1 files changed, 227 insertions, 101 deletions

diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index cf7365581031..9d3a5f29f17f 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -27,8 +27,15 @@
 
 #define NFS_DEFAULT_DELEGATION_WATERMARK (5000U)
 
-static atomic_long_t nfs_active_delegations;
 static unsigned nfs_delegation_watermark = NFS_DEFAULT_DELEGATION_WATERMARK;
+module_param_named(delegation_watermark, nfs_delegation_watermark, uint, 0644);
+
+static struct hlist_head *nfs_delegation_hash(struct nfs_server *server,
+		const struct nfs_fh *fhandle)
+{
+	return server->delegation_hash_table +
+		(nfs_fhandle_hash(fhandle) & server->delegation_hash_mask);
+}
 
 static void __nfs_free_delegation(struct nfs_delegation *delegation)
 {
@@ -37,11 +44,12 @@ static void __nfs_free_delegation(struct nfs_delegation *delegation)
 	kfree_rcu(delegation, rcu);
 }
 
-static void nfs_mark_delegation_revoked(struct nfs_delegation *delegation)
+static void nfs_mark_delegation_revoked(struct nfs_server *server,
+		struct nfs_delegation *delegation)
 {
 	if (!test_and_set_bit(NFS_DELEGATION_REVOKED, &delegation->flags)) {
 		delegation->stateid.type = NFS4_INVALID_STATEID_TYPE;
-		atomic_long_dec(&nfs_active_delegations);
+		atomic_long_dec(&server->nr_active_delegations);
 		if (!test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
 			nfs_clear_verifier_delegated(delegation->inode);
 	}
@@ -59,9 +67,10 @@ static void nfs_put_delegation(struct nfs_delegation *delegation)
 		__nfs_free_delegation(delegation);
 }
 
-static void nfs_free_delegation(struct nfs_delegation *delegation)
+static void nfs_free_delegation(struct nfs_server *server,
+		struct nfs_delegation *delegation)
 {
-	nfs_mark_delegation_revoked(delegation);
+	nfs_mark_delegation_revoked(server, delegation);
 	nfs_put_delegation(delegation);
 }
 
@@ -79,14 +88,14 @@ static void nfs_mark_return_delegation(struct nfs_server *server,
 				       struct nfs_delegation *delegation)
 {
 	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+	set_bit(NFS4SERV_DELEGRETURN, &server->delegation_flags);
 	set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
 }
 
-static bool
-nfs4_is_valid_delegation(const struct nfs_delegation *delegation,
-		fmode_t flags)
+static bool nfs4_is_valid_delegation(const struct nfs_delegation *delegation,
+				     fmode_t type)
 {
-	if (delegation != NULL && (delegation->type & flags) == flags &&
+	if (delegation != NULL && (delegation->type & type) == type &&
 	    !test_bit(NFS_DELEGATION_REVOKED, &delegation->flags) &&
 	    !test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
 		return true;
@@ -103,19 +112,22 @@ struct nfs_delegation *nfs4_get_valid_delegation(const struct inode *inode)
 	return NULL;
 }
 
-static int
-nfs4_do_check_delegation(struct inode *inode, fmode_t flags, bool mark)
+static int nfs4_do_check_delegation(struct inode *inode, fmode_t type,
+				    int flags, bool mark)
 {
 	struct nfs_delegation *delegation;
 	int ret = 0;
 
-	flags &= FMODE_READ|FMODE_WRITE;
+	type &= FMODE_READ|FMODE_WRITE;
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (nfs4_is_valid_delegation(delegation, flags)) {
+	if (nfs4_is_valid_delegation(delegation, type)) {
 		if (mark)
 			nfs_mark_delegation_referenced(delegation);
 		ret = 1;
+		if ((flags & NFS_DELEGATION_FLAG_TIME) &&
+		    !test_bit(NFS_DELEGATION_DELEGTIME, &delegation->flags))
+			ret = 0;
 	}
 	rcu_read_unlock();
 	return ret;
@@ -124,22 +136,23 @@ nfs4_do_check_delegation(struct inode *inode, fmode_t flags, bool mark)
  * nfs4_have_delegation - check if inode has a delegation, mark it
  * NFS_DELEGATION_REFERENCED if there is one.
  * @inode: inode to check
- * @flags: delegation types to check for
+ * @type: delegation types to check for
+ * @flags: various modifiers
  *
  * Returns one if inode has the indicated delegation, otherwise zero.
  */
-int nfs4_have_delegation(struct inode *inode, fmode_t flags)
+int nfs4_have_delegation(struct inode *inode, fmode_t type, int flags)
 {
-	return nfs4_do_check_delegation(inode, flags, true);
+	return nfs4_do_check_delegation(inode, type, flags, true);
 }
 
