1 files changed, 361 insertions, 146 deletions

diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index fb9b1656a287..b9548eb17c77 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -55,6 +55,17 @@
 /* We only care about NFSD_MAY_READ/WRITE for this cache */
 #define NFSD_FILE_MAY_MASK	(NFSD_MAY_READ|NFSD_MAY_WRITE|NFSD_MAY_LOCALIO)
 
+/* If the shrinker runs between calls to list_lru_walk_node() in
+ * nfsd_file_gc(), the "remaining" count will be wrong.  This could
+ * result in premature freeing of some files.  This may not matter much
+ * but is easy to fix with this spinlock which temporarily disables
+ * the shrinker.
+ *
+ * It also serializes callers of nfsd_file_dispose_list_delayed()
+ * against per-net shutdown.
+ */
+static DEFINE_SPINLOCK(nfsd_gc_lock);
+
 static DEFINE_PER_CPU(unsigned long, nfsd_file_cache_hits);
 static DEFINE_PER_CPU(unsigned long, nfsd_file_acquisitions);
 static DEFINE_PER_CPU(unsigned long, nfsd_file_allocations);
@@ -62,16 +73,12 @@ static DEFINE_PER_CPU(unsigned long, nfsd_file_releases);
 static DEFINE_PER_CPU(unsigned long, nfsd_file_total_age);
 static DEFINE_PER_CPU(unsigned long, nfsd_file_evictions);
 
-struct nfsd_fcache_disposal {
-	spinlock_t lock;
-	struct list_head freeme;
-};
-
 static struct kmem_cache		*nfsd_file_slab;
 static struct kmem_cache		*nfsd_file_mark_slab;
 static struct list_lru			nfsd_file_lru;
 static unsigned long			nfsd_file_flags;
 static struct fsnotify_group		*nfsd_file_fsnotify_group;
+static struct fsnotify_group		*nfsd_dir_fsnotify_group;
 static struct delayed_work		nfsd_filecache_laundrette;
 static struct rhltable			nfsd_file_rhltable
 						____cacheline_aligned_in_smp;
@@ -113,7 +120,7 @@ static void
 nfsd_file_schedule_laundrette(void)
 {
 	if (test_bit(NFSD_FILE_CACHE_UP, &nfsd_file_flags))
-		queue_delayed_work(system_unbound_wq, &nfsd_filecache_laundrette,
+		queue_delayed_work(system_dfl_wq, &nfsd_filecache_laundrette,
 				   NFSD_LAUNDRETTE_DELAY);
 }
 
@@ -147,7 +154,7 @@ static void
 nfsd_file_mark_put(struct nfsd_file_mark *nfm)
 {
 	if (refcount_dec_and_test(&nfm->nfm_ref)) {
-		fsnotify_destroy_mark(&nfm->nfm_mark, nfsd_file_fsnotify_group);
+		fsnotify_destroy_mark(&nfm->nfm_mark, nfm->nfm_mark.group);
 		fsnotify_put_mark(&nfm->nfm_mark);
 	}
 }
@@ -155,37 +162,40 @@ nfsd_file_mark_put(struct nfsd_file_mark *nfm)
 static struct nfsd_file_mark *
 nfsd_file_mark_find_or_create(struct inode *inode)
 {
-	int			err;
-	struct fsnotify_mark	*mark;
 	struct nfsd_file_mark	*nfm = NULL, *new;
+	struct fsnotify_group	*group;
+	struct fsnotify_mark	*mark;
+	int			err;
+
+	group = S_ISDIR(inode->i_mode) ? nfsd_dir_fsnotify_group : nfsd_file_fsnotify_group;
 
 	do {
-		fsnotify_group_lock(nfsd_file_fsnotify_group);
-		mark = fsnotify_find_inode_mark(inode,
-						nfsd_file_fsnotify_group);
+		fsnotify_group_lock(group);
+		mark = fsnotify_find_inode_mark(inode, group);
 		if (mark) {
 			nfm = nfsd_file_mark_get(container_of(mark,
 						 struct nfsd_file_mark,
 						 nfm_mark));
-			fsnotify_group_unlock(nfsd_file_fsnotify_group);
+			fsnotify_group_unlock(group);
 			if (nfm) {
 				fsnotify_put_mark(mark);
 				break;
 			}
 			/* Avoid soft lockup race with nfsd_file_mark_put() */
-			fsnotify_destroy_mark(mark, nfsd_file_fsnotify_group);
+			fsnotify_destroy_mark(mark, group);
 			fsnotify_put_mark(mark);
 		} else {
-			fsnotify_group_unlock(nfsd_file_fsnotify_group);
+			fsnotify_group_unlock(group);
 		}
 
