Merge branch 'neighbour-convert-rtm_getneigh-to-rcu-and-make-pneigh-rtnl-free'

Kuniyuki Iwashima says:

====================
neighbour: Convert RTM_GETNEIGH to RCU and make pneigh RTNL-free.

This is kind of v3 of the series below [0] but without NEIGHTBL patches.

Patch 1 ~ 4 and 9 come from the series to convert RTM_GETNEIGH to RCU.

Other patches clean up pneigh_lookup() and convert the pneigh code to
RCU + private mutex so that we can easily remove RTNL from RTM_NEWNEIGH
in the later series.

[0]: https://lore.kernel.org/netdev/20250418012727.57033-1-kuniyu@amazon.com/

v2: https://lore.kernel.org/20250712203515.4099110-1-kuniyu@google.com
v1: https://lore.kernel.org/20250711191007.3591938-1-kuniyu@google.com
====================

Link: https://patch.msgid.link/20250716221221.442239-1-kuniyu@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

include/net/neighbour.h

@@ -176,10 +176,14 @@ struct neigh_ops {
 };
 
 struct pneigh_entry {
-	struct pneigh_entry	*next;
+	struct pneigh_entry	__rcu *next;
 	possible_net_t		net;
 	struct net_device	*dev;
 	netdevice_tracker	dev_tracker;
+	union {
+		struct list_head	free_node;
+		struct rcu_head		rcu;
+	};
 	u32			flags;
 	u8			protocol;
 	bool			permanent;
@@ -236,7 +240,8 @@ struct neigh_table {
 	unsigned long		last_rand;
 	struct neigh_statistics	__percpu *stats;
 	struct neigh_hash_table __rcu *nht;
-	struct pneigh_entry	**phash_buckets;
+	struct mutex		phash_lock;
+	struct pneigh_entry	__rcu **phash_buckets;
 };
 
 static inline int neigh_parms_family(struct neigh_parms *p)
@@ -376,10 +381,10 @@ unsigned long neigh_rand_reach_time(unsigned long base);
 void pneigh_enqueue(struct neigh_table *tbl, struct neigh_parms *p,
 		    struct sk_buff *skb);
 struct pneigh_entry *pneigh_lookup(struct neigh_table *tbl, struct net *net,
-				   const void *key, struct net_device *dev,
-				   int creat);
-struct pneigh_entry *__pneigh_lookup(struct neigh_table *tbl, struct net *net,
-				     const void *key, struct net_device *dev);
+				   const void *key, struct net_device *dev);
+int pneigh_create(struct neigh_table *tbl, struct net *net, const void *key,
+		  struct net_device *dev, u32 flags, u8 protocol,
+		  bool permanent);
 int pneigh_delete(struct neigh_table *tbl, struct net *net, const void *key,
 		  struct net_device *dev);
 

net/core/neighbour.c

@@ -28,6 +28,7 @@
 #include <net/neighbour.h>
 #include <net/arp.h>
 #include <net/dst.h>
+#include <net/ip.h>
 #include <net/sock.h>
 #include <net/netevent.h>
 #include <net/netlink.h>
@@ -53,9 +54,8 @@ static void neigh_timer_handler(struct timer_list *t);
 static void __neigh_notify(struct neighbour *n, int type, int flags,
 			   u32 pid);
 static void neigh_update_notify(struct neighbour *neigh, u32 nlmsg_pid);
-static int pneigh_ifdown_and_unlock(struct neigh_table *tbl,
-				    struct net_device *dev,
-				    bool skip_perm);
+static void pneigh_ifdown(struct neigh_table *tbl, struct net_device *dev,
+			  bool skip_perm);
 
