With two PTP synchronized boards, any body has experience to measure the packet E2E delay by timestamp the packet respectively at sender and receiver boards. I.e. record the sending time point and the receiving time point, then calculate the interval in between.
Maybe the SO_TIMESTAMPING socket option would help?
linuxptp, as well as any other PTP protocol implementation, measures the path delays in order to synchronize the master and slave clocks. Run ptp4l on your board to see the measured path delay which is output by ptp4l by default. See NXP LSDK online documentation on how to run ptp4l on LS1028A-RDB. More details can be found on linuxptp Website.
Thanks for your reply.
Maybe I am not clear on my question. With the perquisite of time synchronized ls1028ardb by linuxptp, I hope to evaluate how long an application packet takes to travel from one end to the another. I.e. on the sending side, the sending ls1028ardb will embed a sending timestamp into packet before sending out and on the receiving side, another ls1028ardb will record the receiving timestamp to compare with the sending timestamp carried in packet. Has anybody done this job or like on the ls1028ardb? It will really be helpful to us.
There are no ready suggestions on measuring a network path delay with non-PTP
packets. All that I can say is, that LS1028 can timestamp any L2 frame. With
that in mind, you can use calculations similar to those done by PTP to establish
the path delay.