• The ZTR_RTTCC algorithm parameters AI and HAI did not support a sufficient range.
• Coalescing regular SX events with SX RTT events under ZTR_RTTCC would keep improper event fields, which would impact congestion control behavior.
• Issue in the ZTR_RTTCC algorithm where probe-abortion handling would behave improperly under high-stress network conditions, affecting proper congestion control and stable traffic performance.
• An assertion failure that would occur with the E-Switch uplink in specific configurations where the e-switch was disabled and Path Migration was active or GVMIs were using SRQ loopback in SQs. The issue occurred because the firmware attempted to perform cleanup operations when the uplink configuration lacked sufficient capacity. Now, when the E-Switch is disabled and no actions are available in the uplink STE, the firmware connects to the uplink STE instead of copying it.
• The ConnectX-8 downstream port failed to send a NACK when rejecting an L1 entry request from the upstream port.
• Invoking the resourcedump tool with segment type DPA_PROCESS_LST returned invalid data when the parameter n1 == 1 and no processes existed on the current vhca_id. The fix adds a proper check, and the resourcedump tool now reports the correct error in this scenario.
• During failover or restart, the SM sending a PortInfo MAD to the HCA firmware triggered reinitialization of port buffers, momentarily halting ingress traffic and causing packet drops. The firmware now avoids reconfiguring port buffers when the new configuration matches the current one.
• MFRL operations would fail due to a timeout.
• Under the ZTR_RTTCC algorithm, a flow that reached its minimum rate due to heavy congestion would not recover its rate once the congestion cleared.
• A missing interrupt from the module IO (Expander) would prevent the module from being raised.
• Reading debug registers would cause link BER (Bit Error Rate) degradation over time.
• Ports connected via 4 or 8 lanes and configured for 200G_2x (using only 2 lanes) would fail to link when using a mix of new firmware (with “Non Tx-Squelch” support) and older firmware versions.

Note: Please make sure on both sides, switch (local device) and Switch/NIC (peer device) you:

• Deploy the new firmware release versions as a matched bundle on both Switch and NIC devices.
• Configure the port to use 2 lanes (instead of 4 or 8 lanes) while keeping the 200G_2x speed setting.
• Issue in Arch diagnostic data counters where the pcie_link_outbound_data_bytes counter was incorrectly returning only zero values.
• Multiple long-running process registers would cause aborted access and timeouts, the internal state is now properly handled.
• Enabling the CCMAD custom header on one PCC probe slot caused other slots to malfunction when multiple slots were configured.

Note: If using firmware versions older than the 40.47.10xx GA release, disable the CCMAD custom header when multiple probe slots are enabled.

• A CQE error with vendor_syndrome RDE_MAL_WQE (0xd6) could cause traffic disruption on the affected QP.

The following fixes have been included in version 40.46.3048:

• Fix for the adapter card dropping NC-SI over MCTP commands when padding bytes were present after the NC-SI checksum. 

This version of the System ROM contains updates aligned with Intel VROC v.9.3 updates.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Support for server Asset Tag length up to 63 characters. This requires a compatible iLO 7 v1.20.00 and above to be installed.

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

None

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

• The ZTR_RTTCC algorithm parameters AI and HAI did not support a sufficient range.
• Coalescing regular SX events with SX RTT events under ZTR_RTTCC would keep improper event fields, which would impact congestion control behavior.
• Issue in the ZTR_RTTCC algorithm where probe-abortion handling would behave improperly under high-stress network conditions, affecting proper congestion control and stable traffic performance.
• An assertion failure that would occur with the E-Switch uplink in specific configurations where the e-switch was disabled and Path Migration was active or GVMIs were using SRQ loopback in SQs. The issue occurred because the firmware attempted to perform cleanup operations when the uplink configuration lacked sufficient capacity. Now, when the E-Switch is disabled and no actions are available in the uplink STE, the firmware connects to the uplink STE instead of copying it.
• The ConnectX-8 downstream port failed to send a NACK when rejecting an L1 entry request from the upstream port.
• Invoking the resourcedump tool with segment type DPA_PROCESS_LST returned invalid data when the parameter n1 == 1 and no processes existed on the current vhca_id. The fix adds a proper check, and the resourcedump tool now reports the correct error in this scenario.
• During failover or restart, the SM sending a PortInfo MAD to the HCA firmware triggered reinitialization of port buffers, momentarily halting ingress traffic and causing packet drops. The firmware now avoids reconfiguring port buffers when the new configuration matches the current one.
• MFRL operations would fail due to a timeout.
• Under the ZTR_RTTCC algorithm, a flow that reached its minimum rate due to heavy congestion would not recover its rate once the congestion cleared.
• A missing interrupt from the module IO (Expander) would prevent the module from being raised.
• Reading debug registers would cause link BER (Bit Error Rate) degradation over time.
• Ports connected via 4 or 8 lanes and configured for 200G_2x (using only 2 lanes) would fail to link when using a mix of new firmware (with “Non Tx-Squelch” support) and older firmware versions.

