Re: [PATCHv2] Bluetooth: btusb: Fix Intel controller hang after shutdown

[Tedd Ho-Jeong An <tedd.an@intel.com>]

Hi Loci,

On Thu, 4 Jun 2015 12:55:32 -0700
"An, Tedd" <tedd.an@intel.com> wrote:

> Hi Loic
> 
> On 6/4/15, 4:59 AM, "Poulain, Loic" <loic.poulain@intel.com> wrote:
> 
> >Hi tedd,
> >
> >On 21/05/2015 21:56, Marcel Holtmann wrote:
> >> Hi Tedd,
> >>
> >>> The vendor command (0xfc3f) in btusb_shutdown_intel causes an internal
> >>> reset with old firmware. After that, upcoming HCI command will hang due
> >>> to a controller bug with the first HCI command sent to it.
> >>> The workaround is to send HCI Reset command first which will reset the
> >>> number of completed commands.
> >>>
> >>> Signed-off-by: Loic Poulain <loic.poulain@intel.com>
> >>> ---
> >>> v2: Send HCI reset from shutdown instead of using HCI_QUIRK_RESET_ON_CLOSE
> >>> drivers/bluetooth/btusb.c | 13 +++++++++++++
> >>> 1 file changed, 13 insertions(+)
> >>>
> >>> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
> >>> index d21f3b4..589f30c 100644
> >>> --- a/drivers/bluetooth/btusb.c
> >>> +++ b/drivers/bluetooth/btusb.c
> >>> @@ -2705,6 +2705,19 @@ static int btusb_shutdown_intel(struct hci_dev *hdev)
> >>> 	}
> >>> 	kfree_skb(skb);
> >>>
> >>> +	/* The 0xfc3f command causes an internal reset with old firmware.
> >>> +	 * In order to avoid the Bug with the first HCI command, we have
> >>> +	 * to send a HCI reset which will reset the number of completed
> >>> +	 * command.
> >>> +	 */
> >>> +	skb = __hci_cmd_sync(hdev, HCI_OP_RESET, 0, NULL, HCI_INIT_TIMEOUT);
> >>> +	if (IS_ERR(skb)) {
> >>> +		BT_ERR("%s: Shutdown HCI reset command failed (%ld)",
> >>> +			hdev->name, PTR_ERR(skb));
> >>> +		return PTR_ERR(skb);
> >>> +	}
> >>> +	kfree_skb(skb);
> >>> +
> >> please test this change and see if it still has the desired effect. ACK or NACK the patch based on your testing.
> >>
> >> Regards
> >>
> >> Marcel
> >>
> >
> >Any chance you test this patch?
> 
> I missed your patch and just got it. I will run this test and let you know the result.
> 

We tested this patch and it re-introduces the LED issue.
The 0xfc3f command turns off the BT LED immediately and BT LED behavior is reset to normal statue (ON)
with HCI_RESET command. So, sending HCI_RESET immediately after 0xfc3f will turn on the LED
back on and it will take another 5 seconds to turn-off again.

However, if HCI_RESET is sent prior to 0xfc3f, it worked ok, if you want to reset the device during
the power off.

> >
> >Regards,
> >Loic
> >
> >-- 
> >Intel Open Source Technology Center
> >http://oss.intel.com/
> >