Power11 System Firmware Fix History - Release levels OB11xx

New Features and Functions

• Support was added for PCIe x16 dual monitor graphics adapter (FC EC90).
• Support was enhanced for Link Layer Discovery Protocol (LLDP) capability when an adapter is in SR-IOV shared mode, allowing collection of network topology information via network switches.
• Support was added for Link Layer Discovery Protocol (LLDP) on IBM Power System E1180 (9080-HEU).
• Support was added for “inband” HMC to OS communication (RMC Connection). This eliminates external network dependencies for RMC for OS versions that are enabled for in-band RMC.
• Support was added for The Linux Integrity Measurement Architecture (IMA). IMA verifies signatures against critical BMC software components, measuring and logging these measurements in the TPM and allowing attestation of these results at runtime.
• Support was added to auto provision vTPM during partition creation, provisioning a signed endorsement key certificate during instantiation to provide that proof of authenticity.

System firmware changes that affect all systems

• A problem was fixed for systems 9824-42A and 9856-42H, with Feature Codes EP3X and EP4C respectively, were configured with an incorrect software tier of P30. The fix will allow the systems to be configured with the correct software tier of P20.
• A problem was fixed to allow the system to attempt to diagnose clocking issues within the redundant clock system that were always flagged as preventative errors that require a technician to come out and check the system. This allows the system to recover gracefully if there is a transient issue in the system, or flag as a preventative error if the issue is not resolved.
• A problem was fixed where error logs were not created for correctable or uncorrectable BMC memory errors.
• A problem was fixed where error SRC - BD561008 is logged for the missing lcd panel even though panel is accessible.
• A problem was fixed where user cannot direct connect to BMC eth0/eth1 when gateway is set to non zero address.
• A security problem was fixed for CVE-2025-38556.
• A change was made to make the BC8A090F error informational until the Power Management complex activation has finished all retries.
• A problem was fixed by providing help text where the Status value of Sensors page confuses the user in ASM GUI.
• A problem was fixed for a system going into a "No Connection" state to the VMI (xxxxx) from the HMC. The HMC lssysconn command will show vmi_state=No connection.
• A problem was fixed for the Virtualization Management Interface where it can generate resource dumps. Platform Event Logs with SRC BD8D6025 and SRC BD8D6029 will also be produced because of the failure.
• A problem was fixed in BMC network where the direct HMC connection using the static network fails. System goes to NoConnection state, as BMC can not route the packets back to HMC.
• A problem was fixed to prevent an SRC BC8A264F.
• A problem was fixed where 12V current share warning error log SRC 110015F1 was incorrectly showing as serviceable PEL.
• A problem was fixed where a user will see any ACF certificate can be uploaded irrespective of the type selected.
• A problem was fixed where BMC ASM GUI button hover was overlapping with loading bar.
• A problem was fixed where BMC ASM PCIe topology page was always empty.
• A problem was fixed where BMC may reboot unexpectedly with an error SRC BD8DC201.
• A problem was fixed where customer was seeing invalid characters on Inventory and LED page.
• A problem was fixed where error log download content is partially empty when it is downloaded from ASMI or any clients.
• A problem was fixed where fabric adapters information was missing in the ASM GUI.
• A problem was fixed where firmware code update from the Host Operating system failed with the SRC BD8DC201.
• A problem was fixed where HTTP 500 error status returned for redfish call to get event logs after BMC reboot.
• A problem was fixed where if BMC reboots when the system is at runtime state an unexpected BMCDUMP may be logged.
• A problem was fixed where Overview page of ASMI shows power cap value when it is disabled.
• A problem was fixed where policy to guard bad hardware was fixed for SRCs with the format BxxxE51x.
• A problem was fixed where ReadOnly account modification via ASM will display the error message.
• A problem was fixed where the CEC would be stuck in the initialization stage during a firmware downgrade.
• A problem was fixed where the client overwhelms the BMC with too many requests and the BMC holds the requests open longer than needed, leading to a slower recovery.
• A problem was fixed where the 'Previously Activated' and 'Entry Check' entries under 'Processor Information' and 'Memory Information' on the Capacity on Demand page were displaying default values.
