7.2 - RHSA-2008:0885

Problem Description:

Updated kernel packages that fix various security issues and several bugs are now available for Red Hat Enterprise Linux 5.

This update has been rated as having important security impact by the Red Hat Security Response Team.

2. Relevant releases/architectures:

Red Hat Enterprise Linux Desktop (v. 5 client) - i386, noarch, x86_64 Red Hat Enterprise Linux (v. 5 server) - i386, ia64, noarch, ppc, s390x, x86_64

3. Description:

The kernel packages contain the Linux kernel, the core of any Linux operating system.

Security fixes:

* a missing capability check was found in the Linux kernel do_change_type routine. This could allow a local unprivileged user to gain privileged access or cause a denial of service. (CVE-2008-2931, Important)

* a flaw was found in the Linux kernel Direct-IO implementation. This could allow a local unprivileged user to cause a denial of service. (CVE-2007-6716, Important)

* Tobias Klein reported a missing check in the Linux kernel Open Sound System (OSS) implementation. This deficiency could lead to a possible information leak. (CVE-2008-3272, Moderate)

* a deficiency was found in the Linux kernel virtual filesystem (VFS) implementation. This could allow a local unprivileged user to attempt file creation within deleted directories, possibly causing a denial of service. (CVE-2008-3275, Moderate)

* a flaw was found in the Linux kernel tmpfs implementation. This could allow a local unprivileged user to read sensitive information from the kernel. (CVE-2007-6417, Moderate)

Bug fixes:

* when copying a small IPoIB packet from the original skb it was received in to a new, smaller skb, all fields in the new skb were not initialized. This may have caused a kernel oops.

* previously, data may have been written beyond the end of an array, causing memory corruption on certain systems, resulting in hypervisor crashes during context switching.

* a kernel crash may have occurred on heavily-used Samba servers after 24 to 48 hours of use.

* under heavy memory pressure, pages may have been swapped out from under the SGI Altix XPMEM driver, causing silent data corruption in the kernel.

* the ixgbe driver is untested, but support was advertised for the Intel 82598 network card. If this card was present when the ixgbe driver was loaded, a NULL pointer dereference and a panic occurred.

* on certain systems, if multiple InfiniBand queue pairs simultaneously fell into an error state, an overrun may have occurred, stopping traffic.

* with bridging, when forward delay was set to zero, setting an interface to the forwarding state was delayed by one or possibly two timers, depending on whether STP was enabled. This may have caused long delays in moving an interface to the forwarding state. This issue caused packet loss when migrating virtual machines, preventing them from being migrated without interrupting applications.

* on certain multinode systems, IPMI device nodes were created in reverse order of where they physically resided.

* process hangs may have occurred while accessing application data files via asynchronous direct I/O system calls.

* on systems with heavy lock traffic, a possible deadlock may have caused anything requiring locks over NFS to stop, or be very slow. Errors such as "lockd: server [IP] not responding, timed out" were logged on client systems.

* unexpected removals of USB devices may have caused a NULL pointer dereference in kobject_get_path.

* on Itanium-based systems, repeatedly creating and destroying Windows guests may have caused Dom0 to crash, due to the "XENMEM_add_to_physmap" hypercall, used by para-virtualized drivers on HVM, being SMP-unsafe.

* when using an MD software RAID, crashes may have occurred when devices were removed or changed while being iterated through. Correct locking is now used.

* break requests had no effect when using "Serial Over Lan" with the Intel 82571 network card. This issue may have caused log in problems.

* on Itanium-based systems, module_free() referred the first parameter before checking it was valid. This may have caused a kernel panic when exiting SystemTap.

Red Hat Enterprise Linux 5 users are advised to upgrade to these updated packages, which contain backported patches to resolve these issues.

4. Solution:

Before applying this update, make sure that all previously-released errata relevant to your system have been applied.

This update is available via Red Hat Network. Details on how to use the Red Hat Network to apply this update are available at http://kbase.redhat.com/faq/FAQ_58_10188

5. Bugs fixed (http://bugzilla.redhat.com/):

426081 - CVE-2007-6417 tmpfs: restore missing clear_highpage (kernels from 2.6.11 up) 447913 - LTC43854-trap 700 Program check on uli05, pc: c000000000323910: .skb_under_panic+0x50/0x68 [rhel-5.2.z] 454388 - CVE-2008-2931 kernel: missing check before setting mount propagation 455768 - Guest OS install causes host machine to crash 456235 - [RHEL5] Kernel panic triggered by smbd 456946 - Silent memory corruption with xpmem 457484 - ixgbe panics system when installing RHEL 5.2 with 82598AT (copper 10 gig) adapter 457858 - CVE-2008-3275 Linux kernel local filesystem DoS 457995 - CVE-2008-3272 kernel snd_seq_oss_synth_make_info leak 458779 - LTC44570-Event Queue overflow on eHCA adapters 458783 - lost packets when live migrating 459071 - LTC41679-IPMI device nodes created in reverse order on multinode systems 459082 - process hangs in async direct IO / possible race between dio_bio_end_aio() and dio_await_one() ? 459083 - deadlock when lockd tries to take f_sema that it already has 459776 - [Stratus 5.2.z bug] kernel NULL pointer dereference in kobject_get_path 459780 - [IA64] Fix SMP-unsafe with XENMEM_add_to_physmap on HVM 460128 - [NEC/Stratus 5.2.z bug] various crashes in md - rdev removed in the middle of ITERATE_RDEV 460509 - SysRq handling issue in serial driver 460639 - kprobes remove causing kernel panic on ia64 with 2.6.18-92.1.10.el5 kernel 461082 - CVE-2007-6716 kernel: dio: zero struct dio with kzalloc instead of manually