 		/* allocate a new nfm */
 		new = kmem_cache_alloc(nfsd_file_mark_slab, GFP_KERNEL);
 		if (!new)
 			return NULL;
-		fsnotify_init_mark(&new->nfm_mark, nfsd_file_fsnotify_group);
+		fsnotify_init_mark(&new->nfm_mark, group);
 		new->nfm_mark.mask = FS_ATTRIB|FS_DELETE_SELF;
 		refcount_set(&new->nfm_ref, 1);
+		mutex_init(&new->nfm_recalc_mutex);
 
 		err = fsnotify_add_inode_mark(&new->nfm_mark, inode, 0);
 
@@ -231,6 +241,9 @@ nfsd_file_alloc(struct net *net, struct inode *inode, unsigned char need,
 	refcount_set(&nf->nf_ref, 1);
 	nf->nf_may = need;
 	nf->nf_mark = NULL;
+	nf->nf_dio_mem_align = 0;
+	nf->nf_dio_offset_align = 0;
+	nf->nf_dio_read_offset_align = 0;
 	return nf;
 }
 
@@ -319,15 +332,17 @@ nfsd_file_check_writeback(struct nfsd_file *nf)
 		mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK);
 }
 
-
-static bool nfsd_file_lru_add(struct nfsd_file *nf)
+static void nfsd_file_lru_add(struct nfsd_file *nf)
 {
-	set_bit(NFSD_FILE_REFERENCED, &nf->nf_flags);
-	if (list_lru_add_obj(&nfsd_file_lru, &nf->nf_lru)) {
+	refcount_inc(&nf->nf_ref);
+	if (list_lru_add_obj(&nfsd_file_lru, &nf->nf_lru))
 		trace_nfsd_file_lru_add(nf);
-		return true;
+	else {
+		refcount_dec(&nf->nf_ref);
+		WARN_ON_ONCE(1);
+		return;
 	}
-	return false;
+	nfsd_file_schedule_laundrette();
 }
 
 static bool nfsd_file_lru_remove(struct nfsd_file *nf)
@@ -363,47 +378,32 @@ nfsd_file_put(struct nfsd_file *nf)
 
 	if (test_bit(NFSD_FILE_GC, &nf->nf_flags) &&
 	    test_bit(NFSD_FILE_HASHED, &nf->nf_flags)) {
-		/*
-		 * If this is the last reference (nf_ref == 1), then try to
-		 * transfer it to the LRU.
-		 */
-		if (refcount_dec_not_one(&nf->nf_ref))
-			return;
-
-		/* Try to add it to the LRU.  If that fails, decrement. */
-		if (nfsd_file_lru_add(nf)) {
-			/* If it's still hashed, we're done */
-			if (test_bit(NFSD_FILE_HASHED, &nf->nf_flags)) {
-				nfsd_file_schedule_laundrette();
-				return;
-			}
-
-			/*
-			 * We're racing with unhashing, so try to remove it from
-			 * the LRU. If removal fails, then someone else already
-			 * has our reference.
-			 */
-			if (!nfsd_file_lru_remove(nf))
-				return;
-		}
+		set_bit(NFSD_FILE_REFERENCED, &nf->nf_flags);
+		set_bit(NFSD_FILE_RECENT, &nf->nf_flags);
 	}
+
 	if (refcount_dec_and_test(&nf->nf_ref))
 		nfsd_file_free(nf);
 }
 
 /**
  * nfsd_file_put_local - put nfsd_file reference and arm nfsd_net_put in caller
- * @nf: nfsd_file of which to put the reference
+ * @pnf: nfsd_file of which to put the reference
  *
  * First save the associated net to return to caller, then put
  * the reference of the nfsd_file.
  */
 struct net *
-nfsd_file_put_local(struct nfsd_file *nf)
+nfsd_file_put_local(struct nfsd_file __rcu **pnf)
 {
-	struct net *net = nf->nf_net;
+	struct nfsd_file *nf;
+	struct net *net = NULL;
 
-	nfsd_file_put(nf);
+	nf = unrcu_pointer(xchg(pnf, NULL));
+	if (nf) {
+		net = nf->nf_net;
+		nfsd_file_put(nf);
+	}
 	return net;
 }
 
@@ -432,25 +432,31 @@ nfsd_file_dispose_list(struct list_head *dispose)
 }
 
 /**
- * nfsd_file_dispose_list_delayed - move list of dead files to net's freeme list
+ * nfsd_file_dispose_list_delayed - queue dead files for nfsd thread disposal
  * @dispose: list of nfsd_files to be disposed
  *
- * Transfers each file to the "freeme" list for its nfsd_net, to eventually
- * be disposed of by the per-net garbage collector.
+ * Transfers each file to the dispose list in its nfsd_net and wakes an nfsd
+ * thread to do the actual close.  This keeps the cost of fput() in the nfsd
+ * threads rather than in the shrinker or GC worker.
+ *
+ * All callers must hold nfsd_gc_lock, so that nfsd_file_cache_shutdown_net()
+ * can synchronize against them before draining the per-net dispose list.
+ * This guarantees nf_net is still live when we call net_generic().
  */
 static void
 nfsd_file_dispose_list_delayed(struct list_head *dispose)
 {
-	while(!list_empty(dispose)) {
+	lockdep_assert_held(&nfsd_gc_lock);
+
+	while (!list_empty(dispose)) {
 		struct nfsd_file *nf = list_first_entry(dispose,
 						struct nfsd_file, nf_gc);
 		struct nfsd_net *nn = net_generic(nf->nf_net, nfsd_net_id);
-		struct nfsd_fcache_disposal *l = nn->fcache_disposal;
 		struct svc_serv *serv;
 
-		spin_lock(&l->lock);
-		list_move_tail(&nf->nf_gc, &l->freeme);
-		spin_unlock(&l->lock);
+		spin_lock(&nn->fcache_dispose_lock);
+		list_move_tail(&nf->nf_gc, &nn->fcache_dispose_list);
+		spin_unlock(&nn->fcache_dispose_lock);
 
 		/*
 		 * The filecache laundrette is shut down after the
@@ -474,21 +480,28 @@ nfsd_file_dispose_list_delayed(struct list_head *dispose)
  */
 void nfsd_file_net_dispose(struct nfsd_net *nn)
 {
-	struct nfsd_fcache_disposal *l = nn->fcache_disposal;
-
-	if (!list_empty(&l->freeme)) {
+	if (!list_empty(&nn->fcache_dispose_list)) {
 		LIST_HEAD(dispose);
 		int i;
 
-		spin_lock(&l->lock);
-		for (i = 0; i < 8 && !list_empty(&l->freeme); i++)
-			list_move(l->freeme.next, &dispose);
-		spin_unlock(&l->lock);
-		if (!list_empty(&l->freeme))
-			/* Wake up another thread to share the work
+		spin_lock(&nn->fcache_dispose_lock);
+		for (i = 0; i < 8 && !list_empty(&nn->fcache_dispose_list); i++)
+			list_move(nn->fcache_dispose_list.next, &dispose);
+		spin_unlock(&nn->fcache_dispose_lock);
+		if (!list_empty(&nn->fcache_dispose_list)) {
+			/*
+			 * Wake up another thread to share the work
 			 * *before* doing any actual disposing.
+			 *
+			 * The filecache laundrette is shut down after
+			 * the nn->nfsd_serv pointer is cleared, but
+			 * before the svc_serv is freed.
 			 */
-			svc_wake_up(nn->nfsd_serv);
+			struct svc_serv *serv = nn->nfsd_serv;
+
+			if (serv)
+				svc_wake_up(serv);
+		}
 		nfsd_file_dispose_list(&dispose);
 	}
 }
@@ -530,13 +543,12 @@ nfsd_file_lru_cb(struct list_head *item, struct list_lru_one *lru,
 	}
 
 	/*
-	 * Put the reference held on behalf of the LRU. If it wasn't the last
-	 * one, then just remove it from the LRU and ignore it.
+	 * Put the reference held on behalf of the LRU if it is the last
+	 * reference, else rotate.
 	 */
-	if (!refcount_dec_and_test(&nf->nf_ref)) {
+	if (!refcount_dec_if_one(&nf->nf_ref)) {
 		trace_nfsd_file_gc_in_use(nf);
-		list_lru_isolate(lru, &nf->nf_lru);
-		return LRU_REMOVED;
+		return LRU_ROTATE;
 	}
 
 	/* Refcount went to zero. Unhash it and queue it to the dispose list */
@@ -548,24 +560,57 @@ nfsd_file_lru_cb(struct list_head *item, struct list_lru_one *lru,
 	return LRU_REMOVED;
 }
 
