Power5 Firmware

Applies to:  OpenPower, i5 and p5 Servers

This document provides information about the installation of Licensed Machine or Licensed Internal Code, which is sometimes referred to generically as microcode or firmware.


Contents


1.0 Systems Affected

This package provides new firmware for OpenPower, i5 and p5 Servers only.  Do not use on any other systems.

The firmware level in this package is:

Note for models 9119-590, 9119-595 and 9406-595
Due to hardware prerequisites which must be verified prior to performing a firmware upgrade from Release Level SF230_xxx to later release levels, customers are advised not to perform this upgrade on their own.

Customers with model 59x servers at firmware level SF230_xxx  who would like to enable Redundant FSP failover capabilities or upgrade their firmware to the latest release level  (SF240_xxx)  for any reason should call 1-800-IBM-SERV and request ECA815.  IBM will dispatch a Service Representative to complete this upgrade for you.

Please note that the upgrade path for 59x servers at release level SF230_xxx bypasses the SF235_xxx release level and goes directly to the SF240_xxx release level.  Refer to the table below:
 
59x Upgrade Path Reference Table
From Level To Level Mechanism
SF230_xxx SF235_xxx Not supported
SF230_xxx SF240_xxx Order ECA 815 and an IBM SSR will perform upgrade
SF235_xxx SF240_xxx Customer Upgradeable, or Order ECA834 and an IBM SSR will perform the upgrade 


2.0 Cautions and Important Notes

Do not attempt to backlevel firmware from any given release level to an earlier release level.   If you feel that it is necessary to backlevel the firmware on your system to an earlier release level, please contact your next level of support.

i5 Systems in an  HSL Cluster

A problem has been found when installing this firmware level on running systems configured  in an HSL Opticonnect cluster.  Customers with systems in this type of configuration must power off the managed system prior to installing this firmware level.  If the managed system is not powered off prior to installing  this firmware level  there is a strong possibility of a system crash with SRC B182E500 being logged.

If your systems have firmware release level SF230 installed, the recommended way to pick up all of the  HSL  Optconnect fixes is to install firmware level SF240_284.  (For 9406-595 systems, ECA 815 must be installed before SF240_284 is installed; contact your service support organization.) This would be a  release upgrade, so the managed system will be powered off  automatically as part of the upgrade process.

If you choose to stay on the SF230 release level, the managed system must be powered off before installing this firmware level.

Special Instructions for Certain Models

If the following models, with the following processor feature codes only, are HMC-managed, this service pack should not be applied with the system in operating state.  The system should be brought to the power off state before applying the service pack.  This special instruction applies only to these specific models and feature codes and this service pack.

Model            Processor Feature Codes

9110-510       7609 or 7610
9111-520       5231
9113-550       5239
9123-710       1939 or 1963
9124-720       1960
9406-520        8950, 8951, 8952, 8953, 8954, or 8972

Installing System Firmware using a CD

Note:  When attempting to install system firmware from a CD-ROM you may receive an HMC error approximately 10 seconds after starting the installation. Here are probable causes for this problem:

HMC-Managed Systems

An HMC running the latest level of V4 R5.0, V5 R1.x, V5 R2.x or V6 R1.x will manage a system with any SF230 firmware release or earlier installed.

Go to the following URLs to access the HMC code packages:

NOTE:  For 9119-590 and 595, 9406-595 and 9118-575,  power subsystem firmware level BP230_155 (or later) must be installed prior to installing system firmware level SF230_156.  You should also make sure that all frames of a multiple 9118-575 configuration are at BP230_155 (or later) before attempting to install system firmware.

NOTE:   You must be logged in as hscroot in order for the firmware installation to complete correctly.

After updating the firmware to this service pack level (SF230_158), do not use the HMC Code Update Advanced Features 'Remove and Activate' function to restore a system to any Release 230 level prior to SF230_150.

Attention:  Before installing firmware on an HMC managed system:

Note: Only 590, 595 and 575 systems will display multiple components. All other system types will only return a single line for each managed system.

