4.51. kernel

Security Fixes

CVE-2013-6381, Important

A buffer overflow flaw was found in the way the qeth_snmp_command() function in the Linux kernel's QETH network device driver implementation handled SNMP IOCTL requests with an out-of-bounds length. A local, unprivileged user could use this flaw to crash the system or, potentially, escalate their privileges on the system.

CVE-2013-4483, Moderate

A flaw was found in the way the ipc_rcu_putref() function in the Linux kernel's IPC implementation handled reference counter decrementing. A local, unprivileged user could use this flaw to trigger an Out of Memory (OOM) condition and, potentially, crash the system.

CVE-2013-4554, Moderate

It was found that the Xen hypervisor implementation did not correctly check privileges of hypercall attempts made by HVM guests, allowing hypercalls to be invoked from protection rings 1 and 2 in addition to ring 0. A local attacker in an HVM guest able to execute code on privilege levels 1 and 2 could potentially use this flaw to further escalate their privileges in that guest. Note: Xen HVM guests running unmodified versions of Red Hat Enterprise Linux and Microsoft Windows are not affected by this issue because they are known to only use protection rings 0 (kernel) and 3 (userspace).

CVE-2013-6383, Moderate

A flaw was found in the way the Linux kernel's Adaptec RAID controller (aacraid) checked permissions of compat IOCTLs. A local attacker could use this flaw to bypass intended security restrictions.

CVE-2013-6885, Moderate

It was found that, under specific circumstances, a combination of write operations to write-combined memory and locked CPU instructions may cause a core hang on certain AMD CPUs (for more information, refer to AMD CPU erratum 793 linked in the References section). A privileged user in a guest running under the Xen hypervisor could use this flaw to cause a denial of service on the host system. This update adds a workaround to the Xen hypervisor implementation, which mitigates the AMD CPU issue. Note: this issue only affects AMD Family 16h Models 00h-0Fh Processors. Non-AMD CPUs are not vulnerable.

CVE-2013-7263, Low

It was found that certain protocol handlers in the Linux kernel's networking implementation could set the addr_len value without initializing the associated data structure. A local, unprivileged user could use this flaw to leak kernel stack memory to user space using the recvmsg, recvfrom, and recvmmsg system calls.

CVE-2013-2929, Low

A flaw was found in the way the get_dumpable() function return value was interpreted in the ptrace subsystem of the Linux kernel. When 'fs.suid_dumpable' was set to 2, a local, unprivileged local user could use this flaw to bypass intended ptrace restrictions and obtain potentially sensitive information.

Bug Fixes

BZ#1044328

Due to a bug in the cifs module, the calculation of the number of virtual circuits was handled incorrectly when establishing SMB sessions. As a consequence in environments with multiple TCP connections between the same SMB client and SMB server, each time a TCP connection was established, all other TCP connections from the client to the server were reset, resulting in an endless loop. With this update, the number of virtual circuits is constantly set to 1, which ensures the correct behavior of the cifs module in this situation.

BZ#1050097

Certain storage device or storage environment failures could cause all SCSI commands and task management functions that were sent to a SCSI target to time out, without any other indication of an error. As a consequence, the Linux SCSI error handling code stopped issuing any I/O operations on the entire HBA adapter until the recovery operations completed. Additionally when using DM Multipath, I/O operations did not fail over to a working path in this situation. To resolve this problem, a new sysfs parameter, "eh_deadline", has been added to the SCSI host object. This parameter allows to set the maximum amount of time for which the SCSI error handling attempts to perform error recovery before resetting the entire HBA adapter. This timeout is disabled by default. The default value of this timeout can be reset for all SCSI HBA adapters on the system using the "eh_deadline" kernel parameter. The described scenario no longer occurs if eh_deadline is properly used.

BZ#1051535

A previous change that corrected a bug preventing communication between NICs using be2net introduced a memory leak in the be2net transmitter (Tx) code path. The memory leak has been fixed by applying a series of patches that corrects handling of socket buffers (SKBs) in the Tx code path.

Enhancement

BZ#1054055

Support for a kernel symbol that allows printing a binary blob of data as a hex dump to syslog has been added to kABI (Kernel Application Binary Interface).

Security Fix

CVE-2012-4398, Moderate

It was found that a deadlock could occur in the Out of Memory (OOM) killer. A process could trigger this deadlock by consuming a large amount of memory, and then causing request_module() to be called. A local, unprivileged user could use this flaw to cause a denial of service (excessive memory consumption).

