NFSv4: Follow a referral

Respond to a moved error on NFS lookup by setting up the referral.
Note: We don't actually follow the referral during lookup/getattr, but
later when we detect fsid mismatch in inode revalidation (similar to the
processing done for cloning submounts). Referrals will have fake attributes
until they are actually followed or traversed.

Signed-off-by: Manoj Naik <manoj@almaden.ibm.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

5 files changed, 60 insertions, 2 deletions

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 0d8302e59d69..ee13cb01b56e 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -888,7 +888,10 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
                                 set_bit(NFS_INO_ADVISE_RDPLUS, &NFS_FLAGS(inode));
                         /* Deal with crossing mountpoints */
                         if (!nfs_fsid_equal(&NFS_SB(sb)->fsid, &fattr->fsid)) {
-                                inode->i_op = &nfs_mountpoint_inode_operations;
+                                if (fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL)
+                                        inode->i_op = &nfs_referral_inode_operations;
+                                else
+                                        inode->i_op = &nfs_mountpoint_inode_operations;
                                 inode->i_fop = NULL;
                         }
                 } else if (S_ISLNK(inode->i_mode))
diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c
index e426516c1116..8ca44b7b25c3 100644
--- a/fs/nfs/namespace.c
+++ b/fs/nfs/namespace.c
@@ -58,7 +58,10 @@ static void * nfs_follow_mountpoint(struct dentry *dentry, struct nameidata *nd)
         if (err != 0)
                 goto out_err;
 
-        mnt = nfs_do_submount(nd->mnt, nd->dentry, &fh, &fattr);
+        if (fattr.valid & NFS_ATTR_FATTR_V4_REFERRAL)
+                mnt = nfs_do_refmount(nd->mnt, nd->dentry);
+        else
+                mnt = nfs_do_submount(nd->mnt, nd->dentry, &fh, &fattr);
         err = PTR_ERR(mnt);
         if (IS_ERR(mnt))
                 goto out_err;
@@ -94,6 +97,10 @@ struct inode_operations nfs_mountpoint_inode_operations = {
         .getattr        = nfs_getattr,
 };
 
+struct inode_operations nfs_referral_inode_operations = {
+        .follow_link    = nfs_follow_mountpoint,
+};
+
 static void nfs_expire_automounts(void *data)
 {
         struct list_head *list = (struct list_head *)data;
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 90ee21a07b3e..3300e35d74ad 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1462,6 +1462,50 @@ out:
         return nfs4_map_errors(status);
 }
 
+/*
+ * Get locations and (maybe) other attributes of a referral.
+ * Note that we'll actually follow the referral later when
+ * we detect fsid mismatch in inode revalidation
+ */
+static int nfs4_get_referral(struct inode *dir, struct qstr *name, struct nfs_fattr *fattr, struct nfs_fh *fhandle)
+{
+        int status = -ENOMEM;
+        struct page *page = NULL;
+        struct nfs4_fs_locations *locations = NULL;
+        struct dentry dentry = {};
+
+        page = alloc_page(GFP_KERNEL);
+        if (page == NULL)
+                goto out;
+        locations = kmalloc(sizeof(struct nfs4_fs_locations), GFP_KERNEL);
+        if (locations == NULL)
+                goto out;
+
+        dentry.d_name.name = name->name;
+        dentry.d_name.len = name->len;
+        status = nfs4_proc_fs_locations(dir, &dentry, locations, page);
+        if (status != 0)
+                goto out;
+        /* Make sure server returned a different fsid for the referral */
+        if (nfs_fsid_equal(&NFS_SERVER(dir)->fsid, &locations->fattr.fsid)) {
+                dprintk("%s: server did not return a different fsid for a referral at %s\n", __FUNCTION__, name->name);
+                status = -EIO;
+                goto out;
+        }
+
+        memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
+        fattr->valid |= NFS_ATTR_FATTR_V4_REFERRAL;
+        if (!fattr->mode)
+                fattr->mode = S_IFDIR;
+        memset(fhandle, 0, sizeof(struct nfs_fh));
+out:
+        if (page)
+                __free_page(page);
+        if (locations)
+                kfree(locations);
+        return status;
+}
+
 static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
 {
         struct nfs4_getattr_arg args = {
@@ -1566,6 +1610,8 @@ static int _nfs4_proc_lookup(struct inode *dir, struct qstr *name,
         
         dprintk("NFS call  lookup %s\n", name->name);
         status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
+        if (status == -NFS4ERR_MOVED)
+                status = nfs4_get_referral(dir, name, fattr, fhandle);
         dprintk("NFS reply lookup: %d\n", status);
         return status;
 }
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 09271b10f9a8..152798949113 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -409,6 +409,7 @@ extern void nfs_unregister_sysctl(void);
  */
 extern struct list_head nfs_automount_list;
 extern struct inode_operations nfs_mountpoint_inode_operations;
+extern struct inode_operations nfs_referral_inode_operations;
 extern int nfs_mountpoint_expiry_timeout;
 extern void nfs_release_automount_timer(void);
 
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index d6eea8348728..7c7320fa51aa 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -63,6 +63,7 @@ struct nfs_fattr {
 #define NFS_ATTR_FATTR          0x0002          /* post-op attributes */
 #define NFS_ATTR_FATTR_V3       0x0004          /* NFSv3 attributes */
 #define NFS_ATTR_FATTR_V4       0x0008          /* NFSv4 change attribute */
+#define NFS_ATTR_FATTR_V4_REFERRAL      0x0010          /* NFSv4 referral */
 
 /*
  * Info on the file system