Re: [Bluez-devel] Re: Problems with anycom usb dongle

[Marcel Holtmann <marcel@holtmann.org>]

Hi Steven,

> >> I think you mean 'need not' not 'must not'.
> > 
> > oh yes, right. Germans (especially I) tend to translate this wrong :)
> 
> I can see why this could be a problem. If "must" means "is required to
> be" then it has two opposites "is required not to be" and "is not required
> to be". That is, does the negation bind to "required" or to "to be".
> 
> You can't easily get to "is not required to be" using must. You end up
> writing something horribly long like "it is not the case that A must be
> B".
> 
> You could have said "doesn't need to be", but "need not be" is
> idiomatic.

I know, I know. It is stupid, but happens sometimes. I am not a native
speaker and from time to time I translate a little bit to literally.

> > And there can't be a SCO and an ACL link with the same handle, right?
> 
> Correct. I don't know if the HCI spec explicitly states this, but it is
> implicit. For example, the Disconnect command takes a connection handle
> as an argument. If SCO and ACL links were able to have the same handle
> then it would be ambiguous which was to be disconnected.

Thanks for clarification.

> > However the problems is that we see a SCO data packet with HCI handle 0
> > and zero data len, which should not be send up to the HCI. Basically I
> > think we must drop this directly in hci_usb driver and not even send it
> > up to BlueZ HCI core.
> 
> It would be valid for the dongle to create a link with HCI SCO handle 0,
> but, if I read the traces correctly, it hasn't done this. So the problem
> is that it's sending SCO packets for a non-existent handle.

Yes and the BlueZ HCI core complains about it.

> It may just be a feature of this dongle that if the isochronous endpoint
> is started up and there's no SCO data to send (or if it hasn't quite
> finished setting up the SCO link internally) then it sends a dummy HCI
> SCO header.
> 
> If the zero length packets are ignored then does the rest of the data
> make sense. If so, silently dropping zero length packets might be
> useful.

Is there anything written about the minimum size of a SCO data packet,
because dropping every zero length SCO packet is quite simply. And it is
also simple to do this only for a specific dongle inside the hci_usb
without touching the BlueZ HCI core.

> >                       Actually I tend to simply mark the SCO support of
> > this dongle broken.
> 
> I think you mean "intend" not "tend". :-)

This becomes an English class ;) Actually one translation of "tend" is
"tendieren" and this is what I meant. However "intend" is also fine.

Regards

Marcel




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. 
http://productguide.itmanagersjournal.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel