[PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Amit K Bag]

From: Chethan T N <chethan.tumkur.narayan@intel.com>

Observed race condition during controller recovery mechanism
resulting the controller not responding to the reset command.

To avoid such race condition need a delay of 2ms soon after the
USB re-enumeration and before sending the Reset command which shall
allow controller to completely recover and process the Reset command.

BUG=b:64035404
TEST=Magic tethering, P0 and P1 santiy test cases, kernel-boot log

Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Amit K Bag <amit.k.bag@intel.com>
---
 drivers/bluetooth/btusb.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index f73a27ea28cc..c7fce9c11889 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -1743,6 +1743,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
 
 	BT_DBG("%s", hdev->name);
 
+	/* Observed race condition during controller recovery mechanism
+	 * resulting the controller not responding to the reset command.
+	 *
+	 * To avoid such race condition need a delay of 2ms soon after the
+	 * USB re-enumeration and before sending the Reset command which shall
+	 * allow controller to completely recover and process the Reset command.
+	 */
+	mdelay(2);
+
 	/* The controller has a bug with the first HCI command sent to it
 	 * returning number of completed commands as zero. This would stall the
 	 * command processing in the Bluetooth core.
-- 
2.7.4

RE: [PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Bag, Amit K]

Hello Marcel,

> Observed race condition during controller recovery mechanism resulting the controller not responding to the reset command.
>
> To avoid such race condition need a delay of 2ms soon after the USB re-enumeration and before sending the Reset command which shall allow controller to completely recover and process the Reset command.
>
> BUG=b:64035404
> TEST=Magic tethering, P0 and P1 santiy test cases, kernel-boot log
>
> Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
> Signed-off-by: Amit K Bag <amit.k.bag@intel.com>
> ---
 > drivers/bluetooth/btusb.c | 9 +++++++++
 > 1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c index f73a27ea28cc..c7fce9c11889 100644
> --- a/drivers/bluetooth/btusb.c
> +++ b/drivers/bluetooth/btusb.c
> @@ -1743,6 +1743,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
> 
> 	BT_DBG("%s", hdev->name);
 >
> +	/* Observed race condition during controller recovery mechanism
> +	 * resulting the controller not responding to the reset command.
> +	 *
> +	 * To avoid such race condition need a delay of 2ms soon after the
> +	 * USB re-enumeration and before sending the Reset command which shall
> +	 * allow controller to completely recover and process the Reset command.
> +	 */
> +	mdelay(2);
> +

I will provide a new patch. Please ignore this patch. 

Thanks & Regards,
Amit Bag

[PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Amit K Bag]

From: Chethan T N <chethan.tumkur.narayan@intel.com>

Observed race condition during controller recovery mechanism
resulting the controller not responding to the reset command.

To avoid such race condition need a delay of 2ms soon after the
USB re-enumeration and before sending the Reset command which shall
allow controller to completely recover and process the Reset command.

BUG=b:64035404
TEST=Magic tethering, P0 and P1 santiy test cases, kernel-boot log

Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
Signed-off-by: Amit K Bag <amit.k.bag@intel.com>
---
 drivers/bluetooth/btusb.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index f73a27ea28cc..2303ac037240 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -1743,6 +1743,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
 
 	BT_DBG("%s", hdev->name);
 
+	/* Observed race condition during controller recovery mechanism
+	 * resulting the controller not responding to the reset command.
+	 *
+	 * To avoid such race condition need a delay of 2ms soon after the
+	 * USB re-enumeration and before sending the Reset command which shall
+	 * allow controller to completely recover and process the Reset command.
+	 */
+	mdelay(5);
+
 	/* The controller has a bug with the first HCI command sent to it
 	 * returning number of completed commands as zero. This would stall the
 	 * command processing in the Bluetooth core.
-- 
2.7.4

Re: [PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Luiz Augusto von Dentz]