 /*
  * nfs4_check_delegation - check if inode has a delegation, do not mark
  * NFS_DELEGATION_REFERENCED if it has one.
  */
-int nfs4_check_delegation(struct inode *inode, fmode_t flags)
+int nfs4_check_delegation(struct inode *inode, fmode_t type)
 {
-	return nfs4_do_check_delegation(inode, flags, false);
+	return nfs4_do_check_delegation(inode, type, 0, false);
 }
 
 static int nfs_delegation_claim_locks(struct nfs4_state *state, const nfs4_stateid *stateid)
@@ -156,8 +169,8 @@ static int nfs_delegation_claim_locks(struct nfs4_state *state, const nfs4_state
 	list = &flctx->flc_posix;
 	spin_lock(&flctx->flc_lock);
 restart:
-	list_for_each_entry(fl, list, fl_list) {
-		if (nfs_file_open_context(fl->fl_file)->state != state)
+	for_each_file_lock(fl, list) {
+		if (nfs_file_open_context(fl->c.flc_file)->state != state)
 			continue;
 		spin_unlock(&flctx->flc_lock);
 		status = nfs4_lock_delegation_recall(fl, state, stateid);
@@ -181,7 +194,6 @@ static int nfs_delegation_claim_opens(struct inode *inode,
 	struct nfs_open_context *ctx;
 	struct nfs4_state_owner *sp;
 	struct nfs4_state *state;
-	unsigned int seq;
 	int err;
 
 again:
@@ -202,12 +214,9 @@ again:
 		sp = state->owner;
 		/* Block nfs4_proc_unlck */
 		mutex_lock(&sp->so_delegreturn_mutex);
-		seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
 		err = nfs4_open_delegation_recall(ctx, state, stateid);
 		if (!err)
 			err = nfs_delegation_claim_locks(state, stateid);
-		if (!err && read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
-			err = -EAGAIN;
 		mutex_unlock(&sp->so_delegreturn_mutex);
 		put_nfs_open_context(ctx);
 		if (err != 0)
@@ -225,40 +234,51 @@ again:
  * @type: delegation type
  * @stateid: delegation stateid
  * @pagemod_limit: write delegation "space_limit"
+ * @deleg_type: raw delegation type
  *
  */
 void nfs_inode_reclaim_delegation(struct inode *inode, const struct cred *cred,
 				  fmode_t type, const nfs4_stateid *stateid,
-				  unsigned long pagemod_limit)
+				  unsigned long pagemod_limit, u32 deleg_type)
 {
 	struct nfs_delegation *delegation;
 	const struct cred *oldcred = NULL;
 
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL) {
-		spin_lock(&delegation->lock);
-		nfs4_stateid_copy(&delegation->stateid, stateid);
-		delegation->type = type;
-		delegation->pagemod_limit = pagemod_limit;
-		oldcred = delegation->cred;
-		delegation->cred = get_cred(cred);
-		clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
-		if (test_and_clear_bit(NFS_DELEGATION_REVOKED,
-				       &delegation->flags))
-			atomic_long_inc(&nfs_active_delegations);
-		spin_unlock(&delegation->lock);
-		rcu_read_unlock();
-		put_cred(oldcred);
-		trace_nfs4_reclaim_delegation(inode, type);
-	} else {
+	if (!delegation) {
 		rcu_read_unlock();
 		nfs_inode_set_delegation(inode, cred, type, stateid,
-					 pagemod_limit);
+					 pagemod_limit, deleg_type);
+		return;
 	}
+
+	spin_lock(&delegation->lock);
+	nfs4_stateid_copy(&delegation->stateid, stateid);
+	delegation->type = type;
+	delegation->pagemod_limit = pagemod_limit;
+	oldcred = delegation->cred;
+	delegation->cred = get_cred(cred);
+	switch (deleg_type) {
+	case NFS4_OPEN_DELEGATE_READ_ATTRS_DELEG:
+	case NFS4_OPEN_DELEGATE_WRITE_ATTRS_DELEG:
+		set_bit(NFS_DELEGATION_DELEGTIME, &delegation->flags);
+		break;
+	default:
+		clear_bit(NFS_DELEGATION_DELEGTIME, &delegation->flags);
+	}
+	clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
+	if (test_and_clear_bit(NFS_DELEGATION_REVOKED, &delegation->flags))
+		atomic_long_inc(&NFS_SERVER(inode)->nr_active_delegations);
+	spin_unlock(&delegation->lock);
+	rcu_read_unlock();
+	put_cred(oldcred);
+	trace_nfs4_reclaim_delegation(inode, type);
 }
 
