On Tue, Oct 01, 2019 at 10:00:42AM +0200, Miroslav Lichvar wrote: > On Mon, Sep 30, 2019 at 07:17:23PM -0700, Richard Cochran wrote: > > The delay between the Tx time stamp on the GM and the Rx time stamp > > must not exceed a small fraction of the sync. period, otherwise the > > servo on the slave becomes unstable. See Eidson 5.2.2. > > I think that depends on the servo and its configuration. I'm sure it's > possible to configure the PI servo to be stable with a delay of up to > an expected number of sync intervals.
Whether it is possible is not the point. The analysis in the book shows that stable PI controllers can become unstable with excessive delay. Since the T in TC stands for "transparent", the expectation is that slaves will synchronize just as well when the GM is reached through the TC. If our TC implements queuing for HW that can't keep up with the Sync rate, then someone will be surprised when their well-tuned slave does not converge suddenly. That would make the linuxptp stack user-unfriendly indeed, and for this reason I won't support it. Thanks, Richard _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
