Re: [PATCH 10/16] xfs: convert xfs_fs_geometry to use mount feature checks

From: "Darrick J. Wong" <djwong@kernel.org>
To: Dave Chinner <david@fromorbit.com>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH 10/16] xfs: convert xfs_fs_geometry to use mount feature checks
Date: Wed, 14 Jul 2021 16:38:41 -0700	[thread overview]
Message-ID: <20210714233841.GS22402@magnolia> (raw)
In-Reply-To: <20210714233554.GC664593@dread.disaster.area>

On Thu, Jul 15, 2021 at 09:35:54AM +1000, Dave Chinner wrote:
> On Wed, Jul 14, 2021 at 04:15:49PM -0700, Darrick J. Wong wrote:
> > On Wed, Jul 14, 2021 at 02:19:06PM +1000, Dave Chinner wrote:
> > > From: Dave Chinner <dchinner@redhat.com>
> > > 
> > > Reporting filesystem features to userspace is currently superblock
> > > based. Now we have a general mount-based feature infrastructure,
> > > switch to using the xfs_mount rather than the superblock directly.
> > > 
> > > This reduces the size of the function by over 300 bytes.
> > > 
> > > $ size -t fs/xfs/built-in.a
> > > 	text    data     bss     dec     hex filename
> > > before	1127855  311352     484 1439691  15f7cb (TOTALS)
> > > after	1127535  311352     484 1439371  15f68b (TOTALS)
> > > 
> > > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> > > ---
> > >  fs/xfs/libxfs/xfs_sb.c | 46 ++++++++++++++++++++++--------------------
> > >  fs/xfs/libxfs/xfs_sb.h |  2 +-
> > >  fs/xfs/xfs_ioctl.c     |  2 +-
> > >  fs/xfs/xfs_ioctl32.c   |  2 +-
> > >  4 files changed, 27 insertions(+), 25 deletions(-)
> > > 
> > > diff --git a/fs/xfs/libxfs/xfs_sb.c b/fs/xfs/libxfs/xfs_sb.c
> > > index baaec7e6a975..5eaf14b6fe3c 100644
> > > --- a/fs/xfs/libxfs/xfs_sb.c
> > > +++ b/fs/xfs/libxfs/xfs_sb.c
> > > @@ -1016,10 +1016,12 @@ xfs_sync_sb_buf(
> > >  
> > >  void
> > >  xfs_fs_geometry(
> > > -	struct xfs_sb		*sbp,
> > > +	struct xfs_mount	*mp,
> > 
> > Hmm.  I /think/ this won't cause problems in userspace, because db
> > passes in &mp->m_sb, and mkfs passes in sbp, which points to &mp->m_sb,
> > and mp itself points to &mbuf...
> 
> Yes, AFAICT that is correct. mkfs.xfs has the same setup so it
> should work there, too.
> 
> > >  	struct xfs_fsop_geom	*geo,
> > >  	int			struct_version)
> > >  {
> > > +	struct xfs_sb		*sbp = &mp->m_sb;
> > > +
> > >  	memset(geo, 0, sizeof(struct xfs_fsop_geom));
> > >  
> > >  	geo->blocksize = sbp->sb_blocksize;
> > > @@ -1050,51 +1052,51 @@ xfs_fs_geometry(
> > >  	geo->flags = XFS_FSOP_GEOM_FLAGS_NLINK |
> > >  		     XFS_FSOP_GEOM_FLAGS_DIRV2 |
> > >  		     XFS_FSOP_GEOM_FLAGS_EXTFLG;
> > > -	if (xfs_sb_version_hasattr(sbp))
> > > +	if (xfs_has_attr(mp))
> > 
> > ...provided those utilities set m_features before we get to this point,
> > right?  And provided that libxfs_init will take care of that, I think
> > only mkfs would need special consideration, right?
> 
> Yup, that's how I read the usrespace situation - the superblock we
> pass to xfs_fs_geometry() is always attached to a mount structure,
> and we'll set up the features field in the mount when we first read
> in the superblock in libxfs_init(). mkfs might require a specific
> call to set up the m_features field after the superblock is
> finalised (i.e in initialise_mount()), but other than that I
> think it all just works...

Ok.  I'm glad you thought about userspace too!

Reviewed-by: Darrick J. Wong <djwong@kernel.org>

--D

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com