Bug Fixes

BZ#995961

A recent patch fixing a problem that prevented communication between NICs using the be2net driver caused the firmware of NICs to become unresponsive, and thus triggered a kernel panic. The problem was caused by unnecessary usage of a hardware workaround that allows skipping VLAN tag insertion. A patch has been applied and the workaround is now used only when the multi-channel configuration is enabled on the NIC. Note that the bug only affected the NICs with firmware version 4.2.xxxx.

BZ#987539

A race condition in the be_open function in the be2net driver could trigger the BUG_ON() macro, which resulted in a kernel panic. A patch addressing this problem has been applied and the race condition is now avoided by enabling polling before enabling interrupts globally. The kernel no longer panics in this situation.

BZ#987244, BZ#978305

Due to a segment register that was not reset after transition to protected mode, a bug could have been triggered in certain older versions of the upstream kernel (the kernel 3.9 - 3.9.4), preventing a guest system from booting and rendering it unresponsive on certain Intel Virtualization Technology (VT) hardware. On the newer kernels, this behavior had a significant impact on the booting speed of virtual machines. This update applies a patch providing early segment setup for the VT feature which allows executing VT under WMware and KVM. Guest machines no longer hang on boot and booting process is now significantly faster when using 64-bit Intel hardware with the VT feature enabled.

BZ#981337

Due to a bug in the networking stack, the kernel could attempt to deference a NULL pointer if a VLAN was configured on top of a GRE tunnel and network packets were transmitted, which resulted in a kernel panic. A patch has been applied to fix this bug by modifying the net driver to test a VLAN hardware header for a NULL value properly. The kernel no longer panics in this scenario.

BZ#967053

A kernel panic could occur in the XEN hypervisor due to a race in the XEN's tracing infrastructure. The race allows an idle vCPU to attempt to log a trace record while another vCPU executes a hypercall to disable the active tracing: for example, when using the xenmon.py performance monitoring utility. To avoid triggering the panic, the respective BUG_ON() routine call in the trace code has been replaced with a simple test condition. The XEN hypervisor no longer crashes due to aforementioned race condition.

BZ#965359

Due to a bug in memory management, a kernel thread process could become unresponsive for a significant amount of time, waiting for a quota of dirty pages to be met and written out, which caused a kernel panic. With this update, memory management allows processes to break out of the throttle loop if there are no more dirty pages available to be written out. This prevents a kernel panic from occurring in this situation.

BZ#957604

A previous change in the port auto-selection code allowed sharing ports with no conflicts extending its usage. Consequently, when binding a socket with the SO_REUSEADDR socket option enabled, the bind(2) function could allocate an ephemeral port that was already used. A subsequent connection attempt failed in such a case with the EADDRNOTAVAIL error code. This update applies a patch that modifies the port auto-selection code so that bind(2) now selects a non-conflict port even with the SO_REUSEADDR option enabled.

BZ#950137

Due to a bug in the be2net driver, events in the RX, TX, and MCC queues were not acknowledged before closing the respective queue. This could cause unpredictable behavior when creating RX rings during the subsequent queue opening. This update applies a patch that corrects this problem and events are now acknowledged as expected in this scenario.

BZ#948317

Incorrect locking around the cl_state_owners list could cause the NFSv4 state reclaimer thread to enter an infinite loop while holding the Big Kernel Lock (BLK). This consequently caused the NFS client to become unresponsive. With this update, safe list iteration is used, which prevents the client from hanging in this scenario.

BZ#947732

When handling requests from Intelligent Platform Management Interface (IPMI) clients, the IPMI driver previously used two different locks for an IPMI request. If two IPMI clients sent their requests at the same time, each request could receive one of the locks and then wait for the second lock to become available. This resulted in a deadlock situation and the system became unresponsive. The problem could occur more likely in environments with many IPMI clients. This update modifies the IPMI driver to handle the received messages using tasklets so the driver now uses a safe locking technique when handling IPMI requests and the mentioned deadlock can no longer occur.

BZ#928098

A bug in the autofs4 mount expiration code could cause the autofs4 module to falsely report a busy tree of NFS mounts as "not in use". Consequently, automount attempted to unmount the tree and failed with a "failed to umount offset" error, leaving the mount tree to appear as empty directories. A patch has been applied to remove an incorrectly used autofs dentry mount check and the aforementioned problem no longer occurs.

