Patch "Revert "NFS: nfs_rename() handle -ERESTARTSYS dentry left behind"" has been added to the 4.11-stable tree

Patch "Revert "NFS: nfs_rename() handle -ERESTARTSYS dentry left behind"" has been added to the 4.11-stable tree

[gregkh]

This is a note to let you know that I've just added the patch titled

    Revert "NFS: nfs_rename() handle -ERESTARTSYS dentry left behind"

to the 4.11-stable tree which can be found at:
    http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
     revert-nfs-nfs_rename-handle-erestartsys-dentry-left-behind.patch
and it can be found in the queue-4.11 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree,
please let <stable@vger.kernel.org> know about it.


>From d9f2950006f110f54444a10442752372ee568289 Mon Sep 17 00:00:00 2001
From: Benjamin Coddington <bcodding@redhat.com>
Date: Fri, 16 Jun 2017 11:12:59 -0400
Subject: Revert "NFS: nfs_rename() handle -ERESTARTSYS dentry left behind"

From: Benjamin Coddington <bcodding@redhat.com>

commit d9f2950006f110f54444a10442752372ee568289 upstream.

This reverts commit 920b4530fb80430ff30ef83efe21ba1fa5623731 which could
call d_move() without holding the directory's i_mutex, and reverts commit
d4ea7e3c5c0e341c15b073016dbf3ab6c65f12f3 "NFS: Fix old dentry rehash after
move", which was a follow-up fix.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Fixes: 920b4530fb80 ("NFS: nfs_rename() handle -ERESTARTSYS dentry left behind")
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 fs/nfs/dir.c |   51 ++++++++++++++++++++++++---------------------------
 1 file changed, 24 insertions(+), 27 deletions(-)

--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -2002,29 +2002,6 @@ nfs_link(struct dentry *old_dentry, stru
 }
 EXPORT_SYMBOL_GPL(nfs_link);
 
-static void
-nfs_complete_rename(struct rpc_task *task, struct nfs_renamedata *data)
-{
-	struct dentry *old_dentry = data->old_dentry;
-	struct dentry *new_dentry = data->new_dentry;
-	struct inode *old_inode = d_inode(old_dentry);
-	struct inode *new_inode = d_inode(new_dentry);
-
-	nfs_mark_for_revalidate(old_inode);
-
-	switch (task->tk_status) {
-	case 0:
-		if (new_inode != NULL)
-			nfs_drop_nlink(new_inode);
-		d_move(old_dentry, new_dentry);
-		nfs_set_verifier(new_dentry,
-					nfs_save_change_attribute(data->new_dir));
-		break;
-	case -ENOENT:
-		nfs_dentry_handle_enoent(old_dentry);
-	}
-}
-
 /*
  * RENAME
  * FIXME: Some nfsds, like the Linux user space nfsd, may generate a
@@ -2055,7 +2032,7 @@ int nfs_rename(struct inode *old_dir, st
 {
 	struct inode *old_inode = d_inode(old_dentry);
 	struct inode *new_inode = d_inode(new_dentry);
-	struct dentry *dentry = NULL;
+	struct dentry *dentry = NULL, *rehash = NULL;
 	struct rpc_task *task;
 	int error = -EBUSY;
 
@@ -2078,8 +2055,10 @@ int nfs_rename(struct inode *old_dir, st
 		 * To prevent any new references to the target during the
 		 * rename, we unhash the dentry in advance.
 		 */
-		if (!d_unhashed(new_dentry))
+		if (!d_unhashed(new_dentry)) {
 			d_drop(new_dentry);
+			rehash = new_dentry;
+		}
 
 		if (d_count(new_dentry) > 2) {
 			int err;
@@ -2096,6 +2075,7 @@ int nfs_rename(struct inode *old_dir, st
 				goto out;
 
 			new_dentry = dentry;
+			rehash = NULL;
 			new_inode = NULL;
 		}
 	}
@@ -2104,8 +2084,7 @@ int nfs_rename(struct inode *old_dir, st
 	if (new_inode != NULL)
 		NFS_PROTO(new_inode)->return_delegation(new_inode);
 
-	task = nfs_async_rename(old_dir, new_dir, old_dentry, new_dentry,
-					nfs_complete_rename);
+	task = nfs_async_rename(old_dir, new_dir, old_dentry, new_dentry, NULL);
 	if (IS_ERR(task)) {
 		error = PTR_ERR(task);
 		goto out;
@@ -2115,9 +2094,27 @@ int nfs_rename(struct inode *old_dir, st
 	if (error == 0)
 		error = task->tk_status;
 	rpc_put_task(task);
+	nfs_mark_for_revalidate(old_inode);
 out:
+	if (rehash)
+		d_rehash(rehash);
 	trace_nfs_rename_exit(old_dir, old_dentry,
 			new_dir, new_dentry, error);
+	if (!error) {
+		if (new_inode != NULL)
+			nfs_drop_nlink(new_inode);
+		/*
+		 * The d_move() should be here instead of in an async RPC completion
+		 * handler because we need the proper locks to move the dentry.  If
+		 * we're interrupted by a signal, the async RPC completion handler
+		 * should mark the directories for revalidation.
+		 */
+		d_move(old_dentry, new_dentry);
+		nfs_set_verifier(new_dentry,
+					nfs_save_change_attribute(new_dir));
+	} else if (error == -ENOENT)
+		nfs_dentry_handle_enoent(old_dentry);
+
 	/* new dentry created? */
 	if (dentry)
 		dput(dentry);


Patches currently in stable-queue which might be from bcodding@redhat.com are

queue-4.11/revert-nfs-nfs_rename-handle-erestartsys-dentry-left-behind.patch
queue-4.11/nfsv4.2-don-t-send-mode-again-in-post-exclusive4_1-setattr-with-umask.patch