2014-06-13 07:22:28

by Johan Hedberg

Subject: [PATCH] Bluetooth: Fix locking of hdev when calling into SMP code

From: Johan Hedberg <[email protected]>

The SMP code expects hdev to be unlocked since e.g. crypto functions
will try to (re)lock it. Therefore, we need to release the lock before
calling into smp.c from mgmt.c. Without this we risk a deadlock whenever
the smp_user_confirm_reply() function is called.

Signed-off-by: Johan Hedberg <[email protected]>
---
net/bluetooth/mgmt.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/bluetooth/mgmt.c b/net/bluetooth/mgmt.c
index 6107e037cd8e..af8e0a6243b7 100644
--- a/net/bluetooth/mgmt.c
+++ b/net/bluetooth/mgmt.c
@@ -3031,8 +3031,13 @@ static int user_pairing_resp(struct sock *sk, struct hci_dev *hdev,
}

if (addr->type == BDADDR_LE_PUBLIC || addr->type == BDADDR_LE_RANDOM) {
- /* Continue with pairing via SMP */
+ /* Continue with pairing via SMP. The hdev lock must be
+ * released as SMP may try to recquire it for crypto
+ * purposes.
+ */
+ hci_dev_unlock(hdev);
err = smp_user_confirm_reply(conn, mgmt_op, passkey);
+ hci_dev_lock(hdev);

if (!err)
err = cmd_complete(sk, hdev->id, mgmt_op,
--
1.9.3

2014-06-13 11:18:33

by Marcel Holtmann

Subject: Re: [PATCH] Bluetooth: Fix locking of hdev when calling into SMP code

Hi Johan,

> The SMP code expects hdev to be unlocked since e.g. crypto functions
> will try to (re)lock it. Therefore, we need to release the lock before
> calling into smp.c from mgmt.c. Without this we risk a deadlock whenever
> the smp_user_confirm_reply() function is called.
>
> Signed-off-by: Johan Hedberg <[email protected]>
> ---
> net/bluetooth/mgmt.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)

patch has been applied to bluetooth tree.

Regards

Marcel

2014-06-13 09:18:38

by Marcel Holtmann

Subject: Re: [PATCH] Bluetooth: Fix locking of hdev when calling into SMP code

Hi Johan,

> The SMP code expects hdev to be unlocked since e.g. crypto functions
> will try to (re)lock it. Therefore, we need to release the lock before
> calling into smp.c from mgmt.c. Without this we risk a deadlock whenever
> the smp_user_confirm_reply() function is called.
>
> Signed-off-by: Johan Hedberg <[email protected]>
> ---
> net/bluetooth/mgmt.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/net/bluetooth/mgmt.c b/net/bluetooth/mgmt.c
> index 6107e037cd8e..af8e0a6243b7 100644
> --- a/net/bluetooth/mgmt.c
> +++ b/net/bluetooth/mgmt.c
> @@ -3031,8 +3031,13 @@ static int user_pairing_resp(struct sock *sk, struct hci_dev *hdev,
> }
>
> if (addr->type == BDADDR_LE_PUBLIC || addr->type == BDADDR_LE_RANDOM) {
> - /* Continue with pairing via SMP */
> + /* Continue with pairing via SMP. The hdev lock must be
> + * released as SMP may try to recquire it for crypto
> + * purposes.
> + */
> + hci_dev_unlock(hdev);
> err = smp_user_confirm_reply(conn, mgmt_op, passkey);
> + hci_dev_lock(hdev);

providing a __smp_user_confirm_reply that operates on a locked hdev and where the crypto functions do not take the hdev lock is not possible.

The lock/unlock seems a bit counterproductive here.

Regards

Marcel

2014-06-13 07:42:53

by Lukasz Rymanowski

Subject: Re: [PATCH] Bluetooth: Fix locking of hdev when calling into SMP code

Hi Johan,

On Fri, Jun 13, 2014 at 9:22 AM, <[email protected]> wrote:
> From: Johan Hedberg <[email protected]>
>
> The SMP code expects hdev to be unlocked since e.g. crypto functions
> will try to (re)lock it. Therefore, we need to release the lock before
> calling into smp.c from mgmt.c. Without this we risk a deadlock whenever
> the smp_user_confirm_reply() function is called.
>
> Signed-off-by: Johan Hedberg <[email protected]>
> ---
> net/bluetooth/mgmt.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/net/bluetooth/mgmt.c b/net/bluetooth/mgmt.c
> index 6107e037cd8e..af8e0a6243b7 100644
> --- a/net/bluetooth/mgmt.c
> +++ b/net/bluetooth/mgmt.c
> @@ -3031,8 +3031,13 @@ static int user_pairing_resp(struct sock *sk, struct hci_dev *hdev,
> }
>
> if (addr->type == BDADDR_LE_PUBLIC || addr->type == BDADDR_LE_RANDOM) {
> - /* Continue with pairing via SMP */
> + /* Continue with pairing via SMP. The hdev lock must be
> + * released as SMP may try to recquire it for crypto
> + * purposes.
> + */
> + hci_dev_unlock(hdev);
> err = smp_user_confirm_reply(conn, mgmt_op, passkey);
> + hci_dev_lock(hdev);
>
> if (!err)
> err = cmd_complete(sk, hdev->id, mgmt_op,
> --
> 1.9.3

Works fine.

Tested-by: Lukasz Rymanowski <[email protected]>

BR
\Łukasz
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html