Bug Fixes

BZ#576540

On Red Hat Enterprise Linux, Apache Tomcat initscripts should be located in the /etc/rc.d/init.d directory. However, the comman initscript was previously located in the /etc/init.d directory due to a mistake in the package specs file. With this update, the specs file has been updated and the conman script is located in the /etc/rc.d/init.d directory along with other initscripts as expected.

BZ#847288

When a web application used its own class loader, a deadlock in Tomcat WebappClassLoader could occur when compiling JSPs due to a synchronization bug. This update fixes the synchronization bug and external class loaders no longer interfere with WebappClassLoader.

BZ#798617

The service status returned an incorrect tomcat6 status when TOMCAT_USER in the /etc/tomcat6/tomcat6.conf file was changed to a user whose UID differed from the user GID due to incorrect logic in retrieving the process details. With this update, the code has been modified and the correct service status is now returned in this scenario.

BZ#785954

When Tomcat attempted to import a non-existing page with JavaScript fragments in the URL parameters, it returned a message that the resource was not available. This update adds HTML filtering to Tomcat and the servlet container now correctly returns the message that the resource is missing in this scenario.

Bug Fixes

BZ#746656

The trace-cmd extract command read a buffer multiple times even after an EOF condition. Consequently, the output of the trace-cmd command contained duplicate data. With this update, the trace-cmd utility has been modified to respect the EOF condition and avoid duplication of data in its output.

BZ#879792

When using the latency tracer, the start_threads() function was not called. Calling the stop_threads() function without first calling start_threads() caused the trace-cmd record command to terminate with a segmentation fault because PIDs were not initialized. Consequently, the trace.dat file was not generated. With this update, stop_threads() is not called unless start_threads() is called first. As a result, the segmentation fault no longer occurs.

Enhancement

BZ#838746

Previously, the trace-cmd record command was able to filter ftrace data based on a single PID only. With this update, multiple PIDs can be specified by using the "-P" option.

Bug Fixes

BZ#907856

Previously, the ktune service did not save readahead values. On startup, it multiplied the current value by a constant and divided the value by the same constant on stop. This could result in a wrong value being set on devices that were added after ktune had been started. Now, the previous readahead values are stored for all devices and the correct values are restored on ktune stop.

BZ#907768

Previously, when multiple devices were added into the system, a udev rule restarted ktune for each new device. This could lead to many restarts in a short period of time. The multiple restarts could trigger a race condition in the kernel, which cannot be currently fixed. The tuned daemon code has been modified not to trigger more than one restart per 10 seconds, thus preventing the race condition from occurring.

Bug Fixes

BZ#714180

Red Hat Enterprise Linux 6.1 and later enters processor power-saving states more aggressively. This could result in a small performance penalty on certain workloads. With this update, the pmqos-static.py daemon has been added to the tuned packages, which allows to set the requested latency using the kernel Power Management QoS interface. It is run when the "latency-performance" profile is activated and it sets cpu_dma_latency=0, which keeps the CPU in C0 state, thus making the system as responsive as possible.

BZ#784308

When the ELEVATOR_TUNE_DEVS option was set to a disk device in the /etc/sysconfig/ktune file instead of providing a disk scheduler control file, the scheduler setting was not written to a disk scheduler control file but directly into the disk device file. Consequently, contents of the disk could become corrupted. With this update, the value of ELEVATOR_TUNE_DEVS is checked and only the disk scheduler control file is allowed for writing. As a result, an invalid value of ELEVATOR_TUNE_DEVS is detected in the described scenario so that the disk contents damage can be prevented.

BZ#801561

When the tuned daemon run with the "enterprise-storage" profile enabled and a non-root, non-boot disk partition from a device with write-back cache was mounted, tuned remounted the partition with the "nobarriers" option. If a power failure occurred at that time, the file system could become corrupted. With this update, tuned can detect usage of write-back cache on devices communicating with kernel via SCSI. In these cases, "nobarriers" is now disabled, thus preventing this bug in the described scenario.

BZ#845336

Previously, when the tuned service was started, the tuned PID file was created with world-writable permissions. This bug has been fixed and the /var/run/tuned/tuned.pid file is now created with correct permissions as expected.

BZ#847445

On a machine with hot-plug disk devices with the "enterprise-storage" profile activated, a new disk device could be added into the system, or the disk could be removed and inserted back. In such a scenario, the scheduler and read-ahead settings from the profile were not applied on the newly-added disks. With this update, a new udev rule has been added, which restarts the ktune daemon whenever a new disk device is added, thus fixing this bug.

BZ#887355

The transparent hugepage kernel thread could interfere with latency-sensitive applications. To lower the latency, the transparent hugepages are now disabled in the latency-performance tuned profile.

BZ#886956

Previously, non-root, non-boot partitions were re-mounted using the "nobarrier" option to improve performance. On virtual guests, this could lead to data corruption if power supply was suddenly interrupted, because there was usually a host cache in transfer. This bug has been fixed and the virtual-guest profile no longer re-mounts partitions using "nobarrier".

