_sig(hdr);
+	u8 peer_id[NODE_ID_LEN] = {0,};
 	u32 dst = msg_dest_domain(hdr);
 	u32 net_id = msg_bc_netid(hdr);
-	u32 self = tipc_own_addr(net);
 	struct tipc_media_addr maddr;
 	u32 src = msg_prevnode(hdr);
 	u32 mtyp = msg_type(hdr);
 	bool dupl_addr = false;
 	bool respond = false;
+	u32 self;
 	int err;
+
+	skb_linearize(skb);
+	hdr = buf_msg(skb);
+
+	if (caps & TIPC_NODE_ID128)
+		memcpy(peer_id, msg_node_id(hdr), NODE_ID_LEN);
+	else
+		sprintf(peer_id, "%x", src);
 
 	err = b->media->msg2addr(b, &maddr, msg_media_addr(hdr));
 	kfree_skb(skb);
···
 		return;
 	if (net_id != tn->net_id)
 		return;
+	if (tipc_disc_addr_trial_msg(b->disc, &maddr, b, dst,
+				     src, sugg, peer_id, mtyp))
+		return;
+	self = tipc_own_addr(net);
+
+	/* Message from somebody using this node's address */
 	if (in_own_node(net, src)) {
 		disc_dupl_alert(b, self, &maddr);
 		return;
···
 		return;
 	if (!tipc_in_scope(legacy, b->domain, src))
 		return;
-
-	tipc_node_check_dest(net, src, b, caps, signature,
+	tipc_node_check_dest(net, src, peer_id, b, caps, signature,
 			     &maddr, &respond, &dupl_addr);
 	if (dupl_addr)
 		disc_dupl_alert(b, src, &maddr);
···
 		return;
 	if (mtyp != DSC_REQ_MSG)
 		return;
-	tipc_disc_msg_xmit(net, DSC_RESP_MSG, src, self, &maddr, b);
+	tipc_disc_msg_xmit(net, DSC_RESP_MSG, src, self, 0, &maddr, b);
 }
 
 /* tipc_disc_add_dest - increment set of discovered nodes
···
 static void tipc_disc_timeout(struct timer_list *t)
 {
 	struct tipc_discoverer *d = from_timer(d, t, timer);
+	struct tipc_net *tn = tipc_net(d->net);
+	u32 self = tipc_own_addr(d->net);
 	struct tipc_media_addr maddr;
 	struct sk_buff *skb = NULL;
-	struct net *net;
+	struct net *net = d->net;
 	u32 bearer_id;
 
 	spin_lock_bh(&d->lock);
···
 		d->timer_intv = TIPC_DISC_INACTIVE;
 		goto exit;
 	}
+
+	/* Did we just leave the address trial period ? */
+	if (!self && !time_before(jiffies, tn->addr_trial_end)) {
+		self = tn->trial_addr;
+		tipc_net_finalize(net, self);
+		msg_set_prevnode(buf_msg(d->skb), self);
+		msg_set_type(buf_msg(d->skb), DSC_REQ_MSG);
+	}
+
 	/* Adjust timeout interval according to discovery phase */
-	d->timer_intv *= 2;
-	if (d->num_nodes && d->timer_intv > TIPC_DISC_SLOW)
-		d->timer_intv = TIPC_DISC_SLOW;
-	else if (!d->num_nodes && d->timer_intv > TIPC_DISC_FAST)
-		d->timer_intv = TIPC_DISC_FAST;
+	if (time_before(jiffies, tn->addr_trial_end)) {
+		d->timer_intv = TIPC_DISC_INIT;
+	} else {
+		d->timer_intv *= 2;
+		if (d->num_nodes && d->timer_intv > TIPC_DISC_SLOW)
+			d->timer_intv = TIPC_DISC_SLOW;
+		else if (!d->num_nodes && d->timer_intv > TIPC_DISC_FAST)
+			d->timer_intv = TIPC_DISC_FAST;
+	}
+
 	mod_timer(&d->timer, jiffies + d->timer_intv);
 	memcpy(&maddr, &d->dest, sizeof(maddr));
 	skb = skb_clone(d->skb, GFP_ATOMIC);
-	net = d->net;
 	bearer_id = d->bearer_id;
 exit:
 	spin_unlock_bh(&d->lock);
···
 int tipc_disc_create(struct net *net, struct tipc_bearer *b,
 		     struct tipc_media_addr *dest, struct sk_buff **skb)
 {
+	struct tipc_net *tn = tipc_net(net);
 	struct tipc_discoverer *d;
 
 	d = kmalloc(sizeof(*d), GFP_ATOMIC);
 	if (!d)
 		return -ENOMEM;
-	d->skb = tipc_buf_acquire(MAX_H_SIZE, GFP_ATOMIC);
+	d->skb = tipc_buf_acquire(MAX_H_SIZE + NODE_ID_LEN, GFP_ATOMIC);
 	if (!d->skb) {
 		kfree(d);
 		return -ENOMEM;
 	}
-
 	tipc_disc_init_msg(net, d->skb, DSC_REQ_MSG, b);
+
+	/* Do we need an address trial period first ? */
+	if (!tipc_own_addr(net)) {
+		tn->addr_trial_end = jiffies + msecs_to_jiffies(1000);
+		msg_set_type(buf_msg(d->skb), DSC_TRIAL_MSG);
+	}
 	memcpy(&d->dest, dest, sizeof(*dest));
 	d->net = net;
 	d->bearer_id = b->identity;