 #ifdef CONFIG_PROC_FS
 static const struct seq_operations neigh_stat_seq_ops;
@@ -436,7 +436,9 @@ static int __neigh_ifdown(struct neigh_table *tbl, struct net_device *dev,
 {
 	write_lock_bh(&tbl->lock);
 	neigh_flush_dev(tbl, dev, skip_perm);
-	pneigh_ifdown_and_unlock(tbl, dev, skip_perm);
+	write_unlock_bh(&tbl->lock);
+
+	pneigh_ifdown(tbl, dev, skip_perm);
 	pneigh_queue_purge(&tbl->proxy_queue, dev ? dev_net(dev) : NULL,
 			   tbl->family);
 	if (skb_queue_empty_lockless(&tbl->proxy_queue))
@@ -719,54 +721,53 @@ static u32 pneigh_hash(const void *pkey, unsigned int key_len)
 	return hash_val;
 }
 
-static struct pneigh_entry *__pneigh_lookup_1(struct pneigh_entry *n,
-					      struct net *net,
-					      const void *pkey,
-					      unsigned int key_len,
-					      struct net_device *dev)
+struct pneigh_entry *pneigh_lookup(struct neigh_table *tbl,
+				   struct net *net, const void *pkey,
+				   struct net_device *dev)
 {
+	struct pneigh_entry *n;
+	unsigned int key_len;
+	u32 hash_val;
+
+	key_len = tbl->key_len;
+	hash_val = pneigh_hash(pkey, key_len);
+	n = rcu_dereference_check(tbl->phash_buckets[hash_val],
+				  lockdep_is_held(&tbl->phash_lock));
+
 	while (n) {
 		if (!memcmp(n->key, pkey, key_len) &&
 		    net_eq(pneigh_net(n), net) &&
 		    (n->dev == dev || !n->dev))
 			return n;
-		n = n->next;
+
+		n = rcu_dereference_check(n->next, lockdep_is_held(&tbl->phash_lock));
 	}
+
 	return NULL;
 }
+EXPORT_IPV6_MOD(pneigh_lookup);
 
-struct pneigh_entry *__pneigh_lookup(struct neigh_table *tbl,
-		struct net *net, const void *pkey, struct net_device *dev)
-{
-	unsigned int key_len = tbl->key_len;
-	u32 hash_val = pneigh_hash(pkey, key_len);
-
-	return __pneigh_lookup_1(tbl->phash_buckets[hash_val],
-				 net, pkey, key_len, dev);
-}
-EXPORT_SYMBOL_GPL(__pneigh_lookup);
-
-struct pneigh_entry * pneigh_lookup(struct neigh_table *tbl,
-				    struct net *net, const void *pkey,
-				    struct net_device *dev, int creat)
+int pneigh_create(struct neigh_table *tbl, struct net *net,
+		  const void *pkey, struct net_device *dev,
+		  u32 flags, u8 protocol, bool permanent)
 {
 	struct pneigh_entry *n;
-	unsigned int key_len = tbl->key_len;
-	u32 hash_val = pneigh_hash(pkey, key_len);
+	unsigned int key_len;
+	u32 hash_val;
+	int err = 0;
 
-	read_lock_bh(&tbl->lock);
-	n = __pneigh_lookup_1(tbl->phash_buckets[hash_val],
-			      net, pkey, key_len, dev);
-	read_unlock_bh(&tbl->lock);
+	mutex_lock(&tbl->phash_lock);
 
-	if (n || !creat)
-		goto out;
-
-	ASSERT_RTNL();
+	n = pneigh_lookup(tbl, net, pkey, dev);
+	if (n)
+		goto update;
 
+	key_len = tbl->key_len;
 	n = kzalloc(sizeof(*n) + key_len, GFP_KERNEL);
-	if (!n)
+	if (!n) {
+		err = -ENOBUFS;
 		goto out;
+	}
 
 	write_pnet(&n->net, net);
 	memcpy(n->key, pkey, key_len);
@@ -776,77 +777,98 @@ struct pneigh_entry * pneigh_lookup(struct neigh_table *tbl,
 	if (tbl->pconstructor && tbl->pconstructor(n)) {
 		netdev_put(dev, &n->dev_tracker);
 		kfree(n);
-		n = NULL;
+		err = -ENOBUFS;
 		goto out;
 	}
 
