Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To:     linux-kernel@vger.kernel.org
Cc:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        stable@vger.kernel.org, Duoming Zhou <duoming@zju.edu.cn>,
        "David S. Miller" <davem@davemloft.net>,
        Sasha Levin <sashal@kernel.org>
Subject: [PATCH 5.16 047/227] ax25: improve the incomplete fix to avoid UAF and NPD bugs
Date:   Mon, 21 Feb 2022 09:47:46 +0100
Message-Id: <20220221084936.444795974@linuxfoundation.org>
In-Reply-To: <20220221084934.836145070@linuxfoundation.org>
References: <20220221084934.836145070@linuxfoundation.org>
User-Agent: quilt/0.66
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Duoming Zhou <duoming@zju.edu.cn>

[ Upstream commit 4e0f718daf97d47cf7dec122da1be970f145c809 ]

The previous commit 1ade48d0c27d ("ax25: NPD bug when detaching
AX25 device") introduce lock_sock() into ax25_kill_by_device to
prevent NPD bug. But the concurrency NPD or UAF bug will occur,
when lock_sock() or release_sock() dereferences the ax25_cb->sock.

The NULL pointer dereference bug can be shown as below:

ax25_kill_by_device()        | ax25_release()
                             |   ax25_destroy_socket()
                             |     ax25_cb_del()
  ...                        |     ...
                             |     ax25->sk=NULL;
  lock_sock(s->sk); //(1)    |
  s->ax25_dev = NULL;        |     ...
  release_sock(s->sk); //(2) |
  ...                        |

The root cause is that the sock is set to null before dereference
site (1) or (2). Therefore, this patch extracts the ax25_cb->sock
in advance, and uses ax25_list_lock to protect it, which can synchronize
with ax25_cb_del() and ensure the value of sock is not null before
dereference sites.

The concurrency UAF bug can be shown as below:

ax25_kill_by_device()        | ax25_release()
                             |   ax25_destroy_socket()
  ...                        |   ...
                             |   sock_put(sk); //FREE
  lock_sock(s->sk); //(1)    |
  s->ax25_dev = NULL;        |   ...
  release_sock(s->sk); //(2) |
  ...                        |

The root cause is that the sock is released before dereference
site (1) or (2). Therefore, this patch uses sock_hold() to increase
the refcount of sock and uses ax25_list_lock to protect it, which
can synchronize with ax25_cb_del() in ax25_destroy_socket() and
ensure the sock wil not be released before dereference sites.

Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/ax25/af_ax25.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/net/ax25/af_ax25.c b/net/ax25/af_ax25.c
index 02f43f3e2c564..44a8730c26acc 100644
--- a/net/ax25/af_ax25.c
+++ b/net/ax25/af_ax25.c
@@ -77,6 +77,7 @@ static void ax25_kill_by_device(struct net_device *dev)
 {
 	ax25_dev *ax25_dev;
 	ax25_cb *s;
+	struct sock *sk;
 
 	if ((ax25_dev = ax25_dev_ax25dev(dev)) == NULL)
 		return;
@@ -85,13 +86,15 @@ static void ax25_kill_by_device(struct net_device *dev)
 again:
 	ax25_for_each(s, &ax25_list) {
 		if (s->ax25_dev == ax25_dev) {
+			sk = s->sk;
+			sock_hold(sk);
 			spin_unlock_bh(&ax25_list_lock);
-			lock_sock(s->sk);
+			lock_sock(sk);
 			s->ax25_dev = NULL;
-			release_sock(s->sk);
+			release_sock(sk);
 			ax25_disconnect(s, ENETUNREACH);
 			spin_lock_bh(&ax25_list_lock);
-
+			sock_put(sk);
 			/* The entry could have been deleted from the
 			 * list meanwhile and thus the next pointer is
 			 * no longer valid.  Play it safe and restart
-- 
2.34.1