Re: [RFC][PATCH] Runtime switching of the idle function [take 2]

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

On Wed, 2005-11-30 at 02:58 +0100, Andi Kleen wrote:
> > > Then you're likely running 32bit. It doesn't use vsyscall gettimeofday
> > > yet, which makes it slower. 64bit would.
> > 
> > Yes, I am.  So it sounds like vsyscall gettimeofday for i386 is in the
> > works?
> 
> John Stultz used to have patches for it, but for some reason he never
> pushed them into mainline. 

Unfortunately it was a pretty ugly patch. Correctness issues with the
existing code have kept focused on my timekeeping rework, however I have
kept it in mind, and I do have a i386 vsyscall gtod patch that applies
ontop of my tod work. I've been maintaining it on the side while I focus
on the core code, but it is much cleaner now. For fun I'll try to
remember to send it out with the next release.

> On i386 it unfortunately needs adding
> a test and branch to the syscall path to be 100% ABI compatible, but I 
> doubt that was the reason he dropped it.

Yea, I didn't know enough about the VDSO/unwind bits to get it to do the
right thing w/ glibc, so that bit was pretty hackish. I'll still need
some help on this bit to make it really something that could be
included.

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/