Hey linux-bluetooth,
We found this bug when reverting some Chromium maintained patches in
our repository that was conditionally dropping LE scan enable commands
if it wasn't toggling between true/false. On some Intel controllers,
disabling LE scan when it's already disabled resulted in a "Command
Disallowed" and this was causing suspend to fail.
On Mon, Jun 1, 2020 at 6:43 PM Manish Mandlik <mmandlik@xxxxxxxxxx> wrote:
>
> Check current scan state by checking HCI_LE_SCAN flag and send scan
> disable command only if scan is already enabled.
>
> Signed-off-by: Manish Mandlik <mmandlik@xxxxxxxxxx>
> ---
>
> net/bluetooth/hci_request.c | 10 ++++++----
> 1 file changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/net/bluetooth/hci_request.c b/net/bluetooth/hci_request.c
> index 1fc55685da62d..1acf5b8e0910c 100644
> --- a/net/bluetooth/hci_request.c
> +++ b/net/bluetooth/hci_request.c
> @@ -998,8 +998,9 @@ static void hci_req_set_event_filter(struct hci_request *req)
>
> static void hci_req_config_le_suspend_scan(struct hci_request *req)
> {
> - /* Can't change params without disabling first */
> - hci_req_add_le_scan_disable(req);
> + /* Before changing params disable scan if enabled */
> + if (hci_dev_test_flag(req->hdev, HCI_LE_SCAN))
> + hci_req_add_le_scan_disable(req);
>
> /* Configure params and enable scanning */
> hci_req_add_le_passive_scan(req);
> @@ -1065,8 +1066,9 @@ void hci_req_prepare_suspend(struct hci_dev *hdev, enum suspended_state next)
> page_scan = SCAN_DISABLED;
> hci_req_add(&req, HCI_OP_WRITE_SCAN_ENABLE, 1, &page_scan);
>
> - /* Disable LE passive scan */
> - hci_req_add_le_scan_disable(&req);
> + /* Disable LE passive scan if enabled */
> + if (hci_dev_test_flag(hdev, HCI_LE_SCAN))
> + hci_req_add_le_scan_disable(&req);
>
> /* Mark task needing completion */
> set_bit(SUSPEND_SCAN_DISABLE, hdev->suspend_tasks);
> --
> 2.27.0.rc2.251.g90737beb825-goog
>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@xxxxxxxxxxxx>
