Ethernet Products
Determine ramifications of Intel® Ethernet products and technologies
5387 Discussions

i219-LM PTP issues after re-connect to a i210

Propa
Beginner
208 Views

Hi all! 

I am having issues with a Precision Time Protocol (PTP; 1588) setup using a i219-LM NIC.

The setup consists of: 

* i210 NIC inside a PC that acts as grandmaster clock (GM)

* i210 NIC inside another PC set up as ordinary clock (OC) connected to GM

* i219-LM NIC inside a PC set up as boundary clock (BC) connected to GM

* a proprietary device supporting PTP as ordinary clock (OC2) connected to BC


The PCs are running debian 11 using e1000e driver with linuxptp installed (ptp4l and phc2sys running as service). 

By now I found a configuration where the time-syncing is working according to the offsets that get reported by pmc. Offsets between GM and OC are +/- 50 nsec, and offsets between GM and BC are +/- 3000nsec. It took some time to find out that this is an already known issue of the i219-LM. Suggestion for improvement: please also note that in the specs!
Offsets between BC and OC2 are also in the thousands due to large offset between GM and BC. Ok on my side as this is still within my required threshold.

I can disconnect ethernet connections between GM and BC and systems are sync'ing fine after re-connection. Also disconnecting, restarting GM system, and then re-connecting GM and BC works fine! 

But I see issues when I restart the GM system while it is connected to BC! During GM start-up something is happening on BC side which causes the BC and also OC2 to get out of sync entirely and never get into sync'ing again.
Ptp4l running on BC then reports the clock jumping backwards and in the next line running forwards as if the clocks on the NIC are stuck! I checked; I don't have multiple services running on any PC. Restarting ptp4l or Phc2sys on BC doesn't solve it. I need to restart BC PC to get it to sync again!

It works when I replace the i219-LM with a i210! Then I can restart GM PC and sync gets re-established again.

-> Is there any driver development of i219-LM still ongoing?
-> Could there be a bug causing the clock to stop ticking when re-connecting to a device?
-> Is there any energy-saving in the i219-LM that might miss some communication during startup of a conncted device?
-> Whats the difference between i219-LM and a i210 in that regard?


I can provide detailed logs if needed. Any advice would be highly appreciated!

 

Thank you!

0 Kudos
0 Replies
Reply