Bug Fixes

BZ#784697

Previously, the /dev/disk/by-id file contained all expected symbolic links for cciss devices, but only one device link was present in the /dev/disk/by-path/ directory. This bug has been fixed and this file now contains all symbolic links as expected.

BZ#790321

The udev(7) man page did not document the hex encoding of blacklisted characters used in the device or symbolic link names. This update adds a paragraph about character encoding into the SYMLINK section of the udev(7) man page.

BZ#829188

Due to a bug in the binutils linker, the libudev library lost the ExecShield (GNU_RELRO) section and was not protected by the ExecShield security mechanism. This update contains libudev with the ExecShield (GNU_RELRO) section included.

BZ#838451

When using multipath devices, the udev utility tried to make a UUID symbolic link for all the different paths, but only the first one succeeded. Consequently, udev wrote several "File exists" error messages to the system log. This update provides a patch to change these messages from being logged as an error message to being logged as an informational message.

BZ#847925

When no medium was inserted in a drive, the cdrom_id utility, which is an udev helper tool, could not read DV and CD-ROM drive profiles. Consequently, the udev properties for the device node of the drive may not have contained all properties describing the capabilities of the drive, which could prevent other software using the udev database from offering all functionality for the drive. This bug has been fixed and all udev properties for the drive, which cdrom_id detects via drive's properties, are now stored for the device as expected.

Enhancement

BZ#826396

Previously, kernel messages showed device names instead of persistent device names provided by udev. As a consequence, device names could point to different devices every boot. This enhancement adds a new feature, which stores the mapping of device names, such as sda or sdb, and persistent device names to kernel messages.

Bug Fixes

BZ#834560

Due to a bug in the libusbredirhost library, handling of timeouts for bulk transfers did not work correctly. Consequently, traffic of USB ACM serial port devices, such as PSTN modems and SmartCard readers, could not be properly redirected. With this update, no timeout is set on the usb-host side for these devices and the traffic redirection works as expected.

BZ#855737

The usbredir code was allocating an unlimited amount of write buffers. Consequently, when a USB webcam produced data faster then it could be written out, the write queue grew boundlessly and the remote-viewer utility used an enormous amount of RAM. The underlying source code has been modified so that usbredir now checks how large the write queue is and drops isochronous data packets when the queue is too long.

Enhancement

BZ#842316

Support for live migration of SPICE USB redirection requires support for state serialization. This update adds this missing support to the libusbredirparser library.

Security Fix

CVE-2013-0157

An information disclosure flaw was found in the way the mount command reported errors. A local attacker could use this flaw to determine the existence of files and directories they do not have access to.

Bug Fixes

BZ#790728

Previously, the blkid utility ignored swap area UUIDs if the first byte was zero. As a consequence, the swap areas could not be addressed by UUIDs; for example, from the /etc/fstab file. The libblkd library has been fixed and now swap partitions are labeled with a valid UUID value if the first byte is zero.

BZ#818621

Previously, the lsblk utility opened block devices to check if the device was in read-only mode, although the information was available in the /sys file system. This resulted in unexpected SELinux alerts and unnecessary open() calls. Now, the lsblk utility does not perform unnecessary opening operations and no longer reads the information from the /sys file system.

BZ#736245

On a non-uniform CPU configuration, for example on a system with two sockets with a different number of cores, the lscpu command failed unexpectedly with a segmentation fault and a core dump was generated. After this update, when executing the lscpu command on such a configuration, the correct result is printed and no core dump is generated.

BZ#837935

On a system with a large number of active processors, the lscpu command failed unexpectedly with a segmentation fault and a core dump was generated. This bug is now fixed and the lscpu command now works as expected on this configuration.

BZ#819945

Executing the hwclock --systz command to reset the system time based on the current time zone caused the clock to be incorrectly adjusted by one hour. This was because hwclock did not adjust the system time during boot according to the "warp clock" semantic described in the settimeofday(2) man page. With this update, hwclock correctly sets the system time when required.

BZ#845477

When SElinux options were specified both in the /etc/fstab file and on the command line, mounting failed and the kernel logged the following error upon running dmesg:

SELinux:  duplicate or incompatible mount options

The handling of SElinux options has been changed so that options on the command line now replace options given in the /etc/fstab file and as a result, devices can be mounted successfully.

BZ#845971

Due to a change in the search order of the mount utility, while reading the /etc/fstab file, the mount command returned a device before a directory. With this update, the search order has been modified and mount now works as expected.

BZ#858009

Previously, any new login or logout sequence by a telnet client caused the /var/run/utmp file to increase by one record on the telnetd machine. As a consequence, the /var/run/utmp file grew without a limit. As a result of trying to search though a huge /var/run/utmp file, the machine running telnetd could experience more severe side-effects over time. For example, the telnetd process could become unresponsive or the overall system performance could degrade. The telnetd now creates a proper record in /var/run/utmp before starting the logging process. As a result, the /var/run/utmp does not grow without a limit on each new login or logout sequence of a telnet session.