net/tipc/link.c

···
  */
 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
 		      int tolerance, char net_plane, u32 mtu, int priority,
-		      int window, u32 session, u32 self, u32 peer,
-		      u16 peer_caps,
+		      int window, u32 session, u32 self,
+		      u32 peer, u8 *peer_id, u16 peer_caps,
 		      struct tipc_link *bc_sndlink,
 		      struct tipc_link *bc_rcvlink,
 		      struct sk_buff_head *inputq,
 		      struct sk_buff_head *namedq,
 		      struct tipc_link **link)
 {
-	char *self_str = tipc_own_id_string(net);
+	char peer_str[NODE_ID_STR_LEN] = {0,};
+	char self_str[NODE_ID_STR_LEN] = {0,};
 	struct tipc_link *l;
 
 	l = kzalloc(sizeof(*l), GFP_ATOMIC);
···
 	*link = l;
 	l->session = session;
 
-	/* Note: peer i/f name is completed by reset/activate message */
-	if (strlen(self_str) > 16)
-		sprintf(l->name, "%x:%s-%x:unknown", self, if_name, peer);
-	else
-		sprintf(l->name, "%s:%s-%x:unknown", self_str, if_name, peer);
+	/* Set link name for unicast links only */
+	if (peer_id) {
+		tipc_nodeid2string(self_str, tipc_own_id(net));
+		if (strlen(self_str) > 16)
+			sprintf(self_str, "%x", self);
+		tipc_nodeid2string(peer_str, peer_id);
+		if (strlen(peer_str) > 16)
+			sprintf(peer_str, "%x", peer);
+	}
+	/* Peer i/f name will be completed by reset/activate message */
+	sprintf(l->name, "%s:%s-%s:unknown", self_str, if_name, peer_str);
+
 	strcpy(l->if_name, if_name);
 	l->addr = peer;
 	l->peer_caps = peer_caps;
···
 	struct tipc_link *l;
 
 	if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, window,
-			      0, ownnode, peer, peer_caps, bc_sndlink,
+			      0, ownnode, peer, NULL, peer_caps, bc_sndlink,
 			      NULL, inputq, namedq, link))
 		return false;
 

net/tipc/link.h

···
 
 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
 		      int tolerance, char net_plane, u32 mtu, int priority,
-		      int window, u32 session, u32 ownnode, u32 peer,
-		      u16 peer_caps,
+		      int window, u32 session, u32 ownnode,
+		      u32 peer, u8 *peer_id, u16 peer_caps,
 		      struct tipc_link *bc_sndlink,
 		      struct tipc_link *bc_rcvlink,
 		      struct sk_buff_head *inputq,

net/tipc/msg.h

···
  */
 #define DSC_REQ_MSG		0
 #define DSC_RESP_MSG		1
+#define DSC_TRIAL_MSG		2
+#define DSC_TRIAL_FAIL_MSG	3
 
 /*
  * Group protocol message types
···
 {
 	msg_set_bits(m, 2, 0, 0xffff, n);
 }
-
 
 /*
  * Word 4
···
 static inline bool msg_is_reset(struct tipc_msg *hdr)
 {
 	return (msg_user(hdr) == LINK_PROTOCOL) && (msg_type(hdr) == RESET_MSG);
+}
+
+static inline u32 msg_sugg_node_addr(struct tipc_msg *m)
+{
+	return msg_word(m, 14);
+}
+
+static inline void msg_set_sugg_node_addr(struct tipc_msg *m, u32 n)
+{
+	msg_set_word(m, 14, n);
+}
+
+static inline void msg_set_node_id(struct tipc_msg *hdr, u8 *id)
+{
+	memcpy(msg_data(hdr), id, 16);
+}
+
+static inline u8 *msg_node_id(struct tipc_msg *hdr)
+{
+	return (u8 *)msg_data(hdr);
 }
 
 struct sk_buff *tipc_buf_acquire(u32 size, gfp_t gfp);

net/tipc/net.c

···
 	}
 	pr_info("Started in network mode\n");
 
-	if (node_id) {
+	if (node_id)
 		tipc_set_node_id(net, node_id);
-		tipc_net_finalize(net, tipc_own_addr(net));
-	}
 	if (addr)
 		tipc_net_finalize(net, addr);
 	return 0;

net/tipc/node.c

···
 	u16 capabilities;
 	u32 signature;
 	u32 link_id;
+	u8 peer_id[16];
 	struct list_head publ_list;
 	struct list_head conn_sks;
 	unsigned long keepalive_intv;
···
 static void tipc_node_timeout(struct timer_list *t);
 static void tipc_node_fsm_evt(struct tipc_node *n, int evt);
 static struct tipc_node *tipc_node_find(struct net *net, u32 addr);
+static struct tipc_node *tipc_node_find_by_id(struct net *net, u8 *id);
 static void tipc_node_put(struct tipc_node *node);
 static bool node_is_up(struct tipc_node *n);
 
