On Thursday 05 June 2014, Karel Zak wrote: > On Thu, Jun 05, 2014 at 12:32:24AM +0200, Ruediger Meier wrote: > > But before doing this I have two questions. > > Maybe we can remove the test at all (or use --force to enable the > test). It was always very problematic test I will try to make it a bit more robust and then see if we still should make it optional. IMO these time sync problems are some kind of science. Could be too much for being solved within one simple shell script. > and I have doubts it's > still necessary as relevant hwclock code is completely different and > more robust now. Ok, then I would do --systohc and --hctosys just one time instead of 10. BTW how could I get the offset between sys and hw clock? What does the last "xxx seconds" in --show output mean? Obviously not the whole offset: $ export LANG=C; export TZ=UTC $ date Thu Jun 5 12:38:21 UTC 2014 $ hwclock --show --utc Thu Jun 5 13:07:28 2014 -0.062807 seconds > > 1. I wonder what is the original purpose of this test. Are we > > really testing hwclock or the kernel or hardware? I mean setting > > and reading > > The purpose is to test how precisely is able hwclock to set HW time. > Long time ago we had problem that each iteration added extra 0.5s to > the time. > > cock 10 times ... could this really discover a bug in our hwclock > > code? > > Yes. > > > 2. Why do we _set_ the time from ntp server at the beginning? > > Wouldnt it be enough to check whether the offset to ntp is the same > > before and after the test? > > Probably yes. Maybe getting sys vs. ntp offset like this: $ ntp_ip=$(dig 0.fedora.pool.ntp.org ANY +short | head -n 1) $ sntp -K /tmp/kod "$ntp_ip" 5 Jun 13:00:38 sntp[6547]: Started sntp 2014-06-05 13:00:38.263771 (-0100) -0.002546 +/- 0.022034 secs Note we have to use fixed IP to get comparable offsets. cu, Rudi -- To unsubscribe from this list: send the line "unsubscribe util-linux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