BZ#924011

Previously, the xdr routines in NFS version 2 and 3 conditionally updated the res->count variable. Read retry attempts after a short NFS read() call could fail to update the res->count variable, resulting in truncated read data being returned. With this update, the res->count variable is updated unconditionally, thus preventing this bug.

BZ#918592

Previously, the NFS Lock Manager (NLM) did not resend blocking lock requests after NFSv3 server reboot recovery. As a consequence, when an application was running on a NFSv3 mount and requested a blocking lock, the application received an -ENOLCK error. This patch ensures that NLM always resend blocking lock requests after the grace period has expired.

BZ#907524

Previously, the be2net code expected the last word of an MCC completion message from the firmware to be transferred by direct memory access (DMA) at once. However, this is not always true, and could therefore cause the BUG_ON() macro to be triggered in the be_mcc_compl_is_new() function, consequently leading to a kernel panic. The BUG_ON() macro has been removed, and a kernel panic no longer occurs in this scenario.

BZ#906909

When a process is opening a file over NFSv4, sometimes an OPEN call can succeed while the following GETATTR operation fails with an NFS4ERR_DELAY error. The NFSv4 code did not handle such a situation correctly and allowed an NFSv4 client to attempt to use the buffer that should contain the GETATTR information. However, the buffer did not contain the valid GETATTR information, which caused the client to return a "-ENOTDIR" error. Consequently, the process failed to open the requested file. This update backports a patch that adds a test condition verifying validity of the GETATTR information. If the GETATTR information is invalid, it is obtained later and the process opens the requested file as expected.

BZ#905190

The IPv4 code did not correctly update the Maximum Transfer Unit (MTU) of the designed interface when receiving ICMP Fragmentation Needed packets. Consequently, a remote host did not respond correctly to ping attempts. With this update, the IPv4 code has been modified so the MTU of the designed interface is adjusted as expected in this situation. The ping command now provides the expected output.

BZ#901547

The size of the buffer used to print the kernel taint output on kernel panic was too small, which resulted in the kernel taint output not being printed completely sometimes. With this update, the size of the buffer has been adjusted and the kernel taint output is now displayed properly.

BZ#894636

Previously, the Generic Receive Offload (GRO) functionality was not enabled by default for VLAN devices. Consequently, certain network adapters, such as Emulex Virtual Fabric Adapter (VFA) II, that use be2net driver, were dropping packets when VLAN tagging was enabled and the 8021q kernel module loaded. This update applies a patch that enables GRO by default for VLAN devices.

BZ#885125

Certain recent Intel input/output memory management unit (IOMMU) systems reported very large numbers of supported mapping domains. Consequently, if the number was too large, booting a system with the intel_iommu kernel parameter enabled (intel_iommu=on) failed with the following error message:

Allocating domain array failed.

With this update, a limit of 4000 domains is set to avoid the described problems.

BZ#881885

Previously, the Xen kernel used the memory size found at the "0x40e" address as the beginning of the Extended BIOS Data Area (EBDA). However, this is not valid on certain machines, such as Dell PowerEdge R710, which caused the system to become unresponsive during boot on these machines. This update modifies the kernel to use the multiboot structure to acquire the correct location of EBDA and the system boot now proceeds as expected in this scenario.

BZ#878316

Previously, race conditions could sometimes occur in interrupt handling on the Emulex BladeEngine 2 (BE2) controllers, causing the network adapter to become unresponsive. This update provides a series of patches for the be2net driver, which prevents the race from occurring. The network cards using BE2 chipsets no longer hang due to incorrectly handled interrupt events.

BZ#878209

Due to a regression introduced by a recent update of the be2net driver, 10Gb NICs configured to use multiple receive queues across multiple CPUs were restricted to use a single receive queue on a single CPU. This resulted in significant performance degradation. With this update, the be2net driver has been corrected to provide support for multiple receive queues on 10Gb NICs as expected.

BZ#877474

A previous change in the tg3 driver corrected a bug causing DMA read engine of the Broadcom BCM5717 Ethernet controller to initiate multiple DMA reads across the PCIe bus. However, the original bug fix used the CHIPREV_ID_5717_A0 macro which is more restrictive so that the DMA read problem was not fixed for the Broadcom BCM5718 Ethernet controller. This update modifies the code to use the ASIC_REV_5717 macro, which corrects the original bug properly.

BZ#876587

The code to print the kernel taint output contained a typographical error. Consequently, the kernel taint output, which is displayed on kernel panic, could not provide taint error messages for unsupported hardware. This update fixes the typo and the kernel taint output is now displayed correctly.