+static enum lru_status
+nfsd_file_gc_cb(struct list_head *item, struct list_lru_one *lru,
+		 void *arg)
+{
+	struct nfsd_file *nf = list_entry(item, struct nfsd_file, nf_lru);
+
+	if (test_and_clear_bit(NFSD_FILE_RECENT, &nf->nf_flags)) {
+		/*
+		 * "REFERENCED" really means "should be at the end of the
+		 * LRU. As we are putting it there we can clear the flag.
+		 */
+		clear_bit(NFSD_FILE_REFERENCED, &nf->nf_flags);
+		trace_nfsd_file_gc_aged(nf);
+		return LRU_ROTATE;
+	}
+	return nfsd_file_lru_cb(item, lru, arg);
+}
+
 static void
 nfsd_file_gc(void)
 {
+	unsigned long ret = 0;
 	LIST_HEAD(dispose);
-	unsigned long ret;
+	int nid;
 
-	ret = list_lru_walk(&nfsd_file_lru, nfsd_file_lru_cb,
-			    &dispose, list_lru_count(&nfsd_file_lru));
-	trace_nfsd_file_gc_removed(ret, list_lru_count(&nfsd_file_lru));
+	spin_lock(&nfsd_gc_lock);
+	for_each_node_state(nid, N_NORMAL_MEMORY) {
+		unsigned long remaining = list_lru_count_node(&nfsd_file_lru, nid);
+
+		while (remaining > 0) {
+			unsigned long nr = min(remaining, NFSD_FILE_GC_BATCH);
+
+			remaining -= nr;
+			ret += list_lru_walk_node(&nfsd_file_lru, nid, nfsd_file_gc_cb,
+						  &dispose, &nr);
+			if (nr)
+				/* walk aborted early */
+				remaining = 0;
+		}
+	}
 	nfsd_file_dispose_list_delayed(&dispose);
+	spin_unlock(&nfsd_gc_lock);
+	trace_nfsd_file_gc_removed(ret, list_lru_count(&nfsd_file_lru));
 }
 
 static void
 nfsd_file_gc_worker(struct work_struct *work)
 {
-	nfsd_file_gc();
 	if (list_lru_count(&nfsd_file_lru))
-		nfsd_file_schedule_laundrette();
+		nfsd_file_gc();
+	nfsd_file_schedule_laundrette();
 }
 
 static unsigned long
@@ -580,10 +625,14 @@ nfsd_file_lru_scan(struct shrinker *s, struct shrink_control *sc)
 	LIST_HEAD(dispose);
 	unsigned long ret;
 
+	if (!spin_trylock(&nfsd_gc_lock))
+		return SHRINK_STOP;
+
 	ret = list_lru_shrink_walk(&nfsd_file_lru, sc,
 				   nfsd_file_lru_cb, &dispose);
-	trace_nfsd_file_shrinker_removed(ret, list_lru_count(&nfsd_file_lru));
 	nfsd_file_dispose_list_delayed(&dispose);
+	spin_unlock(&nfsd_gc_lock);
+	trace_nfsd_file_shrinker_removed(ret, list_lru_count(&nfsd_file_lru));
 	return ret;
 }
 
@@ -656,11 +705,11 @@ nfsd_file_queue_for_close(struct inode *inode, struct list_head *dispose)
 }
 
 /**
- * nfsd_file_close_inode - attempt a delayed close of a nfsd_file
+ * nfsd_file_close_inode - attempt a deferred close of a nfsd_file
  * @inode: inode of the file to attempt to remove
  *
  * Close out any open nfsd_files that can be reaped for @inode. The
- * actual freeing is deferred to the dispose_list_delayed infrastructure.
+ * actual freeing is deferred to the nfsd service threads.
  *
  * This is used by the fsnotify callbacks and setlease notifier.
  */
@@ -669,8 +718,10 @@ nfsd_file_close_inode(struct inode *inode)
 {
 	LIST_HEAD(dispose);
 
+	spin_lock(&nfsd_gc_lock);
 	nfsd_file_queue_for_close(inode, &dispose);
 	nfsd_file_dispose_list_delayed(&dispose);
+	spin_unlock(&nfsd_gc_lock);
 }
 
 /**
@@ -686,17 +737,58 @@ nfsd_file_close_inode(struct inode *inode)
 void
 nfsd_file_close_inode_sync(struct inode *inode)
 {
-	struct nfsd_file *nf;
 	LIST_HEAD(dispose);
 
 	trace_nfsd_file_close(inode);
 
