linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* asm/timex.h
@ 2002-05-24 17:33 Vojtech Pavlik
  2002-05-24 17:49 ` asm/timex.h Christoph Hellwig
  0 siblings, 1 reply; 3+ messages in thread
From: Vojtech Pavlik @ 2002-05-24 17:33 UTC (permalink / raw)
  To: linux-kernel

Hi!

I have several questions about asm/timex.h

1) Who uses it? The kernel certainly doesn't. Perhaps NTP?

2) CLOCK_TICK_RATE is defined on i386 and similar archs
   to be 1193180. This is wrong, because the real frequency
   of the i8253 PIT timer is 14.31818 MHz / 12 = 1.11931816 MHz,
   so CLOCK_TICK_RATE should be 1193182. Should it be fixed?

3) What if an architecture doesn't have a compile-time known
   CLOCK_TICK_RATE? I suppose I cannot just #define it to a variable,
   because the kernel doesn't use it, and that probably means userland
   does ...

Anyone knows the answers?

-- 
Vojtech Pavlik
SuSE Labs

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: asm/timex.h
  2002-05-24 17:33 asm/timex.h Vojtech Pavlik
@ 2002-05-24 17:49 ` Christoph Hellwig
  2002-05-24 18:23   ` asm/timex.h Vojtech Pavlik
  0 siblings, 1 reply; 3+ messages in thread
From: Christoph Hellwig @ 2002-05-24 17:49 UTC (permalink / raw)
  To: Vojtech Pavlik; +Cc: linux-kernel

On Fri, May 24, 2002 at 07:33:45PM +0200, Vojtech Pavlik wrote:
> Hi!
> 
> I have several questions about asm/timex.h
> 
> 1) Who uses it? The kernel certainly doesn't. Perhaps NTP?

The kernel does.  Thanks to the sched.h mess it's even implictly
included in almost any file..

> 3) What if an architecture doesn't have a compile-time known
>    CLOCK_TICK_RATE? I suppose I cannot just #define it to a variable,
>    because the kernel doesn't use it, and that probably means userland
>    does ...

The kernel DOES use it.  grep(1) is your friend.


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: asm/timex.h
  2002-05-24 17:49 ` asm/timex.h Christoph Hellwig
@ 2002-05-24 18:23   ` Vojtech Pavlik
  0 siblings, 0 replies; 3+ messages in thread
From: Vojtech Pavlik @ 2002-05-24 18:23 UTC (permalink / raw)
  To: Christoph Hellwig, Vojtech Pavlik, linux-kernel

On Fri, May 24, 2002 at 06:49:03PM +0100, Christoph Hellwig wrote:
> On Fri, May 24, 2002 at 07:33:45PM +0200, Vojtech Pavlik wrote:
> > Hi!
> > 
> > I have several questions about asm/timex.h
> > 
> > 1) Who uses it? The kernel certainly doesn't. Perhaps NTP?
> 
> The kernel does.  Thanks to the sched.h mess it's even implictly
> included in almost any file..
> 
> > 3) What if an architecture doesn't have a compile-time known
> >    CLOCK_TICK_RATE? I suppose I cannot just #define it to a variable,
> >    because the kernel doesn't use it, and that probably means userland
> >    does ...
> 
> The kernel DOES use it.  grep(1) is your friend.

# grep -rw FINETUNE * | grep -v timex.h
# grep -rw CLOCK_TICK_FACTOR * | grep -v timex.h
# grep -rw CLOCK_TICK_RATE * | grep -v timex.h
drivers/char/ftape/lowlevel/ftape-calibr.c:     return (10000 * count) / ((CLOCK_TICK_RATE + 50) / 100);

The only interesting place where is CLOCK_TICK_RATE used is in linux/timex.h
to define LATCH. Which subsequently is only used in arch specific code
and ftape again.

Ftape's use of it is buggy, because it lacks the needed locking.

So, it seems like CLOCK_TICK_RATE isn't needed by non-arch-specific
code.

And FINETUNE isn't used at all.

But ... the fact that kernel uses it ... probably doesn't mean that
userland doesn't.

-- 
Vojtech Pavlik
SuSE Labs

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2002-05-24 18:23 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-05-24 17:33 asm/timex.h Vojtech Pavlik
2002-05-24 17:49 ` asm/timex.h Christoph Hellwig
2002-05-24 18:23   ` asm/timex.h Vojtech Pavlik

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).