Merge branch 'tipc-de-generealize-topology-server'

Jon Maloy says:

====================
tipc: de-generealize topology server

The topology server is partially based on a template that is much
more generic than what we need. This results in a code that is
unnecessarily hard to follow and keeping bug free.

We now take the consequence of the fact that we only have one such
server in TIPC, - with no prospects for introducing any more, and
adapt the code to the specialized task is really is doing.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

net/tipc/Makefile

@@ -9,7 +9,7 @@ tipc-y	+= addr.o bcast.o bearer.o \
 	   core.o link.o discover.o msg.o  \
 	   name_distr.o  subscr.o monitor.o name_table.o net.o  \
 	   netlink.o netlink_compat.o node.o socket.o eth_media.o \
-	   server.o socket.o group.o
+	   topsrv.o socket.o group.o
 
 tipc-$(CONFIG_TIPC_MEDIA_UDP)	+= udp_media.o
 tipc-$(CONFIG_TIPC_MEDIA_IB)	+= ib_media.o

net/tipc/core.h

@@ -64,7 +64,7 @@ struct tipc_bearer;
 struct tipc_bc_base;
 struct tipc_link;
 struct tipc_name_table;
-struct tipc_server;
+struct tipc_topsrv;
 struct tipc_monitor;
 
 #define TIPC_MOD_VER "2.0.0"
@@ -112,7 +112,7 @@ struct tipc_net {
 	struct list_head dist_queue;
 
 	/* Topology subscription server */
-	struct tipc_server *topsrv;
+	struct tipc_topsrv *topsrv;
 	atomic_t subscription_count;
 };
 
@@ -131,7 +131,7 @@ static inline struct list_head *tipc_nodes(struct net *net)
 	return &tipc_net(net)->node_list;
 }
 
-static inline struct tipc_server *tipc_topsrv(struct net *net)
+static inline struct tipc_topsrv *tipc_topsrv(struct net *net)
 {
 	return tipc_net(net)->topsrv;
 }

net/tipc/group.c

@@ -37,7 +37,7 @@
 #include "addr.h"
 #include "group.h"
 #include "bcast.h"
-#include "server.h"
+#include "topsrv.h"
 #include "msg.h"
 #include "socket.h"
 #include "node.h"

net/tipc/name_table.c

@@ -326,10 +326,10 @@ static struct publication *tipc_nameseq_insert_publ(struct net *net,
 
 	/* Any subscriptions waiting for notification?  */
 	list_for_each_entry_safe(s, st, &nseq->subscriptions, nameseq_list) {
-		tipc_subscrp_report_overlap(s, publ->lower, publ->upper,
-					    TIPC_PUBLISHED, publ->ref,
-					    publ->node, publ->scope,
-					    created_subseq);
+		tipc_sub_report_overlap(s, publ->lower, publ->upper,
+					TIPC_PUBLISHED, publ->ref,
+					publ->node, publ->scope,
+					created_subseq);
 	}
 	return publ;
 }
@@ -397,10 +397,9 @@ static struct publication *tipc_nameseq_remove_publ(struct net *net,
 
 	/* Notify any waiting subscriptions */
 	list_for_each_entry_safe(s, st, &nseq->subscriptions, nameseq_list) {
-		tipc_subscrp_report_overlap(s, publ->lower, publ->upper,
-					    TIPC_WITHDRAWN, publ->ref,
-					    publ->node, publ->scope,
-					    removed_subseq);
+		tipc_sub_report_overlap(s, publ->lower, publ->upper,
+					TIPC_WITHDRAWN, publ->ref, publ->node,
+					publ->scope, removed_subseq);
 	}
 
 	return publ;
@@ -412,33 +411,37 @@ static struct publication *tipc_nameseq_remove_publ(struct net *net,
  * sequence overlapping with the requested sequence
  */
 static void tipc_nameseq_subscribe(struct name_seq *nseq,
-				   struct tipc_subscription *s,
-				   bool status)
+				   struct tipc_subscription *sub)
 {
 	struct sub_seq *sseq = nseq->sseqs;
 	struct tipc_name_seq ns;
+	struct tipc_subscr *s = &sub->evt.s;
+	bool no_status;
 
-	tipc_subscrp_convert_seq(&s->evt.s.seq, s->swap, &ns);
+	ns.type = tipc_sub_read(s, seq.type);
+	ns.lower = tipc_sub_read(s, seq.lower);
+	ns.upper = tipc_sub_read(s, seq.upper);
+	no_status = tipc_sub_read(s, filter) & TIPC_SUB_NO_STATUS;
 
-	tipc_subscrp_get(s);
-	list_add(&s->nameseq_list, &nseq->subscriptions);
+	tipc_sub_get(sub);
+	list_add(&sub->nameseq_list, &nseq->subscriptions);
 
-	if (!status || !sseq)
+	if (no_status || !sseq)
 		return;
 
 	while (sseq != &nseq->sseqs[nseq->first_free]) {
-		if (tipc_subscrp_check_overlap(&ns, sseq->lower, sseq->upper)) {
+		if (tipc_sub_check_overlap(&ns, sseq->lower, sseq->upper)) {
 			struct publication *crs;
 			struct name_info *info = sseq->info;
 			int must_report = 1;
 
 			list_for_each_entry(crs, &info->zone_list, zone_list) {
-				tipc_subscrp_report_overlap(s, sseq->lower,
-							    sseq->upper,
-							    TIPC_PUBLISHED,
-							    crs->ref, crs->node,
-							    crs->scope,
-							    must_report);
+				tipc_sub_report_overlap(sub, sseq->lower,
+							sseq->upper,
+							TIPC_PUBLISHED,
+							crs->ref, crs->node,
+							crs->scope,
+							must_report);
 				must_report = 0;
 			}
 		}
@@ -808,24 +811,27 @@ int tipc_nametbl_withdraw(struct net *net, u32 type, u32 lower, u32 ref,
 /**
  * tipc_nametbl_subscribe - add a subscription object to the name table
  */
-void tipc_nametbl_subscribe(struct tipc_subscription *s, bool status)
+void tipc_nametbl_subscribe(struct tipc_subscription *sub)
 {
-	struct tipc_net *tn = net_generic(s->net, tipc_net_id);
-	u32 type = tipc_subscrp_convert_seq_type(s->evt.s.seq.type, s->swap);
+	struct tipc_net *tn = tipc_net(sub->net);
+	struct tipc_subscr *s = &sub->evt.s;
+	u32 type = tipc_sub_read(s, seq.type);
 	int index = hash(type);
 	struct name_seq *seq;
 	struct tipc_name_seq ns;
 
 	spin_lock_bh(&tn->nametbl_lock);
-	seq = nametbl_find_seq(s->net, type);
+	seq = nametbl_find_seq(sub->net, type);
 	if (!seq)
 		seq = tipc_nameseq_create(type, &tn->nametbl->seq_hlist[index]);
 	if (seq) {
 		spin_lock_bh(&seq->lock);
-		tipc_nameseq_subscribe(seq, s, status);
+		tipc_nameseq_subscribe(seq, sub);
 		spin_unlock_bh(&seq->lock);
 	} else {
-		tipc_subscrp_convert_seq(&s->evt.s.seq, s->swap, &ns);
+		ns.type = tipc_sub_read(s, seq.type);
+		ns.lower = tipc_sub_read(s, seq.lower);
+		ns.upper = tipc_sub_read(s, seq.upper);
 		pr_warn("Failed to create subscription for {%u,%u,%u}\n",
 			ns.type, ns.lower, ns.upper);
 	}
@@ -835,18 +841,19 @@ void tipc_nametbl_subscribe(struct tipc_subscription *s, bool status)
 /**
  * tipc_nametbl_unsubscribe - remove a subscription object from name table
  */
-void tipc_nametbl_unsubscribe(struct tipc_subscription *s)
+void tipc_nametbl_unsubscribe(struct tipc_subscription *sub)
 {
-	struct tipc_net *tn = net_generic(s->net, tipc_net_id);
+	struct tipc_subscr *s = &sub->evt.s;
+	struct tipc_net *tn = tipc_net(sub->net);
 	struct name_seq *seq;
-	u32 type = tipc_subscrp_convert_seq_type(s->evt.s.seq.type, s->swap);
+	u32 type = tipc_sub_read(s, seq.type);
 
 	spin_lock_bh(&tn->nametbl_lock);
-	seq = nametbl_find_seq(s->net, type);
+	seq = nametbl_find_seq(sub->net, type);
 	if (seq != NULL) {
 		spin_lock_bh(&seq->lock);
-		list_del_init(&s->nameseq_list);
-		tipc_subscrp_put(s);
+		list_del_init(&sub->nameseq_list);
+		tipc_sub_put(sub);
 		if (!seq->first_free && list_empty(&seq->subscriptions)) {
 			hlist_del_init_rcu(&seq->ns_list);
 			kfree(seq->sseqs);

net/tipc/name_table.h

@@ -120,7 +120,7 @@ struct publication *tipc_nametbl_insert_publ(struct net *net, u32 type,
 struct publication *tipc_nametbl_remove_publ(struct net *net, u32 type,
 					     u32 lower, u32 node, u32 ref,
 					     u32 key);
-void tipc_nametbl_subscribe(struct tipc_subscription *s, bool status);
+void tipc_nametbl_subscribe(struct tipc_subscription *s);
 void tipc_nametbl_unsubscribe(struct tipc_subscription *s);
 int tipc_nametbl_init(struct net *net);
 void tipc_nametbl_stop(struct net *net);

net/tipc/subscr.c

 /*
  * net/tipc/subscr.c: TIPC network topology service
  *
- * Copyright (c) 2000-2006, Ericsson AB
+ * Copyright (c) 2000-2017, Ericsson AB
  * Copyright (c) 2005-2007, 2010-2013, Wind River Systems
  * All rights reserved.
  *
@@ -38,61 +38,30 @@
 #include "name_table.h"
 #include "subscr.h"
 