Hi Amit,

On Wed, Jun 6, 2018 at 9:22 AM, Amit K Bag <amit.k.bag@intel.com> wrote:
> From: Chethan T N <chethan.tumkur.narayan@intel.com>
>
> Observed race condition during controller recovery mechanism
> resulting the controller not responding to the reset command.
>
> To avoid such race condition need a delay of 2ms soon after the
> USB re-enumeration and before sending the Reset command which shall
> allow controller to completely recover and process the Reset command.
>
> BUG=b:64035404
> TEST=Magic tethering, P0 and P1 santiy test cases, kernel-boot log
>
> Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
> Signed-off-by: Amit K Bag <amit.k.bag@intel.com>
> ---
>  drivers/bluetooth/btusb.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
>
> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
> index f73a27ea28cc..2303ac037240 100644
> --- a/drivers/bluetooth/btusb.c
> +++ b/drivers/bluetooth/btusb.c
> @@ -1743,6 +1743,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
>
>         BT_DBG("%s", hdev->name);
>
> +       /* Observed race condition during controller recovery mechanism
> +        * resulting the controller not responding to the reset command.
> +        *
> +        * To avoid such race condition need a delay of 2ms soon after the
> +        * USB re-enumeration and before sending the Reset command which shall
> +        * allow controller to completely recover and process the Reset command.
> +        */
> +       mdelay(5);

Looks like the commit message and comment are not longer in sync with
the actual value here. Also perhaps this could be a parameter so any
hardware that has this problem just set how much delay it needs before
setup can be called.

>         /* The controller has a bug with the first HCI command sent to it
>          * returning number of completed commands as zero. This would stall the
>          * command processing in the Bluetooth core.
> --
> 2.7.4
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Luiz Augusto von Dentz

[PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Amit K Bag]

From: Chethan T N <chethan.tumkur.narayan@intel.com>

Observed race condition during controller recovery mechanism
resulting the controller not responding to the reset command.

To avoid such race condition need a delay of 2ms soon after the
USB re-enumeration and before sending the Reset command which shall
allow controller to completely recover and process the Reset command.

Change-Id: Ifbc387d4377bc79081244d444d45ba6effc95f23
Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
---
 drivers/bluetooth/btusb.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index f73a27ea28cc..844b4145ff76 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -1741,6 +1741,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
 	int disable_patch, err;
 	struct intel_version ver;
 
+	/* Observed race condition during controller recovery mechanism
+	 * resulting the controller not responding to the reset command.
+	 *
+	 * To avoid such race condition need a delay of 2ms soon after the
+	 * USB re-enumeration and before sending the Reset command which shall
+	 * allow controller to completely recover and process the Reset command.
+	 */
+	mdelay(2);
+
 	BT_DBG("%s", hdev->name);
 
 	/* The controller has a bug with the first HCI command sent to it
-- 
2.7.4

Re: [PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Marcel Holtmann]

Hi Amit,

> Observed race condition during controller recovery mechanism
> resulting the controller not responding to the reset command.
> 
> To avoid such race condition need a delay of 2ms soon after the
> USB re-enumeration and before sending the Reset command which shall
> allow controller to completely recover and process the Reset command.
> 
> Change-Id: Ifbc387d4377bc79081244d444d45ba6effc95f23
> Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>
> ---
> drivers/bluetooth/btusb.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
> index f73a27ea28cc..844b4145ff76 100644
> --- a/drivers/bluetooth/btusb.c
> +++ b/drivers/bluetooth/btusb.c
> @@ -1741,6 +1741,15 @@ static int btusb_setup_intel(struct hci_dev *hdev)
> 	int disable_patch, err;
> 	struct intel_version ver;
> 
> +	/* Observed race condition during controller recovery mechanism
> +	 * resulting the controller not responding to the reset command.
> +	 *
> +	 * To avoid such race condition need a delay of 2ms soon after the
> +	 * USB re-enumeration and before sending the Reset command which shall
> +	 * allow controller to completely recover and process the Reset command.
> +	 */
> +	mdelay(2);
> +
> 	BT_DBG("%s", hdev->name);

can you leave the BT_DBG() at the top and move the mdelay + comment after it.

Regards

Marcel

Re: [PATCH] Bluetooth: btusb: Avoid race condition during controller recovery

[Marcel Holtmann]

Hi Amit,

> Observed race condition during controller recovery mechanism
> resulting the controller not responding to the reset command.
> 
> To avoid such race condition need a delay of 2ms soon after the
> USB re-enumeration and before sending the Reset command which shall
> allow controller to completely recover and process the Reset command.
> 
> Change-Id: Ifbc387d4377bc79081244d444d45ba6effc95f23
> Signed-off-by: Chethan T N <chethan.tumkur.narayan@intel.com>

oh, and no Change-Id please and you also need to add your Signed-off-by here.

Regards

Marcel