[LU-6357] Kernel update [RHEL6.6 2.6.32-504.12.2.el6]

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.8.0
• Severity: 3
• Assignee: Bob Glossman (Inactive)
• Reporter: Bob Glossman (Inactive)

Description

• It was found that the Linux kernel's Infiniband subsystem did not
  properly sanitize input parameters while registering memory regions from
  user space via the (u)verbs API. A local user with access to a
  /dev/infiniband/uverbsX device could use this flaw to crash the system or,
  potentially, escalate their privileges on the system. (CVE-2014-8159,
  Important)

• A flaw was found in the way the Linux kernel's splice() system call
  validated its parameters. On certain file systems, a local, unprivileged
  user could use this flaw to write past the maximum file size, and thus
  crash the system. (CVE-2014-7822, Moderate)

• A flaw was found in the way the Linux kernel's netfilter subsystem
  handled generic protocol tracking. As demonstrated in the Stream Control
  Transmission Protocol (SCTP) case, a remote attacker could use this flaw to
  bypass intended iptables rule restrictions when the associated connection
  tracking module was not loaded on the system. (CVE-2014-8160, Moderate)

• It was found that the fix for CVE-2014-3601 was incomplete: the Linux
  kernel's kvm_iommu_map_pages() function still handled IOMMU mapping
  failures incorrectly. A privileged user in a guest with an assigned host
  device could use this flaw to crash the host. (CVE-2014-8369, Moderate)

Bug fixes:

• The maximum amount of entries in the IPv6 route table
  (net.ipv6.route.max_size) was 4096, and every route towards this maximum
  size limit was counted. Communication to more systems was impossible when
  the limit was exceeded. Now, only cached routes are counted, which
  guarantees that the kernel does not run out of memory, but the user can now
  install as many routes as the memory allows until the kernel indicates it
  can no longer handle the amount of memory and returns an error message.

In addition, the default "net.ipv6.route.max_size" value has been increased
to 16384 for performance improvement reasons. (BZ#1177581)

• When the user attempted to scan for an FCOE-served Logical Unit Number
  (LUN), after an initial LUN scan, a kernel panic occurred in
  bnx2fc_init_task. System scanning for LUNs is now stable after LUNs have
  been added. (BZ#1179098)

• Under certain conditions, such as when attempting to scan the network for
  LUNs, a race condition in the bnx2fc driver could trigger a kernel panic in
  bnx2fc_init_task. A patch fixing a locking issue that caused the race
  condition has been applied, and scanning the network for LUNs no longer
  leads to a kernel panic. (BZ#1179098)

• Previously, it was not possible to boot the kernel on Xen hypervisor in
  PVHVM mode if more than 32 vCPUs were specified in the guest configuration.
  Support for more than 32 vCPUs has been added, and the kernel now boots
  successfully in the described situation. (BZ#1179343)

• When the NVMe driver allocated a namespace queue, it indicated that it
  was a request-based driver when it was actually a block I/O-based driver.
  Consequently, when NVMe driver was loaded along with a request-based dm
  device, the system could terminate unexpectedly or become unresponsive when
  attempting to access data. The NVMe driver no longer sets the
  QUEUE_FLAG_STACKABLE bit when allocating a namespace queue and
  device-mapper no longer perceives NVMe driver as request-based; system
  hangs or crashes no longer occur. (BZ#1180555)

• If a user attempted to apply an NVRAM firmware update when running the
  tg3 module provided with Red Hat Enterprise Linux 6.6 kernels, the update
  could fail. As a consequence, the Network Interface Card (NIC) could stay
  in an unusable state and this could prevent the entire system from booting.
  The tg3 module has been updated to correctly apply firmware updates.
  (BZ#1182903)

• Support for key sizes of 256 and 192 bits has been added to AES-NI.
  (BZ#1184332)

Bugs fixed (https://bugzilla.redhat.com/):

1156518 - CVE-2014-8369 kernel: kvm: excessive pages un-pinning in kvm_iommu_map error path
1163792 - CVE-2014-7822 kernel: splice: lack of generic write checks
1181166 - CVE-2014-8159 kernel: infiniband: uverbs: unprotected physical memory access
1182059 - CVE-2014-8160 kernel: iptables restriction bypass if a protocol handler kernel module not loaded