The first patch fixes some _maxsz that are now wrong since a patch of
mine which changed for format of NFSv4.0 open-owners.

I think this inconsistency might have been easier to avoid if the
#define was closer to the code that it describes.  So the second patch
moves most of these #defines close to the code.  That requires moving
all the "encode" function after the "decode" functions.

So maybe the second patch is too intrusive to be worth it - it is just
an RFC patch.  The first patch is much less intrusive.

Thanks,
NeilBrown

 [PATCH 1/2] NFS: fix open_owner_id_maxsz and related fields.
 [PATCH 2/2] NFS: move _maxsz and _sz #defines to the function which