mnt: fs_fully_visible test the proper mount for MNT_LOCKED

commit d71ed6c930ac7d8f88f3cef6624a7e826392d61f upstream.

MNT_LOCKED implies on a child mount implies the child is locked to the
parent.  So while looping through the children the children should be
tested (not their parent).

Typically an unshare of a mount namespace locks all mounts together
making both the parent and the slave as locked but there are a few
corner cases where other things work.

Fixes: ceeb0e5d39fc ("vfs: Ignore unlocked mounts in fs_fully_visible")
Reported-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/fs/namespace.c b/fs/namespace.c
index 0570729c87fd22dd299c50ee0303f43c384e4932..efb396f41fcdbbc1ba3c289d6d4e8c38a5561863 100644 (file)
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -3262,7 +3262,7 @@ static bool fs_fully_visible(struct file_system_type *type, int *new_mnt_flags)
                list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) {
                        struct inode *inode = child->mnt_mountpoint->d_inode;
                        /* Only worry about locked mounts */
-                       if (!(mnt_flags & MNT_LOCKED))
+                       if (!(child->mnt.mnt_flags & MNT_LOCKED))
                                continue;
                        /* Is the directory permanetly empty? */
                        if (!is_empty_dir_inode(inode))