Merge remote branch 'fsnotify/for-next'

8 files changed, 107 insertions, 80 deletions

diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index 42c7fafc8bfe..e33b9a00b3bb 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -13,7 +13,6 @@
 #include <linux/sched.h>
 #include <linux/file.h>
 #include <linux/fs.h>
-#include <linux/fsnotify.h>
 #include <linux/quotaops.h>
 #include <linux/xattr.h>
 #include <linux/mount.h>
diff --git a/fs/cachefiles/xattr.c b/fs/cachefiles/xattr.c
index e18b183b47e1..6e050686e9b3 100644
--- a/fs/cachefiles/xattr.c
+++ b/fs/cachefiles/xattr.c
@@ -13,7 +13,6 @@
 #include <linux/sched.h>
 #include <linux/file.h>
 #include <linux/fs.h>
-#include <linux/fsnotify.h>
 #include <linux/quotaops.h>
 #include <linux/xattr.h>
 #include <linux/slab.h>
diff --git a/fs/namei.c b/fs/namei.c
index 5362af9b7372..4ff7ca530533 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -1748,6 +1748,9 @@ struct file *do_filp_open(int dfd, const char *pathname,
 	if (!(open_flag & O_CREAT))
 		mode = 0;
 
+	/* Must never be set by userspace */
+	open_flag &= ~FMODE_NONOTIFY;
+
 	/*
 	 * O_SYNC is implemented as __O_SYNC|O_DSYNC.  As many places only
 	 * check for O_DSYNC if the need any syncing at all we enforce it's
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 184938fcff04..5f30270ae644 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -20,7 +20,6 @@
 #include <linux/fcntl.h>
 #include <linux/namei.h>
 #include <linux/delay.h>
-#include <linux/fsnotify.h>
 #include <linux/posix_acl_xattr.h>
 #include <linux/xattr.h>
 #include <linux/jhash.h>
@@ -924,7 +923,6 @@ nfsd_vfs_read(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
 		nfsdstats.io_read += host_err;
 		*count = host_err;
 		err = 0;
-		fsnotify_access(file);
 	} else 
 		err = nfserrno(host_err);
 out:
@@ -1035,7 +1033,6 @@ nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp, struct file *file,
 		goto out_nfserr;
 	*cnt = host_err;
 	nfsdstats.io_write += host_err;
-	fsnotify_modify(file);
 
 	/* clear setuid/setgid flag after write */
 	if (inode->i_mode & (S_ISUID | S_ISGID))
diff --git a/fs/notify/fanotify/fanotify.c b/fs/notify/fanotify/fanotify.c
index b04f88eed09e..c2ba86a972e1 100644
--- a/fs/notify/fanotify/fanotify.c
+++ b/fs/notify/fanotify/fanotify.c
@@ -92,7 +92,11 @@ static int fanotify_get_response_from_access(struct fsnotify_group *group,
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
-	wait_event(group->fanotify_data.access_waitq, event->response);
+	wait_event(group->fanotify_data.access_waitq, event->response ||
+				atomic_read(&group->fanotify_data.bypass_perm));
+
+	if (!event->response) /* bypass_perm set */
+		return 0;
 
 	/* userspace responded, convert to something usable */
 	spin_lock(&event->lock);
@@ -180,13 +184,6 @@ static bool fanotify_should_send_event(struct fsnotify_group *group,
 		marks_mask = (vfsmnt_mark->mask | inode_mark->mask);
 		marks_ignored_mask = (vfsmnt_mark->ignored_mask | inode_mark->ignored_mask);
 	} else if (inode_mark) {
-		/*
-		 * if the event is for a child and this inode doesn't care about
-		 * events on the child, don't send it!
-		 */
-		if ((event_mask & FS_EVENT_ON_CHILD) &&
-		    !(inode_mark->mask & FS_EVENT_ON_CHILD))
-			return false;
 		marks_mask = inode_mark->mask;
 		marks_ignored_mask = inode_mark->ignored_mask;
 	} else if (vfsmnt_mark) {
@@ -200,10 +197,21 @@ static bool fanotify_should_send_event(struct fsnotify_group *group,
 	    (marks_ignored_mask & FS_ISDIR))
 		return false;
 
-	if (event_mask & marks_mask & ~marks_ignored_mask)
-		return true;
+	/*
+	 * if the event is for a child and this inode doesn't care about
+	 * events on the child, don't send it!
+	 */
+	if ((event_mask & FS_EVENT_ON_CHILD) &&
+	    !(marks_mask & FS_EVENT_ON_CHILD))
+		return false;
 
