On Sat, 26 Apr 2008, H. Peter Anvin wrote:

quoted text
> dean gaudet wrote:
> > On Sat, 26 Apr 2008, H. Peter Anvin wrote:
> > 
> > > dean gaudet wrote:
> > > > i might be too late... but shouldn't these #defines be PR_SET_RDTSC and
> > > > PR_GET_RDTSC or something like that?
> > > > 
> > > > to me calling them PR_SET_TSC/PR_GET_TSC just seem like alternative ways
> > > > to
> > > > change/get the TSC (and could even reduce to portable TSC
> > > > implementations...
> > > > since such registers do exist on other architectures).
> > > > 
> > > I would argue no, the flag is "is the TSC available".  RDTSC is an
> > > x86-specific name and would map poorly onto other architectures.
> > 
> > yeah but "SET TSC" to me reads as "set the TSC".... i read nothing about
> > making some instruction available or not.
> > 
> > although clock_gettime/clock_settime could more natural APIs for such
> > things.
> > 
> 
> PR_SET/PR_GET are the common prefixes, though.

that's not what's at issue... "PR_SET_TSC" reads "set the TSC" ... it 
doesn't read "set the ability to use the rdtsc instruction". 
PR_SET_RDTSC_ENABLE would seem to be more accurate... or even 
PR_SET_CR4_TSD which is what it really does :)

-dean
--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/