Re: [PATCH] Provide an interface for getting the current tick length

[Andrew Morton <akpm@osdl.org>]

Paul Mackerras <paulus@samba.org> wrote:
>
> Andrew Morton writes:
> 
> > >  +	if ((time_adjust_step = time_adjust) != 0 ) {
> > 
> > <mutters something about coding style>
> 
> Seems perfectly plain to me, but if you don't like it I can change it.
> 

[random space before right-paren?]

	time_adjust_step = time_adjust;
	if (time_adjust_step) {

would be more conventional.  Some like the `!= 0' as well - personally I
find that a net distraction.

Ah, you copied-and-pasted from update_wall_time_one_tick().

Can we share that code?

> > 
> > >  +		/*
> > >  +		 * Limit the amount of the step to be in the range
> > >  +		 * -tickadj .. +tickadj
> > >  +		 */
> > >  +		time_adjust_step = min(time_adjust_step, (long)tickadj);
> > >  +		time_adjust_step = max(time_adjust_step, (long)-tickadj);
> > >  +	}
> > >  +	delta_nsec = tick_nsec + time_adjust_step * 1000;
> > 
> > Is that going to overflow if sizeof(long) == 4?
> 
> No.  time_adjust_step is in microseconds and is restricted to the
> range -tickadj .. tickadj, and tickadj is between 1 and 10 (assuming
> HZ >= 50).  tick_nsec is around 1e9 / HZ.  There's no way delta_nsec
> could end up less than 0 or larger than around 20 million for any
> reasonable HZ value (i.e. >= 50).
> 

ok...