Merge branch 'tcp-scale-connect-under-pressure'

Eric Dumazet says:

====================
tcp: scale connect() under pressure

Adoption of bhash2 in linux-6.1 made some operations almost twice
more expensive, because of additional locks.

This series adds RCU in __inet_hash_connect() to help the
case where many attempts need to be made before finding
an available 4-tuple.

This brings a ~200 % improvement in this experiment:

Server:
ulimit -n 40000; neper/tcp_crr -T 200 -F 30000 -6 --nolog

Client:
ulimit -n 40000; neper/tcp_crr -T 200 -F 30000 -6 --nolog -c -H server

Before series:

  utime_start=0.288582
  utime_end=1.548707
  stime_start=20.637138
  stime_end=2002.489845
  num_transactions=484453
  latency_min=0.156279245
  latency_max=20.922042756
  latency_mean=1.546521274
  latency_stddev=3.936005194
  num_samples=312537
  throughput=47426.00

perf top on the client:

 49.54%  [kernel]       [k] _raw_spin_lock
 25.87%  [kernel]       [k] _raw_spin_lock_bh
  5.97%  [kernel]       [k] queued_spin_lock_slowpath
  5.67%  [kernel]       [k] __inet_hash_connect
  3.53%  [kernel]       [k] __inet6_check_established
  3.48%  [kernel]       [k] inet6_ehashfn
  0.64%  [kernel]       [k] rcu_all_qs

After this series:

  utime_start=0.271607
  utime_end=3.847111
  stime_start=18.407684
  stime_end=1997.485557
  num_transactions=1350742
  latency_min=0.014131929
  latency_max=17.895073144
  latency_mean=0.505675853   # Nice reduction of latency metrics
  latency_stddev=2.125164772
  num_samples=307884
  throughput=139866.80       # 194 % increase

perf top on client:

 56.86%  [kernel]       [k] __inet6_check_established
 17.96%  [kernel]       [k] __inet_hash_connect
 13.88%  [kernel]       [k] inet6_ehashfn
  2.52%  [kernel]       [k] rcu_all_qs
  2.01%  [kernel]       [k] __cond_resched
  0.41%  [kernel]       [k] _raw_spin_lock
====================

Link: https://patch.msgid.link/20250302124237.3913746-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

include/net/inet_hashtables.h

@@ -89,6 +89,7 @@ struct inet_bind_bucket {
 	bool			fast_ipv6_only;
 	struct hlist_node	node;
 	struct hlist_head	bhash2;
+	struct rcu_head		rcu;
 };
 
 struct inet_bind2_bucket {
@@ -226,8 +227,7 @@ struct inet_bind_bucket *
 inet_bind_bucket_create(struct kmem_cache *cachep, struct net *net,
 			struct inet_bind_hashbucket *head,
 			const unsigned short snum, int l3mdev);
-void inet_bind_bucket_destroy(struct kmem_cache *cachep,
-			      struct inet_bind_bucket *tb);
+void inet_bind_bucket_destroy(struct inet_bind_bucket *tb);
 
 bool inet_bind_bucket_match(const struct inet_bind_bucket *tb,
 			    const struct net *net, unsigned short port,
@@ -529,7 +529,8 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 			struct sock *sk, u64 port_offset,
 			int (*check_established)(struct inet_timewait_death_row *,
 						 struct sock *, __u16,
-						 struct inet_timewait_sock **));
+						 struct inet_timewait_sock **,
+						 bool rcu_lookup));
 
 int inet_hash_connect(struct inet_timewait_death_row *death_row,
 		      struct sock *sk);

net/ipv4/inet_connection_sock.c

@@ -157,12 +157,10 @@ static bool inet_use_bhash2_on_bind(const struct sock *sk)
 {
 #if IS_ENABLED(CONFIG_IPV6)
 	if (sk->sk_family == AF_INET6) {
-		int addr_type = ipv6_addr_type(&sk->sk_v6_rcv_saddr);
-
-		if (addr_type == IPV6_ADDR_ANY)
+		if (ipv6_addr_any(&sk->sk_v6_rcv_saddr))
 			return false;
 