-	return false;
+	/*
+	 * It might seem logical to check:
+	 * if (event_mask & marks_mask & ~marks_ignored_mask)
+	 * 	return true;
+	 * but we we know this was true from the caller so just return true.
+	 */
+	return true;
 }
 
 static void fanotify_free_group_priv(struct fsnotify_group *group)
diff --git a/fs/notify/fanotify/fanotify_user.c b/fs/notify/fanotify/fanotify_user.c
index 063224812b7e..5d2e68370950 100644
--- a/fs/notify/fanotify/fanotify_user.c
+++ b/fs/notify/fanotify/fanotify_user.c
@@ -62,6 +62,7 @@ static int create_fd(struct fsnotify_group *group, struct fsnotify_event *event)
 	struct dentry *dentry;
 	struct vfsmount *mnt;
 	struct file *new_file;
+	unsigned int flags;
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
@@ -83,12 +84,22 @@ static int create_fd(struct fsnotify_group *group, struct fsnotify_event *event)
 	mnt = mntget(event->path.mnt);
 	/* it's possible this event was an overflow event.  in that case dentry and mnt
 	 * are NULL;  That's fine, just don't call dentry open */
-	if (dentry && mnt)
-		new_file = dentry_open(dentry, mnt,
-				       group->fanotify_data.f_flags | FMODE_NONOTIFY,
-				       current_cred());
-	else
+	if (dentry && mnt) {
+		flags = group->fanotify_data.f_flags;
+		new_file = dentry_open(dentry, mnt, flags, current_cred());
+		/*
+		 * Attempt fallback to read-only access if writable was not possible
+		 * in order to at least provide something to the listener.
+		 */
+		if (IS_ERR(new_file) && group->fanotify_data.readonly_fallback) {
+			flags &= ~O_ACCMODE;
+			flags |= O_RDONLY;
+			new_file = dentry_open(dentry, mnt, flags,
+					       current_cred());
+		}
+	} else {
 		new_file = ERR_PTR(-EOVERFLOW);
+	}
 	if (IS_ERR(new_file)) {
 		/*
 		 * we still send an event even if we can't open the file.  this
@@ -106,10 +117,12 @@ static int create_fd(struct fsnotify_group *group, struct fsnotify_event *event)
 	return client_fd;
 }
 
-static ssize_t fill_event_metadata(struct fsnotify_group *group,
+static int fill_event_metadata(struct fsnotify_group *group,
 				   struct fanotify_event_metadata *metadata,
 				   struct fsnotify_event *event)
 {
+	int ret = 0;
+
 	pr_debug("%s: group=%p metadata=%p event=%p\n", __func__,
 		 group, metadata, event);
 
@@ -117,9 +130,15 @@ static ssize_t fill_event_metadata(struct fsnotify_group *group,
 	metadata->vers = FANOTIFY_METADATA_VERSION;
 	metadata->mask = event->mask & FAN_ALL_OUTGOING_EVENTS;
 	metadata->pid = pid_vnr(event->tgid);
-	metadata->fd = create_fd(group, event);
+	if (unlikely(event->mask & FAN_Q_OVERFLOW))
+		metadata->fd = FAN_NOFD;
+	else {
+		metadata->fd = create_fd(group, event);
+		if (metadata->fd < 0)
+			ret = metadata->fd;
+	}
 
-	return metadata->fd;
+	return ret;
 }
 
 #ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
@@ -199,14 +218,6 @@ static int prepare_for_access_response(struct fsnotify_group *group,
 	re->fd = fd;
 
 	mutex_lock(&group->fanotify_data.access_mutex);
-
-	if (group->fanotify_data.bypass_perm) {
-		mutex_unlock(&group->fanotify_data.access_mutex);
-		kmem_cache_free(fanotify_response_event_cache, re);
-		event->response = FAN_ALLOW;
-		return 0;
-	}
-		
 	list_add_tail(&re->list, &group->fanotify_data.access_list);
 	mutex_unlock(&group->fanotify_data.access_mutex);
 
@@ -257,10 +268,11 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
-	fd = fill_event_metadata(group, &fanotify_event_metadata, event);
-	if (fd < 0)
-		return fd;
+	ret = fill_event_metadata(group, &fanotify_event_metadata, event);
+	if (ret < 0)
+		goto out;
 
