Re: BCM2048 bluetooth connected over OMAP serial

["sre@debian.org" <sre@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2921 bytes --]

Hi,

On Wed, Dec 10, 2014 at 09:56:22PM +0100, Pavel Machek wrote:
> On Wed 2014-12-10 18:42:03, Mark Rutland wrote:
> > On Wed, Dec 10, 2014 at 05:02:42PM +0000, Arnd Bergmann wrote:
> > > On Wednesday 10 December 2014 17:43:33 Pavel Machek wrote:
> > > > 
> > > > So, there's bluetooth chip that's connected to the SoC by UART and some
> > > > GPIOs. What would be right representation in the device tree?
> > > > Something like this?
> > > > 
> > > >         bluetooth {
> > > >                   compatible = "broadcom,bcm2048";
> > > >                   uart = <&uart2>;
> > > >                   reset-gpios = <&gpio3 27 GPIO_ACTIVE_HIGH>; /* want 91 */
> > > >                   host-wakeup-gpios = <&gpio4 5 GPIO_ACTIVE_HIGH>; /* want 101 */
> > > >                   bluetooth-wakeup-gpios = <&gpio2 5 GPIO_ACTIVE_HIGH>; /* want 37 */
> > > >                   chip-type = >;
> > > >                   bt-sysclk = <2>;
> > > >                   reset-gpio-shared = <0>;
> > > >         };
> > > > 
> > > > Is there some way to prevent OMAP tty driver from binding to the
> > > > device and exporting the device to userspace?
> > > 
> > > I think from the driver perspective, you want this to be a tty line
> > > discipline rather than a driver that attaches to the physical
> > > uart.
> > > 
> > > For the DT representation, I fear we haven't got a precedent. A uart
> > > phandle sounds reasonable, but there might be other ways to do it
> > > and we should consider if there are better alternatives. It could
> > > possibly be a child node of the uart, but that would require other
> > > infrastructure in the kernel because we don't currently create
> > > devices for those.
> > 
> > I think the child node is the way to go; that would match what we do for
> > I2C and SPI. We might need new infrastructure, but I don't think we
> > should treat this differently simlpy because we don't have that yet.
> 
> Well, uart in this case looks more like a GPIO than an I2C (no
> addressing, just few wires). And we do phandle for GPIOs.

Right and the devices use I2C for full communication and GPIOs as
helpers. I guess UART counts as full communication and not as helper.

phandle vs child node is not a matter of adressing and btw where is
the difference between "5th gpio on 1st gpio controller" and "5th
address on 1st i2c controller"?

> Actually, the chip also has PCM, analog audio, and "pc compatible?"
> connections, plus some connection to WIFI. So we may need more
> phandles there....

This is much harder to solve. I think we don't have a DT binding for
a device, which uses two communication interfaces as the bcm2048
(uart & i2c). OTOH we may just add a slave device for the fm-radio
part like this:

uart {
    bcm2048 {
        stuff;
    };
};

i2c {
    bcm2048-radio {
        master = <&bcm2048>;
    };
};

-- Sebastian

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