Re: Firmware for Bluetooth (and wifi)

[Hans de Goede <hdegoede-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>]

Hi,

On 01/28/2014 11:11 AM, Arend van Spriel wrote:
> On 01/27/2014 11:12 AM, Tomasz Figa wrote:
>>> The brcmfmac driver that consumes these DT nodes will have a closer look
>>> at the device obtaining the chipid during the probe and determine if it
>>> can support it. So the compatible string indicates that the device needs
>>> a so-called fullmac wireless driver opposed to a mac80211 aka. softmac
>>> wireless driver.
>>
>> The compatible string should guarantee that the chip ID register holds a
>> valid value, so just "wifi-fullmac" or "brcmfmac" sounds too generic to
> 
> I am not sure I understand this requirement. Is the DT node claimed
> somehow after of_find_matching_node() and unavailable to other drivers.
> 
>> me. The string must specify the family of chips with this chip ID scheme
>> in a reasonably precise way. "brcm,bcm43xx-fmac" maybe? I still see a
>> risk of, say, BCM43999 showing up, which would be a completely different
>> chip. while having the model matching the pattern.
> 
> If a completely different chip, ie. BCM43999, shows up in a board the
> device tree should not use "brcm,bcm43xx-fmac". That would be an error
> in the dts file, right? All the devices listed in your bindings patch
> are treated the same, ie. *compatible* on DT level and hence can have
> the same compatible property.
> 
> In my opinion that is what the compatible property is about. It
> identifies how a specific category of devices is accessed/configured. As
> an example please see [1]. It shows one compatible string for a binding
> that is used for different MPIC controllers.
> 
> Just to be clear, I like your suggestion to use "brcm,bcm43xx-fmac", but
> felt you did not so added my explanation/point of view.

The usual way to solve this is to have the dts file have a list of
compatibility strings going from specific to more generic, so for ie the
wifi on the cubietruck the dts file would contain:

compatible = "brcm,bcm43362", "brcm,bcm43xx-fmac";

And then the brcmfmac driver will contain .compatible = "brcm,bcm43xx-fmac"

If we then ever need to have some specific quirks in the driver the driver
can use of_device_is_compatible(dev->of_node, "brcm,bcm43362") to check for
the 43362.

Their could even be a completely separate driver for the "brcm,bcm43362",
with brcmfmac still claiming "brcm,bcm43xx-fmac", as matching is done
from left to right, so if there is a specific driver and a more generic
one the specific driver will win (assuming both are built-in / loaded
at probe time).

TL;DR: dts file should have:
  compatible = "brcm,bcm43362", "brcm,bcm43xx-fmac";
brcmfmac should have:
  .compatible = "brcm,bcm43xx-fmac",
So that we can add device specific quirks later (if necessary).

Regards,

Hans