git: 4c0bc591466f - main - man9: add hz(9) and hardclock(9)

[Konstantin Belousov]

On Fri, Jun 18, 2021 at 02:43:49PM +0000, Warner Losh wrote:
> The branch main has been updated by imp:
> 
> URL: https://cgit.FreeBSD.org/src/commit/?id=4c0bc591466fd2731ba892269260b7dab74cfbad
> +.Sh DESCRIPTION
> +The
> +.Fn hardclock
> +function is called
> +.Xr hz 9
> +times per second.
This is not true, I believe.

> +It implements the real-time system clock.
> +The argument
> +.Va cnt
> +is the estimated number of ticks since the last call to
> +.Fn hardclock .
This is more accurate.

> +The argument
> +.Va usermode
> +is none-zero when
> +.Fn hardclock
> +is called from a user-mode context.
hardclock() is never called from user-mode context, it is a kernel function.
Perhaps you mean to say that hardclock() is executing in the context of
the handler that interrupted usermode.

> +Implement software
> +.Xr watchdog 9
> +processing.
> +.It
> +Enqueue
> +.Xr epoch 9
> +processing.
> +.El
> +.Sh SEE ALSO
> +.Xr adjtime 2 ,
> +.Xr ntp_adjtime 2 ,
> +.Xr signal 3 ,
> +.Xr ntpd 8 ,
> +.Xr callout 9 ,
Callout processing is not handled by hardclock().  It is done directly
by timer eventhandler.

> +.Pp
> +The main clock is used to update the system's notion of time via
> +.Xr timecounters 9
> +and to pace periodic system callbacks via
> +.Xr callout 9 ,
> +.Xr epoch 9 ,
> +and other methods documented in
> +.Xr hardclock 9 .
Again, callouts are not managed through hardclock().

> +That routine will be called approximately
> +.Va hz
> +times per second.
And this is not true.

> +.Pp
> +The second clock, running at either