-		if (addr_type != IPV6_ADDR_MAPPED)
+		if (!ipv6_addr_v4mapped(&sk->sk_v6_rcv_saddr))
 			return true;
 	}
 #endif
@@ -600,7 +598,7 @@ fail_unlock:
 		if (bhash2_created)
 			inet_bind2_bucket_destroy(hinfo->bind2_bucket_cachep, tb2);
 		if (bhash_created)
-			inet_bind_bucket_destroy(hinfo->bind_bucket_cachep, tb);
+			inet_bind_bucket_destroy(tb);
 	}
 	if (head2_lock_acquired)
 		spin_unlock(&head2->lock);

net/ipv4/inet_hashtables.c

@@ -76,7 +76,7 @@ struct inet_bind_bucket *inet_bind_bucket_create(struct kmem_cache *cachep,
 		tb->fastreuse = 0;
 		tb->fastreuseport = 0;
 		INIT_HLIST_HEAD(&tb->bhash2);
-		hlist_add_head(&tb->node, &head->chain);
+		hlist_add_head_rcu(&tb->node, &head->chain);
 	}
 	return tb;
 }
@@ -84,11 +84,11 @@ struct inet_bind_bucket *inet_bind_bucket_create(struct kmem_cache *cachep,
 /*
  * Caller must hold hashbucket lock for this tb with local BH disabled
  */
-void inet_bind_bucket_destroy(struct kmem_cache *cachep, struct inet_bind_bucket *tb)
+void inet_bind_bucket_destroy(struct inet_bind_bucket *tb)
 {
 	if (hlist_empty(&tb->bhash2)) {
-		__hlist_del(&tb->node);
-		kmem_cache_free(cachep, tb);
+		hlist_del_rcu(&tb->node);
+		kfree_rcu(tb, rcu);
 	}
 }
 
@@ -201,7 +201,7 @@ static void __inet_put_port(struct sock *sk)
 	}
 	spin_unlock(&head2->lock);
 
-	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
+	inet_bind_bucket_destroy(tb);
 	spin_unlock(&head->lock);
 }
 
@@ -285,7 +285,7 @@ bhash2_find:
 
 error:
 	if (created_inet_bind_bucket)
-		inet_bind_bucket_destroy(table->bind_bucket_cachep, tb);
+		inet_bind_bucket_destroy(tb);
 	spin_unlock(&head2->lock);
 	spin_unlock(&head->lock);
 	return -ENOMEM;
@@ -537,7 +537,8 @@ EXPORT_SYMBOL_GPL(__inet_lookup_established);
 /* called with local bh disabled */
 static int __inet_check_established(struct inet_timewait_death_row *death_row,
 				    struct sock *sk, __u16 lport,
-				    struct inet_timewait_sock **twp)
+				    struct inet_timewait_sock **twp,
+				    bool rcu_lookup)
 {
 	struct inet_hashinfo *hinfo = death_row->hashinfo;
 	struct inet_sock *inet = inet_sk(sk);
@@ -551,11 +552,24 @@ static int __inet_check_established(struct inet_timewait_death_row *death_row,
 	unsigned int hash = inet_ehashfn(net, daddr, lport,
 					 saddr, inet->inet_dport);
 	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
-	spinlock_t *lock = inet_ehash_lockp(hinfo, hash);
-	struct sock *sk2;
-	const struct hlist_nulls_node *node;
 	struct inet_timewait_sock *tw = NULL;
+	const struct hlist_nulls_node *node;
+	struct sock *sk2;
+	spinlock_t *lock;
 
+	if (rcu_lookup) {
+		sk_nulls_for_each(sk2, node, &head->chain) {
+			if (sk2->sk_hash != hash ||
+			    !inet_match(net, sk2, acookie, ports, dif, sdif))
+				continue;
+			if (sk2->sk_state == TCP_TIME_WAIT)
+				break;
+			return -EADDRNOTAVAIL;
+		}
+		return 0;
+	}
+
+	lock = inet_ehash_lockp(hinfo, hash);
 	spin_lock(lock);
 
 	sk_nulls_for_each(sk2, node, &head->chain) {
@@ -994,7 +1008,8 @@ static u32 *table_perturb;
 int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 		struct sock *sk, u64 port_offset,
 		int (*check_established)(struct inet_timewait_death_row *,
-			struct sock *, __u16, struct inet_timewait_sock **))
+			struct sock *, __u16, struct inet_timewait_sock **,
+			bool rcu_lookup))
 {
 	struct inet_hashinfo *hinfo = death_row->hashinfo;
 	struct inet_bind_hashbucket *head, *head2;
@@ -1012,7 +1027,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
 
