Re: [RFC][patch 00/12] clocksource / timekeeping rework V2

[john stultz <johnstul@us.ibm.com>]

On Thu, 2009-07-30 at 06:49 -0700, Daniel Walker wrote:
> On Thu, 2009-07-30 at 15:04 +0200, Martin Schwidefsky wrote:
> > On Thu, 30 Jul 2009 05:49:33 -0700
> > Daniel Walker <dwalker@fifo99.com> wrote:
> > 
> > > On Thu, 2009-07-30 at 12:53 +0200, Martin Schwidefsky wrote:
> > > 
> > > > > I'm not sure allowing that type of override a good idea tho .. I don't
> > > > > think it's considered a usable clock when the rating goes to zero.
> > > > 
> > > > Override as the root user -> your foot, no? The whole override stuff is
> > > > there for the case that the clocksource selection picked a broken clock
> > > > and you want to force the system into a semi-working state. Ideally the
> > > > whole override would go away, but that is probably wishful thinking..
> > > 
> > > I would agree if the system doesn't crash as a result, if it just starts
> > > to operate funny then that's maybe acceptable. If you keep the change
> > > rating function, you could potentially remove the unregister path..
> > 
> > Why shouldn't it be possible to have a clocksource as a module? I think
> > that the unregister path should stay. To really make it work we'd need
> > a function to force the system out of the one-shot mode though.
> 
> Because I don't think there is a sane reason to allow it. It should be
> more like if someone has a need for it, then let them add back the
> unregister path and explain why they need it.

Clocksources as modules was one of the initial design goals I had way
back. The benefit being that an older distro kernel could be made to
support newer stranger hardware via a clocksource driver. While the
hardware vendors have for the most part consolidated on HPET/ACPI PM
which has mostly avoided the need, I still think its worth preserving.

thanks
-john