Power6 High-End System Firmware

08/31/09

System firmware changes that affect all systems

• DEFERRED:  This fix corrects the handling of a specific processor instruction sequence that was generated on a particular heavily-tuned High Performance Computing (HPC) application. This specific instruction sequence has the potential to produce an incorrect result. This instruction sequence has only been observed in a single HPC application.  However, it is strongly recommended that you apply this fix. 
• HIPER:  A problem was fixed that caused the migration of a partition using shared processors to fail with a reason code of 4180043, or caused the source system to hang or crash.
• A problem was fixed that caused SRC 1000911B to be erroneously logged during a reset/reload of the service processor.

• On systems with 7311-D11, 7314-G30, 5790, or 5796 19" drawers attached, a problem was fixed that caused SRC 10009138 to be erroneously logged.

Concurrent maintenance (CM) firmware fixes

• A problem was fixed that caused SRC B7005603 to be erroneously logged when a F/C 5802 or 5877 19" drawer was concurrently added to the system. 

05/18/09

System firmware changes that affect all systems:

• DEFERRED:  A problem was fixed that caused the advanced system management interface (ASMI) menus to become unresponsive, and the system to appear to hang, when a GX adapter slot reservation was attempted when the system was at service processor standby.
• The firmware was enhanced to improve the service processor's capability to recover from bad bits in the flash memory.  A predictive error, or an unrecoverable error, will be logged against the card that contains the system firmware if the number of correctable or uncorrectable errors exceeds the threshold.
• A problem was fixed that prevented the service processor from automatically booting from the permanent (or P) side if the temporary (or T) side of the firmware flash was corrupted.  When the problem occurred, the service processor stopped instead of booting from the P side.
• The firmware was enhanced so that SRC B1xxE458 (with word 6=0000E42B) will be logged as informational instead of generating a call home.
• A problem was fixed that caused non-terminating SRCs (such as B1818A1E) that indicate registry read errors to be logged during a disruptive installation of system firmware.
• The firmware was enhanced to improve the field replaceable unit (FRU) callouts when a clock failure occurs.
•  A problem was fixed that caused a partition being migrated to become unresponsive on the target system when firmware-assisted dump was enabled.
•  The callouts for SRC B181E6ED, which is logged when a system is booted with service processor redundancy disabled, were improved to indicate that redundancy was disabled rather than calling out a firmware failure.
• A problem was fixed that caused hardware to be deconfigured when the system encountered network errors, even though the SRCs were being logged as informational.
• A problem was fixed that caused the detailed data at the end of an "early power off warning type 5" AIX error log entry to be filled with invalid data instead of zeros.
• A problem was fixed that caused a partition being migrated to crash on the target system.
• A problem was fixed that might cause a system to crash with SRC B170E504 when a processor was dynamically deconfigured.
• The firmware was enhanced such that when data is written to the VPD (Anchor) card, the results are verified, resulting in fewer VPD cards being replaced.
• A problem was fixed that prevented all of the necessary files from being synchronized between the primary and the secondary system controllers.  One possible symptom of this problem was the time-of-day clocks being out of synch after a system controller failover.
• A problem was fixed that caused SRC B1818601 to be logged, and a service processor dump to be generated, at runtime.

• In systems using InfiniBand switches for processor clustering, a problem was fixed that caused packets to be dropped under certain circumstances.
• On systems with five or more nodes, a problem was fixed that prevented the identify LED function from turning on the correct node's LED.

