Re: [Xen-devel] [PATCH 06 of 10] xenpaging: improve performance in policy_choose_victim [and 1 more messages]

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Ian Jackson writes ("Re: [Xen-devel] [PATCH 00 of 10] tools/xenpaging: cleanups 
and performance improvements"):
> Olaf Hering writes ("[Xen-devel] [PATCH 00 of 10] tools/xenpaging: cleanups 
> and performance improvements"):
> > This series adjusts the error reporting in the various code paths, with
> > the intention that fatal errors can be detected by callers and handled
> > properly. During my performance analysis with callgrind I found and
> > fixed a few bottlenecks in the page-in code paths.
> 
> Since these are all xenpaging changes, I'm inclined to follow your
> lead and commit them soon.  If anyone has any comments, please shout!

Olaf, can you please rebase to current tip ?  The other recent
xenpaging changes conflict with these.

And then while you're there you should obviously incorporate this
change:

Olaf Hering writes ("Re: [Xen-devel] [PATCH 06 of 10] xenpaging: improve 
performance in policy_choose_victim"):
> On Mon, Jan 30, Olaf Hering wrote:
> > +        if ( test_bit(current_gfn, bitmap) )
> > +            continue;
> > +
> > +        /* gfn already tested */
> > +        if ( test_bit(current_gfn, bitmap) )
> > +            continue;
> > +
> > +        /* gfn found */
> > +        break;
> 
> The second one should be "unconsumed" instead of "bitmap", otherwise
> wraparounds lead to endless loop in caller..

Thanks,
Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xensource.com/xen-devel