+	fd = fanotify_event_metadata.fd;
 	ret = prepare_for_access_response(group, event, fd);
 	if (ret)
 		goto out_close_fd;
@@ -274,7 +286,15 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 out_kill_access_response:
 	remove_access_response(group, event, fd);
 out_close_fd:
-	sys_close(fd);
+	if (fd != FAN_NOFD)
+		sys_close(fd);
+out:
+#ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
+	if (event->mask & FAN_ALL_PERM_EVENTS) {
+		event->response = FAN_DENY;
+		wake_up(&group->fanotify_data.access_waitq);
+	}
+#endif
 	return ret;
 }
 
@@ -382,7 +402,7 @@ static int fanotify_release(struct inode *ignored, struct file *file)
 
 	mutex_lock(&group->fanotify_data.access_mutex);
 
-	group->fanotify_data.bypass_perm = true;
+	atomic_inc(&group->fanotify_data.bypass_perm);
 
 	list_for_each_entry_safe(re, lre, &group->fanotify_data.access_list, list) {
 		pr_debug("%s: found group=%p re=%p event=%p\n", __func__, group,
@@ -586,11 +606,10 @@ static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
 {
 	struct fsnotify_mark *fsn_mark;
 	__u32 added;
+	int ret = 0;
 
 	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
 	if (!fsn_mark) {
-		int ret;
-
 		if (atomic_read(&group->num_marks) > group->fanotify_data.max_marks)
 			return -ENOSPC;
 
@@ -600,17 +619,16 @@ static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
 
 		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
 		ret = fsnotify_add_mark(fsn_mark, group, NULL, mnt, 0);
-		if (ret) {
-			fanotify_free_mark(fsn_mark);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
-	fsnotify_put_mark(fsn_mark);
+
 	if (added & ~mnt->mnt_fsnotify_mask)
 		fsnotify_recalc_vfsmount_mask(mnt);
-
-	return 0;
+err:
+	fsnotify_put_mark(fsn_mark);
+	return ret;
 }
 
 static int fanotify_add_inode_mark(struct fsnotify_group *group,
@@ -619,6 +637,7 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 {
 	struct fsnotify_mark *fsn_mark;
 	__u32 added;
+	int ret = 0;
 
 	pr_debug("%s: group=%p inode=%p\n", __func__, group, inode);
 
@@ -634,8 +653,6 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 
 	fsn_mark = fsnotify_find_inode_mark(group, inode);
 	if (!fsn_mark) {
-		int ret;
-
 		if (atomic_read(&group->num_marks) > group->fanotify_data.max_marks)
 			return -ENOSPC;
 
@@ -645,16 +662,16 @@ static int fanotify_add_inode_mark(struct fsnotify_group *group,
 
 		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
 		ret = fsnotify_add_mark(fsn_mark, group, inode, NULL, 0);
-		if (ret) {
-			fanotify_free_mark(fsn_mark);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
-	fsnotify_put_mark(fsn_mark);
+
 	if (added & ~inode->i_fsnotify_mask)
 		fsnotify_recalc_inode_mask(inode);
-	return 0;
+err:
+	fsnotify_put_mark(fsn_mark);
+	return ret;
 }
 
 /* fanotify syscalls */
@@ -687,17 +704,20 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
 
 	/* fsnotify_alloc_group takes a ref.  Dropped in fanotify_release */
 	group = fsnotify_alloc_group(&fanotify_fsnotify_ops);
-	if (IS_ERR(group))
+	if (IS_ERR(group)) {
+		free_uid(user);
 		return PTR_ERR(group);
+	}
 
 	group->fanotify_data.user = user;
 	atomic_inc(&user->fanotify_listeners);
 
-	group->fanotify_data.f_flags = event_f_flags;
+	group->fanotify_data.f_flags = event_f_flags | FMODE_NONOTIFY;
 #ifdef CONFIG_FANOTIFY_ACCESS_PERMISSIONS
 	mutex_init(&group->fanotify_data.access_mutex);
 	init_waitqueue_head(&group->fanotify_data.access_waitq);
 	INIT_LIST_HEAD(&group->fanotify_data.access_list);
+	atomic_set(&group->fanotify_data.bypass_perm, 0);
 #endif
 	switch (flags & FAN_ALL_CLASS_BITS) {
 	case FAN_CLASS_NOTIF:
@@ -732,6 +752,14 @@ SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
 		group->fanotify_data.max_marks = FANOTIFY_DEFAULT_MAX_MARKS;
 	}
 