-/**
- * struct tipc_subscriber - TIPC network topology subscriber
- * @kref: reference counter to tipc_subscription object
- * @conid: connection identifier to server connecting to subscriber
- * @lock: control access to subscriber
- * @subscrp_list: list of subscription objects for this subscriber
- */
-struct tipc_subscriber {
-	struct kref kref;
-	int conid;
-	spinlock_t lock;
-	struct list_head subscrp_list;
-};
-
-static void tipc_subscrb_put(struct tipc_subscriber *subscriber);
-
-/**
- * htohl - convert value to endianness used by destination
- * @in: value to convert
- * @swap: non-zero if endianness must be reversed
- *
- * Returns converted value
- */
-static u32 htohl(u32 in, int swap)
-{
-	return swap ? swab32(in) : in;
-}
-
-static void tipc_subscrp_send_event(struct tipc_subscription *sub,
-				    u32 found_lower, u32 found_upper,
-				    u32 event, u32 port_ref, u32 node)
+static void tipc_sub_send_event(struct tipc_subscription *sub,
+				u32 found_lower, u32 found_upper,
+				u32 event, u32 port, u32 node)
 {
-	struct tipc_net *tn = net_generic(sub->net, tipc_net_id);
-	struct tipc_subscriber *subscriber = sub->subscriber;
-	struct kvec msg_sect;
+	struct tipc_event *evt = &sub->evt;
 
-	msg_sect.iov_base = (void *)&sub->evt;
-	msg_sect.iov_len = sizeof(struct tipc_event);
-	sub->evt.event = htohl(event, sub->swap);
-	sub->evt.found_lower = htohl(found_lower, sub->swap);
-	sub->evt.found_upper = htohl(found_upper, sub->swap);
-	sub->evt.port.ref = htohl(port_ref, sub->swap);
-	sub->evt.port.node = htohl(node, sub->swap);
-	tipc_conn_sendmsg(tn->topsrv, subscriber->conid, NULL,
-			  msg_sect.iov_base, msg_sect.iov_len);
+	if (sub->inactive)
+		return;
+	tipc_evt_write(evt, event, event);
+	tipc_evt_write(evt, found_lower, found_lower);
+	tipc_evt_write(evt, found_upper, found_upper);
+	tipc_evt_write(evt, port.ref, port);
+	tipc_evt_write(evt, port.node, node);
+	tipc_topsrv_queue_evt(sub->net, sub->conid, event, evt);
 }
 
 /**
- * tipc_subscrp_check_overlap - test for subscription overlap with the
+ * tipc_sub_check_overlap - test for subscription overlap with the
  * given values
  *
  * Returns 1 if there is overlap, otherwise 0.
  */
-int tipc_subscrp_check_overlap(struct tipc_name_seq *seq, u32 found_lower,
-			       u32 found_upper)
+int tipc_sub_check_overlap(struct tipc_name_seq *seq, u32 found_lower,
+			   u32 found_upper)
 {
 	if (found_lower < seq->lower)
 		found_lower = seq->lower;
@@ -103,298 +72,98 @@ int tipc_subscrp_check_overlap(struct tipc_name_seq *seq, u32 found_lower,
 	return 1;
 }
 
-u32 tipc_subscrp_convert_seq_type(u32 type, int swap)
-{
-	return htohl(type, swap);
-}
-
-void tipc_subscrp_convert_seq(struct tipc_name_seq *in, int swap,
-			      struct tipc_name_seq *out)
-{
-	out->type = htohl(in->type, swap);
-	out->lower = htohl(in->lower, swap);
-	out->upper = htohl(in->upper, swap);
-}
-
-void tipc_subscrp_report_overlap(struct tipc_subscription *sub, u32 found_lower,
-				 u32 found_upper, u32 event, u32 port_ref,
-				 u32 node, u32 scope, int must)
+void tipc_sub_report_overlap(struct tipc_subscription *sub,
+			     u32 found_lower, u32 found_upper,
+			     u32 event, u32 port, u32 node,
+			     u32 scope, int must)
 {
-	u32 filter = htohl(sub->evt.s.filter, sub->swap);
+	struct tipc_subscr *s = &sub->evt.s;
+	u32 filter = tipc_sub_read(s, filter);
 	struct tipc_name_seq seq;
 
-	tipc_subscrp_convert_seq(&sub->evt.s.seq, sub->swap, &seq);
-	if (!tipc_subscrp_check_overlap(&seq, found_lower, found_upper))
+	seq.type = tipc_sub_read(s, seq.type);
+	seq.lower = tipc_sub_read(s, seq.lower);
+	seq.upper = tipc_sub_read(s, seq.upper);
+
+	if (!tipc_sub_check_overlap(&seq, found_lower, found_upper))
 		return;
+
 	if (!must && !(filter & TIPC_SUB_PORTS))
 		return;
 	if (filter & TIPC_SUB_CLUSTER_SCOPE && scope == TIPC_NODE_SCOPE)
 		return;
 	if (filter & TIPC_SUB_NODE_SCOPE && scope != TIPC_NODE_SCOPE)
 		return;
-
-	tipc_subscrp_send_event(sub, found_lower, found_upper, event, port_ref,
-				node);
+	spin_lock(&sub->lock);
+	tipc_sub_send_event(sub, found_lower, found_upper,
+			    event, port, node);
+	spin_unlock(&sub->lock);
 }
 
-static void tipc_subscrp_timeout(struct timer_list *t)
+static void tipc_sub_timeout(struct timer_list *t)
 {
 	struct tipc_subscription *sub = from_timer(sub, t, timer);
-	struct tipc_subscriber *subscriber = sub->subscriber;
-
-	spin_lock_bh(&subscriber->lock);
-	tipc_nametbl_unsubscribe(sub);
-	list_del(&sub->subscrp_list);
-	spin_unlock_bh(&subscriber->lock);
-
-	/* Notify subscriber of timeout */
-	tipc_subscrp_send_event(sub, sub->evt.s.seq.lower, sub->evt.s.seq.upper,
-				TIPC_SUBSCR_TIMEOUT, 0, 0);
-
-	tipc_subscrp_put(sub);
-}
-
-static void tipc_subscrb_kref_release(struct kref *kref)
-{
-	kfree(container_of(kref,struct tipc_subscriber, kref));
-}
-
-static void tipc_subscrb_put(struct tipc_subscriber *subscriber)
-{
-	kref_put(&subscriber->kref, tipc_subscrb_kref_release);
-}
+	struct tipc_subscr *s = &sub->evt.s;
 
-static void tipc_subscrb_get(struct tipc_subscriber *subscriber)
-{
-	kref_get(&subscriber->kref);
+	spin_lock(&sub->lock);
+	tipc_sub_send_event(sub, s->seq.lower, s->seq.upper,
+			    TIPC_SUBSCR_TIMEOUT, 0, 0);
+	sub->inactive = true;
+	spin_unlock(&sub->lock);
 }
 
-static void tipc_subscrp_kref_release(struct kref *kref)
+static void tipc_sub_kref_release(struct kref *kref)
 {
-	struct tipc_subscription *sub = container_of(kref,
-						     struct tipc_subscription,
-						     kref);
-	struct tipc_net *tn = net_generic(sub->net, tipc_net_id);
-	struct tipc_subscriber *subscriber = sub->subscriber;
-
-	atomic_dec(&tn->subscription_count);
-	kfree(sub);
-	tipc_subscrb_put(subscriber);
+	kfree(container_of(kref, struct tipc_subscription, kref));
 }
 
-void tipc_subscrp_put(struct tipc_subscription *subscription)
+void tipc_sub_put(struct tipc_subscription *subscription)
 {
-	kref_put(&subscription->kref, tipc_subscrp_kref_release);
+	kref_put(&subscription->kref, tipc_sub_kref_release);
 }
 
-void tipc_subscrp_get(struct tipc_subscription *subscription)
+void tipc_sub_get(struct tipc_subscription *subscription)
 {
 	kref_get(&subscription->kref);
 }
 
