[PATCH] Bluetooth: Notify about device registration before power on

[PATCH] Bluetooth: Notify about device registration before power on

[Marcel Holtmann]

It is important that the monitor interface gets notified about
a new device before its power on procedure has been started.

For some reason that is no longer working as expected and the power
on procedure runs first. It is safe to just notify about device
registration and trigger the power on procedure afterwards.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
---
 net/bluetooth/hci_core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index 8a0ce70..a0a2f97 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -1754,11 +1754,11 @@ int hci_register_dev(struct hci_dev *hdev)
 	if (hdev->dev_type != HCI_AMP)
 		set_bit(HCI_AUTO_OFF, &hdev->dev_flags);
 
-	schedule_work(&hdev->power_on);
-
 	hci_notify(hdev, HCI_DEV_REG);
 	hci_dev_hold(hdev);
 
+	schedule_work(&hdev->power_on);
+
 	return id;
 
 err_wqueue:
-- 
1.7.11.7

Re: [PATCH] Bluetooth: Notify about device registration before power on

[Johan Hedberg]

Hi Marcel,

On Tue, Oct 30, 2012, Marcel Holtmann wrote:
> It is important that the monitor interface gets notified about
> a new device before its power on procedure has been started.
> 
> For some reason that is no longer working as expected and the power
> on procedure runs first. It is safe to just notify about device
> registration and trigger the power on procedure afterwards.
> 
> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
> ---
>  net/bluetooth/hci_core.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Acked-by: Johan Hedberg <johan.hedberg@intel.com>

Johan

Re: [PATCH] Bluetooth: Notify about device registration before power on

[Gustavo Padovan]

Hi Marcel,

* Marcel Holtmann <marcel@holtmann.org> [2012-10-30 01:35:40 -0700]:

> It is important that the monitor interface gets notified about
> a new device before its power on procedure has been started.
> 
> For some reason that is no longer working as expected and the power
> on procedure runs first. It is safe to just notify about device
> registration and trigger the power on procedure afterwards.
> 
> Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
> ---
>  net/bluetooth/hci_core.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Patch has been applied to bluetooth-next. Thanks.

	Gustavo

Re: [PATCH] Bluetooth: Notify about device registration before power on

[Johan Hedberg]

Hi Gustavo,

On Wed, Oct 31, 2012, Gustavo Padovan wrote:
> > It is important that the monitor interface gets notified about
> > a new device before its power on procedure has been started.
> > 
> > For some reason that is no longer working as expected and the power
> > on procedure runs first. It is safe to just notify about device
> > registration and trigger the power on procedure afterwards.
> > 
> > Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
> > ---
> >  net/bluetooth/hci_core.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> Patch has been applied to bluetooth-next. Thanks.

I guess you mean bluetooth.git? (that's where the patch is right now -
not bluetooth-next). Also, I don't think the Cc: stable addition was
necessary since this scheduling behavior seems to be something only
introduced for 3.7 (not 100% sure about this though).

Johan

Re: [PATCH] Bluetooth: Notify about device registration before power on

[Marcel Holtmann]

Hi Gustavo,

> > > It is important that the monitor interface gets notified about
> > > a new device before its power on procedure has been started.
> > > 
> > > For some reason that is no longer working as expected and the power
> > > on procedure runs first. It is safe to just notify about device
> > > registration and trigger the power on procedure afterwards.
> > > 
> > > Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
> > > ---
> > >  net/bluetooth/hci_core.c | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > Patch has been applied to bluetooth-next. Thanks.
> 
> I guess you mean bluetooth.git? (that's where the patch is right now -
> not bluetooth-next). Also, I don't think the Cc: stable addition was
> necessary since this scheduling behavior seems to be something only
> introduced for 3.7 (not 100% sure about this though).

I did not send this to stable, because I think it is 3.7-rc only
behavior. Feel free to test it. With the help of btmon it is really easy
to see if the New Index event arrives before the first HCI command or
not. Remember that this only applies for newly attached devices and
btmon already running.

Regards

Marcel