Memory Reference Code Not Cleared for Auxiliary Processors Resulting in False SPX86-8001-U5 UE Error

Asset ID:	1-72-1557001.1
Update Date:	2016-11-15
Keywords:

Solution Type Problem Resolution Sure

Solution 1557001.1 : Memory Reference Code Not Cleared for Auxiliary Processors Resulting in False SPX86-8001-U5 UE Error

Applies to:

Sun Fire X4170 M2 Server - Version Not Applicable and later
Sun Fire X4470 Server - Version Not Applicable and later
Sun Blade X6270 M2 Server Module - Version Not Applicable and later
Sun Blade X6270 Server Module - Version Not Applicable and later
Sun Fire X4270 M2 Server - Version Not Applicable and later
Information in this document applies to any platform.

Symptoms

The Intel microprocessor supports logging of errors using the MCA (machine check architecture).

MCA will indite memory error if an error occurs and report the errors to iLOM for diagnosis output.

There is an issue with early platform firmware that results in false SPX86-8001-U5 errors being reported by iLOM under certain conditions.

The conditions are as follows:

The platform tests Intel CPU's during POST (Power On Self Test) and reports any errors found to iLOM. However during this time, only the BSP (Boot Strap Processor) is active.

All other processors (AP's - Auxiliary Processors) are run through a series of tests but the test execution is stored in a special scratchpad area for reporting later in the boot cycle.

BIOS (Basic Input/Output System) typically reads and clears this scratchpad area to diagnose issues.

The condition that results in a possible SPX86-8001-U5 is seen when a platform is reset during initialisation which does not allow for the scratchpad area to be cleared by BIOS and reports the scratchpad test execution as real errors on the next platform power-on.

On the next power-on BIOS will check the MC registers (Memory reference Code) and send the error information to iLOM if the MC registers were set.

iLOM will then report the error as failing DIMMS with a UE condition (Uncorrectable Error).

The scratchpad was not cleared correctly so the error is reported falsely.

Error signature in iLOM SEL (system event log):

1 | 1/1/2012 | 00:00:01 | System Boot Initiated | Initiated by warm reset | Asserted
2 | 1/1/2012 | 00:00:01 | System Firmware Progress | Memory initialization | Asserted
3 | 1/1/2012 | 00:00:01 | System Boot Initiated | Initiated by warm reset | Asserted
4 | 1/1/2012 | 00:00:01 | System Firmware Progress | Memory initialization | Asserted
5 | 1/1/2012 | 00:00:01 | System Firmware Progress | Primary CPU initialization | Asserted
6 | 1/1/2012 | 00:00:01 | System Firmware Progress | Management controller initialization | Asserted
7 | 1/1/2012 | 00:00:01 | System Firmware Progress | Secondary CPU Initialization | Asserted
8 | 1/1/2012 | 00:00:01 | Memory | Uncorrectable Error | Asserted | OEM Data-2 0x00 OEM Data-3 0x08

Error signature with iLOM Fault Diagnostics / ASR:

ASR:Memory Uncorrectable ECC Fault

Critical alert on faulty component MB/P0/D8. A system component faulted due to fault in memory intel dimm_ue.

Event Time = Tue Dec 1 00:00:01 UTC 2012
Fault Message ID = SPX86-8001-U5
Fault UUID = e1111111-d222-c333-d444-e55555555555
Knowledge Article URL = http://www.sun.com/msg/SPX86-8001-U5
Fault Description = Not available
Fault Severity = other
Product Manufacturer = ORACLE CORPORATION
Product Name = SUN FIRE X4170 M2 SERVER
Product Serial Number = SOMESERNU
Chassis Manufacturer = SOMECHAMU
Chassis Name = SOMECHANA
Chassis Serial Number = SOMECHANU
Chassis Part Number = SOMECHAPA
DiagEntity = SOMEDIAGEN
SystemIdentifier = Not available

Changes

Platform reset / rebooted

Cause

System scratchpad not cleared

Solution

Platform firmware was updated to resolve this issue however, since there are multiple platforms involved, the minimum firmware for each platform is being described here.

Please clear this error as detailed below then update your platform with the following firmware version or higher.

set /SYS/MB/PX/DX clear_fault_action=true

(where PX/DX represents the processor/DIMM your error appeared on e.g. P0/D8)

Then install the following version of firmware or above if another is available.

This was the latest version available during the writing of this document:

Sun Fire X4170 <Patch: 16838772> - ILOM 3.0.16.15.c, BIOS 07.06.03.07 (SW 2.6.3)

Sun Fire X4270 <Patch: 16838826> - ILOM 3.0.16.15.c, BIOS 07.06.03.07 (SW 2.6.3)

Sun Fire X4275 <Patch: 16838852> - ILOM 3.0.16.15.c, BIOS 07.06.03.07 (SW 2.6.3)

Sun Fire X4470 <Patch: 16405063> - ILOM 3.0.16.13.b, BIOS 09.05.02.00 (SW 1.5)

Sun Fire X4170 M2 <Patch: 16832629> - ILOM 3.1.2.20.a, BIOS 08.14.01.03 (SW 1.7.1)

Sun Fire X4270 M2 <Patch: 16832680> - ILOM 3.1.2.20.a, BIOS 08.14.01.03 (SW 1.7.1)

Sun Fire X4470 M2 (X2-4) <Patch: 16404931> - ILOM: 3.1.2.24.b, BIOS 16.04.02.00 (SW 1.4)

Sun Blade X6270 Server Module <Patch: 15944807> - ILOM 3.0.16.17a, BIOS 07.07.01.03 (SW 2.5.1)

Sun Blade X6270 M2 Server Module <Patch: 14568638> - ILOM 3.0.16.11h, BIOS 08.07.03.06 (SW 1.4.1)

The latest firmware for each platform can alway be found on the Oracle Firmware Page

For Oracle EXALOGIC X2-2 and Oracle EXALOGIC X3-2, your platform bundle will include the required firmware to resolve this issue.

Consult your platform service manual for full instructions on the firmware update procedure.

Bug References:

<Bug: 15772044>

<Bug: 15770526> (Duplicate)

See also SPX86-8001-ME - Memory DIMMs are not Populated <Document: 1021412.1>

Product Issues Pages:

Sun Fire X4170 X4270 X4275 Current Product Issues <Document: 1345581.1>

Sun Fire X4170 M2, X4270 M2 Current Product Issues <Document:1345623.1>

Oracle EXALOGIC Current Product Issues X2-2 , X3-2 <Document:1360310.1>

Sun Blade X6270 Server Module Current Product Issues <Document:1363764.1>

Sun Blade X6275 M2 (Varu+) Server Module Current Product Issues <Document:1363762.1>

Sun Netra X6270 M2 Server Module Current Product Issues <Document:1363768.1>

References

<BUG:15772044> - SUNBT7144259 EXALOGIC: ILOM REPORTING BAD MEMORY DIMMS
<BUG:15770526> - SUNBT7142212 LYNX+ EXHIBITING FALSE UE FAULTS ON SYSTEM INITIALIZATION WITH LATE
<NOTE:1557029.1> - Memory Reference Code Not Cleared for AP's Resulting in False SPX86-8001-ME Population Errors

Attachments

This solution has no attachment