From 1ce31c9e08997ea0fa62be0a7437f868be173f13 Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Tue, 29 Sep 2015 07:42:44 -0700
Subject: inet: constify __inet_inherit_port() sock argument

socket is not touched, make it const.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/inet_hashtables.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'net/ipv4/inet_hashtables.c')

diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 89120196a949..56742e995dd3 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -126,7 +126,7 @@ void inet_put_port(struct sock *sk)
 }
 EXPORT_SYMBOL(inet_put_port);
 
-int __inet_inherit_port(struct sock *sk, struct sock *child)
+int __inet_inherit_port(const struct sock *sk, struct sock *child)
 {
 	struct inet_hashinfo *table = sk->sk_prot->h.hashinfo;
 	unsigned short port = inet_sk(child)->inet_num;
-- 
cgit v1.2.3


From 079096f103faca2dd87342cca6f23d4b34da8871 Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Fri, 2 Oct 2015 11:43:32 -0700
Subject: tcp/dccp: install syn_recv requests into ehash table

In this patch, we insert request sockets into TCP/DCCP
regular ehash table (where ESTABLISHED and TIMEWAIT sockets
are) instead of using the per listener hash table.

ACK packets find SYN_RECV pseudo sockets without having
to find and lock the listener.

In nominal conditions, this halves pressure on listener lock.

Note that this will allow for SO_REUSEPORT refinements,
so that we can select a listener using cpu/numa affinities instead
of the prior 'consistent hash', since only SYN packets will
apply this selection logic.

We will shrink listen_sock in the following patch to ease
code review.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ying Cai <ycai@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/inet_connection_sock.h |   4 --
 include/net/inet_hashtables.h      |   1 +
 include/net/request_sock.h         |   4 --
 include/net/tcp.h                  |   3 -
 net/core/request_sock.c            |  28 +--------
 net/dccp/ipv4.c                    |  64 +++++++-------------
 net/dccp/ipv6.c                    |  72 +++++++----------------
 net/ipv4/inet_connection_sock.c    | 103 +++++++-------------------------
 net/ipv4/inet_diag.c               |  96 +++---------------------------
 net/ipv4/inet_hashtables.c         |  14 ++++-
 net/ipv4/syncookies.c              |   4 ++
 net/ipv4/tcp_input.c               |   2 +-
 net/ipv4/tcp_ipv4.c                | 117 +++++++++++--------------------------
 net/ipv6/inet6_connection_sock.c   |  67 ---------------------
 net/ipv6/tcp_ipv6.c                |  82 ++++++++++++--------------
 15 files changed, 160 insertions(+), 501 deletions(-)

(limited to 'net/ipv4/inet_hashtables.c')

diff --git a/include/net/inet_connection_sock.h b/include/net/inet_connection_sock.h
index b2e2e30befa9..730aa034cd3d 100644
--- a/include/net/inet_connection_sock.h
+++ b/include/net/inet_connection_sock.h
@@ -258,10 +258,6 @@ inet_csk_rto_backoff(const struct inet_connection_sock *icsk,
 
 struct sock *inet_csk_accept(struct sock *sk, int flags, int *err);
 
-struct request_sock *inet_csk_search_req(struct sock *sk,
-					 const __be16 rport,
-					 const __be32 raddr,
-					 const __be32 laddr);
 int inet_csk_bind_conflict(const struct sock *sk,
 			   const struct inet_bind_bucket *tb, bool relax);
 int inet_csk_get_port(struct sock *sk, unsigned short snum);
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index 3fb778d7c875..6683ada25fef 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -205,6 +205,7 @@ void inet_put_port(struct sock *sk);
 
 void inet_hashinfo_init(struct inet_hashinfo *h);
 
+int inet_ehash_insert(struct sock *sk, struct sock *osk);
 void __inet_hash_nolisten(struct sock *sk, struct sock *osk);
 void __inet_hash(struct sock *sk, struct sock *osk);
 void inet_hash(struct sock *sk);
diff --git a/include/net/request_sock.h b/include/net/request_sock.h
index 97c1ba61ed2d..e1850923c4f5 100644
--- a/include/net/request_sock.h
+++ b/include/net/request_sock.h
@@ -266,8 +266,4 @@ static inline int reqsk_queue_is_full(const struct request_sock_queue *queue)
 	return reqsk_queue_len(queue) >> queue->listen_opt->max_qlen_log;
 }
 
-void reqsk_queue_hash_req(struct request_sock_queue *queue,
-			  u32 hash, struct request_sock *req,
-			  unsigned long timeout);
-
 #endif /* _REQUEST_SOCK_H */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index a26341d2ad67..225e9561af35 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1618,7 +1618,6 @@ static inline bool tcp_stream_is_thin(struct tcp_sock *tp)
 /* /proc */
 enum tcp_seq_states {
 	TCP_SEQ_STATE_LISTENING,
-	TCP_SEQ_STATE_OPENREQ,
 	TCP_SEQ_STATE_ESTABLISHED,
 };
 
@@ -1717,8 +1716,6 @@ struct tcp_request_sock_ops {
 	int (*send_synack)(const struct sock *sk, struct dst_entry *dst,
 			   struct flowi *fl, struct request_sock *req,
 			   u16 queue_mapping, struct tcp_fastopen_cookie *foc);
-	void (*queue_hash_add)(struct sock *sk, struct request_sock *req,
-			       const unsigned long timeout);
 };
 
 #ifdef CONFIG_SYN_COOKIES
diff --git a/net/core/request_sock.c b/net/core/request_sock.c
index 5ca624cea04c..a4b305d8ca2b 100644
--- a/net/core/request_sock.c
+++ b/net/core/request_sock.c
@@ -99,35 +99,9 @@ static inline struct listen_sock *reqsk_queue_yank_listen_sk(
 
 void reqsk_queue_destroy(struct request_sock_queue *queue)
 {
-	/* make all the listen_opt local to us */
 	struct listen_sock *lopt = reqsk_queue_yank_listen_sk(queue);
 
-	if (reqsk_queue_len(queue) != 0) {
-		unsigned int i;
-
-		for (i = 0; i < lopt->nr_table_entries; i++) {
-			struct request_sock *req;
-
-			spin_lock_bh(&queue->syn_wait_lock);
-			while ((req = lopt->syn_table[i]) != NULL) {
-				lopt->syn_table[i] = req->dl_next;
-				/* Because of following del_timer_sync(),
-				 * we must release the spinlock here
-				 * or risk a dead lock.
-				 */
-				spin_unlock_bh(&queue->syn_wait_lock);
-				atomic_dec(&queue->qlen);
-				if (del_timer_sync(&req->rsk_timer))
-					reqsk_put(req);
-				reqsk_put(req);
-				spin_lock_bh(&queue->syn_wait_lock);
-			}
-			spin_unlock_bh(&queue->syn_wait_lock);
-		}
-	}
-
-	if (WARN_ON(reqsk_queue_len(queue) != 0))
-		pr_err("qlen %u\n", reqsk_queue_len(queue));
+	/* cleaning is done by req timers */
 	kvfree(lopt);
 }
 
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index 5b7818c63cec..8910c9567719 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -444,36 +444,6 @@ put_and_exit:
 }
 EXPORT_SYMBOL_GPL(dccp_v4_request_recv_sock);
 
-static struct sock *dccp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
-{
-	const struct dccp_hdr *dh = dccp_hdr(skb);
-	const struct iphdr *iph = ip_hdr(skb);
-	struct sock *nsk;
-	/* Find possible connection requests. */
-	struct request_sock *req = inet_csk_search_req(sk, dh->dccph_sport,
-						       iph->saddr, iph->daddr);
-	if (req) {
-		nsk = dccp_check_req(sk, skb, req);
-		if (!nsk)
-			reqsk_put(req);
-		return nsk;
-	}
-	nsk = inet_lookup_established(sock_net(sk), &dccp_hashinfo,
-				      iph->saddr, dh->dccph_sport,
-				      iph->daddr, dh->dccph_dport,
-				      inet_iif(skb));
-	if (nsk != NULL) {
-		if (nsk->sk_state != DCCP_TIME_WAIT) {
-			bh_lock_sock(nsk);
-			return nsk;
-		}
-		inet_twsk_put(inet_twsk(nsk));
-		return NULL;
-	}
-
-	return sk;
-}
-
 static struct dst_entry* dccp_v4_route_skb(struct net *net, struct sock *sk,
 					   struct sk_buff *skb)
 {
@@ -705,18 +675,6 @@ int dccp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 	 * NOTE: the check for the packet types is done in
 	 *	 dccp_rcv_state_process
 	 */
-	if (sk->sk_state == DCCP_LISTEN) {
-		struct sock *nsk = dccp_v4_hnd_req(sk, skb);
-
-		if (nsk == NULL)
-			goto discard;
-
-		if (nsk != sk) {
-			if (dccp_child_process(sk, nsk, skb))
-				goto reset;
-			return 0;
-		}
-	}
 
 	if (dccp_rcv_state_process(sk, skb, dh, skb->len))
 		goto reset;
@@ -724,7 +682,6 @@ int dccp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 
 reset:
 	dccp_v4_ctl_send_reset(sk, skb);
-discard:
 	kfree_skb(skb);
 	return 0;
 }
@@ -868,6 +825,27 @@ static int dccp_v4_rcv(struct sk_buff *skb)
 		goto no_dccp_socket;
 	}
 
