Re: 057c:3800 BlueFRITZ! Bluetooth Stick broken since 2.6.something

[Marcel Holtmann <marcel@holtmann.org>]

Hi Johan,

> > > > On Fri, Nov 18, 2011, Michael Schmitt wrote:
> > > > > < HCI Command: Read Local Version Information (0x04|0x0001) plen 0
> > > > > > HCI Event: Command Complete (0x0e) plen 12
> > > > >     Read Local Version Information (0x04|0x0001) ncmd 1
> > > > >     status 0x00
> > > > >     HCI Version: 1.2 (0x2) HCI Revision: 0x2006
> > > > >     LMP Version: 1.2 (0x2) LMP Subversion: 0x1806
> > > > >     Manufacturer: AVM Berlin (31)
> > > > 
> > > > Ok, so this is a 1.2 adapter.
> > > > 
> > > > > < HCI Command: Read Local Supported Commands (0x04|0x0002) plen 0
> > > > > > HCI Event: Command Status (0x0f) plen 4
> > > > >     Read Local Supported Commands (0x04|0x0002) status 0x01 ncmd 1
> > > > >     Error: Unknown HCI Command
> > > > 
> > > > This is the reason why you're getting a timeout. Since the adapter
> > > > claims to support Bluetooth version 1.2 it should also support this HCI
> > > > command, so from that perspective it's not conforming to the
> > > 
> > > Apparently we check for "Bluetooth Core Specification 1.1" 
> > > 
> > > Shall we change to:
> > > 
> > > diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
> > > index f0fbb02..ed55b33 100644
> > > --- a/net/bluetooth/hci_event.c
> > > +++ b/net/bluetooth/hci_event.c
> > > @@ -562,7 +562,7 @@ static void hci_setup(struct hci_dev *hdev)
> > >  {
> > >         hci_setup_event_mask(hdev);
> > >  
> > > -       if (hdev->lmp_ver > 1)
> > > +       if (hdev->lmp_ver > 2)
> > >                 hci_send_cmd(hdev, HCI_OP_READ_LOCAL_COMMANDS, 0, NULL);
> > 
> > 0x00 is 1.0b and 0x01 is 1.1. So you are changing the check from 1.2 to
> > 2.0. The command was introduced with 1.2 actually.
> > 
> > The only real change that needs to be done here is to check hci_ver
> > instead of lmp_ver. This command is a HCI feature and not a LMP feature
> > actually.
> 
> That doesn't help much with this particular dongle though since the LMP
> and HCI versions are the same. What I'd propose is the patch which I
> already attached to an earlier email in this thread and then either a
> quirk for this adapter or just generally ignoring the
> HCI_Read_Local_Commands failure status.

using lmp_ver instead of hci_ver is still a bug, but fair enough it is a
different bug.

I am fine just allowing Read_Local_Commands to fail. However that will
serious limit some features since we have to be using the result more
often in the future. But luckily that does not matter for 1.2 controller
at all.

Regards

Marcel