-	write_lock_bh(&tbl->lock);
+	hash_val = pneigh_hash(pkey, key_len);
 	n->next = tbl->phash_buckets[hash_val];
-	tbl->phash_buckets[hash_val] = n;
-	write_unlock_bh(&tbl->lock);
+	rcu_assign_pointer(tbl->phash_buckets[hash_val], n);
+update:
+	WRITE_ONCE(n->flags, flags);
+	n->permanent = permanent;
+	WRITE_ONCE(n->protocol, protocol);
 out:
-	return n;
+	mutex_unlock(&tbl->phash_lock);
+	return err;
 }
-EXPORT_SYMBOL(pneigh_lookup);
 
+static void pneigh_destroy(struct rcu_head *rcu)
+{
+	struct pneigh_entry *n = container_of(rcu, struct pneigh_entry, rcu);
+
+	netdev_put(n->dev, &n->dev_tracker);
+	kfree(n);
+}
 
 int pneigh_delete(struct neigh_table *tbl, struct net *net, const void *pkey,
 		  struct net_device *dev)
 {
-	struct pneigh_entry *n, **np;
-	unsigned int key_len = tbl->key_len;
-	u32 hash_val = pneigh_hash(pkey, key_len);
+	struct pneigh_entry *n, __rcu **np;
+	unsigned int key_len;
+	u32 hash_val;
 
-	write_lock_bh(&tbl->lock);
-	for (np = &tbl->phash_buckets[hash_val]; (n = *np) != NULL;
+	key_len = tbl->key_len;
+	hash_val = pneigh_hash(pkey, key_len);
+
+	mutex_lock(&tbl->phash_lock);
+
+	for (np = &tbl->phash_buckets[hash_val];
+	     (n = rcu_dereference_protected(*np, 1)) != NULL;
 	     np = &n->next) {
 		if (!memcmp(n->key, pkey, key_len) && n->dev == dev &&
 		    net_eq(pneigh_net(n), net)) {
-			*np = n->next;
-			write_unlock_bh(&tbl->lock);
+			rcu_assign_pointer(*np, n->next);
+
+			mutex_unlock(&tbl->phash_lock);
+
 			if (tbl->pdestructor)
 				tbl->pdestructor(n);
-			netdev_put(n->dev, &n->dev_tracker);
-			kfree(n);
+
+			call_rcu(&n->rcu, pneigh_destroy);
 			return 0;
 		}
 	}
-	write_unlock_bh(&tbl->lock);
+
+	mutex_unlock(&tbl->phash_lock);
 	return -ENOENT;
 }
 
-static int pneigh_ifdown_and_unlock(struct neigh_table *tbl,
-				    struct net_device *dev,
-				    bool skip_perm)
+static void pneigh_ifdown(struct neigh_table *tbl, struct net_device *dev,
+			  bool skip_perm)
 {
-	struct pneigh_entry *n, **np, *freelist = NULL;
+	struct pneigh_entry *n, __rcu **np;
+	LIST_HEAD(head);
 	u32 h;
 
+	mutex_lock(&tbl->phash_lock);
+
 	for (h = 0; h <= PNEIGH_HASHMASK; h++) {
 		np = &tbl->phash_buckets[h];
-		while ((n = *np) != NULL) {
+		while ((n = rcu_dereference_protected(*np, 1)) != NULL) {
 			if (skip_perm && n->permanent)
 				goto skip;
 			if (!dev || n->dev == dev) {
-				*np = n->next;
-				n->next = freelist;
-				freelist = n;
+				rcu_assign_pointer(*np, n->next);
+				list_add(&n->free_node, &head);
 				continue;
 			}
 skip:
 			np = &n->next;
 		}
 	}
-	write_unlock_bh(&tbl->lock);
-	while ((n = freelist)) {
-		freelist = n->next;
-		n->next = NULL;
+
+	mutex_unlock(&tbl->phash_lock);
+
+	while (!list_empty(&head)) {
+		n = list_first_entry(&head, typeof(*n), free_node);
+		list_del(&n->free_node);
+
 		if (tbl->pdestructor)
 			tbl->pdestructor(n);
-		netdev_put(n->dev, &n->dev_tracker);
-		kfree(n);
+
+		call_rcu(&n->rcu, pneigh_destroy);
 	}
-	return -ENOENT;
 }
 