Attention:  (Applicable to 9119-590 and 595, 9406-595 and 9118-575 systems only) After installing the System and/or Power Subsystem Firmware, proceed to the HMC console and rebuild the managed server and/or frame using the following procedure:

1. If System Firmware was just installed then reset the connection to the server:

From the "Navigation Area" window pane on the HMC Console:

Next, from the "Server and Partition: Frame Management" window pane: 2. If Power Subsystem Firmware was just installed then reset the connection to the frame:

From the "Navigation Area" window pane on the HMC Console:

Next, from the "Server and Partition: Frame Management" window pane: 3. If BOTH the System Firmware and the Power Subsystem Firmware were just installed then reset the connections to both the server and the frame using steps 1) and 2) above after both sets of firmware have been loaded.

Memory Considerations for Firmware Upgrades

The increase in memory used by the firmware is due to the additional functionality in later firmware releases.

3.0 Firmware Information and Description

IBM introduced the Concurrent Firmware Maintenance (CFM) function on p5 systems in system firmware level 01SF230_126_120, which was released on June 16, 2005.  This function supports nondisruptive system firmware service packs to be applied to the system concurrently (without requiring an IPL to activate changes).

For systems that are not managed by an HMC, the installation of system firmware is always disruptive.

Note:  The concurrent levels of system firmware may, on occasion, contain fixes that are known as deferred. These deferred fixes can be installed concurrently, but will not be activated until the next IPL.  Deferred fixes, if any, will be identified in the "Firmware Update Descriptions" table of this document.  For deferred fixes within a service pack, only the fixes in the service pack which cannot be concurrently activated are deferred.

Use the following example as a reference to determine whether your installation will be concurrent or disruptive.

Note:  The file names and service pack levels used in the following examples are for clarification only, and are not
             necessarily levels that have been, or will be released.

 System firmware file naming convention:

     01SFXXX_YYY_ZZZ

NOTE:  Values of service pack and last disruptive service pack  level (YYY and ZZZ) are only unique within
a release level (XXX).  For example, 01SF235_250_120 and 01SF240_250_200 are different service packs.

An installation is disruptive if:

               Example:  Currently installed release is SF225, new release is SF230                Example:  SF230_120_120 is disruptive, no matter what level of SF230 is currently
                                   installed on the system                  Example:  Currently installed service pack is SF230_120_120 and
                                     new service pack is SF230_152_130

An installation is concurrent if:

          Example: Currently installed service pack  is SF230_126_120 and
                             new service pack is SF230_143_120
 

Firmware File Information and Update Description
 

Filename Size Checksum
01SF230_158_120.rpm 13256062 54899
 
SF230
SF230_158_120

09/25/06

Impact:  Function       Severity:  Hiper

System firmware changes that affect systems using HSL Opticonnect

  • A problem was fixed that caused a system in an Opticonnect cluster to crash with SRC B182E500 when partitions were activated.
System firmware changes that affect the model 575, 590, and 595, other HMC-controlled servers, and clustered systems
  • HIPER:  On model 575, 590, and 595 systems, a problem was fixed that caused a memory failure with SRC B123E500 after the concurrent repair of a CEC DCA. 
  • On model 590 and 595 systems, the FRU callouts were enhanced to improve system reliability after an uncorrectable memory error was repaired.
  • A problem was fixed that was causing enhanced error handling (EEH) error codes to be erroneously generated on 7040-61D I/O drawers when certain adapter card configurations were heavily stressed by the application code.
System firmware changes that affect certain p5, i5, and OpenPower systems
  • HIPER:  On model 561, 570, 590 and model 595 systems, a problem was fixed that caused memory to be deconfigured due to an erroneous memory controller failure, with SRC B121E500 and word 8 = 00EA0035.
  • On model 561 and 570 systems with more than one processor drawer, a problem was fixed that was preventing enclosure VPD keywords from being changed in the second, third, and fourth processor drawers using the ASMI menus.
  • A problem was fixed that caused the service processor to run out of memory with SRC B110F139.
System firmware changes that affect all p5 and i5 systems
  • HIPER:  A problem was fixed that caused prevented correctable L3 errors from being recovered, with the result that an error threshold was reached, and an erroneous FRU callout was made.
  • HIPER:  A problem was fixed that caused the system or partition to crash when corruption was detected in low memory.
  • HIPER:  A problem was fixed that was preventing AIX system dumps with SRC D200A200 after a partition hang.
  • A problem was fixed that caused dynamic logical partitioning (DLPAR) operations to intermittently hang partitions running Linux.
  • A problem was fixed that caused AIX to incorrectly report errors that occurred during booting, and were then passed to the AIX error log.
  • A problem was fixed that caused the "ping" command in the system management services (SMS) to fail during a NIM installation.
  • A problem was fixed that caused multiple B7005190 SRCs to be generated when virtual Ethernet adapters were configured.
  • A problem was fixed that kept an already-connected uninterruptible power supply (UPS) from being detected correctly after the service processor performed a reset/reload operation. 
SF230_156_120

