[linux-sunxi] Re: [PATCH 4/5] ARM: dts: sun5i: Add dts file for the Empire Electronix D709 tablet

[Hans de Goede]

Hi,

On 22-12-15 05:29, Chen-Yu Tsai wrote:
> On Tue, Dec 22, 2015 at 3:36 AM, Hans de Goede <hdegoede at redhat.com> wrote:
>> Hi,
>>
>> On 21-12-15 17:11, Maxime Ripard wrote:
>>>
>>> Hi,x
>>>
>>> On Mon, Dec 21, 2015 at 11:46:18AM +0100, Hans de Goede wrote:
>>>>>
>>>>> On the side, any thoughts on how to handle the differences between
>>>>> various "Q8"
>>>>> tablets, like different I2C-based sensors and WiFi chips?
>>>>
>>>>
>>>> For i2c based sensors the plan is to use devicetree overlays + an in
>>>> kernel
>>>> overlay manager which probes the i2c bus (checking known touchscreen /
>>>> accelerometer
>>>> addresses) and then picks the right touchscreen + accelerometer overlays.
>>>>
>>>> Wifi is somewhat more tricky I must admit, esp. since there seem to be q8
>>>> a23 based
>>>> tablet variants with usb wifi and others with sdio wifi. Since both
>>>> busses are
>>>> discoverable I'm tempted to just enable both in devicetree, and let the
>>>> kernel probe
>>>> and see what is actually there. This assume that the way the wifichip is
>>>> powered
>>>> is the same on all boards, or at least that it is safe to enable the
>>>> necessary
>>>> regulators on all boards ...
>>>>
>>>>> I'm asking because with Maxime's couple-regulator we should be able to
>>>>> get the
>>>>> RTL8723BS on the Q8 A23/33 v1.5 working.
>>>>
>>>>
>>>> So this means enabled the sdio controller (should be safe on all boards?)
>>>> and
>>>> enabling 2 regulators to power the wifi-chip. I think it will be safe to
>>>> do this
>>>> even on boards where those regulators are not used, what do you think ?
>>>
>>>
>>> Wouldn't that introduce some useless power drain on those boards?
>>
>>
>> If nothing is attached to those regulators (which I expect to be the case
>> when
>> they are not used to power wifi) then I would expect the drain to be
>> minimal,
>> my biggest worry is some board having tied these to ground, but I don't
>> think
>> that is very likely.
>
> AFAIK the AXP datasheets mention that unused DCDC outputs should be left
> floating. Not sure if this applies to LDO outputs as well. But any used
> outputs would have a bypass capacitor.
>
>  From the board designs I've seen, this seems to be the common case.
> I'm not an electrical engineer, but I think we're covered here.
>
> Another thing we need to deal with is the different power sequencing
> requirements for the different SDIO chips.

In my experience there are not that much sequences, as simply making sure
all necessary resources are one before probing the sdio bus, I have yet to
see a case where the order in which the resources are enabled matters.

Regards,

Hans