tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: rfc: high-resolution timer framework

On Tue, Aug 05, 2008 at 11:17:34PM +0300, Alexander Shishkin wrote:
> On Fri, Aug 01, 2008 at 05:35:04PM +0000, Christos Zoulas wrote:
> > >Well, I was rather hoping that this code receives a fair amount of
> > >testing before it can take the place of CLOCK_REALTIME.
> > 
> > Well, it will definitely get more testing if it replaces CLOCK_REALTIME :-)
> The other thing is, I'm not entirely sure if it makes sense to
> completely get rid of current implementation based on the timer wheel. I
> think this should be properly thought through and discussed. Probably I
> can come up with a way for both to coexist.

The 'timer wheel' implementation is very efficient for the majority of
timers, especially for 'guard' timers in protocol code that are started
and then usually cancelled (often almost immediately) and rarely expire.

For such timers you don't want to sort them completly.

If you are not careful then device drivers will have to start implementing
their own timer subsystem because the kernel-provide one is not appropriate.

I do realise that, in some cases (maybe streaming audio/video) the current
timers are inadequate - especially with a 10ms tick.


David Laight:

Home | Main Index | Thread Index | Old Index