[PATCH] xfs/259: handle minimum block size more precisely

[Dave Chinner]

On Thu, Apr 07, 2016 at 04:48:37PM -0700, Christoph Hellwig wrote:
> On Fri, Apr 08, 2016 at 07:32:31AM +1000, Dave Chinner wrote:
> > > diff --git a/tests/xfs/259 b/tests/xfs/259
> > > index 16c1935..3150ff3 100755
> > > --- a/tests/xfs/259
> > > +++ b/tests/xfs/259
> > > @@ -51,9 +51,7 @@ testfile=$TEST_DIR/259.image
> > >  # Test various sizes slightly less than 4 TB. Need to handle different
> > >  # minimum block sizes for CRC enabled filesystems, but use a small log so we
> > >  # don't write lots of zeros unnecessarily.
> > > -xfs_info $TEST_DIR | _filter_mkfs 2> $tmp.mkfs > /dev/null
> > > -. $tmp.mkfs
> > 
> > This tests the configuration of the test device, which is not
> > controlled by the test harness, so can be different to the
> > configuration being used for the scratch device.
> > 
> > > -if [ $_fs_has_crcs -eq 1 ]; then
> > > +if [ $XFS_MKFS_CRC_DEFAULT -eq 1 ]; then
> > 
> > IOWs, this is not an not equivalent test.
> 
> And I think that's the whole point of this change :)
> 
> Previously it tested what the TEST_DIR did, which was wrong for this
> test.  Now it tests what mkfs does by default (including for the scratch
> dev), which is what we really want here.

Which is not at all clear from the patch description.

Seriously, though, this does not belong in common/config. We already
have a helper function to check what mkfs supports (i.e.
 _scratch_mkfs_xfs_supported()), and if we just want a bare check
then factor this into a _mkfs_xfs_supported() and supply the
parameters specific to the test.

Indeed, this is basically what we do with _require_xfs_mkfs_crc();
the same thing should be done, but without the "notrun" if -m crc
s not supported...

Cheers,

Dave.
-- 
Dave Chinner
david at fromorbit.com