Re: nanosleep

Re: nanosleep

[mark manning]

hrm - im confiused now - how can you do a n NANO second delay when the resolution is 10 mili seconds ?

forgive my ignorance and my persistant "stupid" questions :)

----- Original Message -----
From: "H. Peter Anvin" <hpa@zytor.com>
Date: 	9 Apr 2002 22:47:42 -0700
To: linux-kernel@vger.kernel.org
Subject: Re: nanosleep


> Followup to:  <20020410044243.2916.qmail@fastermail.com>
> By author:    "mark manning" <mark.manning@fastermail.com>
> In newsgroup: linux.dev.kernel
> >
> > thanx - how much of a difference should i expect - i know the
> > syscall is asking for at least the required ammount but that the
> > task switcher might not give me control back for a while after the
> > requested delay but i was expecting to be a little closer to what i
> > had asked for - this isnt critical of corse but i would like to know
> > what to expect.
> > 
> 
> Read the man page:
> 
> BUGS
>        The current implementation of nanosleep is  based  on  the
>        normal  kernel  timer mechanism, which has a resolution of
>        1/HZ s (i.e, 10 ms on Linux/i386 and 1 ms on Linux/Alpha).
>        Therefore, nanosleep pauses always for at least the speci
>        fied time, however it can take up to  10  ms  longer  than
>        specified  until  the  process becomes runnable again. For
>        the same reason, the value returned in case of a delivered
>        signal  in *rem is usually rounded to the next larger mul
>        tiple of 1/HZ s.
> 
>        As some applications  require  much  more  precise  pauses
>        (e.g.,  in  order to control some time-critical hardware),
>        nanosleep is also capable of short high-precision  pauses.
>        If  the process is scheduled under a real-time policy like
>        SCHED_FIFO or SCHED_RR, then pauses of up to 2 ms will  be
>        performed as busy waits with microsecond precision.
> 
> 	-hpa
> -- 
> <hpa@transmeta.com> at work, <hpa@zytor.com> in private!
> "Unix gives you enough rope to shoot yourself in the foot."
> http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>
> -
> 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/
> 

-- 

_______________________________________________
Get your free email from http://www.fastermail.com

Powered by Outblaze

Re: nanosleep

[Robert Love]

On Wed, 2002-04-10 at 01:57, mark manning wrote:
> hrm - im confiused now - how can you do a n NANO second delay when the
> resolution is 10 mili seconds ?

Uh, you can't - that was his point.

You can try, and you will certainly sleep at least that long, but any
time given modulo 10ms is out the door ...

Like I said, check out the high resolution timers project.

	Robert Love

Re: nanosleep

[andrew may]

On Wed, Apr 10, 2002 at 02:01:35AM -0400, Robert Love wrote:
> On Wed, 2002-04-10 at 01:57, mark manning wrote:
> > hrm - im confiused now - how can you do a n NANO second delay when the
> > resolution is 10 mili seconds ?
> 
> Uh, you can't - that was his point.

Well there is a cheap trick if you catch what the man page says.
 
> You can try, and you will certainly sleep at least that long, but any
> time given modulo 10ms is out the door ...

Make all your calls to nanasleep be less than 2ms, and loop through as
many as you need until you are under 2ms.

Don't do it for too long because you get no other use out of your machine
while your doing this but it does work.

Re: nanosleep

[H. Peter Anvin]

Followup to:  <20020413145306.A29006@ecam.san.rr.com>
By author:    andrew may <acmay@acmay.homeip.net>
In newsgroup: linux.dev.kernel
> 
> Make all your calls to nanasleep be less than 2ms, and loop through as
> many as you need until you are under 2ms.
> 
> Don't do it for too long because you get no other use out of your machine
> while your doing this but it does work.
> 

If you want to busy-wait, you might as well do it in userspace (use
RDTSC.)

Maybe a future version of the kernel will have more flexible
scheduling.

	-hpa

-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>

Re: nanosleep

[mark manning]

thanx - how much of a difference should i expect - i know the syscall is asking for at least the required ammount but that the task switcher might not give me control back for a while after the requested delay but i was expecting to be a little closer to what i had asked for - this isnt critical of corse but i would like to know what to expect.


