Upgrade Sun SPARC T5-x systems to system firmware 9.1.0.b (or later) to prevent C2C Link Training and PCIe Root Complex related faults

Asset ID:	1-79-1582207.1
Update Date:	2017-05-23
Keywords:

Solution Type Predictive Self-Healing Sure

Solution 1582207.1 : Upgrade Sun SPARC T5-x systems to system firmware 9.1.0.b (or later) to prevent C2C Link Training and PCIe Root Complex related faults

Applies to:

SPARC T5-8
SPARC T5-1B
SPARC T5-4
SPARC T5-2
Netra SPARC T5-1B Server Module
SPARC

Purpose

This document is intended to alert customers and support center personnel of issues that can be avoided or resolved by loading updated system firmware

Details

Sun SPARC T5-x systems without system firmware 9.1.0.b (or later) may experience C2C Link Training and PCIe Root Complex faults, or prompt unnecessary hardware replacement, which can be prevented by upgrading to System Firmware 9.1.0.b.(or later)

These issues are addressed in the following releases:

SPARC Platform

System Firmware 9.1.0.b or later, as delivered in the following patches:

    SPARC T5-8 Server with patch 17264131 or later
    SPARC T5-4 Server with patch 17264131 or later
    SPARC T5-2 Server with patch 17264122 or later
    SPARC T5-1B Netra Blade with patch 17264110 or later
    SPARC T5-1B Blade with patch 17264114 or later

The following is additional data for the C2C Link Training Issue, PCIe Root Complex Issue, and additional bugs resolved with firmware 9.1.0.b

A) C2C link training issue

msg-id = SPSUN4V-8000-2S

fault class = fault.cpu.generic-sparc.c2c

B) PCIe root complex issue

msg-id:

PCIEX-8000-KP PCIEX subsystem problem
PCIEX-8000-3S PCIEX subsystem problem
PCIEX-8000-EE PCIEX subsystem problem
PCIEX-8000-FX PCIEX subsystem problem
PCIEX-8000-J5 PCIEX subsystem problem
PCIEX-8000-YJ excessive PCI Express recovered bus errors detected

fault class:
fault.io.pciex.bus-linkerr-corr
fault.io.pciex.device-interr-corr

C) Additional CR's resolved with firmware 9.1.0.b

Bug 16962705 RETRY PCIE LINK TRAINING BETWEEN FIRST LEVEL SWITCH AND ENDPOINT ON T5 PLATFORMS

Associated Bugs: 16636491, 16456646, 16422934

applies to: T5-8, T5-4 and T5-2.

There are problems with PCIe link training between the IDT switch and its downstream endpoint link partner . The problems manifest as a link in L0 with a reduced width and/or speed.

Bug 17087161 PCIE LINK WIDTH REDUCTION TO XGBE ON T5-4/8

Associated Bugs: 17166158, 17166243, 17166289

applies to: PCIe Link width reduction to XGBE on T5-4/8

Failure mode seen in RQT where the PCIe link width to Net0/XGBE0 twinville reduces from x8 to x4.

Steps for resolution:

1) clear fault

#fmadm repair UUID

2) upgrade firmware to 9.1.0.b

    SPARC T5-4 Server with patch 17264131 or later
    SPARC T5-4 Server with patch 17264131 or later
    SPARC T5-2 Server with patch 17264122 or later
    SPARC T5-1B Netra Blade with patch 17264110 or later
    SPARC T5-1B Blade with patch 17264114 or later

NOTE: For systems that continue to experience C2C Link Training or PCIe Root Complex issues after firmware is upgraded please perform AC power cycle as described in the following step 3

3) AC power cycle 60 seconds with cords unplugged. Details can be seen in Performing an AC power cycle on the T3/T4/T5 Servers (Doc ID 1571054.1)

4) after system comes back up check for any new faults

#fmadm faulty

Hardware replacement should only be considered if the above steps prove to be unsuccessful

Attachments

This solution has no attachment