12/05/08

System firmware changes that affect all systems:

• DEFERRED and HIPER:  The system initialization settings were changed to reduce the likelihood of a system crash under  extremely rare circumstances.
• HIPER:  A problem was fixed that caused a system to fail to reboot after a B1xxE504 SRC was logged, due to a processor interconnection bus failure.  The same SRC, B1xxE504, was logged when the reboot failed.
• A problem was fixed that caused SRC 11001D1x to be erroneously logged during system boot.
• A problem was fixed that might, if a platform dump occurred, have caused a reset/reload of the service processor, and the platform dump to be corrupted.
• A problem was fixed that caused incorrect field replaceable unit (FRU) part numbers to be returned for the BPF scroll assembly, UEPO panel and the CEC MDA scroll assembly.
• A problem was fixed that prevented the system from rebooting if an error occurred during a memory-preserving IPL.
• The firmware was enhanced so that if a system with redundant system controllers is booted with redundancy disabled, a call home error will be logged.
• The firmware was enhanced so that a call home will be made if the hypervisor issues a "terminate immediate" interrupt.
• A problem was fixed that prevented service processor and hypervisor error log entries from being reported to the operating system after a successful partition migration.  This problem only affected the partition that was migrated.
• On systems running AIX or Linux, a problem was fixed that, under certain rare circumstances, might cause the operating system to crash.
• A problem was fixed that, in certain configurations, caused the removal of a host Ethernet adapter (HEA) port to fail when using a dynamic LPAR (DLPAR) operation.
• A problem was fixed that, under certain rare circumstances, caused the hypervisor to crash when it was booting with SRC B6000103 being logged.
• A problem was fixed that, under certain circumstances, prevented the operating system from recovering a PCI-E adapter on which a temporary enhanced error handling (EEH) error occurred.
• A problem was fixed that, under certain rarely occurring circumstances, caused the system to crash if an L2 or L3 cache failure is not discovered and repaired when it initially occurs.
• A problem was fixed that caused the service processor diagnostics to call out a processor as the failing item, instead of the memory DIMMs, when a large number of memory error correction coding (ECC) errors occurred.
• A problem was fixed that prevented the system from powering on after the "reset to factory settings" option was selected in the advanced system management interface (ASMI) menus.
• A problem was fixed that caused the wrong field replaceable unit (FRU) to be called out when SRC B152F109, which indicates a problem with the NVRAM in a bulk power controller (BPC), was logged.
•  (picked up under feature 683162):  A problem was fixed that prevented service processor and hypervisor error log entries from being reported to the operating system after a successful partition migration.  This problem only affected the partition that was migrated.
• A problem was fixed that might cause a default catch to occur when booting from an iSCSI device.

• On systems with a host Ethernet adapter (HEA) or host channel adapter (HCA) assigned to a Linux partition, a problem was fixed that prevented the partition from booting if 512 GB, 1 TB, or 1.5 TB of memory was assigned to the partition.  When this problem occurred, SRC B700F105 was logged.
• In systems with clustered processors, various problems were fixed in the InfiniBand interconnection networks.
• A problem was fixed that, under certain circumstances, caused an AIX or Linux partition to fail to boot with SRC D200E0AF being logged.
• On systems with external I/O frames, a problem was fixed that might have prevented the firmware from "unthrottling" processors after entering power save mode.

08/28/08

System firmware changes that affect all systems:

