Re: [PATCH 2/3] TTY: add slave driver to power-on device via a regulator.

[Sebastian Reichel <sre@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2362 bytes --]

Hi Marcel,

On Fri, Dec 12, 2014 at 01:46:33AM +0100, Marcel Holtmann wrote:
> > NACK. The compatible value should describe the connected device. You
> > did not connect a regulator, but a bluetooth chip! DT should look
> > like this:
> > 
> > &uart1 {
> >    bluetooth {
> >        compatible = "vendor,bluetooth-chip";
> >        vdd-supply = <&vaux4>;
> >    };
> > };
> > 
> > I think it would be ok to use your generic driver to handle the
> > specific compatible value, though.
> > 
> > Having the proper compatible value means, that there can be a more
> > specific driver later, that other operating systems can expose the
> > device as some kind of /dev/bluetooth instead of /dev/ttyXY, that
> > userspace is able to know there is a bluetooth device connected to
> > /dev/ttyXY by parsing the DT and results in easier-to-understand
> > DTS.
> 
> I think that is up to udev to be able to make this a nice device
> node. However the device node name is pretty much irrelevant. 

Ah that was not about Linux, but about $random-os using DT. Just
ignore the device name its not important for the argument.

> What you want is enough meta data to tell that there is Bluetooth
> chip behind this TTY.

Yes, that is what I tried to say :)

> From a Bluetooth stack perspective only hciattach needs to be run
> to get the N_HCI line discipline up and running for that chip.

Yes, but with the metadata from Neil's proposed binding you do not
know, that there is a bluetooth chip connected, so you need to do
this manually. Not very nice, when it could be done automatically.

> However what would be interesting for hciattach would be a
> possibility to expose the Bluetooth manufacturer. Since the init
> routine is different for each manufacturer.

Yes and it may be, that a manufacturer breaks its init routine at
some point. The problem is not much different from other devices
and normally solved through the compatible string.

> So if that could be encoded in the DT, then that would be
> certainly helpful.

All needed data should be encoded when specified as in the following
example:

&uart1 {
    bluetooth {
        compatible = "wi2wi,w2cbw003";
        vdd-supply = <&vaux4>;
    };
};

Note: wi2wi is not yet in
Documentation/devicetree/bindings/vendor-prefixes.txt

-- Sebastian

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]