-static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation, int issync)
+static int nfs_do_return_delegation(struct inode *inode,
+				    struct nfs_delegation *delegation,
+				    int issync)
 {
 	const struct cred *cred;
 	int res = 0;
@@ -267,9 +287,8 @@ static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *
 		spin_lock(&delegation->lock);
 		cred = get_cred(delegation->cred);
 		spin_unlock(&delegation->lock);
-		res = nfs4_proc_delegreturn(inode, cred,
-				&delegation->stateid,
-				issync);
+		res = nfs4_proc_delegreturn(inode, cred, &delegation->stateid,
+					    delegation, issync);
 		put_cred(cred);
 	}
 	return res;
@@ -297,7 +316,8 @@ nfs_start_delegation_return_locked(struct nfs_inode *nfsi)
 	if (delegation == NULL)
 		goto out;
 	spin_lock(&delegation->lock);
-	if (!test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
+	if (delegation->inode &&
+	    !test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
 		clear_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags);
 		/* Refcount matched in nfs_end_delegation_return() */
 		ret = nfs_get_delegation(delegation);
@@ -321,14 +341,16 @@ nfs_start_delegation_return(struct nfs_inode *nfsi)
 }
 
 static void nfs_abort_delegation_return(struct nfs_delegation *delegation,
-					struct nfs_client *clp, int err)
+					struct nfs_server *server, int err)
 {
-
 	spin_lock(&delegation->lock);
 	clear_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
 	if (err == -EAGAIN) {
 		set_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags);
-		set_bit(NFS4CLNT_DELEGRETURN_DELAYED, &clp->cl_state);
+		set_bit(NFS4SERV_DELEGRETURN_DELAYED,
+			&server->delegation_flags);
+		set_bit(NFS4CLNT_DELEGRETURN_DELAYED,
+			&server->nfs_client->cl_state);
 	}
 	spin_unlock(&delegation->lock);
 }
@@ -342,6 +364,8 @@ nfs_detach_delegation_locked(struct nfs_inode *nfsi,
 		rcu_dereference_protected(nfsi->delegation,
 				lockdep_is_held(&clp->cl_lock));
 
+	trace_nfs4_detach_delegation(&nfsi->vfs_inode, delegation->type);
+
 	if (deleg_cur == NULL || delegation != deleg_cur)
 		return NULL;
 
@@ -350,6 +374,7 @@ nfs_detach_delegation_locked(struct nfs_inode *nfsi,
 		spin_unlock(&delegation->lock);
 		return NULL;
 	}
+	hlist_del_init_rcu(&delegation->hash);
 	list_del_rcu(&delegation->super_list);
 	delegation->inode = NULL;
 	rcu_assign_pointer(nfsi->delegation, NULL);
@@ -397,7 +422,8 @@ nfs_update_delegation_cred(struct nfs_delegation *delegation,
 }
 