BZ#730891, BZ#783514, BZ#809139, BZ#820183, BZ#839281

Man pages of several utilities included in the package have been updated to fix minor mistakes and add entries for previously undocumented functionalities.

Enhancements

BZ#719927

A new --compare option for hwclock to compare the offset between system time and hardware clock has been added due to a discontinued distribution of adjtimex in Red Hat Enterprise Linux 6.0 and later, which had previously provided this option.

BZ#809449

The lsblk command now supports a new option, --inverse, used to print dependencies between block devices in reverse order. This feature is required to properly reboot or shut down systems with a configured cluster.

BZ#823008

The lscpu utility, which displays detailed information about the available CPUs, has been updated to include numerous new features. Also, a new utility, chcpu, has been added, which allows the user to change the CPU state (online or offline, standby or active, and other states), disable and enable CPUs, and configure specified CPUs. For more information about these utilities, refer to the lscpu(1) and chcpu(8) man pages.

Bug Fixes

BZ#730303

When running a large program under valgrind, the "Valgrind: FATAL: VG_N_SEGNAMES is too low." error messages could be returned. With this update, the compile time constants have been increased (VG_N_SEGMENTS to 50000, VG_N_SEGNAMES to 25000) and these errors no longer occur.

BZ#862795

Previously, the valgrind gdbserver did not properly report exit or a fatal-signal process termination to the gdb debugger. Consequently, the "Remote connection closed" error messages were returned. This bug has been fixed in the code and the process termination is now properly reported in gdb.

BZ#816244

On IBM S/390 architecture, valgrind could report a "Conditional jump or move depends on uninitialized value(s)" warning message for the tsearch() function in glibc. This update includes a standard suppression for these warning messages, which are no longer reported.

Enhancement

BZ#672959

The embedded gdbserver has been added to allow integration of valgrind with the gdb debugger.

Bug Fix

BZ#840087

Previously, an attempt to boot a Red Hat Enterprise Virtualization Hypervisor ISO in a virtual machine was unsuccessful. The boot menu appeared but then stopped responding. The underlying source code has been modified and the virtual machine now works as expected in the described scenario.

Bug Fixes

BZ#750421

Prior to this update, a Windows Server 2003 guest could become suspended when rebooting if the balloon size was changed before, due to a lack of free memory. This update releases the memory to the guest before processing the power management request.

BZ#760022

Prior to this update, the virtio-win floppy disk did not contain NDIS drivers for Windows XP and Windows 7 platforms due to a lack of free space on VFD media. This update modifies the underlying code and switches to 2.88 MB media instead of 1.44 MB.

BZ#768795

Prior to this update, the work items processed the inflate and deflate requests. As a consequence, a stop error could occur when several requests were executed simultaneously. This update uses a dedicated thread instead of work items to process the inflate and deflate requests in sequence.

BZ#805423

Prior to this update, the port surprise-removal handler did not stop and purge the write and read queues. As a consequence, requests could be send to already removed devices. This update modifies the underlying code to stop and purge the write and read queues as expected and requests are no longer sent to removed devices.

BZ#807967, BZ#875155

Prior to this update, the initialization sequence of the virtio-net driver did not work properly after disabling and enabling virtio-net, or when resetting the power management. As a consequence, the first packets that were sent through the DHCP client could, under certain circumstances, become suspended in the queue and the DHCP client did not receive the IP address. With this update, the initialization sequence has been fixed and virtio-net now works as expected in the described scenario.

BZ#814896

Prior to this update, the virtio queue was not correctly reinitialized during the resume routine. As a consequence, ports could not handle the read requests correctly. This update adds the correct virtual queue for re-initialization when resuming after hibernation.

BZ#815295

On Microsoft Windows 7 operating system, a driver disregarded platform requests to indicate only a certain number of packets during one DPC (Deferred Procedure Call). As a consequence, the Windows Hardware Quality Labs (WHQL) test failed and the platform did not moderate the driver workload for the RX path. This update modifies the underlying code to implement packet indication moderation and the WHQL certification now passes in the described scenario.

BZ#824814

Prior to this update, the viostor driver did not handle configuration change events as expected. Consequently, when a relevant image was resized on-line, viostor left it unattended. This update modifies the underlying code to reset the bus sequence when changing the configurations and the driver can now recognize that media has been resized.

BZ#831570

Prior to this update, the work items processed the inflate and deflate requests. As a consequence, the inflate and deflate requests could be executed simultaneously with PnP and Power management (PM) handlers. This update uses a dedicated thread instead of work items to process the PnP and PM requests only after all other pending requests are completed.

BZ#839143

Prior to this update, the balloon driver failed to keep the current balloon size between hibernation-resume cycles and restarts. This update keeps the current balloon size between restarts and hibernation-resume cycles and adjusts the balloon size according to this value.

Enhancements

BZ#782268

This update introduces the vioscsi.sys driver to virtio-win packages to provide virtio-scsi functionality to Microsoft Windows platforms.

BZ#828275

This update adds support for the virtio control queue to offload RX filtering to the host.