+	fd = -EINVAL;
+	if (flags & FAN_READONLY_FALLBACK) {
+		if ((event_f_flags & O_ACCMODE) == O_RDWR)
+			group->fanotify_data.readonly_fallback = true;
+		else
+			goto out_put_group;
+	}
+
 	fd = anon_inode_getfd("[fanotify]", &fanotify_fops, group, f_flags);
 	if (fd < 0)
 		goto out_put_group;
@@ -764,8 +792,10 @@ SYSCALL_DEFINE(fanotify_mark)(int fanotify_fd, unsigned int flags,
 	if (flags & ~FAN_ALL_MARK_FLAGS)
 		return -EINVAL;
 	switch (flags & (FAN_MARK_ADD | FAN_MARK_REMOVE | FAN_MARK_FLUSH)) {
-	case FAN_MARK_ADD:
+	case FAN_MARK_ADD:		/* fallthrough */
 	case FAN_MARK_REMOVE:
+		if (!mask)
+			return -EINVAL;
 	case FAN_MARK_FLUSH:
 		break;
 	default:
diff --git a/fs/notify/inotify/inotify_user.c b/fs/notify/inotify/inotify_user.c
index 444c305a468c..4cd5d5d78f9f 100644
--- a/fs/notify/inotify/inotify_user.c
+++ b/fs/notify/inotify/inotify_user.c
@@ -752,6 +752,7 @@ SYSCALL_DEFINE1(inotify_init1, int, flags)
 	if (ret >= 0)
 		return ret;
 
+	fsnotify_put_group(group);
 	atomic_dec(&user->inotify_devs);
 out_free_uid:
 	free_uid(user);
diff --git a/fs/notify/mark.c b/fs/notify/mark.c
index 325185e514bb..28b64eb03e33 100644
--- a/fs/notify/mark.c
+++ b/fs/notify/mark.c
@@ -103,17 +103,6 @@ static DEFINE_SPINLOCK(destroy_lock);
 static LIST_HEAD(destroy_list);
 static DECLARE_WAIT_QUEUE_HEAD(destroy_waitq);
 
-void fsnotify_get_mark(struct fsnotify_mark *mark)
-{
-	atomic_inc(&mark->refcnt);
-}
-
-void fsnotify_put_mark(struct fsnotify_mark *mark)
-{
-	if (atomic_dec_and_test(&mark->refcnt))
-		mark->free_mark(mark);
-}
-
 /*
  * Any time a mark is getting freed we end up here.
  * The caller had better be holding a reference to this mark so we don't actually
@@ -217,7 +206,7 @@ int fsnotify_add_mark(struct fsnotify_mark *mark,
 		      struct fsnotify_group *group, struct inode *inode,
 		      struct vfsmount *mnt, int allow_dups)
 {
-	int ret = 0;
+	int ret;
 
 	BUG_ON(inode && mnt);
 	BUG_ON(!inode && !mnt);
@@ -232,23 +221,20 @@ int fsnotify_add_mark(struct fsnotify_mark *mark,
 	spin_lock(&group->mark_lock);
 
 	mark->flags |= FSNOTIFY_MARK_FLAG_ALIVE;
-
 	mark->group = group;
 	list_add(&mark->g_list, &group->marks_list);
-	atomic_inc(&group->num_marks);
 	fsnotify_get_mark(mark); /* for i_list and g_list */
+	atomic_inc(&group->num_marks);
 
-	if (inode) {
+	ret = 0;
+	if (inode)
 		ret = fsnotify_add_inode_mark(mark, group, inode, allow_dups);
-		if (ret)
-			goto err;
-	} else if (mnt) {
+	else if (mnt)
 		ret = fsnotify_add_vfsmount_mark(mark, group, mnt, allow_dups);
-		if (ret)
-			goto err;
-	} else {
+	else
 		BUG();
-	}
+	if (ret)
+		goto err;
 
 	spin_unlock(&group->mark_lock);
 
@@ -260,7 +246,7 @@ int fsnotify_add_mark(struct fsnotify_mark *mark,
 	if (inode)
 		__fsnotify_update_child_dentry_flags(inode);
 
-	return ret;
+	return 0;
 err:
 	mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
 	list_del_init(&mark->g_list);
@@ -346,6 +332,10 @@ static int fsnotify_mark_destroy(void *ignored)
 
 		synchronize_srcu(&fsnotify_mark_srcu);
 
+		/*
+		 * at this point we cannot be found via the i_list or g_list so
+		 * drop that reference.
+		 */
 		list_for_each_entry_safe(mark, next, &private_destroy_list, destroy_list) {
 			list_del_init(&mark->destroy_list);
 			fsnotify_put_mark(mark);