 	nfsd_file_queue_for_close(inode, &dispose);
-	while (!list_empty(&dispose)) {
-		nf = list_first_entry(&dispose, struct nfsd_file, nf_gc);
-		list_del_init(&nf->nf_gc);
-		nfsd_file_free(nf);
-	}
+	nfsd_file_dispose_list(&dispose);
+}
+
+/**
+ * nfsd_file_close_export - close cached file handles for an export
+ * @net: net namespace in which to operate
+ * @path: export path whose cached files should be closed
+ *
+ * Close out GC-managed nfsd_file entries whose underlying file is on
+ * the same filesystem as, and a descendant of, @path.  nfsd_file
+ * entries do not carry an export reference, so the check uses the
+ * file's dentry ancestry.  False positives (closing a cached handle
+ * that did not originate from the target export) are harmless -- the
+ * handle is simply reopened on the next access.
+ *
+ * Called from the NFSD_CMD_UNLOCK_EXPORT handler before revoking
+ * NFSv4 state, to ensure that cached file handles do not hold the
+ * filesystem busy.
+ */
+void nfsd_file_close_export(struct net *net, const struct path *path)
+{
+	struct rhashtable_iter iter;
+	struct nfsd_file *nf;
+	LIST_HEAD(dispose);
+
+	rhltable_walk_enter(&nfsd_file_rhltable, &iter);
+	do {
+		rhashtable_walk_start(&iter);
+
+		nf = rhashtable_walk_next(&iter);
+		while (!IS_ERR_OR_NULL(nf)) {
+			if (nf->nf_net == net &&
+			    test_bit(NFSD_FILE_GC, &nf->nf_flags) &&
+			    nf->nf_file &&
+			    file_inode(nf->nf_file)->i_sb ==
+					path->dentry->d_sb &&
+			    is_subdir(nf->nf_file->f_path.dentry,
+				      path->dentry))
+				nfsd_file_cond_queue(nf, &dispose);
+			nf = rhashtable_walk_next(&iter);
+		}
+
+		rhashtable_walk_stop(&iter);
+	} while (nf == ERR_PTR(-EAGAIN));
+	rhashtable_walk_exit(&iter);
+
+	nfsd_file_dispose_list(&dispose);
 }
 
 static int
@@ -741,12 +833,36 @@ nfsd_file_fsnotify_handle_event(struct fsnotify_mark *mark, u32 mask,
 	return 0;
 }
 
+#ifdef CONFIG_NFSD_V4
+static int
+nfsd_dir_fsnotify_handle_event(struct fsnotify_group *group, u32 mask,
+			       const void *data, int data_type, struct inode *dir,
+			       const struct qstr *name, u32 cookie,
+			       struct fsnotify_iter_info *iter_info)
+{
+	return nfsd_handle_dir_event(mask, dir, data, data_type, name);
+}
+#else
+static int
+nfsd_dir_fsnotify_handle_event(struct fsnotify_group *group, u32 mask,
+			       const void *data, int data_type, struct inode *dir,
+			       const struct qstr *name, u32 cookie,
+			       struct fsnotify_iter_info *iter_info)
+{
+	return 0;
+}
+#endif
 
 static const struct fsnotify_ops nfsd_file_fsnotify_ops = {
 	.handle_inode_event = nfsd_file_fsnotify_handle_event,
 	.free_mark = nfsd_file_mark_free,
 };
 
+static const struct fsnotify_ops nfsd_dir_fsnotify_ops = {
+	.handle_event = nfsd_dir_fsnotify_handle_event,
+	.free_mark = nfsd_file_mark_free,
+};
+
 int
 nfsd_file_cache_init(void)
 {
@@ -798,8 +914,7 @@ nfsd_file_cache_init(void)
 		goto out_shrinker;
 	}
 
-	nfsd_file_fsnotify_group = fsnotify_alloc_group(&nfsd_file_fsnotify_ops,
-							0);
+	nfsd_file_fsnotify_group = fsnotify_alloc_group(&nfsd_file_fsnotify_ops, 0);
 	if (IS_ERR(nfsd_file_fsnotify_group)) {
 		pr_err("nfsd: unable to create fsnotify group: %ld\n",
 			PTR_ERR(nfsd_file_fsnotify_group));
@@ -808,11 +923,23 @@ nfsd_file_cache_init(void)
 		goto out_notifier;
 	}
 
+	nfsd_dir_fsnotify_group = fsnotify_alloc_group(&nfsd_dir_fsnotify_ops, 0);
+	if (IS_ERR(nfsd_dir_fsnotify_group)) {
+		pr_err("nfsd: unable to create fsnotify group: %ld\n",
+			PTR_ERR(nfsd_dir_fsnotify_group));
+		ret = PTR_ERR(nfsd_dir_fsnotify_group);
+		nfsd_dir_fsnotify_group = NULL;
+		goto out_notify_group;
+	}
+
 	INIT_DELAYED_WORK(&nfsd_filecache_laundrette, nfsd_file_gc_worker);
 out:
 	if (ret)
 		clear_bit(NFSD_FILE_CACHE_UP, &nfsd_file_flags);
 	return ret;
+out_notify_group:
+	fsnotify_put_group(nfsd_file_fsnotify_group);
+	nfsd_file_fsnotify_group = NULL;
 out_notifier:
 	lease_unregister_notifier(&nfsd_file_lease_notifier);
 out_shrinker:
@@ -869,42 +996,14 @@ __nfsd_file_cache_purge(struct net *net)
 	nfsd_file_dispose_list(&dispose);
 }
 