BZ#834175

This update supports all possible offload combinations and offload parities between IPv4 and IPv6 for Windows certification 2012 and Windows 8 certification.

BZ#838005

This update adds offloads for IPv6 to virtio packages.

BZ#908163

This update adds virtual floppy drive drivers for Windows Server 2008 R2 guests to virtio packages.

Bug Fixes

BZ#802639

Previously, the live migration dialog box of the virt-manager tool incorrectly described the unit of bandwidth as "Mbps" instead of "MB/s". With this update, the migration dialog has been changed to provide correct information on bandwidth units.

BZ#824275

Prior to this update, an unnecessary reboot occurred after the virt-manager tool created a new guest virtual machine by importing an existing disk image. With this update, a backported upstream patch has been provided, and virt-manager no longer restarts after importing an existing disk image.

BZ#872611

Due to differences in dependency solving between the yum and rpm programs, the virt-manager package failed to update from "noarch" to newer architecture version. With this update, a patch has been provided to mark the noarch version as obsolete. As a result, the noarch package can now be updated without complications.

Enhancement

BZ#878946

The "Delete Associated storage files" option is enabled by default in the virt-manager tool. A warning message is displayed prior to file deletion to notify the user about this configuration.

Bug Fixes

BZ#807176

Prior to this update, the "-o" (sort) option was not properly described in the output of the "virt-top --help" command. Four of the possible sort parameters were not mentioned in the description. This bug has been fixed and the full range of sort parameters is now shown in the virt-top --help message.

BZ#834208

Previously, the column names of the virt-top summary table were not explained in the virt-top man page. The manual page has been updated, and the headings are now properly documented in the "COLUMN HEADINGS" section.

Enhancements

BZ#825627

The copyright information has been updated in the virt-top man pages and help documents.

BZ#835547

This update adds a separate man page for the processcsv.py script, which was previously documented only in the virt-top man page.

BZ#841759

With this update, the "virt-top -1" command has been enhanced to separately display the usage of virtual CPUs. Two numbers are now shown under each domain column; the first is the percentage of the physical CPU used by the domain and the hypervisor together, the second is the percentage used just by the domain. This information is important for performance tuning and other tasks.

Bug Fixes

BZ#794680

The virt-v2v packages used to rename block devices in various guest configuration files during conversion, including the /etc/fstab file. Consequently, the virt-v2v utility returned a redundant warning message when a guest's /etc/fstab file referenced to the /etc/fd0 file as the block device did not know it. To fix this bug, warning messages concerning floppy devices have been explicitly suppressed and virt-v2v no longer returns warning messages in this situation.

BZ#803629

When reading a libvirt guest, virt-v2v uses libvirt metadata to determine the on-disk format, considering only those of "dir", "fs", and "netfs" types as meaningful. If a guest used a different type of storage pool, virt-v2v interpreted these data as a format type, which was unable to convert by libvirt guests. To address this bug, virt-v2v now only uses volume format metadata from storage pools of type "dir", "fs", and "netfs", but also all other storage pools can only hold raw data, so the format is assumed to be "raw". As a result, virt-v2v can now convert libvirt guests using any supported storage pool type.

BZ#838057

When creating a new libvirt guest, virt-v2v failed to disable caching for disks as recommended. As a consequence, guests created by virt-v2v used caching for their disks, unless explicitly disabled by the user after conversion. To address this bug, virt-v2v now explicitly disables caching for all disks when creating a new libvirt guest, and guests created by virt-v2v now have caching disabled for all disks. The user can enable it again if required after conversion.

BZ#868405

Virt-v2v failed when attempting to perform an on-disk format conversion when reading a guest using the libvirtxml input method. A patch has been provided to fix this bug and virt-v2v can now perform format conversions on guests using libvirtxml.

Enhancement

BZ#682945

With this update, virt-v2v can do an on-disk format conversion while converting a remote libvirt guest. Note that when doing this kind of format conversion, virt-v2v must make an intermediate copy of the guest storage data on the conversion server. Other types of conversion do not require any intermediate storage on the conversion server. The user must ensure that the TMPDIR temporary directory has sufficient space for this intermediate copy.

Bug Fixes

BZ#814150

The remote-viewer and the virt-viewer tools, both use the same constant to print their usage message. Consequently, when the user used an unknown command option with the remote-viewer command, the error message referred to the virt-viewer --help command instead of the remote-viewer --help command. With this update, the remote-viewer and virt-viewer code has been modified so that the commands now return the correct error message when used with an unknown option.

BZ#822794

When connected to a guest using the virt-viewer -v command and the console was closed, the command prompt was printed at the end of the last line instead of the new line. This update fixes this bug and the command prompt is printed correctly.

BZ#832121

If the XML listen attribute contained a string consisting of the colon (":") and zero ("0") characters, virt-viewer did not treat the string as a wildcard address and did not create an appropriate remote host address as expected. Consequently, an attempt to connect to a remote host with such an address led to the connection failure. This update modifies the underlying source code to treat the aforementioned characters as wildcards and virt-viewer now successfully connects to a remote host in the described scenario.

BZ#854318

Due to changes in the latest upstream version of the spice-gtk packages, virt-viewer stopped working with the new spice-gtk module. With this update, the virt-viewer packages have been rebuilt to work properly with this new version of spice-gtk.

BZ#856610

Previously, the automatic window resize option did not work correctly with the remote-viewer client. When disabling and then re-enabling the automatic window size, the resized window was smaller then expected. This update provides a patch to fix this bug and the automatic window resize option now works properly.

BZ#856678

Within certain non-US keyboard layouts, keyboard shortcuts using the "Alt" key and another character worked even if they were disabled in a virtual machine. This update applies a patch that fixes this bug and keyboard shortcuts are now disabled as expected.

BZ#864929

Previously, when the virt-viewer client was sized to the full screen, the virt-viewer size resolution could not be set to a higher resolution than the monitor's native resolution. With this update, the user is now able to set a higher resolution than the monitor's native resolution.

BZ#867248

When connecting to a SPICE guest and the user input an incorrect graphic password first, a later attempt to connect using the correct password was unsuccessful, and the virt-viewer tool terminated unexpectedly. This update modifies the underlying code so that virt-viewer no longer crashes in the described scenario.

BZ#867459

When connecting to the Red Hat Enterprise Virtualization portal and the remote-viewer client was started from the XPI plug-in, the client terminated unexpectedly with a segmentation fault. This update modifies the underlying code and applies a patch to fix this bug so that remote-viewer now works as expected in this situation.

BZ#881020

Previously, when using remote-viewer to display multiple screens of a virtual machine with multiple physical displays, under certain circumstances, remote-viewer could display only one screen in single remote-viewer window and the other screens were disconnected. With this update, the underlying code has been modified so that all physical displays are now properly displayed in the respective remote-viewer windows.

Enhancements

BZ#828339

This enhancement provides the new --title option which allows the user to specify a title displayed in the remote-viewer window title bar.

BZ#842305

With this update, the virt-viewer tool supports the SpiceMonitorsConfig display message.

BZ#865793

The virt-viewer tool is now able to handle requests from the Red Hat Enterprise Virtualization portal to enable or disable passing of the Ctrl+Alt+Delete key combination to the guest operating system.

BZ#875126

Screenshots can be currently saved only in the PNG format. With this update, the ".png" suffix is automatically added to the screenshot file name if is it missing.

Enhancement

BZ#829427

This enhancement adds support for the Virtage hardware partitioning system to the virt-what utility.

Bug Fixes

BZ#825215

Previously, when running the virt-who service, unregistering a Red Hat Enterprise Virtualization Hypervisor host from the Subscription Asset Manager (SAM) server caused the service to be terminated with the following message:

SubscriptionManagerError: No such file or directoryError in communication with candlepin, trying to recoverUnable to read certificate, system is not registered or you are not root

Only the last line of the aforementioned message should have been displayed. This bug has been fixed, and the traceback errors are now saved to the log file and not printed on the screen.

BZ#866890

When a snapshot of a virtual machine (VM) was created in Microsoft Hyper-V Server, the virt-who agent replaced the UUID of the VM file with the UUID of the snapshot. This bug has been fixed, and the UUID is not changed in the described case. Additionally, in certain cases, the virt-who agent running with the "--hyperv" command-line option terminated with the following message:

AttributeError: HyperV instance has no attribute 'ping'

This bug has been fixed and the aforementioned error no longer occurs.

BZ#869960

Previously, the virt-who agent failed to function correctly when a URL, which was set in the VIRTWHO_ESX_SERVER parameter, was missing the initial "https://" string. With this update, virt-who has been modified, and "https://" is no longer required in VIRTWHO_ESX_SERVER.

Enhancements

BZ#808060

With this update, the virt-who agent has been modified to start as a foreground process and to print error messages or debugging output (the "-d" command-line option) to the standard error output. Moreover, the following command-line options have been enhanced: the "-o" option provides the one-shot mode and exits after sending the list of guests; the "-b" option and the "service virt-who start" command equivalently start on the background and send data to the /var/log/ directory.

BZ#846788

The virt-who agent has been modified to support Red Hat Enterprise Virtualization Manager polling.

BZ#860854

With this update, the virt-who agent has been modified to correctly recognize guest virtual machines, which are installed on top of Microsoft Hyper-V Server.

BZ#868149

The virt-who manual pages and the output of the "virt-who --help" command have been enhanced with clarifying information. In addition, a typographical error has been corrected in both texts.

Bug Fix

BZ#852332

Due to the broken %postun scriptlet, an attempt to uninstall wdaemon caused an error message to appear. This error message also occurred during the wdaemon update because the old package is removed during the update. Consequently, the wdaemon service was not restarted after the update. The %postun scriptlet has been fixed and wdaemon works as expected in this situation.

Enhancement

BZ#838752

This enhancement adds support for emulation of the Wacom Intuos5 tablet series.

BZ#754168

Prior to this update, the wget package contained a redundant URL to the wget upstream project. This update modifies the specification file to list the correct http://www.gnu.org/software/wget/.