+	if (sk->sk_state == DCCP_NEW_SYN_RECV) {
+		struct request_sock *req = inet_reqsk(sk);
+		struct sock *nsk = NULL;
+
+		sk = req->rsk_listener;
+		if (sk->sk_state == DCCP_LISTEN)
+			nsk = dccp_check_req(sk, skb, req);
+		if (!nsk) {
+			reqsk_put(req);
+			goto discard_it;
+		}
+		if (nsk == sk) {
+			sock_hold(sk);
+			reqsk_put(req);
+		} else if (dccp_child_process(sk, nsk, skb)) {
+			dccp_v4_ctl_send_reset(sk, skb);
+			goto discard_it;
+		} else {
+			return 0;
+		}
+	}
 	/*
 	 * RFC 4340, sec. 9.2.1: Minimum Checksum Coverage
 	 *	o if MinCsCov = 0, only packets with CsCov = 0 are accepted
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index e8753aa3b7a4..1361a3f45df7 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -290,37 +290,6 @@ static struct request_sock_ops dccp6_request_sock_ops = {
 	.syn_ack_timeout = dccp_syn_ack_timeout,
 };
 
-static struct sock *dccp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
-{
-	const struct dccp_hdr *dh = dccp_hdr(skb);
-	const struct ipv6hdr *iph = ipv6_hdr(skb);
-	struct request_sock *req;
-	struct sock *nsk;
-
-	req = inet6_csk_search_req(sk, dh->dccph_sport, &iph->saddr,
-				   &iph->daddr, inet6_iif(skb));
-	if (req) {
-		nsk = dccp_check_req(sk, skb, req);
-		if (!nsk)
-			reqsk_put(req);
-		return nsk;
-	}
-	nsk = __inet6_lookup_established(sock_net(sk), &dccp_hashinfo,
-					 &iph->saddr, dh->dccph_sport,
-					 &iph->daddr, ntohs(dh->dccph_dport),
-					 inet6_iif(skb));
-	if (nsk != NULL) {
-		if (nsk->sk_state != DCCP_TIME_WAIT) {
-			bh_lock_sock(nsk);
-			return nsk;
-		}
-		inet_twsk_put(inet_twsk(nsk));
-		return NULL;
-	}
-
-	return sk;
-}
-
 static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 {
 	struct request_sock *req;
@@ -398,7 +367,7 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 	if (dccp_v6_send_response(sk, req))
 		goto drop_and_free;
 
-	inet6_csk_reqsk_queue_hash_add(sk, req, DCCP_TIMEOUT_INIT);
+	inet_csk_reqsk_queue_hash_add(sk, req, DCCP_TIMEOUT_INIT);
 	return 0;
 
 drop_and_free:
@@ -641,24 +610,6 @@ static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 	 * NOTE: the check for the packet types is done in
 	 *	 dccp_rcv_state_process
 	 */
-	if (sk->sk_state == DCCP_LISTEN) {
-		struct sock *nsk = dccp_v6_hnd_req(sk, skb);
-
-		if (nsk == NULL)
-			goto discard;
-		/*
-		 * Queue it on the new socket if the new socket is active,
-		 * otherwise we just shortcircuit this and continue with
-		 * the new socket..
-		 */
-		if (nsk != sk) {
-			if (dccp_child_process(sk, nsk, skb))
-				goto reset;
-			if (opt_skb != NULL)
-				__kfree_skb(opt_skb);
-			return 0;
-		}
-	}
 
 	if (dccp_rcv_state_process(sk, skb, dccp_hdr(skb), skb->len))
 		goto reset;
@@ -732,6 +683,27 @@ static int dccp_v6_rcv(struct sk_buff *skb)
 		goto no_dccp_socket;
 	}
 
+	if (sk->sk_state == DCCP_NEW_SYN_RECV) {
+		struct request_sock *req = inet_reqsk(sk);
+		struct sock *nsk = NULL;
+
+		sk = req->rsk_listener;
+		if (sk->sk_state == DCCP_LISTEN)
+			nsk = dccp_check_req(sk, skb, req);
+		if (!nsk) {
+			reqsk_put(req);
+			goto discard_it;
+		}
+		if (nsk == sk) {
+			sock_hold(sk);
+			reqsk_put(req);
+		} else if (dccp_child_process(sk, nsk, skb)) {
+			dccp_v6_ctl_send_reset(sk, skb);
+			goto discard_it;
+		} else {
+			return 0;
+		}
+	}
 	/*
 	 * RFC 4340, sec. 9.2.1: Minimum Checksum Coverage
 	 *	o if MinCsCov = 0, only packets with CsCov = 0 are accepted
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index e62f04775c93..80904df02187 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -476,65 +476,12 @@ no_route:
 }
 EXPORT_SYMBOL_GPL(inet_csk_route_child_sock);
 
-static inline u32 inet_synq_hash(const __be32 raddr, const __be16 rport,
-				 const u32 rnd, const u32 synq_hsize)
-{
-	return jhash_2words((__force u32)raddr, (__force u32)rport, rnd) & (synq_hsize - 1);
-}
-
 #if IS_ENABLED(CONFIG_IPV6)
 #define AF_INET_FAMILY(fam) ((fam) == AF_INET)
 #else
 #define AF_INET_FAMILY(fam) true
 #endif
 
-/* Note: this is temporary :
- * req sock will no longer be in listener hash table
-*/
-struct request_sock *inet_csk_search_req(struct sock *sk,
-					 const __be16 rport,
-					 const __be32 raddr,
-					 const __be32 laddr)
-{
-	struct inet_connection_sock *icsk = inet_csk(sk);
-	struct listen_sock *lopt = icsk->icsk_accept_queue.listen_opt;
-	struct request_sock *req;
-	u32 hash = inet_synq_hash(raddr, rport, lopt->hash_rnd,
-				  lopt->nr_table_entries);
-
-	spin_lock(&icsk->icsk_accept_queue.syn_wait_lock);
-	for (req = lopt->syn_table[hash]; req != NULL; req = req->dl_next) {
-		const struct inet_request_sock *ireq = inet_rsk(req);
-
-		if (ireq->ir_rmt_port == rport &&
-		    ireq->ir_rmt_addr == raddr &&
-		    ireq->ir_loc_addr == laddr &&
-		    AF_INET_FAMILY(req->rsk_ops->family)) {
-			atomic_inc(&req->rsk_refcnt);
-			WARN_ON(req->sk);
-			break;
-		}
-	}
-	spin_unlock(&icsk->icsk_accept_queue.syn_wait_lock);
-
-	return req;
-}
-EXPORT_SYMBOL_GPL(inet_csk_search_req);
-
-void inet_csk_reqsk_queue_hash_add(struct sock *sk, struct request_sock *req,
-				   unsigned long timeout)
-{
-	struct inet_connection_sock *icsk = inet_csk(sk);
-	struct listen_sock *lopt = icsk->icsk_accept_queue.listen_opt;
-	const u32 h = inet_synq_hash(inet_rsk(req)->ir_rmt_addr,
-				     inet_rsk(req)->ir_rmt_port,
-				     lopt->hash_rnd, lopt->nr_table_entries);
-
-	reqsk_queue_hash_req(&icsk->icsk_accept_queue, h, req, timeout);
-	inet_csk_reqsk_queue_added(sk);
-}
-EXPORT_SYMBOL_GPL(inet_csk_reqsk_queue_hash_add);
-
 /* Only thing we need from tcp.h */
 extern int sysctl_tcp_synack_retries;
 
@@ -571,26 +518,20 @@ int inet_rtx_syn_ack(const struct sock *parent, struct request_sock *req)
 }
 EXPORT_SYMBOL(inet_rtx_syn_ack);
 
-/* return true if req was found in the syn_table[] */
+/* return true if req was found in the ehash table */
 static bool reqsk_queue_unlink(struct request_sock_queue *queue,
 			       struct request_sock *req)
 {
-	struct listen_sock *lopt = queue->listen_opt;
-	struct request_sock **prev;
-	bool found = false;
+	struct inet_hashinfo *hashinfo = req_to_sk(req)->sk_prot->h.hashinfo;
+	spinlock_t *lock;
+	bool found;
 
-	spin_lock(&queue->syn_wait_lock);
+	lock = inet_ehash_lockp(hashinfo, req->rsk_hash);
 
-	for (prev = &lopt->syn_table[req->rsk_hash]; *prev != NULL;
-	     prev = &(*prev)->dl_next) {
-		if (*prev == req) {
-			*prev = req->dl_next;
-			found = true;
-			break;
-		}
-	}
+	spin_lock(lock);
+	found = __sk_nulls_del_node_init_rcu(req_to_sk(req));
+	spin_unlock(lock);
 