04/03/06

Impact:  Function       Severity:  Hiper 

Two fixes that were deferred in earlier levels are concurrent in this level:

1.  In firmware level SF230_150, there was a deferred fix that affects all p5 and i5 systems: 

HIPER:  The initialization settings of the memory buffers were changed to eliminate spurious checkstops that were causing system and partition crashes.

In SF230_156, this fix is CONCURRENT. 

2.  In firmware level SF230_153 (which is also part of ECA 823), there was a deferred fix that affects model 59x systems: 

HIPER:  The hardware initialization settings were changed to fix an incorrect timeout setting in the non-cacheable unit (NCU). The current settings may cause a system checkstop, which could result in a system or partition crash.  This is also known as the  "EMQ hang" problem. 

In SF230_156, this fix is CONCURRENT.  Also, the recovery time for the EMQ hang was improved to benefit HACMP configurations. 
 

System firmware changes that affect all p5 and i5 systems:

  • HIPER: An incorrect initialization setting in the shared memory interface  (SMI) may cause a system checkstop, which could result in a system or partition crash. This is also known as the "SMI parity error" issue. 
  • HIPER:  Fixed a firmware problem that caused the system to interpret a machine condition as requiring special attention handling, where no special attention existed. This could result in a system checkstop (with error codes B181F647 and B110F139).  This problem is also known as the "master attention alone" issue.
System firmware changes that affect the model 575, 59x, other HMC-controlled servers, and clustered systems:
 
  • DEFERRED and HIPER:  A potential server checkstop issue exists in clusters using p5 models p575, p590, and p595, when these models are used to execute compute-intensive applications tuned with the November Power5 Fortran compiler.  Clusters executing applications that are compiled for optimal performance by utilizing the software pre-fetch function are exposed to a server checkstop (B1xxE500). Hardware initialization settings were changed to fix this problem and  the performance tools modified to take these new settings into account.
  • HIPER  On systems with a high performance switch (HPS) network adapter installed, fixed a problem that caused a checkstop when memory was moved from a partition that did not own an HPS network adapter. problem was fixed that prevented I/O towers from being concurrently added to i5 systems.
  • A problem was fixed that was preventing partitions with large amounts of memory from booting successfully.
  • The firmware was changed so that the time it takes to remove a large amount of memory from a partition using a DLPAR (dynamic LPAR) operation is reduced.
  • A problem was fixed that was causing enhanced error handling (EEH) error codes to be erroneously generated on 7040-61D I/O drawers when certain adapter card configurations were heavily stressed by the application code.
System firmware changes that affect p5 systems not in 24 inch racks:
  • On p5 systems that are not managed by an HMC, the timeout for firmware installations performed by the operating system ("in-band" installations) was increased.
SF230_150_120

11/02/05

Impact:  Serviceability        Severity:  Special Attention 

Attention: An IPL is required to activate a deferred fix, the system must be powered off via the HMC, (i.e. HMC state=Power OFF) and then powered back on. 

  • DEFERRED Changed the initialization settings of the memory buffers to eliminate spurious checkstops that were causing system and partition crashes.
  • DEFERRED  On model 590 and 595 systems, fixed a problem that was causing DDR II memory cards in slots C10 and C12 to fail the memory test during a slow-mode boot. 
  • DEFERRED  Fixed a problem that was not allowing the system enclosure serial number to be rewritten using the ASMI menus.
  • DEFERRED  Fixed a problem that causes the default boot list and the stored boot list to be swapped in the system management services (SMS) firmware when booting an AIX or Linux partition. 
  • Several problems were fixed that caused a failure to boot with SRCs B1817201, C1001014 and/or C1001020.
  • On an HMC, fixed a problem that caused a blank screen when reopening a VTERM after closing it.
  • Fixed a problem that was causing a platform dump to be incomplete under certain circumstances.
  • Fixed a problem that caused a "no connection" state between the HMC and the service processor after the service processor attempted a dump operation.
  • Added support for ARP (address request protocol) to the System Management Services (SMS) firmware for AIX and Linux partitions.
  • Fixed a problem in the server firmware that was allowing the HMC to allocate more memory to a partition than was available.
SF230_147_120

