help for sja1105 gPTP test problem !!!!

[jinmei]

l have two board(each board have one sja1105 on it), and connect the two board with one network cable. two board is point to point  conect to each other. and  two boards can ping each other ok.

then l run  "ptp4l -i eth0 -f automotive-master.cfg --step_threshold=1 -m &" on the master board,

and run  "ptp4l -i eth0 -f automotive-slave.cfg --step_threshold=1 -m -H &"  on the slave board,

l get  "received SYNC without timestamp"  error on the slave board, how to solve this problem? anyone help me?   please check attached file for detail

[davidtosenovjan]

I am forwarding expert response:

"From the information given in this ticket I can't really tell where the problem is coming from. Can you please request more information on the used setup and how the hardware is configured? Maybe also get a tcpdump of all relevant traffic."

[jinmei]

hello  davidtosenovjan

    detail info is as below:

    l have two circuit board,  one is used for PTP master , one is used for PTP slave soc,  the two board connected each other directly with network cable . the two board can ping each other succefully,   board master have no switch(sja1105q) on it,  but board slave have one switch(sja1105q) on it, the switch have 5 ports ,  port0 is connect to slave soc, so the switch(sja1105) is between master and slave soc.

ptp.pngptp.png

when l run ptp4l  and phc2sys on both of the two board,  and attached file is a tcpdump of all relevant traffic. you can use wireshark to watch the attached file content.

l can see Sync and Sync FollowUp message one the content, but can not found  (PDelay Req) Peer_Delay_Req_Message.  do you know why no PDelay Req? 

l found the slave soc can get Sync and  Sync FollowUp message from master board.   so Sync and  Sync FollowUp message can forward though the switch(sja1105q) to the slave soc side.

 PTP1.png

[davidtosenovjan]

There seems to be 2 issues:

1. ) They receive SYNC frames without timestamp --> So it seems the SJA1105PQRS does not add the timestamp This can be caused by a wrong configuration of the switch. The switch needs to recognize the gPTP frames as gPTP frames in order to update the timestamp. Please refer the customer to the AH1704 Application Hints of SJA1105PQRS. Chapter "7.3.1 Frame Trapping" describes how they can configure the switch so that it does recognize (trapping) the gPTP frames. Furthermore please tell them to look at chapter 7.3.3 as well. This describes how to use the IGNORE2STF flag in the switch configuration for either 1-step or 2-step SYNC frames. The customers TCP dump shows that they use 2-step SYNC frames. Chapter 7.6.6 shows PTP configuration examples.

2. ) They do not see Pdelay and Pdelay_req messages This is most likely not an issue with the SJA1105PQRS device but with the gPTP stack running on one of the uC. The switch does not generate gPTP frames.