Sun Fire V20z Server
 
Red Hat Enterprise Linux x64
 
Red Hat Enterprise Linux x86
 
Sun Fire V40z Server
 

GCS>Sun Microsystems>Servers>x64 Servers
 

The system remains functional and the errors repeat throughout the syslog.

The errors are similar to the following example:

 Oct 13 11:19:14 srdopt3 kernel: CPU 0: Silent Northbridge MCE
Oct 13 11:19:14 srdopt3 kernel: Northbridge status a60000010005001b
Oct 13 11:19:14 srdopt3 kernel:     GART TLB error generic level generic
Oct 13 11:19:14 srdopt3 kernel:     extended error gart error
Oct 13 11:19:14 srdopt3 kernel:     link number 0
Oct 13 11:19:14 srdopt3 kernel:     err cpu1
Oct 13 11:19:14 srdopt3 kernel:     processor context corrupt
Oct 13 11:19:14 srdopt3 kernel:     error address valid
Oct 13 11:19:14 srdopt3 kernel:     error uncorrected
Oct 13 11:19:14 srdopt3 kernel:     previous error lost
Oct 13 11:19:14 srdopt3 kernel:     error address 00000000e7f6a050

If, after the kernel update the error message still appears in the "messages", in the BIOS Advanced menu, there is an option named "No Spec. TLB Reload". By default, this setting is disabled, which allows TLB reload.

If the above error message is encountered ensure that "No Spec. TLB reload" is ENABLED in the Advance BIOS setting.

To accomplish this, reboot the server then press F2 to enter BIOS setup.

Then proceed to Avanced -> Chipset Configuration BIOS menu

Finally use the arrow keys to scroll down and modify the "No Spec. TLB reload" field to Enabled. Press Esc and then F10 to save your changes.

  PhoenixBIOS Setup Utility
Advanced
+------------------------------------------------------------------------------+
|              Chipset Configuration                 |   Item Specific Help    |
|----------------------------------------------------+-------------------------|
|                                                    |                         |
|   Setting items on this menu to incorrect         ^| Override                |
|   values may cause your system to malfunction.    :| Speculative TLB Reload  |
|                                                   :| (Disable to permit TLB  |
|   SRAT Table             [Enabled]                :| speculative reloads).   |
|   Node Interleave:       [Disabled]               :|                         |
|   Bank Interleave:       [Auto]                   :| For best performance,   |
|                                                   :| leave this disabled.    |
|   ECC:                   [Enabled]                :|                         |
|   Dram ECC:              [Enabled]                :|                         |
|   ECC Scrub Redirection  [Enabled]                :|                         |
|   Chip-Kill:             [Enabled]                :|                         |
|   DCACHE ECC Scrub CTL:  [5.12 us]                :|                         |
|   L2 ECC Scrub CTL:      [10.2 us]                :|                         |
|   Dram ECC Scrub CTL:    [163.8 us]               :|                         |
|   No Spec. TLB Reload:   [Enabled]                :|                         |
|                                                    |                         |
+------------------------------------------------------------------------------+
Fsc  Exit  <>  Select Menu  Enter  Select > Sub-Menu  F10  Save and Exit  

This will disallow TLB reloading and avoid the error message.

If the kernel is updated, and the BIOS setting is changed, and still the errors appear in the "messages", the "IOMMU" needs to be booted without AGP support.

After adding "iommu=noagp" (no quotes) as parameter to the kernel boot options (either at boot time or in /boot/grub/menu.lst) reboot the server to check that the message disappeared.

For more information regarding decoding machine check errors. Understanding and Decoding Machine Check Errors on Sun Fire[TM] V20z/Sun Fire[TM] V40z running Red Hat OS

Several different but connected solutions for this problem exist.

Please try one after the other for a sure solution.

This solution has no attachment