10/27/05
 

Impact:  Serviceability        Severity:  Special Attention 
  • Fixed a problem that caused system processors to get into a locked state.
  • Changed the initialization settings of the memory buffers to eliminate spurious checkstops that were causing system and partition crashes.
SF230_145_120

08/24/05

Impact:  Serviceability        Severity:  Special Attention 
  • Fixed a problem on i5 systems with 5074 I/O towers that causes a partition to hang with SRC B6000255 when an I/O adapter in the 5074 is moved using dynamic LPAR.
  • Fixed a problem on i5 and p5 systems that causes a continuous stream of phantom interrupts on partitions with shared processors.
SF230_143_120

08/09/05

Impact:  Serviceability        Severity:  Special Attention 

Attention: An IPL is required to activate a deferred fix, the system must be powered off via the HMC, (i.e. HMC state=Power OFF) and then powered back on. 

  • DEFERRED  On model 575 and 59x systems in which an IBM High Performance Switch (HPS) is installed, fixed a problem that was causing an invalid high performance switch adapter failure to be reported after a switch failure. 
  • DEFERRED  On model SR110000 systems only, fixed a problem that was causing the I/O in the CEC not to show up after an IPL. 
  • DEFERRED  On systems that are managed by an HMC, a problem was fixed that was causing concurrent firmware update to fail with error code ACT01724 on the HMC. 
  • DEFERRED  Fixed a problem involving correctable and uncorrectable memory errors that may have resulted in memory being deconfigured during a slow mode IPL on 9119-59x systems with F/C 7814 (4 GB memory cards) installed. 
  • Support for 16-way model 575 systems was added to the SF230 release. 
  • Fixed a problem that was causing the firmware to erroneously report a failure of the High Performance Switch (HPS) adapter card (with error code B181F62A) in a 16-way model 575 system. 
  • On model 59x systems, a memory leak was fixed that could cause the system to reset during boot.
  • Fixed false error code B1812033 introduced in code level SF230_120 on all p5 and i5 systems model 570 and lower. The B1812033 error code may still be valid on 575 and 59x systems. 
  • Fixed intermittent boot problem introduced in code level SF230_120 on all p5 and i5 systems model 550 and lower. Error codes related to this boot problem include B1701004, B181F12E or B150F22A. 
  • On model 575 and 59x systems in which an IBM High Performance Switch (HPS) is installed, a problem in the system firmware was fixed that was causing duplicate errors for the same problem to be reported in the service processor error log and the HPS Network Manager error log. 
  • On model 575 and 59x systems in which an IBM High Performance Switch (HPS) is installed, a problem in the system firmware was fixed that was causing some switch network interface (SNI) adapters to fail a wrap test. 
  • On systems that are managed by an HMC, fixed a problem that was preventing some types of errors that were logged in the service processor's error log from being sent to the HMC. 
  • Fixed a problem that caused a very early reset/reload of the service processor with error codes of B1817201, B1817209 or similar after a firmware update. 
  • On systems that shipped with a firmware level that begins with SF210, a problem was fixed that was causing the firmware installation to the T side to fail with error code B1817201. 
  • Fixed a problem that was causing platform system dumps to be corrupted. 
  • Fixed a problem that was causing concurrent firmware maintenance to fail.
SF230_126_120

06/16/05

Impact:  Function        Severity:  Hiper 
  • Fixed a problem that causes the media bay to lose power, which makes the DVD drives in the bay inoperable on 9117-570 and 9406-570 systems.
  • Fixed most problems introduced in SF230_120 that causes the white power on/off button to intermittently fail to boot the system, terminating with error codes B181F12E, B1701004 or B150F22A, on 9111-520, 9113-550, 9405-520, 9406-520 and 9406-550 systems.
  • Fixed a problem that causes various types of planar and RIO cable failures in the 7040-61D I/O subsystem to be reported on 9118-575, 9119-590, 9119-595 and 9406-595.
SF230_120_120

