On Sun, 26 Feb 2017, Gary E. Miller wrote: > On Sun, 26 Feb 2017 12:42:37 -0800 (PST) > Fred Wright <f...@fwright.net> wrote: [...] > > From drivers/pps/Kconfig: > > > > --------------------------------------------------------------------- > > config NTP_PPS [...] > Oh, that interface. NTP does not use that interface at all. If you are > going to connect your PPS directly to the kernel then you do not use ntpd > at all.
It can be enabled via some flags bits in ntp.conf (IIRC only in the relevant refclock driver(s)), which may or may not have survived the ntp.org->ntpsec transformation. It's not a full replacement for ntpd, since it doesn't provide coarse-time synchronization. The NTP documentation claims that ntpd's own discipline works better, though I'm skeptical that that's true in *all* cases. > Image if the kernel and ntpd where both trying to steer the clock with > slightly different input data and control loops... But supposedly ntpd knows when it's using that feature. Fred Wright _______________________________________________ devel mailing list devel@ntpsec.org http://lists.ntpsec.org/mailman/listinfo/devel