[pnfs] pnfs - new version of xdr for pNFS Linux client/server

[William A.(Andy) Adamson]

hi rahul

i just looked at http://www.nfsv4-editor.org/draft-05/nfs41_prot.x - and i 
find that FATTR4_FS_LAYOUT_TYPE is defined as 62. 64 is FATTR4_LAYOUT_TYPE 
which is the per-file layout type, and which we don't have implemented.

so, FS_LAYOUT_TYPES is changed from 56 to 62, and will fit inside the bitmap 
of length 2.

i suggest we make this change:

---

 pnfs-andros/include/linux/nfs4.h |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

diff -puN include/linux/nfs4.h~fs_layout_types include/linux/nfs4.h
--- pnfs/include/linux/nfs4.h~fs_layout_types   2006-09-05 13:53:44.000000000 
-0400
+++ pnfs-andros/include/linux/nfs4.h    2006-09-05 13:55:20.000000000 -0400
@@ -344,7 +344,7 @@ enum lock_type4 {
 #define FATTR4_WORD1_TIME_MODIFY        (1UL << 21)
 #define FATTR4_WORD1_TIME_MODIFY_SET    (1UL << 22)
 #define FATTR4_WORD1_MOUNTED_ON_FILEID  (1UL << 23)
-#define FATTR4_WORD1_FS_LAYOUT_TYPES   (1UL << 24)
+#define FATTR4_WORD1_FS_LAYOUT_TYPES   (1UL << 30)

 #define NFSPROC4_NULL 0
 #define NFSPROC4_COMPOUND 1
_

-->Andy

Rahul.Iyer at netapp.com said:
> And finally, there's the change of the FS_LAYOUT_TYPES from 56 to 64. This
> poses a larger problem as the value 64 would need a new word since the 2
> words defined (FATTR4_WORD0 and FATTR4_WORD1) would not be enough. We would
> have to introduce a new FATTR4_WORD2. This however might break things like
> GETATTR and would require a greater amount of testing. I was unsure of
> whether we should do this now, and so have not made that change. 


from draft 05 nfs41_prot.x:

const FATTR4_RECV_IMPL_ID	= 59;
const FATTR4_ABSENT		= 60;
const FATTR4_FS_STATUS		= 61;
const FATTR4_FS_LAYOUT_TYPE	= 62;
const FATTR4_LAYOUT_HINT	= 63;
const FATTR4_LAYOUT_TYPE	= 64;
const FATTR4_LAYOUT_BLKSIZE	= 65;
const FATTR4_LAYOUT_ALIGNMENT	= 66;
const FATTR4_FS_LOCATIONS_INFO	= 67;
const FATTR4_MDSTHRESHOLD	= 68;