-/* tipc_subscrb_subscrp_delete - delete a specific subscription or all
- * subscriptions for a given subscriber.
- */
-static void tipc_subscrb_subscrp_delete(struct tipc_subscriber *subscriber,
-					struct tipc_subscr *s)
-{
-	struct list_head *subscription_list = &subscriber->subscrp_list;
-	struct tipc_subscription *sub, *temp;
-	u32 timeout;
-
-	spin_lock_bh(&subscriber->lock);
-	list_for_each_entry_safe(sub, temp, subscription_list,  subscrp_list) {
-		if (s && memcmp(s, &sub->evt.s, sizeof(struct tipc_subscr)))
-			continue;
-
-		timeout = htohl(sub->evt.s.timeout, sub->swap);
-		if (timeout == TIPC_WAIT_FOREVER || del_timer(&sub->timer)) {
-			tipc_nametbl_unsubscribe(sub);
-			list_del(&sub->subscrp_list);
-			tipc_subscrp_put(sub);
-		}
-
-		if (s)
-			break;
-	}
-	spin_unlock_bh(&subscriber->lock);
-}
-
-static struct tipc_subscriber *tipc_subscrb_create(int conid)
-{
-	struct tipc_subscriber *subscriber;
-
-	subscriber = kzalloc(sizeof(*subscriber), GFP_ATOMIC);
-	if (!subscriber) {
-		pr_warn("Subscriber rejected, no memory\n");
-		return NULL;
-	}
-	INIT_LIST_HEAD(&subscriber->subscrp_list);
-	kref_init(&subscriber->kref);
-	subscriber->conid = conid;
-	spin_lock_init(&subscriber->lock);
-
-	return subscriber;
-}
-
-static void tipc_subscrb_delete(struct tipc_subscriber *subscriber)
-{
-	tipc_subscrb_subscrp_delete(subscriber, NULL);
-	tipc_subscrb_put(subscriber);
-}
-
-static void tipc_subscrp_cancel(struct tipc_subscr *s,
-				struct tipc_subscriber *subscriber)
-{
-	tipc_subscrb_get(subscriber);
-	tipc_subscrb_subscrp_delete(subscriber, s);
-	tipc_subscrb_put(subscriber);
-}
-
-static struct tipc_subscription *tipc_subscrp_create(struct net *net,
-						     struct tipc_subscr *s,
-						     int swap)
+struct tipc_subscription *tipc_sub_subscribe(struct net *net,
+					     struct tipc_subscr *s,
+					     int conid)
 {
-	struct tipc_net *tn = net_generic(net, tipc_net_id);
+	u32 filter = tipc_sub_read(s, filter);
 	struct tipc_subscription *sub;
-	u32 filter = htohl(s->filter, swap);
+	u32 timeout;
 
-	/* Refuse subscription if global limit exceeded */
-	if (atomic_read(&tn->subscription_count) >= TIPC_MAX_SUBSCRIPTIONS) {
-		pr_warn("Subscription rejected, limit reached (%u)\n",
-			TIPC_MAX_SUBSCRIPTIONS);
+	if ((filter & TIPC_SUB_PORTS && filter & TIPC_SUB_SERVICE) ||
+	    (tipc_sub_read(s, seq.lower) > tipc_sub_read(s, seq.upper))) {
+		pr_warn("Subscription rejected, illegal request\n");
 		return NULL;
 	}
-
-	/* Allocate subscription object */
 	sub = kmalloc(sizeof(*sub), GFP_ATOMIC);
 	if (!sub) {
 		pr_warn("Subscription rejected, no memory\n");
 		return NULL;
 	}
-
-	/* Initialize subscription object */
 	sub->net = net;
-	if (((filter & TIPC_SUB_PORTS) && (filter & TIPC_SUB_SERVICE)) ||
-	    (htohl(s->seq.lower, swap) > htohl(s->seq.upper, swap))) {
-		pr_warn("Subscription rejected, illegal request\n");
-		kfree(sub);
-		return NULL;
-	}
-
-	sub->swap = swap;
+	sub->conid = conid;
+	sub->inactive = false;
 	memcpy(&sub->evt.s, s, sizeof(*s));
-	atomic_inc(&tn->subscription_count);
+	spin_lock_init(&sub->lock);
 	kref_init(&sub->kref);
-	return sub;
-}
-
-static int tipc_subscrp_subscribe(struct net *net, struct tipc_subscr *s,
-				  struct tipc_subscriber *subscriber, int swap,
-				  bool status)
-{
-	struct tipc_subscription *sub = NULL;
-	u32 timeout;
-
-	sub = tipc_subscrp_create(net, s, swap);
-	if (!sub)
-		return -1;
-
-	spin_lock_bh(&subscriber->lock);
-	list_add(&sub->subscrp_list, &subscriber->subscrp_list);
-	sub->subscriber = subscriber;
-	tipc_nametbl_subscribe(sub, status);
-	tipc_subscrb_get(subscriber);
-	spin_unlock_bh(&subscriber->lock);
-
-	timer_setup(&sub->timer, tipc_subscrp_timeout, 0);
-	timeout = htohl(sub->evt.s.timeout, swap);
-
+	tipc_nametbl_subscribe(sub);
+	timer_setup(&sub->timer, tipc_sub_timeout, 0);
+	timeout = tipc_sub_read(&sub->evt.s, timeout);
 	if (timeout != TIPC_WAIT_FOREVER)
 		mod_timer(&sub->timer, jiffies + msecs_to_jiffies(timeout));
-	return 0;
-}
-
-/* Handle one termination request for the subscriber */
-static void tipc_subscrb_release_cb(int conid, void *usr_data)
-{
-	tipc_subscrb_delete((struct tipc_subscriber *)usr_data);
-}
-
-/* Handle one request to create a new subscription for the subscriber */
-static int tipc_subscrb_rcv_cb(struct net *net, int conid,
-			       struct sockaddr_tipc *addr, void *usr_data,
-			       void *buf, size_t len)
-{
-	struct tipc_subscriber *subscriber = usr_data;
-	struct tipc_subscr *s = (struct tipc_subscr *)buf;
-	bool status;
-	int swap;
-
-	/* Determine subscriber's endianness */
-	swap = !(s->filter & (TIPC_SUB_PORTS | TIPC_SUB_SERVICE |
-			      TIPC_SUB_CANCEL));
-
-	/* Detect & process a subscription cancellation request */
-	if (s->filter & htohl(TIPC_SUB_CANCEL, swap)) {
-		s->filter &= ~htohl(TIPC_SUB_CANCEL, swap);
-		tipc_subscrp_cancel(s, subscriber);
-		return 0;
-	}
-	status = !(s->filter & htohl(TIPC_SUB_NO_STATUS, swap));
-	return tipc_subscrp_subscribe(net, s, subscriber, swap, status);
-}
-
-/* Handle one request to establish a new subscriber */
-static void *tipc_subscrb_connect_cb(int conid)
-{
-	return (void *)tipc_subscrb_create(conid);
-}
-
-int tipc_topsrv_start(struct net *net)
-{
-	struct tipc_net *tn = net_generic(net, tipc_net_id);
-	const char name[] = "topology_server";
-	struct tipc_server *topsrv;
-	struct sockaddr_tipc *saddr;
-
-	saddr = kzalloc(sizeof(*saddr), GFP_ATOMIC);
-	if (!saddr)
-		return -ENOMEM;
-	saddr->family			= AF_TIPC;
-	saddr->addrtype			= TIPC_ADDR_NAMESEQ;
-	saddr->addr.nameseq.type	= TIPC_TOP_SRV;
-	saddr->addr.nameseq.lower	= TIPC_TOP_SRV;
-	saddr->addr.nameseq.upper	= TIPC_TOP_SRV;
-	saddr->scope			= TIPC_NODE_SCOPE;
-
-	topsrv = kzalloc(sizeof(*topsrv), GFP_ATOMIC);
-	if (!topsrv) {
-		kfree(saddr);
-		return -ENOMEM;
-	}
-	topsrv->net			= net;
-	topsrv->saddr			= saddr;
-	topsrv->imp			= TIPC_CRITICAL_IMPORTANCE;
-	topsrv->type			= SOCK_SEQPACKET;
-	topsrv->max_rcvbuf_size		= sizeof(struct tipc_subscr);
-	topsrv->tipc_conn_recvmsg	= tipc_subscrb_rcv_cb;
-	topsrv->tipc_conn_new		= tipc_subscrb_connect_cb;
-	topsrv->tipc_conn_release	= tipc_subscrb_release_cb;
-
-	strncpy(topsrv->name, name, strlen(name) + 1);
-	tn->topsrv = topsrv;
-	atomic_set(&tn->subscription_count, 0);
-
-	return tipc_server_start(topsrv);
+	return sub;
 }
 
-void tipc_topsrv_stop(struct net *net)
+void tipc_sub_unsubscribe(struct tipc_subscription *sub)
 {
-	struct tipc_net *tn = net_generic(net, tipc_net_id);
-	struct tipc_server *topsrv = tn->topsrv;
-
-	tipc_server_stop(topsrv);
-	kfree(topsrv->saddr);
-	kfree(topsrv);
+	tipc_nametbl_unsubscribe(sub);
+	if (sub->evt.s.timeout != TIPC_WAIT_FOREVER)
+		del_timer_sync(&sub->timer);
+	list_del(&sub->sub_list);
+	tipc_sub_put(sub);
 }

net/tipc/subscr.h

 /*
  * net/tipc/subscr.h: Include file for TIPC network topology service
  *
- * Copyright (c) 2003-2006, Ericsson AB
+ * Copyright (c) 2003-2017, Ericsson AB
  * Copyright (c) 2005-2007, 2012-2013, Wind River Systems
  * All rights reserved.
  *
@@ -37,48 +37,72 @@
 #ifndef _TIPC_SUBSCR_H
 #define _TIPC_SUBSCR_H
 
