Bug 1323956 (CVE-2016-3961, xsa174)

ISSUE DESCRIPTION
=================

Huge (2Mb) pages are generally unavailable to PV guests.  Since x86
Linux pvops-based kernels are generally multi purpose, they would
normally be built with hugetlbfs support enabled.  Use of that
functionality by an application in a PV guest would cause an
infinite page fault loop, and an OOPS to occur upon an attempt to
terminate the hung application.

IMPACT
======

Depending on the guest kernel configuration, the OOPS could result
in a kernel crash (guest DoS).

VULNERABLE SYSTEMS
==================

All upstream x86 Linux versions operating as PV Xen guests are
vulnerable.

ARM systems are not vulnerable.  x86 HVM guests are not vulnerable.

x86 Linux versions derived from linux-2.6.18-xen.hg (XenoLinux) are not
vulnerable.

Oracle Unbreakable Enterprise Kernels are not vulnerable.

We believe that non-Linux guests are not vulnerable, as we are not
aware of any with an analogous bug.

MITIGATION
==========

Running only HVM guests will avoid this issue.

Not enabling hugetlbfs use, by not altering the boot time default value
of zero in /proc/sys/vm/nr_hugepages (which can only be written by the
root user) will avoid this issue.

It is possible that disabling (or not enabling) the "panic on OOPS"
behavior (via use of the "oops=panic" command line option or the
"panic_on_oops" sysctl) will also avoid this issue, by limiting the
effect to an application crash.  We are not currently sure whether
this is an effective mitigation, as we are not sure whether any locks
or mutexes are held at the point of the crash.

External References:

http://xenbits.xen.org/xsa/advisory-174.html

Acknowledgements:

Name: the Xen project