Note: Please make sure on both sides, switch (local device) and Switch/NIC (peer device) you:

• Deploy the new firmware release versions as a matched bundle on both Switch and NIC devices.
• Configure the port to use 2 lanes (instead of 4 or 8 lanes) while keeping the 200G_2x speed setting.
• Issue in Arch diagnostic data counters where the pcie_link_outbound_data_bytes counter was incorrectly returning only zero values.
• Multiple long-running process registers would cause aborted access and timeouts, the internal state is now properly handled.
• Enabling the CCMAD custom header on one PCC probe slot caused other slots to malfunction when multiple slots were configured.

Note: If using firmware versions older than the 40.47.10xx GA release, disable the CCMAD custom header when multiple probe slots are enabled.

• A CQE error with vendor_syndrome RDE_MAL_WQE (0xd6) could cause traffic disruption on the affected QP.

The following fixes have been included in version 40.46.3048:

• Fix for the adapter card dropping NC-SI over MCTP commands when padding bytes were present after the NC-SI checksum. 

SPDM - Get Certificate response returns out of bound values

• Configuring a small MTU led to fragmentation of packets critical for the PXE boot process. As a result, the PXE boot filters mistakenly discarded these packets, causing the PXE boot to fail.

• Configuring PHY_RATE_MASK for 10G in NV settings incorrectly disabled 10G capabilities.
• RTT packets with any destination MAC address were incorrectly treated as having a valid destination MAC. The new firmware now discards RTT packets if their destination MAC does not match the port's MAC.

• Configuring a small MTU led to fragmentation of packets critical for the PXE boot process. As a result, the PXE boot filters mistakenly discarded these packets, causing the PXE boot to fail.

• PCC_CNP_COUNT could not be reset using the pcc_counter.sh script in the DOCA tools.
• VQoS algorithm issue related to learning when an element is active and when it begins sending traffic.
• A race condition that would prevent the application from transmitting when VQoS was enabled.
• Bandwidth would drop when unbinding multiple VFs with VQoS enabled.
• RTT packets with any destination MAC address were incorrectly treated as having a valid destination MAC. The new firmware now discards RTT packets if their destination MAC does not match the port's MAC.

This version of the System ROM contains updates aligned with Intel VROC v.9.3 updates.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Support for server Asset Tag length up to 63 characters. This requires a compatible iLO 7 v1.20.00 and above to be installed.

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

None

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected” in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration” error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

This version of the System ROM contains updates aligned with Intel VROC v.9.3 updates.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Support for server Asset Tag length up to 63 characters. This requires a compatible iLO 7 v1.20.00 and above to be installed.

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

None

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

This version of the System ROM contains updates aligned with Intel VROC v.9.3 updates.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Support for server Asset Tag length up to 63 characters. This requires a compatible iLO 7 v1.20.00 and above to be installed.

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

None

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

This version of the System ROM contains updates aligned with Intel Birch Stream UPLR2 and UPLR2.5 BKC updates.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system configured with Mirrored Memory in Advanced ECC mode could become unresponsive after performing Memory Address Translation via Redfish.

Addressed an issue where the system secure boot status could appear inconsistent between RBSU and iLO.

Addressed an issue where the BootProgress property’s OSRunning status in Redfish could be report incorrectly.

Addressed an issue where an error message might pop up when accessing Encrypted Device Page (System Utility > System Configurations > RBSU> Server Security > Device Encryption Option > Device Encryptions Settings > Encrypted Device)

Addressed an issue where the system could encounter insufficient resources when GPUs were configured in Shared Direct mode.

Addressed an issue where the system might encounter a Red Screen of Death (RSOD) after installing a SMART Modular CXL card.

Known Issues:

•     This firmware revision improves data integrity and reliability by optimizing adjacent track refresh parameters and management of emergency power loss table structure
•     The risk of not upgrading to this firmware is the increased possibility of data corruption in emergency power loss events
•     Several maintenance items are included in this firmware revision that reduce the probability of hangs and provide some minor performance improvements

