mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: john stultz <johnstul@us.ibm.com>
To: Roman Zippel <zippel@linux-m68k.org>
Cc: lkml <linux-kernel@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	George Anzinger <george@mvista.com>,
	Ulrich Windl <ulrich.windl@rz.uni-regensburg.de>
Subject: Re: [RFC][PATCH 2/2] Reduced NTP rework (part 2)
Date: Thu, 29 Sep 2005 12:33:42 -0700	[thread overview]
Message-ID: <1128022423.8195.302.camel@cog.beaverton.ibm.com> (raw)
In-Reply-To: <Pine.LNX.4.61.0509291658130.3728@scrub.home>

On Thu, 2005-09-29 at 20:43 +0200, Roman Zippel wrote:
> On Tue, 27 Sep 2005, john stultz wrote:
> > The idea being:
> > 
> > update_wall_clock():
> > 	ticks = jiffies - wall_jiffies
> > 	while (ticks):
> > 		ticks--
> > 		xtime += tick_nsec + ntp_adjustment
> > 
> > 
> > isn't that different from:
> > 
> > timekeeping_periodic_hook():
> > 	now = timesource_read(ts)
> > 	delta_cycle = now - last
> > 	while (delta_cycle > interval_cycle):
> > 		delta_cycle -= interval_cycle
> > 		system_time += interval_nsec
> 
> BTW that's not what you do in the first part of the patch:
> 
> +static void ntp_advance(unsigned long interval_ns)

Well, I was trying to describe what I am going to follow the NTP patches
with. 

So yes, the reduced NTP rework patches are not discussed in the above
(but ntp_advance() does have a place in the above, I just left it out to
shorten the comparison), but they allow the two examples above to look
similar. 

For clarity here's the ntp details included

update_wall_clock():
	ticks = jiffies - wall_jiffies
	while (ticks):
		ticks--
		xtime += tick_nsec + ntp_adjustment
		ntp_advance(tick_nsec)


timekeeping_periodic_hook():
	now = timesource_read(ts)
	delta_cycle = now - last
	while (delta_cycle > interval_cycle):
		delta_cycle -= interval_cycle
		system_time += interval_nsec
		ntp_advance(interval_nsec)


> I'm quite sure that the interval_ns is wrong, it's important to advance 
> the ntp state in constant intervals (i.e. interval_cycle). Your patch 
> already includes time adjustments and e.g. the "while (interval_ns >= 
> tick_nsec)" loop is not executed anymore, once time_adjust_step becomes 
> negative.

Commenting the specific code would help clarify this. If I'm
understanding you, you're talking about the following logic:

static void ntp_advance(unsigned long interval_ns):
	static unsigned long interval_sum;

	/* increment the interval sum */
	interval_sum += interval_ns

	/* calculate the per tick singleshot adjtime adjustment step */
	while (interval_ns >= tick_nsec):
		time_adjust_step = time_adjust
		if (time_adjust_step):
			time_adjust_step = min(time_adjust_step, tickadj)
			time_adjust_step = max(time_adjust_step, -tickadj)
			time_adjust -= time_adjust_step
		interval_ns -= tick_nsec

I'm not sure I understand the problem if time_adjust_step becomes
negative.

> In general I would prefer it if we could finalize the basic design first, 
> before doing such changes, otherwise I'm afraid we need a cleanup of the 
> cleanup.

Well, that's evolution. :)  But really, the reduced ntp rework stuff
isn't a cleanup. Its just the bare minimum changes to NTP that I'm
needing for my generic timeofday code. Hopefully the reduced changes
will clarify what exactly I need (or think I need ;) from the NTP
subsystem to get my timekeeping code to function properly. Then maybe
you (or anyone else - don't let Roman have all the fun) can point to a
better way. 


> > The only difference between continuous and tick based systems would then
> > be in gettimeofday() (which really could be the same with a simple
> > #define)
> > 
> > continuous_gettimeofday():
> > 	now = timesource_read(ts)
> > 	delta_cycle = now - last
> > 	delta_nsec = cyc2ns(timesource, delta_cycle)
> > 	return system_time + delta_nsec
> > 
> > tick_gettime():
> > 	now = timesource_read(jiffes_timesource)
> > 	delta_cycle = now - last
> > 	delta_nsec = cyc2ns(timesource, delta_cycle)
> > 	delta_nsec += arch_get_offset()
> > 	return system_time + delta_nsec
> 
> The basic idea of gettimeofday is of course always the same: "base + 
> get_offset() * mult". I can understand the temptation to unify the 
> implementation, but please accept the current reality that we have 
> different gettimeofday implementations (for whatever reasons), so unifying 
> them would be a premature change. If the situation changes later we can 
> still do that unification.

I'm sorta going at it from the other way (call me optimistic :), where
I'm trying to unify what I can until I hit the exception. Then I'll
happily break out an arch specific gettimeofday implementation.

Once again, I do appreciate your feedback. Hopefully I'll have patches
out later today for you to look at.

thanks
-john




  reply	other threads:[~2005-09-29 19:34 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-09-22 19:58 [RFC][PATCH 1/2] Reduced NTP rework (part 1) john stultz
2005-09-22 19:59 ` [RFC][PATCH 2/2] Reduced NTP rework (part 2) john stultz
2005-09-27 16:37   ` Roman Zippel
2005-09-27 20:19     ` john stultz
2005-09-27 21:57       ` Thomas Gleixner
2005-09-29 18:43       ` Roman Zippel
2005-09-29 19:33         ` john stultz [this message]
2005-10-10 12:39           ` Roman Zippel
2005-10-10 20:46             ` john stultz
2005-10-04  7:58     ` Ulrich Windl
2005-10-04 18:30       ` john stultz

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1128022423.8195.302.camel@cog.beaverton.ibm.com \
    --to=johnstul@us.ibm.com \
    --cc=george@mvista.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=tglx@linutronix.de \
    --cc=ulrich.windl@rz.uni-regensburg.de \
    --cc=zippel@linux-m68k.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®