Re: [Patch V7 02/10] ASoC: qcom: Document LPASS CPU bindings

[Kumar Gala <galak@codeaurora.org>]

On Mar 6, 2015, at 4:03 PM, Kenneth Westfield <kwestfie@codeaurora.org> wrote:

> On Fri, Mar 06, 2015 at 10:07:01AM -0600, Kumar Gala wrote:
>> On Mar 5, 2015, at 7:51 PM, Kenneth Westfield <kwestfie@codeaurora.org> wrote:
>>> On Thu, Mar 05, 2015 at 12:52:30PM -0600, Kumar Gala wrote:
>>>> On Mar 3, 2015, at 6:21 PM, Kenneth Westfield <kwestfie@codeaurora.org> wrote:
>>>> 
>>>>> +++ b/Documentation/devicetree/bindings/sound/qcom,lpass-cpu.txt
>>>>> @@ -0,0 +1,49 @@
>>>>> +* Qualcomm Technologies LPASS CPU DAI
>>>>> +
>>>>> +Required subnodes:
>>>>> +
>>>>> +- qcom,adsp		: Audio DSP sub-node
>>>>> +
>>>>> +Optional Audio DSP subnode properties:
>>>>> +
>>>>> +- status		: "disabled" indicates the adsp is not available.
>>>>> +
>>>> 
>>>> What is the intent of this subnode?
>>>> 
>>> 
>>> From the cover letter:
>>> Even though the ipq806x LPASS does not contain an audio DSP, other SOCs
>>> do have one.  For those SOCs, the audio DSP typically controls the
>>> hardware blocks in the LPASS.  Hence, different CPU DAI driver(s) would
>>> need to be used in order to facilitate audio with the DSP.  As such, the
>>> LPASS DT contains an adsp subnode, which is disabled for this SOC.  The
>>> same subnode should be enabled and populated for other SOCs that do
>>> contain an audio DSP.  Not using the audio DSP would require different
>>> CPU DAI driver(s), in addition to possible bootloader and/or firmware
>>> changes.
>>> 
>>> This was the result of a request from Mark.  See here:
>>> http://thread.gmane.org/gmane.linux.drivers.devicetree/109331/focus=11633
>> 
>> Two quick comments before I read Mark?s comments.
>> 
>> 1. Its not normal practice to put something into a DT that does not exist.  Having a node, but marking it disabled implies existence.
> 
> Will change the DT definition to optional.
> 
>> 2. How would one normally address the audio DSP if it did exist.  I?m just wondering if having a subnode is the proper solution vs maybe a phandle
> 
> The audio DSP is, in fact, contained within the audio subsystem.  The
> representation of that relationship in the DT, I believe, would be a subnode.
> OTOH, if there is a strong sentiment towards using a phandle, that would be
> fine with me.

Just depends on how we communicate with the DSP.  If its mostly via MMIO access than a sub node makes sense.  If its via some other RPC/communication mechanism than possibly a phandle.  Trying to understand a bit more to than see what I’d recommend.

- k

-- 
Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project