[El-errata] ELSA-2011-2029 Important: Oracle Linux 5 Unbreakable Enterprise kernel security update

Mon Oct 10 11:37:14 PDT 2011

Oracle Linux Security Advisory ELSA-2011-2029

The following updated rpms for Oracle Linux 5 have been uploaded to the 
Unbreakable Linux Network:

i386:
kernel-uek-2.6.32-200.20.1.el5uek.i686.rpm
kernel-uek-debug-2.6.32-200.20.1.el5uek.i686.rpm
kernel-uek-debug-devel-2.6.32-200.20.1.el5uek.i686.rpm
kernel-uek-devel-2.6.32-200.20.1.el5uek.i686.rpm
kernel-uek-doc-2.6.32-200.20.1.el5uek.noarch.rpm
kernel-uek-firmware-2.6.32-200.20.1.el5uek.noarch.rpm
ofa-2.6.32-200.20.1.el5uek-1.5.1-4.0.53.i686.rpm
ofa-2.6.32-200.20.1.el5uekdebug-1.5.1-4.0.53.i686.rpm

x86_64:
kernel-uek-firmware-2.6.32-200.20.1.el5uek.noarch.rpm
kernel-uek-doc-2.6.32-200.20.1.el5uek.noarch.rpm
kernel-uek-2.6.32-200.20.1.el5uek.x86_64.rpm
kernel-uek-headers-2.6.32-200.20.1.el5uek.x86_64.rpm
kernel-uek-devel-2.6.32-200.20.1.el5uek.x86_64.rpm
kernel-uek-debug-devel-2.6.32-200.20.1.el5uek.x86_64.rpm
kernel-uek-debug-2.6.32-200.20.1.el5uek.x86_64.rpm
ofa-2.6.32-200.20.1.el5uek-1.5.1-4.0.53.x86_64.rpm
ofa-2.6.32-200.20.1.el5uekdebug-1.5.1-4.0.53.x86_64.rpm

SRPMS:
http://oss.oracle.com/ol5/SRPMS-updates/kernel-uek-2.6.32-200.20.1.el5uek.src.rpm
http://oss.oracle.com/ol5/SRPMS-updates/ofa-2.6.32-200.20.1.el5uek-1.5.1-4.0.53.src.rpm

Users with Oracle Linux Premier Support can now use Ksplice to patch 
against this
Security Advisory.

We recommend that all users of  Oracle Linux 5 install these updates.

Users of Ksplice Uptrack can install these updates by running :

# /usr/sbin/uptrack-upgrade -y

On systems that have "autoinstall = yes" in /etc/uptrack/uptrack.conf,
these updates will be installed automatically and you do not need to
take any additional action.

Description of changes:

* CVE-2011-2898: Information leak in packet subsystem

Uninitialized struct padding in the packet subsystem led to an
information leak of two bytes of kernel memory to userspace.

* CVE-2011-2723: Remote denial of service vulnerability in gro.

The skb_gro_header_slow function in the Linux kernel had a bug which
allowed a remote attacker to put certain gro fields in an inconsistent
state, resulting in a denial of service.

* CVE-2011-2496: Local denial of service in mremap().

Robert Swiecki discovered that mremap() could be abused for local denial of
service by triggering a BUG_ON assert.

* CVE-2011-2484: Denial of service in taskstats subsystem.

The add_del_listener function in kernel/taskstats.c in the Linux kernel
did not prevent multiple registrations of exit handlers, which allowed
local users to cause a denial of service (memory and CPU consumption),
and bypass the OOM Killer, via a crafted application.

* CVE-2011-1833: Information disclosure in eCryptfs.

Vasiliy Kulikov of Openwall and Dan Rosenberg discovered that eCryptfs
incorrectly validated permissions on the requested source directory. A
local attacker could use this flaw to mount an arbitrary directory,
possibly leading to information disclosure.

kernel-uek:

[2.6.32-200.20.1.el5uek]
- af_packet: prevent information leak {CVE-2011-2898}
- gro: Only reset frag0 when skb can be pulled {CVE-2011-2723}
- vm: fix vm_pgoff wrap in stack expansion {CVE-2011-2496}
- vm: fix vm_pgoff wrap in upward expansion {CVE-2011-2496}
- taskstats: don't allow duplicate entries in listener mode {CVE-2011-2484}
- Ecryptfs: Add mount option to check uid of device being mounted 
{CVE-2011-1833}

ofa:

[1.5.1-4.0.53]
- Don't test rds_ib_ring_empty or ring_log without proper locks held

[1.5.1-4.0.52]
- Don't use RCU on the bind hash table to avoid close delays

[1.5.1-4.0.50]
- Fix crashes when rdma_resolve_route fails

[1.5.1-4.0.49]
- Add EL5.6 build support.

[1.5.1-4.0.48]
- RDS: Fix deadlock on congestion updates
- RDS: Fix memory use-after-free on RDS sockets

[1.5.1-4.0.47]
- RDS: disable the arp verify patch

[1.5.1-4.0.46]
- RDS: Destory the rdma id later in connection reset to avoid use after free

[1.5.1-4.0.45]
- RDS: sock close hang on RDS_MSG_MAPPED
- RDS: sk_alloc use GFP_KERNEL
- RDS: bad addr crash

[1.5.1-4.0.44]
- RDS: RDS_MSG_MAPPED race
- RDS: completion shutdown race
- RDS: sock rca hashtable race

[1.5.1-4.0.43]
- RDS: don't resched recv refill unless we've done at least 128 buffers

[1.5.1-4.0.42]
- RDS: resched recv refill as needed (Chris Mason)
- RDS: add rds-specific sock destruct function

[1.5.1-4.0.41]
- IPOIB: work around a possible lost interrupt (Eli 
Cohen<eli at mallanox.co.il>)

[1.5.1-4.0.40]
- RDS: Make sure rds_send_xmit catches pending congestion updates

[1.5.1-4.0.39]
- RDS: don't leak rdma ids that are new

[1.5.1-4.0.38]
- RDS: deal with ENOMEM in rds_ib_recv_refill

[1.5.1-4.0.37]
- RDS: don't destroy the rdma id on half formed connections

[1.5.1-4.0.36]
- RDS: don't trust the LL_SEND_FULL bit

[1.5.1-4.0.35]
- RDS: give up on half formed connections after 15s

[1.5.1-4.0.34]
- Bring back the LL_SEND_FULL bit, otherwise we don't make progress
   on the send queue once the ring empties

[1.5.1-4.0.33]
- Don't schedule inside rds_send_xmit, it is called with locks held

[1.5.1-4.0.32]
- Replace mismatched arp fix with more effective one
- Don't set the LL_SEND_FULL bit when the send ring is full

[1.5.1-4.0.31]
- Fix mismatched arp entries during connection reset
- Fix softlockups in rds_send_xmit

[1.5.1-4.0.29]
- Add fix for enodev errors from sendmsg during resets