-	spin_unlock(&queue->syn_wait_lock);
 	if (timer_pending(&req->rsk_timer) && del_timer_sync(&req->rsk_timer))
 		reqsk_put(req);
 	return found;
@@ -616,10 +557,8 @@ static void reqsk_timer_handler(unsigned long data)
 	int max_retries, thresh;
 	u8 defer_accept;
 
-	if (sk_listener->sk_state != TCP_LISTEN || !lopt) {
-		reqsk_put(req);
-		return;
-	}
+	if (sk_listener->sk_state != TCP_LISTEN || !lopt)
+		goto drop;
 
 	max_retries = icsk->icsk_syn_retries ? : sysctl_tcp_synack_retries;
 	thresh = max_retries;
@@ -669,36 +608,36 @@ static void reqsk_timer_handler(unsigned long data)
 		mod_timer_pinned(&req->rsk_timer, jiffies + timeo);
 		return;
 	}
+drop:
 	inet_csk_reqsk_queue_drop(sk_listener, req);
 	reqsk_put(req);
 }
 
-void reqsk_queue_hash_req(struct request_sock_queue *queue,
-			  u32 hash, struct request_sock *req,
-			  unsigned long timeout)
+static void reqsk_queue_hash_req(struct request_sock *req,
+				 unsigned long timeout)
 {
-	struct listen_sock *lopt = queue->listen_opt;
-
 	req->num_retrans = 0;
 	req->num_timeout = 0;
 	req->sk = NULL;
 
 	setup_timer(&req->rsk_timer, reqsk_timer_handler, (unsigned long)req);
 	mod_timer_pinned(&req->rsk_timer, jiffies + timeout);
-	req->rsk_hash = hash;
 
+	inet_ehash_insert(req_to_sk(req), NULL);
 	/* before letting lookups find us, make sure all req fields
 	 * are committed to memory and refcnt initialized.
 	 */
 	smp_wmb();
 	atomic_set(&req->rsk_refcnt, 2);
+}
 
-	spin_lock(&queue->syn_wait_lock);
-	req->dl_next = lopt->syn_table[hash];
-	lopt->syn_table[hash] = req;
-	spin_unlock(&queue->syn_wait_lock);
+void inet_csk_reqsk_queue_hash_add(struct sock *sk, struct request_sock *req,
+				   unsigned long timeout)
+{
+	reqsk_queue_hash_req(req, timeout);
+	inet_csk_reqsk_queue_added(sk);
 }
-EXPORT_SYMBOL(reqsk_queue_hash_req);
+EXPORT_SYMBOL_GPL(inet_csk_reqsk_queue_hash_add);
 
 /**
  *	inet_csk_clone_lock - clone an inet socket, and lock its clone
diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c
index 0ac1d68dc8a6..ab9f8a66615d 100644
--- a/net/ipv4/inet_diag.c
+++ b/net/ipv4/inet_diag.c
@@ -730,91 +730,21 @@ static void twsk_build_assert(void)
 #endif
 }
 
-static int inet_diag_dump_reqs(struct sk_buff *skb, struct sock *sk,
-			       struct netlink_callback *cb,
-			       const struct inet_diag_req_v2 *r,
-			       const struct nlattr *bc)
-{
-	struct inet_connection_sock *icsk = inet_csk(sk);
-	struct inet_sock *inet = inet_sk(sk);
-	struct inet_diag_entry entry;
-	int j, s_j, reqnum, s_reqnum;
-	struct listen_sock *lopt;
-	int err = 0;
-
-	s_j = cb->args[3];
-	s_reqnum = cb->args[4];
-
-	if (s_j > 0)
-		s_j--;
-
-	entry.family = sk->sk_family;
-
-	spin_lock(&icsk->icsk_accept_queue.syn_wait_lock);
-
-	lopt = icsk->icsk_accept_queue.listen_opt;
-	if (!lopt || !reqsk_queue_len(&icsk->icsk_accept_queue))
-		goto out;
-
-	if (bc) {
-		entry.sport = inet->inet_num;
-		entry.userlocks = sk->sk_userlocks;
-	}
-
-	for (j = s_j; j < lopt->nr_table_entries; j++) {
-		struct request_sock *req, *head = lopt->syn_table[j];
-
-		reqnum = 0;
-		for (req = head; req; reqnum++, req = req->dl_next) {
-			struct inet_request_sock *ireq = inet_rsk(req);
-
-			if (reqnum < s_reqnum)
-				continue;
-			if (r->id.idiag_dport != ireq->ir_rmt_port &&
-			    r->id.idiag_dport)
-				continue;
-
-			if (bc) {
-				/* Note: entry.sport and entry.userlocks are already set */
-				entry_fill_addrs(&entry, req_to_sk(req));
-				entry.dport = ntohs(ireq->ir_rmt_port);
-
-				if (!inet_diag_bc_run(bc, &entry))
-					continue;
-			}
-
-			err = inet_req_diag_fill(req_to_sk(req), skb,
-						 NETLINK_CB(cb->skb).portid,
-						 cb->nlh->nlmsg_seq,
-						 NLM_F_MULTI, cb->nlh);
-			if (err < 0) {
-				cb->args[3] = j + 1;
-				cb->args[4] = reqnum;
-				goto out;
-			}
-		}
-
-		s_reqnum = 0;
-	}
-
-out:
-	spin_unlock(&icsk->icsk_accept_queue.syn_wait_lock);
-
-	return err;
-}
-
 void inet_diag_dump_icsk(struct inet_hashinfo *hashinfo, struct sk_buff *skb,
 			 struct netlink_callback *cb,
 			 const struct inet_diag_req_v2 *r, struct nlattr *bc)
 {
 	struct net *net = sock_net(skb->sk);
 	int i, num, s_i, s_num;
+	u32 idiag_states = r->idiag_states;
 
+	if (idiag_states & TCPF_SYN_RECV)
+		idiag_states |= TCPF_NEW_SYN_RECV;
 	s_i = cb->args[1];
 	s_num = num = cb->args[2];
 
 	if (cb->args[0] == 0) {
-		if (!(r->idiag_states & (TCPF_LISTEN | TCPF_SYN_RECV)))
+		if (!(idiag_states & TCPF_LISTEN))
 			goto skip_listen_ht;
 
 		for (i = s_i; i < INET_LHTABLE_SIZE; i++) {
@@ -844,21 +774,11 @@ void inet_diag_dump_icsk(struct inet_hashinfo *hashinfo, struct sk_buff *skb,
 				    r->id.idiag_sport)
 					goto next_listen;
 
-				if (!(r->idiag_states & TCPF_LISTEN) ||
-				    r->id.idiag_dport ||
+				if (r->id.idiag_dport ||
 				    cb->args[3] > 0)
-					goto syn_recv;
-
-				if (inet_csk_diag_dump(sk, skb, cb, r, bc) < 0) {
-					spin_unlock_bh(&ilb->lock);
-					goto done;
-				}
-
-syn_recv:
-				if (!(r->idiag_states & TCPF_SYN_RECV))
 					goto next_listen;
 
-				if (inet_diag_dump_reqs(skb, sk, cb, r, bc) < 0) {
+				if (inet_csk_diag_dump(sk, skb, cb, r, bc) < 0) {
 					spin_unlock_bh(&ilb->lock);
 					goto done;
 				}
@@ -879,7 +799,7 @@ skip_listen_ht:
 		s_i = num = s_num = 0;
 	}
 
-	if (!(r->idiag_states & ~(TCPF_LISTEN | TCPF_SYN_RECV)))
+	if (!(idiag_states & ~TCPF_LISTEN))
 		goto out;
 
 	for (i = s_i; i <= hashinfo->ehash_mask; i++) {
@@ -906,7 +826,7 @@ skip_listen_ht:
 				goto next_normal;
 			state = (sk->sk_state == TCP_TIME_WAIT) ?
 				inet_twsk(sk)->tw_substate : sk->sk_state;
-			if (!(r->idiag_states & (1 << state)))
+			if (!(idiag_states & (1 << state)))
 				goto next_normal;
 			if (r->sdiag_family != AF_UNSPEC &&
 			    sk->sk_family != r->sdiag_family)
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 56742e995dd3..bed8886a4b6c 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -398,14 +398,18 @@ static u32 inet_sk_port_offset(const struct sock *sk)
 					  inet->inet_dport);
 }
 
-void __inet_hash_nolisten(struct sock *sk, struct sock *osk)
+/* insert a socket into ehash, and eventually remove another one
+ * (The another one can be a SYN_RECV or TIMEWAIT
+ */
+int inet_ehash_insert(struct sock *sk, struct sock *osk)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
 	struct hlist_nulls_head *list;
 	struct inet_ehash_bucket *head;
 	spinlock_t *lock;
+	int ret = 0;
 
