On Thu, Jun 02, 2022 at 03:26:52PM +0530, Devasish Dey wrote: > > I'd suggest to call tsproc_set_delay() in handle_state_decision_event(), > > which already does that for the initial_delay option. The condition > > could be extended like this: > > > > if (!tmv_is_zero(c->initial_delay) || best_dm == DM_NO_MECHANISM) > > tsproc_set_delay(c->tsproc, c->initial_delay); > > > This is not going to work. As the delay mechanism is per port configuration > so cannot be applied at the clock level.
I don't see why it wouldn't work. Can you elaborate? There is only one clock, which is reset when the slave port is switched in the handle_state_decision_event() function. > TS_PROC is also a per port configuration. Only the clock's tsproc should be active with no DM. > > Also, I'd suggest to rename the DM constant and user setting to > > DM_NONE and "none" to make it shorter and less ugly. I don't think we > > need to follow the spec here literally. > > > We would like to follow it as per spec as it is easy for everyone to > understand the code and easily co-relate with the standards. > Since E2E/P2P was named as DM_XXX, we were in perception that you had > followed the standard while adding the enums. I think E2E and P2P are good names, but 1588-2019 broke consistency with the NO_MECHANISM. It would be nice to keep at least the user setting consistent. -- Miroslav Lichvar _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
