On Mon, Jun 12, 2017 at 06:01:32PM -0600, Brian Olson wrote: > I just started using linuxptp version 1.8, so this may be a misunderstanding > on my part. I’m seeing slightly better performance from phc2sys using -E > linreg. The problem is that when I reboot, it seems to semi-stabilize at an > offset near -2 seconds. If I restart only phc2sys without changing anything > else, it works properly. Here are some logs from a reboot, I added some > spaces to align columns a bit better. > > if I drop the -E linreg (using PI loop) I don’t have this issue. Any > thoughts? I’m using "-a -r -N 25 -E linreg" > > I added some extra logs in clockadj.c to log ticks and freq, I can include > those if it would be helpful. > > 2017-06-12T19:00:17.551561-04:00 bco2 phc2sys: [25.183] failed to adjust the > clock: Invalid argument
This is odd. > 2017-06-12T19:00:18.551709-04:00 bco2 phc2sys: [26.183] reconfiguring after > port state change > 2017-06-12T19:00:18.551721-04:00 bco2 phc2sys: [26.183] selecting > CLOCK_REALTIME for synchronization > 2017-06-12T19:00:18.551726-04:00 bco2 phc2sys: [26.183] selecting eth2 as the > master clock > 2017-06-12T19:00:18.551790-04:00 bco2 phc2sys: [26.183] phc offset > 196930623 s0 freq +1000000000 delay 51 Hm, the initial frequency offset is 1, so the servo thinks the clock is stopped. This explains the weird 2 second offset. What kernel, arch, and libc is this? Can you please check what value has the realtime_hz variable in clockadj.c? -- Miroslav Lichvar ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