-	WARN_ON(!sk_unhashed(sk));
+	WARN_ON_ONCE(!sk_unhashed(sk));
 
 	sk->sk_hash = sk_ehashfn(sk);
 	head = inet_ehash_bucket(hashinfo, sk->sk_hash);
@@ -419,6 +423,12 @@ void __inet_hash_nolisten(struct sock *sk, struct sock *osk)
 		sk_nulls_del_node_init_rcu(osk);
 	}
 	spin_unlock(lock);
+	return ret;
+}
+
+void __inet_hash_nolisten(struct sock *sk, struct sock *osk)
+{
+	inet_ehash_insert(sk, osk);
 	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 }
 EXPORT_SYMBOL_GPL(__inet_hash_nolisten);
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 6b97b5f6457c..729ceb5f63c6 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -284,6 +284,10 @@ bool cookie_ecn_ok(const struct tcp_options_received *tcp_opt,
 }
 EXPORT_SYMBOL(cookie_ecn_ok);
 
+/* On input, sk is a listener.
+ * Output is listener if incoming packet would not create a child
+ *           NULL if memory could not be allocated.
+ */
 struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
 {
 	struct ip_options *opt = &TCP_SKB_CB(skb)->header.h4.opt;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 8b0ce73c2049..a56912772354 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6241,7 +6241,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 			goto drop_and_free;
 
 		tcp_rsk(req)->tfo_listener = false;
-		af_ops->queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
+		inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
 	}
 	tcp_reqsk_record_syn(sk, req, skb);
 
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index a33101616215..bfe9d39ee87d 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1224,7 +1224,6 @@ static const struct tcp_request_sock_ops tcp_request_sock_ipv4_ops = {
 	.route_req	=	tcp_v4_route_req,
 	.init_seq	=	tcp_v4_init_sequence,
 	.send_synack	=	tcp_v4_send_synack,
-	.queue_hash_add =	inet_csk_reqsk_queue_hash_add,
 };
 
 int tcp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
@@ -1343,34 +1342,11 @@ put_and_exit:
 }
 EXPORT_SYMBOL(tcp_v4_syn_recv_sock);
 
-static struct sock *tcp_v4_hnd_req(struct sock *sk, struct sk_buff *skb)
+static struct sock *tcp_v4_cookie_check(struct sock *sk, struct sk_buff *skb)
 {
+#ifdef CONFIG_SYN_COOKIES
 	const struct tcphdr *th = tcp_hdr(skb);
-	const struct iphdr *iph = ip_hdr(skb);
-	struct request_sock *req;
-	struct sock *nsk;
-
-	req = inet_csk_search_req(sk, th->source, iph->saddr, iph->daddr);
-	if (req) {
-		nsk = tcp_check_req(sk, skb, req, false);
-		if (!nsk || nsk == sk)
-			reqsk_put(req);
-		return nsk;
-	}
-
-	nsk = inet_lookup_established(sock_net(sk), &tcp_hashinfo, iph->saddr,
-			th->source, iph->daddr, th->dest, inet_iif(skb));
-
-	if (nsk) {
-		if (nsk->sk_state != TCP_TIME_WAIT) {
-			bh_lock_sock(nsk);
-			return nsk;
-		}
-		inet_twsk_put(inet_twsk(nsk));
-		return NULL;
-	}
 
-#ifdef CONFIG_SYN_COOKIES
 	if (!th->syn)
 		sk = cookie_v4_check(sk, skb);
 #endif
@@ -1409,10 +1385,10 @@ int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 		goto csum_err;
 
 	if (sk->sk_state == TCP_LISTEN) {
-		struct sock *nsk = tcp_v4_hnd_req(sk, skb);
+		struct sock *nsk = tcp_v4_cookie_check(sk, skb);
+
 		if (!nsk)
 			goto discard;
-
 		if (nsk != sk) {
 			sock_rps_save_rxhash(nsk, skb);
 			sk_mark_napi_id(nsk, skb);
@@ -1603,6 +1579,29 @@ process:
 	if (sk->sk_state == TCP_TIME_WAIT)
 		goto do_time_wait;
 
+	if (sk->sk_state == TCP_NEW_SYN_RECV) {
+		struct request_sock *req = inet_reqsk(sk);
+		struct sock *nsk = NULL;
+
+		sk = req->rsk_listener;
+		if (tcp_v4_inbound_md5_hash(sk, skb))
+			goto discard_and_relse;
+		if (sk->sk_state == TCP_LISTEN)
+			nsk = tcp_check_req(sk, skb, req, false);
+		if (!nsk) {
+			reqsk_put(req);
+			goto discard_it;
+		}
+		if (nsk == sk) {
+			sock_hold(sk);
+			reqsk_put(req);
+		} else if (tcp_child_process(sk, nsk, skb)) {
+			tcp_v4_send_reset(nsk, skb);
+			goto discard_it;
+		} else {
+			return 0;
+		}
+	}
 	if (unlikely(iph->ttl < inet_sk(sk)->min_ttl)) {
 		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
 		goto discard_and_relse;
@@ -1830,35 +1829,7 @@ static void *listening_get_next(struct seq_file *seq, void *cur)
 	++st->num;
 	++st->offset;
 
-	if (st->state == TCP_SEQ_STATE_OPENREQ) {
-		struct request_sock *req = cur;
-
-		icsk = inet_csk(st->syn_wait_sk);
-		req = req->dl_next;
-		while (1) {
-			while (req) {
-				if (req->rsk_ops->family == st->family) {
-					cur = req;
-					goto out;
-				}
-				req = req->dl_next;
-			}
-			if (++st->sbucket >= icsk->icsk_accept_queue.listen_opt->nr_table_entries)
-				break;
-get_req:
-			req = icsk->icsk_accept_queue.listen_opt->syn_table[st->sbucket];
-		}
-		sk	  = sk_nulls_next(st->syn_wait_sk);
-		st->state = TCP_SEQ_STATE_LISTENING;
-		spin_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
-	} else {
-		icsk = inet_csk(sk);
-		spin_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
-		if (reqsk_queue_len(&icsk->icsk_accept_queue))
-			goto start_req;
-		spin_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
-		sk = sk_nulls_next(sk);
-	}
+	sk = sk_nulls_next(sk);
 get_sk:
 	sk_nulls_for_each_from(sk, node) {
 		if (!net_eq(sock_net(sk), net))
@@ -1868,15 +1839,6 @@ get_sk:
 			goto out;
 		}
 		icsk = inet_csk(sk);
-		spin_lock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
-		if (reqsk_queue_len(&icsk->icsk_accept_queue)) {
-start_req:
-			st->syn_wait_sk = sk;
-			st->state	= TCP_SEQ_STATE_OPENREQ;
-			st->sbucket	= 0;
-			goto get_req;
-		}
-		spin_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
 	}
 	spin_unlock_bh(&ilb->lock);
 	st->offset = 0;
@@ -2008,7 +1970,6 @@ static void *tcp_seek_last_pos(struct seq_file *seq)
 	void *rc = NULL;
 
 	switch (st->state) {
-	case TCP_SEQ_STATE_OPENREQ:
 	case TCP_SEQ_STATE_LISTENING:
 		if (st->bucket >= INET_LHTABLE_SIZE)
 			break;
@@ -2067,7 +2028,6 @@ static void *tcp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 	}
 
 	switch (st->state) {
-	case TCP_SEQ_STATE_OPENREQ:
 	case TCP_SEQ_STATE_LISTENING:
 		rc = listening_get_next(seq, v);
 		if (!rc) {
@@ -2092,11 +2052,6 @@ static void tcp_seq_stop(struct seq_file *seq, void *v)
 	struct tcp_iter_state *st = seq->private;
 
 	switch (st->state) {
-	case TCP_SEQ_STATE_OPENREQ:
-		if (v) {
-			struct inet_connection_sock *icsk = inet_csk(st->syn_wait_sk);
-			spin_unlock_bh(&icsk->icsk_accept_queue.syn_wait_lock);
-		}
 	case TCP_SEQ_STATE_LISTENING:
 		if (v != SEQ_START_TOKEN)
 			spin_unlock_bh(&tcp_hashinfo.listening_hash[st->bucket].lock);
@@ -2269,18 +2224,12 @@ static int tcp4_seq_show(struct seq_file *seq, void *v)
 	}
 	st = seq->private;
 
-	switch (st->state) {
-	case TCP_SEQ_STATE_LISTENING:
-	case TCP_SEQ_STATE_ESTABLISHED:
-		if (sk->sk_state == TCP_TIME_WAIT)
-			get_timewait4_sock(v, seq, st->num);
-		else
-			get_tcp4_sock(v, seq, st->num);
-		break;
-	case TCP_SEQ_STATE_OPENREQ:
+	if (sk->sk_state == TCP_TIME_WAIT)
+		get_timewait4_sock(v, seq, st->num);
+	else if (sk->sk_state == TCP_NEW_SYN_RECV)
 		get_openreq4(v, seq, st->num);
-		break;
-	}
+	else
+		get_tcp4_sock(v, seq, st->num);
 out:
 	seq_pad(seq, '\n');
 	return 0;
diff --git a/net/ipv6/inet6_connection_sock.c b/net/ipv6/inet6_connection_sock.c
index ea915aa5e4e2..5d1c7cee2cb2 100644
--- a/net/ipv6/inet6_connection_sock.c
+++ b/net/ipv6/inet6_connection_sock.c
@@ -94,73 +94,6 @@ struct dst_entry *inet6_csk_route_req(const struct sock *sk,
 }
 EXPORT_SYMBOL(inet6_csk_route_req);
 
