Re: [PATCH v2 3/5] ASoC: Intel: Add Geminilake Dialog Maxim machine driver support

[Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>]

On 4/11/18 11:34 AM, Mark Brown wrote:
> On Wed, Apr 11, 2018 at 11:20:10AM -0500, Pierre-Louis Bossart wrote:
> 
>>> +static char glk_spk_dai_name[10];
>>> +static char glk_hs_dai_name[10];
> 
>> Off-by-one? "SSPx-Codec" would be 11 chars if you include null termination.
> 
> Or avoid needing global statics entirely...
> 
>>>    static int broxton_audio_probe(struct platform_device *pdev)
>>>    {
>>>    	struct bxt_card_private *ctx;
>>> +	int dai_index = 8;
> 
>> maybe better to do an explicit search that hard-code values?
>> This will actually not work if you have an additional FE for headset on GLK,
>> or you are assuming a dependency on a patch added later in the series - not
>> good.
> 
> Definitely, this is just a recipe for fragility.

Which reminds me btw that we talked about adding a helper to dynamically 
change fields in a dailink (codec/cpu/dai names) instead of doing this 
in multiple machine drivers, e.g.

snd_soc_fixup_dailink(bxt_dailink, device_name, CPU_DAI_NAME, "Pin1");
snd_soc_fixup_dailink(bxt_dailink, device_name, DAI_NAME, "SSP5-Codec");
snd_soc_fixup_dailink(bxt_dailink, device_name, CODEC_NAME, hid-name);