xfs
[Top] [All Lists]

Re: [PATCH 06/32] xfs: use btree block initialisation functions in growf

To: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Subject: Re: [PATCH 06/32] xfs: use btree block initialisation functions in growfs
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Sat, 24 Nov 2012 08:25:44 +1100
Cc: xfs@xxxxxxxxxxx
In-reply-to: <20121123124015.GB12150@xxxxxxxxxxxxx>
References: <1352721264-3700-1-git-send-email-david@xxxxxxxxxxxxx> <1352721264-3700-7-git-send-email-david@xxxxxxxxxxxxx> <20121123124015.GB12150@xxxxxxxxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Fri, Nov 23, 2012 at 07:40:15AM -0500, Christoph Hellwig wrote:
> On Mon, Nov 12, 2012 at 10:53:58PM +1100, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> > 
> > Factor xfs_btree_init_block() to be independent of the btree cursor,
> > and use the function to initialise btree blocks in the growfs code.
> > This makes adding support for different format btree blocks simple.
> > 
> > Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
> > ---
> >  fs/xfs/xfs_btree.c |   33 ++++++++++++++++++++++++---------
> >  fs/xfs/xfs_btree.h |   11 +++++++++++
> >  fs/xfs/xfs_fsops.c |   37 +++++++++++++------------------------
> >  3 files changed, 48 insertions(+), 33 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_btree.c b/fs/xfs/xfs_btree.c
> > index e53e317..121ea99 100644
> > --- a/fs/xfs/xfs_btree.c
> > +++ b/fs/xfs/xfs_btree.c
> > @@ -853,18 +853,22 @@ xfs_btree_set_sibling(
> >     }
> >  }
> >  
> > -STATIC void
> > +void
> >  xfs_btree_init_block(
> > -   struct xfs_btree_cur    *cur,
> > -   int                     level,
> > -   int                     numrecs,
> > -   struct xfs_btree_block  *new)   /* new block */
> > +   struct xfs_mount *mp,
> > +   struct xfs_buf  *bp,
> 
> Do we need the mount argument here?  bp->b_mount should always
> be initialized.

Possible.

> > +   __u32           magic,
> > +   __u16           level,
> > +   __u16           numrecs,
> > +   unsigned int    flags)
> >  {
> > -   new->bb_magic = cpu_to_be32(xfs_magics[cur->bc_btnum]);
> > +   struct xfs_btree_block  *new = XFS_BUF_TO_BLOCK(bp);
> 
> Any reaosn not to pass the btree_block directly instead of the buffer?

CRC additions require the block number to be put into the structure,
so we need to pass the buffer.

> 
> > +STATIC void
> > +xfs_btree_init_block_cur(
> > +   struct xfs_btree_cur    *cur,
> > +   int                     level,
> > +   int                     numrecs,
> > +   struct xfs_buf          *bp)
> > +{
> > +   xfs_btree_init_block(cur->bc_mp, bp, xfs_magics[cur->bc_btnum],
> > +                          level, numrecs, cur->bc_flags);
> 
> I'd scrapt this helper.

I only used it to avoid changing existing code. I can remove it when
CRCs are introduced if you want, as this touches all this code
again.

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>