On Wed, Apr 07, 2021 at 05:25:48PM +0300, Jarkko Nikula wrote:
> On 4/2/21 1:30 PM, William Breathitt Gray wrote:
> > On Thu, Apr 01, 2021 at 06:32:28PM +0300, Jarkko Nikula wrote:
> >> +static const struct counter_device_ext intel_qep_ext[] = {
> >> + INTEL_QEP_COUNTER_EXT_RW(noise),
> >> + INTEL_QEP_COUNTER_EXT_RW(preset_enable)
> >> +};
> >
> > "noise" is a new attribute so you'll need to provide an entry in
> > Documentation/ABI/testing/sysfs-bus-counter explaining it.
> >
> Should this noise actually be visible as seconds (ns, µs, ms) instead of
> plain register value? From the spec "It selects the maximum glitch width
> to remove in terms of peripheral clock cycles: PCLK_CYCLES = MAX_COUNT + 2".
>
> I think it should since for a person configuring the system plain
> register value is more or less guesswork.
>
> Jarkko
If this clock is actually temporal in nature then denominating this
value in a temporal unit such as nanonseconds should be fine as long as
such values have a well-defined behavior for this device. I do recommend
giving this attribute a more fitting name however because it is not
immediately obvious what an attribute called "noise" controls.
William Breathitt Gray
Attachment:
signature.asc
Description: PGP signature