• A problem was fixed where the user had to use ServiceLogin Certificate option to upload Admin reset certificate in ASM GUI.
• A problem was fixed where a user logged in as a read-only user did not receive a general error message along with the 'User Unauthorized' message when attempting actions on the ASM Server Power Operations page or Firmware page.
• A problem was fixed where user was able to click the BMC GUI table filter checkbox name instead of the actual checkbox rendering table unfiltered.
• A problem was fixed where user was not able to clear guard records using the ASMI GUI. As a workaround, the customer can use the Guard command line tool to clear the core guard records.
• Changes were made to support only for the following MACS: hmac-sha2-256-etm@openssh.com, hmac-sha2-512-etm@openssh.com, hmac-sha2-256, hmac-sha2-512.
• A change was made to correctly report DDR5 memory characteristics to OS inventory. No functional impact, effective after system IPL.
• BMC support was added for Linux IMA (Integrity Measurement Architecture) to create informational PELs when a binary on the BMC is executed and its Linux IMA signature verification fails.
• BMC support was added for Linux IMA (Integrity Measurement Architecture) will create informational PELs when a binary on the BMC is executed and its Linux IMA signature verification fails.
• Support was added so that an informational PEL will be created when a BMC factory reset is requested.
• A problem was fixed when removing a system from Power Enterprise Pool 2.0 when there are guarded cores in the system.
• A problem was fixed for a failure or partial completion for a processor config change for a partition. As a workaround, the processor config change can be retried.
• A problem was fixed where B400FF04 SRCs are seen on the system after non-concurrently replacing an SR-IOV adapter while it is in shared mode and has logical ports configured.
• A problem was fixed where the error message "HSCL146A The I/O adapter: null is not valid." may be displayed by the management console when transitioning an SR-IOV adapter from shared to dedicated mode even though the operation was successful.
• The version of OpenSSL used by PowerVM has been updated to OpenSSL 3.0.18.
• A security problem was fixed for CVE-2025-52497 and CVE-2025-49087.
• A security problem was fixed for CVE-2025-36194.
• A security problem was fixed for CVE-2025-36238.
• A change was made to move SR-IOV adapters up to the latest level of adapter firmware. This update contains important reliability enhancements and performance fixes for adapters in promiscuous mode. This change updates the adapter firmware to XX.43.3608 for the following Feature Codes and CCIN: #EC71/EC72 with CCIN 2CF9, #EC75,EC76 with CCIN 2CFB, #EC85,EC86 with CCIN EC2C, and #EN24,EN26 with CCIN EC2A. Applying the update will cause a re-initialization of the specific SR-IOV adapter environment, causing a brief outage. Update instructions: https://www.ibm.com/docs/en/power10?topic=adapters-updating-sr-iov-adapter-firmware
• A change was made to suppress logging TPM failures with SRC B7009009 during TPM health checks.
• A problem has been fixed where a combination of communication fails to the power IC devices on a DDIMM could result in an IPL crash, with no log indicating which parts caused the fail. This fix will properly call out and deconfigure the failing DDIMM.
• A problem has been fixed to improve the firmware selection of processor cores added to an active dedicated processor partition.
• A problem was fixed for a multi-port adapter failing to configure or encountering performance impacts. The problem only pertains to multi-port adapters in an ENZ0 PCIe4 expansion drawer. The problem does not pertain to adapters running in SR-IOV shared mode. After installing the firmware service pack containing this fix, the partition(s) owning the impacted adapter(s) will need to be rebooted or the adapter(s) will need to be dynamically removed and added back to the partition(s) for the fix to activate for the impacted adapter(s).
• A problem was fixed for a possible system termination with SRC B700F105 after changing the I/O configuration or deleting a partition that is in the failed state with the SRC B2001230.
• A problem was fixed for a possible system termination with SRC B700F105 when running KVM in a PowerVM LPAR.
• A problem was fixed for a system state of incomplete on the management console which is not recoverable with a "Recover system" option of the managed system. The problem occurs when a new partition is created on the system, but the internal representation of the partition is not completed. The fix allows for the system firmware to validate partition creation is complete and will synchronize inconsistencies in the partition representation.
• A problem was fixed for network adapters with Feature Codes EC24/EC26 where traffic loss for SR-IOV logical ports configured on the adapter may occur and SRC B400FF02 platform event log will be created. This problem may intermittently occur when the adapter is configured in shared mode and a manually initiated resource dump is initiated from the HMC or ASMI targeting the adapter.
• A problem was fixed in firmware that prevented the external stand-alone USB DVD (Feature EUA5) from being selected as a boot device via the SMS menu and the SMS macros.
• A problem was fixed in the Power Hypervisor code where resources assigned to Virtual Functions may be reallocated even when there are configured VF resources for partitions. This could happen during hardware repair or replacement (FRU Repair or Exchange) if the VF resources weren’t removed from the operating system before starting the procedure.
• A problem was fixed where a concurrent firmware update would fail and the system would go to Incomplete state on the HMC.
• A problem was fixed that prevented the HMC from displaying daily notifications on the GUI and sending daily email notifications for a system in a Power Enterprise Pool 2.0 (PEP 2.0) whose PEP 2.0 Capacity on Demand code was within 7 days of expiration.
• A problem was fixed to continue to configure all the PCIe Switch slots under a PCIe Switch when one of the slot initialization operations fails under the PCIe Switch. The problem occurs only when there is an issue with one or more slots in IPL under PCIe Switch. This fix is applicable for all PCIe Switches in System Units and in IO Expansion Units.
• A problem was fixed where a B2006002 SRC may be seen when re-IPLing a system with 2 or more SR-IOV adapters in shared mode, even though no adapters were replaced. The failed SR-IOV adapter can be powered on again and it should boot correctly.
• A problem was fixed where a serviceable event for a processor issue was logged with an incorrect call out.
• A problem was fixed where AIX Update Access Key (UAK) expiration HMC notifications are logged along with A7004763 and A7004764 SRCs on IBMi partitions on systems with no AIX LP installed. With the fix, these notices will be suppressed if no AIX partition exists on the system.
• A problem was fixed where booting to the SMS menu and executing a specific SMS key sequence could cause undefined behavior. The problem may occur if a user executes the following key sequence upon boot to the SMS menu: 5 (Select Boot Options) ->1 (Select/Install boot device) ->5 (List all Devices).
• A problem was fixed where firmware might get out of sync after a runtime code update followed by a memory-preserving IPL. The fix addresses the issue and ensures proper security checks pass with the new code.
• A problem was fixed where if all TPM (trusted platform module) hardware has been marked as failed in the hypervisor, the hypervisor will block live partition migrations with the system key and capacity on demand activation codes from being applied.
• A problem was fixed where long running CoD based firmware operations, such as the processing of applied entitlements, may time-out. This results in the server state being set to incomplete on the HMC, which will require a rebuild of the server.
• A problem was fixed where Power Enterprise Pool 2.0 (PEP 2.0) and virtual serial number activation codes applied on the HMC may time-out.
• A problem was fixed where the hypervisor could allow memory to be over-configured on the system, leading to inconsistent memory values and errors on the management console.
• A problem was fixed where the hypervisor was reserving too much memory for SR-IOV adapters that are configured in shared mode. Activating this service pack will automatically resolve the issue and free up memory on the system.
• A problem was fixed where the SMS boot position (index) may not populate properly. SMS options: 5 (Select Boot Options) -> 1 (Select/Install boot device) ->5 (List all Devices). Workaround: SMS options: 5 (Select Boot Options) -> 2 (Configure Boot Device Order) -> [ESC] -> 5 (Select Boot Options) -> 1 (Select/Install boot device) -> 5 (List all Devices).
• A problem was fixed where unnecessary SBE Updates were being performed on Cold IPLs (ie, the system is powered off before IPLing/booting). The solution better determines if a SBE Update is really necessary on a Cold IPL and saves IPL time by bypassing any unnecessary SBE Updates.
• A problem was fixed where preventative errors were being flagged incorrectly and unnecessarily during testing. This fix reconfigures the flags that are set for watching for these preventative errors.

• A problem was fixed in the BMC network where the direct HMC connection using the static network fails.  System goes to NoConnection state, as BMC cannot route the packets back to HMC.
• A problem was fixed that could cause reduced system performance due to incorrect core state handling on certain processor configurations, including field core override settings and guarded processors.z