Re: kern/53016: Clock not stable

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

The following reply was made to PR kern/53016; it has been noted by GNATS.

From: Mouse <mouse%Rodents-Montreal.ORG@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/53016: Clock not stable
Date: Wed, 16 May 2018 21:25:01 -0400 (EDT)

 Quite possibly not relevant, but I have a machine that, under NetBSD as
 old as 5.2, keeps _really_ bad time if I leave it set on the default
 clock source.  (Really bad, as in, the difference between time
 intervals measured by the machine and time intervals measured by
 wallclock time is a factor of at least 2 and I think it was more like
 10.)  I have to change kern.timecounter.hardware to get decent clock
 behaviour (I set it to hpet0 each boot, and the clock then keeps
 perfectly acceptable time then; ntp.drift contains "3.605").
 
 I forget what the default timecounter is, and now is not a good time to
 reboot it to see, but I suspect it's TSC, based on searching dmesg.boot
 for timecounter:
 
 timecounter: Timecounters tick every 10.000 msec
 timecounter: Timecounter "i8254" frequency 1193182 Hz quality 100
 timecounter: Timecounter "ACPI-Fast" frequency 3579545 Hz quality 1000
 timecounter: Timecounter "hpet0" frequency 23999999 Hz quality 2000
 timecounter: Timecounter "clockinterrupt" frequency 100 Hz quality 0
 timecounter: Timecounter "TSC" frequency 11664676160 Hz quality 3000
 
 This is, of course, related to the problem described only in that they
 both involve bad timekeeping, and the behaviour changes with choice of
 timecounter.  But it does make me slightly suspicious that some aspect
 of it may be longstanding.
 
 /~\ The ASCII				  Mouse
 \ / Ribbon Campaign
  X  Against HTML		mouse%rodents-montreal.org@localhost
 / \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B