netdev - [PATCH net 1/2] Revert "net/sctp: fix race condition in sctp_destroy

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <4dc7d7dd4bcf7122604ccb52a5c747c3fb9101c5.1619989856.git.lucien.xin@gmail.com>
Date:   Mon,  3 May 2021 05:11:41 +0800
From:   Xin Long <lucien.xin@...il.com>
To:     network dev <netdev@...r.kernel.org>, linux-sctp@...r.kernel.org
Cc:     davem@...emloft.net, kuba@...nel.org,
        Marcelo Ricardo Leitner <marcelo.leitner@...il.com>,
        Or Cohen <orcohen@...oaltonetworks.com>
Subject: [PATCH net 1/2] Revert "net/sctp: fix race condition in sctp_destroy_sock"

This reverts commit b166a20b07382b8bc1dcee2a448715c9c2c81b5b.

This one has to be reverted as it introduced a dead lock, as
syzbot reported:

       CPU0                    CPU1
       ----                    ----
  lock(&net->sctp.addr_wq_lock);
                               lock(slock-AF_INET6);
                               lock(&net->sctp.addr_wq_lock);
  lock(slock-AF_INET6);

CPU0 is the thread of sctp_addr_wq_timeout_handler(), and CPU1
is that of sctp_close().

The original issue this commit fixed will be fixed in the next
patch.

Reported-by: syzbot+959223586843e69a2674@...kaller.appspotmail.com
Signed-off-by: Xin Long <lucien.xin@...il.com>
---
 net/sctp/socket.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/net/sctp/socket.c b/net/sctp/socket.c
index b7b9013..76a388b5 100644
--- a/net/sctp/socket.c
+++ b/net/sctp/socket.c
@@ -1520,9 +1520,11 @@ static void sctp_close(struct sock *sk, long timeout)
 
 	/* Supposedly, no process has access to the socket, but
 	 * the net layers still may.
+	 * Also, sctp_destroy_sock() needs to be called with addr_wq_lock
+	 * held and that should be grabbed before socket lock.
 	 */
-	local_bh_disable();
-	bh_lock_sock(sk);
+	spin_lock_bh(&net->sctp.addr_wq_lock);
+	bh_lock_sock_nested(sk);
 
 	/* Hold the sock, since sk_common_release() will put sock_put()
 	 * and we have just a little more cleanup.
@@ -1531,7 +1533,7 @@ static void sctp_close(struct sock *sk, long timeout)
 	sk_common_release(sk);
 
 	bh_unlock_sock(sk);
-	local_bh_enable();
+	spin_unlock_bh(&net->sctp.addr_wq_lock);
 
 	sock_put(sk);
 
@@ -4991,6 +4993,9 @@ static int sctp_init_sock(struct sock *sk)
 	sk_sockets_allocated_inc(sk);
 	sock_prot_inuse_add(net, sk->sk_prot, 1);
 
+	/* Nothing can fail after this block, otherwise
+	 * sctp_destroy_sock() will be called without addr_wq_lock held
+	 */
 	if (net->sctp.default_auto_asconf) {
 		spin_lock(&sock_net(sk)->sctp.addr_wq_lock);
 		list_add_tail(&sp->auto_asconf_list,
@@ -5025,9 +5030,7 @@ static void sctp_destroy_sock(struct sock *sk)
 
 	if (sp->do_auto_asconf) {
 		sp->do_auto_asconf = 0;
-		spin_lock_bh(&sock_net(sk)->sctp.addr_wq_lock);
 		list_del(&sp->auto_asconf_list);
-		spin_unlock_bh(&sock_net(sk)->sctp.addr_wq_lock);
 	}
 	sctp_endpoint_free(sp->ep);
 	local_bh_disable();
-- 
2.1.0