Re: RE: [PATCH net v2 1/1] net: fec: fix the PTP periodic output sysfs interface

[Buday Csaba]

>>>> I don't understand why you think that we should be exposing 4 channels
>>>> to the user when they can only use one. And have the user guess which
>>>> one should be programmed. Please explain (and Buday will have to update
>>>> the commit message), to me v1 looked like a much better fix.
>>>
>>> As I explained in v1, I think the "channel index" parameter of
>>> "/sys/class/ptp/ptp<N>/period" refers to the hardware channel number,
>>> rather than the software mapping to the hardware channel.
>>
>> Can you cite any documentation or examples?
> 
> There is no documentation that explicitly specifies whether it must be
> the hardware channel or the software-defined channel. So this leads
> to different people having different understandings of it.
> 

What about updating the documentation first?
Like:
--- a/Documentation/ABI/testing/sysfs-ptp
+++ b/Documentation/ABI/testing/sysfs-ptp
@@ -59,7 +59,8 @@ Date:         September 2010
  Contact:       Richard Cochran <richardcochran@xxxxxxxxx>
  Description:
                 This file contains the number of programmable periodic
-               output channels offered by the PTP hardware clock.
+               output channels offered by the PTP hardware clock, and
+               supported by the driver.

  What:          /sys/class/ptp/ptp<N>/n_pins
  Date:          March 2014

Same for the period.

>> What would be the benefit of exposing the ID within the MAC IP (not
>> even the board) to the user?
> 
>  From my perspective, I expect the "channel index" to be consistent with
> "fsl,pps-channel" property, otherwise this may confuse some users.
> 
> Hi Buday,
> 
> If you want to revert to the first version, I suggest removing the check.
> Because the check in period_store() already guarantees that
> req.perout.index must be 0.
> 
> 

Okay, I will do that.