Merge tag 'nfs-for-3.11-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull second set of NFS client updates from Trond Myklebust:
 "This mainly contains some small readdir optimisations that had
  dependencies on Al Viro's readdir rewrite.  There is also a fix for a
  nasty deadlock which surfaced earlier in this merge window.

  Highlights include:
   - Fix an_rpc pipefs regression that causes a deadlock on mount
   - Readdir optimisations by Scott Mayhew and Jeff Layton
   - clean up the rpc_pipefs dentry operation setup"

* tag 'nfs-for-3.11-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  SUNRPC: Fix a deadlock in rpc_client_register()
  rpc_pipe: rpc_dir_inode_operations can be static
  NFS: Allow nfs_updatepage to extend a write under additional circumstances
  NFS: Make nfs_readdir revalidate less often
  NFS: Make nfs_attribute_cache_expired() non-static
  rpc_pipe: set dentry operations at d_alloc time
  nfs: set verifier on existing dentries in nfs_prime_dcache

6 files changed, 58 insertions, 23 deletions

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 0fac2cb1ea18..e474ca2b2bfe 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -450,6 +450,7 @@ void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry)
 	dentry = d_lookup(parent, &filename);
 	if (dentry != NULL) {
 		if (nfs_same_file(dentry, entry)) {
+			nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
 			status = nfs_refresh_inode(dentry->d_inode, entry->fattr);
 			if (!status)
 				nfs_setsecurity(dentry->d_inode, entry->fattr, entry->label);
@@ -817,7 +818,7 @@ static int nfs_readdir(struct file *file, struct dir_context *ctx)
 	nfs_readdir_descriptor_t my_desc,
 			*desc = &my_desc;
 	struct nfs_open_dir_context *dir_ctx = file->private_data;
-	int res;
+	int res = 0;
 
 	dfprintk(FILE, "NFS: readdir(%s/%s) starting at cookie %llu\n",
 			dentry->d_parent->d_name.name, dentry->d_name.name,
@@ -839,7 +840,8 @@ static int nfs_readdir(struct file *file, struct dir_context *ctx)
 	desc->plus = nfs_use_readdirplus(inode, ctx) ? 1 : 0;
 
 	nfs_block_sillyrename(dentry);
-	res = nfs_revalidate_mapping(inode, file->f_mapping);
+	if (ctx->pos == 0 || nfs_attribute_cache_expired(inode))
+		res = nfs_revalidate_mapping(inode, file->f_mapping);
 	if (res < 0)
 		goto out;
 
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index c93639e6cf68..af6e806044d7 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -936,7 +936,7 @@ int nfs_attribute_timeout(struct inode *inode)
 	return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
 }
 
-static int nfs_attribute_cache_expired(struct inode *inode)
+int nfs_attribute_cache_expired(struct inode *inode)
 {
 	if (nfs_have_delegated_attributes(inode))
 		return 0;
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index a2c7c28049d5..f1bdb7254776 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -888,6 +888,28 @@ out:
 	return PageUptodate(page) != 0;
 }
 
+/* If we know the page is up to date, and we're not using byte range locks (or
+ * if we have the whole file locked for writing), it may be more efficient to
+ * extend the write to cover the entire page in order to avoid fragmentation
+ * inefficiencies.
+ *
+ * If the file is opened for synchronous writes or if we have a write delegation
+ * from the server then we can just skip the rest of the checks.
+ */
+static int nfs_can_extend_write(struct file *file, struct page *page, struct inode *inode)
+{
+	if (file->f_flags & O_DSYNC)
+		return 0;
+	if (NFS_PROTO(inode)->have_delegation(inode, FMODE_WRITE))
+		return 1;
+	if (nfs_write_pageuptodate(page, inode) && (inode->i_flock == NULL ||
+			(inode->i_flock->fl_start == 0 &&
+			inode->i_flock->fl_end == OFFSET_MAX &&
+			inode->i_flock->fl_type != F_RDLCK)))
+		return 1;
+	return 0;
+}
+
 /*
  * Update and possibly write a cached page of an NFS file.
  *
@@ -908,14 +930,7 @@ int nfs_updatepage(struct file *file, struct page *page,
 		file->f_path.dentry->d_name.name, count,
 		(long long)(page_file_offset(page) + offset));
 
