This is what the start of the leap second looked like under SuSE Linux 9.3
(kernel 2.6.11.4):
# gettimeofday gettimeofday REALTIME
MONOTONIC PROCESS THREAD TSC
6381 2005-12-31T23:59:59Z 1136073599.974488 1136073599.974491000
1340270.017237736 1776.279262510 1776.279262780 000e3f9bb62ed949
6382 2005-12-31T23:59:59Z 1136073599.983681 1136073599.983684000
1340270.026430736 1776.288451136 1776.288451409 000e3f9bb7d29b96
6383 2005-12-31T23:59:59Z 1136073599.992797 1136073599.992800000
1340270.035547736 1776.297562953 1776.297563193 000e3f9bb972dd85
6384 2005-12-31T23:59:59Z 1136073599.002817 1136073599.002821000
1340270.045567736 1776.307580061 1776.307580717 000e3f9bbb3c66b1
6385 2005-12-31T23:59:59Z 1136073599.011952 1136073599.011955000
1340270.054701736 1776.316708835 1776.316709098 000e3f9bbcdd7fc0
6386 2005-12-31T23:59:59Z 1136073599.021069 1136073599.021072000
1340270.063818736 1776.325821195 1776.325821536 000e3f9bbe7dc731
^^^
Both gettimeofday() and clock_gettime(REALTIME, ...) jump back by
one second at the start of the leap second, i.e. they run through
23:59:59 twice, messing up causality a little bit. On the other hand,
clock_gettime(MONOTONIC, ...) and the process clocks run through
just fine.
Good night,
Markus
--
Markus Kuhn, Computer Laboratory, University of Cambridge
http://www.cl.cam.ac.uk/~mgk25/ || CB3 0FD, Great Britain
_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions
