Merge tag 'fsnotify_for_v6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs

Pull fsnotify updates from Jan Kara:
 "A couple of smaller random fsnotify fixes"

* tag 'fsnotify_for_v6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  fsnotify: Fix ordering of iput() and watched_objects decrement
  fsnotify: fix sending inotify event with unexpected filename
  fanotify: allow reporting errors on failure to open fd
  fsnotify, lsm: Decouple fsnotify from lsm

Author: torvalds

fs/notify/fanotify/Kconfig

@@ -15,7 +15,6 @@ config FANOTIFY
 config FANOTIFY_ACCESS_PERMISSIONS
 	bool "fanotify permissions checking"
 	depends on FANOTIFY
-	depends on SECURITY
 	default n
 	help
 	   Say Y here is you want fanotify listeners to be able to make permissions

fs/notify/fanotify/fanotify_user.c

@@ -265,13 +265,6 @@ static int create_fd(struct fsnotify_group *group, const struct path *path,
 			       group->fanotify_data.f_flags | __FMODE_NONOTIFY,
 			       current_cred());
 	if (IS_ERR(new_file)) {
-		/*
-		 * we still send an event even if we can't open the file.  this
-		 * can happen when say tasks are gone and we try to open their
-		 * /proc files or we try to open a WRONLY file like in sysfs
-		 * we just send the errno to userspace since there isn't much
-		 * else we can do.
-		 */
 		put_unused_fd(client_fd);
 		client_fd = PTR_ERR(new_file);
 	} else {
@@ -662,7 +655,7 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 	unsigned int info_mode = FAN_GROUP_FLAG(group, FANOTIFY_INFO_MODES);
 	unsigned int pidfd_mode = info_mode & FAN_REPORT_PIDFD;
 	struct file *f = NULL, *pidfd_file = NULL;
-	int ret, pidfd = FAN_NOPIDFD, fd = FAN_NOFD;
+	int ret, pidfd = -ESRCH, fd = -EBADF;
 
 	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
 
@@ -690,10 +683,39 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 	if (!FAN_GROUP_FLAG(group, FANOTIFY_UNPRIV) &&
 	    path && path->mnt && path->dentry) {
 		fd = create_fd(group, path, &f);
-		if (fd < 0)
-			return fd;
+		/*
+		 * Opening an fd from dentry can fail for several reasons.
+		 * For example, when tasks are gone and we try to open their
+		 * /proc files or we try to open a WRONLY file like in sysfs
+		 * or when trying to open a file that was deleted on the
+		 * remote network server.
+		 *
+		 * For a group with FAN_REPORT_FD_ERROR, we will send the
+		 * event with the error instead of the open fd, otherwise
+		 * Userspace may not get the error at all.
+		 * In any case, userspace will not know which file failed to
+		 * open, so add a debug print for further investigation.
+		 */
+		if (fd < 0) {
+			pr_debug("fanotify: create_fd(%pd2) failed err=%d\n",
+				 path->dentry, fd);
+			if (!FAN_GROUP_FLAG(group, FAN_REPORT_FD_ERROR)) {
+				/*
+				 * Historically, we've handled EOPENSTALE in a
+				 * special way and silently dropped such
+				 * events. Now we have to keep it to maintain
+				 * backward compatibility...
+				 */
+				if (fd == -EOPENSTALE)
+					fd = 0;
+				return fd;
+			}
+		}
 	}
-	metadata.fd = fd;
+	if (FAN_GROUP_FLAG(group, FAN_REPORT_FD_ERROR))
+		metadata.fd = fd;
+	else
+		metadata.fd = fd >= 0 ? fd : FAN_NOFD;
 
 	if (pidfd_mode) {
 		/*
@@ -708,18 +730,16 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 		 * The PIDTYPE_TGID check for an event->pid is performed
 		 * preemptively in an attempt to catch out cases where the event
 		 * listener reads events after the event generating process has
-		 * already terminated. Report FAN_NOPIDFD to the event listener
-		 * in those cases, with all other pidfd creation errors being
-		 * reported as FAN_EPIDFD.
+		 * already terminated.  Depending on flag FAN_REPORT_FD_ERROR,
+		 * report either -ESRCH or FAN_NOPIDFD to the event listener in
+		 * those cases with all other pidfd creation errors reported as
+		 * the error code itself or as FAN_EPIDFD.
 		 */
-		if (metadata.pid == 0 ||
-		    !pid_has_task(event->pid, PIDTYPE_TGID)) {
-			pidfd = FAN_NOPIDFD;
-		} else {
+		if (metadata.pid && pid_has_task(event->pid, PIDTYPE_TGID))
 			pidfd = pidfd_prepare(event->pid, 0, &pidfd_file);
-			if (pidfd < 0)
-				pidfd = FAN_EPIDFD;
-		}
+
+		if (!FAN_GROUP_FLAG(group, FAN_REPORT_FD_ERROR) && pidfd < 0)
+			pidfd = pidfd == -ESRCH ? FAN_NOPIDFD : FAN_EPIDFD;
 	}
 
 	ret = -EFAULT;
@@ -736,9 +756,6 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 	buf += FAN_EVENT_METADATA_LEN;
 	count -= FAN_EVENT_METADATA_LEN;
 
-	if (fanotify_is_perm_event(event->mask))
-		FANOTIFY_PERM(event)->fd = fd;
-
 	if (info_mode) {
 		ret = copy_info_records_to_user(event, info, info_mode, pidfd,
 						buf, count);
@@ -752,15 +769,18 @@ static ssize_t copy_event_to_user(struct fsnotify_group *group,
 	if (pidfd_file)
 		fd_install(pidfd, pidfd_file);
 
+	if (fanotify_is_perm_event(event->mask))
+		FANOTIFY_PERM(event)->fd = fd;
+
 	return metadata.event_len;
 
 out_close_fd:
-	if (fd != FAN_NOFD) {
+	if (f) {
 		put_unused_fd(fd);
 		fput(f);
 	}
 
-	if (pidfd >= 0) {
+	if (pidfd_file) {
 		put_unused_fd(pidfd);
 		fput(pidfd_file);
 	}
@@ -827,15 +847,6 @@ static ssize_t fanotify_read(struct file *file, char __user *buf,
 		}
 
 		ret = copy_event_to_user(group, event, buf, count);
-		if (unlikely(ret == -EOPENSTALE)) {
-			/*
-			 * We cannot report events with stale fd so drop it.
-			 * Setting ret to 0 will continue the event loop and
-			 * do the right thing if there are no more events to
-			 * read (i.e. return bytes read, -EAGAIN or wait).
-			 */
-			ret = 0;
-		}
 
 		/*
 		 * Permission events get queued to wait for response.  Other
@@ -844,7 +855,7 @@ static ssize_t fanotify_read(struct file *file, char __user *buf,
 		if (!fanotify_is_perm_event(event->mask)) {
 			fsnotify_destroy_event(group, &event->fse);
 		} else {
-			if (ret <= 0) {
+			if (ret <= 0 || FANOTIFY_PERM(event)->fd < 0) {
 				spin_lock(&group->notification_lock);
 				finish_permission_event(group,
 					FANOTIFY_PERM(event), FAN_DENY, NULL);
@@ -1941,7 +1952,7 @@ static int __init fanotify_user_setup(void)
 				     FANOTIFY_DEFAULT_MAX_USER_MARKS);
 
 	BUILD_BUG_ON(FANOTIFY_INIT_FLAGS & FANOTIFY_INTERNAL_GROUP_FLAGS);
-	BUILD_BUG_ON(HWEIGHT32(FANOTIFY_INIT_FLAGS) != 12);
+	BUILD_BUG_ON(HWEIGHT32(FANOTIFY_INIT_FLAGS) != 13);
 	BUILD_BUG_ON(HWEIGHT32(FANOTIFY_MARK_FLAGS) != 11);
 
 	fanotify_mark_cache = KMEM_CACHE(fanotify_mark,

fs/notify/fsnotify.c

@@ -333,16 +333,19 @@ static int fsnotify_handle_event(struct fsnotify_group *group, __u32 mask,
 	if (!inode_mark)
 		return 0;
 
