Bug 202362 - ntp: restore refclocks selection (10.2-RELEASE regression)
Summary: ntp: restore refclocks selection (10.2-RELEASE regression)
Status: New
Alias: None
Product: Base System
Classification: Unclassified
Component: bin (show other bugs)
Version: 10.2-RELEASE
Hardware: Any Any
: --- Affects Some People
Assignee: freebsd-bugs mailing list
URL: https://lists.freebsd.org/pipermail/f...
Keywords: needs-qa, patch, regression
Depends on:
Blocks:
 
Reported: 2015-08-16 15:40 UTC by Christian Weisgerber
Modified: 2015-08-17 13:37 UTC (History)
4 users (show)

See Also:
koobs: mfc-stable10?


Attachments
restore refclocks selection (4.60 KB, patch)
2015-08-16 15:40 UTC, Christian Weisgerber
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Christian Weisgerber freebsd_committer 2015-08-16 15:40:25 UTC
Created attachment 159922 [details]
restore refclocks selection

The panicked update to ntp 4.2.8p1 in r280849 did not follow the configure instructions in FREEBSD-upgrade. In particular, the selection of clock drivers was not maintained and instead overwritten with the upstream defaults, enabling numerous obscure drivers but disabling the comparatively popular RAWDCF refclock.

To make matters more complicated, the previous clock selection in ntp/config.h was already partially out of sync with FREEBSD-upgrade.

The attached patch restores the clock driver selection to that of r268351 and updates the corresponding instructions in FREEBSD-upgrade.

A further discrepancy is --disable-debugging vs. #define DEBUG 1. The patch does not address this.

This affects 11.0-CURRENT, 10.2-STABLE, and 10.2-RELEASE. I noticed the problem when updating a box to 10.2R and ntpd only threw the cryptic error message "PARSE receiver #0: parse_start: parse_setfmt() FAILED." instead of attaching to my DCF77 time signal receiver.