Re: [patch] hrtimer: round up relative start time on low-res arches

[Ingo Molnar]

* Roman Zippel <zippel@xxxxxxxxxxxxxx> wrote:

> On Tue, 14 Feb 2006, Ingo Molnar wrote:
> 
> > CONFIG_TIME_LOW_RES is a temporary way for architectures to signal that 
> > they simply return xtime in do_gettimeoffset(). In this corner-case we 
> > want to round up by resolution when starting a relative timer, to avoid 
> > short timeouts. This will go away with the GTOD framework.
> 
> This fixes the worst cases. Even the common case should somehow 
> reflect that the relative start time should be rounded up in the same 
> way (even if not by that much), e.g. due to rounding the current 
> get_time() (at least for the non TIME_INTERPOLATION case) has a 1usec 
> resolution, which should be added there.

yeah, agreed. That will be accurately fixed via GTOD's per-hwclock 
resolution values. It will have another advantage as well: e.g. the 
whole of m68k wont be penalized via CONFIG_TIME_LOW_RES for having a 
handful of sub-arches (Apollo, Sun3x, Q40) that dont have a higher 
resolution timer - every clock can define its own resolution. You could 
help that effort by porting m68k to use GTOD ;-)

	Ingo
-
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/