BZ#872531

The cxgb4 driver previously did not clear data structures used for firmware requests. Consequently, when initializing some Chelsio's Terminator 4 (T4) adapters, a probe request could fail because the request was incompatible with the adapter's firmware. This update modifies the cxgb4 driver to properly initialize firmware request structures before sending a request to the firmware and the problem no longer occurs.

BZ#865095

The memory management code specific to the AMD64 and Intel 64 architectures previously did not contain proper memory barriers in the smp_invalidate_interrupt() routine. As a consequence, CPUs on AMD64 and Intel 64 systems containing modulo 8 number of CPUs (8, 16, 24 and so on) could sometimes heavily compete for spinlock resources, spending most of the CPU time by attempts to acquire spinlocks. Such systems could therefore rarely appear to be unresponsive with a very slow computing progress. This update applies a patch introducing proper memory barriers in the smp_invalidate_interrupt() routine so the problem can no longer occur.

BZ#864648

Previously, the kernel's futex wait code used timeouts that had granularity in milliseconds. Also, when passing these timeouts to system calls, the kernel converted the timeouts to "jiffies". Consequently, programs could time out inaccurately which could lead to significant latency problems in certain environments. This update modifies the futex wait code to use a high-resolution timer (hrtimer) so the timeout granularity is now in microseconds. Timeouts are no longer converted to "jiffies" when passed to system calls. Timeouts passed to programs are now accurate and the programs time out as expected.

BZ#862865

A boot-time memory allocation pool (the DMI heap) is used to keep the list of Desktop Management Interface (DMI) devices during the system boot. Previously, the size of the DMI heap was only 2048 bytes on the AMD64 and Intel 64 architectures and the DMI heap space could become easily depleted on some systems, such as the IBM System x3500 M2. A subsequent OOM failure could, under certain circumstances, lead to a NULL pointer entry being stored in the DMI device list. Consequently, scanning of such a corrupted DMI device list resulted in a kernel panic. The boot-time memory allocation pool for the AMD64 and Intel 64 architectures has been enlarged to 4096 bytes and the routines responsible for populating the DMI device list have been modified to skip entries if their name string is NULL. The kernel no longer panics in this scenario.

BZ#862520

Due to a bug in the be2net driver, the receive completion queue (RX CQ) could report completions with an incorrect fragment ID (frag_idx). This triggered a BUG_ON() macro that resulted in a kernel panic. A patch has been applied to the be2net driver ensuring that partially coalesced CQ entries are properly flushed when completion coalescing is enabled on a CQ. The kernel no longer panics in this situation.

BZ#859194

The generic allocator (genalloc) could, under certain circumstances, incorrectly allocate memory for the gen_pool structure. This could result in memory corruption where genalloc attempted to set the bits it had not allocated. A patch has been applied that ensures proper byte allocation and the memory corruption problem no longer occurs when allocating a generic memory pool.

BZ#853145

Previously, an NFS client could sometimes cache negative dentries until the page cache was flushed or the directory listing operation was performed on the parent directory. As a consequence, an incorrect dentry was never normally revalidated and a stat call always failed, providing incorrect results. This was caused by an incorrect resolution of an attribute indicating a cache change (cache_change_attribute) along with insufficient flushing of cached directories. A series of patches has been backported to resolve this problem so the cache_change_attribute is now updated properly and the cached directories are flushed more readily.

BZ#845447

Previously, when hot-unplugging a USB serial adapter device, the USB serial driver did not properly clean up used serial ports. Therefore, when hot-plugging the USB serial device again, the USB serial driver allocated new port IDs instead of using previously used ports. This update modifies the USB serial driver to clean up open ports correctly so that the ports can be reused next time the device is plugged in.

BZ#843473

With Red Hat Enterprise Linux 5.9, a patch that fixed IGMP reporting bug in a network bridge was backported to the bonding code from Red Hat Enterprise Linux 6. However, two other patches related to the problem were not included. This update backports these patches from Red Hat Enterprise Linux 6. Specifically, the first patch fixing a NULL pointer deference that could occur if the master bond was not a network bridge. The patch adds a testing condition which prevents the code from dereferencing a NULL pointer. The second patch introduces a hook that allows to identify which bridge port is used for the master bridge interface and modifies the bonding code to use new functions to determine whether the used bond is a network bridge.

