Re: gettimeofday order of magnitude slower with pmtimer, which is default

[john stultz <johnstul@us.ibm.com>]

On Wed, 2006-03-22 at 21:12 +0200, Avi Kivity wrote:
> OGAWA Hirofumi wrote:
> > john stultz <johnstul@us.ibm.com> writes:
> >
> >   
> >> In my TOD rework I've dropped the triple read, figuring if a problem
> >> arose we could blacklist the specific box. This patch covers that, so it
> >> looks like a good idea to me.
> >>
> >> I've not tested it myself, but if you feel good about it, please send it
> >> to Andrew.
> >>     
> >
> > Current patch is the following. If I'm missing something, or you have
> > some comment, please tell me. (Since I don't have ICH4, ICH4 detection
> > is untested)
> >   
> Doesn't it make sense to mark the port as user accessible in the I/O 
> permissions bitmap and export it as a vsyscall? that would save the 
> syscall overhead.

i386 doesn't yet have a vsyscall gtod. I had some code around earlier
for it, but it still needed a good bit of work before it would be ready
for mainline. If you're interested in working on this, I'd be happy to
send them to you.

thanks
-john

-
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/