[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <87a6nzrqe0.fsf@factor-ts.ru>
Date: Wed, 09 Jun 2021 18:47:51 +0300
From: Peter Kosyh <p.kosyh@...il.com>
To: netdev@...r.kernel.org
Cc: Peter Kosyh <p.kosyh@...il.com>,
'Mikhail Smirnov' <smirnov@...tor-ts.ru>,
"David S. Miller" <davem@...emloft.net>
Subject: [PATCH] udp: compute_score and sk_bound_dev_if regression
udp: commit 6da5b0f027a825df2aebc1927a27bda185dc03d4
introduced regression in compute_score() Previously for addr_any sockets an
interface bound socket had a higher priority than an unbound socket that
seems right. For example, this feature is used in dhcprelay daemon and now it
is broken.
So, this patch returns the old behavior and gives higher score for sk_bound_dev_if sockets.
Signed-off-by: Peter Kosyh <p.kosyh@...il.com>
---
net/ipv4/udp.c | 3 ++-
net/ipv6/udp.c | 3 ++-
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 15f5504adf5b..4239ffa93c6f 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -390,7 +390,8 @@ static int compute_score(struct sock *sk, struct net *net,
dif, sdif);
if (!dev_match)
return -1;
- score += 4;
+ if (sk->sk_bound_dev_if)
+ score += 4;
if (READ_ONCE(sk->sk_incoming_cpu) == raw_smp_processor_id())
score++;
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 199b080d418a..c2f88b5def25 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -133,7 +133,8 @@ static int compute_score(struct sock *sk, struct net *net,
dev_match = udp_sk_bound_dev_eq(net, sk->sk_bound_dev_if, dif, sdif);
if (!dev_match)
return -1;
- score++;
+ if (sk->sk_bound_dev_if)
+ score++;
if (READ_ONCE(sk->sk_incoming_cpu) == raw_smp_processor_id())
score++;
--
2.31.1
Powered by blists - more mailing lists