On 13-04-16 02:43 AM, Niels de Vos wrote:
> That was exactly the next thing I was going to suggest :) I think the
> fhandle can be different sizes though, but if it is always round-up to
> have only full blocks of 4 bytes, it should work just fine.
That's one way to do it :)
> Could you pass the patch that you've used? I'll review and verify it
> a little too.
Attached. Nothing to complex, just extending the GlusterFS magic at the
beginning of a FH to 4B.
> However, you should open a service request at Oracle to let them know
> that their DNFS client should encode the fhandle in correct XDR (with
> roundup bytes to make full 4 byte blocks).
Haha… yes, yes I should. If only we were the end customer :/
M.
--
Michael Brown | `One of the main causes of the fall of
Systems Consultant | the Roman Empire was that, lacking zero,
Net Direct Inc. | they had no way to indicate successful
☎: +1 519 883 1172 x5106 | termination of their C programs.' - Firth
diff -ur orig-glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.c glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.c
--- orig-glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.c 2012-10-11 13:38:25.000000000 -0400
+++ glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.c 2013-04-15 14:27:13.006534649 -0400
@@ -46,6 +46,12 @@
if (fh->ident[1] != GF_NFSFH_IDENT1)
return 0;
+ if (fh->ident[2] != GF_NFSFH_IDENT2)
+ return 0;
+
+ if (fh->ident[3] != GF_NFSFH_IDENT3)
+ return 0;
+
return 1;
}
@@ -58,6 +64,8 @@
fh->ident[0] = GF_NFSFH_IDENT0;
fh->ident[1] = GF_NFSFH_IDENT1;
+ fh->ident[2] = GF_NFSFH_IDENT2;
+ fh->ident[3] = GF_NFSFH_IDENT3;
uuid_copy (fh->gfid, buf->ia_gfid);
}
@@ -161,6 +169,8 @@
newfh->ident[0] = GF_NFSFH_IDENT0;
newfh->ident[1] = GF_NFSFH_IDENT1;
+ newfh->ident[2] = GF_NFSFH_IDENT2;
+ newfh->ident[3] = GF_NFSFH_IDENT3;
uuid_copy (newfh->gfid, inode->gfid);
uuid_copy (newfh->exportid, exportid);
return 0;
diff -ur orig-glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.h glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.h
--- orig-glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.h 2012-10-11 13:38:25.000000000 -0400
+++ glusterfs-3.3.1/xlators/nfs/server/src/nfs3-fh.h 2013-04-15 14:27:30.895205267 -0400
@@ -35,7 +35,9 @@
* handles for now. This will change if and when we need v4. */
#define GF_NFSFH_IDENT0 ':'
#define GF_NFSFH_IDENT1 'O'
-#define GF_NFSFH_IDENT_SIZE (sizeof(char) * 2)
+#define GF_NFSFH_IDENT2 'G'
+#define GF_NFSFH_IDENT3 'L'
+#define GF_NFSFH_IDENT_SIZE (sizeof(char) * 4)
#define GF_NFSFH_STATIC_SIZE (GF_NFSFH_IDENT_SIZE + (2*sizeof (uuid_t)))
#define nfs3_fh_exportid_to_index(exprtid) ((uint16_t)exprtid[15])
@@ -45,9 +47,9 @@
struct nfs3_fh {
/* Used to ensure that a bunch of bytes are actually a GlusterFS NFS
- * file handle. Should contain ":O"
+ * file handle. Should contain ":OGL"
*/
- char ident[2];
+ char ident[4];
/* UUID that identifies an export. The value stored in exportid
* depends on the usage of gluster nfs. If the DVM is enabled using