 static inline void neigh_parms_put(struct neigh_parms *parms)
@@ -1783,6 +1805,7 @@ void neigh_table_init(int index, struct neigh_table *tbl)
 		WARN_ON(tbl->entry_size % NEIGH_PRIV_ALIGN);
 
 	rwlock_init(&tbl->lock);
+	mutex_init(&tbl->phash_lock);
 
 	INIT_DEFERRABLE_WORK(&tbl->gc_work, neigh_periodic_work);
 	queue_delayed_work(system_power_efficient_wq, &tbl->gc_work,
@@ -1999,22 +2022,13 @@ static int neigh_add(struct sk_buff *skb, struct nlmsghdr *nlh,
 	if (tb[NDA_PROTOCOL])
 		protocol = nla_get_u8(tb[NDA_PROTOCOL]);
 	if (ndm_flags & NTF_PROXY) {
-		struct pneigh_entry *pn;
-
 		if (ndm_flags & (NTF_MANAGED | NTF_EXT_VALIDATED)) {
 			NL_SET_ERR_MSG(extack, "Invalid NTF_* flag combination");
 			goto out;
 		}
 
-		err = -ENOBUFS;
-		pn = pneigh_lookup(tbl, net, dst, dev, 1);
-		if (pn) {
-			pn->flags = ndm_flags;
-			pn->permanent = !!(ndm->ndm_state & NUD_PERMANENT);
-			if (protocol)
-				pn->protocol = protocol;
-			err = 0;
-		}
+		err = pneigh_create(tbl, net, dst, dev, ndm_flags, protocol,
+				    !!(ndm->ndm_state & NUD_PERMANENT));
 		goto out;
 	}
 
@@ -2643,13 +2657,15 @@ static int pneigh_fill_info(struct sk_buff *skb, struct pneigh_entry *pn,
 	u32 neigh_flags, neigh_flags_ext;
 	struct nlmsghdr *nlh;
 	struct ndmsg *ndm;
+	u8 protocol;
 
 	nlh = nlmsg_put(skb, pid, seq, type, sizeof(*ndm), flags);
 	if (nlh == NULL)
 		return -EMSGSIZE;
 
-	neigh_flags_ext = pn->flags >> NTF_EXT_SHIFT;
-	neigh_flags     = pn->flags & NTF_OLD_MASK;
+	neigh_flags = READ_ONCE(pn->flags);
+	neigh_flags_ext = neigh_flags >> NTF_EXT_SHIFT;
+	neigh_flags &= NTF_OLD_MASK;
 
 	ndm = nlmsg_data(nlh);
 	ndm->ndm_family	 = tbl->family;
@@ -2663,7 +2679,8 @@ static int pneigh_fill_info(struct sk_buff *skb, struct pneigh_entry *pn,
 	if (nla_put(skb, NDA_DST, tbl->key_len, pn->key))
 		goto nla_put_failure;
 
-	if (pn->protocol && nla_put_u8(skb, NDA_PROTOCOL, pn->protocol))
+	protocol = READ_ONCE(pn->protocol);
+	if (protocol && nla_put_u8(skb, NDA_PROTOCOL, protocol))
 		goto nla_put_failure;
 	if (neigh_flags_ext && nla_put_u32(skb, NDA_FLAGS_EXT, neigh_flags_ext))
 		goto nla_put_failure;
@@ -2770,12 +2787,12 @@ static int pneigh_dump_table(struct neigh_table *tbl, struct sk_buff *skb,
 	if (filter->dev_idx || filter->master_idx)
 		flags |= NLM_F_DUMP_FILTERED;
 
