Re: [Linuxptp-users] gPTP, pDelayResp and pDelayRespFollowUp correction fields

[Kat Y <aps...@gm...>]

Hello Richard, thank you very much!
My switch is adding correction field to both Path_delay_req and
path_delay_response.

Can I ask one more thing..
Since Path_delay_response_follow_up gets the correction field of
path_delay_request, I cannot have the correction on Path_delay_response too
and I should set it to 0 upon receiving the message am I right?

Regards,
Katarina

sri, 24. lip 2020. u 14:39 Richard Cochran <ric...@gm...>
napisao je:

> On Wed, Jun 24, 2020 at 10:08:14AM +0200, Kat Y wrote:
> > I can see that both pDelayResp and pDelayRespFollowUp get corrected
> though.
>
> The linuxptp stack does not "correct" these messages.
>
> For 2-step peer delay, the pDelayRespFollowUp contains a copy of the
> correction field from the original pDelayReq.  This allows the
> requestor to know the value of the correction, and this behavior
> follows the 1588 standard exactly.
>
> > I think that linuxptp is adding correction to pDelayRespFollowUp message
> > not the switch.
>
> No, not adding, but rather copying.
>
> > Is this wrong, is it expected of linuxptp? Summed up correction will be
> > doubled. Should I zero one of the correction fields or divide summed up
> > correction by 2?
>
> Sounds like your switch is changing the correction field of the
> pDelayResp.  Use tcpdump, and you will see that linuxptp sends a zero
> in the pDelayResp.correction field.
>
> > Also, can someone recommend a document (or shortly explain if it's not a
> > problem) on nrate and nrate calculations because I cannot seem to find
> > anything online?
>
> See 802.1AS-2011.
>
> Thanks,
> Richard
>