-#include "server.h"
+#include "topsrv.h"
 
-#define TIPC_MAX_SUBSCRIPTIONS	65535
-#define TIPC_MAX_PUBLICATIONS	65535
+#define TIPC_MAX_SUBSCR         65535
+#define TIPC_MAX_PUBLICATIONS   65535
 
 struct tipc_subscription;
-struct tipc_subscriber;
+struct tipc_conn;
 
 /**
  * struct tipc_subscription - TIPC network topology subscription object
  * @subscriber: pointer to its subscriber
  * @seq: name sequence associated with subscription
- * @net: point to network namespace
  * @timer: timer governing subscription duration (optional)
  * @nameseq_list: adjacent subscriptions in name sequence's subscription list
- * @subscrp_list: adjacent subscriptions in subscriber's subscription list
- * @swap: indicates if subscriber uses opposite endianness in its messages
+ * @sub_list: adjacent subscriptions in subscriber's subscription list
  * @evt: template for events generated by subscription
  */
 struct tipc_subscription {
 	struct kref kref;
-	struct tipc_subscriber *subscriber;
 	struct net *net;
 	struct timer_list timer;
 	struct list_head nameseq_list;
-	struct list_head subscrp_list;
-	int swap;
+	struct list_head sub_list;
 	struct tipc_event evt;
+	int conid;
+	bool inactive;
+	spinlock_t lock; /* serialize up/down and timer events */
 };
 
-int tipc_subscrp_check_overlap(struct tipc_name_seq *seq, u32 found_lower,
-			       u32 found_upper);
-void tipc_subscrp_report_overlap(struct tipc_subscription *sub,
-				 u32 found_lower, u32 found_upper, u32 event,
-				 u32 port_ref, u32 node, u32 scope, int must);
-void tipc_subscrp_convert_seq(struct tipc_name_seq *in, int swap,
-			      struct tipc_name_seq *out);
-u32 tipc_subscrp_convert_seq_type(u32 type, int swap);
+struct tipc_subscription *tipc_sub_subscribe(struct net *net,
+					     struct tipc_subscr *s,
+					     int conid);
+void tipc_sub_unsubscribe(struct tipc_subscription *sub);
+
+int tipc_sub_check_overlap(struct tipc_name_seq *seq, u32 found_lower,
+			   u32 found_upper);
+void tipc_sub_report_overlap(struct tipc_subscription *sub,
+			     u32 found_lower, u32 found_upper,
+			     u32 event, u32 port, u32 node,
+			     u32 scope, int must);
 int tipc_topsrv_start(struct net *net);
 void tipc_topsrv_stop(struct net *net);
 
-void tipc_subscrp_put(struct tipc_subscription *subscription);
-void tipc_subscrp_get(struct tipc_subscription *subscription);
+void tipc_sub_put(struct tipc_subscription *subscription);
+void tipc_sub_get(struct tipc_subscription *subscription);
+
+#define TIPC_FILTER_MASK (TIPC_SUB_PORTS | TIPC_SUB_SERVICE | TIPC_SUB_CANCEL)
+
+/* tipc_sub_read - return field_ of struct sub_ in host endian format
+ */
+#define tipc_sub_read(sub_, field_)					\
+	({								\
+		struct tipc_subscr *sub__ = sub_;			\
+		u32 val__ = (sub__)->field_;				\
+		int swap_ = !((sub__)->filter & TIPC_FILTER_MASK);	\
+		(swap_ ? swab32(val__) : val__);			\
+	})
+
+/* tipc_evt_write - write val_ to field_ of struct evt_ in user endian format
+ */
+#define tipc_evt_write(evt_, field_, val_)				\
+	({								\
+		struct tipc_event *evt__ = evt_;			\
+		u32 val__ = val_;					\
+		int swap_ = !((evt__)->s.filter & (TIPC_FILTER_MASK));	\
+		(evt__)->field_ = swap_ ? swab32(val__) : val__;	\
+	})
 
 #endif

net/tipc/server.c → net/tipc/topsrv.c

@@ -2,6 +2,7 @@
  * net/tipc/server.c: TIPC server infrastructure
  *
  * Copyright (c) 2012-2013, Wind River Systems
+ * Copyright (c) 2017-2018, Ericsson AB
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -33,7 +34,8 @@
  * POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include "server.h"
+#include "subscr.h"
+#include "topsrv.h"
 #include "core.h"
 #include "socket.h"
 #include "addr.h"
@@ -47,7 +49,37 @@
 #define CF_CONNECTED		1
 #define CF_SERVER		2
 
-#define sock2con(x) ((struct tipc_conn *)(x)->sk_user_data)
+#define TIPC_SERVER_NAME_LEN	32
+
+/**
+ * struct tipc_topsrv - TIPC server structure
+ * @conn_idr: identifier set of connection
+ * @idr_lock: protect the connection identifier set
+ * @idr_in_use: amount of allocated identifier entry
+ * @net: network namspace instance
+ * @rcvbuf_cache: memory cache of server receive buffer
+ * @rcv_wq: receive workqueue
+ * @send_wq: send workqueue
+ * @max_rcvbuf_size: maximum permitted receive message length
+ * @tipc_conn_new: callback will be called when new connection is incoming
+ * @tipc_conn_release: callback will be called before releasing the connection
+ * @tipc_conn_recvmsg: callback will be called when message arrives
+ * @name: server name
+ * @imp: message importance
+ * @type: socket type
+ */
+struct tipc_topsrv {
+	struct idr conn_idr;
+	spinlock_t idr_lock; /* for idr list */
+	int idr_in_use;
+	struct net *net;
+	struct work_struct awork;
+	struct workqueue_struct *rcv_wq;
+	struct workqueue_struct *send_wq;
+	int max_rcvbuf_size;
+	struct socket *listener;
+	char name[TIPC_SERVER_NAME_LEN];
+};
 
 /**
  * struct tipc_conn - TIPC connection structure
@@ -56,12 +88,13 @@
  * @sock: socket handler associated with connection
  * @flags: indicates connection state
  * @server: pointer to connected server
+ * @sub_list: lsit to all pertaing subscriptions
+ * @sub_lock: lock protecting the subscription list
+ * @outqueue_lock: control access to the outqueue
  * @rwork: receive work item
- * @usr_data: user-specified field
  * @rx_action: what to do when connection socket is active
  * @outqueue: pointer to first outbound message in queue
  * @outqueue_lock: control access to the outqueue
- * @outqueue: list of connection objects for its server
  * @swork: send work item
  */
 struct tipc_conn {
@@ -69,50 +102,51 @@ struct tipc_conn {
 	int conid;
 	struct socket *sock;
 	unsigned long flags;
-	struct tipc_server *server;
+	struct tipc_topsrv *server;
+	struct list_head sub_list;
+	spinlock_t sub_lock; /* for subscription list */
 	struct work_struct rwork;
-	int (*rx_action) (struct tipc_conn *con);
-	void *usr_data;
 	struct list_head outqueue;
-	spinlock_t outqueue_lock;
+	spinlock_t outqueue_lock; /* for outqueue */
 	struct work_struct swork;
 };
 
 /* An entry waiting to be sent */
 struct outqueue_entry {
+	bool inactive;
+	struct tipc_event evt;
 	struct list_head list;
-	struct kvec iov;
-	struct sockaddr_tipc dest;
 };
 
-static void tipc_recv_work(struct work_struct *work);
-static void tipc_send_work(struct work_struct *work);
-static void tipc_clean_outqueues(struct tipc_conn *con);
+static void tipc_conn_recv_work(struct work_struct *work);
+static void tipc_conn_send_work(struct work_struct *work);
+static void tipc_topsrv_kern_evt(struct net *net, struct tipc_event *evt);
+static void tipc_conn_delete_sub(struct tipc_conn *con, struct tipc_subscr *s);
+
+static bool connected(struct tipc_conn *con)
+{
+	return con && test_bit(CF_CONNECTED, &con->flags);
+}
 
 static void tipc_conn_kref_release(struct kref *kref)
 {
 	struct tipc_conn *con = container_of(kref, struct tipc_conn, kref);
-	struct tipc_server *s = con->server;
-	struct sockaddr_tipc *saddr = s->saddr;
-	struct socket *sock = con->sock;
-	struct sock *sk;
+	struct tipc_topsrv *s = con->server;
+	struct outqueue_entry *e, *safe;
 
-	if (sock) {
-		sk = sock->sk;
-		if (test_bit(CF_SERVER, &con->flags)) {
-			__module_get(sock->ops->owner);
-			__module_get(sk->sk_prot_creator->owner);
-		}
-		saddr->scope = -TIPC_NODE_SCOPE;
-		kernel_bind(sock, (struct sockaddr *)saddr, sizeof(*saddr));
-		sock_release(sock);
-		con->sock = NULL;
-	}
 	spin_lock_bh(&s->idr_lock);
 	idr_remove(&s->conn_idr, con->conid);
 	s->idr_in_use--;
 	spin_unlock_bh(&s->idr_lock);
-	tipc_clean_outqueues(con);
+	if (con->sock)
+		sock_release(con->sock);
+
+	spin_lock_bh(&con->outqueue_lock);
+	list_for_each_entry_safe(e, safe, &con->outqueue, list) {
+		list_del(&e->list);
+		kfree(e);
+	}
+	spin_unlock_bh(&con->outqueue_lock);
 	kfree(con);
 }
 