-	read_lock_bh(&tbl->lock);
-
 	for (h = s_h; h <= PNEIGH_HASHMASK; h++) {
 		if (h > s_h)
 			s_idx = 0;
-		for (n = tbl->phash_buckets[h], idx = 0; n; n = n->next) {
+		for (n = rcu_dereference(tbl->phash_buckets[h]), idx = 0;
+		     n;
+		     n = rcu_dereference(n->next)) {
 			if (idx < s_idx || pneigh_net(n) != net)
 				goto next;
 			if (neigh_ifindex_filtered(n->dev, filter->dev_idx) ||
@@ -2784,16 +2801,13 @@ static int pneigh_dump_table(struct neigh_table *tbl, struct sk_buff *skb,
 			err = pneigh_fill_info(skb, n, NETLINK_CB(cb->skb).portid,
 					       cb->nlh->nlmsg_seq,
 					       RTM_NEWNEIGH, flags, tbl);
-			if (err < 0) {
-				read_unlock_bh(&tbl->lock);
+			if (err < 0)
 				goto out;
-			}
 		next:
 			idx++;
 		}
 	}
 
-	read_unlock_bh(&tbl->lock);
 out:
 	cb->args[3] = h;
 	cb->args[4] = idx;
@@ -2910,64 +2924,58 @@ static int neigh_dump_info(struct sk_buff *skb, struct netlink_callback *cb)
 	return err;
 }
 
-static int neigh_valid_get_req(const struct nlmsghdr *nlh,
-			       struct neigh_table **tbl,
-			       void **dst, int *dev_idx, u8 *ndm_flags,
-			       struct netlink_ext_ack *extack)
+static struct ndmsg *neigh_valid_get_req(const struct nlmsghdr *nlh,
+					 struct nlattr **tb,
+					 struct netlink_ext_ack *extack)
 {
-	struct nlattr *tb[NDA_MAX + 1];
 	struct ndmsg *ndm;
 	int err, i;
 
 	ndm = nlmsg_payload(nlh, sizeof(*ndm));
 	if (!ndm) {
 		NL_SET_ERR_MSG(extack, "Invalid header for neighbor get request");
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 
 	if (ndm->ndm_pad1  || ndm->ndm_pad2  || ndm->ndm_state ||
 	    ndm->ndm_type) {
 		NL_SET_ERR_MSG(extack, "Invalid values in header for neighbor get request");
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 
 	if (ndm->ndm_flags & ~NTF_PROXY) {
 		NL_SET_ERR_MSG(extack, "Invalid flags in header for neighbor get request");
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (!(ndm->ndm_flags & NTF_PROXY) && !ndm->ndm_ifindex) {
+		NL_SET_ERR_MSG(extack, "No device specified");
+		return ERR_PTR(-EINVAL);
 	}
 
 	err = nlmsg_parse_deprecated_strict(nlh, sizeof(struct ndmsg), tb,
 					    NDA_MAX, nda_policy, extack);
 	if (err < 0)
-		return err;
-
-	*ndm_flags = ndm->ndm_flags;
-	*dev_idx = ndm->ndm_ifindex;
-	*tbl = neigh_find_table(ndm->ndm_family);
-	if (*tbl == NULL) {
-		NL_SET_ERR_MSG(extack, "Unsupported family in header for neighbor get request");
-		return -EAFNOSUPPORT;
-	}
+		return ERR_PTR(err);
 
 	for (i = 0; i <= NDA_MAX; ++i) {
-		if (!tb[i])
-			continue;
-
 		switch (i) {
 		case NDA_DST:
-			if (nla_len(tb[i]) != (int)(*tbl)->key_len) {
-				NL_SET_ERR_MSG(extack, "Invalid network address in neighbor get request");
-				return -EINVAL;
+			if (!tb[i]) {
+				NL_SET_ERR_ATTR_MISS(extack, NULL, NDA_DST);
+				return ERR_PTR(-EINVAL);
 			}
-			*dst = nla_data(tb[i]);
 			break;
 		default:
+			if (!tb[i])
+				continue;
+
 			NL_SET_ERR_MSG(extack, "Unsupported attribute in neighbor get request");
-			return -EINVAL;
+			return ERR_PTR(-EINVAL);
 		}
 	}
 