-/*
- * request_sock (formerly open request) hash tables.
- */
-static u32 inet6_synq_hash(const struct in6_addr *raddr, const __be16 rport,
-			   const u32 rnd, const u32 synq_hsize)
-{
-	u32 c;
-
-	c = jhash_3words((__force u32)raddr->s6_addr32[0],
-			 (__force u32)raddr->s6_addr32[1],
-			 (__force u32)raddr->s6_addr32[2],
-			 rnd);
-
-	c = jhash_2words((__force u32)raddr->s6_addr32[3],
-			 (__force u32)rport,
-			 c);
-
-	return c & (synq_hsize - 1);
-}
-
-struct request_sock *inet6_csk_search_req(struct sock *sk,
-					  const __be16 rport,
-					  const struct in6_addr *raddr,
-					  const struct in6_addr *laddr,
-					  const int iif)
-{
-	struct inet_connection_sock *icsk = inet_csk(sk);
-	struct listen_sock *lopt = icsk->icsk_accept_queue.listen_opt;
-	struct request_sock *req;
-	u32 hash = inet6_synq_hash(raddr, rport, lopt->hash_rnd,
-				   lopt->nr_table_entries);
-
-	spin_lock(&icsk->icsk_accept_queue.syn_wait_lock);
-	for (req = lopt->syn_table[hash]; req != NULL; req = req->dl_next) {
-		const struct inet_request_sock *ireq = inet_rsk(req);
-
-		if (ireq->ir_rmt_port == rport &&
-		    req->rsk_ops->family == AF_INET6 &&
-		    ipv6_addr_equal(&ireq->ir_v6_rmt_addr, raddr) &&
-		    ipv6_addr_equal(&ireq->ir_v6_loc_addr, laddr) &&
-		    (!ireq->ir_iif || ireq->ir_iif == iif)) {
-			atomic_inc(&req->rsk_refcnt);
-			WARN_ON(req->sk != NULL);
-			break;
-		}
-	}
-	spin_unlock(&icsk->icsk_accept_queue.syn_wait_lock);
-
-	return req;
-}
-EXPORT_SYMBOL_GPL(inet6_csk_search_req);
-
-void inet6_csk_reqsk_queue_hash_add(struct sock *sk,
-				    struct request_sock *req,
-				    const unsigned long timeout)
-{
-	struct inet_connection_sock *icsk = inet_csk(sk);
-	struct listen_sock *lopt = icsk->icsk_accept_queue.listen_opt;
-	const u32 h = inet6_synq_hash(&inet_rsk(req)->ir_v6_rmt_addr,
-				      inet_rsk(req)->ir_rmt_port,
-				      lopt->hash_rnd, lopt->nr_table_entries);
-
-	reqsk_queue_hash_req(&icsk->icsk_accept_queue, h, req, timeout);
-	inet_csk_reqsk_queue_added(sk);
-}
-EXPORT_SYMBOL_GPL(inet6_csk_reqsk_queue_hash_add);
-
 void inet6_csk_addr2sockaddr(struct sock *sk, struct sockaddr *uaddr)
 {
 	struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *) uaddr;
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index cadb44a2d34e..a215614cfb2b 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -727,7 +727,6 @@ static const struct tcp_request_sock_ops tcp_request_sock_ipv6_ops = {
 	.route_req	=	tcp_v6_route_req,
 	.init_seq	=	tcp_v6_init_sequence,
 	.send_synack	=	tcp_v6_send_synack,
-	.queue_hash_add =	inet6_csk_reqsk_queue_hash_add,
 };
 
 static void tcp_v6_send_response(const struct sock *sk, struct sk_buff *skb, u32 seq,
@@ -938,37 +937,11 @@ static void tcp_v6_reqsk_send_ack(const struct sock *sk, struct sk_buff *skb,
 }
 
 
-static struct sock *tcp_v6_hnd_req(struct sock *sk, struct sk_buff *skb)
+static struct sock *tcp_v6_cookie_check(struct sock *sk, struct sk_buff *skb)
 {
+#ifdef CONFIG_SYN_COOKIES
 	const struct tcphdr *th = tcp_hdr(skb);
-	struct request_sock *req;
-	struct sock *nsk;
-
-	/* Find possible connection requests. */
-	req = inet6_csk_search_req(sk, th->source,
-				   &ipv6_hdr(skb)->saddr,
-				   &ipv6_hdr(skb)->daddr, tcp_v6_iif(skb));
-	if (req) {
-		nsk = tcp_check_req(sk, skb, req, false);
-		if (!nsk || nsk == sk)
-			reqsk_put(req);
-		return nsk;
-	}
-	nsk = __inet6_lookup_established(sock_net(sk), &tcp_hashinfo,
-					 &ipv6_hdr(skb)->saddr, th->source,
-					 &ipv6_hdr(skb)->daddr, ntohs(th->dest),
-					 tcp_v6_iif(skb));
-
-	if (nsk) {
-		if (nsk->sk_state != TCP_TIME_WAIT) {
-			bh_lock_sock(nsk);
-			return nsk;
-		}
-		inet_twsk_put(inet_twsk(nsk));
-		return NULL;
-	}
 
-#ifdef CONFIG_SYN_COOKIES
 	if (!th->syn)
 		sk = cookie_v6_check(sk, skb);
 #endif
@@ -1258,15 +1231,11 @@ static int tcp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 		goto csum_err;
 
 	if (sk->sk_state == TCP_LISTEN) {
-		struct sock *nsk = tcp_v6_hnd_req(sk, skb);
+		struct sock *nsk = tcp_v6_cookie_check(sk, skb);
+
 		if (!nsk)
 			goto discard;
 
-		/*
-		 * Queue it on the new socket if the new socket is active,
-		 * otherwise we just shortcircuit this and continue with
-		 * the new socket..
-		 */
 		if (nsk != sk) {
 			sock_rps_save_rxhash(nsk, skb);
 			sk_mark_napi_id(nsk, skb);
@@ -1402,6 +1371,33 @@ process:
 	if (sk->sk_state == TCP_TIME_WAIT)
 		goto do_time_wait;
 
+	if (sk->sk_state == TCP_NEW_SYN_RECV) {
+		struct request_sock *req = inet_reqsk(sk);
+		struct sock *nsk = NULL;
+
+		sk = req->rsk_listener;
+		tcp_v6_fill_cb(skb, hdr, th);
+		if (tcp_v6_inbound_md5_hash(sk, skb)) {
+			reqsk_put(req);
+			goto discard_it;
+		}
+		if (sk->sk_state == TCP_LISTEN)
+			nsk = tcp_check_req(sk, skb, req, false);
+		if (!nsk) {
+			reqsk_put(req);
+			goto discard_it;
+		}
+		if (nsk == sk) {
+			sock_hold(sk);
+			reqsk_put(req);
+			tcp_v6_restore_cb(skb);
+		} else if (tcp_child_process(sk, nsk, skb)) {
+			tcp_v6_send_reset(nsk, skb);
+			goto discard_it;
+		} else {
+			return 0;
+		}
+	}
 	if (hdr->hop_limit < inet6_sk(sk)->min_hopcount) {
 		NET_INC_STATS_BH(net, LINUX_MIB_TCPMINTTLDROP);
 		goto discard_and_relse;
@@ -1765,18 +1761,12 @@ static int tcp6_seq_show(struct seq_file *seq, void *v)
 	}
 	st = seq->private;
 
-	switch (st->state) {
-	case TCP_SEQ_STATE_LISTENING:
-	case TCP_SEQ_STATE_ESTABLISHED:
-		if (sk->sk_state == TCP_TIME_WAIT)
-			get_timewait6_sock(seq, v, st->num);
-		else
-			get_tcp6_sock(seq, v, st->num);
-		break;
-	case TCP_SEQ_STATE_OPENREQ:
+	if (sk->sk_state == TCP_TIME_WAIT)
+		get_timewait6_sock(seq, v, st->num);
+	else if (sk->sk_state == TCP_NEW_SYN_RECV)
 		get_openreq6(seq, v, st->num);
-		break;
-	}
+	else
+		get_tcp6_sock(seq, v, st->num);
 out:
 	return 0;
 }
-- 
cgit v1.2.3


From 70da268b569d32a9fddeea85dc18043de9d89f89 Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Thu, 8 Oct 2015 19:33:21 -0700
Subject: net: SO_INCOMING_CPU setsockopt() support

SO_INCOMING_CPU as added in commit 2c8c56e15df3 was a getsockopt() command
to fetch incoming cpu handling a particular TCP flow after accept()

This commits adds setsockopt() support and extends SO_REUSEPORT selection
logic : If a TCP listener or UDP socket has this option set, a packet is
delivered to this socket only if CPU handling the packet matches the specified
one.

