lkml.org 
[lkml]   [2004]   [Aug]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: boot time, process start time, and NOW time
From
Date
On Mon, 2004-08-16 at 23:56, Tim Schmielau wrote:
> On Tue, 16 Aug 2004, Albert Cahalan wrote:
>
> > If you're interested in reducing (not solving)
> > the problem for the 2.6.x series, you might change
> > HZ to something that works better with the PIT.
>
> No, that's not needed anymore. We've already started to account for the
> difference, e.g. by using TICK_NSEC in jiffies64_to_clock_t().

Well, unfortunately TICK_NSEC just gives the *current* tick length as
requested by ntpd. So it won't work over an long interval of jiffies
where TICK_NSEC might have changed.

> Problem is, we are only halfway through the attempt to remove the use
> of "jiffies" as a clock, so currently to incompatible time sources get
> mixed
> up.
>
> The other problem seems to be that this move away from "jiffies" seems to
> happen on an ad-hoc basis whenever we encounter a problem, rather than
> with a big picture in mind.

Indeed you are correct. Since timer interrupts are not precisely or
accurately delivered, a timer interrupt counter (jiffies), cannot be
used as a reliable time source (except where there is not other time
source). The problem is that it is difficult to discern where jiffies is
just being used as a timer subsystem counter, or where its being used as
a time of day time source.

Even worse, this is a userspace visible usage of jiffies as a time
stamp, so in this case we have to preserve the interface and find a way
to emulate it. So proc_pid_stats() may need to do the reverse of what
procps is doing.

> John Stultz once laid out a concept for a (coordinated) rewrite in 2.7,
> and I think this still is a good idea.

Yep, I've been working like crazy on just this (well, when my work isn't
swamping me). Unfortunately it is a major overhaul and causes cascading
changes (removal of xtime), so its not going as quickly as I'd like.
However I feel the design is quite good and I will attach a copy of it
and the first pass of the code in a reply to this email.

thanks
-john

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:05    [W:0.521 / U:0.624 seconds]
©2003-2014 Jasper Spaans. Advertise on this site