-	return 0;
+	return ndm;
 }
 
 static inline size_t neigh_nlmsg_size(void)
@@ -2981,27 +2989,6 @@ static inline size_t neigh_nlmsg_size(void)
 	       + nla_total_size(1); /* NDA_PROTOCOL */
 }
 
-static int neigh_get_reply(struct net *net, struct neighbour *neigh,
-			   u32 pid, u32 seq)
-{
-	struct sk_buff *skb;
-	int err = 0;
-
-	skb = nlmsg_new(neigh_nlmsg_size(), GFP_KERNEL);
-	if (!skb)
-		return -ENOBUFS;
-
-	err = neigh_fill_info(skb, neigh, pid, seq, RTM_NEWNEIGH, 0);
-	if (err) {
-		kfree_skb(skb);
-		goto errout;
-	}
-
-	err = rtnl_unicast(skb, net, pid);
-errout:
-	return err;
-}
-
 static inline size_t pneigh_nlmsg_size(void)
 {
 	return NLMSG_ALIGN(sizeof(struct ndmsg))
@@ -3010,85 +2997,91 @@ static inline size_t pneigh_nlmsg_size(void)
 	       + nla_total_size(1); /* NDA_PROTOCOL */
 }
 
-static int pneigh_get_reply(struct net *net, struct pneigh_entry *neigh,
-			    u32 pid, u32 seq, struct neigh_table *tbl)
-{
-	struct sk_buff *skb;
-	int err = 0;
-
-	skb = nlmsg_new(pneigh_nlmsg_size(), GFP_KERNEL);
-	if (!skb)
-		return -ENOBUFS;
-
-	err = pneigh_fill_info(skb, neigh, pid, seq, RTM_NEWNEIGH, 0, tbl);
-	if (err) {
-		kfree_skb(skb);
-		goto errout;
-	}
-
-	err = rtnl_unicast(skb, net, pid);
-errout:
-	return err;
-}
-
 static int neigh_get(struct sk_buff *in_skb, struct nlmsghdr *nlh,
 		     struct netlink_ext_ack *extack)
 {
 	struct net *net = sock_net(in_skb->sk);
+	u32 pid = NETLINK_CB(in_skb).portid;
+	struct nlattr *tb[NDA_MAX + 1];
 	struct net_device *dev = NULL;
-	struct neigh_table *tbl = NULL;
+	u32 seq = nlh->nlmsg_seq;
+	struct neigh_table *tbl;
 	struct neighbour *neigh;
-	void *dst = NULL;
-	u8 ndm_flags = 0;
-	int dev_idx = 0;
+	struct sk_buff *skb;
+	struct ndmsg *ndm;
+	void *dst;
 	int err;
 
-	err = neigh_valid_get_req(nlh, &tbl, &dst, &dev_idx, &ndm_flags,
-				  extack);
-	if (err < 0)
-		return err;
+	ndm = neigh_valid_get_req(nlh, tb, extack);
+	if (IS_ERR(ndm))
+		return PTR_ERR(ndm);
 
-	if (dev_idx) {
-		dev = __dev_get_by_index(net, dev_idx);
+	if (ndm->ndm_flags & NTF_PROXY)
+		skb = nlmsg_new(neigh_nlmsg_size(), GFP_KERNEL);
+	else
+		skb = nlmsg_new(pneigh_nlmsg_size(), GFP_KERNEL);
+	if (!skb)
+		return -ENOBUFS;
+
+	rcu_read_lock();
+
+	tbl = neigh_find_table(ndm->ndm_family);
+	if (!tbl) {
+		NL_SET_ERR_MSG(extack, "Unsupported family in header for neighbor get request");
+		err = -EAFNOSUPPORT;
+		goto err_unlock;
+	}
+
+	if (nla_len(tb[NDA_DST]) != (int)tbl->key_len) {
+		NL_SET_ERR_MSG(extack, "Invalid network address in neighbor get request");
+		err = -EINVAL;
+		goto err_unlock;
+	}
+
+	dst = nla_data(tb[NDA_DST]);
+
+	if (ndm->ndm_ifindex) {
+		dev = dev_get_by_index_rcu(net, ndm->ndm_ifindex);
 		if (!dev) {
 			NL_SET_ERR_MSG(extack, "Unknown device ifindex");
-			return -ENODEV;
+			err = -ENODEV;
+			goto err_unlock;
 		}
 	}
 
