Re: [PATCH v2 1/2] Bluetooth: Introduce HCI_CONN_FLAG_DEVICE_PRIVACY device flag

[Marcel Holtmann <marcel@holtmann.org>]

Hi Luiz,

> This introduces HCI_CONN_FLAG_DEVICE_PRIVACY which can be used by
> userspace to indicate to the controller to use Device Privacy Mode to a
> specific device.
> 
> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
> ---
> v2: Fix supported flags not actually checking if the hdev really
> supports the flags.
> 
> include/net/bluetooth/hci_core.h |  4 ++++
> net/bluetooth/mgmt.c             | 30 ++++++++++++++++++++++++++----
> 2 files changed, 30 insertions(+), 4 deletions(-)
> 
> diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/hci_core.h
> index 2560cfe80db8..42ba40df6e20 100644
> --- a/include/net/bluetooth/hci_core.h
> +++ b/include/net/bluetooth/hci_core.h
> @@ -160,6 +160,7 @@ struct bdaddr_list_with_flags {
> 
> enum hci_conn_flags {
> 	HCI_CONN_FLAG_REMOTE_WAKEUP,
> +	HCI_CONN_FLAG_DEVICE_PRIVACY,
> 	HCI_CONN_FLAG_MAX
> };
> 
> @@ -1465,6 +1466,9 @@ void hci_conn_del_sysfs(struct hci_conn *conn);
> #define use_ll_privacy(dev) (ll_privacy_capable(dev) && \
> 			     hci_dev_test_flag(dev, HCI_ENABLE_LL_PRIVACY))
> 
> +#define privacy_mode_capable(dev) (use_ll_privacy(dev) && \
> +				   (hdev->commands[39] & 0x04))
> +
> /* Use enhanced synchronous connection if command is supported */
> #define enhanced_sco_capable(dev) ((dev)->commands[29] & 0x08)
> 
> diff --git a/net/bluetooth/mgmt.c b/net/bluetooth/mgmt.c
> index f8f74d344297..d82d1a62754a 100644
> --- a/net/bluetooth/mgmt.c
> +++ b/net/bluetooth/mgmt.c
> @@ -4349,7 +4349,22 @@ static int set_exp_feature(struct sock *sk, struct hci_dev *hdev,
> 			       MGMT_STATUS_NOT_SUPPORTED);
> }
> 
> -#define SUPPORTED_DEVICE_FLAGS() ((1U << HCI_CONN_FLAG_MAX) - 1)
> +static u32 supported_device_flags(struct hci_dev *hdev)
> +{
> +	u32 flags = 0;
> +
> +	/* Check if adapter can wakeup the system */
> +	if (hdev->wakeup && hdev->wakeup(hdev))
> +		flags |= BIT(HCI_CONN_FLAG_REMOTE_WAKEUP);

I would do this change as a separate patch since it has nothing to do with the device privacy setting.

Do we have to call hdev->wakeup() as well here? Isn’t the existence of the callback enough indication.

That also said, doesn’t it make sense to store the supported_flags in the device params struct. It would make it certainly easy to return. Potentially we have a few calls to Get Device Flags and its notifications that have to rebuild this flags field over and over again.

Regards

Marcel