Re: [PATCH 0/5] clocksource patches

[Ingo Molnar <mingo@elte.hu>]

* Thomas Gleixner <tglx@linutronix.de> wrote:

> > For example above you bascially only state that your clock event source 
> > is superior and the correct way of doing this without any explanation why 
> > (and the "No, thanks." doesn't exactly imply that you're even interested 
> > in alternatives). 
> 
> The question arises, who is not interested in alternatives. You are 
> well aware about the efforts others made, but you don't even think 
> about working together with them. Do you really expect people to jump 
> on your train, when you entirely ignore their work and efforts and 
> just propose your own view of the world?
> 
> I did nowhere say that I'm not interested in alternative solutions.  
> You interpret it into my words for whatever reason.

just to explain it to everyone: the code Thomas refers to and which we 
are working on is John's GTOD patchset with Thomas' high-resolution 
timers patches ontop of it. [all of that (and more) is glued together in 
the -rt tree as well].

Thomas' hrtimers queue (ontop of 2.6.16) is a practical, working 
implementation of the clock-event design Thomas is talking about, 
resulting in a working high-resolution timers solution that spans all 
the relevant Linux APIs: nanosleep() and POSIX timers. So Thomas' 
arguments derive straight from that experience.

for more details, the latest hrtimers code can be found at:

  http://tglx.de/projects/hrtimers

the merge of the hrtimers subsystem into 2.6.16 was just the first step, 
and the next steps are expressed in the patches above.

	Ingo