> Date: Wed, 15 Aug 2012 11:58:28 -0400 > From: Ted Unangst <[email protected]> > > On Wed, Aug 15, 2012 at 17:36, Mike Belopuhov wrote: > > On Wed, Aug 15, 2012 at 5:02 PM, Ted Unangst <[email protected]> wrote: > >> The acpihpet timer is, in my testing, lots better than the acpitimer. > >> Faster to read and more precise. They should not have the same quality > >> value. Double acpihpet. > >> > > > > as long as acpi subsystem attaches acpitimer earlier we don't > > need that since acpihpet always prevails.
Ah yes. Since the frequency of the timecounter is known, it makes sense to prefer the one with the highest frequency as it is more likely to provide better resolution. > There is still a user presentation issue. When I run sysctl > kern.timecounter, I don't want to see acpitimer and acpihpet listed > with the same quality. acpihpet will always have a higher frequency, > so it will always be picked, but users should not have to do digging > through the source of tc_init to discover why. Well, PHK's paper suggests that the quality of timecounters is much more about stability than it is about precision. Combined with the fact that the timecounter code does take the frequency of the timecounter into account when selecting the timecounter to use, I'd argue that the frequency shouldn't matter at all when assigning the quality.