• DEFERRED and HIPER:  A problem was fixed that, under certain rarely occurring circumstances, an application could cause a processor to go into an error state, and the system to crash.
• HIPER:  A problem was fixed that caused the system to terminate abnormally with SRC B131E504.
• HIPER:  A problem was fixed that might cause a partition to crash during a partition migration before the migration was complete.
• DEFERRED:  Enhancements were made to the system firmware to reduce the system boot time on power up.
• DEFERRED:  A problem was fixed such that under certain rare circumstances, if a system controller failover occurred, the new secondary system controller was not able to communicate with the system.
• DEFERRED:  A problem was fixed that caused SRC B1608CB0 to be logged if a separate I/O frame is attached to the CEC frame.
• A problem was fixed that caused multiple instances of SRC B1818A03 and B1818A0A to be logged erroneously, and multiple calls home to be made, during a frame connection reset.
• A problem was fixed that caused SRC B1819506 to be erroneously generated, and a call home to be made, when service processor (or system controller) error log entries were generated faster than they could be processed.
• A problem was fixed that caused the hardware management console (HMC) to show an "Incomplete" state after it attempted to read a file with an incorrect size from the service processor (or system controller).  This problem also occurred if the "factory configuration" option was used on the advanced system management interface (ASMI) menus.
• Enhancements were made to the firmware to improve the FRU callouts for certain types of failures of the time-of-day clock circuitry.
• A problem was fixed that prevented a dump file larger than 4 GB from being successfully off-loaded to the hardware management console (HMC).
• On systems with redundant bulk power controllers, a problem was fixed that caused the hardware management console (HMC) to get stuck at "Pending Authentication" for one of the bulk power controllers (BPCs).
• On systems with I/O drawers attached, a problem was fixed that might have caused some I/O slots in the drawers not to be configured when the system was booted.
• In systems with clustered processors, various problems were fixed in the InfiniBand interconnection networks.
• A problem was fixed that caused the location codes of the external InfiniBand ports on a 5791 I/O drawer with the InfiniBand interface to be reported incorrectly on the HMC.
• A problem was fixed that caused SRC B7006971 to be generated because the firmware was  incorrectly performing operations on PCI-Express I/O adapters during dynamic LPAR (DLPAR) operations on memory.
• A problem was fixed the might have caused an out-of-memory condition in the hypervisor, with SRC B7000200 being logged.
• A problem was fixed in the thermal management firmware that caused SRCs B1812635 and B1812636 to be logged, and the system or node to run in low power mode when it should have been in nominal, or nominal when it should have been in low power mode.
• A problem was fixed that caused SRC B1818A10 to be erroneously generated after a successful installation of system firmware.
• A problem was fixed that caused the AIX commands "lsmcode" and "diag" to fail after a partition migration.
• A problem was fixed that caused the message "BA330000malloc error!" to be displayed on the operating system console after a partition migration, even though SRC BA330000 had not been logged.  When this problem occurred, the partition migration appeared to be successful.  However, a process within the partition was either hung or had failed, and in most cased the partition had to be rebooted to fully recover.
• A problem was fixed that caused the status of the connection between the hardware management console (HMC) and the service processor to be set to an invalid state.  This might cause problems when the HMC and service processor tried to communicate.
• A problem was fixed that caused partitions that were being rebooted to hang at D200E0AF after a concurrent firmware update under certain circumstances.
• A problem was fixed that prevented the replacement of a system controller from completing successfully if the system controller had been guarded out prior to it replacement.
• A problem was fixed that caused the system controller to go through an unnecessary reset/reload cycle when a checkstop occurred or the system was powered off.
• Enhancements were made to the firmware to improve the FRU callouts for certain types of failures of the node controller.
• A problem was fixed that caused predictive SRC B181EF88 to be logged when, under certain circumstances, a system controller failover occurred at runtime.
• A problem was fixed such that if redundancy was disabled, and the emergency power off (EPO) switch was then used to power off the system, redundancy was erroneously enabled when the system came back up.
• Enhancements were made to the firmware to improve the FRU callouts for certain types of failures of a node controller.
• A problem was fixed such that caused the service processor (or system controller) to lose its communication link with the hypervisor, and SRC A181D000 to be logged, under certain rare circumstances.
• On systems using virtual shared processor pools (VSPP), a problem was fixed that caused the number of processors assigned to the partitions to be reduced after a memory-preserving IPL. 

06/10/08

This level is a disruptive update from the prior level, EH330_018.  The system should be powered off before installing this level of system firmware.  If this level is installed when the system is running, the CECs will be rebooted, causing all partitions to be terminated, and a reboot will be required. 

