Re: [PATCH 9/9] xfs: factor buffer reading from xfs_dir2_leaf_getdents

From: Dave Chinner <david@fromorbit.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: xfs@oss.sgi.com
Subject: Re: [PATCH 9/9] xfs: factor buffer reading from xfs_dir2_leaf_getdents
Date: Wed, 20 Jun 2012 17:41:17 +1000	[thread overview]
Message-ID: <20120620074117.GJ30705@dastard> (raw)
In-Reply-To: <20120620072709.GA27358@infradead.org>

On Wed, Jun 20, 2012 at 03:27:10AM -0400, Christoph Hellwig wrote:
> > +struct _map_info {
> > +	xfs_bmbt_irec_t	*map;		/* map vector for blocks */
> > +	xfs_extlen_t	map_blocks;	/* number of fsbs in map */
> > +	xfs_dablk_t	map_off;	/* last mapped file offset */
> > +	int		map_size;	/* total entries in *map */
> > +	int		map_valid;	/* valid entries in *map */
> > +	int		nmap;		/* mappings to ask xfs_bmapi */
> > +	xfs_dir2_db_t	curdb;		/* db for current block */
> > +};
> > +
> > +struct _ra_info {
> 
> Can you give these structure names xfs_dir2_leaf prefixes, please?

Sure.

> Also any reason the ra_info structure is kept entirely separate?

Logical grouping, easy to validate.

> While
> I see a bit of a point to have a logical grouping, it still semes more
> useful to just embedd it into the map_info.

Ok, I can do that.

> > +	map_info.map_size = howmany(bufsize + mp->m_dirblksize,
> > +				     mp->m_sb.sb_blocksize);
> > +	map_info.map = kmem_zalloc(map_info.map_size *
> > +					sizeof(struct xfs_bmbt_irec), KM_SLEEP);
> 
> I'd be tempted to say that the map field in the map_info should be a
> variable sized array the end, and the whole structure should be
> dynamically allocated to get a couple more variables off the stack.

Makes sense, even though we aren't usually under stack pressure
here.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs