SupportConsoleDevelopersStart a trialContact

7. Architecture Specific Support

7.1. i386

  • In a virtual environment, timekeeping for Red Hat Enterprise Linux 64-bit kernels can be problematic, since time is kept by counting timer interrupts. De- and re-scheduling the virtual machine can cause a delay in these interrupts, resulting in a timekeeping discrepancy. This kernel release reconfigures the timekeeping algorithm to keep time based on a time-elapsed counter. (Bugzilla #463573)
  • It was found that, if their stacks exceed the combined size of ~4GB, 64-bit threaded applications slowed down drastically in pthread_create(). This is because glibc uses MAP_32BIT to allocate those stacks. As the use of MAP_32BIT is a legacy implementation, this update adds a new flag (MAP_STACK mmap) to the kernel to avoid constraining 64-bit applications. (Bugzilla #459321)
  • The update includes a feature bit that encourages TSCs to keep running in deep-C states. This bit NONSTOP_TSC acts in conjunction with CONSTANT_TSC. CONSTANT_TSC indicates that the TSC runs at constant frequency irrespective of P/T- states, and NONSTOP_TSC indicates that TSC does not stop in deep C-states. (Bugzilla #474091)
  • This update includes a patch to include asm-x86_64 headers in kernel-devel packages built on or for i386, i486, i586 and i686 architectures. (Bugzilla #491775)
  • This update includes a fix to ensure that specifying memmap=X$Y as a boot parameter on i386 architectures yields a new BIOS map. (Bugzilla #464500)
  • This update adds a patch to correct a problem with the Non-Maskable Interrupt (NMI) that appeared in previous kernel releases. The problem appeared to affect various Intel processors and caused the system to report the NMI watchdog was 'stuck'. New parameters in the NMI code correct this issue. (Bugzilla #500892)
  • This release re-introduces PCI Domain support for HP xw9400 and xw9300 systems. (Bugzilla #474891)
  • Functionality has been corrected to export module powernow-k8 parameters to /sys/modules. This information was previously not exported.(Bugzilla #492010)

7.2. x86_64

  • An optimization error was found in linux-2.6-misc-utrace-update.patch. When running 32-bit processes on a 64-bit machine systems didn't return ENOSYS on missing (out of table range) system calls. This kernel release includes a patch to correct this. (Bugzilla #481682)
  • Some cluster systems were found to boot with an unstable time source. It was determined that this was a result of kernel code not checking for a free performance counter (PERFCTR) when calibrating the TSC (Time Stamp Clock) during the boot process. This resulted, in a small percentage of cases, in the system defaulting to a busy PERFCTR and getting unreliable calibrations.
    A fix was implemented to correct this by ensuring the system checked for a free PERFCTR before defaulting (Bugzilla #467782). This fix, however, cannot satisfy all possible contingencies as it is possible that all PERFCTRs will be busy when required for TSC calibration. Another patch has been included to initiate a kernel panic in the unlikely event (fewer than 1% of cases) that this scenario occurs. (Bugzilla #472523).

7.3. PPC

  • This kernel release includes various patches to update the spufs (Synergistic Processing Units file system) for Cell processors. (Bugzilla #475620)
  • An issue was identified wherein /proc/cpuinfo would list logical PVR Power7 processor architecture as "unknown" when show_cpuinfo() was run. This update adds a patch to have show_cpuinfo() identify Power7 architectures as Power6. (Bugzilla #486649)
  • This update includes several patches that are required to add/improve MSI-X (Message Signaled Interrupts) support on machines using System P processors. (Bugzilla #492580)
  • A patch has been added to this release to enable the functionality of the previously problematic power button on Cell Blades machines. (Bugzilla #475658)

7.4. s390x

Red Hat Enterprise Linux introduces a wide range of new features for IBM System z machines, most notably:
  • Utilizing Named Saved Segments (NSS), the z/VM hypervisor makes operating system code in shared real memory pages available to z/VM guest virtual machines. With this update, multiple Red Hat Enterprise Linux guest operating systems on the z/VM can boot from the NSS and be run from a single copy of the Linux kernel in memory. (BZ#474646)
  • Device driver support has been added in this update for the new IBM System z PCI cryptography accelerators, utilizing the same interfaces as prior versions. (BZ#488496)
  • Red Hat Enterprise Linux 5.4 adds support for processor degradation, which allows processor speed to be reduced in some circumstances (i.e. system overheating). (BZ#474664) This new feature allows automation software to observe the machine state and act based on defined policies.