 static void
-nfs_update_inplace_delegation(struct nfs_delegation *delegation,
+nfs_update_inplace_delegation(struct nfs_server *server,
+		struct nfs_delegation *delegation,
 		const struct nfs_delegation *update)
 {
 	if (nfs4_stateid_is_newer(&update->stateid, &delegation->stateid)) {
@@ -410,7 +436,7 @@ nfs_update_inplace_delegation(struct nfs_delegation *delegation,
 			nfs_update_delegation_cred(delegation, update->cred);
 			/* smp_mb__before_atomic() is implicit due to xchg() */
 			clear_bit(NFS_DELEGATION_REVOKED, &delegation->flags);
-			atomic_long_inc(&nfs_active_delegations);
+			atomic_long_inc(&server->nr_active_delegations);
 		}
 	}
 }
@@ -422,13 +448,13 @@ nfs_update_inplace_delegation(struct nfs_delegation *delegation,
  * @type: delegation type
  * @stateid: delegation stateid
  * @pagemod_limit: write delegation "space_limit"
+ * @deleg_type: raw delegation type
  *
  * Returns zero on success, or a negative errno value.
  */
 int nfs_inode_set_delegation(struct inode *inode, const struct cred *cred,
-				  fmode_t type,
-				  const nfs4_stateid *stateid,
-				  unsigned long pagemod_limit)
+			     fmode_t type, const nfs4_stateid *stateid,
+			     unsigned long pagemod_limit, u32 deleg_type)
 {
 	struct nfs_server *server = NFS_SERVER(inode);
 	struct nfs_client *clp = server->nfs_client;
@@ -448,6 +474,12 @@ int nfs_inode_set_delegation(struct inode *inode, const struct cred *cred,
 	delegation->cred = get_cred(cred);
 	delegation->inode = inode;
 	delegation->flags = 1<<NFS_DELEGATION_REFERENCED;
+	switch (deleg_type) {
+	case NFS4_OPEN_DELEGATE_READ_ATTRS_DELEG:
+	case NFS4_OPEN_DELEGATE_WRITE_ATTRS_DELEG:
+		delegation->flags |= BIT(NFS_DELEGATION_DELEGTIME);
+	}
+	delegation->test_gen = 0;
 	spin_lock_init(&delegation->lock);
 
 	spin_lock(&clp->cl_lock);
@@ -459,7 +491,7 @@ int nfs_inode_set_delegation(struct inode *inode, const struct cred *cred,
 	if (nfs4_stateid_match_other(&old_delegation->stateid,
 				&delegation->stateid)) {
 		spin_lock(&old_delegation->lock);
-		nfs_update_inplace_delegation(old_delegation,
+		nfs_update_inplace_delegation(server, old_delegation,
 				delegation);
 		spin_unlock(&old_delegation->lock);
 		goto out;
@@ -505,19 +537,26 @@ add_new:
 	spin_unlock(&inode->i_lock);
 
 	list_add_tail_rcu(&delegation->super_list, &server->delegations);
+	hlist_add_head_rcu(&delegation->hash,
+			nfs_delegation_hash(server, &NFS_I(inode)->fh));
 	rcu_assign_pointer(nfsi->delegation, delegation);
 	delegation = NULL;
 
-	atomic_long_inc(&nfs_active_delegations);
+	atomic_long_inc(&server->nr_active_delegations);
 
 	trace_nfs4_set_delegation(inode, type);
+
+	/* If we hold writebacks and have delegated mtime then update */
+	if (deleg_type == NFS4_OPEN_DELEGATE_WRITE_ATTRS_DELEG &&
+	    nfs_have_writebacks(inode))
+		nfs_update_delegated_mtime(inode);
 out:
 	spin_unlock(&clp->cl_lock);
 	if (delegation != NULL)
 		__nfs_free_delegation(delegation);
 	if (freeme != NULL) {
 		nfs_do_return_delegation(inode, freeme, 0);
-		nfs_free_delegation(freeme);
+		nfs_free_delegation(server, freeme);
 	}
 	return status;
 }
@@ -527,7 +566,7 @@ out:
  */
 static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation *delegation, int issync)
 {
-	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	struct nfs_server *server = NFS_SERVER(inode);
 	unsigned int mode = O_WRONLY | O_RDWR;
 	int err = 0;
 
@@ -549,11 +588,11 @@ static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation
 		/*
 		 * Guard against state recovery
 		 */
-		err = nfs4_wait_clnt_recover(clp);
+		err = nfs4_wait_clnt_recover(server->nfs_client);
 	}
 
 	if (err) {
-		nfs_abort_delegation_return(delegation, clp, err);
+		nfs_abort_delegation_return(delegation, server, err);
 		goto out;
 	}
 
@@ -568,19 +607,10 @@ static bool nfs_delegation_need_return(struct nfs_delegation *delegation)
 {
 	bool ret = false;
 
+	trace_nfs_delegation_need_return(delegation);
+
 	if (test_and_clear_bit(NFS_DELEGATION_RETURN, &delegation->flags))
 		ret = true;
-	else if (test_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags)) {
-		struct inode *inode;
-
-		spin_lock(&delegation->lock);
-		inode = delegation->inode;
-		if (inode && list_empty(&NFS_I(inode)->open_files))
-			ret = true;
-		spin_unlock(&delegation->lock);
-	}
-	if (ret)
-		clear_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
 	if (test_bit(NFS_DELEGATION_RETURNING, &delegation->flags) ||
 	    test_bit(NFS_DELEGATION_RETURN_DELAYED, &delegation->flags) ||
 	    test_bit(NFS_DELEGATION_REVOKED, &delegation->flags))
@@ -599,6 +629,9 @@ static int nfs_server_return_marked_delegations(struct nfs_server *server,
 	struct nfs_delegation *place_holder_deleg = NULL;
 	int err = 0;
 
