[Linuxptp-users] FW: Rogue peer delay response

[Christian L. <chr...@ch...>]

Hi

I have direct Ethernet connection Meinberg M500 - TI AM57x evalboard (ARM A15, Linux Kernel 4.9.41, PTP4L V1.8).

Normally everything is working fine.

Occasionally however I get "rogue peer delay response" from evalboard after starting ptp4l. Even sticky sometimes: it never quit loop "rogue peer delay response -  wait 16s - rogue peer delay response wait 16s...". This even after rebooting the eval boards.  It **only** happens in the start-up phase, i.e. when detecting a new master and switching through states. It never happens once it's in sync.

I was looking into the code a little bit:
"rogue peer delay response" is printed when there is no peer delay response expected because there was no peer delay request sent previously (p->peer_delay_req == NULL).

However there are some situations during resync when e.g. flush_peer_delay() is called which sets  p->peer_delay_req = NULL.
Could there be some race conditions when ptp4l sends a Pdelay.req, a state change happens,  flush_peer_delay() is called and then a Pdelay.resp arrives from M500?

Regards, Chris