fs/namespace.c in the Linux kernel before 4.9 does not restrict how many mounts may exist in a mount namespace, which allows local users to cause a denial of service (memory consumption and deadlock) via MS_BIND mount system calls, as demonstrated by a loop that triggers exponential growth in the number of mounts.
https://github.com/torvalds/linux/commit/d29216842a85c7970c536108e093963f02714498
https://bugzilla.redhat.com/show_bug.cgi?id=1356471
https://access.redhat.com/errata/RHSA-2017:2077
https://access.redhat.com/errata/RHSA-2017:1842
http://www.securityfocus.com/bid/91754