Re: [Linuxptp-users] Synchronization fault with stmmac

[Keller, J. E <jac...@in...>]

I would rather have this check in the ptp core, because it wouldn't rely
on drivers doing it, and we'd only have to code it in one place.

On Wed, 2014-03-26 at 08:57 +0100, Mohamed Belaouad wrote:
> I agree with you. Either that or the drivers could return an error in such a case.
> 
> Best Regards,
> Mohamed
> 
> ----- Mail original -----
> > On Tue, 2014-03-25 at 15:29 +0100, Richard Cochran wrote:
> > > On Tue, Mar 25, 2014 at 02:38:29PM +0100, Mohamed Belaouad wrote:
> > > 
> > > > Considering that the drivers sets the max adj to 62,500,000, I
> > > > should be able to set the ppb ajustment up/down to -+62,500,000 from
> > > > my understanding.
> > > 
> > > Yes, that is correct.
> > > 
> > > > However, when printing the value passed to the driver with testptp I get:
> > > >    Value passed to testptp | Value printed at the entry of the driver
> > > >    function
> > > >        > +32,768,000       |   -32,768,000
> > > >        < -32,768,000       |   -32,768,000
> > > 
> > > I guess you are on a 32 bit platform. In that case, the size of the
> > > adjustment is limited by the timex.freq field. See linuxptp/phc.c.
> > > 
> > > > Please excuse me if it is a misunderstanding on my part.
> > > 
> > > I thought you meant that the kernel should check that the dialed
> > > frequency offset is within the bounds of [-max_adj, +max_adj].
> > > That is a different issue.
> > > 
> > 
> > I think that change to the kernel would be a good thing, since most
> > drivers assume that the value passed to them is within range.
> > 
> > Regards,
> > Jake
> > 
> > > Thanks,
> > > Richard
> > > 
> > 
> > 
> >