BZ#814208

Prior to this update, the wget utility did not previously work as intended with the "-T, --timeout" option set when http server did not answer the SSL handshake. Wget source code has been patched, to ensure that wget aborts the connection when using --timeout option correctly.

BZ#714893

Prior to this update, the wget utility source code was lacking check of the HTTP response parsing function return value. In some cases, when HTTP response header was malformed (fuzzed), the parsing function returned error. Because the returned value was not checked, it then resulted in Segmentation Fault. This update adds check of the HTTP response parsing function return value in the wget source code. Now when HTTP response header is malformed (fuzzed) and the parsing function returns error, the following error message is thrown and wget retries the request:

2012-10-01 10:13:44 ERROR -1: Malformed status line.

Bug Fixes

BZ#813579

When roaming from one Access Point (AP) to another and the connection was disrupted, NetworkManager did not always automatically reconnect. This update includes a number of backported upstream patches to improve Proactive Key Caching (PKC), also known as Opportunistic Key Caching (OKC). As a result, WPA connections now roam more reliably.

BZ#837402

Previously, the supplicant would attempt to roam to slightly stronger access points, increasing the chance of a disconnection. This bug has been fixed and the supplicant now only attempts to roam to a stronger access point when the current signal is significantly degraded.

Enhancement

BZ#672976

The "wpa_gui" program was removed from "wpa_supplicant" in the 6.0 release as per BZ#553349, however, the man page was still being installed. This upgrade removes the man page.

Bug Fix

BZ#801139

Prior to this update, the x3270 emulator failed to support the double-byte character set (DBCS). As a consequence, the character sets option for Japanese was disabled. This update modifies the underlying code to enable DBCS and adds the icu packages as a dependency. Now, Japanese character sets are again available.

Bug Fix

BZ#860454

With Red Hat Enterprise Linux 6.4, XFS has been enhanced to allow the use of the 32-bit project quota ID feature. However, the top 16 bits of a 32-bit project quota ID were not properly saved and restored using the xfsdump and xfsrestore utilities. This caused the data to be saved and restored with an incorrect 16-bit project quota ID. With this update, the underlying source code has been fixed so that all 32 bits of the project quota ID are properly saved and restored by xsfdump and xfsrestore.

Bug Fixes

BZ#730433

When the manual geometry of the mkfs.xfs utility was specified for striping as well as calculating of the allocation group counts and size, mkfs.xfs could emit confusing error messages on failure. With this update, more standardized and informative error messages are returned.

BZ#836433

When the sector size was not specified by the "-f" option, the mkfs.xfs utility used the 512 byte sector size by default even for drives with 4 Kb physical sectors. With this update, mkfs.xfs correctly recognizes the sector size in the described scenario, which fixes this bug.

BZ#878859

When attempting to set a 32-bit quota project ID on an XFS file system which did not have this feature enabled, the command returned success, but truncated the project ID to the lower 16 bits. With this update, a project ID of more than 16 bits cannot be set unless the 32-bit project ID feature is enabled.

Enhancement

BZ#827186

With this update, mkfs.xfs can enable 32-bit project quota IDs on a file system with the "-i projid32bit=1" parameter specified. Without this parameter, mkfs.xfs defaults to 16-bit project quota IDs. The 32-bit project quota IDs can be enabled on existing file systems by using the "xfs_admin -p" command.

Security Fix

CVE-2012-0862

When xinetd services are configured with the "TCPMUX" or "TCPMUXPLUS" type, and the tcpmux-server service is enabled, those services are accessible via port 1. It was found that enabling the tcpmux-server service (it is disabled by default) allowed every xinetd service, including those that are not configured with the "TCPMUX" or "TCPMUXPLUS" type, to be accessible via port 1. This could allow a remote attacker to bypass intended firewall restrictions.

Bug Fixes

BZ#790036

Prior to this update, a file descriptor array in the service.c source file was not handled as expected. As a consequence, some of the descriptors remained open when xinetd was under heavy load. Additionally, the system log was filled with a large number of messages that took up a lot of disk space over time. This update modifies the xinetd code to handle the file descriptors correctly and messages no longer fill the system log.

BZ#809271

Prior to this update, services were disabled permanently when their CPS limit was reached. As a consequence, a failed bind operation could occur when xinetd attempted to restart the service. This update adds additional logic that attempts to restart the service. Now, the service is only disabled if xinetd cannot restart the service after 30 attempts.

Bug Fixes

BZ#883578

Due to overlapping memory areas, remote-viewer became unresponsive after a migration of a guest playing a video. This update adjusts the monitors_config pointer to fix this issue, and migration of a guest, which is displaying video, works as expected.

BZ#896005

This update disables "surfaces" by default due to a performance regression with the rendering support.

Bug Fixes

BZ#859851

Due to a bug in the input driver, covering the Expresskeys on the Wacom Intuos5 graphics tablet caused a spurious stylus jump to the upper left corner (0,0). This bug has been fixed and the described issue no longer occurs.

