fanotify: support watching filesystems and mounts inside userns

An unprivileged user is allowed to create an fanotify group and add
inode marks, but not filesystem, mntns and mount marks.

Add limited support for setting up filesystem, mntns and mount marks by
an unprivileged user under the following conditions:

1. User has CAP_SYS_ADMIN in the user ns where the group was created
2.a. User has CAP_SYS_ADMIN in the user ns where the sb was created
OR (in case setting up a mntns mark)
2.b. User has CAP_SYS_ADMIN in the user ns associated with the mntns

Signed-off-by: Amir Goldstein <[e-mail]>
Signed-off-by: Jan Kara <[e-mail]>
Link: https://patch.msgid.link/[e-mail]

fs/notify/fanotify/fanotify.c

···
 
 static void fanotify_free_group_priv(struct fsnotify_group *group)
 {
+	put_user_ns(group->user_ns);
 	kfree(group->fanotify_data.merge_hash);
 	if (group->fanotify_data.ucounts)
 		dec_ucount(group->fanotify_data.ucounts,

fs/notify/fanotify/fanotify_user.c

···
 /* fanotify syscalls */
 SYSCALL_DEFINE2(fanotify_init, unsigned int, flags, unsigned int, event_f_flags)
 {
+	struct user_namespace *user_ns = current_user_ns();
 	struct fsnotify_group *group;
 	int f_flags, fd;
 	unsigned int fid_mode = flags & FANOTIFY_FID_BITS;
···
 		/*
 		 * An unprivileged user can setup an fanotify group with
 		 * limited functionality - an unprivileged group is limited to
-		 * notification events with file handles and it cannot use
-		 * unlimited queue/marks.
+		 * notification events with file handles or mount ids and it
+		 * cannot use unlimited queue/marks.
 		 */
-		if ((flags & FANOTIFY_ADMIN_INIT_FLAGS) || !fid_mode)
+		if ((flags & FANOTIFY_ADMIN_INIT_FLAGS) ||
+		    !(flags & (FANOTIFY_FID_BITS | FAN_REPORT_MNT)))
 			return -EPERM;
 
 		/*
···
 	}
 
 	/* Enforce groups limits per user in all containing user ns */
-	group->fanotify_data.ucounts = inc_ucount(current_user_ns(),
-						  current_euid(),
+	group->fanotify_data.ucounts = inc_ucount(user_ns, current_euid(),
 						  UCOUNT_FANOTIFY_GROUPS);
 	if (!group->fanotify_data.ucounts) {
 		fd = -EMFILE;
···
 
 	group->fanotify_data.flags = flags | internal_flags;
 	group->memcg = get_mem_cgroup_from_mm(current->mm);
+	group->user_ns = get_user_ns(user_ns);
 
 	group->fanotify_data.merge_hash = fanotify_alloc_merge_hash();
 	if (!group->fanotify_data.merge_hash) {
···
 	struct fsnotify_group *group;
 	struct path path;
 	struct fan_fsid __fsid, *fsid = NULL;
+	struct user_namespace *user_ns = NULL;
+	struct mnt_namespace *mntns;
 	u32 valid_mask = FANOTIFY_EVENTS | FANOTIFY_EVENT_FLAGS;
 	unsigned int mark_type = flags & FANOTIFY_MARK_TYPE_BITS;
 	unsigned int mark_cmd = flags & FANOTIFY_MARK_CMD_BITS;
···
 	}
 
 	/*
-	 * An unprivileged user is not allowed to setup mount nor filesystem
-	 * marks.  This also includes setting up such marks by a group that
-	 * was initialized by an unprivileged user.
+	 * A user is allowed to setup sb/mount/mntns marks only if it is
+	 * capable in the user ns where the group was created.
 	 */
-	if ((!capable(CAP_SYS_ADMIN) ||
-	     FAN_GROUP_FLAG(group, FANOTIFY_UNPRIV)) &&
+	if (!ns_capable(group->user_ns, CAP_SYS_ADMIN) &&
 	    mark_type != FAN_MARK_INODE)
 		return -EPERM;
 
···
 		fsid = &__fsid;
 	}
 
-	/* inode held in place by reference to path; group by fget on fd */
+	/*
+	 * In addition to being capable in the user ns where group was created,
+	 * the user also needs to be capable in the user ns associated with
+	 * the filesystem or in the user ns associated with the mntns
+	 * (when marking mntns).
+	 */
 	if (obj_type == FSNOTIFY_OBJ_TYPE_INODE) {
 		inode = path.dentry->d_inode;
 		obj = inode;
 	} else if (obj_type == FSNOTIFY_OBJ_TYPE_VFSMOUNT) {
+		user_ns = path.mnt->mnt_sb->s_user_ns;
 		obj = path.mnt;
 	} else if (obj_type == FSNOTIFY_OBJ_TYPE_SB) {
+		user_ns = path.mnt->mnt_sb->s_user_ns;
 		obj = path.mnt->mnt_sb;
 	} else if (obj_type == FSNOTIFY_OBJ_TYPE_MNTNS) {
-		obj = mnt_ns_from_dentry(path.dentry);
+		mntns = mnt_ns_from_dentry(path.dentry);
+		user_ns = mntns->user_ns;
+		obj = mntns;
 	}
+
+	ret = -EPERM;
+	if (user_ns && !ns_capable(user_ns, CAP_SYS_ADMIN))
+		goto path_put_and_out;
 
 	ret = -EINVAL;
 	if (!obj)

include/linux/fanotify.h

···
 					 FAN_REPORT_PIDFD | \
 					 FAN_REPORT_FD_ERROR | \
 					 FAN_UNLIMITED_QUEUE | \
-					 FAN_UNLIMITED_MARKS | \
-					 FAN_REPORT_MNT)
+					 FAN_UNLIMITED_MARKS)
 
 /*
  * fanotify_init() flags that are allowed for user without CAP_SYS_ADMIN.
···
  * so one of the flags for reporting file handles is required.
  */
 #define FANOTIFY_USER_INIT_FLAGS	(FAN_CLASS_NOTIF | \
-					 FANOTIFY_FID_BITS | \
+					 FANOTIFY_FID_BITS | FAN_REPORT_MNT | \
 					 FAN_CLOEXEC | FAN_NONBLOCK)
 
 #define FANOTIFY_INIT_FLAGS	(FANOTIFY_ADMIN_INIT_FLAGS | \

include/linux/fsnotify_backend.h

···
 						 * full */
 
 	struct mem_cgroup *memcg;	/* memcg to charge allocations */
+	struct user_namespace *user_ns;	/* user ns where group was created */
 
 	/* groups can define private fields here or use the void *private */
 	union {