On Tue, 9 Jan 2007, KIYOHARA Takashi wrote:

> > Sure.  I will write this in several days. ;-)
>
> ppp(4) was able to be done with btuart on my gumstix.

Excellent work!

(some general comments :)

I don't think it is necessary to move COMUNIT defines out of dev/ic/com.c
since you only use it for informational output (and I think the printf
can't be right anyway, its not just "tty%02d")

why BTH4DISC but btuart.c ? I think naming should be consistent

btuartvar.h should just be btuart.h and exported to userland since
btuartd.c will need it?

> My LSI vendor is ericsson.  Besides, I think that the initialization of
> CSR is also possible. However, I do not have the device of CSR.
[...]
> I am referring to hciattach(8) of bluez.  The data sheet of ericsson
> and CSR was seen.  However, the rest was not able to be found.

Hmm.. hciattach(8) does the initialisation of the device from userland,
would that not work better? Seems that it makes it easier to upgrade
device support if that is not in the kernel..

> I referred to the following URL.
>
>   http://wiki.netbsd.se/index.php/Bluetooth
>
> However, SCO audio did not work. Is this the same reason as bt3c(4)?

Most likely - I admit this is sucky and should be better. Do you know if
that controller sends NUM_COMPL_PKTS events for SCO packets?  I think the
reason it won't work is that we don't handle those, because the USB
dongles do not, and we rely instead on the USB isoc timing to keep pace.
(because the bt3c is a v1.1 device, I had no inclination to fix it in that
case but with the advent of this, maybe it should be done)

iain