Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <730128a9-46ce-02b2-f88d-c9982fff2e69@opensource.cirrus.com>
Date:   Tue, 15 Nov 2022 11:41:36 +0000
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.12.0
Subject: Re: [PATCH 2/4] soundwire: Provide build stubs for common functions
Content-Language: en-US
To:     Charles Keepax <ckeepax@opensource.cirrus.com>,
        Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
CC:     <vkoul@kernel.org>, <yung-chuan.liao@linux.intel.com>,
        <sanyog.r.kale@intel.com>, <alsa-devel@alsa-project.org>,
        <linux-kernel@vger.kernel.org>, <patches@opensource.cirrus.com>
References: <20221114102956.914468-1-ckeepax@opensource.cirrus.com>
 <20221114102956.914468-3-ckeepax@opensource.cirrus.com>
 <90b01c76-fc65-a57f-9247-fae78241342c@linux.intel.com>
 <20221115110345.GM10437@ediswmail.ad.cirrus.com>
From:   Richard Fitzgerald <rf@opensource.cirrus.com>
In-Reply-To: <20221115110345.GM10437@ediswmail.ad.cirrus.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 15/11/2022 11:03, Charles Keepax wrote:
> On Mon, Nov 14, 2022 at 10:13:07AM -0600, Pierre-Louis Bossart wrote:
>>
>>
>> On 11/14/22 04:29, Charles Keepax wrote:
>>> Provide stub functions when CONFIG_SOUNDWIRE is not set for functions
>>> that are quite likely to be used from common code on devices supporting
>>> multiple control buses.
>>
>> So far this case has been covered by splitting SoundWire related code
>> away from, say I2C, and with a clear 'depends on SOUNDWIRE'. This is the
>> case for rt5682, max98373, etc.
>>
>> Is this not good enough?
>>
>> I am not against this patch, just wondering if allowing code for
>> different interfaces to be part of the same file will lead to confusions
>> with e.g. register offsets or functionality exposed with different
>> registers.
>>
> 
> I guess this is a bit of a grey area this one. Both work, I guess
> the reason I was leaning this way is that in order to avoid a
> circular dependency if I put all the soundwire DAI handling into
> the soundwire code then I have to duplicate the snd_soc_dai_driver
> structure into both the sdw and i2c specific code (worth noting
> the I2S DAIs are still usable when the part is sdw to the host). But
> there are also downsides to this approach in that it will likely have
> some small impact on driver size when soundwire is not built in.
> 

I think we should just add the stubs. Other subsystems use stubs to help
with code that references stuff that might not be available.

Splitting all the soundwire-specifics out into a separate module works
for simple chips that are either I2S or soundwire. but can get messy for
a complex codec. I used the separate file method for CS42L42, but for a
driver I'm working on I abandoned that and put both DAIs in the core
code. I didn't notice the missing stubs because my defconfig that was
intended to omit soundwire apparently has something that is selecting
it anyway.