@@ -126,76 +160,17 @@ static void conn_get(struct tipc_conn *con)
 	kref_get(&con->kref);
 }
 
-static struct tipc_conn *tipc_conn_lookup(struct tipc_server *s, int conid)
-{
-	struct tipc_conn *con;
-
-	spin_lock_bh(&s->idr_lock);
-	con = idr_find(&s->conn_idr, conid);
-	if (con) {
-		if (!test_bit(CF_CONNECTED, &con->flags) ||
-		    !kref_get_unless_zero(&con->kref))
-			con = NULL;
-	}
-	spin_unlock_bh(&s->idr_lock);
-	return con;
-}
-
-static void sock_data_ready(struct sock *sk)
-{
-	struct tipc_conn *con;
-
-	read_lock_bh(&sk->sk_callback_lock);
-	con = sock2con(sk);
-	if (con && test_bit(CF_CONNECTED, &con->flags)) {
-		conn_get(con);
-		if (!queue_work(con->server->rcv_wq, &con->rwork))
-			conn_put(con);
-	}
-	read_unlock_bh(&sk->sk_callback_lock);
-}
-
-static void sock_write_space(struct sock *sk)
-{
-	struct tipc_conn *con;
-
-	read_lock_bh(&sk->sk_callback_lock);
-	con = sock2con(sk);
-	if (con && test_bit(CF_CONNECTED, &con->flags)) {
-		conn_get(con);
-		if (!queue_work(con->server->send_wq, &con->swork))
-			conn_put(con);
-	}
-	read_unlock_bh(&sk->sk_callback_lock);
-}
-
-static void tipc_register_callbacks(struct socket *sock, struct tipc_conn *con)
+static void tipc_conn_close(struct tipc_conn *con)
 {
-	struct sock *sk = sock->sk;
-
-	write_lock_bh(&sk->sk_callback_lock);
-
-	sk->sk_data_ready = sock_data_ready;
-	sk->sk_write_space = sock_write_space;
-	sk->sk_user_data = con;
-
-	con->sock = sock;
-
-	write_unlock_bh(&sk->sk_callback_lock);
-}
-
-static void tipc_close_conn(struct tipc_conn *con)
-{
-	struct tipc_server *s = con->server;
 	struct sock *sk = con->sock->sk;
 	bool disconnect = false;
 
 	write_lock_bh(&sk->sk_callback_lock);
 	disconnect = test_and_clear_bit(CF_CONNECTED, &con->flags);
+
 	if (disconnect) {
 		sk->sk_user_data = NULL;
-		if (con->conid)
-			s->tipc_conn_release(con->conid, con->usr_data);
+		tipc_conn_delete_sub(con, NULL);
 	}
 	write_unlock_bh(&sk->sk_callback_lock);
 
@@ -205,23 +180,26 @@ static void tipc_close_conn(struct tipc_conn *con)
 
 	/* Don't flush pending works, -just let them expire */
 	kernel_sock_shutdown(con->sock, SHUT_RDWR);
+
 	conn_put(con);
 }
 
-static struct tipc_conn *tipc_alloc_conn(struct tipc_server *s)
+static struct tipc_conn *tipc_conn_alloc(struct tipc_topsrv *s)
 {
 	struct tipc_conn *con;
 	int ret;
 
-	con = kzalloc(sizeof(struct tipc_conn), GFP_ATOMIC);
+	con = kzalloc(sizeof(*con), GFP_ATOMIC);
 	if (!con)
 		return ERR_PTR(-ENOMEM);
 
 	kref_init(&con->kref);
 	INIT_LIST_HEAD(&con->outqueue);
+	INIT_LIST_HEAD(&con->sub_list);
 	spin_lock_init(&con->outqueue_lock);
-	INIT_WORK(&con->swork, tipc_send_work);
-	INIT_WORK(&con->rwork, tipc_recv_work);
+	spin_lock_init(&con->sub_lock);
+	INIT_WORK(&con->swork, tipc_conn_send_work);
+	INIT_WORK(&con->rwork, tipc_conn_recv_work);
 
 	spin_lock_bh(&s->idr_lock);
 	ret = idr_alloc(&s->conn_idr, con, 0, 0, GFP_ATOMIC);
@@ -240,260 +218,353 @@ static struct tipc_conn *tipc_alloc_conn(struct tipc_server *s)
 	return con;
 }
 
-static int tipc_receive_from_sock(struct tipc_conn *con)
+static struct tipc_conn *tipc_conn_lookup(struct tipc_topsrv *s, int conid)
 {
-	struct tipc_server *s = con->server;
-	struct sock *sk = con->sock->sk;
-	struct sockaddr_tipc addr;
-	struct msghdr msg = {};
-	struct kvec iov;
-	void *buf;
-	int ret;
+	struct tipc_conn *con;
 
-	buf = kmem_cache_alloc(s->rcvbuf_cache, GFP_ATOMIC);
-	if (!buf) {
-		ret = -ENOMEM;
-		goto out_close;
-	}
+	spin_lock_bh(&s->idr_lock);
+	con = idr_find(&s->conn_idr, conid);
+	if (!connected(con) || !kref_get_unless_zero(&con->kref))
+		con = NULL;
+	spin_unlock_bh(&s->idr_lock);
+	return con;
+}
 
-	iov.iov_base = buf;
-	iov.iov_len = s->max_rcvbuf_size;
-	msg.msg_name = &addr;
-	iov_iter_kvec(&msg.msg_iter, READ | ITER_KVEC, &iov, 1, iov.iov_len);
-	ret = sock_recvmsg(con->sock, &msg, MSG_DONTWAIT);
-	if (ret <= 0) {
-		kmem_cache_free(s->rcvbuf_cache, buf);
-		goto out_close;
+/* tipc_conn_delete_sub - delete a specific or all subscriptions
+ * for a given subscriber
+ */
+static void tipc_conn_delete_sub(struct tipc_conn *con, struct tipc_subscr *s)
+{
+	struct tipc_net *tn = tipc_net(con->server->net);
+	struct list_head *sub_list = &con->sub_list;
+	struct tipc_subscription *sub, *tmp;
+
+	spin_lock_bh(&con->sub_lock);
+	list_for_each_entry_safe(sub, tmp, sub_list, sub_list) {
+		if (!s || !memcmp(s, &sub->evt.s, sizeof(*s))) {
+			tipc_sub_unsubscribe(sub);
+			atomic_dec(&tn->subscription_count);
+		} else if (s) {
+			break;
+		}
 	}
-
-	read_lock_bh(&sk->sk_callback_lock);
-	if (test_bit(CF_CONNECTED, &con->flags))
-		ret = s->tipc_conn_recvmsg(sock_net(con->sock->sk), con->conid,
-					   &addr, con->usr_data, buf, ret);
-	read_unlock_bh(&sk->sk_callback_lock);
-	kmem_cache_free(s->rcvbuf_cache, buf);
-	if (ret < 0)
-		tipc_conn_terminate(s, con->conid);
-	return ret;
-
-out_close:
-	if (ret != -EWOULDBLOCK)
-		tipc_close_conn(con);
-	else if (ret == 0)
-		/* Don't return success if we really got EOF */
-		ret = -EAGAIN;
-
-	return ret;
+	spin_unlock_bh(&con->sub_lock);
 }
 
-static int tipc_accept_from_sock(struct tipc_conn *con)
+static void tipc_conn_send_to_sock(struct tipc_conn *con)
 {
-	struct tipc_server *s = con->server;
-	struct socket *sock = con->sock;
-	struct socket *newsock;
-	struct tipc_conn *newcon;
+	struct list_head *queue = &con->outqueue;
+	struct tipc_topsrv *srv = con->server;
+	struct outqueue_entry *e;
+	struct tipc_event *evt;
+	struct msghdr msg;
+	struct kvec iov;
+	int count = 0;
 	int ret;
 
-	ret = kernel_accept(sock, &newsock, O_NONBLOCK);
-	if (ret < 0)
-		return ret;
+	spin_lock_bh(&con->outqueue_lock);
 
-	newcon = tipc_alloc_conn(con->server);
-	if (IS_ERR(newcon)) {
-		ret = PTR_ERR(newcon);
-		sock_release(newsock);
-		return ret;
-	}
+	while (!list_empty(queue)) {
+		e = list_first_entry(queue, struct outqueue_entry, list);
+		evt = &e->evt;
+		spin_unlock_bh(&con->outqueue_lock);
 
-	newcon->rx_action = tipc_receive_from_sock;
-	tipc_register_callbacks(newsock, newcon);
+		if (e->inactive)
+			tipc_conn_delete_sub(con, &evt->s);
 
-	/* Notify that new connection is incoming */
-	newcon->usr_data = s->tipc_conn_new(newcon->conid);
-	if (!newcon->usr_data) {
-		sock_release(newsock);
-		conn_put(newcon);
-		return -ENOMEM;
-	}
+		memset(&msg, 0, sizeof(msg));
+		msg.msg_flags = MSG_DONTWAIT;
+		iov.iov_base = evt;
+		iov.iov_len = sizeof(*evt);
+		msg.msg_name = NULL;
 