 	if (port) {
 		local_bh_disable();
-		ret = check_established(death_row, sk, port, NULL);
+		ret = check_established(death_row, sk, port, NULL, false);
 		local_bh_enable();
 		return ret;
 	}
@@ -1048,6 +1063,21 @@ other_parity_scan:
 			continue;
 		head = &hinfo->bhash[inet_bhashfn(net, port,
 						  hinfo->bhash_size)];
+		rcu_read_lock();
+		hlist_for_each_entry_rcu(tb, &head->chain, node) {
+			if (!inet_bind_bucket_match(tb, net, port, l3mdev))
+				continue;
+			if (tb->fastreuse >= 0 || tb->fastreuseport >= 0) {
+				rcu_read_unlock();
+				goto next_port;
+			}
+			if (!check_established(death_row, sk, port, &tw, true))
+				break;
+			rcu_read_unlock();
+			goto next_port;
+		}
+		rcu_read_unlock();
+
 		spin_lock_bh(&head->lock);
 
 		/* Does not bother with rcv_saddr checks, because
@@ -1057,12 +1087,12 @@ other_parity_scan:
 			if (inet_bind_bucket_match(tb, net, port, l3mdev)) {
 				if (tb->fastreuse >= 0 ||
 				    tb->fastreuseport >= 0)
-					goto next_port;
+					goto next_port_unlock;
 				WARN_ON(hlist_empty(&tb->bhash2));
 				if (!check_established(death_row, sk,
-						       port, &tw))
+						       port, &tw, false))
 					goto ok;
-				goto next_port;
+				goto next_port_unlock;
 			}
 		}
 
@@ -1076,8 +1106,9 @@ other_parity_scan:
 		tb->fastreuse = -1;
 		tb->fastreuseport = -1;
 		goto ok;
-next_port:
+next_port_unlock:
 		spin_unlock_bh(&head->lock);
+next_port:
 		cond_resched();
 	}
 
@@ -1149,7 +1180,7 @@ error:
 
 	spin_unlock(&head2->lock);
 	if (tb_created)
-		inet_bind_bucket_destroy(hinfo->bind_bucket_cachep, tb);
+		inet_bind_bucket_destroy(tb);
 	spin_unlock(&head->lock);
 
 	if (tw)

net/ipv4/inet_timewait_sock.c

@@ -39,7 +39,7 @@ void inet_twsk_bind_unhash(struct inet_timewait_sock *tw,
 	tw->tw_tb = NULL;
 	tw->tw_tb2 = NULL;
 	inet_bind2_bucket_destroy(hashinfo->bind2_bucket_cachep, tb2);
-	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
+	inet_bind_bucket_destroy(tb);
 
 	__sock_put((struct sock *)tw);
 }

net/ipv6/inet6_hashtables.c

@@ -263,7 +263,8 @@ EXPORT_SYMBOL_GPL(inet6_lookup);
 
 static int __inet6_check_established(struct inet_timewait_death_row *death_row,
 				     struct sock *sk, const __u16 lport,
-				     struct inet_timewait_sock **twp)
+				     struct inet_timewait_sock **twp,
+				     bool rcu_lookup)
 {
 	struct inet_hashinfo *hinfo = death_row->hashinfo;
 	struct inet_sock *inet = inet_sk(sk);
@@ -276,11 +277,25 @@ static int __inet6_check_established(struct inet_timewait_death_row *death_row,
 	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
 						inet->inet_dport);
 	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
-	spinlock_t *lock = inet_ehash_lockp(hinfo, hash);
-	struct sock *sk2;
-	const struct hlist_nulls_node *node;
 	struct inet_timewait_sock *tw = NULL;
+	const struct hlist_nulls_node *node;
+	struct sock *sk2;
+	spinlock_t *lock;
 
+	if (rcu_lookup) {
+		sk_nulls_for_each(sk2, node, &head->chain) {
+			if (sk2->sk_hash != hash ||
+			    !inet6_match(net, sk2, saddr, daddr,
+					 ports, dif, sdif))
+				continue;
+			if (sk2->sk_state == TCP_TIME_WAIT)
+				break;
+			return -EADDRNOTAVAIL;
+		}
+		return 0;
+	}
+
+	lock = inet_ehash_lockp(hinfo, hash);
 	spin_lock(lock);
 
 	sk_nulls_for_each(sk2, node, &head->chain) {