Description

SPARC M5-32 and M6-32 Servers with Sun System Firmware 9.4.2.d, 9.4.2.e, 9.5.1.c, or 9.5.3, may misidentify faulty components or fail to diagnose faulty components in PDOMs (physical domains) configured with "expandable=false". Service Processor based fault diagnosis may mistakenly identify faulty hardware contained in DCU1 as being located in DCU2. Faulty hardware in DCU2 or DCU3 may not be properly diagnosed.

Occurrence

This issue can occur on the following platforms:

SPARC Platform:

• SPARC M5-32 Servers with Firmware version 9.4.2.d, 9.4.2.e, 9.5.1.c, or 9.5.3
• SPARC M6-32 Servers with Firmware version 9.4.2.d, 9.4.2.e, 9.5.1.c, or 9.5.3

Note 1: This issue only occurs on HOSTs with expandable property set to "false" AND assigned DCU1, DCU2, or DCU3.  DCU0 hardware is not impacted by this issue.

To determine if the expandable property is set to "false", use the following command:

    -> show -level 3 /Servers/PDomains/ expandable dcus_assigned
       /Servers/PDomains/PDomain_0/HOST
         Properties:
           dcus_assigned = /SYS/DCU0
           expandable = false 

      /Servers/PDomains/PDomain_1/HOST
        Properties:
          dcus_assigned = /SYS/DCU1
          expandable = false

     /Servers/PDomains/PDomain_2/HOST
       Properties:
         dcus_assigned = (none)
         expandable = true

     /Servers/PDomains/PDomain_3/HOST
       Properties:
         dcus_assigned = /SYS/DCU2 /SYS/DCU3
         expandable = true

In the above example, the failure can occur on PDomain_1 (aka HOST1).

Note 2: To determine the firmware version installed on the system, use the following ILOM command from the active SP:

    -> show /System sysfw_version
       /System
         Properties:
           sysfw_version = Sun System Firmware 9.5.3 2015/11/25 09:17

Symptoms

 If the described issue occurs, FMA will misidentify or fail to identify the faulty hardware. Oracle snapshot analysis is required to confirm the bug.

Workaround

 There is no work around for this issue. Please see the Resolution section below.

Resolution

This issue is addressed on the following platforms:

SPARC Platform:

• SPARC M5-32 Servers with Firmware version 9.5.4.b (as delivered in patch 22982110) or later
• SPARC M6-32 Servers with Firmware version 9.5.4.b (as delivered in patch 22982110) or later

Patches

 <Patch 22982110>

History

03-May-2016: Document released, status Resolved
11-May-2016: Updated the Description, Occurrence, and Symptoms sections
16-May-2016: Minor update, no change in content

Internal Section: Comments:

This regression was caused by the putback for 19451946

Ereports for DCU0 will always be correct.
Some ereports for DCUs 2 and 3 will be incomplete, and therefore unrecognized.
Some ereports for DCU1 will be incorrectly attributed to DCU2.
In this case:

-The ereport will not be seen on SPP2 (in fmdump output)
-The ereport will be seen on SPP1, and on the master SP (in fmdump output)
-The NAC name in the ereport class will indicate a device on DCU2, e.g.
    ereport.cpu.generic-sparc.mem-cs@/SYS/CMU8/CMP0/D1003/RANK0/DRAM3/ROW6087/COL898
-The correct ereport class would have been:
    ereport.cpu.generic-sparc.mem-cs@/SYS/CMU4/CMP0/D1003/RANK0/DRAM3/ROW6087/COL898

Questions regarding this document should be addressed to
sunalertpublication_us_grp@oracle.com and copy the
submitter/responsible Engineer listed below.

Internal Contributor/Submitter: jack.hayward@oracle.com
Internal Eng Responsible Engineer: jack.hayward@oracle.com
Oracle Knowledge Analyst: jeff.folla@oracle.com
Internal Eng Business Unit Group: Systems RPE
Internal Associated SRs: 3-12084156811, 3-12224020537, 3-12349373301, 3-12503370781, 3-12512203491
Internal Pending Patches:
Internal Resolution Patches: 22982110

References

This solution has no attachment