Re: [PATCH 11/16] GFS: mount and tuning options

[Greg KH <greg@kroah.com>]

On Mon, Oct 10, 2005 at 04:14:29PM -0500, David Teigland wrote:
> On Mon, Oct 10, 2005 at 02:01:08PM -0700, Greg KH wrote:
> > On Mon, Oct 10, 2005 at 12:10:52PM -0500, David Teigland wrote:
> > > +static ssize_t statfs_show(struct gfs2_sbd *sdp, char *buf)
> > > +{
> > > +	struct gfs2_statfs_change sc;
> > > +	int rv;
> > > +
> > > +	if (gfs2_tune_get(sdp, gt_statfs_slow))
> > > +		rv = gfs2_statfs_slow(sdp, &sc);
> > > +	else
> > > +		rv = gfs2_statfs_i(sdp, &sc);
> > > +
> > > +	if (rv)
> > > +		goto out;
> > > +
> > > +	rv += sprintf(buf + rv, "bsize %u\n", sdp->sd_sb.sb_bsize);
> > > +	rv += sprintf(buf + rv, "total %lld\n", sc.sc_total);
> > > +	rv += sprintf(buf + rv, "free %lld\n", sc.sc_free);
> > > +	rv += sprintf(buf + rv, "dinodes %lld\n", sc.sc_dinodes);
> > 
> > No, 1 value per sysfs file please.
> 
> I'm aware of that rule and have followed it everywhere else.  This is a
> special case where the one statfs produces three results.

Then why not have 4 different files, for the result of the last "statfs"
command?

thanks,

greg k-h