+	if (!test_and_clear_bit(NFS4SERV_DELEGRETURN,
+				&server->delegation_flags))
+		return 0;
 restart:
 	/*
 	 * To avoid quadratic looping we hold a reference
@@ -627,6 +660,9 @@ restart:
 				prev = delegation;
 			continue;
 		}
+		inode = nfs_delegation_grab_inode(delegation);
+		if (inode == NULL)
+			continue;
 
 		if (prev) {
 			struct inode *tmp = nfs_delegation_grab_inode(prev);
@@ -637,12 +673,6 @@ restart:
 			}
 		}
 
-		inode = nfs_delegation_grab_inode(delegation);
-		if (inode == NULL) {
-			rcu_read_unlock();
-			iput(to_put);
-			goto restart;
-		}
 		delegation = nfs_start_delegation_return_locked(NFS_I(inode));
 		rcu_read_unlock();
 
@@ -653,6 +683,7 @@ restart:
 		cond_resched();
 		if (!err)
 			goto restart;
+		set_bit(NFS4SERV_DELEGRETURN, &server->delegation_flags);
 		set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
 		goto out;
 	}
@@ -667,6 +698,9 @@ static bool nfs_server_clear_delayed_delegations(struct nfs_server *server)
 	struct nfs_delegation *d;
 	bool ret = false;
 
+	if (!test_and_clear_bit(NFS4SERV_DELEGRETURN_DELAYED,
+				&server->delegation_flags))
+		goto out;
 	list_for_each_entry_rcu (d, &server->delegations, super_list) {
 		if (!test_bit(NFS_DELEGATION_RETURN_DELAYED, &d->flags))
 			continue;
@@ -674,6 +708,7 @@ static bool nfs_server_clear_delayed_delegations(struct nfs_server *server)
 		clear_bit(NFS_DELEGATION_RETURN_DELAYED, &d->flags);
 		ret = true;
 	}
+out:
 	return ret;
 }
 
@@ -733,7 +768,7 @@ void nfs_inode_evict_delegation(struct inode *inode)
 		set_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
 		set_bit(NFS_DELEGATION_INODE_FREEING, &delegation->flags);
 		nfs_do_return_delegation(inode, delegation, 1);
-		nfs_free_delegation(delegation);
+		nfs_free_delegation(NFS_SERVER(inode), delegation);
 	}
 }
 
@@ -764,6 +799,43 @@ int nfs4_inode_return_delegation(struct inode *inode)
 }
 
 /**
+ * nfs4_inode_set_return_delegation_on_close - asynchronously return a delegation
+ * @inode: inode to process
+ *
+ * This routine is called to request that the delegation be returned as soon
+ * as the file is closed. If the file is already closed, the delegation is
+ * immediately returned.
+ */
+void nfs4_inode_set_return_delegation_on_close(struct inode *inode)
+{
+	struct nfs_delegation *delegation;
+	struct nfs_delegation *ret = NULL;
+
+	if (!inode)
+		return;
+	rcu_read_lock();
+	delegation = nfs4_get_valid_delegation(inode);
+	if (!delegation)
+		goto out;
+	spin_lock(&delegation->lock);
+	if (!delegation->inode)
+		goto out_unlock;
+	if (list_empty(&NFS_I(inode)->open_files) &&
+	    !test_and_set_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
+		/* Refcount matched in nfs_end_delegation_return() */
+		ret = nfs_get_delegation(delegation);
+	} else
+		set_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
+out_unlock:
+	spin_unlock(&delegation->lock);
+	if (ret)
+		nfs_clear_verifier_delegated(inode);
+out:
+	rcu_read_unlock();
+	nfs_end_delegation_return(inode, ret, 0);
+}
+
+/**
  * nfs4_inode_return_delegation_on_close - asynchronously return a delegation
  * @inode: inode to process
  *
@@ -782,7 +854,8 @@ void nfs4_inode_return_delegation_on_close(struct inode *inode)
 	if (!delegation)
 		goto out;
 	if (test_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags) ||
-	    atomic_long_read(&nfs_active_delegations) >= nfs_delegation_watermark) {
+	    atomic_long_read(&NFS_SERVER(inode)->nr_active_delegations) >=
+	    nfs_delegation_watermark) {
 		spin_lock(&delegation->lock);
 		if (delegation->inode &&
 		    list_empty(&NFS_I(inode)->open_files) &&
@@ -824,11 +897,25 @@ int nfs4_inode_make_writeable(struct inode *inode)
 	return nfs4_inode_return_delegation(inode);
 }
 