···
 	return node;
 }
 
+/* tipc_node_find_by_id - locate specified node object by its 128-bit id
+ * Note: this function is called only when a discovery request failed
+ * to find the node by its 32-bit id, and is not time critical
+ */
+static struct tipc_node *tipc_node_find_by_id(struct net *net, u8 *id)
+{
+	struct tipc_net *tn = tipc_net(net);
+	struct tipc_node *n;
+	bool found = false;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(n, &tn->node_list, list) {
+		read_lock_bh(&n->lock);
+		if (!memcmp(id, n->peer_id, 16) &&
+		    kref_get_unless_zero(&n->kref))
+			found = true;
+		read_unlock_bh(&n->lock);
+		if (found)
+			break;
+	}
+	rcu_read_unlock();
+	return found ? n : NULL;
+}
+
 static void tipc_node_read_lock(struct tipc_node *n)
 {
 	read_lock_bh(&n->lock);
···
 	}
 }
 
-struct tipc_node *tipc_node_create(struct net *net, u32 addr, u16 capabilities)
+struct tipc_node *tipc_node_create(struct net *net, u32 addr,
+				   u8 *peer_id, u16 capabilities)
 {
 	struct tipc_net *tn = net_generic(net, tipc_net_id);
 	struct tipc_node *n, *temp_node;
···
 		goto exit;
 	}
 	n->addr = addr;
+	memcpy(&n->peer_id, peer_id, 16);
 	n->net = net;
 	n->capabilities = capabilities;
 	kref_init(&n->kref);
···
 	n->signature = INVALID_NODE_SIG;
 	n->active_links[0] = INVALID_BEARER_ID;
 	n->active_links[1] = INVALID_BEARER_ID;
-	if (!tipc_link_bc_create(net, tipc_own_addr(net), n->addr,
-				 U16_MAX,
+	if (!tipc_link_bc_create(net, tipc_own_addr(net),
+				 addr, U16_MAX,
 				 tipc_link_window(tipc_bc_sndlink(net)),
 				 n->capabilities,
 				 &n->bc_entry.inputq1,
···
 	return retval;
 }
 
-void tipc_node_check_dest(struct net *net, u32 onode,
-			  struct tipc_bearer *b,
+static u32 tipc_node_suggest_addr(struct net *net, u32 addr)
+{
+	struct tipc_node *n;
+
+	addr ^= tipc_net(net)->random;
+	while ((n = tipc_node_find(net, addr))) {
+		tipc_node_put(n);
+		addr++;
+	}
+	return addr;
+}
+
+/* tipc_node_try_addr(): Check if addr can be used by peer, suggest other if not
+ */
+u32 tipc_node_try_addr(struct net *net, u8 *id, u32 addr)
+{
+	struct tipc_net *tn = tipc_net(net);
+	struct tipc_node *n;
+
+	/* Suggest new address if some other peer is using this one */
+	n = tipc_node_find(net, addr);
+	if (n) {
+		if (!memcmp(n->peer_id, id, NODE_ID_LEN))
+			addr = 0;
+		tipc_node_put(n);
+		if (!addr)
+			return 0;
+		return tipc_node_suggest_addr(net, addr);
+	}
+
+	/* Suggest previously used address if peer is known */
+	n = tipc_node_find_by_id(net, id);
+	if (n) {
+		addr = n->addr;
+		tipc_node_put(n);
+	}
+	/* Even this node may be in trial phase */
+	if (tn->trial_addr == addr)
+		return tipc_node_suggest_addr(net, addr);
+
+	return addr;
+}
+
+void tipc_node_check_dest(struct net *net, u32 addr,
+			  u8 *peer_id, struct tipc_bearer *b,
 			  u16 capabilities, u32 signature,
 			  struct tipc_media_addr *maddr,
 			  bool *respond, bool *dupl_addr)
···
 	*dupl_addr = false;
 	*respond = false;
 
-	n = tipc_node_create(net, onode, capabilities);
+	n = tipc_node_create(net, addr, peer_id, capabilities);
 	if (!n)
 		return;
 
···
 		if (!tipc_link_create(net, if_name, b->identity, b->tolerance,
 				      b->net_plane, b->mtu, b->priority,
 				      b->window, mod(tipc_net(net)->random),
-				      tipc_own_addr(net), onode,
+				      tipc_own_addr(net), addr, peer_id,
 				      n->capabilities,
 				      tipc_bc_sndlink(n->net), n->bc_entry.link,
 				      &le->inputq,

net/tipc/node.h

···
 #define INVALID_BEARER_ID -1
 
 void tipc_node_stop(struct net *net);
-void tipc_node_check_dest(struct net *net, u32 onode,
+u32 tipc_node_try_addr(struct net *net, u8 *id, u32 addr);
+void tipc_node_check_dest(struct net *net, u32 onode, u8 *peer_id128,
 			  struct tipc_bearer *bearer,
 			  u16 capabilities, u32 signature,
 			  struct tipc_media_addr *maddr,