-	if (!dst) {
-		NL_SET_ERR_MSG(extack, "Network address not specified");
-		return -EINVAL;
-	}
-
-	if (ndm_flags & NTF_PROXY) {
+	if (ndm->ndm_flags & NTF_PROXY) {
 		struct pneigh_entry *pn;
 
-		pn = pneigh_lookup(tbl, net, dst, dev, 0);
+		pn = pneigh_lookup(tbl, net, dst, dev);
 		if (!pn) {
 			NL_SET_ERR_MSG(extack, "Proxy neighbour entry not found");
-			return -ENOENT;
+			err = -ENOENT;
+			goto err_unlock;
 		}
-		return pneigh_get_reply(net, pn, NETLINK_CB(in_skb).portid,
-					nlh->nlmsg_seq, tbl);
+
+		err = pneigh_fill_info(skb, pn, pid, seq, RTM_NEWNEIGH, 0, tbl);
+		if (err)
+			goto err_unlock;
+	} else {
+		neigh = neigh_lookup(tbl, dst, dev);
+		if (!neigh) {
+			NL_SET_ERR_MSG(extack, "Neighbour entry not found");
+			err = -ENOENT;
+			goto err_unlock;
+		}
+
+		err = neigh_fill_info(skb, neigh, pid, seq, RTM_NEWNEIGH, 0);
+		neigh_release(neigh);
+		if (err)
+			goto err_unlock;
 	}
 
-	if (!dev) {
-		NL_SET_ERR_MSG(extack, "No device specified");
-		return -EINVAL;
-	}
-
-	neigh = neigh_lookup(tbl, dst, dev);
-	if (!neigh) {
-		NL_SET_ERR_MSG(extack, "Neighbour entry not found");
-		return -ENOENT;
-	}
-
-	err = neigh_get_reply(net, neigh, NETLINK_CB(in_skb).portid,
-			      nlh->nlmsg_seq);
-
-	neigh_release(neigh);
+	rcu_read_unlock();
 
+	return rtnl_unicast(skb, net, pid);
+err_unlock:
+	rcu_read_unlock();
+	kfree_skb(skb);
 	return err;
 }
 
@@ -3295,9 +3288,10 @@ static struct pneigh_entry *pneigh_get_first(struct seq_file *seq)
 
 	state->flags |= NEIGH_SEQ_IS_PNEIGH;
 	for (bucket = 0; bucket <= PNEIGH_HASHMASK; bucket++) {
-		pn = tbl->phash_buckets[bucket];
+		pn = rcu_dereference(tbl->phash_buckets[bucket]);
+
 		while (pn && !net_eq(pneigh_net(pn), net))
-			pn = pn->next;
+			pn = rcu_dereference(pn->next);
 		if (pn)
 			break;
 	}
@@ -3315,15 +3309,17 @@ static struct pneigh_entry *pneigh_get_next(struct seq_file *seq,
 	struct neigh_table *tbl = state->tbl;
 
 	do {
-		pn = pn->next;
+		pn = rcu_dereference(pn->next);
 	} while (pn && !net_eq(pneigh_net(pn), net));
 
