A long time error is in usr.sbin/timed/measure.c where someone
deleted the htonl conversion of the time value. It does not matter
on big endians but small (little-endian) machines get it all wrong when
timedc is used to measure clock differences.. (timed itself gets problems too)
Here are the diff(s):
usr.sbin/timed/measure.c
155c155
< 			oicp->icmp_otime = ((tcur.tv_sec % SECDAY) * 1000
---
> 			oicp->icmp_otime = htonl((tcur.tv_sec % SECDAY) * 1000
regards, Gerard.
-- 
Gerard J van der Grinten           pa0gri@net.pa0gri.ampr.org [44.137.1.1]
Elzenlaan 8                        gvdg@nlr.nl         (temporary qrl)
3467 TJ Driebruggen                gvdg@fridley.pa0gri.ampr.org (home)
Netherlands       (+031)-34871606  Home. (+031)-52748435 Qrl.

------------------------------------------------------------------------------