Dear ptp4l developers, this message is FYI, a "problem report from the wild"...
I have the privilege to play around with your software, on a Skylake-based platform containing an Intel i219LM (alongside i210 which seems to give comparably mediocre results). I also have the privilege to test your software against a latest-generation Meinberg GrandMaster (HPS100). I first tried your software on the stock kernel in Debian 8 (3.16.something if memory serves) - and the "timed out while ..." errors prompted me to upgrade to the latest vanilla, which happens to be 4.13.12 at the time of this writing. The errors are now less frequent, but they do occur. I've noticed this debate and patch: https://sourceforge.net/p/linuxptp/mailman/message/35810919/ https://patchwork.ozlabs.org/patch/758160/ I can see in the source code of 4.13.12 that the patch has been applied = it's present in the vanilla. Attached you'll find an example of my error from the ptp4l output. Probably not much new in there. And, a GNUplot graph of the offset+delay (gnawed from ptp4l output). The needles on "offset" (the green trace) are a result of ptp4l resetting the port upon each occurrence of the bug, during which time it stops disciplining the clock for a while... Notice how the errors become more frequent after 9 a.m. - I came to the machine and started a PCAP sniffer on that same port. Yesterday I have uninstalled the NetworkManager (which had sneaked in as a "recommended" dependency of the X desktop) so the network ports are no longer messed around with by dhclient. I have static addresses in /etc/network/interfaces. And the machine ran overnight without any PCAP activity, just the static IP address on eth0 and ptp4l querying the GM every second. The one thing that has stunned me in a nice way: I first ran ptp4l against the GM through two managed switches, not very loaded, but unaware of PTP. That resulted in the "offset" jumping within +/- 700 "units". After I got the idea to plug the client straight into the GM, the "offset" now wanders within +/- 10 "units". That's an improvement of two decimal orders :-D Wonderful :-) What are those units BTW? PPB? Nanoseconds? Or, ticks of the PHC oscillator ? Any comment welcome... Makes me think of ASPM, I haven't checked that yet in the BIOS, and it's possibly not tweakable in this "embedded/desktop" machine, but I guess the Intel NIC drivers now disable ASPM by default in the NIC, so I don't expect to find any low-hanging fruit in there... My attempts at recording the raw PTP traffic are obviously motivated by a desire to be able to browse/analyze the logs afterwards. As for the disruptive effect of the capturing activity, the particular application using libpcap doesn't make a difference (tcpdump, wireshark or ptpTrackHound - all the same). The Debian on the box is using systemd, if that should make any difference... (I don't think so). Thanks for the software gem that you have created, and for including support for Intel hardware. Frank Rysanek
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.
---- File information -----------
File: ptp4l_intel_bug.txt
Date: 13 Nov 2017, 13:22
Size: 3323 bytes.
Type: Text
ptp4l_intel_bug.txt
Description: Binary data
The following section of this message contains a file attachment
prepared for transmission using the Internet MIME message format.
If you are using Pegasus Mail, or any other MIME-compliant system,
you should be able to save it or view it from within your mailer.
If you cannot, please ask your system administrator for assistance.
---- File information -----------
File: ptp.png
Date: 13 Nov 2017, 13:24
Size: 39555 bytes.
Type: Unknown
ptp.png
Description: Binary data
------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
