Re: [PATCH 2/2] xfs: verify icount in superblock write

[Brian Foster <bfoster@redhat.com>]

On Sun, Jul 29, 2018 at 02:39:00PM +1000, Dave Chinner wrote:
> On Fri, Jul 27, 2018 at 06:44:40AM -0400, Brian Foster wrote:
> > That aside.. since these values shouldn't change often I'm wondering if
> > it's worth calculating the global min/max once at mount time (we'd have
> 
> The values are needed during the initial read of the superblock
> before any "for the life of the mount" calculations can be made,
> and they can change across log recovery, too, if recovery replays a
> growfs transaction. Hence they really need to be point in time
> calculations based on the current superblock values during
> superblock reads....
> 

I'm not so worried about storing the calculated total as much as
simplifying the implementation. All but the last iteration in the
current loop calculate a fixed value, so this can trivially be
simplified to something like:

	xfs_agino_range(mp, --agcount, first, last);
	max = last - first + 1;
	if (agcount) {
		xfs_agino_range(mp, agcount - 1, first, last);
		max += (last - first + 1) * agcount;
	}

... which (if it mattered) could probably simplify even further if we
factored out the additional portion of the calculation that is fixed:

	xfs_agino_range(mp, --agcount, first, last);
	max = last - first + 1;
	if (agcount)
		max += agcount * mp->m_aginodes;

I.e., with an ->m_aginodes field in xfs_mount that is analogous to
sb_agblocks.

Brian

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html