On Fri, 07 Jul 2023 15:35:29 +0200, Mark Brown wrote: > > On Fri, Jul 07, 2023 at 03:30:48PM +0200, Takashi Iwai wrote: > > Mark Brown wrote: > > > > It's moderately common - typically in these cases the control is > > > described in the datasheet as an attenuation control rather than a gain, > > > and this usually corresponds to the physical implementation being only > > > able to make signals smaller relative to the reference. > > > Yeah, I see the use case. The problem is, however, that we're using > > the very same dB info for both gain and attenuation. That means, > > application has no idea how to interpret those dB values -- to be > > added or to be subtracted. > > > We should have defined a new TLV type for attenuation to > > differentiate, and define the TLV macro to give proper min/max. > > The ASoC generic control stuff supports inverting the value prior to > presentation to userspace so it's masked there (instead of writing the > number userspace sees to the register we subtract the number from the > maximum value and write that to the register), pulling that up further > to the ALSA core might be nice I guess? I believe yes. Though, I'm still not sure how we can improve the mismatch of dB min/max. The dB values of those inverted controls reflect the result of subtraction, no? Takashi
