home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!think.com!spool.mu.edu!agate!ucbvax!maths.tcd.ie!ajudge
- From: ajudge@maths.tcd.ie (Alan Judge)
- Newsgroups: comp.protocols.time.ntp
- Subject: Problems with xntp and MIPS RISC/os 5.01
- Message-ID: <9301111451.aa09054@salmon.maths.tcd.ie>
- Date: 11 Jan 93 14:51:25 GMT
- Sender: usenet@ucbvax.BERKELEY.EDU
- Distribution: inet
- Organization: The Internet
- Lines: 36
-
- [Sorry to those who have already seen this in news.]
-
- We are trying to run xntp 3 under MIPS RISC/os 5.01 on an R4000 based
- MIPS box. We already have xntp 3 running fine on RISC/os 4.51 R3000
- machines; these machines will be upgraded soon, and one of them is a
- fairly well used Stratum 2, hence the urgency of the problem.
-
- Anyway, the main problem is that settimeofday seems to work oddly
- under RISC/os 5. To quote:
- The implementation of settimeofday ignores the tv_usec
- field of tp. If the time needs to be set with better
- than one second accuracy, call settimeofday for the
- seconds and then adjtime for finer accuracy.
-
- This seems to cause xntp a good deal of indigestion and it never
- settles down. The time drifts quite a lot off (like 5-10 seconds and
- more).
-
-
- Since the standard code didn't work, I had a rummage in the code and
- noticed SLEWALWAYS. I tried defining this, and despite using several
- different tickadj values, xntp never worked properly and in some cases
- went total wild (drifting over a minute off in a few hours).
-
- I've even tried some other hacks in ntp_loopfilter.c (which doesn't
- seem to depend on SLEWALWAYS, even though it looks like it should),
- but have had no success.
-
-
- So, can anyone tell me:
- Does the SLEWALWAYS code work?
- Do I need to do anything else?
-
- Any suggestions that I could try?
- --
- Alan
-