PCI errors with BPE3 and PCIE Gen2 card (43W4391) - IBM BladeCenter HS22

Resolving The Problem

Source

RETAIN tip: H195347

Symptom

When using a PCIe 2.0 (Gen 2) compatible expansion card in BladeCenter PCI Express I/O Expansion Unit, Option 43W4391 (Field Replaceable Unit (FRU) 43W4390), the blade may restart or shutdown unexpectedly. The Advanced Management Module (AMM) may record the following errors in the event log:

Group 2 (PCIs) bus uncorrectable error
Group 2, connector (One of PCI Error) PCI slot 0 fault
PCI bus 1, connector (PCI 2) PCI slot 0 fault
Chassis (NMI State) software NMI

Affected configurations

The system may be any of the following IBM servers:

• BladeCenter HS22, Type 1936, any model
• BladeCenter HS22, Type 7870, any model

The system is configured with one or more of the following IBM Options:

• BladeCenter PCI Express I/O Expansion Unit, Option 43W4391, FRU 43W4390

This tip is not software specific.

The following system BIOS level(s) are affected:

• 1.08 and below

The system has the symptom described above.

Solution

This behavior is corrected in the 1.09 (46A) version of Unified Extensible Firmware Interface (UEFI) firmware.

The file is available from the IBM System x Support web site at the following URL:

• http://www.ibm.com/systems/support/

Workaround

Use a PCIe Gen 1 speed expansion card in the BladeCenter PCI Express I/O Expansion Unit (BPE3).

Additional information

The BPE3 does not support Gen2 speed even though BladeCenter HS22 and the expansion card may. The HS22 blade can be set to Gen1 speed in the UEFI. However, UEFI 1.08 and below do not correctly save Gen1/Gen2 speeds. As a result, during operation, the card may request to move up to Gen2 speed and have that request honored by the HS22. The blade is then likely to have a Peripheral Component Interconnect (PCI) error because the BPE3 cannot support the higher speed.