Intel Xeon 5600 series processor C6-State refresh erratum - IBM System x

Resolving The Problem

Source

RETAIN tip: H207019

Symptom

Systems that have Advanced Configuration and Power Interface (ACPI) C3 (often known as Sleep) enabled may be exposed to a memory uncorrectable error asserted by the Intel 5600 series processor.

Users will see the following example Error log and memory will get disabled during runtime:

Affected configurations

The system may be any of the following IBM servers:

• System x3400 M3, type 7379, any model
  
• System x3500 M3, type 7380, any model
  
• System x3550 M3, type 4254, any model
  
• System x3550 M3, type 7944, any model
  
• System x3620 M3, type 7376, any model
  
• System x3630 M3, type 7377, any model
  
• System x3650 M3, type 4255, any model
  
• System x3650 M3, type 7945, any model
  
• iDataPlex dx360 M3 Server, type 6391, any model

This tip is not software specific.

This tip is not option specific.

The following system BIOS or UEFI level(s) are affected:

• Later than Unified Extensible Firmware Interface (UEFI) Version 1.15 - x3550 M3 or x3650 M3

Solution

This behavior was corrected in a release of Unified Extensible Firmware Interface (UEFI) firmware Version 1.16 Build ID: D6E158A.

The file is available by selecting the appropriate Product Group, type of System, Product name, Product machine type, and Operating system on IBM Support's Fix Central web page, at the following URL:

• http://www.ibm.com/support/fixcentral/

Workaround

The system Unified Extensible Firmware Interface (UEFI) setting of memory static driver impedance is disabled as default. If ACPI C3 is found to be enabled, then users can change the memory static driver impedance using the F1 Setup menu as shown:

Select System Settings --> Memory --> Static Driver Impedance = Enabled

The system UEFI setting of the ACPI C3 is disabled as default. If ACPI C3 is found to be enabled, change the C-state by the F1 Setup menu as shown:

Select System Settings --> Processors --> Central Processing Unit (CPU) C-states = Disabled

If users install with a Linux Base Operation System, such as Red Hat Enterprise Linux (RHEL) 6 or SUSE Linux Enterprise Client (SLEC) 11, the Linux kernel version 2.6.35, or later will use its ACPI table to enable the Intel CPU C-state by the native hardware driver.

Follow this workaround to disable the CPU C-state by adding the 'intel_idle.max_cstate=0' kernel parameter when starting up the system:

1. Start up the system and edit the following file:
   

   /start/GRUB/menu.lst

   
   
2. Add the following parameter to the kernel command line:
   

   intel_idle.max_cstate=0

   
   For example:
   

   kernel /vmlinuz-2.6.32-131.0.15.el6.i686 ro root=/dev/mapper/VolGroup-lv_root rd_LVM_LV=VolGroup/lv_root rd_LVM_LV=VolGroup/lv_swap rd_NO_LUKS rd_NO_MD rd_NO_DM LANG=en_US.UTF-8 SYSFONT=latarcyrheb-sun16 KEYBOARDTYPE=pc KEYTABLE=us crashkernel=auto rhgb quiet intel_idle.max_cstate=0

   
   
3. Save and exit the file.
4. Restart the system.

Additional information

Certain memory control signals may be driven incorrectly during package C6 memory self-refresh. Although the system set the ACPI C3 as disable in default, there is still a chance the user enables CPU C-state or cause by new Linux kernel.

The Linux kernel with the 'intel_idle' driver in certain newer operating systems can engage C-states even when C-states and C1E are disabled explicitly.

This setting should be disabled only when C-states and Logical Processor options also are set to Disabled.