-static struct nfsd_fcache_disposal *
-nfsd_alloc_fcache_disposal(void)
-{
-	struct nfsd_fcache_disposal *l;
-
-	l = kmalloc(sizeof(*l), GFP_KERNEL);
-	if (!l)
-		return NULL;
-	spin_lock_init(&l->lock);
-	INIT_LIST_HEAD(&l->freeme);
-	return l;
-}
-
-static void
-nfsd_free_fcache_disposal(struct nfsd_fcache_disposal *l)
-{
-	nfsd_file_dispose_list(&l->freeme);
-	kfree(l);
-}
-
-static void
-nfsd_free_fcache_disposal_net(struct net *net)
-{
-	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
-	struct nfsd_fcache_disposal *l = nn->fcache_disposal;
-
-	nfsd_free_fcache_disposal(l);
-}
-
 int
 nfsd_file_cache_start_net(struct net *net)
 {
 	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
 
-	nn->fcache_disposal = nfsd_alloc_fcache_disposal();
-	return nn->fcache_disposal ? 0 : -ENOMEM;
+	spin_lock_init(&nn->fcache_dispose_lock);
+	INIT_LIST_HEAD(&nn->fcache_dispose_list);
+	return 0;
 }
 
 /**
@@ -923,8 +1022,18 @@ nfsd_file_cache_purge(struct net *net)
 void
 nfsd_file_cache_shutdown_net(struct net *net)
 {
+	struct nfsd_net *nn = net_generic(net, nfsd_net_id);
+
 	nfsd_file_cache_purge(net);
-	nfsd_free_fcache_disposal_net(net);
+	/*
+	 * Ensure any in-progress shrinker, GC, or fsnotify/lease callback
+	 * (all of which hold nfsd_gc_lock while calling
+	 * nfsd_file_dispose_list_delayed()) has fully completed before
+	 * draining the per-net dispose list.
+	 */
+	spin_lock(&nfsd_gc_lock);
+	spin_unlock(&nfsd_gc_lock);
+	nfsd_file_dispose_list(&nn->fcache_dispose_list);
 }
 
 void
@@ -948,6 +1057,8 @@ nfsd_file_cache_shutdown(void)
 	rcu_barrier();
 	fsnotify_put_group(nfsd_file_fsnotify_group);
 	nfsd_file_fsnotify_group = NULL;
+	fsnotify_put_group(nfsd_dir_fsnotify_group);
+	nfsd_dir_fsnotify_group = NULL;
 	kmem_cache_destroy(nfsd_file_slab);
 	nfsd_file_slab = NULL;
 	fsnotify_wait_marks_destroyed();
@@ -1027,12 +1138,41 @@ nfsd_file_is_cached(struct inode *inode)
 }
 
 static __be32
+nfsd_file_get_dio_attrs(const struct svc_fh *fhp, struct nfsd_file *nf)
+{
+	struct inode *inode = file_inode(nf->nf_file);
+	struct kstat stat;
+	__be32 status;
+
+	/* Currently only need to get DIO alignment info for regular files */
+	if (!S_ISREG(inode->i_mode))
+		return nfs_ok;
+
+	status = fh_getattr(fhp, &stat);
+	if (status != nfs_ok)
+		return status;
+
+	trace_nfsd_file_get_dio_attrs(inode, &stat);
+
+	if (stat.result_mask & STATX_DIOALIGN) {
+		nf->nf_dio_mem_align = stat.dio_mem_align;
+		nf->nf_dio_offset_align = stat.dio_offset_align;
+	}
+	if (stat.result_mask & STATX_DIO_READ_ALIGN)
+		nf->nf_dio_read_offset_align = stat.dio_read_offset_align;
+	else
+		nf->nf_dio_read_offset_align = nf->nf_dio_offset_align;
+
+	return nfs_ok;
+}
+
+static __be32
 nfsd_file_do_acquire(struct svc_rqst *rqstp, struct net *net,
 		     struct svc_cred *cred,
 		     struct auth_domain *client,
 		     struct svc_fh *fhp,
 		     unsigned int may_flags, struct file *file,
-		     struct nfsd_file **pnf, bool want_gc)
+		     umode_t type, bool want_gc, struct nfsd_file **pnf)
 {
 	unsigned char need = may_flags & NFSD_FILE_MAY_MASK;
 	struct nfsd_file *new, *nf;
@@ -1043,13 +1183,13 @@ nfsd_file_do_acquire(struct svc_rqst *rqstp, struct net *net,
 	int ret;
 
