linux-kernel - [PATCH v1] can: j1939: prevent deadlock by moving j1939_sk

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>] [day] [month] [year] [list]

Message-Id: <20230324130141.2132787-1-o.rempel@pengutronix.de>
Date:   Fri, 24 Mar 2023 14:01:41 +0100
From:   Oleksij Rempel <o.rempel@...gutronix.de>
To:     Robin van der Gracht <robin@...tonic.nl>,
        Oliver Hartkopp <socketcan@...tkopp.net>,
        Marc Kleine-Budde <mkl@...gutronix.de>
Cc:     Oleksij Rempel <o.rempel@...gutronix.de>,
        syzbot+ee1cd780f69483a8616b@...kaller.appspotmail.com,
        Hillf Danton <hdanton@...a.com>, kernel@...gutronix.de,
        linux-can@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: [PATCH v1] can: j1939: prevent deadlock by moving j1939_sk_errqueue()

This commit addresses a deadlock situation that can occur in certain
scenarios, such as when running data TP/ETP transfer and subscribing to
the error queue while receiving a net down event. The deadlock involves
locks in the following order:

3
  j1939_session_list_lock ->  active_session_list_lock
  j1939_session_activate
  ...
  j1939_sk_queue_activate_next -> sk_session_queue_lock
  ...
  j1939_xtp_rx_eoma_one

2
  j1939_sk_queue_drop_all  ->  sk_session_queue_lock
  ...
  j1939_sk_netdev_event_netdown -> j1939_socks_lock
  j1939_netdev_notify

1
  j1939_sk_errqueue -> j1939_socks_lock
  __j1939_session_cancel -> active_session_list_lock
  j1939_tp_rxtimer

       CPU0                    CPU1
       ----                    ----
  lock(&priv->active_session_list_lock);
                               lock(&jsk->sk_session_queue_lock);
                               lock(&priv->active_session_list_lock);
  lock(&priv->j1939_socks_lock);

The solution implemented in this commit is to move the
j1939_sk_errqueue() call out of the active_session_list_lock context,
thus preventing the deadlock situation.

Reported-by: syzbot+ee1cd780f69483a8616b@...kaller.appspotmail.com
Fixes: 5b9272e93f2e ("can: j1939: extend UAPI to notify about RX status")
Co-developed-by: Hillf Danton <hdanton@...a.com>
Signed-off-by: Hillf Danton <hdanton@...a.com>
Signed-off-by: Oleksij Rempel <o.rempel@...gutronix.de>
---
 net/can/j1939/transport.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/net/can/j1939/transport.c b/net/can/j1939/transport.c
index fce9b9ebf13f..fb92c3609e17 100644
--- a/net/can/j1939/transport.c
+++ b/net/can/j1939/transport.c
@@ -1124,8 +1124,6 @@ static void __j1939_session_cancel(struct j1939_session *session,
 
 	if (session->sk)
 		j1939_sk_send_loop_abort(session->sk, session->err);
-	else
-		j1939_sk_errqueue(session, J1939_ERRQUEUE_RX_ABORT);
 }
 
 static void j1939_session_cancel(struct j1939_session *session,
@@ -1140,6 +1138,9 @@ static void j1939_session_cancel(struct j1939_session *session,
 	}
 
 	j1939_session_list_unlock(session->priv);
+
+	if (!session->sk)
+		j1939_sk_errqueue(session, J1939_ERRQUEUE_RX_ABORT);
 }
 
 static enum hrtimer_restart j1939_tp_txtimer(struct hrtimer *hrtimer)
@@ -1253,6 +1254,9 @@ static enum hrtimer_restart j1939_tp_rxtimer(struct hrtimer *hrtimer)
 			__j1939_session_cancel(session, J1939_XTP_ABORT_TIMEOUT);
 		}
 		j1939_session_list_unlock(session->priv);
+
+		if (!session->sk)
+			j1939_sk_errqueue(session, J1939_ERRQUEUE_RX_ABORT);
 	}
 
 	j1939_session_put(session);
-- 
2.30.2