[PATCH] Bluetooth: btusb: request wake pin with NOAUTOEN

[PATCH] Bluetooth: btusb: request wake pin with NOAUTOEN

[Brian Norris]

Badly-designed systems might have (for example) active-high wake pins
that default to high (e.g., because of external pull ups) until they
have an active firmware which starts driving it low. This can cause an
interrupt storm in the time between request_irq() and disable_irq().

We don't support shared interrupts here, so let's just pre-configure the
interrupt to avoid auto-enabling it.

Fixes: fd913ef7ce61 ("Bluetooth: btusb: Add out-of-band wakeup support")
Signed-off-by: Brian Norris <briannorris@chromium.org>
---
 drivers/bluetooth/btusb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
index 4761499db9ee..470ee68555d9 100644
--- a/drivers/bluetooth/btusb.c
+++ b/drivers/bluetooth/btusb.c
@@ -2885,6 +2885,7 @@ static int btusb_config_oob_wake(struct hci_dev *hdev)
 		return 0;
 	}
 
+	irq_set_status_flags(irq, IRQ_NOAUTOEN);
 	ret = devm_request_irq(&hdev->dev, irq, btusb_oob_wake_handler,
 			       0, "OOB Wake-on-BT", data);
 	if (ret) {
@@ -2899,7 +2900,6 @@ static int btusb_config_oob_wake(struct hci_dev *hdev)
 	}
 
 	data->oob_wake_irq = irq;
-	disable_irq(irq);
 	bt_dev_info(hdev, "OOB Wake-on-BT configured at IRQ %u", irq);
 	return 0;
 }
-- 
2.21.0.rc0.258.g878e2cd30e-goog

Re: [PATCH] Bluetooth: btusb: request wake pin with NOAUTOEN

[Matthias Kaehlcke]

On Thu, Feb 21, 2019 at 04:30:51PM -0800, Brian Norris wrote:
> Badly-designed systems might have (for example) active-high wake pins
> that default to high (e.g., because of external pull ups) until they
> have an active firmware which starts driving it low. This can cause an
> interrupt storm in the time between request_irq() and disable_irq().
> 
> We don't support shared interrupts here, so let's just pre-configure the
> interrupt to avoid auto-enabling it.
> 
> Fixes: fd913ef7ce61 ("Bluetooth: btusb: Add out-of-band wakeup support")
> Signed-off-by: Brian Norris <briannorris@chromium.org>
> ---
>  drivers/bluetooth/btusb.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/bluetooth/btusb.c b/drivers/bluetooth/btusb.c
> index 4761499db9ee..470ee68555d9 100644
> --- a/drivers/bluetooth/btusb.c
> +++ b/drivers/bluetooth/btusb.c
> @@ -2885,6 +2885,7 @@ static int btusb_config_oob_wake(struct hci_dev *hdev)
>  		return 0;
>  	}
>  
> +	irq_set_status_flags(irq, IRQ_NOAUTOEN);
>  	ret = devm_request_irq(&hdev->dev, irq, btusb_oob_wake_handler,
>  			       0, "OOB Wake-on-BT", data);
>  	if (ret) {
> @@ -2899,7 +2900,6 @@ static int btusb_config_oob_wake(struct hci_dev *hdev)
>  	}
>  
>  	data->oob_wake_irq = irq;
> -	disable_irq(irq);
>  	bt_dev_info(hdev, "OOB Wake-on-BT configured at IRQ %u", irq);
>  	return 0;
>  }

Reviewed-by: Matthias Kaehlcke <mka@chromium.org>

Re: [PATCH] Bluetooth: btusb: request wake pin with NOAUTOEN

[Brian Norris]

On Thu, Feb 21, 2019 at 4:30 PM Brian Norris <briannorris@chromium.org> wrote:
>
> Badly-designed systems might have (for example) active-high wake pins
> that default to high (e.g., because of external pull ups) until they
> have an active firmware which starts driving it low. This can cause an
> interrupt storm in the time between request_irq() and disable_irq().
>
> We don't support shared interrupts here, so let's just pre-configure the
> interrupt to avoid auto-enabling it.
>
> Fixes: fd913ef7ce61 ("Bluetooth: btusb: Add out-of-band wakeup support")
> Signed-off-by: Brian Norris <briannorris@chromium.org>

Ping! v5.1 is enabling wake support on Scarlet, and without this
patch, that causes interrupt storms at boot. Please apply this patch
for v5.1-rc!

See also:
https://lkml.kernel.org/lkml/CA+ASDXMePkQDRfaSwNGnRYyGdsuvfUCXBtDK79o2mP=1hdNQUA@mail.gmail.com

Brian

Re: [PATCH] Bluetooth: btusb: request wake pin with NOAUTOEN

[Brian Norris]

+ akpm, in case he can help out

Another week, another buggy 5.1 RC.

On Tue, Mar 26, 2019 at 11:48 AM Brian Norris <briannorris@chromium.org> wrote:
>
> On Thu, Feb 21, 2019 at 4:30 PM Brian Norris <briannorris@chromium.org> wrote:
> >
> > Badly-designed systems might have (for example) active-high wake pins
> > that default to high (e.g., because of external pull ups) until they
> > have an active firmware which starts driving it low. This can cause an
> > interrupt storm in the time between request_irq() and disable_irq().
> >
> > We don't support shared interrupts here, so let's just pre-configure the
> > interrupt to avoid auto-enabling it.
> >
> > Fixes: fd913ef7ce61 ("Bluetooth: btusb: Add out-of-band wakeup support")
> > Signed-off-by: Brian Norris <briannorris@chromium.org>
>
> Ping! v5.1 is enabling wake support on Scarlet, and without this
> patch, that causes interrupt storms at boot. Please apply this patch
> for v5.1-rc!
>
> See also:
> https://lkml.kernel.org/lkml/CA+ASDXMePkQDRfaSwNGnRYyGdsuvfUCXBtDK79o2mP=1hdNQUA@mail.gmail.com

Ping x 2!

Marcel merged my other series first, which enabled this feature,
without applying the bugfix (which was sent first). So right now, my
RK3399/Scarlet board is in very bad shape. Can somebody please merge
this bugfix for 5.1? If not, I'll probably just send Linus a pull
request myself.

Original patch link, in case this helps someone:

http://lkml.kernel.org/lkml/20190222003051.127006-1-briannorris@chromium.org
https://lore.kernel.org/patchwork/patch/1044896/

Thanks,
Brian