All of lore.kernel.org
 help / color / mirror / Atom feed
From: Luiz Augusto von Dentz <luiz.dentz@gmail.com>
To: linux-bluetooth@vger.kernel.org
Subject: [PATCH v3 2/2] Bluetooth: L2CAP: Don't offer deferred channels before security is met
Date: Fri, 20 Nov 2026 10:48:16 -0500	[thread overview]
Message-ID: <20261120154816.1745952-2-luiz.dentz@gmail.com> (raw)
In-Reply-To: <20261120154816.1745952-1-luiz.dentz@gmail.com>

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

A BR/EDR channel is moved to BT_CONNECT2 as soon as the connection
request is answered as pending, no matter if it is waiting for the
information exchange, for the link security or for userspace to
authorize it, and the socket state follows the channel state.

bt_accept_poll() assumes that a child in BT_CONNECT2 of a listener
using BT_DEFER_SETUP is waiting to be authorized, and
bt_accept_dequeue() returns any child of such a listener, so although
the listener is only notified once the security is met, a listener
that doesn't wait for it can accept() the child and then authorize it
with recvmsg() while authentication is still pending.
l2cap_sock_defer() then moves the channel to BT_CONFIG and responds
with L2CAP_CR_SUCCESS, and if authentication fails afterwards the
channel is left open since l2cap_security_cfm() doesn't close channels
in BT_CONFIG.

Fix this by only moving the socket to BT_CONNECT2 when the channel is
offered to the listener with l2cap_sock_defer_cb(), which is only
called once the security is met, and by only letting accept() return a
child of a listener using BT_DEFER_SETUP once it is in BT_CONNECT2, as
bt_accept_poll() does. The child stays in the accept queue so the
backlog limit and the cleanup done when the listener is closed are not
affected.

Fixes: f66dc81f44d9 ("Bluetooth: Add support for deferring L2CAP connection setup")
Assisted-by: OpenCode:claude-fable-5.2
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
---
v3: Only let accept() return children in BT_CONNECT2, as
    bt_accept_poll() does, instead of only skipping the ones in
    BT_OPEN.
v2: New patch.
---
 net/bluetooth/af_bluetooth.c |  6 +++++-
 net/bluetooth/l2cap_sock.c   | 10 +++++++++-
 2 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/net/bluetooth/af_bluetooth.c b/net/bluetooth/af_bluetooth.c
index 7758e9ea3848..103d61d13fbf 100644
--- a/net/bluetooth/af_bluetooth.c
+++ b/net/bluetooth/af_bluetooth.c
@@ -330,8 +330,12 @@ struct sock *bt_accept_dequeue(struct sock *parent, struct socket *newsock)
 			continue;
 		}
 
+		/* With deferred setup only a child in BT_CONNECT2 has been
+		 * offered to the listener, see bt_accept_poll().
+		 */
 		if (sk->sk_state == BT_CONNECTED || !newsock ||
-		    test_bit(BT_SK_DEFER_SETUP, &bt_sk(parent)->flags)) {
+		    (test_bit(BT_SK_DEFER_SETUP, &bt_sk(parent)->flags) &&
+		     sk->sk_state == BT_CONNECT2)) {
 			bt_accept_unlink(sk);
 			if (newsock)
 				sock_graft(sk, newsock);
diff --git a/net/bluetooth/l2cap_sock.c b/net/bluetooth/l2cap_sock.c
index d5663a4642a9..57ada00f76eb 100644
--- a/net/bluetooth/l2cap_sock.c
+++ b/net/bluetooth/l2cap_sock.c
@@ -1812,7 +1812,13 @@ static void l2cap_sock_state_change_cb(struct l2cap_chan *chan, int state,
 
 	lock_sock(sk);
 
-	sk->sk_state = state;
+	/* BT_CONNECT2 also covers a channel still waiting for the information
+	 * exchange or link security. Only l2cap_sock_defer_cb() moves the
+	 * socket to BT_CONNECT2, once the channel can be offered to the
+	 * listener.
+	 */
+	if (state != BT_CONNECT2)
+		sk->sk_state = state;
 
 	if (err)
 		sk->sk_err = err;
@@ -1880,6 +1886,8 @@ static void l2cap_sock_defer_cb(struct l2cap_chan *chan)
 
 	lock_sock(sk);
 
+	sk->sk_state = BT_CONNECT2;
+
 	parent = bt_sk(sk)->parent;
 	if (parent)
 		parent->sk_data_ready(parent);
-- 
2.56.0


  reply	other threads:[~2026-11-20 15:48 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-20 15:48 [PATCH v3 1/2] Bluetooth: L2CAP: Fix accepting channels with insufficient security Luiz Augusto von Dentz
2026-11-20 15:48 ` Luiz Augusto von Dentz [this message]
2026-11-20 21:37 ` [v3,1/2] " bluez.test.bot

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20261120154816.1745952-2-luiz.dentz@gmail.com \
    --to=luiz.dentz@gmail.com \
    --cc=linux-bluetooth@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.