This allows to build very efficient TCP servers, using one listener per
RX queue, as the associated TCP listener should only accept flows handled
in softirq by the same cpu.
This provides optimal NUMA behavior and keep cpu caches hot.

Note that __inet_lookup_listener() still has to iterate over the list of
all listeners. Following patch puts sk_refcnt in a different cache line
to let this iteration hit only shared and read mostly cache lines.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/sock.h          | 10 ++++------
 net/core/sock.c             |  5 +++++
 net/ipv4/inet_hashtables.c  |  2 ++
 net/ipv4/udp.c              |  6 +++++-
 net/ipv6/inet6_hashtables.c |  2 ++
 net/ipv6/udp.c              | 11 +++++++----
 6 files changed, 25 insertions(+), 11 deletions(-)

(limited to 'net/ipv4/inet_hashtables.c')

diff --git a/include/net/sock.h b/include/net/sock.h
index 9322cafd191b..cf54739f30d5 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -150,6 +150,7 @@ typedef __u64 __bitwise __addrpair;
  *	@skc_node: main hash linkage for various protocol lookup tables
  *	@skc_nulls_node: main hash linkage for TCP/UDP/UDP-Lite protocol
  *	@skc_tx_queue_mapping: tx queue number for this connection
+ *	@skc_incoming_cpu: record/match cpu processing incoming packets
  *	@skc_refcnt: reference count
  *
  *	This is the minimal network layer representation of sockets, the header
@@ -212,6 +213,8 @@ struct sock_common {
 		struct hlist_nulls_node skc_nulls_node;
 	};
 	int			skc_tx_queue_mapping;
+	int			skc_incoming_cpu;
+
 	atomic_t		skc_refcnt;
 	/* private: */
 	int                     skc_dontcopy_end[0];
@@ -274,7 +277,6 @@ struct cg_proto;
   *	@sk_rcvtimeo: %SO_RCVTIMEO setting
   *	@sk_sndtimeo: %SO_SNDTIMEO setting
   *	@sk_rxhash: flow hash received from netif layer
-  *	@sk_incoming_cpu: record cpu processing incoming packets
   *	@sk_txhash: computed flow hash for use on transmit
   *	@sk_filter: socket filtering instructions
   *	@sk_timer: sock cleanup timer
@@ -331,6 +333,7 @@ struct sock {
 #define sk_v6_daddr		__sk_common.skc_v6_daddr
 #define sk_v6_rcv_saddr	__sk_common.skc_v6_rcv_saddr
 #define sk_cookie		__sk_common.skc_cookie
+#define sk_incoming_cpu		__sk_common.skc_incoming_cpu
 
 	socket_lock_t		sk_lock;
 	struct sk_buff_head	sk_receive_queue;
@@ -353,11 +356,6 @@ struct sock {
 #ifdef CONFIG_RPS
 	__u32			sk_rxhash;
 #endif
-	u16			sk_incoming_cpu;
-	/* 16bit hole
-	 * Warned : sk_incoming_cpu can be set from softirq,
-	 * Do not use this hole without fully understanding possible issues.
-	 */
 
 	__u32			sk_txhash;
 #ifdef CONFIG_NET_RX_BUSY_POLL
diff --git a/net/core/sock.c b/net/core/sock.c
index 33957776cc1a..dcc7d62654d5 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -988,6 +988,10 @@ set_rcvbuf:
 					 sk->sk_max_pacing_rate);
 		break;
 
+	case SO_INCOMING_CPU:
+		sk->sk_incoming_cpu = val;
+		break;
+
 	default:
 		ret = -ENOPROTOOPT;
 		break;
@@ -2379,6 +2383,7 @@ void sock_init_data(struct socket *sock, struct sock *sk)
 
 	sk->sk_max_pacing_rate = ~0U;
 	sk->sk_pacing_rate = ~0U;
+	sk->sk_incoming_cpu = -1;
 	/*
 	 * Before updating sk_refcnt, we must commit prior changes to memory
 	 * (Documentation/RCU/rculist_nulls.txt for details)
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index bed8886a4b6c..08643a3616af 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -185,6 +185,8 @@ static inline int compute_score(struct sock *sk, struct net *net,
 				return -1;
 			score += 4;
 		}
+		if (sk->sk_incoming_cpu == raw_smp_processor_id())
+			score++;
 	}
 	return score;
 }
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index e1fc129099ea..24ec14f9825c 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -375,7 +375,8 @@ static inline int compute_score(struct sock *sk, struct net *net,
 			return -1;
 		score += 4;
 	}
-
+	if (sk->sk_incoming_cpu == raw_smp_processor_id())
+		score++;
 	return score;
 }
 
@@ -419,6 +420,9 @@ static inline int compute_score2(struct sock *sk, struct net *net,
 		score += 4;
 	}
 
+	if (sk->sk_incoming_cpu == raw_smp_processor_id())
+		score++;
+
 	return score;
 }
 
diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index 6ac8dad0138a..21ace5a2bf7c 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -114,6 +114,8 @@ static inline int compute_score(struct sock *sk, struct net *net,
 				return -1;
 			score++;
 		}
+		if (sk->sk_incoming_cpu == raw_smp_processor_id())
+			score++;
 	}
 	return score;
 }
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 0aba654f5b91..01bcb49619ee 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -182,10 +182,12 @@ static inline int compute_score(struct sock *sk, struct net *net,
 		score++;
 	}
 
+	if (sk->sk_incoming_cpu == raw_smp_processor_id())
+		score++;
+
 	return score;
 }
 
-#define SCORE2_MAX (1 + 1 + 1)
 static inline int compute_score2(struct sock *sk, struct net *net,
 				 const struct in6_addr *saddr, __be16 sport,
 				 const struct in6_addr *daddr,
@@ -223,6 +225,9 @@ static inline int compute_score2(struct sock *sk, struct net *net,
 		score++;
 	}
 
+	if (sk->sk_incoming_cpu == raw_smp_processor_id())
+		score++;
+
 	return score;
 }
 
@@ -251,8 +256,7 @@ begin:
 				hash = udp6_ehashfn(net, daddr, hnum,
 						    saddr, sport);
 				matches = 1;
-			} else if (score == SCORE2_MAX)
-				goto exact_match;
+			}
 		} else if (score == badness && reuseport) {
 			matches++;
 			if (reciprocal_scale(hash, matches) == 0)
@@ -269,7 +273,6 @@ begin:
 		goto begin;
 
 	if (result) {
-exact_match:
 		if (unlikely(!atomic_inc_not_zero_hint(&result->sk_refcnt, 2)))
 			result = NULL;
 		else if (unlikely(compute_score2(result, net, saddr, sport,
-- 
cgit v1.2.3


From c2f34a65a61cd1ace3b53c93e8b38d2f79f4ff0d Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Wed, 14 Oct 2015 05:58:38 -0700
Subject: tcp/dccp: fix potential NULL deref in __inet_inherit_port()

As we no longer hold listener lock in fast path, it is possible that a
child is created right after listener freed its bound port, if a close()
is done while incoming packets are processed.

__inet_inherit_port() must detect this and return an error,
so that caller can free the child earlier.

Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets")
Fixes: 079096f103fa ("tcp/dccp: install syn_recv requests into ehash table")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/inet_hashtables.c | 4 ++++
 1 file changed, 4 insertions(+)

(limited to 'net/ipv4/inet_hashtables.c')

diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 08643a3616af..958728a22001 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -137,6 +137,10 @@ int __inet_inherit_port(const struct sock *sk, struct sock *child)
 
 	spin_lock(&head->lock);
 	tb = inet_csk(sk)->icsk_bind_hash;
+	if (unlikely(!tb)) {
+		spin_unlock(&head->lock);
+		return -ENOENT;
+	}
 	if (tb->port != port) {
 		/* NOTE: using tproxy and redirecting skbs to a proxy
 		 * on a different listener port breaks the assumption
-- 
cgit v1.2.3


From 5e0724d027f0548511a2165a209572d48fe7a4c8 Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Thu, 22 Oct 2015 08:20:46 -0700
Subject: tcp/dccp: fix hashdance race for passive sessions

Multiple cpus can process duplicates of incoming ACK messages
matching a SYN_RECV request socket. This is a rare event under
normal operations, but definitely can happen.

Only one must win the race, otherwise corruption would occur.

To fix this without adding new atomic ops, we use logic in
inet_ehash_nolisten() to detect the request was present in the same
ehash bucket where we try to insert the new child.

If request socket was not found, we have to undo the child creation.

This actually removes a spin_lock()/spin_unlock() pair in
reqsk_queue_unlink() for the fast path.