-static void nfs_mark_return_if_closed_delegation(struct nfs_server *server,
-		struct nfs_delegation *delegation)
+static void
+nfs_mark_return_if_closed_delegation(struct nfs_server *server,
+				     struct nfs_delegation *delegation)
 {
-	set_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
-	set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state);
+	struct inode *inode;
+
+	if (test_bit(NFS_DELEGATION_RETURN, &delegation->flags) ||
+	    test_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags))
+		return;
+	spin_lock(&delegation->lock);
+	inode = delegation->inode;
+	if (!inode)
+		goto out;
+	if (list_empty(&NFS_I(inode)->open_files))
+		nfs_mark_return_delegation(server, delegation);
+	else
+		set_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags);
+out:
+	spin_unlock(&delegation->lock);
 }
 
 static bool nfs_server_mark_return_all_delegations(struct nfs_server *server)
@@ -944,7 +1031,7 @@ static void nfs_revoke_delegation(struct inode *inode,
 		}
 		spin_unlock(&delegation->lock);
 	}
-	nfs_mark_delegation_revoked(delegation);
+	nfs_mark_delegation_revoked(NFS_SERVER(inode), delegation);
 	ret = true;
 out:
 	rcu_read_unlock();
@@ -952,13 +1039,6 @@ out:
 		nfs_inode_find_state_and_recover(inode, stateid);
 }
 
-void nfs_remove_bad_delegation(struct inode *inode,
-		const nfs4_stateid *stateid)
-{
-	nfs_revoke_delegation(inode, stateid);
-}
-EXPORT_SYMBOL_GPL(nfs_remove_bad_delegation);
-
 void nfs_delegation_mark_returned(struct inode *inode,
 		const nfs4_stateid *stateid)
 {
@@ -983,7 +1063,12 @@ void nfs_delegation_mark_returned(struct inode *inode,
 			delegation->stateid.seqid = stateid->seqid;
 	}
 
-	nfs_mark_delegation_revoked(delegation);
+	nfs_mark_delegation_revoked(NFS_SERVER(inode), delegation);
+	clear_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
+	spin_unlock(&delegation->lock);
+	if (nfs_detach_delegation(NFS_I(inode), delegation, NFS_SERVER(inode)))
+		nfs_put_delegation(delegation);
+	goto out_rcu_unlock;
 
 out_clear_returning:
 	clear_bit(NFS_DELEGATION_RETURNING, &delegation->flags);
@@ -996,6 +1081,24 @@ out_rcu_unlock:
 }
 
 /**
+ * nfs_remove_bad_delegation - handle delegations that are unusable
+ * @inode: inode to process
+ * @stateid: the delegation's stateid
+ *
+ * If the server ACK-ed our FREE_STATEID then clean
+ * up the delegation, else mark and keep the revoked state.
+ */
+void nfs_remove_bad_delegation(struct inode *inode,
+		const nfs4_stateid *stateid)
+{
+	if (stateid && stateid->type == NFS4_FREED_STATEID_TYPE)
+		nfs_delegation_mark_returned(inode, stateid);
+	else
+		nfs_revoke_delegation(inode, stateid);
+}
+EXPORT_SYMBOL_GPL(nfs_remove_bad_delegation);
+
+/**
  * nfs_expire_unused_delegation_types
  * @clp: client to process
  * @flags: delegation types to expire
@@ -1073,11 +1176,12 @@ static struct inode *
 nfs_delegation_find_inode_server(struct nfs_server *server,
 				 const struct nfs_fh *fhandle)
 {
+	struct hlist_head *head = nfs_delegation_hash(server, fhandle);
 	struct nfs_delegation *delegation;
 	struct super_block *freeme = NULL;
 	struct inode *res = NULL;
 