BZ#862939

Previously, the xorg.conf configuration file with two devices containing the same input node caused a double free error and subsequent failure of the X server. With this update, xorg.conf has been fixed, and the server crash is now prevented.

Enhancements

BZ#838751

With this update, support for the Wacom Intuos5 series graphics tablets has been added to the xorg-x11-drv-wacom package.

BZ#857088

With this update, support for the Wacom Cintiq 22HD series graphics tablets has been added to the xorg-x11-drv-wacom package.

Bug Fixes

BZ#608076

When the GNOME sound volume applet was configured to pop up after pressing the "mute", "volume up", or "volume down" hardware buttons, doing so caused a graphical glitch to appear in a dual monitor configuration. Now, the screen glitch no longer appears.

BZ#745033

When spice-client was opened in full-screen mode, the client screen contained a static image which was not refreshed until it was switched back to window mode. Now, the static image no longer appears when opened in full-screen mode.

BZ#816347

When the screen saver started to fade, pressing keys did not interrupt the fade and did not immediately display the unlock screen. Now, pressing keys stops the screen from fading.

BZ#829321

A NULL pointer dereference caused X.Org to terminate unexpectedly with a segmentation fault on certain servers. The error is fixed and X.Org no longer crashes on those servers.

BZ#837073

An invalid pointer dereference in the server caused the server to unexpectedly terminate with a segmentation fault when the mouse was moved over the VNC window. Crashes no longer occur when moving the mouse over the VNC window.

BZ#853236

The KVM process could not access the X server because the "/usr/bin/Xorg" binary was unreadable for non-root users. Now, all users can read the binary and KVM guests can access host operating systems.

BZ#858005

A transformation matrix is used to bind a device to a specific area on the screen. An uninitialized device transformation matrix caused the pointer to jump to the top-left corner of the screen on some devices. With this update, the transformation matrix is properly initialized and pointer device movement works as expected.

BZ#863913

An X Input Extension (XI 1.x) grab on a disabled device led to a NULL pointer dereference error which caused the server to terminate unexpectedly. Currently, the XI 1.x grab functions normally and the X server no longer crashes.

BZ#864054

When screens are reconfigured, the server updates some internal fields to adjust input device coordinate scaling if the device is bound to a specific screen. The NVIDIA binary driver did not have access to these internal methods, and was not able to update these fields when it changed output configurations. A new API is now exported for the driver and the NVIDIA driver is now able to update the server-internal fields.

BZ#868054

Pointer screen crossings for non-Xinerama setups caused the mouse pointer to wrap around on the first screen instead of moving to the second screen. Now, the mouse pointer can move between both screens on non-Xinerama setups.

BZ#883206

Running xrestop on servers that used Intel, ATI or Nouveau drivers caused the server to terminate unexpectedly with a segmentation fault. Now, users are able to run xrestop on those servers without crashes.

Bug Fix

BZ#871579

Prior to this update, the yaboot loader used by default a maximum block size of 512 bytes in the fdisk partition table. As a consequence, yaboot could not load a kernel in a disk that was formatted using 4 kilobytes partitions. This update extends the MAX_BLOCK_SIZE value to 4 kilobytes to allow for disks that use the advanced format.

Enhancement

BZ#822657

This update adds VLAN Tag support for network boot and installation to allow multiple VLANs in a bridged network to share the same physical network link but maintain isolation.

Bug Fix

BZ#647495

Prior to this update, ypbind started too late in the boot sequence, which caused problems in some environments, where it needed to be started before netfs. This update changes the priority of the ypbind service. Now, ypbind starts as expected.

Bug Fixes

BZ#790812

Prior to this update, the NIS server was returning "0" (YP_FALSE) instead of "-1" (YP_NOMAP) after a request for a database not present in the server's domain. This behavior caused the autofs mount attempts to fail on Solaris clients. With this update, the return value has been fixed and the autofs mounts no longer fail on Solaris clients.

BZ#816981

Previously, when the crypt() function returned NULL, the yppasswd utility did not properly recognize the return value. This bug has been fixed, and the NULL return values of crypt() are now recognized and reported correctly by yppaswd.

BZ#845283

Previously, the ypserv utility allocated large amounts of virtual memory when parsing XDR requests, but failed to free that memory in case the request was not parsed successfully. Consequently, memory leaks occurred. With this update, a patch has been provided to free the already allocated memory when parsing of a request fails. As a result, the memory leaks no longer occur.

BZ#863952

Previously, the yppush(8) man page did not describe how to change settings of the yppush utility. The manual page has been amended to specify that the settings can be changed in the /var/yp/Makefile file.

Bug Fixes

BZ#789092

Previously, yum-rhn-plugin ignored the timeout value set for yum. In some scenarios with slow networking, this could cause yum to timeout when communicating with Red Hat Network. Now, yum-rhnplugin abides by the timeout set for all yum repositories.

BZ#802636

Previously, the check-update utility could in certain cases incorrectly return a 0 error code if an error occurred. With this update, "1" is returned if an error occurs.