BZ#839839

Under certain circumstances, a race between certain asynchronous operations, such as "silly rename" and "silly delete", and the invalidate_inodes() function could occur when unmounting an NFS file system. Due to this race, the system could become unresponsive, or a kernel oops or data corruption could occur if an inode was removed from the list of inodes while the invalidate_inodes() function performed an iteration on the inode. This update modifies the NFS code to wait until the asynchronous operations are finished before performing inode clean-up. The race condition no longer occurs and an NFS file system is unmounted as expected.

BZ#839334

Previously on system boot, devices with associated Reserved Memory Region Reporting (RMRR) information had lost their RMRR information after they were removed from the static identity (SI) domain. Consequently, a system unexpectedly terminated in an endless loop due to unexpected NMIs triggered by DMA errors. This problem was observed on HP ProLiant Generation 7 (G7) and 8 (Gen8) systems. This update prevents non-USB devices that have RMRR information associated with them from being placed into the SI domain during system boot. HP ProLiant G7 and Gen8 systems that contain devices with the RMRR information now boot as expected.

BZ#831330

Previously, GFS2 did not properly free directory hash table memory from cache when the directory was removed from cache. If the same GFS2 inode was later reused as another directory, the stale directory hash table was reused instead of reading the correct information from the media. If the GFS2 hash table was not reused, a small amount of memory was lost until the next reboot. If the hash table was reused, the directory could become corrupt. Later, GFS2 could discover the file system inconsistency and withdraw from the file system, making it unavailable until the system was rebooted. This update applies a patch to the kernel that frees the directory hash table correctly from cache and prevents this file system corruption.

BZ#795550

The qla2xxx driver creates optrom and optrom_ctl files in sysfs which are used by some tools such as the scli command line tool from QLogic. However, the functions which implement these pseudo-files have race conditions. It will crash the kernel when multiple tools using these files at the same time. Users can work around this issue by making sure only 1 such process is running at a given point of time.

BZ#731531

Switching the FPU context was not properly handled in certain environments, such as systems with multi-core AMD processors using the 32-bit kernel. When running multiple instances of the applications using the FPU frequently, data corruption could occur because processes could often be restored with the context of another instance. This update applies series of patches that modifies the kernel's FPU behavior: the "lazy" FPU context switch is temporarily disabled after 5 consecutive context switches using the FPU, and restored again after the context is switched 256 times. The aforementioned data corruption problem no longer occurs.

BZ#595184

Previously, if a target sent multiple local port logout (LOGO) events, the fc_rport_work() function in the Fibre Channel library module (libfc) tried to process all of them, irrespective of the status of processing prior to the LOGO events. Consequently, fc_rport_work() terminated unexpectedly with a stack trace. This update simplifies the remote port (rport) restart logic by making the decision to restart after deleting the transport rport. Now, all I/O operations run as expected and fc_rport_work() no longer crashes in the described scenario.

BZ#918952

Previously, the NFSv3 server incorrectly converted 64-bit cookies to 32-bit. Consequently, the cookies became invalid, which affected all file system operations depending on these cookies, such as the READDIR operation that is used to read entries from a directory. This led to various problems, such as exported directories being empty or displayed incorrectly, or an endless loop of the READDIRPLUS procedure which could potentially cause a buffer overflow. This update modifies knfsd code so that 64-bit cookies are now handled correctly and all file system operations work as expected.

Enhancements

BZ#796912

The ALSA HDA audio driver has been updated to support Creative Recon3D audio cards.

BZ#873514

The "unregister_lro_netdev" and "register_lro_netdev" kABI symbols have been added to the kernel. These symbols allow Large Receive Offload (LRO) to be disabled by the kernel stack.

BZ#918279

The Red Hat Enterprise Linux 5.10 kernel includes a new panic_on_io_nmi option (configured using the /proc/sys/kernel/panic_on_io_nmi file), which allows the kernel to panic when a non-maskable interrupt (NMI) occurs that is caused by an I/O error.

BZ#919633

The cciss driver has been updated to the latest version, which adds support for ProLiant servers with the latest HP SAS Smart Array controllers.

Security Fixes

CVE-2012-3511, Moderate

This update fixes the following security issues:

* A use-after-free flaw was found in the madvise() system call implementation in the Linux kernel. A local, unprivileged user could use this flaw to cause a denial of service or, potentially, escalate their privileges.

CVE-2013-4162, Moderate

A flaw was found in the way the Linux kernel's TCP/IP protocol suite implementation handled IPv6 sockets that used the UDP_CORK option. A local, unprivileged user could use this flaw to cause a denial of service.

