Re: [PATCH 3/5] lockstat: core infrastructure

[Daniel Walker <dwalker@mvista.com>]

On Wed, 2007-05-30 at 19:16 +0200, Peter Zijlstra wrote:

> > >From the architecture perspective there are two low level clock hooks to
> > implement one is sched_clock() , and at least one clocksource structure.
> > Both do essentially the same thing. With timekeepings clocksource
> > structure actually being easier to implement cause the math is built in.
> 
> I think you are mistaken here; the two are similar but not identical.
> 
> I see sched_clock() as fast first, accurate second. Whereas the
> clocksource thing is accurate first, fast second.

This is true .. However, if there is a speed different it's small.
In the past I've replace sched_clock() with a clocksource, and there was
no noticeable speed different .. Just recently I replaced x86's
sched_clock() math with the clocksource math with no noticable
difference .. At least not from my benchmarks ..

> There is room for both of them.

There is room, but we don't need sched_clock() .. Certainly we shouldn't
force architectures to implement sched_clock() by calling it a "bug" if
it's lowres.

Daniel