BZ#824193

Prior to this update, applying automatic updates with the yum-rhn-plugin utility on Red Hat Enterprise Linux 6 system could fail with an "empty transaction" error message. This was because the cached version of yum-rhn-plugin metadata was not up-to-date. With this update, yum-rhn-plugin downloads new metadata if available, ensuring that all packages are available for download.

BZ#830219

Previously, the messaging in yum-rhn-plugin was specific only to Red Hat Network Classic scenarios. This update clarifies what source yum-rhn-plugin is receiving updates from to reduce confusion.

BZ#831234

Prior to this update, yum-rhn-plugin did not correctly try the alternate server URLs provided if the first option failed. This update ensures that fail-over situations are handled correctly.

Bug Fixes

BZ#674756

When running the yum localinstall command, various requires, obsoletes, and conflicts situations were not handled properly and resulted in inconsistent package installations using different Yum commands. The underlying source code has been modified and Yum resolves all the aforementioned situations properly.

BZ#727553

When trying to execute the yum update --skip-broken command on the command line, the package dependency resolution never ended. This bug is now fixed and dependencies are resolved successfully after executing the yum update --skip-broken command.

BZ#802462

After creating a new yum history file, the yum history stats command failed with a traceback instead of reporting an actual error. This bug is now fixed and when the yum history stats command fails after creating a new yum history file, it displays an error message.

BZ#815568

Previously, when running the yum makecache command, followed by the yum -C updateinfo command, the second command failed to execute because although the updateinfo file had been downloaded by yum makecache it was uncompressed and treated as unavailable by yum -C updateinfo. This bug is now fixed and yum -C updateinfo works as expected in this scenario.

BZ#834159

Previously, when trying to install an obsoleted package from a repository, Yum reported Nothing to do instead of providing an "obsolete" error message. This bug is now fixed and Yum now correctly warns users about obsolete packages.

BZ#840543

Previously, when the yum upgrade command failed to execute, Yum displayed a misleading Protected multilib versions error message instead of the accurate one. This bug is now fixed and if Yum fails, it displays the correct error message.

BZ#872518

When Yum was executed by a regular user, Yum downloaded metadata even if the "root" metadata were up-to-date. This bug is now fixed and Yum does not download unnecessary data if the "root" metadata are up-to-date.

BZ#809117

A typo in the yum(8) man page has been fixed.

BZ#878335

After a rebase update of the createrepo utility, execution of the createrepo --update command took significantly longer. This update reduces the time for executing the createrepo --update command.

BZ#737173

Previously, when the yum updateinfo command, provided by the yum-security plug-in, was used, Yum did not merge the version information from multiple repositories. This could prevent the latest version of a package that was present in multiple repositories to not be installed. Now, when installing packages from multiple repositories, Yum installs only the latest packages available.

BZ#819522

Previously, when trying to reinstall an unavailable package and the execution failed, the exit code had the value of 0. This bug is now fixed and when reinstallation of an unavailable package fails, it returns an exit code with the value of 1.

BZ#820674

Previously, when the yum-debug-restore command was used to restore multiple installonly packages, Yum tried to keep a limit of packages that were installed simultaneously and removed packages that were present in the system. Also, Yum restored multiple packages but assumed that just one would be installed. Yum's installonly_limit configuration now determines what to install and remove correctly when multiple items are installed at once. This is most noticeable when using commands like yum shell and yum-debug-restore.

BZ#858844

When using the yum.yumBase().update() function to specify a package name, version, and/or release of a certain package, the function terminated and failed to update the aforementioned variables. This bug is now fixed and the yum.yumBase().update() function can be used successfully to specify a package name, version, and release.

BZ#868840

Previously, when trying to resolve dependencies while updating packages that had dependencies, Yum entered a loop and no packages were installed after execution of the yum update command. After this update, the yum update command now handles packages with obsoleting dependencies as expected.

BZ#880968

Due to an incorrect prioritization of actions performed by Yum after entering a command with a syntactically incorrect subcommand, Yum performed a series of unnecessary actions before acknowledging the typo. This bug has been fixed and Yum performs an immediate syntax check in the described scenario.

BZ#887935

When updateinfo.xml was generated via the update_md.UpdateNotice() method, the yum API only accounted for the "issued date" element and ignored the updated date element. Now, the yum API accounts for the "updated date" element and the "updated date" element is displayed in the XML file.

BZ#885159

Previously, users were not notified that different certificate files with the same basename were treated as identical, which could lead to various problems. With this update, Yum checks certificate files for such duplicates and displays an error message appropriately.

Enhancements

BZ#684859

Yum plug-ins are now able to set exit codes on any Yum operations.

BZ#744335

With this update, yum-cron is now documented in the yum-cron(8) man page.

BZ#748054

Support for the installonlypkgs functionality in rhev-hypervisor packages has been added.

Bug Fix

BZ#754694

Due to missing information about the zlib version, some applications using zlib could not work properly. The zlib.map version script, which provides version information, has been added to the underlying source code and zlib now works as expected.

Enhancement

BZ#823007

This enhancement optimizes the zlib compression library for IBM System z.