-	/* Wake up receive process in case of 'SYN+' message */
-	newsock->sk->sk_data_ready(newsock->sk);
-	return ret;
+		if (con->sock) {
+			ret = kernel_sendmsg(con->sock, &msg, &iov,
+					     1, sizeof(*evt));
+			if (ret == -EWOULDBLOCK || ret == 0) {
+				cond_resched();
+				return;
+			} else if (ret < 0) {
+				return tipc_conn_close(con);
+			}
+		} else {
+			tipc_topsrv_kern_evt(srv->net, evt);
+		}
+
+		/* Don't starve users filling buffers */
+		if (++count >= MAX_SEND_MSG_COUNT) {
+			cond_resched();
+			count = 0;
+		}
+		spin_lock_bh(&con->outqueue_lock);
+		list_del(&e->list);
+		kfree(e);
+	}
+	spin_unlock_bh(&con->outqueue_lock);
 }
 
-static struct socket *tipc_create_listen_sock(struct tipc_conn *con)
+static void tipc_conn_send_work(struct work_struct *work)
 {
-	struct tipc_server *s = con->server;
-	struct socket *sock = NULL;
-	int ret;
+	struct tipc_conn *con = container_of(work, struct tipc_conn, swork);
 
-	ret = sock_create_kern(s->net, AF_TIPC, SOCK_SEQPACKET, 0, &sock);
-	if (ret < 0)
-		return NULL;
-	ret = kernel_setsockopt(sock, SOL_TIPC, TIPC_IMPORTANCE,
-				(char *)&s->imp, sizeof(s->imp));
-	if (ret < 0)
-		goto create_err;
-	ret = kernel_bind(sock, (struct sockaddr *)s->saddr, sizeof(*s->saddr));
-	if (ret < 0)
-		goto create_err;
+	if (connected(con))
+		tipc_conn_send_to_sock(con);
+
+	conn_put(con);
+}
 
-	switch (s->type) {
-	case SOCK_STREAM:
-	case SOCK_SEQPACKET:
-		con->rx_action = tipc_accept_from_sock;
+/* tipc_conn_queue_evt() - interrupt level call from a subscription instance
+ * The queued work is launched into tipc_send_work()->tipc_send_to_sock()
+ */
+void tipc_topsrv_queue_evt(struct net *net, int conid,
+			   u32 event, struct tipc_event *evt)
+{
+	struct tipc_topsrv *srv = tipc_topsrv(net);
+	struct outqueue_entry *e;
+	struct tipc_conn *con;
 
-		ret = kernel_listen(sock, 0);
-		if (ret < 0)
-			goto create_err;
-		break;
-	case SOCK_DGRAM:
-	case SOCK_RDM:
-		con->rx_action = tipc_receive_from_sock;
-		break;
-	default:
-		pr_err("Unknown socket type %d\n", s->type);
-		goto create_err;
-	}
+	con = tipc_conn_lookup(srv, conid);
+	if (!con)
+		return;
 
-	/* As server's listening socket owner and creator is the same module,
-	 * we have to decrease TIPC module reference count to guarantee that
-	 * it remains zero after the server socket is created, otherwise,
-	 * executing "rmmod" command is unable to make TIPC module deleted
-	 * after TIPC module is inserted successfully.
-	 *
-	 * However, the reference count is ever increased twice in
-	 * sock_create_kern(): one is to increase the reference count of owner
-	 * of TIPC socket's proto_ops struct; another is to increment the
-	 * reference count of owner of TIPC proto struct. Therefore, we must
-	 * decrement the module reference count twice to ensure that it keeps
-	 * zero after server's listening socket is created. Of course, we
-	 * must bump the module reference count twice as well before the socket
-	 * is closed.
-	 */
-	module_put(sock->ops->owner);
-	module_put(sock->sk->sk_prot_creator->owner);
-	set_bit(CF_SERVER, &con->flags);
+	if (!connected(con))
+		goto err;
 
-	return sock;
+	e = kmalloc(sizeof(*e), GFP_ATOMIC);
+	if (!e)
+		goto err;
+	e->inactive = (event == TIPC_SUBSCR_TIMEOUT);
+	memcpy(&e->evt, evt, sizeof(*evt));
+	spin_lock_bh(&con->outqueue_lock);
+	list_add_tail(&e->list, &con->outqueue);
+	spin_unlock_bh(&con->outqueue_lock);
 
-create_err:
-	kernel_sock_shutdown(sock, SHUT_RDWR);
-	sock_release(sock);
-	return NULL;
+	if (queue_work(srv->send_wq, &con->swork))
+		return;
+err:
+	conn_put(con);
 }
 
-static int tipc_open_listening_sock(struct tipc_server *s)
+/* tipc_conn_write_space - interrupt callback after a sendmsg EAGAIN
+ * Indicates that there now is more space in the send buffer
+ * The queued work is launched into tipc_send_work()->tipc_conn_send_to_sock()
+ */
+static void tipc_conn_write_space(struct sock *sk)
 {
-	struct socket *sock;
 	struct tipc_conn *con;
 
-	con = tipc_alloc_conn(s);
-	if (IS_ERR(con))
-		return PTR_ERR(con);
-
-	sock = tipc_create_listen_sock(con);
-	if (!sock) {
-		idr_remove(&s->conn_idr, con->conid);
-		s->idr_in_use--;
-		kfree(con);
-		return -EINVAL;
+	read_lock_bh(&sk->sk_callback_lock);
+	con = sk->sk_user_data;
+	if (connected(con)) {
+		conn_get(con);
+		if (!queue_work(con->server->send_wq, &con->swork))
+			conn_put(con);
 	}
+	read_unlock_bh(&sk->sk_callback_lock);
+}
+
+static int tipc_conn_rcv_sub(struct tipc_topsrv *srv,
+			     struct tipc_conn *con,
+			     struct tipc_subscr *s)
+{
+	struct tipc_net *tn = tipc_net(srv->net);
+	struct tipc_subscription *sub;
 
-	tipc_register_callbacks(sock, con);
+	if (tipc_sub_read(s, filter) & TIPC_SUB_CANCEL) {
+		tipc_conn_delete_sub(con, s);
+		return 0;
+	}
+	if (atomic_read(&tn->subscription_count) >= TIPC_MAX_SUBSCR) {
+		pr_warn("Subscription rejected, max (%u)\n", TIPC_MAX_SUBSCR);
+		return -1;
+	}
+	sub = tipc_sub_subscribe(srv->net, s, con->conid);
+	if (!sub)
+		return -1;
+	atomic_inc(&tn->subscription_count);
+	spin_lock_bh(&con->sub_lock);
+	list_add(&sub->sub_list, &con->sub_list);
+	spin_unlock_bh(&con->sub_lock);
 	return 0;
 }
 
-static struct outqueue_entry *tipc_alloc_entry(void *data, int len)
+static int tipc_conn_rcv_from_sock(struct tipc_conn *con)
 {
-	struct outqueue_entry *entry;
-	void *buf;
-
-	entry = kmalloc(sizeof(struct outqueue_entry), GFP_ATOMIC);
-	if (!entry)
-		return NULL;
+	struct tipc_topsrv *srv = con->server;
+	struct sock *sk = con->sock->sk;
+	struct msghdr msg = {};
+	struct tipc_subscr s;
+	struct kvec iov;
+	int ret;
 
-	buf = kmemdup(data, len, GFP_ATOMIC);
-	if (!buf) {
-		kfree(entry);
-		return NULL;
+	iov.iov_base = &s;
+	iov.iov_len = sizeof(s);
+	msg.msg_name = NULL;
+	iov_iter_kvec(&msg.msg_iter, READ | ITER_KVEC, &iov, 1, iov.iov_len);
+	ret = sock_recvmsg(con->sock, &msg, MSG_DONTWAIT);
+	if (ret == -EWOULDBLOCK)
+		return -EWOULDBLOCK;
+	if (ret > 0) {
+		read_lock_bh(&sk->sk_callback_lock);
+		ret = tipc_conn_rcv_sub(srv, con, &s);
+		read_unlock_bh(&sk->sk_callback_lock);
 	}
+	if (ret < 0)
+		tipc_conn_close(con);
 
-	entry->iov.iov_base = buf;
-	entry->iov.iov_len = len;
-
-	return entry;
+	return ret;
 }
 
-static void tipc_free_entry(struct outqueue_entry *e)
+static void tipc_conn_recv_work(struct work_struct *work)
 {
-	kfree(e->iov.iov_base);
-	kfree(e);
-}
+	struct tipc_conn *con = container_of(work, struct tipc_conn, rwork);
+	int count = 0;
 
-static void tipc_clean_outqueues(struct tipc_conn *con)
-{
-	struct outqueue_entry *e, *safe;
+	while (connected(con)) {
+		if (tipc_conn_rcv_from_sock(con))
+			break;
 
-	spin_lock_bh(&con->outqueue_lock);
-	list_for_each_entry_safe(e, safe, &con->outqueue, list) {
-		list_del(&e->list);
-		tipc_free_entry(e);
+		/* Don't flood Rx machine */
+		if (++count >= MAX_RECV_MSG_COUNT) {
+			cond_resched();
+			count = 0;
+		}
 	}
-	spin_unlock_bh(&con->outqueue_lock);
+	conn_put(con);
 }
 