----- Original Message -----
From: "H. Peter Anvin" <hpa@zytor.com>
Date: Tue, 09 Apr 2002 21:17:03 -0700
To: mark manning <mark.manning@fastermail.com>
Subject: Re: nanosleep


> mark manning wrote:
> > doh - i think something is still wrong, i ask for 1000 ms and i get a second but if i do a 500 itteration loop asking for 1 ms i get 5 seconds.  i am also starting to distrust my elapsed time display which is using the gettimeofday syscall
> > 
> 
> It doesn't work that way.  Each call to nanosleep() gives you a 
> *MINIMUM* time to delay.  The kernel may decide to schedule you away and 
> pick your process up when it suits it.
> 
> 	-hpa
> 
> 
> 

-- 

_______________________________________________
Get your free email from http://www.fastermail.com

Powered by Outblaze

Re: nanosleep

[H. Peter Anvin]

Followup to:  <20020410044243.2916.qmail@fastermail.com>
By author:    "mark manning" <mark.manning@fastermail.com>
In newsgroup: linux.dev.kernel
>
> thanx - how much of a difference should i expect - i know the
> syscall is asking for at least the required ammount but that the
> task switcher might not give me control back for a while after the
> requested delay but i was expecting to be a little closer to what i
> had asked for - this isnt critical of corse but i would like to know
> what to expect.
> 

Read the man page:

BUGS
       The current implementation of nanosleep is  based  on  the
       normal  kernel  timer mechanism, which has a resolution of
       1/HZ s (i.e, 10 ms on Linux/i386 and 1 ms on Linux/Alpha).
       Therefore, nanosleep pauses always for at least the speci­
       fied time, however it can take up to  10  ms  longer  than
       specified  until  the  process becomes runnable again. For
       the same reason, the value returned in case of a delivered
       signal  in *rem is usually rounded to the next larger mul­
       tiple of 1/HZ s.

       As some applications  require  much  more  precise  pauses
       (e.g.,  in  order to control some time-critical hardware),
       nanosleep is also capable of short high-precision  pauses.
       If  the process is scheduled under a real-time policy like
       SCHED_FIFO or SCHED_RR, then pauses of up to 2 ms will  be
       performed as busy waits with microsecond precision.

	-hpa
-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>

Re: nanosleep

[mark manning]

thanx - how much of a difference should i expect - i know the syscall is asking for at least the required ammount but that the task switcher might not give me control back for a while after the requested delay but i was expecting to be a little closer to what i had asked for - this isnt critical of corse but i would like to know what to expect.


----- Original Message -----
From: "H. Peter Anvin" <hpa@zytor.com>
Date: Tue, 09 Apr 2002 21:17:03 -0700
To: mark manning <mark.manning@fastermail.com>
Subject: Re: nanosleep


> mark manning wrote:
> > doh - i think something is still wrong, i ask for 1000 ms and i get a second but if i do a 500 itteration loop asking for 1 ms i get 5 seconds.  i am also starting to distrust my elapsed time display which is using the gettimeofday syscall
> > 
> 
> It doesn't work that way.  Each call to nanosleep() gives you a 
> *MINIMUM* time to delay.  The kernel may decide to schedule you away and 
> pick your process up when it suits it.
> 
> 	-hpa
> 
> 
> 

-- 

_______________________________________________
Get your free email from http://www.fastermail.com

Powered by Outblaze

Re: nanosleep

[Robert Love]

On Wed, 2002-04-10 at 00:41, mark manning wrote:

> thanx - how much of a difference should i expect - i know the syscall is
> asking for at least the required ammount but that the task switcher might
> not give me control back for a while after the requested delay but i was
> expecting to be a little closer to what i had asked for - this isnt critical
> of corse but i would like to know what to expect.

The minimum granularity of the timer is 1/HZ, which on a i386 is only
10ms.

If you want high-resolution timers, check out the high-res-timers
project at http://high-res-timers.sf.net/ - they implement POSIX timers
(which include a nanosleep call) with very high resolution (1/cpu
clock).

	Robert Love

Re: nanosleep

[mark manning]

[-- Attachment #1: Type: text/plain, Size: 822 bytes --]

hi - yes i read the man pages, im already using nanosleep for doing 1 second delays and i chose nanosleep because a signal can interrupt the delay which might be useful (and might not :)

----- Original Message -----
From: Mark Hahn <hahn@physics.mcmaster.ca>
Date: Tue, 9 Apr 2002 20:32:03 -0400 (EDT)
To: mark manning <mark.manning@fastermail.com>
Subject: Re: nanosleep


> > could someone show me how to do a delay of N miliseconds using nanosleep ?
> > (given n delay n mili seconds etc) - i cant seem to get my code to work
> > doh! 
> 
> did you read the fine man page, esp the section on resolution
> and busy-waits?  also, using select for its timeout is probably
> more portable.
> 
> 

-- 

_______________________________________________
Get your free email from http://www.fastermail.com

Powered by Outblaze

nanosleep

[mark manning]

could someone show me how to do a delay of N miliseconds using nanosleep ? (given n delay n mili seconds etc) - i cant seem to get my code to work doh!
-- 

_______________________________________________
Get your free email from http://www.fastermail.com

Powered by Outblaze