-	list_for_each_entry_rcu(delegation, &server->delegations, super_list) {
+	hlist_for_each_entry_rcu(delegation, head, hash) {
 		spin_lock(&delegation->lock);
 		if (delegation->inode != NULL &&
 		    !test_bit(NFS_DELEGATION_REVOKED, &delegation->flags) &&
@@ -1164,7 +1268,6 @@ static int nfs_server_reap_unclaimed_delegations(struct nfs_server *server,
 	struct inode *inode;
 restart:
 	rcu_read_lock();
-restart_locked:
 	list_for_each_entry_rcu(delegation, &server->delegations, super_list) {
 		if (test_bit(NFS_DELEGATION_INODE_FREEING,
 					&delegation->flags) ||
@@ -1175,13 +1278,13 @@ restart_locked:
 			continue;
 		inode = nfs_delegation_grab_inode(delegation);
 		if (inode == NULL)
-			goto restart_locked;
+			continue;
 		delegation = nfs_start_delegation_return_locked(NFS_I(inode));
 		rcu_read_unlock();
 		if (delegation != NULL) {
 			if (nfs_detach_delegation(NFS_I(inode), delegation,
 						server) != NULL)
-				nfs_free_delegation(delegation);
+				nfs_free_delegation(server, delegation);
 			/* Match nfs_start_delegation_return_locked */
 			nfs_put_delegation(delegation);
 		}
@@ -1218,6 +1321,7 @@ static void nfs_mark_test_expired_delegation(struct nfs_server *server,
 		return;
 	clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
 	set_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags);
+	set_bit(NFS4SERV_DELEGATION_EXPIRED, &server->delegation_flags);
 	set_bit(NFS4CLNT_DELEGATION_EXPIRED, &server->nfs_client->cl_state);
 }
 
@@ -1294,24 +1398,30 @@ static int nfs_server_reap_expired_delegations(struct nfs_server *server,
 	struct inode *inode;
 	const struct cred *cred;
 	nfs4_stateid stateid;
+	unsigned long gen = ++server->delegation_gen;
+
+	if (!test_and_clear_bit(NFS4SERV_DELEGATION_EXPIRED,
+				&server->delegation_flags))
+		return 0;
 restart:
 	rcu_read_lock();
-restart_locked:
 	list_for_each_entry_rcu(delegation, &server->delegations, super_list) {
 		if (test_bit(NFS_DELEGATION_INODE_FREEING,
 					&delegation->flags) ||
 		    test_bit(NFS_DELEGATION_RETURNING,
 					&delegation->flags) ||
 		    test_bit(NFS_DELEGATION_TEST_EXPIRED,
-					&delegation->flags) == 0)
+					&delegation->flags) == 0 ||
+			delegation->test_gen == gen)
 			continue;
 		inode = nfs_delegation_grab_inode(delegation);
 		if (inode == NULL)
-			goto restart_locked;
+			continue;
 		spin_lock(&delegation->lock);
 		cred = get_cred_rcu(delegation->cred);
 		nfs4_stateid_copy(&stateid, &delegation->stateid);
 		spin_unlock(&delegation->lock);
+		delegation->test_gen = gen;
 		clear_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags);
 		rcu_read_unlock();
 		nfs_delegation_test_free_expired(inode, &stateid, cred);
@@ -1322,6 +1432,9 @@ restart_locked:
 			goto restart;
 		}
 		nfs_inode_mark_test_expired_delegation(server,inode);
+		set_bit(NFS4SERV_DELEGATION_EXPIRED, &server->delegation_flags);
+		set_bit(NFS4CLNT_DELEGATION_EXPIRED,
+			&server->nfs_client->cl_state);
 		iput(inode);
 		return -EAGAIN;
 	}
@@ -1476,4 +1589,17 @@ out:
 	return ret;
 }
 
-module_param_named(delegation_watermark, nfs_delegation_watermark, uint, 0644);
+int nfs4_delegation_hash_alloc(struct nfs_server *server)
+{
+	int delegation_buckets, i;
+
+	delegation_buckets = roundup_pow_of_two(nfs_delegation_watermark / 16);
+	server->delegation_hash_mask = delegation_buckets - 1;
+	server->delegation_hash_table = kmalloc_array(delegation_buckets,
+			sizeof(*server->delegation_hash_table), GFP_KERNEL);
+	if (!server->delegation_hash_table)
+		return -ENOMEM;
+	for (i = 0; i < delegation_buckets; i++)
+		INIT_HLIST_HEAD(&server->delegation_hash_table[i]);
+	return 0;
+}