Re: [PATCH 15/21] implement generic xfs_btree_rshift

[Dave Chinner <david@xxxxxxxxxxxxx>]

On Fri, Aug 01, 2008 at 09:49:14PM +0200, Christoph Hellwig wrote:
> > > +                           XFS_BB_NUM_BITS, &first, &last);
> > > +         xfs_trans_log_buf(cur->bc_tp, bp, first, last);
> > > + } else {
> > > +         /* XXX(hch): maybe factor out into a method? */
> > > +         xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
> > > +                 XFS_ILOG_FBROOT(cur->bc_private.b.whichfork));
> > 
> > I don't think it is necessary at this point.
> 
> It's the only leakage of the detailed inode root implementation into
> the generic code, so I'm still wondering whether a method would be
> better.

Ah, right. yes, it probably would be cleaner to do it as a
separate method, but Ǐ don't think it's that important right now.

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx