Re: [PATCH 1/9] Bluetooth: Fix persistency of the HCI_MGMT flag

[Johan Hedberg <johan.hedberg@gmail.com>]

Hi Marcel,

On Thu, Jan 05, 2012, Marcel Holtmann wrote:
> > The HCI_MGMT flag needs to remain set even after closing an HCI device.
> > 
> > Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
> > ---
> >  net/bluetooth/hci_core.c |    4 ++--
> >  1 files changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
> > index 6d38d80..52edc11 100644
> > --- a/net/bluetooth/hci_core.c
> > +++ b/net/bluetooth/hci_core.c
> > @@ -670,8 +670,8 @@ static int hci_dev_do_close(struct hci_dev *hdev)
> >  	mgmt_powered(hdev, 0);
> >  	hci_dev_unlock(hdev);
> >  
> > -	/* Clear flags */
> > -	hdev->flags = 0;
> > +	/* Clear flags, except persistent ones like HCI_MGMT */
> > +	hdev->flags &= BIT(HCI_MGMT);
> >  
> >  	hci_req_unlock(hdev);
> >  
> 
> can we just move this away from hdev->flags into a hdev->dev_flags.

Yes, that should be done (and not just for HCI_MGMT but also for other
related flags) but it should imo be in a separate patch. Simply moving
to dev_flags won't fix this particular bug since there's a similar
"hdev->dev_flags = 0;" inside hci_cc_reset() in hci_event.c.

Johan