-int tipc_conn_sendmsg(struct tipc_server *s, int conid,
-		      struct sockaddr_tipc *addr, void *data, size_t len)
+/* tipc_conn_data_ready - interrupt callback indicating the socket has data
+ * The queued work is launched into tipc_recv_work()->tipc_conn_rcv_from_sock()
+ */
+static void tipc_conn_data_ready(struct sock *sk)
 {
-	struct outqueue_entry *e;
 	struct tipc_conn *con;
 
-	con = tipc_conn_lookup(s, conid);
-	if (!con)
-		return -EINVAL;
-
-	if (!test_bit(CF_CONNECTED, &con->flags)) {
-		conn_put(con);
-		return 0;
+	read_lock_bh(&sk->sk_callback_lock);
+	con = sk->sk_user_data;
+	if (connected(con)) {
+		conn_get(con);
+		if (!queue_work(con->server->rcv_wq, &con->rwork))
+			conn_put(con);
 	}
+	read_unlock_bh(&sk->sk_callback_lock);
+}
 
-	e = tipc_alloc_entry(data, len);
-	if (!e) {
-		conn_put(con);
-		return -ENOMEM;
-	}
+static void tipc_topsrv_accept(struct work_struct *work)
+{
+	struct tipc_topsrv *srv = container_of(work, struct tipc_topsrv, awork);
+	struct socket *lsock = srv->listener;
+	struct socket *newsock;
+	struct tipc_conn *con;
+	struct sock *newsk;
+	int ret;
 
-	if (addr)
-		memcpy(&e->dest, addr, sizeof(struct sockaddr_tipc));
+	while (1) {
+		ret = kernel_accept(lsock, &newsock, O_NONBLOCK);
+		if (ret < 0)
+			return;
+		con = tipc_conn_alloc(srv);
+		if (IS_ERR(con)) {
+			ret = PTR_ERR(con);
+			sock_release(newsock);
+			return;
+		}
+		/* Register callbacks */
+		newsk = newsock->sk;
+		write_lock_bh(&newsk->sk_callback_lock);
+		newsk->sk_data_ready = tipc_conn_data_ready;
+		newsk->sk_write_space = tipc_conn_write_space;
+		newsk->sk_user_data = con;
+		con->sock = newsock;
+		write_unlock_bh(&newsk->sk_callback_lock);
+
+		/* Wake up receive process in case of 'SYN+' message */
+		newsk->sk_data_ready(newsk);
+	}
+}
 
-	spin_lock_bh(&con->outqueue_lock);
-	list_add_tail(&e->list, &con->outqueue);
-	spin_unlock_bh(&con->outqueue_lock);
+/* tipc_toprsv_listener_data_ready - interrupt callback with connection request
+ * The queued job is launched into tipc_topsrv_accept()
+ */
+static void tipc_topsrv_listener_data_ready(struct sock *sk)
+{
+	struct tipc_topsrv *srv;
 
-	if (!queue_work(s->send_wq, &con->swork))
-		conn_put(con);
-	return 0;
+	read_lock_bh(&sk->sk_callback_lock);
+	srv = sk->sk_user_data;
+	if (srv->listener)
+		queue_work(srv->rcv_wq, &srv->awork);
+	read_unlock_bh(&sk->sk_callback_lock);
 }
 
-void tipc_conn_terminate(struct tipc_server *s, int conid)
+static int tipc_topsrv_create_listener(struct tipc_topsrv *srv)
 {
-	struct tipc_conn *con;
+	int imp = TIPC_CRITICAL_IMPORTANCE;
+	struct socket *lsock = NULL;
+	struct sockaddr_tipc saddr;
+	struct sock *sk;
+	int rc;
 
-	con = tipc_conn_lookup(s, conid);
-	if (con) {
-		tipc_close_conn(con);
-		conn_put(con);
-	}
+	rc = sock_create_kern(srv->net, AF_TIPC, SOCK_SEQPACKET, 0, &lsock);
+	if (rc < 0)
+		return rc;
+
+	srv->listener = lsock;
+	sk = lsock->sk;
+	write_lock_bh(&sk->sk_callback_lock);
+	sk->sk_data_ready = tipc_topsrv_listener_data_ready;
+	sk->sk_user_data = srv;
+	write_unlock_bh(&sk->sk_callback_lock);
+
+	rc = kernel_setsockopt(lsock, SOL_TIPC, TIPC_IMPORTANCE,
+			       (char *)&imp, sizeof(imp));
+	if (rc < 0)
+		goto err;
+
+	saddr.family	                = AF_TIPC;
+	saddr.addrtype		        = TIPC_ADDR_NAMESEQ;
+	saddr.addr.nameseq.type	        = TIPC_TOP_SRV;
+	saddr.addr.nameseq.lower	= TIPC_TOP_SRV;
+	saddr.addr.nameseq.upper	= TIPC_TOP_SRV;
+	saddr.scope			= TIPC_NODE_SCOPE;
+
+	rc = kernel_bind(lsock, (struct sockaddr *)&saddr, sizeof(saddr));
+	if (rc < 0)
+		goto err;
+	rc = kernel_listen(lsock, 0);
+	if (rc < 0)
+		goto err;
+
+	/* As server's listening socket owner and creator is the same module,
+	 * we have to decrease TIPC module reference count to guarantee that
+	 * it remains zero after the server socket is created, otherwise,
+	 * executing "rmmod" command is unable to make TIPC module deleted
+	 * after TIPC module is inserted successfully.
+	 *
+	 * However, the reference count is ever increased twice in
+	 * sock_create_kern(): one is to increase the reference count of owner
+	 * of TIPC socket's proto_ops struct; another is to increment the
+	 * reference count of owner of TIPC proto struct. Therefore, we must
+	 * decrement the module reference count twice to ensure that it keeps
+	 * zero after server's listening socket is created. Of course, we
+	 * must bump the module reference count twice as well before the socket
+	 * is closed.
+	 */
+	module_put(lsock->ops->owner);
+	module_put(sk->sk_prot_creator->owner);
+
+	return 0;
+err:
+	sock_release(lsock);
+	return -EINVAL;
 }
 
 bool tipc_topsrv_kern_subscr(struct net *net, u32 port, u32 type, u32 lower,
 			     u32 upper, u32 filter, int *conid)
 {
-	struct tipc_subscriber *scbr;
 	struct tipc_subscr sub;
-	struct tipc_server *s;
 	struct tipc_conn *con;
+	int rc;
 
 	sub.seq.type = type;
 	sub.seq.lower = lower;
@@ -502,42 +573,33 @@ bool tipc_topsrv_kern_subscr(struct net *net, u32 port, u32 type, u32 lower,
 	sub.filter = filter;
 	*(u32 *)&sub.usr_handle = port;
 
-	con = tipc_alloc_conn(tipc_topsrv(net));
+	con = tipc_conn_alloc(tipc_topsrv(net));
 	if (IS_ERR(con))
 		return false;
 
 	*conid = con->conid;
-	s = con->server;
-	scbr = s->tipc_conn_new(*conid);
-	if (!scbr) {
-		conn_put(con);
-		return false;
-	}
-
-	con->usr_data = scbr;
 	con->sock = NULL;
-	s->tipc_conn_recvmsg(net, *conid, NULL, scbr, &sub, sizeof(sub));
-	return true;
+	rc = tipc_conn_rcv_sub(tipc_topsrv(net), con, &sub);
+	if (rc < 0)
+		tipc_conn_close(con);
+	return !rc;
 }
 
 void tipc_topsrv_kern_unsubscr(struct net *net, int conid)
 {
 	struct tipc_conn *con;
-	struct tipc_server *srv;
 
 	con = tipc_conn_lookup(tipc_topsrv(net), conid);
 	if (!con)
 		return;
 
 	test_and_clear_bit(CF_CONNECTED, &con->flags);
-	srv = con->server;
-	if (con->conid)
-		srv->tipc_conn_release(con->conid, con->usr_data);
+	tipc_conn_delete_sub(con, NULL);
 	conn_put(con);
 	conn_put(con);
 }
 
-static void tipc_send_kern_top_evt(struct net *net, struct tipc_event *evt)
+static void tipc_topsrv_kern_evt(struct net *net, struct tipc_event *evt)
 {
 	u32 port = *(u32 *)&evt->s.usr_handle;
 	u32 self = tipc_own_addr(net);
@@ -555,95 +617,7 @@ static void tipc_send_kern_top_evt(struct net *net, struct tipc_event *evt)
 	tipc_sk_rcv(net, &evtq);
 }
 
