Merge tag 'fsnotify_for_v5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs

Pull fsnotify fixes from Jan Kara:
 "Two fsnotify fixes"

* tag 'fsnotify_for_v5.2-rc1' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  fsnotify: fix unlink performance regression
  fsnotify: Clarify connector assignment in fsnotify_add_mark_list()

2 files changed, 46 insertions, 0 deletions

diff --git a/fs/notify/fsnotify.c b/fs/notify/fsnotify.c
index 5433e37fb0c5..8c7cbac7183c 100644
--- a/fs/notify/fsnotify.c
+++ b/fs/notify/fsnotify.c
@@ -108,6 +108,47 @@ void fsnotify_sb_delete(struct super_block *sb)
 }
 
 /*
+ * fsnotify_nameremove - a filename was removed from a directory
+ *
+ * This is mostly called under parent vfs inode lock so name and
+ * dentry->d_parent should be stable. However there are some corner cases where
+ * inode lock is not held. So to be on the safe side and be reselient to future
+ * callers and out of tree users of d_delete(), we do not assume that d_parent
+ * and d_name are stable and we use dget_parent() and
+ * take_dentry_name_snapshot() to grab stable references.
+ */
+void fsnotify_nameremove(struct dentry *dentry, int isdir)
+{
+	struct dentry *parent;
+	struct name_snapshot name;
+	__u32 mask = FS_DELETE;
+
+	/* d_delete() of pseudo inode? (e.g. __ns_get_path() playing tricks) */
+	if (IS_ROOT(dentry))
+		return;
+
+	if (isdir)
+		mask |= FS_ISDIR;
+
+	parent = dget_parent(dentry);
+	/* Avoid unneeded take_dentry_name_snapshot() */
+	if (!(d_inode(parent)->i_fsnotify_mask & FS_DELETE) &&
+	    !(dentry->d_sb->s_fsnotify_mask & FS_DELETE))
+		goto out_dput;
+
+	take_dentry_name_snapshot(&name, dentry);
+
+	fsnotify(d_inode(parent), mask, d_inode(dentry), FSNOTIFY_EVENT_INODE,
+		 &name.name, 0);
+
+	release_dentry_name_snapshot(&name);
+
+out_dput:
+	dput(parent);
+}
+EXPORT_SYMBOL(fsnotify_nameremove);
+
+/*
  * Given an inode, first check if we care what happens to our children.  Inotify
  * and dnotify both tell their parents about events.  If we care about any event
  * on a child we run all of our children and set a dentry flag saying that the
diff --git a/fs/notify/mark.c b/fs/notify/mark.c
index 22acb0a79b53..b251105f646f 100644
--- a/fs/notify/mark.c
+++ b/fs/notify/mark.c
@@ -619,6 +619,11 @@ restart:
 	/* mark should be the last entry.  last is the current last entry */
 	hlist_add_behind_rcu(&mark->obj_list, &last->obj_list);
 added:
+	/*
+	 * Since connector is attached to object using cmpxchg() we are
+	 * guaranteed that connector initialization is fully visible by anyone
+	 * seeing mark->connector set.
+	 */
 	WRITE_ONCE(mark->connector, conn);
 out_err:
 	spin_unlock(&conn->lock);