[PATCH 5/5] afs: use d_duplicate()

[NeilBrown]

From: NeilBrown <neil at brown.name>

To prepare for d_alloc_parallel() being permitted without a directory
lock, use d_duplicate() when duplicating a dentry in order to perform a
silly-rename.  This creates on in-lookup dentry which will contine to block
concurrent lookups after the original dentry is silly-renamed, and until
the final rename happens.

As the resulting dentry is in-lookup, we need to call d_lookup_done()
before d_put().

Signed-off-by: NeilBrown <neil at brown.name>
---
 fs/afs/dir.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/afs/dir.c b/fs/afs/dir.c
index b8773adc4e3e..dfd1930e2282 100644
--- a/fs/afs/dir.c
+++ b/fs/afs/dir.c
@@ -2075,7 +2075,10 @@ static void afs_rename_put(struct afs_operation *op)
 	store_release_wake_up(&op->dentry->d_flags,
 			      op->dentry->d_flags &~ DCACHE_BLOCKED);
 	spin_unlock(&op->dentry->d_lock);
-	dput(op->rename.tmp);
+	if (op->rename.tmp) {
+		d_lookup_done(op->rename.tmp);
+		dput(op->rename.tmp);
+	}
 }
 
 static const struct afs_operation_ops afs_rename_operation = {
@@ -2213,8 +2216,7 @@ static int afs_rename(struct mnt_idmap *idmap, struct inode *old_dir,
 			if (d_count(new_dentry) > 2) {
 				spin_unlock(&new_dentry->d_lock);
 				/* copy the target dentry's name */
-				op->rename.tmp = d_alloc(new_dentry->d_parent,
-							 &new_dentry->d_name);
+				op->rename.tmp = d_duplicate(new_dentry);
 				if (!op->rename.tmp) {
 					afs_op_nomem(op);
 					goto error;
-- 
2.50.0.107.gf914562f5916.dirty