Fixes: e994b2f0fb92 ("tcp: do not lock listener to process SYN packets")
Fixes: 079096f103fa ("tcp/dccp: install syn_recv requests into ehash table")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/inet_connection_sock.h |  7 ++++++-
 include/net/inet_hashtables.h      |  4 ++--
 include/net/tcp.h                  |  4 +++-
 net/dccp/dccp.h                    |  4 +++-
 net/dccp/ipv4.c                    |  6 ++++--
 net/dccp/ipv6.c                    |  9 ++++++---
 net/dccp/minisocks.c               | 14 +++++++-------
 net/ipv4/inet_connection_sock.c    | 33 +++++++++++++++++++++++++-------
 net/ipv4/inet_hashtables.c         | 39 ++++++++++++++++++++++++--------------
 net/ipv4/syncookies.c              |  4 +++-
 net/ipv4/tcp_fastopen.c            |  4 +++-
 net/ipv4/tcp_ipv4.c                |  6 ++++--
 net/ipv4/tcp_minisocks.c           | 11 ++++-------
 net/ipv6/tcp_ipv6.c                |  9 ++++++---
 14 files changed, 102 insertions(+), 52 deletions(-)

(limited to 'net/ipv4/inet_hashtables.c')

diff --git a/include/net/inet_connection_sock.h b/include/net/inet_connection_sock.h
index 63615709839d..481fe1c9044c 100644
--- a/include/net/inet_connection_sock.h
+++ b/include/net/inet_connection_sock.h
@@ -43,7 +43,9 @@ struct inet_connection_sock_af_ops {
 	int	    (*conn_request)(struct sock *sk, struct sk_buff *skb);
 	struct sock *(*syn_recv_sock)(const struct sock *sk, struct sk_buff *skb,
 				      struct request_sock *req,
-				      struct dst_entry *dst);
+				      struct dst_entry *dst,
+				      struct request_sock *req_unhash,
+				      bool *own_req);
 	u16	    net_header_len;
 	u16	    net_frag_header_len;
 	u16	    sockaddr_len;
@@ -272,6 +274,9 @@ void inet_csk_reqsk_queue_add(struct sock *sk, struct request_sock *req,
 			      struct sock *child);
 void inet_csk_reqsk_queue_hash_add(struct sock *sk, struct request_sock *req,
 				   unsigned long timeout);
+struct sock *inet_csk_complete_hashdance(struct sock *sk, struct sock *child,
+					 struct request_sock *req,
+					 bool own_req);
 
 static inline void inet_csk_reqsk_queue_added(struct sock *sk)
 {
diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index 6683ada25fef..de2e3ade6102 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -205,8 +205,8 @@ void inet_put_port(struct sock *sk);
 
 void inet_hashinfo_init(struct inet_hashinfo *h);
 
-int inet_ehash_insert(struct sock *sk, struct sock *osk);
-void __inet_hash_nolisten(struct sock *sk, struct sock *osk);
+bool inet_ehash_insert(struct sock *sk, struct sock *osk);
+bool inet_ehash_nolisten(struct sock *sk, struct sock *osk);
 void __inet_hash(struct sock *sk, struct sock *osk);
 void inet_hash(struct sock *sk);
 void inet_unhash(struct sock *sk);
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 11e320412216..f80e74c5ad18 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -457,7 +457,9 @@ struct sock *tcp_create_openreq_child(const struct sock *sk,
 void tcp_ca_openreq_child(struct sock *sk, const struct dst_entry *dst);
 struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 				  struct request_sock *req,
-				  struct dst_entry *dst);
+				  struct dst_entry *dst,
+				  struct request_sock *req_unhash,
+				  bool *own_req);
 int tcp_v4_do_rcv(struct sock *sk, struct sk_buff *skb);
 int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
 int tcp_connect(struct sock *sk);
diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h
index 923f5a180134..b0e28d24e1a7 100644
--- a/net/dccp/dccp.h
+++ b/net/dccp/dccp.h
@@ -278,7 +278,9 @@ int dccp_v4_do_rcv(struct sock *sk, struct sk_buff *skb);
 
 struct sock *dccp_v4_request_recv_sock(const struct sock *sk, struct sk_buff *skb,
 				       struct request_sock *req,
-				       struct dst_entry *dst);
+				       struct dst_entry *dst,
+				       struct request_sock *req_unhash,
+				       bool *own_req);
 struct sock *dccp_check_req(struct sock *sk, struct sk_buff *skb,
 			    struct request_sock *req);
 
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index 59bc180b02d8..5684e14932bd 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -393,7 +393,9 @@ static inline u64 dccp_v4_init_sequence(const struct sk_buff *skb)
 struct sock *dccp_v4_request_recv_sock(const struct sock *sk,
 				       struct sk_buff *skb,
 				       struct request_sock *req,
-				       struct dst_entry *dst)
+				       struct dst_entry *dst,
+				       struct request_sock *req_unhash,
+				       bool *own_req)
 {
 	struct inet_request_sock *ireq;
 	struct inet_sock *newinet;
@@ -426,7 +428,7 @@ struct sock *dccp_v4_request_recv_sock(const struct sock *sk,
 
 	if (__inet_inherit_port(sk, newsk) < 0)
 		goto put_and_exit;
-	__inet_hash_nolisten(newsk, NULL);
+	*own_req = inet_ehash_nolisten(newsk, req_to_sk(req_unhash));
 
 	return newsk;
 
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index d9cc731f2619..ef4e48ce9143 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -380,7 +380,9 @@ drop:
 static struct sock *dccp_v6_request_recv_sock(const struct sock *sk,
 					      struct sk_buff *skb,
 					      struct request_sock *req,
-					      struct dst_entry *dst)
+					      struct dst_entry *dst,
+					      struct request_sock *req_unhash,
+					      bool *own_req)
 {
 	struct inet_request_sock *ireq = inet_rsk(req);
 	struct ipv6_pinfo *newnp;
@@ -393,7 +395,8 @@ static struct sock *dccp_v6_request_recv_sock(const struct sock *sk,
 		/*
 		 *	v6 mapped
 		 */
-		newsk = dccp_v4_request_recv_sock(sk, skb, req, dst);
+		newsk = dccp_v4_request_recv_sock(sk, skb, req, dst,
+						  req_unhash, own_req);
 		if (newsk == NULL)
 			return NULL;
 
@@ -511,7 +514,7 @@ static struct sock *dccp_v6_request_recv_sock(const struct sock *sk,
 		dccp_done(newsk);
 		goto out;
 	}
-	__inet_hash(newsk, NULL);
+	*own_req = inet_ehash_nolisten(newsk, req_to_sk(req_unhash));
 
 	return newsk;
 
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index d10aace43672..1994f8af646b 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -143,6 +143,7 @@ struct sock *dccp_check_req(struct sock *sk, struct sk_buff *skb,
 {
 	struct sock *child = NULL;
 	struct dccp_request_sock *dreq = dccp_rsk(req);
+	bool own_req;
 
 	/* Check for retransmitted REQUEST */
 	if (dccp_hdr(skb)->dccph_type == DCCP_PKT_REQUEST) {
@@ -182,14 +183,13 @@ struct sock *dccp_check_req(struct sock *sk, struct sk_buff *skb,
 	if (dccp_parse_options(sk, dreq, skb))
 		 goto drop;
 
-	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
-	if (child == NULL)
+	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL,
+							 req, &own_req);
+	if (!child)
 		goto listen_overflow;
 
-	inet_csk_reqsk_queue_drop(sk, req);
-	inet_csk_reqsk_queue_add(sk, req, child);
-out:
-	return child;
+	return inet_csk_complete_hashdance(sk, child, req, own_req);
+
 listen_overflow:
 	dccp_pr_debug("listen_overflow!\n");
 	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_TOO_BUSY;
@@ -198,7 +198,7 @@ drop:
 		req->rsk_ops->send_reset(sk, skb);
 
 	inet_csk_reqsk_queue_drop(sk, req);
-	goto out;
+	return NULL;
 }
 
 EXPORT_SYMBOL_GPL(dccp_check_req);
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 8430bc8ccd58..1feb15f23de8 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -523,15 +523,15 @@ static bool reqsk_queue_unlink(struct request_sock_queue *queue,
 			       struct request_sock *req)
 {
 	struct inet_hashinfo *hashinfo = req_to_sk(req)->sk_prot->h.hashinfo;
-	spinlock_t *lock;
-	bool found;
+	bool found = false;
 
-	lock = inet_ehash_lockp(hashinfo, req->rsk_hash);
-
-	spin_lock(lock);
-	found = __sk_nulls_del_node_init_rcu(req_to_sk(req));
-	spin_unlock(lock);
+	if (sk_hashed(req_to_sk(req))) {
+		spinlock_t *lock = inet_ehash_lockp(hashinfo, req->rsk_hash);
 
