On Tue, Feb 01, 2011 at 07:30:09PM -0600, Alex Elder wrote:
> On Tue, 2011-01-25 at 14:58 +1100, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> > test 042 generates a worst-case fragmented filesystem and uses it to
> > test xfs_fsr. It uses small 4k files to generate the hole-space-hole
> > pattern that fragments free space badly. It is much faster to
> > generate the same pattern by creating a single large file and
> > punching holes in it. Also, instead of writing large files to
> > create unfragmented space, just use preallocation so we don't have
> > to write the data to disk.
> > These changes reduce the runtime of the test on a single SATA drive
> > from 106s to 27s.
> > Signed-off-by: Dave Chinner <dchinner@xxxxxxxxxx>
> Perhaps your test system is configured different
> from mine, but I don't see the speedup you see.
> In fact, it might have slowed it down. I really
> haven't experimented much though and will report
> back if I find anything more constructive to say.
That may be because the resvsp call at the moment is a sync
transaction. I've got a patch to make that async which should solve
the problem for you. Mind you, the above numbers come from a Vm that
wasn't running that patch, so it probably depends mostly on how
efficient your storage is a processing barriers to the difference
you see right now...