Subject: Re: [vserver] xfs and vserver 2.3.0.35.6pre
From: "Michael S. Zick" <mszick@morethan.org>
Date: Mon, 13 Oct 2008 09:07:11 -0500

On Mon October 13 2008, kristian wrote:
> On Mon, Oct 13, 2008 at 5:54 PM, Michael S. Zick <mszick@morethan.org>wrote:
> 
> > On Mon October 13 2008, kristian wrote:
> > > hello,
> > >
> > > I did apply (after ONE minor fix) vserver 2.3.0.35.6pre to the kernel
> > > 2.6.27-rc9 without failures.
> > >
> >
> > Did the same to 2.6.27-stable this weekend -
> > Ignoring the expected Makefile reject, I had one hunk
> > fail to apply to fs/dcache.c
> >
> > I "corrected" (quoted since I don't have a clue what
> > I am doing here) with:
> >
> > --- fs/dcache.c.orig    2008-10-09 17:13:53.000000000 -0500
> > +++ fs/dcache.c 2008-10-12 10:49:04.000000000 -0500
> > @@ -962,6 +970,7 @@
> >        if (parent)
> >                list_add(&dentry->d_u.d_child, &parent->d_subdirs);
> >        dentry_stat.nr_dentry++;
> > +       vx_dentry_inc(dentry);
> >        spin_unlock(&dcache_lock);
> >
> > Is that same/similar to your "one minor fix"?
> >
> 
> this was already in the original patch,  but I did change the next part
> (also not really knowing what I do, but it all looks sane) for fs/dcache.c
> 
> @@ -1412,4 +1421,5 @@ struct dentry * __d_lookup(struct dentry
> 
>                 atomic_inc(&dentry->d_count);
> +               vx_dentry_inc(dentry);
>                 found = dentry;
>                 spin_unlock(&dentry->d_lock);
>

I probably just snipped the wrong fragment out of the difference
I made and posted it.  I should have looked 500 lines deeper. ;)

I am certain we are talking about the same change (the one you posted).

Somebody made a gratuitous inversion of the logic to have a single
exit point for several tests.  Someone should mention to them that 
gcc is an optimizing compiler - I am guessing it generates the same 
code regardless of how you write it. ;)  
(I didn't actually check.)

Mike
 
> with regards
> Kristian
> 
> 
> > > than I realized that xfs is not
> > > compiling. after removing the xfs support from the kernel
> > > configuration everything went smooth and works OK so far.
> > >
> >
> > Same here - had to disable XFS and just get on with my life.
> >
> > > since I work on a project where we want to use vserver on a kernel
> > > 2.6.27 in the near future (rather sooner than later), I have two
> > > questions:
> > >
> > > * are there already people using the 2.3.0.35 patch on a production
> > system ?
> > >
> >
> > Not used in a production system here but I am making massive use of
> > COW to replicate and track changes made in a large (3Gbyte) build system.
> >
> > What can I say?  It hasn't crashed and burned (yet).
> >
> > Mike
> > > * are we the only ones using xfs ? if not, is there a patchset which
> > > compiles - I am happy to test
> > > it.
> > >
> > > maybe I can help somehow (fixing xfs patch is unfortunatley beyond my
> > > capabilities) ?
> > >
> > > thanx for any comments,
> > >
> > > with regards
> > > Kristian
> > >
> > >
> >
> >
> >
> 
>