Re: [PATCH RFC 2/3] ASoC: hdmi-codec: add support for bclk_ratio

[Jyri Sarha <jsarha@ti.com>]

On 27/02/2019 13:47, Russell King - ARM Linux admin wrote:
> On Mon, Feb 25, 2019 at 10:58:34PM +0200, Jyri Sarha wrote:
>> I think there are other options too. The obvious one would be passing
>> the blck_ratio callback trough HDMI-codec to HDMI-bridge-driver, but
>> yes, I do not like that either.
>>
>> The other would be changing the implicit bclk_ratio to 2 x sample-width,
>> and accepting blck_ratios of 36 and 40 in tda998x, and set CTS_N_M = 3
>> and CTS_N_K = 2 for them too.
> To put a bit of further information out there...
> 
> I really don't like that idea - what if we have a transmitter that
> really does use a bclk_ratio of 36 or 40?  That would mess up the
> CTS calculation in the TDA998x.
> 
> The equation I've come up to fit what we know for TDA998x is:
> 
> 	k = m * bclk_ratio / 128
> 
> where k and m are parameters that we values we select for TDA998x.
> This reflects the entire clock regeneration system including the sink.
> The possible values of k are 1, 2, 3, 4, or 8.  m are 1, 2, 4, or 8.
> I also have this equation which defines the fs regenerated at the sink:
> 
> 	fso = bclk_ratio * fsi * m / (128 * k)
> 
> If we did have a ratio of 36 or 40, the first equation above fails
> since k is not one of the possible integers.  If we round down and use
> e.g. 2 for the 36fs case, then we'd actually end up with a sample
> frequency on the sink of 1.125x faster than it should be, and the sink
> would suffer starvation of audio samples.  If we rounded up to 3, then
> the sample frequency will be 3/4 of it's true value, so the sink will
> overflow and would have to discard audio samples.
> 
> We know this happens, because that's the root of the problem at hand:
> wrongly setting the m and k values for the bclk_ratio being supplied
> to the TDA998x causes audio to be corrupted when reproduced by the
> sink.
> 
> Given that, we do need some way to validate the bclk_ratio when it is
> set, and not during hw_params which would (a) lead to ALSA devices
> failing when userspace is negotiating the parameters and (b) gives no
> way in the kernel for set_bclk_ratio() to discover whether a particular
> ratio is supported by the codec.
> 
> So, I think there's three possible ways around this:
> 1. adding a set_bclk_ratio() method to hdmi_codec_ops
> 2. calling hw_params() when our set_bclk_ratio() method is called
>    (what if the rest of the format isn't set or incompatible, which
>    may cause the hw_params() method to fail?)
> 3. adding a list of acceptable bclk_ratio values to hdmi_codec_pdata

Or leaving the bclk_ratio field zero in struct hdmi_codec_params if
set_bclk_ratio() is not called and leaving the bridge driver to decide
what to do in such a situation.

And in tda998x_audio_hw_params() doing something like this:

if (audio.bclk_ratio == 0)
	audio.bclk_ratio = DIV_ROUND_UP(params->sample_width, 8) * 8;

But then again I think it would be quite sane option to set bclk_ration
in hdmi-codec to 2*sample_width and simply refuse the ratios of 36 or 40
in tda998x.

Best regards,
Jyri


-- 
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki