Re: [PATCH v1 09/17] xfs: extent transaction reservations for parent attributes

From: Alli <allison.henderson@oracle.com>
To: Dave Chinner <david@fromorbit.com>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH v1 09/17] xfs: extent transaction reservations for parent attributes
Date: Fri, 17 Jun 2022 17:31:59 -0700	[thread overview]
Message-ID: <fcdbb58be768a82e2dac8204301439d2b27572ac.camel@oracle.com> (raw)
In-Reply-To: <20220616053807.GC227878@dread.disaster.area>

On Thu, 2022-06-16 at 15:38 +1000, Dave Chinner wrote:
> On Sat, Jun 11, 2022 at 02:41:52AM -0700, Allison Henderson wrote:
> > We need to add, remove or modify parent pointer attributes during
> > create/link/unlink/rename operations atomically with the dirents in
> > the
> > parent directories being modified. This means they need to be
> > modified
> > in the same transaction as the parent directories, and so we need
> > to add
> > the required space for the attribute modifications to the
> > transaction
> > reservations.
> > 
> > [achender: rebased, added xfs_sb_version_hasparent stub]
> > 
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> > Signed-off-by: Allison Henderson <allison.henderson@oracle.com>
> > ---
> >  fs/xfs/libxfs/xfs_format.h     |   5 ++
> >  fs/xfs/libxfs/xfs_trans_resv.c | 103 +++++++++++++++++++++++++++
> > ------
> >  fs/xfs/libxfs/xfs_trans_resv.h |   1 +
> >  3 files changed, 90 insertions(+), 19 deletions(-)
> > 
> > diff --git a/fs/xfs/libxfs/xfs_format.h
> > b/fs/xfs/libxfs/xfs_format.h
> > index afdfc8108c5f..96976497306c 100644
> > --- a/fs/xfs/libxfs/xfs_format.h
> > +++ b/fs/xfs/libxfs/xfs_format.h
> > @@ -390,6 +390,11 @@ xfs_sb_has_incompat_feature(
> >  	return (sbp->sb_features_incompat & feature) != 0;
> >  }
> >  
> > +static inline bool xfs_sb_version_hasparent(struct xfs_sb *sbp)
> > +{
> > +	return false; /* We'll enable this at the end of the set */
> > +}
> 
> Just noticed this in passing - I have not looked at the reservation
> calculations at all yet. We don't use "xfs_sb_version" feature
> checks anymore. This goes into xfs_mount.h as a feature flag and we
> use xfs_has_parent_pointers(mp) to check for the feature rather that
> superblock bits.
Sure, will move.  Thx for the reviews!

Allison
> 
> Cheers,
> 
> Dave.