-static void tipc_send_to_sock(struct tipc_conn *con)
-{
-	struct tipc_server *s = con->server;
-	struct outqueue_entry *e;
-	struct tipc_event *evt;
-	struct msghdr msg;
-	int count = 0;
-	int ret;
-
-	spin_lock_bh(&con->outqueue_lock);
-	while (test_bit(CF_CONNECTED, &con->flags)) {
-		e = list_entry(con->outqueue.next, struct outqueue_entry, list);
-		if ((struct list_head *) e == &con->outqueue)
-			break;
-
-		spin_unlock_bh(&con->outqueue_lock);
-
-		if (con->sock) {
-			memset(&msg, 0, sizeof(msg));
-			msg.msg_flags = MSG_DONTWAIT;
-			if (s->type == SOCK_DGRAM || s->type == SOCK_RDM) {
-				msg.msg_name = &e->dest;
-				msg.msg_namelen = sizeof(struct sockaddr_tipc);
-			}
-			ret = kernel_sendmsg(con->sock, &msg, &e->iov, 1,
-					     e->iov.iov_len);
-			if (ret == -EWOULDBLOCK || ret == 0) {
-				cond_resched();
-				goto out;
-			} else if (ret < 0) {
-				goto send_err;
-			}
-		} else {
-			evt = e->iov.iov_base;
-			tipc_send_kern_top_evt(s->net, evt);
-		}
-		/* Don't starve users filling buffers */
-		if (++count >= MAX_SEND_MSG_COUNT) {
-			cond_resched();
-			count = 0;
-		}
-
-		spin_lock_bh(&con->outqueue_lock);
-		list_del(&e->list);
-		tipc_free_entry(e);
-	}
-	spin_unlock_bh(&con->outqueue_lock);
-out:
-	return;
-
-send_err:
-	tipc_close_conn(con);
-}
-
-static void tipc_recv_work(struct work_struct *work)
-{
-	struct tipc_conn *con = container_of(work, struct tipc_conn, rwork);
-	int count = 0;
-
-	while (test_bit(CF_CONNECTED, &con->flags)) {
-		if (con->rx_action(con))
-			break;
-
-		/* Don't flood Rx machine */
-		if (++count >= MAX_RECV_MSG_COUNT) {
-			cond_resched();
-			count = 0;
-		}
-	}
-	conn_put(con);
-}
-
-static void tipc_send_work(struct work_struct *work)
-{
-	struct tipc_conn *con = container_of(work, struct tipc_conn, swork);
-
-	if (test_bit(CF_CONNECTED, &con->flags))
-		tipc_send_to_sock(con);
-
-	conn_put(con);
-}
-
-static void tipc_work_stop(struct tipc_server *s)
-{
-	destroy_workqueue(s->rcv_wq);
-	destroy_workqueue(s->send_wq);
-}
-
-static int tipc_work_start(struct tipc_server *s)
+static int tipc_topsrv_work_start(struct tipc_topsrv *s)
 {
 	s->rcv_wq = alloc_ordered_workqueue("tipc_rcv", 0);
 	if (!s->rcv_wq) {
@@ -661,50 +635,68 @@ static int tipc_work_start(struct tipc_server *s)
 	return 0;
 }
 
-int tipc_server_start(struct tipc_server *s)
+static void tipc_topsrv_work_stop(struct tipc_topsrv *s)
 {
-	int ret;
+	destroy_workqueue(s->rcv_wq);
+	destroy_workqueue(s->send_wq);
+}
 
-	spin_lock_init(&s->idr_lock);
-	idr_init(&s->conn_idr);
-	s->idr_in_use = 0;
+int tipc_topsrv_start(struct net *net)
+{
+	struct tipc_net *tn = tipc_net(net);
+	const char name[] = "topology_server";
+	struct tipc_topsrv *srv;
+	int ret;
 
-	s->rcvbuf_cache = kmem_cache_create(s->name, s->max_rcvbuf_size,
-					    0, SLAB_HWCACHE_ALIGN, NULL);
-	if (!s->rcvbuf_cache)
+	srv = kzalloc(sizeof(*srv), GFP_ATOMIC);
+	if (!srv)
 		return -ENOMEM;
 
-	ret = tipc_work_start(s);
-	if (ret < 0) {
-		kmem_cache_destroy(s->rcvbuf_cache);
-		return ret;
-	}
-	ret = tipc_open_listening_sock(s);
-	if (ret < 0) {
-		tipc_work_stop(s);
-		kmem_cache_destroy(s->rcvbuf_cache);
+	srv->net = net;
+	srv->max_rcvbuf_size = sizeof(struct tipc_subscr);
+	INIT_WORK(&srv->awork, tipc_topsrv_accept);
+
+	strncpy(srv->name, name, strlen(name) + 1);
+	tn->topsrv = srv;
+	atomic_set(&tn->subscription_count, 0);
+
+	spin_lock_init(&srv->idr_lock);
+	idr_init(&srv->conn_idr);
+	srv->idr_in_use = 0;
+
+	ret = tipc_topsrv_work_start(srv);
+	if (ret < 0)
 		return ret;
-	}
+
+	ret = tipc_topsrv_create_listener(srv);
+	if (ret < 0)
+		tipc_topsrv_work_stop(srv);
+
 	return ret;
 }
 
-void tipc_server_stop(struct tipc_server *s)
+void tipc_topsrv_stop(struct net *net)
 {
+	struct tipc_topsrv *srv = tipc_topsrv(net);
+	struct socket *lsock = srv->listener;
 	struct tipc_conn *con;
 	int id;
 
-	spin_lock_bh(&s->idr_lock);
-	for (id = 0; s->idr_in_use; id++) {
-		con = idr_find(&s->conn_idr, id);
+	spin_lock_bh(&srv->idr_lock);
+	for (id = 0; srv->idr_in_use; id++) {
+		con = idr_find(&srv->conn_idr, id);
 		if (con) {
-			spin_unlock_bh(&s->idr_lock);
-			tipc_close_conn(con);
-			spin_lock_bh(&s->idr_lock);
+			spin_unlock_bh(&srv->idr_lock);
+			tipc_conn_close(con);
+			spin_lock_bh(&srv->idr_lock);
 		}
 	}
-	spin_unlock_bh(&s->idr_lock);
-
-	tipc_work_stop(s);
-	kmem_cache_destroy(s->rcvbuf_cache);
-	idr_destroy(&s->conn_idr);
+	__module_get(lsock->ops->owner);
+	__module_get(lsock->sk->sk_prot_creator->owner);
+	sock_release(lsock);
+	srv->listener = NULL;
+	spin_unlock_bh(&srv->idr_lock);
+	tipc_topsrv_work_stop(srv);
+	idr_destroy(&srv->conn_idr);
+	kfree(srv);
 }

net/tipc/server.h → net/tipc/topsrv.h

@@ -2,6 +2,7 @@
  * net/tipc/server.h: Include file for TIPC server code
  *
  * Copyright (c) 2012-2013, Wind River Systems
+ * Copyright (c) 2017, Ericsson AB
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -36,68 +37,18 @@
 #ifndef _TIPC_SERVER_H
 #define _TIPC_SERVER_H
 
-#include <linux/idr.h>
-#include <linux/tipc.h>
-#include <net/net_namespace.h>
+#include "core.h"
 
 #define TIPC_SERVER_NAME_LEN	32
 #define TIPC_SUB_CLUSTER_SCOPE  0x20
 #define TIPC_SUB_NODE_SCOPE     0x40
 #define TIPC_SUB_NO_STATUS      0x80
 
-/**
- * struct tipc_server - TIPC server structure
- * @conn_idr: identifier set of connection
- * @idr_lock: protect the connection identifier set
- * @idr_in_use: amount of allocated identifier entry
- * @net: network namspace instance
- * @rcvbuf_cache: memory cache of server receive buffer
- * @rcv_wq: receive workqueue
- * @send_wq: send workqueue
- * @max_rcvbuf_size: maximum permitted receive message length
- * @tipc_conn_new: callback will be called when new connection is incoming
- * @tipc_conn_release: callback will be called before releasing the connection
- * @tipc_conn_recvmsg: callback will be called when message arrives
- * @saddr: TIPC server address
- * @name: server name
- * @imp: message importance
- * @type: socket type
- */
-struct tipc_server {
-	struct idr conn_idr;
-	spinlock_t idr_lock;
-	int idr_in_use;
-	struct net *net;
-	struct kmem_cache *rcvbuf_cache;
-	struct workqueue_struct *rcv_wq;
-	struct workqueue_struct *send_wq;
-	int max_rcvbuf_size;
-	void *(*tipc_conn_new)(int conid);
-	void (*tipc_conn_release)(int conid, void *usr_data);
-	int (*tipc_conn_recvmsg)(struct net *net, int conid,
-				 struct sockaddr_tipc *addr, void *usr_data,
-				 void *buf, size_t len);
-	struct sockaddr_tipc *saddr;
-	char name[TIPC_SERVER_NAME_LEN];
-	int imp;
-	int type;
-};
-
-int tipc_conn_sendmsg(struct tipc_server *s, int conid,
-		      struct sockaddr_tipc *addr, void *data, size_t len);
+void tipc_topsrv_queue_evt(struct net *net, int conid,
+			   u32 event, struct tipc_event *evt);
 
 bool tipc_topsrv_kern_subscr(struct net *net, u32 port, u32 type, u32 lower,
 			     u32 upper, u32 filter, int *conid);
 void tipc_topsrv_kern_unsubscr(struct net *net, int conid);
 
-/**
- * tipc_conn_terminate - terminate connection with server
- *
- * Note: Must call it in process context since it might sleep
- */
-void tipc_conn_terminate(struct tipc_server *s, int conid);
-int tipc_server_start(struct tipc_server *s);
-
-void tipc_server_stop(struct tipc_server *s);
-
 #endif