06/03/05

Impact:  Function        Severity:  Hiper 

New function: 

  • Adds code to notify the user when the VPD card is not in its original system, and to prevent that system from booting. 
  • Adds support for i5/OS hosting virtual Linux and AIX partitions on i5 systems that are not managed by an HMC. 
  • Adds support for F/C 7910 on model 9118-575 systems the switch network interface  adapter, which provides the interface to the pSeries high performance switch (HPS). 
  • Adds support for F/C 7817 on 9119-590, 9119-595, and 9406-595 systems, the switch network interface (SNI) adapter, which provides the interface to the pSeries high performance switch (HPS). 
  • Adds support for concurrent firmware maintenance on systems managed by an HMC. 
  • Adds support for 32GB memory cards on 9119-590, 9119-595, and 9406-595 systems. 
  • Adds support for F/C 7894, 8GB (4x2GB) feature (2GB DDR2 DIMMs, CCIN 30F3) on machine type 9117-570.
Problems resolved: 
  • LEDs on GX bus adapter cards remain on after the node is powered off for 9119-590, 9119-595, and 9406-595 systems. 
  • The service processor is not able to place a "call home" through a modem. 
  • On systems 9118-575, 9119-590, 9119-595, and 9406-595: a firmware upgrade from SF222_075 to SF225_096 fails with error code ACT01724 on the HMC. 
  • Fixed a problem that causes the default boot list and the stored boot list to be swapped in the system management services (SMS) code when booting an AIX or Linux partition. 
  • Fixed a highly intermittent problem that causes the service processor to perform a dump when an SRC is displayed on the control panel. 
  • On  9118-575, 9119-590, 9119-595, and 9406-595 systems, updates the firmware so that the system will continue to boot when certain types of failures occur on bus adapter cards. 
  • Fixed a problem on  9118-575, 9119-590, 9119-595, and 9406-595 systems which caused a reboot to fail with error code B7006900 under certain conditions. 
  • Fixed a problem with moving a CD-ROM device between two logical partitions using DLPAR commands on the HMC. 
  • On 9117-570 and 9406-570 systems:  fixed a problem a that causes a DMA_ERR with error code 1000 0003 on the integrated Ethernet port. 
  • DS8000 storage subsystem:  multiple firmware fixes. 
  • Fixed a problem that causes the system clock to revert to December 31, 1969 and the system to crash with error code B7000103. 
  • On 9119-590 systems, fixed a problem that keeps large configurations (12 or more I/O drawers with both planars installed) from booting. 
  • Fixed a problem that prevents the 'lscfg -vp' command from reporting platform-specific data on all p5 systems.
  • Fixed a problem that intermittently causes a partition or the system to hang when the lsvpd or lscfg command is run on all p5 systems.  Also, 'the lsmcode' command may return "NOT DETERMINABLE".
  • If two power supplies are installed, the system may fail to power on if one power supply fails in the 9110-510, 9111-520, 9113-550, 9117-570, 9123-xxx, 9124-xxx, 9405-520, 9406-520, 9406-550, 9406-570 systems.
  • Fixed a problem that prevents a virtual SCSI client from connecting to a virtual SCSI server adapter on 9118-575, 9119-590, and 9119-595 systems.

  •  


4.0 How to Determine Currently Installed Firmware Levels

Instructions for determining the current level of firmware can be found at http://publib.boulder.ibm.com/eserver/v1r3s/index.jsp?topic=/ipha5/viewexistlevel.htm


5.0 Downloading the Firmware Package

The firmware is located at the web site

http://www14.software.ibm.com/webapp/set2/firmware/gjsn

Follow the instructions on this web page. You must read and agree to the license agreement to obtain the firmware packages.

In the drop-down box, choose the entry for your specific machine type and model.

You may download the rpm file for system firmware from this location to your server, an ftp server, or a CD-ROM. If your system is HMC-managed, you will also need to download the xml file located on the final download page.  Make sure the file names have the format 01SF2xx_yyy_zzz, with an extension of .rpm and .xml, before copying them to your server, an ftp server or CD-ROM.  If using a CD-ROM, copy the .rpm and .xml files to the CD-ROM using a local CD-ROM burner utility.