 retry:
-	if (rqstp) {
-		status = fh_verify(rqstp, fhp, S_IFREG,
+	if (rqstp)
+		status = fh_verify(rqstp, fhp, type,
 				   may_flags|NFSD_MAY_OWNER_OVERRIDE);
-	} else {
-		status = fh_verify_local(net, cred, client, fhp, S_IFREG,
+	else
+		status = fh_verify_local(net, cred, client, fhp, type,
 					 may_flags|NFSD_MAY_OWNER_OVERRIDE);
-	}
+
 	if (status != nfs_ok)
 		return status;
 	inode = d_inode(fhp->fh_dentry);
@@ -1058,16 +1198,8 @@ retry:
 	nf = nfsd_file_lookup_locked(net, current_cred(), inode, need, want_gc);
 	rcu_read_unlock();
 
-	if (nf) {
-		/*
-		 * If the nf is on the LRU then it holds an extra reference
-		 * that must be put if it's removed. It had better not be
-		 * the last one however, since we should hold another.
-		 */
-		if (nfsd_file_lru_remove(nf))
-			refcount_dec(&nf->nf_ref);
+	if (nf)
 		goto wait_for_construction;
-	}
 
 	new = nfsd_file_alloc(net, inode, need, want_gc);
 	if (!new) {
@@ -1130,15 +1262,16 @@ out:
 
 open_file:
 	trace_nfsd_file_alloc(nf);
+
 	nf->nf_mark = nfsd_file_mark_find_or_create(inode);
 	if (nf->nf_mark) {
-		if (file) {
+		if (file && (file->f_mode & FMODE_OPENED)) {
 			get_file(file);
 			nf->nf_file = file;
 			status = nfs_ok;
 			trace_nfsd_file_opened(nf, status);
 		} else {
-			ret = nfsd_open_verified(fhp, may_flags, &nf->nf_file);
+			ret = nfsd_open_verified(fhp, type, may_flags, &nf->nf_file);
 			if (ret == -EOPENSTALE && stale_retry) {
 				stale_retry = false;
 				nfsd_file_unhash(nf);
@@ -1152,6 +1285,8 @@ open_file:
 			}
 			status = nfserrno(ret);
 			trace_nfsd_file_open(nf, status);
+			if (status == nfs_ok)
+				status = nfsd_file_get_dio_attrs(fhp, nf);
 		}
 	} else
 		status = nfserr_jukebox;
@@ -1161,6 +1296,9 @@ open_file:
 	 */
 	if (status != nfs_ok || inode->i_nlink == 0)
 		nfsd_file_unhash(nf);
+	else if (want_gc)
+		nfsd_file_lru_add(nf);
+
 	clear_and_wake_up_bit(NFSD_FILE_PENDING, &nf->nf_flags);
 	if (status == nfs_ok)
 		goto out;
@@ -1195,7 +1333,7 @@ nfsd_file_acquire_gc(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		     unsigned int may_flags, struct nfsd_file **pnf)
 {
 	return nfsd_file_do_acquire(rqstp, SVC_NET(rqstp), NULL, NULL,
-				    fhp, may_flags, NULL, pnf, true);
+				    fhp, may_flags, NULL, S_IFREG, true, pnf);
 }
 
 /**
@@ -1220,7 +1358,7 @@ nfsd_file_acquire(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		  unsigned int may_flags, struct nfsd_file **pnf)
 {
 	return nfsd_file_do_acquire(rqstp, SVC_NET(rqstp), NULL, NULL,
-				    fhp, may_flags, NULL, pnf, false);
+				    fhp, may_flags, NULL, S_IFREG, false, pnf);
 }
 
 /**
@@ -1263,8 +1401,8 @@ nfsd_file_acquire_local(struct net *net, struct svc_cred *cred,
 	const struct cred *save_cred = get_current_cred();
 	__be32 beres;
 