-	if (mask & FS_EVENT_ON_CHILD) {
-		/*
-		 * Some events can be sent on both parent dir and child marks
-		 * (e.g. FS_ATTRIB).  If both parent dir and child are
-		 * watching, report the event once to parent dir with name (if
-		 * interested) and once to child without name (if interested).
-		 * The child watcher is expecting an event without a file name
-		 * and without the FS_EVENT_ON_CHILD flag.
-		 */
-		mask &= ~FS_EVENT_ON_CHILD;
+	/*
+	 * Some events can be sent on both parent dir and child marks (e.g.
+	 * FS_ATTRIB).  If both parent dir and child are watching, report the
+	 * event once to parent dir with name (if interested) and once to child
+	 * without name (if interested).
+	 *
+	 * In any case regardless whether the parent is watching or not, the
+	 * child watcher is expecting an event without the FS_EVENT_ON_CHILD
+	 * flag. The file name is expected if and only if this is a directory
+	 * event.
+	 */
+	mask &= ~FS_EVENT_ON_CHILD;
+	if (!(mask & ALL_FSNOTIFY_DIRENT_EVENTS)) {
 		dir = NULL;
 		name = NULL;
 	}

fs/notify/mark.c

@@ -138,8 +138,11 @@ static void fsnotify_get_sb_watched_objects(struct super_block *sb)
 
 static void fsnotify_put_sb_watched_objects(struct super_block *sb)
 {
-	if (atomic_long_dec_and_test(fsnotify_sb_watched_objects(sb)))
-		wake_up_var(fsnotify_sb_watched_objects(sb));
+	atomic_long_t *watched_objects = fsnotify_sb_watched_objects(sb);
+
+	/* the superblock can go away after this decrement */
+	if (atomic_long_dec_and_test(watched_objects))
+		wake_up_var(watched_objects);
 }
 
 static void fsnotify_get_inode_ref(struct inode *inode)
@@ -150,8 +153,11 @@ static void fsnotify_get_inode_ref(struct inode *inode)
 
 static void fsnotify_put_inode_ref(struct inode *inode)
 {
-	fsnotify_put_sb_watched_objects(inode->i_sb);
+	/* read ->i_sb before the inode can go away */
+	struct super_block *sb = inode->i_sb;
+
 	iput(inode);
+	fsnotify_put_sb_watched_objects(sb);
 }
 
 /*

fs/open.c

@@ -929,6 +929,10 @@ static int do_dentry_open(struct file *f,
 	if (error)
 		goto cleanup_all;
 
+	error = fsnotify_open_perm(f);
+	if (error)
+		goto cleanup_all;
+
 	error = break_lease(file_inode(f), f->f_flags);
 	if (error)
 		goto cleanup_all;

include/linux/fanotify.h

@@ -36,6 +36,7 @@
 #define FANOTIFY_ADMIN_INIT_FLAGS	(FANOTIFY_PERM_CLASSES | \
 					 FAN_REPORT_TID | \
 					 FAN_REPORT_PIDFD | \
+					 FAN_REPORT_FD_ERROR | \
 					 FAN_UNLIMITED_QUEUE | \
 					 FAN_UNLIMITED_MARKS)
 

include/uapi/linux/fanotify.h

@@ -60,6 +60,7 @@
 #define FAN_REPORT_DIR_FID	0x00000400	/* Report unique directory id */
 #define FAN_REPORT_NAME		0x00000800	/* Report events with name */
 #define FAN_REPORT_TARGET_FID	0x00001000	/* Report dirent target id  */
+#define FAN_REPORT_FD_ERROR	0x00002000	/* event->fd can report error */
 
 /* Convenience macro - FAN_REPORT_NAME requires FAN_REPORT_DIR_FID */
 #define FAN_REPORT_DFID_NAME	(FAN_REPORT_DIR_FID | FAN_REPORT_NAME)

security/security.c

@@ -19,7 +19,6 @@
 #include <linux/kernel.h>
 #include <linux/kernel_read_file.h>
 #include <linux/lsm_hooks.h>
-#include <linux/fsnotify.h>
 #include <linux/mman.h>
 #include <linux/mount.h>
 #include <linux/personality.h>
@@ -3103,13 +3102,7 @@ int security_file_receive(struct file *file)
  */
 int security_file_open(struct file *file)
 {
-	int ret;
-
-	ret = call_int_hook(file_open, file);
-	if (ret)
-		return ret;
-
-	return fsnotify_open_perm(file);
+	return call_int_hook(file_open, file);
 }
 
 /**