Another method is to download the ISO image and create a CD-ROM to use with your HMC.  The ISO image contains both the system and power subsystem firmware files.

Note: If your HMC is not internet-connected you will need to download the new firmware level to a CD-ROM or ftp server.
 


6.0 Installing the Firmware

The method used to install new firmware will depend on the release level of firmware which is currently installed on your server. The release level can be determined by the prefix of the new firmware's filename.

Example: 01SFXXX_YYY_ZZZ

Where XXX =  release level

Instructions for installing firmware updates and upgrades can be found at
http://publib.boulder.ibm.com/infocenter/eserver/v1r3s/index.jsp?topic=/ipha5/fix_serv_firm_kick.htm

Choose the options that match your server setup, and follow the instructions provided.

Note: For systems without an HMC, the update and upgrade procedure is the same.
 


7.0 Firmware History

SF225
SF225_096_096

03/12/05

Impact:  Function   Severity:  Hiper 
  • Fixed a problem in the service processor firmware that made it look like the T-side image was being lost after a "reset to factory configuration" was done using the ASMI menus. 
SF225_080_080

02/18/05

Impact:  New        Severity:  New 
  • Added support for 9118-575, 9110-510, and 9123-710.
SF222
SF222_081_081

02/07/05

Impact:  Function   Severity:  Hiper
  • Fixed a problem that caused FSCSI errors to show up erroneously in the AIX error log when an I/O adapter failed.
  • Fixed a problem that caused the processor runtime diagnostics to call out the wrong DIMM (or other memory component) when a memory failure occurred.
  • Fixed a problem that caused a platform dump to run out of storage space.
  • Fixed a problem that prevented the LEDs on I/O drawers from identifying or "rolling up" to the frame enclosure LED.
  • Fixed a problem that caused the service processor to generate error log entries every few seconds when attempting to service a recoverable CEC hardware error on systems that have processors with no memory associated with them.
SF222_075_075

12/17/04

Impact:  Function   Severity:  Hiper 
  • Fixed a problem with the AMSI menu option that allows system VPD keywords to be set.
  • Fixed a problem that kept the system from continuing to boot when an 8GB memory DIMM is deallocated because an uncorrectable error was generated by the DIMM during boot.
  • Fixed a problem that caused a partition's time to appear to drift (approximately 15 seconds/month) following a system IPL.
  • Support for partitions running i5/OS on p5 systems was added.
  • Fixed a problem that caused the default boot list and the stored boot list in the SMS menus to be swapped.
SF222_071_071

11/24/04

Impact:  New        Severity:  New 
  • Added support for systems 9119-590, 9119-595, and 9406-595.
SF220
SF220_051_051

11/01/04

Impact:  Function   Severity:  Hiper 
  • Fixed a problem that kept feature code 1827, a UPS sense cable, from being detected. 
SF220_049_049

09/28/04

Impact:  Function   Severity:  Hiper 
  • Fixed a problem that kept the serial ports on the processor drawer from functioning in 9117-570 and 9406-570 systems with multiple processor drawers.
  • Fixed a problem that kept the user from accessing the advanced system management interface (ASMI) via the web interface.
SF220_046_046

09/13/04

Impact:  Function   Severity:  Hiper 
  • Fixed a problem that caused the kept the service processor from getting to standby when an Ethernet cable was not plugged into it.
SF220_045_045

08/03/04

Impact:  Function   Severity: Hiper 
  • Various corrections and updates.
SF220_043_043

08/31/04

Impact:  New       Severity:  New 
  • Added support for new models:  9111-520, 9113-550, 9406-550, 9124-720, and 9117-570.

8.0  Change History

 
DATE Description
Aug 3, 2007 Revised Section 5.0 to reflect changes in microcode download website.
Jan 15, 2007 Rearranged information in Section 2.0. 
Jan 2, 2007 Added information in Section 2.0 about i5 systems in HSL clusters.
Sept 28, 2006 Added information in Section 2.0 concerning memory usage when upgrading firmware.
Sept 26, 2006 Added information pertaining to upgrading firmware to release level 240_xxx.