Re: long delays (possibly infinite) in time_interpolator_get_counter

[Christoph Lameter]

On Mon, 1 Aug 2005, Bjorn Helgaas wrote:

> > If it is really synchronized then you can run with "nojitter" without any 
> > issue. Then you wont have to deal with the cmpxchg and everything is fine. 
> > But I suspect that the ITC are NOT truly synchronized (it has an 
> > "offset" that may be nonzero right?) otherwise you would have used nojitter.
> 
> And why should everyday users have to be concerned with "nojitter"?

Because so far ITCs were not truly synchronized.
 
> > Extra timer hardware is necessary to fix the ITC deficiency. You are at 
> > the source of the problem. Fix the damn hardware to provide a standardized 
> > synchronized clock or provide a truly synchronized ITC.
> 
> The "ITC deficiency" is a platform design issue.  Most small SMP platforms
> *do* synchronize the clocks of all processors.  Obviously that's difficult
> on large boxes, and then you may need extra timers in the platform.

I have no objections to making nojitter the default. There is already some 
PAL information that provides information on ITC reliability. If you can 
insure that this is accurate (currently it indicates no drift even if 
the ITC's have an offset) then simply switch on nojitter for small 
boxes.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/