+		spin_lock(lock);
+		found = __sk_nulls_del_node_init_rcu(req_to_sk(req));
+		spin_unlock(lock);
+	}
 	if (timer_pending(&req->rsk_timer) && del_timer_sync(&req->rsk_timer))
 		reqsk_put(req);
 	return found;
@@ -811,6 +811,25 @@ void inet_csk_reqsk_queue_add(struct sock *sk, struct request_sock *req,
 }
 EXPORT_SYMBOL(inet_csk_reqsk_queue_add);
 
+struct sock *inet_csk_complete_hashdance(struct sock *sk, struct sock *child,
+					 struct request_sock *req, bool own_req)
+{
+	if (own_req) {
+		inet_csk_reqsk_queue_drop(sk, req);
+		reqsk_queue_removed(&inet_csk(sk)->icsk_accept_queue, req);
+		inet_csk_reqsk_queue_add(sk, req, child);
+		/* Warning: caller must not call reqsk_put(req);
+		 * child stole last reference on it.
+		 */
+		return child;
+	}
+	/* Too bad, another child took ownership of the request, undo. */
+	bh_unlock_sock(child);
+	sock_put(child);
+	return NULL;
+}
+EXPORT_SYMBOL(inet_csk_complete_hashdance);
+
 /*
  *	This routine closes sockets which have been at least partially
  *	opened, but not yet accepted.
diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
index 958728a22001..ccc5980797fc 100644
--- a/net/ipv4/inet_hashtables.c
+++ b/net/ipv4/inet_hashtables.c
@@ -407,13 +407,13 @@ static u32 inet_sk_port_offset(const struct sock *sk)
 /* insert a socket into ehash, and eventually remove another one
  * (The another one can be a SYN_RECV or TIMEWAIT
  */
-int inet_ehash_insert(struct sock *sk, struct sock *osk)
+bool inet_ehash_insert(struct sock *sk, struct sock *osk)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
 	struct hlist_nulls_head *list;
 	struct inet_ehash_bucket *head;
 	spinlock_t *lock;
-	int ret = 0;
+	bool ret = true;
 
 	WARN_ON_ONCE(!sk_unhashed(sk));
 
@@ -423,30 +423,41 @@ int inet_ehash_insert(struct sock *sk, struct sock *osk)
 	lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
 
 	spin_lock(lock);
-	__sk_nulls_add_node_rcu(sk, list);
 	if (osk) {
-		WARN_ON(sk->sk_hash != osk->sk_hash);
-		sk_nulls_del_node_init_rcu(osk);
+		WARN_ON_ONCE(sk->sk_hash != osk->sk_hash);
+		ret = sk_nulls_del_node_init_rcu(osk);
 	}
+	if (ret)
+		__sk_nulls_add_node_rcu(sk, list);
 	spin_unlock(lock);
 	return ret;
 }
 
-void __inet_hash_nolisten(struct sock *sk, struct sock *osk)
+bool inet_ehash_nolisten(struct sock *sk, struct sock *osk)
 {
-	inet_ehash_insert(sk, osk);
-	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
+	bool ok = inet_ehash_insert(sk, osk);
+
+	if (ok) {
+		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
+	} else {
+		percpu_counter_inc(sk->sk_prot->orphan_count);
+		sk->sk_state = TCP_CLOSE;
+		sock_set_flag(sk, SOCK_DEAD);
+		inet_csk_destroy_sock(sk);
+	}
+	return ok;
 }
-EXPORT_SYMBOL_GPL(__inet_hash_nolisten);
+EXPORT_SYMBOL_GPL(inet_ehash_nolisten);
 
 void __inet_hash(struct sock *sk, struct sock *osk)
 {
 	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
 	struct inet_listen_hashbucket *ilb;
 
-	if (sk->sk_state != TCP_LISTEN)
-		return __inet_hash_nolisten(sk, osk);
-
+	if (sk->sk_state != TCP_LISTEN) {
+		inet_ehash_nolisten(sk, osk);
+		return;
+	}
 	WARN_ON(!sk_unhashed(sk));
 	ilb = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
 
@@ -567,7 +578,7 @@ ok:
 		inet_bind_hash(sk, tb, port);
 		if (sk_unhashed(sk)) {
 			inet_sk(sk)->inet_sport = htons(port);
-			__inet_hash_nolisten(sk, (struct sock *)tw);
+			inet_ehash_nolisten(sk, (struct sock *)tw);
 		}
 		if (tw)
 			inet_twsk_bind_unhash(tw, hinfo);
@@ -584,7 +595,7 @@ ok:
 	tb  = inet_csk(sk)->icsk_bind_hash;
 	spin_lock_bh(&head->lock);
 	if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
-		__inet_hash_nolisten(sk, NULL);
+		inet_ehash_nolisten(sk, NULL);
 		spin_unlock_bh(&head->lock);
 		return 0;
 	} else {
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 4c0892badb8b..4cbe9f0a4281 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -221,8 +221,10 @@ struct sock *tcp_get_cookie_sock(struct sock *sk, struct sk_buff *skb,
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct sock *child;
+	bool own_req;
 
-	child = icsk->icsk_af_ops->syn_recv_sock(sk, skb, req, dst);
+	child = icsk->icsk_af_ops->syn_recv_sock(sk, skb, req, dst,
+						 NULL, &own_req);
 	if (child) {
 		atomic_set(&req->rsk_refcnt, 1);
 		sock_rps_save_rxhash(child, skb);
diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c
index 93396bf7b475..55be6ac70cff 100644
--- a/net/ipv4/tcp_fastopen.c
+++ b/net/ipv4/tcp_fastopen.c
@@ -133,12 +133,14 @@ static struct sock *tcp_fastopen_create_child(struct sock *sk,
 	struct request_sock_queue *queue = &inet_csk(sk)->icsk_accept_queue;
 	struct sock *child;
 	u32 end_seq;
+	bool own_req;
 
 	req->num_retrans = 0;
 	req->num_timeout = 0;
 	req->sk = NULL;
 
-	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
+	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL,
+							 NULL, &own_req);
 	if (!child)
 		return NULL;
 
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 30dd45c1f568..1c2648bbac4b 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1247,7 +1247,9 @@ EXPORT_SYMBOL(tcp_v4_conn_request);
  */
 struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 				  struct request_sock *req,
-				  struct dst_entry *dst)
+				  struct dst_entry *dst,
+				  struct request_sock *req_unhash,
+				  bool *own_req)
 {
 	struct inet_request_sock *ireq;
 	struct inet_sock *newinet;
@@ -1323,7 +1325,7 @@ struct sock *tcp_v4_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 
 	if (__inet_inherit_port(sk, newsk) < 0)
 		goto put_and_exit;
-	__inet_hash_nolisten(newsk, NULL);
+	*own_req = inet_ehash_nolisten(newsk, req_to_sk(req_unhash));
 
 	return newsk;
 
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 1fd5d413a664..3575dd1e5b67 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -580,6 +580,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 	const struct tcphdr *th = tcp_hdr(skb);
 	__be32 flg = tcp_flag_word(th) & (TCP_FLAG_RST|TCP_FLAG_SYN|TCP_FLAG_ACK);
 	bool paws_reject = false;
+	bool own_req;
 
 	tmp_opt.saw_tstamp = 0;
 	if (th->doff > (sizeof(struct tcphdr)>>2)) {
@@ -767,18 +768,14 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
 	 * ESTABLISHED STATE. If it will be dropped after
 	 * socket is created, wait for troubles.
 	 */
-	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
+	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL,
+							 req, &own_req);
 	if (!child)
 		goto listen_overflow;
 
 	sock_rps_save_rxhash(child, skb);
 	tcp_synack_rtt_meas(child, req);
-	inet_csk_reqsk_queue_drop(sk, req);
-	inet_csk_reqsk_queue_add(sk, req, child);
-	/* Warning: caller must not call reqsk_put(req);
-	 * child stole last reference on it.
-	 */
-	return child;
+	return inet_csk_complete_hashdance(sk, child, req, own_req);
 
 listen_overflow:
 	if (!sysctl_tcp_abort_on_overflow) {
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index f495d189f5e0..714bc5ad096e 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -965,7 +965,9 @@ drop:
 
 static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *skb,
 					 struct request_sock *req,
-					 struct dst_entry *dst)
+					 struct dst_entry *dst,
+					 struct request_sock *req_unhash,
+					 bool *own_req)
 {
 	struct inet_request_sock *ireq;
 	struct ipv6_pinfo *newnp;
@@ -984,7 +986,8 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *
 		 *	v6 mapped
 		 */
 
-		newsk = tcp_v4_syn_recv_sock(sk, skb, req, dst);
+		newsk = tcp_v4_syn_recv_sock(sk, skb, req, dst,
+					     req_unhash, own_req);
 
 		if (!newsk)
 			return NULL;
@@ -1145,7 +1148,7 @@ static struct sock *tcp_v6_syn_recv_sock(const struct sock *sk, struct sk_buff *
 		tcp_done(newsk);
 		goto out;
 	}
-	__inet_hash(newsk, NULL);
+	*own_req = inet_ehash_nolisten(newsk, req_to_sk(req_unhash));
 
 	return newsk;
 
-- 
cgit v1.2.3