Re: [bluetooth] linux-3.x regression (bisected)

[Gustavo Padovan <padovan@profusion.mobi>]

Hi David,

* David Herrmann <dh.herrmann@googlemail.com> [2011-12-28 17:04:11 +0100]:

> On Wed, Dec 28, 2011 at 4:52 PM, Gustavo Padovan <padovan@profusion.mobi>=
 wrote:
> ..snip..
> > I think this patch can do handling, let's see what others think.
> >
> > =A0 =A0 =A0 =A0Gustavo
> >
> >
> > ---
> > Author: Gustavo F. Padovan <padovan@profusion.mobi>
> > Date: =A0 Wed Dec 28 13:40:02 2011 -0200
> >
> > =A0 =A0Bluetooth: Fix lmp_host_le_capable() check for broken devices
> >
> > =A0 =A0Some dongles reports a wrong Local Extended Features leading the=
 kernel
> > =A0 =A0think that dongle support LE while it don't.
> >
> > =A0 =A0The fix here is just rely on a bit in Local Features (LE Capable=
) to tell
> > =A0 =A0us if the device really supports LE.
> >
> > =A0 =A0LE Host Capable is the only bit used from Local Extended Feature=
s in our
> > =A0 =A0kernel.
> >
> > =A0 =A0Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>
> >
> > diff --git a/include/net/bluetooth/hci_core.h b/include/net/bluetooth/h=
ci_core.h
> > index 5e2e984..c693111 100644
> > --- a/include/net/bluetooth/hci_core.h
> > +++ b/include/net/bluetooth/hci_core.h
> > @@ -676,7 +676,11 @@ void hci_conn_del_sysfs(struct hci_conn *conn);
> > =A0#define lmp_le_capable(dev) =A0 =A0 =A0 =A0((dev)->features[4] & LMP=
_LE)
> >
> > =A0/* ----- Extended LMP capabilities ----- */
> > -#define lmp_host_le_capable(dev) =A0 ((dev)->extfeatures[0] & LMP_HOST=
_LE)
> > +/* Some crap dongles does not report a proper Local Extended Features =
causing
> > + * the kernel to wrongly init it as a LE device. So first check if it =
is LE
> > + * capable (controller) which is a info from the Local Features */
> > +#define lmp_host_le_capable(dev) =A0 =A0( lmp_le_capable(dev) && \
> > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 (dev)->extfeatures[0] & LMP_HOST_LE)
> >
> > =A0/* ----- HCI protocols ----- */
> > =A0static inline int hci_proto_connect_ind(struct hci_dev *hdev, bdaddr=
_t *bdaddr,
>=20
> Next time we use the extfeatures for something else, we will break
> this device again. Can't we fix dev->extfeatures on initialization
> based on a blacklist/quirklist? Or is dev->extfeatures modified after
> initialzation of the device somewhere?

Sure, we will, but I don't see us using another bit of extfeatures in the n=
ear
future (after the next Core Spec release maybe) and I don't wanna got that =
far
in this fix right now.

	Gustavo