[LU-6405] kernel update [RHEL7.1 3.10.0-229.1.2.el7]

• 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 use-after-free flaw was found in the way the Linux kernel's SCTP
  implementation handled authentication key reference counting during INIT
  collisions. A remote attacker could use this flaw to crash the system or,
  potentially, escalate their privileges on the system. (CVE-2015-1421,
  Important)

This update also fixes the following bugs:

• In certain systems with multiple CPUs, when a crash was triggered on one
  CPU with an interrupt handler and this CPU sent Non-Maskable Interrupt
  (NMI) to another CPU, and, at the same time, ioapic_lock had already been
  acquired, a deadlock occurred in ioapic_lock. As a consequence, the kdump
  service could become unresponsive. This bug has been fixed and kdump now
  works as expected. (BZ#1197742)

• On Lenovo X1 Carbon 3rd Gen, X250, and T550 laptops, the thinkpad_acpi
  module was not properly loaded, and thus the function keys and radio
  switches did not work. This update applies a new string pattern of BIOS
  version, which fixes this bug, and function keys and radio switches now
  work as intended. (BZ#1197743)

• During a heavy file system load involving many worker threads, all worker
  threads in the pool became blocked on a resource, and no manager thread
  existed to create more workers. As a consequence, the running processes
  became unresponsive. With this update, the logic around manager creation
  has been changed to assure that the last worker thread becomes a manager
  thread and does not start executing work items. Now, a manager thread
  exists, spawns new workers as needed, and processes no longer hang.
  (BZ#1197744)

• If a thin-pool's metadata enters read-only or fail mode, for example, due
  to thin-pool running out of metadata or data space, any attempt to make
  metadata changes such as creating a thin device or snapshot thin device
  should error out cleanly. However, previously, the kernel code returned
  verbose and alarming error messages to the user. With this update, due to
  early trapping of attempt to make metadata changes, informative errors are
  displayed, no longer unnecessarily alarming the user. (BZ#1197745)

• When running Red Hat Enterprise Linux as a guest on Microsoft Hyper-V
  hypervisor, the storvsc module did not return the correct error code for
  the upper level Small Computer System Interface (SCSI) subsystem. As a
  consequence, a SCSI command failed and storvsc did not handle such a
  failure properly under some conditions, for example, when RAID devices were
  created on top of storvsc devices. An upstream patch has been applied to
  fix this bug, and storvsc now returns the correct error code in the
  described situation. (BZ#1197749)

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

1181166 - CVE-2014-8159 kernel: infiniband: uverbs: unprotected physical memory access
1196581 - CVE-2015-1421 kernel: net: slab corruption from use after free on INIT collisions