 	while (!pn) {
 		if (++state->bucket > PNEIGH_HASHMASK)
 			break;
-		pn = tbl->phash_buckets[state->bucket];
+
+		pn = rcu_dereference(tbl->phash_buckets[state->bucket]);
+
 		while (pn && !net_eq(pneigh_net(pn), net))
-			pn = pn->next;
+			pn = rcu_dereference(pn->next);
 		if (pn)
 			break;
 	}
@@ -3887,7 +3883,7 @@ static const struct rtnl_msg_handler neigh_rtnl_msg_handlers[] __initconst = {
 	{.msgtype = RTM_NEWNEIGH, .doit = neigh_add},
 	{.msgtype = RTM_DELNEIGH, .doit = neigh_delete},
 	{.msgtype = RTM_GETNEIGH, .doit = neigh_get, .dumpit = neigh_dump_info,
-	 .flags = RTNL_FLAG_DUMP_UNLOCKED},
+	 .flags = RTNL_FLAG_DOIT_UNLOCKED | RTNL_FLAG_DUMP_UNLOCKED},
 	{.msgtype = RTM_GETNEIGHTBL, .dumpit = neightbl_dump_info},
 	{.msgtype = RTM_SETNEIGHTBL, .doit = neightbl_set},
 };

net/ipv4/arp.c

@@ -864,7 +864,7 @@ static int arp_process(struct net *net, struct sock *sk, struct sk_buff *skb)
 			    (arp_fwd_proxy(in_dev, dev, rt) ||
 			     arp_fwd_pvlan(in_dev, dev, rt, sip, tip) ||
 			     (rt->dst.dev != dev &&
-			      pneigh_lookup(&arp_tbl, net, &tip, dev, 0)))) {
+			      pneigh_lookup(&arp_tbl, net, &tip, dev)))) {
 				n = neigh_event_ns(&arp_tbl, sha, &sip, dev);
 				if (n)
 					neigh_release(n);
@@ -1089,9 +1089,7 @@ static int arp_req_set_public(struct net *net, struct arpreq *r,
 	if (mask) {
 		__be32 ip = ((struct sockaddr_in *)&r->arp_pa)->sin_addr.s_addr;
 
-		if (!pneigh_lookup(&arp_tbl, net, &ip, dev, 1))
-			return -ENOBUFS;
-		return 0;
+		return pneigh_create(&arp_tbl, net, &ip, dev, 0, 0, false);
 	}
 
 	return arp_req_set_proxy(net, dev, 1);

net/ipv6/ip6_output.c

@@ -563,7 +563,7 @@ int ip6_forward(struct sk_buff *skb)
 
 	/* XXX: idev->cnf.proxy_ndp? */
 	if (READ_ONCE(net->ipv6.devconf_all->proxy_ndp) &&
-	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
+	    pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev)) {
 		int proxied = ip6_forward_proxy_check(skb);
 		if (proxied > 0) {
 			/* It's tempting to decrease the hop limit

net/ipv6/ndisc.c

@@ -768,11 +768,9 @@ static int pndisc_is_router(const void *pkey,
 	struct pneigh_entry *n;
 	int ret = -1;
 
-	read_lock_bh(&nd_tbl.lock);
-	n = __pneigh_lookup(&nd_tbl, dev_net(dev), pkey, dev);
+	n = pneigh_lookup(&nd_tbl, dev_net(dev), pkey, dev);
 	if (n)
-		ret = !!(n->flags & NTF_ROUTER);
-	read_unlock_bh(&nd_tbl.lock);
+		ret = !!(READ_ONCE(n->flags) & NTF_ROUTER);
 
 	return ret;
 }
@@ -1100,7 +1098,7 @@ static enum skb_drop_reason ndisc_recv_na(struct sk_buff *skb)
 		if (lladdr && !memcmp(lladdr, dev->dev_addr, dev->addr_len) &&
 		    READ_ONCE(net->ipv6.devconf_all->forwarding) &&
 		    READ_ONCE(net->ipv6.devconf_all->proxy_ndp) &&
-		    pneigh_lookup(&nd_tbl, net, &msg->target, dev, 0)) {
+		    pneigh_lookup(&nd_tbl, net, &msg->target, dev)) {
 			/* XXX: idev->cnf.proxy_ndp */
 			goto out;
 		}