    Note

    Processor degradation is supported on z990, z890 and later systems and is observed through SCLP system service event type 4 event qualifier 3. STSI will report the new capacity of the processor in the file: /sys/devices/system/cpu/cpuN/capability.
  • Control Program Identification (CPI) descriptive data is used to identify individual systems on the Hardware Management Console (HMC). With this update, CPI data can now be associated with a Red Hat Enterprise Linux instance. (BZ#475820)
    For more information on CPI refer to the Device Drivers, Features, and Commands document
  • Fibre Channel Protocol (FCP) performance data can now be measured on Red Hat Enterprise Linux instances on the IBM System z platform. (BZ#475334) Metrics that are collected and reported on include:
    • Performance relevant data on stack components such as Linux devices, Small Computer System Interface (SCSI) Logical Unit Numbers (LUNs) and Host Bus Adapter (HBA) storage controller information.
    • Per stack component: current values of relevant measurements as throughput, utilization and other applicable measurements.
    • Statistical aggregations (minimum, maximum, averages and histogram) of data associated with I/O requests including size, latency per component and totals.
  • Support has been added to the kernel to issue EMC Symmetrix Control I/O. This update provides the ability to manage EMC Symmetrix storage arrays with Red Hat Enterprise Linux on the IBM System z platform. (BZ#461288)
  • A new feature has been implemented in the kernel to perform an Initial Program Load (IPL) on a Red Hat Enterprise Linux virtual machine immediately following a kernel panic and dump.(BZ#474688)
  • Hardware that supports the configuration topology facility passes the system CPU topology information to the scheduler, allowing it to make load balancing decisions. On machines where I/O interrupts are unevenly distributed, CPUs that are grouped together and get more I/O interrupts than others will tend to have a higher average load, creating performance issues in some cases.
    Previously, CPU topology support was enabled by default. With this update, CPU topology support is disabled by default, and the kernel parameter "topology=on" has been added to allow this feature to be enabled. (BZ#475797)
  • New kernel options can now be added using the IPL command without modifying the content of the CMS parmfile, allowing for temporary overwriting of kernel options that are already provided by the parmfile. The entire boot command line can be replaced with the VM parameter string, bypassing any kernel options from the parmfile. Furthermore, customers can create new Linux Named Saved Systems (NSS) on the CP/CMS command line. (BZ#475530)
  • The qeth driver has been updated with HiperSockets Layer3 support for IPv6. (BZ#475572) For further details on this feature, refer to the "qeth device driver for OSA-Express (QDIO) and HiperSockets" chapter in IBM's "Device Drivers, Features, and Commands" book located at: http://www.ibm.com/developerworks/linux/linux390/october2005_documentation.html
  • Starting with z9 HiperSocket firmware returns the version string in a different format. This change resulted in missing mcl_level information in the qeth status message issued during online setting of the device. The updated qeth driver now correctly reads the new version string format of HiperSockets, allowing for a standardization of output format. (BZ#479881)
  • In Red Hat Enterprise Linux 5.4, the s390utils package has been rebased to version 1.8.1. For a full list of features that this rebase provides, please refer to the Package Updates section of the Technical Notes. (BZ#477189)
  • In the kernel, a sysfs interface has been implemented to associate actions to shutdown triggers. For more details on this feature, refer to the "Shutdown actions" chapter in IBM's "Device Drivers, Features, and Commands" book located at: http://www.ibm.com/developerworks/linux/linux390/development_documentation.html
Red Hat logoGithubRedditYoutubeTwitter

Learn

Try, buy, & sell

Communities

About Red Hat Documentation

We help Red Hat users innovate and achieve their goals with our products and services with content they can trust.

Making open source more inclusive

Red Hat is committed to replacing problematic language in our code, documentation, and web properties. For more details, see the Red Hat Blog.

About Red Hat

We deliver hardened solutions that make it easier for enterprises to work across platforms and environments, from the core datacenter to the network edge.

© 2024 Red Hat, Inc.