CVE-2013-2141, Low

An information leak flaw in the Linux kernel could allow a local, unprivileged user to leak kernel memory to user-space.

Bug Fixes

BZ#983864

A bug in the be2net driver prevented communication between NICs using be2net. This update applies a patch addressing this problem along with several other upstream patches that fix various other problems. Traffic between NICs using the be2net driver now proceeds as expected.

BZ#999819

A recent patch fixing a problem that prevented communication between NICs using the be2net driver caused the firmware of NICs to become unresponsive, and thus triggered a kernel panic. The problem was caused by unnecessary usage of a hardware workaround that allows skipping VLAN tag insertion. A patch has been applied and the workaround is now used only when the multi-channel configuration is enabled on the NIC. Note that the bug only affected the NICs with firmware version 4.2.xxxx.

BZ#1001488

A bug in the autofs4 mount expiration code could cause the autofs4 module to falsely report a busy tree of NFS mounts as "not in use". Consequently, automount attempted to unmount the tree and failed with a "failed to umount offset" error, leaving the mount tree to appear as empty directories. A patch has been applied to remove an incorrectly used autofs dentry mount check and the aforementioned problem no longer occurs.

BZ#1005239

A race condition in the be_open function in the be2net driver could trigger the BUG_ON() macro, which resulted in a kernel panic. A patch addressing this problem has been applied and the race condition is now avoided by enabling polling before enabling interrupts globally. The kernel no longer panics in this situation.

Security Fixes

CVE-2012-5515, Moderate

It was found that the Xen hypervisor implementation did not perform range checking on the guest provided values in multiple hypercalls. A privileged guest user could use this flaw to trigger long loops, leading to a denial of service (Xen hypervisor hang).

CVE-2012-1568, Low

It was found that when running a 32-bit binary that uses a large number of shared libraries, one of the libraries would always be loaded at a predictable address in memory. An attacker could use this flaw to bypass the Address Space Layout Randomization (ASLR) security feature.

CVE-2012-4444, Low

A flaw was found in the way the Linux kernel's IPv6 implementation handled overlapping, fragmented IPv6 packets. A remote attacker could potentially use this flaw to bypass protection mechanisms (such as a firewall or intrusion detection system (IDS)) when sending network packets to a target system.

Security Fixes

CVE-2013-2206, Important

This update fixes the following security issues:

A flaw was found in the way the Linux kernel's Stream Control Transmission Protocol (SCTP) implementation handled duplicate cookies. If a local user queried SCTP connection information at the same time a remote attacker has initialized a crafted SCTP connection to the system, it could trigger a NULL pointer dereference, causing the system to crash.

CVE-2013-2224, Important

It was found that the fix for CVE-2012-3552 released via RHSA-2012:1540 introduced an invalid free flaw in the Linux kernel's TCP/IP protocol suite implementation. A local, unprivileged user could use this flaw to corrupt kernel memory via crafted sendmsg() calls, allowing them to cause a denial of service or, potentially, escalate their privileges on the system.

CVE-2013-2232, Moderate

An invalid pointer dereference flaw was found in the Linux kernel's TCP/IP protocol suite implementation. A local, unprivileged user could use this flaw to crash the system or, potentially, escalate their privileges on the system by using sendmsg() with an IPv6 socket connected to an IPv4 destination.

CVE-2013-2164, , Low, CVE-2013-2147, , Low, CVE-2013-2234, , Low, CVE-2013-2237, , Low

Information leak flaws in the Linux kernel could allow a privileged, local user to leak kernel memory to user-space.

Security Fix

CVE-2012-3400, Low

Buffer overflow flaws were found in the udf_load_logicalvol() function in the Universal Disk Format (UDF) file system implementation in the Linux kernel. An attacker with physical access to a system could use these flaws to cause a denial of service or escalate their privileges.

Bug Fixes

BZ#884704

Previously, race conditions could sometimes occur in interrupt handling on the Emulex BladeEngine 2 (BE2) controllers, causing the network adapter to become unresponsive. This update provides a series of patches for the be2net driver, which prevents the race from occurring. The network cards using BE2 chipsets no longer hang due to incorrectly handled interrupt events.

BZ#902683

A boot-time memory allocation pool (the DMI heap) is used to keep the list of Desktop Management Interface (DMI) devices during the system boot. Previously, the size of the DMI heap was only 2048 bytes on the AMD64 and Intel 64 architectures and the DMI heap space could become easily depleted on some systems, such as the IBM System x3500 M2. A subsequent OOM failure could, under certain circumstances, lead to a NULL pointer entry being stored in the DMI device list. Consequently, scanning of such a corrupted DMI device list resulted in a kernel panic. The boot-time memory allocation pool for the AMD64 and Intel 64 architectures has been enlarged to 4096 bytes and the routines responsible for populating the DMI device list have been modified to skip entries if their name string is NULL. The kernel no longer panics in this scenario.

