Re: [RFC PATCH] ASoC: pcm512x: Implement the set_bclk_ratio interface

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

On 1/15/19 6:19 AM, Dimitris Papavasiliou wrote:
> On Tue, Jan 15, 2019 at 10:08 AM Peter Rosin <peda@axentia.se> wrote:
>> On 2019-01-14 18:36, Pierre-Louis Bossart wrote:
>>>> You should perhaps check if (ratio >= 1 && ratio <= 256) prior to
>>>> accepting a divider that can't be programmed? But maybe that's enforced
>>>> somewhere else? And perhaps the sanity check should be even stricter?
>>> Yes it should be stricter with a power of two only. I tried really hard
>>> to make this codec work with ratios of 50 (and a 19.2 MHz mclk) on an Up
>>> board and it's just not possible, probably not supported by hardware.
>> Disallowing anything but powers of two just because 50 doesn't work seems
>> pretty wild. According to docs I think 48 should work just fine.
>>
>> Section 8.3.2 Audio Data Interface page 16.
>> http://www.ti.com/lit/ds/symlink/pcm5142.pdf
>>
>> Or?
> According to the datasheet[1], you can program any divider between 1
> and 256, via register 33 (page 80), but, as Peter points out, in the
> description of the audio data interface (page 15), only the values 32,
> 48, 64, 128, 256 are mentioned as permissible.  So one approach would
> be to restrict the values accepted by the callback to this set.  Since
> the datasheet is not terribly clear on whether only these are allowed,
> and since the ratio is set explicitly in the machine driver, it could
> be argued that it would be preferable to allow all values that can be
> set in the register, to minimize the risk of needlessly rejecting
> valid configurations.
>
> [1] http://www.ti.com/lit/ds/symlink/pcm5121.pdf
I meant multiple of 16, sorry. multiples of 25/50 commonly used with 
other codecs will not work.