VMware cannot finish booting when IOMMU is enabled - IBM System x3755 M3 (7164)

Resolving The Problem

Source

RETAIN tip: H201769

Symptom

Users using an IBM System x3755 M3, type 7164, with the ServeRAID M1015 SAS/SATA Controller, Option part number 46M0831, installed may cause VMware ESX to not load properly or hang when the I/O Memory Management Unit (IOMMU) is enabled in the the Unified Extensible Firmware Interface (UEFI) Basic Input/Output System (BIOS) settings.

Affected configurations

The system may be any of the following IBM servers:

• System x3755 M3, type 7164, any model

The system is configured with at least one of the following:

• VMware ESX Server 4.1, any Update
• VMware ESXi 4.1 Embedded, Base Install
• VMware ESXi 4.1 Installable, Base Install

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

• ServeRAID M1015 SAS/SATA Controller, Option part number 46M0831, replacement part number (CRU) 46M0861

The following system BIOS/uEFI level(s) are affected:

• 28A, 34A

Note: This does not imply that the network operating system will work under all combinations of hardware and software.

Please see the compatibility page for more information: http://www.ibm.com/servers/eserver/serverproven/compat/us/

Solution

A new ServeRAID M1015 firmware v20.10.1-0022 or higher is now available to resolve this issue.

The referenced file will be available from the "(Critical update) IBM ServeRAID M1015 SAS/SATA Controller Firmware v20.10.1-0022 for Microsoft Windows Server 2003/2008 - IBM System x" web page, at the following URL:

• http://www.ibm.com/support/entry/portal/docdisplay?lndocid=MIGR-5082826

Workaround

Disable the IOMMU in the UEFI setup using the following steps:

1. Reboot the server.
2. Press F1 when the system is at the IBM splash screen.
3. Select System Settings, Processors, and IOMMU Mode.
   
   Note: Restoring the default UEFI settings will also disable the IOMMU function. If IOMMU is preferred, then obtain the latest M1015 firmware update.

Additional information

The IOMMU feature when enabled by the UEFI is not properly interpreted by VMware, which causes an incorrect I/O memory mapping issue that eventually hangs the operating system.

When the IOMMU is disabled, the operating system handles the I/O memory mapping correctly and does not crash the operating system.

The new ServeRAID M1015 firmware allows the controller to be used correctly when the IOMMU is enabled.