2018-04-26 02:34:38

by Richard Cochran

[permalink] [raw]
Subject: Re: [PATCH] staging: fsl-dpaa2/eth: Add support for hardware timestamping

On Wed, Apr 25, 2018 at 05:17:49PM +0800, Yangbo Lu wrote:

> +/* PTP nominal frequency 1GHz */
> +#define DPAA2_PTP_NOMINAL_FREQ_PERIOD_NS 1

Nit: Frequency is the inverse of the period. It can be one or the
other, not both.

Why not call it simply DPAA2_PTP_CLK_PERIOD_NS?

You haven't implemented the ethtool_get_ts_info() method, but you need
to do so. Show us how the user can related these MAC time stamps to
the PHC device in your other patch series.

Thanks,
Richard


2018-04-26 10:24:19

by Yangbo Lu

[permalink] [raw]
Subject: RE: [PATCH] staging: fsl-dpaa2/eth: Add support for hardware timestamping

Hi Richard,


> -----Original Message-----
> From: Richard Cochran [mailto:[email protected]]
> Sent: Thursday, April 26, 2018 10:33 AM
> To: Y.b. Lu <[email protected]>
> Cc: [email protected]; [email protected]; Greg
> Kroah-Hartman <[email protected]>; [email protected];
> Ruxandra Ioana Ciocoi Radulescu <[email protected]>
> Subject: Re: [PATCH] staging: fsl-dpaa2/eth: Add support for hardware
> timestamping
>
> On Wed, Apr 25, 2018 at 05:17:49PM +0800, Yangbo Lu wrote:
>
> > +/* PTP nominal frequency 1GHz */
> > +#define DPAA2_PTP_NOMINAL_FREQ_PERIOD_NS 1
>
> Nit: Frequency is the inverse of the period. It can be one or the other, not
> both.
>
> Why not call it simply DPAA2_PTP_CLK_PERIOD_NS?

[Y.b. Lu] Thanks. Let me use DPAA2_PTP_CLK_PERIOD_NS in next version.

>
> You haven't implemented the ethtool_get_ts_info() method, but you need to
> do so. Show us how the user can related these MAC time stamps to the PHC
> device in your other patch series.

[Y.b. Lu] I will add patch for ethtool_get_ts_info in next version. Thanks!
But I'm not sure whether ethtool should show hw timestamping always ignoring phc enabled or not.
Actually the hw timestamping is supported always since ptp clock had been initialized in MC firmware.
Let me just send out the patch for reviewing.

>
> Thanks,
> Richard