netdev - [PATCH net] net: Do not clear the socket TX queue in sock

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-next>] [day] [month] [year] [list]

Message-Id: <1592748544-41555-1-git-send-email-tariqt@mellanox.com>
Date:   Sun, 21 Jun 2020 17:09:04 +0300
From:   Tariq Toukan <tariqt@...lanox.com>
To:     "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>
Cc:     netdev@...r.kernel.org, Saeed Mahameed <saeedm@...lanox.com>,
        Boris Pismenny <borisp@...lanox.com>,
        Tariq Toukan <tariqt@...lanox.com>
Subject: [PATCH net] net: Do not clear the socket TX queue in sock_orphan()

sock_orphan() call to sk_set_socket() implies clearing the sock TX queue.
This might cause unexpected out-of-order transmit, as outstanding packets
can pick a different TX queue and bypass the ones already queued.
This is undesired in general. More specifically, it breaks the in-order
scheduling property guarantee for device-offloaded TLS sockets.

Introduce a function variation __sk_set_socket() that does not clear
the TX queue, and call it from sock_orphan().
All other callers of sk_set_socket() do not operate on an active socket,
so they do not need this change.

Fixes: e022f0b4a03f ("net: Introduce sk_tx_queue_mapping")
Signed-off-by: Tariq Toukan <tariqt@...lanox.com>
Reviewed-by: Boris Pismenny <borisp@...lanox.com>
---
 include/net/sock.h | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

Please queue for -stable.

diff --git a/include/net/sock.h b/include/net/sock.h
index c53cc42b5ab9..23e43f3d79f0 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -1846,10 +1846,15 @@ static inline int sk_rx_queue_get(const struct sock *sk)
 }
 #endif
 
+static inline void __sk_set_socket(struct sock *sk, struct socket *sock)
+{
+	sk->sk_socket = sock;
+}
+
 static inline void sk_set_socket(struct sock *sk, struct socket *sock)
 {
 	sk_tx_queue_clear(sk);
-	sk->sk_socket = sock;
+	__sk_set_socket(sk, sock);
 }
 
 static inline wait_queue_head_t *sk_sleep(struct sock *sk)
@@ -1868,7 +1873,7 @@ static inline void sock_orphan(struct sock *sk)
 {
 	write_lock_bh(&sk->sk_callback_lock);
 	sock_set_flag(sk, SOCK_DEAD);
-	sk_set_socket(sk, NULL);
+	__sk_set_socket(sk, NULL);
 	sk->sk_wq  = NULL;
 	write_unlock_bh(&sk->sk_callback_lock);
 }
-- 
1.8.3.1