netdev - [PATCH bpf-next 5/8] tcp: seq_file: Replace listening

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Fri, 25 Jun 2021 13:05:17 -0700
From:   Martin KaFai Lau <kafai@...com>
To:     <bpf@...r.kernel.org>
CC:     Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Eric Dumazet <edumazet@...gle.com>, <kernel-team@...com>,
        Neal Cardwell <ncardwell@...gle.com>, <netdev@...r.kernel.org>,
        Yonghong Song <yhs@...com>, Yuchung Cheng <ycheng@...gle.com>
Subject: [PATCH bpf-next 5/8] tcp: seq_file: Replace listening_hash with lhash2

This patch moves the tcp seq_file iteration on listeners
from the port only listening_hash to the port+addr lhash2.

When iterating from the bpf iter, the next patch will need to
lock the socket such that the bpf iter with netadmin cap can do
setsockopt (e.g. to change the TCP_CONGESTION).  To avoid locking the
bucket and then locking the sock, the bpf iter will first batch some
sockets from the same bucket and then unlock the bucket.  If the bucket
size is small (which usually is), it is easier to batch the whole
bucket such that it is less likely to miss a setsockopt on a socket
due to changes in the bucket.

However, the port only listening_hash could have many listeners
hashed to a bucket (e.g. many individual VIP(s):443 and also
multiple by the number of SO_REUSEPORT).  We have seen bucket size in
tens of thousands range.  Also, the chance of having changes
in some popular port buckets (e.g. 443) is also high.

The port+addr lhash2 was introduced to solve this large listener bucket
issue.  Also, the listening_hash usage has already been replaced with
lhash2 in the fast path inet[6]_lookup_listener().  This patch follows
the same direction on moving to lhash2 and iterates the lhash2
instead of listening_hash.

Signed-off-by: Martin KaFai Lau <kafai@...com>
---
 include/net/inet_hashtables.h |  6 ++++++
 net/ipv4/tcp_ipv4.c           | 35 ++++++++++++++++++-----------------
 2 files changed, 24 insertions(+), 17 deletions(-)

diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
index ca6a3ea9057e..f72ec113ae56 100644
--- a/include/net/inet_hashtables.h
+++ b/include/net/inet_hashtables.h
@@ -160,6 +160,12 @@ struct inet_hashinfo {
 					____cacheline_aligned_in_smp;
 };
 
+#define inet_lhash2_for_each_icsk_continue(__icsk) \
+	hlist_for_each_entry_continue(__icsk, icsk_listen_portaddr_node)
+
+#define inet_lhash2_for_each_icsk(__icsk, list) \
+	hlist_for_each_entry(__icsk, list, icsk_listen_portaddr_node)
+
 #define inet_lhash2_for_each_icsk_rcu(__icsk, list) \
 	hlist_for_each_entry_rcu(__icsk, list, icsk_listen_portaddr_node)
 
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 693bda155876..0d851289a89e 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2296,21 +2296,22 @@ static void *listening_get_first(struct seq_file *seq)
 	struct tcp_iter_state *st = seq->private;
 
 	st->offset = 0;
-	for (; st->bucket < INET_LHTABLE_SIZE; st->bucket++) {
-		struct inet_listen_hashbucket *ilb;
-		struct hlist_nulls_node *node;
+	for (; st->bucket <= tcp_hashinfo.lhash2_mask; st->bucket++) {
+		struct inet_listen_hashbucket *ilb2;
+		struct inet_connection_sock *icsk;
 		struct sock *sk;
 
-		ilb = &tcp_hashinfo.listening_hash[st->bucket];
-		if (hlist_nulls_empty(&ilb->nulls_head))
+		ilb2 = &tcp_hashinfo.lhash2[st->bucket];
+		if (hlist_empty(&ilb2->head))
 			continue;
 
-		spin_lock(&ilb->lock);
-		sk_nulls_for_each(sk, node, &ilb->nulls_head) {
+		spin_lock(&ilb2->lock);
+		inet_lhash2_for_each_icsk(icsk, &ilb2->head) {
+			sk = (struct sock *)icsk;
 			if (seq_sk_match(seq, sk))
 				return sk;
 		}
-		spin_unlock(&ilb->lock);
+		spin_unlock(&ilb2->lock);
 	}
 
 	return NULL;
@@ -2324,22 +2325,22 @@ static void *listening_get_first(struct seq_file *seq)
 static void *listening_get_next(struct seq_file *seq, void *cur)
 {
 	struct tcp_iter_state *st = seq->private;
-	struct inet_listen_hashbucket *ilb;
-	struct hlist_nulls_node *node;
+	struct inet_listen_hashbucket *ilb2;
+	struct inet_connection_sock *icsk;
 	struct sock *sk = cur;
 
 	++st->num;
 	++st->offset;
 
-	sk = sk_nulls_next(sk);
-
-	sk_nulls_for_each_from(sk, node) {
+	icsk = inet_csk(sk);
+	inet_lhash2_for_each_icsk_continue(icsk) {
+		sk = (struct sock *)icsk;
 		if (seq_sk_match(seq, sk))
 			return sk;
 	}
 
-	ilb = &tcp_hashinfo.listening_hash[st->bucket];
-	spin_unlock(&ilb->lock);
+	ilb2 = &tcp_hashinfo.lhash2[st->bucket];
+	spin_unlock(&ilb2->lock);
 	++st->bucket;
 	return listening_get_first(seq);
 }
@@ -2456,7 +2457,7 @@ static void *tcp_seek_last_pos(struct seq_file *seq)
 
 	switch (st->state) {
 	case TCP_SEQ_STATE_LISTENING:
-		if (st->bucket >= INET_LHTABLE_SIZE)
+		if (st->bucket > tcp_hashinfo.lhash2_mask)
 			break;
 		st->state = TCP_SEQ_STATE_LISTENING;
 		rc = listening_get_first(seq);
@@ -2541,7 +2542,7 @@ void tcp_seq_stop(struct seq_file *seq, void *v)
 	switch (st->state) {
 	case TCP_SEQ_STATE_LISTENING:
 		if (v != SEQ_START_TOKEN)
-			spin_unlock(&tcp_hashinfo.listening_hash[st->bucket].lock);
+			spin_unlock(&tcp_hashinfo.lhash2[st->bucket].lock);
 		break;
 	case TCP_SEQ_STATE_ESTABLISHED:
 		if (v)
-- 
2.30.2