Re: [PATCH v2] soundwire: qcom: adjust autoenumeration timeout

From: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
To: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>,
	vkoul@kernel.org
Cc: alsa-devel@alsa-project.org, linux-kernel@vger.kernel.org,
	bard.liao@intel.com,
	Srinivasa Rao Mandadapu <quic_srivasam@quicinc.com>
Subject: Re: [PATCH v2] soundwire: qcom: adjust autoenumeration timeout
Date: Mon, 9 May 2022 15:42:33 +0100	[thread overview]
Message-ID: <b127f25f-51ea-f5d1-baef-9a0f026a2e05@linaro.org> (raw)
In-Reply-To: <0d645ee0-a54f-d9b4-9392-06ea1c37b29a@linux.intel.com>


On 09/05/2022 15:36, Pierre-Louis Bossart wrote:
> 
>>>> Even if enumeration timeout, we will not access the registers because
>>>> the ASoC codec is not registered yet from WCD938x component master.
>>>
>>> What happens when the codec is registered then? the autoenumeration
>> Codec is only registered after reset and when both TX and RX components
>> are probed.
>>
>>> still didn't complete, so what prevents the read/writes from failing
>>> then?
>> If codec is reset and registered and for some reason autoenum took more
>> than 100ms which will be hw bug then :-).
>> In this case register read/writes will fail.
> 
> Does this reset result in the 'bus reset' in the SoundWire sence and
> restart hence the autoenumeration?
The reset am referring here is codec reset gpio line. on reset device 
will show up on the bus which should trigger an autoenumeration.

> 
> It looks like you have a race between different components and starting
> the bus before it's needed, no?WCD938x is bit of a odd hw configuration, that is why we are using 
Component framework for this codec. This does ensure most part of it.

--srini