[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <Pine.LNX.4.64.0703102132240.10330@woody.linux-foundation.org>
Date: Sat, 10 Mar 2007 21:35:54 -0800 (PST)
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Davide Libenzi <davidel@...ilserver.org>
cc: Nicholas Miell <nmiell@...cast.net>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [patch 6/9] signalfd/timerfd v1 - timerfd core ...
On Sat, 10 Mar 2007, Davide Libenzi wrote:
> On Sat, 10 Mar 2007, Linus Torvalds wrote:
>
> > (That said, using "struct itimerspec" might be a good idea. That would
> > also obviate the need for TFD_TIMER_SEQ, since an itimerspec automatically
> > has both "base" and "incremental" parts).
>
> But TFD_TIMER_SEQ is a simple auto-rearm case of TFD_TIMER_REL. So the
> timespec is sufficent too (in all three cases we just need *one* time).
Well, people actually do use itimers like "give me a timer every second,
starting five seconds from now".
> Actually, the only place where I can find the itimerspec usefull, is
> indeed with TFD_TIMER_SEQ. In cases where you want you clock starting at a
> given time (it_value) *and* with the given frequency (it_interval).
.. and this is where itimerspec is even better: once you have absolute
time, *and* a process that might miss ticks (because it does something
else), the "absolute time start + interval" thing can avoid drifting
(which a "relative interval" has a really hard time doing).
So if you want a "timer tick every second, *on* the second" kind of
interface, you really do want a absolute time starting point, and then a
fixed interval. Two different times.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists