Re: [Linuxptp-devel] [PATCH 0/5] Add minimal gPTP (802.1 AS) support

[Delio B. <dbr...@au...>]

On Jul 16, 2012, at 8:33 PM, Richard Cochran wrote:

> On Mon, Jul 16, 2012 at 10:58:14AM +0200, Delio Brignoli wrote:
>> 
>> On Jul 14, 2012, at 12:49 PM, Richard Cochran wrote:
>> 
>>> On Sat, Jul 14, 2012 at 11:11:39AM +0200, Richard Cochran wrote:
>>>> 
>>>> I studied 802.1AS-2011 a bit, and I think that the ptp4l program could
>>>> also function as a "time-aware end station" (but not as a bridge). But
>>>> I wonder whether there will be conformance issues, for example when we
>>>> have ClockMaster = ClockSlave = LocalClock.
>>> 
>>> And just to get the discussion going, here is a list of what I think
>>> needs to get done in order to support gPTP minimally.
>>> 
>>> - config option for transportSpecific field.
>>> - config option for L2 destination MAC addresses
>>> - generic TLV support (need this for 1588-2008 management anyhow)
>>> - path trace TLV
>>> - alternate BCM algorithm state machine (controlled by config file option)
>>> - transmit hooks on general messages (controlled by config file options)
>>> - use the cumulativeScaledRateOffset when slave
>>> - send a reasonable looking Follow_Up information TLV when grand master
>>> - handle signaling message with Message Interval Request TLV
>>> 
>>> (and maybe more)
>> 
>> The above looks good, we also need at least:
>> - config option to ignore twoStepFlag and force two step sync
>> 
>> and I would add:
>> - calculate/use neighborRateRatio for pdelayRateRatio calculation
>> - optionally use exp averaging filter for pdelayRateRatio as described in 11.1.2 of 802.1AS
> 
> Before you had said that, with the patches you posted, gPTP was
> already working for you. But not synchronization? Do you really need
> the neighborRateRatio and cumulativeScaledRateOffset, or can these be
> ignored (as the AS standard seems to suggest in a few places)?
> 
> Can you please clarify this for me?

It seems that neighborRateRatio and cumulativeScaledRateOffset are not
necessary for an implementation to work [*], but if implemented, they may improve
the sync.

Regards
--
Delio

[*] While trying to get gPTP to work I hacked in neighborRateRatio to see 
if would make a difference with the issue I was seeing, but in the end the bug was
elsewhere and neighborRateRatio did not make an obvious difference.