BZ#905829

The size of the buffer used to print the kernel taint output on kernel panic was too small, which resulted in the kernel taint output not being printed completely sometimes. With this update, the size of the buffer has been adjusted and the kernel taint output is now displayed properly.

BZ#885063

The code to print the kernel taint output contained a typographical error. Consequently, the kernel taint output, which is displayed on kernel panic, could not provide taint error messages for unsupported hardware. This update fixes the typo and the kernel taint output is now displayed correctly.

Security Fixes

CVE-2012-6544, , Low, CVE-2012-6545, , Low, CVE-2013-3222, , Low, CVE-2013-3224, , Low, CVE-2013-3231, , Low, CVE-2013-3235, , Low

Information leaks in the Linux kernel could allow a local, unprivileged user to leak kernel memory to user-space.

CVE-2013-0914, Low

An information leak was found in the Linux kernel's POSIX signals implementation. A local, unprivileged user could use this flaw to bypass the Address Space Layout Randomization (ASLR) security feature.

CVE-2013-1929, Low

A heap-based buffer overflow in the way the tg3 Ethernet driver parsed the vital product data (VPD) of devices could allow an attacker with physical access to a system to cause a denial of service or, potentially, escalate their privileges.

Bug Fixes

BZ#957606

Previously on system boot, devices with associated Reserved Memory Region Reporting (RMRR) information had lost their RMRR information after they were removed from the static identity (SI) domain. Consequently, a system unexpectedly terminated in an endless loop due to unexpected NMIs triggered by DMA errors. This problem was observed on HP ProLiant Generation 7 (G7) and 8 (Gen8) systems. This update prevents non-USB devices that have RMRR information associated with them from being placed into the SI domain during system boot. HP ProLiant G7 and Gen8 systems that contain devices with the RMRR information now boot as expected.

BZ#958021

Previously, the kernel's futex wait code used timeouts that had granularity in milliseconds. Also, when passing these timeouts to system calls, the kernel converted the timeouts to "jiffies". Consequently, programs could time out inaccurately which could lead to significant latency problems in certain environments. This update modifies the futex wait code to use a high-resolution timer (hrtimer) so the timeout granularity is now in microseconds. Timeouts are no longer converted to "jiffies" when passed to system calls. Timeouts passed to programs are now accurate and the programs time out as expected.

BZ#966878

A recent change modified the size of the task_struct structure in the floating point unit (fpu) counter. However, on Intel Itanium systems, this change caused the kernel Application Binary Interface (kABI) to stop working properly when a previously compiled module was loaded, resulting in a kernel panic. With this update the change causing this bug has been reverted so the bug can no longer occur.

BZ#971872

The cxgb4 driver previously did not clear data structures used for firmware requests. Consequently, when initializing some Chelsio's Terminator 4 (T4) adapters, a probe request could fail because the request was incompatible with the adapter's firmware. This update modifies the cxgb4 driver to properly initialize firmware request structures before sending a request to the firmware and the problem no longer occurs.

Security Fix

CVE-2013-0153, Moderate

A flaw was found in the way the Xen hypervisor AMD IOMMU driver handled interrupt remapping entries. By default, a single interrupt remapping table is used, and old interrupt remapping entries are not cleared, potentially allowing a privileged guest user in a guest that has a passed-through, bus-mastering capable PCI device to inject interrupt entries into others guests, including the privileged management domain (Dom0), leading to a denial of service.

Bug Fixes

BZ#947736

When a process is opening a file over NFSv4, sometimes an OPEN call can succeed while the following GETATTR operation fails with an NFS4ERR_DELAY error. The NFSv4 code did not handle such a situation correctly and allowed an NFSv4 client to attempt to use the buffer that should contain the GETATTR information. However, the buffer did not contain the valid GETATTR information, which caused the client to return a "-ENOTDIR" error. Consequently, the process failed to open the requested file. This update backports a patch that adds a test condition verifying validity of the GETATTR information. If the GETATTR information is invalid, it is obtained later and the process opens the requested file as expected.

