On 1/13/2020 8:27 AM, Peter Fischer wrote: > Hello fellow linuxptp users, >
Hello. > > I am fairly new to linuxptp and I am currently trying to synchronize 3 > devices, 2 custom boards (on which I do not have any information) and > one Intel-Atom based embedded computer with a network card with an > Intel-82574 GbE LAN controller. The devices are connected via an aruba > 2930M switch. > You probably want to gather further information about those devices, including what driver they operate with. You can gather some information using both ethtool -T and ethtool -i. > When I start ptp4l on all devices for the first time after booting, I > usually get very poor results (offsets > 100us) on at least one device, > which can probably be attributed to the unstable and high delays. To > give you an impression: > ptp4l[390.901]: master offset -885335 s2 freq +2931837 path > delay 1822 > ptp4l[390.909]: master offset -904739 s2 freq +2834634 path > delay 1822 > ptp4l[390.917]: master offset -924167 s2 freq +2736521 path > delay 1822 > ptp4l[390.948]: master offset -1006909 s2 freq +2444842 path > delay 1822 > ptp4l[390.952]: port 1: minimum delay request interval 2^0 > ptp4l[390.952]: port 1: minimum delay request interval 2^-3 > ptp4l[390.956]: master offset -1023747 s2 freq +2350212 path > delay 3050 > ptp4l[390.964]: master offset -889030 s2 freq +2716197 path > delay 3050 > ptp4l[390.971]: master offset -743951 s2 freq +3119528 path > delay 3050 > ptp4l[390.987]: master offset -449849 s2 freq +3982740 path > delay 3050 > ptp4l[391.003]: master offset -499055 s2 freq +3813577 path > delay 3050 > ptp4l[391.011]: master offset -525148 s2 freq +3712652 path > delay 3050 > ptp4l[391.019]: master offset -551199 s2 freq +3610730 path > delay 3050 > ptp4l[391.026]: master offset -576386 s2 freq +3510316 path > delay 3050 > ptp4l[391.050]: master offset -655698 s2 freq +3244059 path > delay 3050 > ptp4l[391.074]: master offset -728610 s2 freq +2993866 path > delay 3050 > ptp4l[391.081]: master offset -748295 s2 freq +2902559 path > delay 3050 > ptp4l[391.089]: master offset -767749 s2 freq +2811106 path > delay 3050 > ptp4l[391.097]: master offset -786740 s2 freq +2720225 path > delay 3050 > ptp4l[391.152]: master offset 279573 s2 freq +5932237 path > delay 3050 > ptp4l[391.160]: master offset 249062 s2 freq +5851403 path > delay 3050 > ptp4l[391.168]: master offset 345180 s2 freq +6154719 path > delay 3050 > ptp4l[391.184]: master offset 591803 s2 freq +6920319 path > delay 3050 > ptp4l[391.187]: port 1: minimum delay request interval 2^0 > ptp4l[391.187]: port 1: minimum delay request interval 2^-3 > ptp4l[391.191]: master offset 708371 s2 freq +7300649 path > delay -1470 > ptp4l[391.199]: master offset 817418 s2 freq +7663107 path > delay -1470 > ptp4l[391.207]: master offset 922357 s2 freq +8017757 path > delay -1470 > ptp4l[391.215]: master offset 1026266 s2 freq +8373792 path > delay -1470 > It stays like this and does not converge. This would make me think that there is a driver bug. > > The weird thing is, when kill ptp4l on that device and restart it again, > I get perfect values: > ptp4l[1988.969]: master offset -1 s2 freq -352 path > delay 240 > ptp4l[1988.977]: master offset -2 s2 freq -355 path > delay 240 > ptp4l[1989.001]: master offset -2 s2 freq -355 path > delay 240 > ptp4l[1989.166]: master offset -7 s2 freq -370 path > delay 240 > ptp4l[1989.174]: master offset -8 s2 freq -374 path > delay 240 > ptp4l[1989.276]: master offset 4 s2 freq -338 path > delay 240 > ptp4l[1989.284]: master offset 3 s2 freq -340 path > delay 240 > ptp4l[1989.291]: master offset 6 s2 freq -331 path > delay 240 > And this *really* makes me suspect a driver bug. The fact that it fails to work at the start, and restarting things after it's already performed some steps does work.. I assume these outputs are for the custom devices, and not the Intel networking card? > Furthermore, I get a lot of these messages (don't know if related): > ptp4l[422.686]: port 1: received SYNC without timestamp > ptp4l[426.874]: port 1: received DELAY_REQ without timestamp > > Does anybody have an Idea what might cause this behavior and how to get > good values from the first run? > These messages indicate that the device is not timestamping some of the received packets. > > Best Regards, > > Peter Fischer _______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