-	/* If we're not using byte range locks, and we know the page
-	 * is up to date, it may be more efficient to extend the write
-	 * to cover the entire page in order to avoid fragmentation
-	 * inefficiencies.
-	 */
-	if (nfs_write_pageuptodate(page, inode) &&
-			inode->i_flock == NULL &&
-			!(file->f_flags & O_DSYNC)) {
+	if (nfs_can_extend_write(file, page, inode)) {
 		count = max(count + offset, nfs_page_length(page));
 		offset = 0;
 	}
diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
index 0b176297aaf6..7125cef74164 100644
--- a/include/linux/nfs_fs.h
+++ b/include/linux/nfs_fs.h
@@ -348,6 +348,7 @@ extern int nfs_permission(struct inode *, int);
 extern int nfs_open(struct inode *, struct file *);
 extern int nfs_release(struct inode *, struct file *);
 extern int nfs_attribute_timeout(struct inode *inode);
+extern int nfs_attribute_cache_expired(struct inode *inode);
 extern int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode);
 extern int __nfs_revalidate_inode(struct nfs_server *, struct inode *);
 extern int nfs_revalidate_mapping(struct inode *inode, struct address_space *mapping);
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index f0339ae9bf37..aa401560777b 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -290,7 +290,7 @@ static int rpc_client_register(const struct rpc_create_args *args,
 	struct rpc_auth *auth;
 	struct net *net = rpc_net_ns(clnt);
 	struct super_block *pipefs_sb;
-	int err = 0;
+	int err;
 
 	pipefs_sb = rpc_get_sb_net(net);
 	if (pipefs_sb) {
@@ -299,6 +299,10 @@ static int rpc_client_register(const struct rpc_create_args *args,
 			goto out;
 	}
 
+	rpc_register_client(clnt);
+	if (pipefs_sb)
+		rpc_put_sb_net(net);
+
 	auth = rpcauth_create(args->authflavor, clnt);
 	if (IS_ERR(auth)) {
 		dprintk("RPC:       Couldn't create auth handle (flavor %u)\n",
@@ -306,16 +310,14 @@ static int rpc_client_register(const struct rpc_create_args *args,
 		err = PTR_ERR(auth);
 		goto err_auth;
 	}
-
-	rpc_register_client(clnt);
+	return 0;
+err_auth:
+	pipefs_sb = rpc_get_sb_net(net);
+	__rpc_clnt_remove_pipedir(clnt);
 out:
 	if (pipefs_sb)
 		rpc_put_sb_net(net);
 	return err;
-
-err_auth:
-	__rpc_clnt_remove_pipedir(clnt);
-	goto out;
 }
 
 static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, struct rpc_xprt *xprt)
diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c
index 4679df5a6d50..61239a2cb786 100644
--- a/net/sunrpc/rpc_pipe.c
+++ b/net/sunrpc/rpc_pipe.c
@@ -480,6 +480,23 @@ static const struct dentry_operations rpc_dentry_operations = {
 	.d_delete = rpc_delete_dentry,
 };
 
+/*
+ * Lookup the data. This is trivial - if the dentry didn't already
+ * exist, we know it is negative.
+ */
+static struct dentry *
+rpc_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
+{
+	if (dentry->d_name.len > NAME_MAX)
+		return ERR_PTR(-ENAMETOOLONG);
+	d_add(dentry, NULL);
+	return NULL;
+}
+
+static const struct inode_operations rpc_dir_inode_operations = {
+	.lookup		= rpc_lookup,
+};
+
 static struct inode *
 rpc_get_inode(struct super_block *sb, umode_t mode)
 {
@@ -492,7 +509,7 @@ rpc_get_inode(struct super_block *sb, umode_t mode)
 	switch (mode & S_IFMT) {
 	case S_IFDIR:
 		inode->i_fop = &simple_dir_operations;
-		inode->i_op = &simple_dir_inode_operations;
+		inode->i_op = &rpc_dir_inode_operations;
 		inc_nlink(inode);
 	default:
 		break;
@@ -666,11 +683,8 @@ static struct dentry *__rpc_lookup_create_exclusive(struct dentry *parent,
 		if (!dentry)
 			return ERR_PTR(-ENOMEM);
 	}
-	if (dentry->d_inode == NULL) {
-		if (!dentry->d_op)
-			d_set_d_op(dentry, &rpc_dentry_operations);
+	if (dentry->d_inode == NULL)
 		return dentry;
-	}
 	dput(dentry);
 	return ERR_PTR(-EEXIST);
 }
@@ -1117,6 +1131,7 @@ rpc_fill_super(struct super_block *sb, void *data, int silent)
 	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
 	sb->s_magic = RPCAUTH_GSSMAGIC;
 	sb->s_op = &s_ops;
+	sb->s_d_op = &rpc_dentry_operations;
 	sb->s_time_gran = 1;
 
 	inode = rpc_get_inode(sb, S_IFDIR | S_IRUGO | S_IXUGO);