-	beres = nfsd_file_do_acquire(NULL, net, cred, client,
-				     fhp, may_flags, NULL, pnf, false);
+	beres = nfsd_file_do_acquire(NULL, net, cred, client, fhp, may_flags,
+				     NULL, S_IFREG, false, pnf);
 	put_cred(revert_creds(save_cred));
 	return beres;
 }
@@ -1274,12 +1412,12 @@ nfsd_file_acquire_local(struct net *net, struct svc_cred *cred,
  * @rqstp: the RPC transaction being executed
  * @fhp: the NFS filehandle of the file just created
  * @may_flags: NFSD_MAY_ settings for the file
- * @file: cached, already-open file (may be NULL)
+ * @file: cached, already-open file (may be NULL or not yet opened)
  * @pnf: OUT: new or found "struct nfsd_file" object
  *
  * Acquire a nfsd_file object that is not GC'ed. If one doesn't already exist,
- * and @file is non-NULL, use it to instantiate a new nfsd_file instead of
- * opening a new one.
+ * and @file has FMODE_OPENED set, use it to instantiate a new nfsd_file
+ * instead of opening a new one.
  *
  * Return values:
  *   %nfs_ok - @pnf points to an nfsd_file with its reference
@@ -1293,7 +1431,33 @@ nfsd_file_acquire_opened(struct svc_rqst *rqstp, struct svc_fh *fhp,
 			 struct nfsd_file **pnf)
 {
 	return nfsd_file_do_acquire(rqstp, SVC_NET(rqstp), NULL, NULL,
-				    fhp, may_flags, file, pnf, false);
+				    fhp, may_flags, file, S_IFREG, false, pnf);
+}
+
+/**
+ * nfsd_file_acquire_dir - Get a struct nfsd_file with an open directory
+ * @rqstp: the RPC transaction being executed
+ * @fhp: the NFS filehandle of the file to be opened
+ * @pnf: OUT: new or found "struct nfsd_file" object
+ *
+ * The nfsd_file_object returned by this API is reference-counted
+ * but not garbage-collected. The object is unhashed after the
+ * final nfsd_file_put(). This opens directories only, and only
+ * in O_RDONLY mode.
+ *
+ * Return values:
+ *   %nfs_ok - @pnf points to an nfsd_file with its reference
+ *   count boosted.
+ *
+ * On error, an nfsstat value in network byte order is returned.
+ */
+__be32
+nfsd_file_acquire_dir(struct svc_rqst *rqstp, struct svc_fh *fhp,
+		      struct nfsd_file **pnf)
+{
+	return nfsd_file_do_acquire(rqstp, SVC_NET(rqstp), NULL, NULL, fhp,
+				    NFSD_MAY_READ|NFSD_MAY_64BIT_COOKIE,
+				    NULL, S_IFDIR, false, pnf);
 }
 
 /*
@@ -1348,3 +1512,54 @@ int nfsd_file_cache_stats_show(struct seq_file *m, void *v)
 		seq_printf(m, "mean age (ms): -\n");
 	return 0;
 }
+
+/**
+ * nfsd_fsnotify_recalc_mask - recalculate the fsnotify mask for a nfsd_file
+ * @nf: nfsd_file to recalculate the mask on
+ *
+ * When a directory nfsd_file has a delegation added or removed, that may
+ * change the events that nfsd requires from the VFS layer. This function
+ * recalculates the fsnotify mask based on the leases present.
+ */
+void nfsd_fsnotify_recalc_mask(struct nfsd_file *nf)
+{
+	struct inode *inode = file_inode(nf->nf_file);
+	u32 lease_mask, set = 0, clear = 0;
+	struct fsnotify_mark *mark;
+
+	/* This is only needed when adding or removing dir delegs */
+	if (!S_ISDIR(inode->i_mode) || !nf->nf_mark)
+		return;
+
+	mark = &nf->nf_mark->nfm_mark;
+
+	/*
+	 * The mark is shared by every nfsd_file on this inode, so concurrent
+	 * delegation add/remove on the same directory can recalc it in
+	 * parallel. Serialize the read of the lease state and the update of
+	 * the mark so that a recalc working from a stale snapshot of the
+	 * lease list can't clobber a concurrent recalc's update.
+	 */
+	mutex_lock(&nf->nf_mark->nfm_recalc_mutex);
+
+	/* Set up notifications for any ignored delegation events */
+	lease_mask = inode_lease_ignore_mask(inode);
+
+	if (lease_mask & FL_IGN_DIR_CREATE)
+		set |= FS_CREATE | FS_MOVED_TO;
+	else
+		clear |= FS_CREATE | FS_MOVED_TO;
+
+	if (lease_mask & FL_IGN_DIR_DELETE)
+		set |= FS_DELETE | FS_MOVED_FROM;
+	else
+		clear |= FS_DELETE | FS_MOVED_FROM;
+
+	if (lease_mask & FL_IGN_DIR_RENAME)
+		set |= FS_RENAME;
+	else
+		clear |= FS_RENAME;
+
+	fsnotify_modify_mark_mask(mark, set, clear);
+	mutex_unlock(&nf->nf_mark->nfm_recalc_mutex);
+}