BZ#952098

Previously, the xdr routines in NFS version 2 and 3 conditionally updated the res->count variable. Read retry attempts after a short NFS read() call could fail to update the res->count variable, resulting in truncated read data being returned. With this update, the res->count variable is updated unconditionally so this bug can no longer occur.

BZ#953435

When handling requests from Intelligent Platform Management Interface (IPMI) clients, the IPMI driver previously used two different locks for an IPMI request. If two IPMI clients sent their requests at the same time, each request could receive one of the locks and then wait for the second lock to become available. This resulted in a deadlock situation and the system became unresponsive. The problem could occur more likely in environments with many IPMI clients. This update modifies the IPMI driver to handle the received messages using tasklets so the driver now uses a safe locking technique when handling IPMI requests and the mentioned deadlock can no longer occur.

BZ#954296

Incorrect locking around the cl_state_owners list could cause the NFSv4 state reclaimer thread to enter an infinite loop while holding the Big Kernel Lock (BLK). As a consequence, the NFSv4 client became unresponsive. With this update, safe list iteration is used, which prevents the NFSv4 client from hanging in this scenario.

Security Fixes

CVE-2013-0216, Moderate

A flaw was found in the Xen netback driver implementation in the Linux kernel. A privileged guest user with access to a para-virtualized network device could use this flaw to cause a long loop in netback, leading to a denial of service that could potentially affect the entire system.

CVE-2013-0231, Moderate

A flaw was found in the Xen PCI device back-end driver implementation in the Linux kernel. A privileged guest user in a guest that has a PCI passthrough device could use this flaw to cause a denial of service that could potentially affect the entire system.

CVE-2013-1826, Moderate

A NULL pointer dereference flaw was found in the IP packet transformation framework (XFRM) implementation in the Linux kernel. A local user who has the CAP_NET_ADMIN capability could use this flaw to cause a denial of service.

CVE-2012-6537, Low

Information leak flaws were found in the XFRM implementation in the Linux kernel. A local user who has the CAP_NET_ADMIN capability could use these flaws to leak kernel stack memory to user-space.

CVE-2012-6542, Low

An information leak flaw was found in the logical link control (LLC) implementation in the Linux kernel. A local, unprivileged user could use this flaw to leak kernel stack memory to user-space.

CVE-2012-6546, Low

Two information leak flaws were found in the Linux kernel's Asynchronous Transfer Mode (ATM) subsystem. A local, unprivileged user could use these flaws to leak kernel stack memory to user-space.

CVE-2012-6547, Low

An information leak flaw was found in the TUN/TAP device driver in the Linux kernel's networking implementation. A local user with access to a TUN/TAP virtual interface could use this flaw to leak kernel stack memory to user-space.

Bug Fixes

BZ#923353

The IPv4 code did not correctly update the Maximum Transfer Unit (MTU) of the designed interface when receiving ICMP Fragmentation Needed packets. Consequently, a remote host did not respond correctly to ping attempts. With this update, the IPv4 code has been modified so the MTU of the designed interface is adjusted as expected in this situation. The ping command now provides the expected output.

BZ#923910

Previously, the be2net code expected the last word of an MCC completion message from the firmware to be transferred by direct memory access (DMA) at once. However, this is not always true, and could therefore cause the BUG_ON() macro to be triggered in the be_mcc_compl_is_new() function, consequently leading to a kernel panic. The BUG_ON() macro has been removed from be_mcc_compl_is_new(), and the kernel panic no longer occurs in this scenario.

BZ#924087

Previously, the NFSv3 server incorrectly converted 64-bit cookies to 32-bit. Consequently, the cookies became invalid, which affected all file system operations depending on these cookies, such as the READDIR operation that is used to read entries from a directory. This led to various problems, such as exported directories being empty or displayed incorrectly, or an endless loop of the READDIRPLUS procedure which could potentially cause a buffer overflow. This update modifies knfsd code so that 64-bit cookies are now handled correctly and all file system operations work as expected.

Security Fixes

CVE-2013-0268, Important

A flaw was found in the way file permission checks for the "/dev/cpu/[x]/msr" files were performed in restricted root environments (for example, when using a capability-based security model). A local user with the ability to write to these files could use this flaw to escalate their privileges to kernel level, for example, by writing to the SYSENTER_EIP_MSR register.

CVE-2013-0871, Important

A race condition was found in the way the Linux kernel's ptrace implementation handled PTRACE_SETREGS requests when the debuggee was woken due to a SIGKILL signal instead of being stopped. A local, unprivileged user could use this flaw to escalate their privileges.