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]
Date:   Wed, 11 Mar 2020 19:20:14 -0700
From:   Manish Mandlik <mmandlik@...gle.com>
To:     marcel@...tmann.org
Cc:     Alain Michaud <alainm@...omium.org>,
        linux-bluetooth@...r.kernel.org,
        Miao-chen Chou <mcchou@...omium.org>,
        Joseph Hwang <josephsih@...omium.org>,
        Yoni Shavit <yshavit@...omium.org>,
        Manish Mandlik <mmandlik@...gle.com>,
        "David S. Miller" <davem@...emloft.net>,
        Johan Hedberg <johan.hedberg@...il.com>,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        Jakub Kicinski <kuba@...nel.org>
Subject: [PATCH v2] Bluetooth: clean up connection in hci_cs_disconnect

From: Joseph Hwang <josephsih@...omium.org>

In bluetooth core specification 4.2,
Vol 2, Part E, 7.8.9 LE Set Advertise Enable Command, it says

    The Controller shall continue advertising until ...
    or until a connection is created or ...
    In these cases, advertising is then disabled.

Hence, advertising would be disabled before a connection is
established. In current kernel implementation, advertising would
be re-enabled when all connections are terminated.

The correct disconnection flow looks like

  < HCI Command: Disconnect

  > HCI Event: Command Status
      Status: Success

  > HCI Event: Disconnect Complete
      Status: Success

Specifically, the last Disconnect Complete Event would trigger a
callback function hci_event.c:hci_disconn_complete_evt() to
cleanup the connection and re-enable advertising when proper.

However, sometimes, there might occur an exception in the controller
when disconnection is being executed. The disconnection flow might
then look like

  < HCI Command: Disconnect

  > HCI Event: Command Status
      Status: Unknown Connection Identifier

  Note that "> HCI Event: Disconnect Complete" is missing when such an
exception occurs. This would result in advertising staying disabled
forever since the connection in question is not cleaned up correctly.

To fix the controller exception issue, we need to do some connection
cleanup when the disconnect command status indicates an error.

Signed-off-by: Joseph Hwang <josephsih@...omium.org>
Signed-off-by: Manish Mandlik <mmandlik@...gle.com>
---

Changes in v2:
- Moved "u8 type" declaration inside if block

 net/bluetooth/hci_event.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
index a40ed31f6eb8f..a116114279107 100644
--- a/net/bluetooth/hci_event.c
+++ b/net/bluetooth/hci_event.c
@@ -2202,10 +2202,22 @@ static void hci_cs_disconnect(struct hci_dev *hdev, u8 status)
 	hci_dev_lock(hdev);
 
 	conn = hci_conn_hash_lookup_handle(hdev, __le16_to_cpu(cp->handle));
-	if (conn)
+	if (conn) {
+		u8 type = conn->type;
+
 		mgmt_disconnect_failed(hdev, &conn->dst, conn->type,
 				       conn->dst_type, status);
 
+		/* If the disconnection failed for any reason, the upper layer
+		 * does not retry to disconnect in current implementation.
+		 * Hence, we need to do some basic cleanup here and re-enable
+		 * advertising if necessary.
+		 */
+		hci_conn_del(conn);
+		if (type == LE_LINK)
+			hci_req_reenable_advertising(hdev);
+	}
+
 	hci_dev_unlock(hdev);
 }
 
-- 
2.25.1.481.gfbce0eb801-goog

Powered by blists - more mailing lists