System firmware changes that affect all systems:

• HIPER:  A problem was fixed that caused a concurrent firmware installation to hang with SRC BA00E840 being logged.  This problem may also cause a partition migration to hang, under certain circumstances, with the same SRC, BA00E840, being logged.  This SRC will be logged when this level of firmware is installed and will generate a call home; it should be ignored.  It will not be logged during subsequent installations.
• HIPER: The processor initialization settings were changed to reduce the likelihood of a processor going into an error state and causing a checkstop or system crash.
• HIPER: A problem was fixed that, under certain circumstances, caused a system termination during a service processor failover.
• HIPER: A problem was fixed that caused large numbers of enhanced error handling (EEH) errors to be logged against the 4-port gigabit Ethernet adapter, F/C 5740, under certain circumstances.
• HIPER:  On systems with a redundant system controllers installed and enabled, a problem was fixed that might cause a communications hang between the two system controllers.  When this occurred, it triggered a reset/reload of the primary system controller, and the resulting fail-over to the secondary system controller failed in such a way that the system crashed. 
• Several problems were fixed that might cause one or both of the clock cards to be deconfigured, and erroneously called out as bad, when the system boots up from the power-off state. 
• A problem was fixed that caused the /tmp directory on the system controllers and the service processor in the bulk power controller (BPC) to fill up, which results in an out-of-memory condition.  When this problem occurred, the system controllers or service processor in the BPC usually performed a reset/reload.  This is one possible cause of SRC B1817201 being logged. 
• A problem was fixed that prevented the "i5/OS enable/disable" setting (in the ASMI power on/off menu) from taking effect when the system is booted.  This solution requires the system to be booted up to hypervisor standby twice after the setting is changed to "enabled". This will be fixed in a future service pack to remove the requirement for the second boot to hypervisor standby. 
• A problem was fixed that caused the firmware  to receive a false error indication when reading the registers on the LED controller.  SRC B1811340 was logged when this happened.
• A problem was fixed that prevented an error fail-over to the secondary system controller from completing successfully.
• A problem was fixed that might have caused a system firmware installation to fail with SRC B18138B7 being logged.
• A problem was fixed that caused an error log to be generated that called out system controller A (Un-P1-C2), instead of the correct callout, which was system controller B (Un-P2-C5).
• A problem was fixed that caused the P1 LED on the front light strip to be on when it should have been off.
• A problem was fixed that caused the wrong memory DIMM location to be called out when certain types of failures occurred.
• A problem was fixed that might have caused cache chip failures when the system is operating in Power Save mode.  Error log entries that might indicate that this problem is occurring include correctable errors and uncorrectable errors in L2, i-cache and d-cache memory, parity errors, and SRC B181E504. 
• The firmware was enhanced so that the IDs "celogin1" and "celogin2" allow an authorized service provider to log into the bulk power controller (BPC).
• A problem was fixed that caused a partition using a host channel adapter (HCA) or host Ethernet adapter (HEA) to appear to hang (with progress code D200C1FF being displayed) before successfully shutting down.  The amount of time the partition appeared to hang depended on the amount of memory assigned to the partition and the usage of HCA or HEA.
• A problem was fixed that prevented the HMC from connecting to the managed system if the HMC's DHCP server IP range is changed when the managed system is running.
• The error logging and FRU callout firmware was enhanced so that if a failure occurs on one or both clock cards, only one will get deconfigured, and the system will continue to try to boot instead of terminating.
• The firmware was enhanced to improve the system memory error recovery.
• The firmware was enhanced so that the contents of  the /tmp directory are included when a service processor dump is taken.
• A problem was fixed in the hypervisor that might cause a partition migration to fail.
• The firmware was enhanced so that:
• A failure when writing VPD to a P6 processor will cause the node to be deconfigured rather than terminating the system.
• The failure of a VPD write operation will  not corrupt the VPD table, which may lead to unnecessary system down-time and unnecessary FRU replacement.

• On systems using QLogic InfiniBand switches, a problem was fixed that caused the PortInfo:linkWidthActive and PortInfo:linkSpeedActive to be inaccurately stored and displayed on the display of subnet parameters.

05/13/08

• GA Level