• The ZTR_RTTCC algorithm parameters AI and HAI did not support a sufficient range.
• Coalescing regular SX events with SX RTT events under ZTR_RTTCC would keep improper event fields, which would impact congestion control behavior.
• Issue in the ZTR_RTTCC algorithm where probe-abortion handling would behave improperly under high-stress network conditions, affecting proper congestion control and stable traffic performance.
• An assertion failure that would occur with the E-Switch uplink in specific configurations where the e-switch was disabled and Path Migration was active or GVMIs were using SRQ loopback in SQs. The issue occurred because the firmware attempted to perform cleanup operations when the uplink configuration lacked sufficient capacity. Now, when the E-Switch is disabled and no actions are available in the uplink STE, the firmware connects to the uplink STE instead of copying it.
• MCTP SMBus configuration issue which affected proper initialization and reliable communication between firmware components using the SMBus transport.
• During failover or restart, the SM sending a PortInfo MAD to the HCA firmware triggered reinitialization of port buffers, momentarily halting ingress traffic and causing packet drops. The firmware now avoids reconfiguring port buffers when the new configuration matches the current one.
• Under the ZTR_RTTCC algorithm, a flow that reached its minimum rate due to heavy congestion would not recover its rate once the congestion cleared.
• Destroying or modifying a DPA partition from a non-owner VHCA was incorrectly allowed, such actions are now properly disallowed.
• PTP was not supported when the port speed was configured to 1G.

• The adapter card would drop NC-SI over MCTP commands when padding bytes were present after the NC-SI checksum.

For more information, refer to HPE Customer Advisory at the following URL: https://support.hpe.com/hpesc/public/docDisplay?docId=a00141941en_us 

• OFED installation caused CIFS to break in RHEL 8.4 and above. A dummy module was added so that CIFS will be disabled after OFED installation in RHEL 8.4 and above.
• Performance degradation on older kernel versions using RX cache, particularly on slower ARM CPUs with larger RX buffers. The issue was caused by the driver attempting to allocate new RX pages too quickly, leading to head-of-line blocking in the RX cache. The fix improves RX cache usage by triggering page allocation for a bulk of at least 2 WQEs, allowing the application more time to process packets and return buffers to the RX cache, thereby reducing blocking and enhancing performance.
• VF behavior was not aligned with /sys/class/net/<interface-name>/device/sriov_numvfs (silently ignore attempts to set the same number of VFs, and prevent changing the number of VFs until existing VFs are removed).
• On BlueField-3 devices running linux-bluefield kernel versions 5.15.0-1050 or 5.15.0-1060, a kernel crash occurred due to a NULL pointer dereference in the cls_api network scheduler.

• Buffer initialization became a performance bottleneck during the allocation of large buffers, typically when using a high number of QPs with large message sizes. The root cause was the inefficient use of rand(). This has been resolved by replacing it with a faster pseudo-random algorithm.
• A potential deadlock that could occur during the handling of peer memory registration failures.
• The following error occurred when switching to SwitchDev mode: "mlx5_core 0000:03:00.0: mlx5_cmd_out_err:835:(pid 24362): CREATE_FLOW_GROUP(0x933) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x4065f0), err(-22); mlx5_core 0000:03:00.0: E-Switch: Failed to create peer miss flow group err(-22)"
• A sysfs issue that occurred when accessing hardware counters from within a namespace.
• "ibstat" would fail and crash when encountering a non-RoCE/IB device, preventing it from displaying information for the remaining valid RoCE/IB devices.
• Bandwidth degradation in cases of high number of QPs, where polling only 16 CQEs per iteration may not be sufficient to process all completions in time.
• A race condition between firmware syndrome report and driver initialization during boot.
• The driver failed to load when a firmware syndrome was detected during boot.
• The mlnx_tune -l command did not list several operating systems that were in fact supported.
• The kernel did not define TCA_TUNNEL_KEY_ENC_SRC_PORT. To align offload behavior with non-offload, the OVS community introduced a commit ( netdev-offload-tc: Fix offload of tunnel key tp_src) that causes offload to fail if this tunnel attribute is used. Now, any rule with a tunnel set action that includes a tunnel source port can no longer be offloaded.
• LLDP traffic from VFs or BF host PFs was not reaching the representor kernel interfaces.
• Flows where both the inner and outer destination (dst) and/or source (src) MAC addresses were rewritten to the same value, the outer MAC address rewrite was ignored, leading to an outer MAC address of 00:00:00:00:00:00.
• Enabling sFlow with OVN caused OVS to crash.
• OVS crashed unexpectedly after DPUs repeatedly broadcast the error message “packet with own source address.”
• Changing the hw-offload setting from true to false while ports were configured led to errors reported in the OVS log.

• Fixed an issue by adding the handling of EAP packet with Identifier Value 0 in iLO.

• Fixed a file descriptor leak in iLO CLI.

• Fixed an issue where iLO is taking 20-30 minutes to be accessible after Aux Power Cycle during a firmware flash on HPE ProLiant Compute XD685 with NVIDIA H200 GPU trays.

• Fixed an issue where NVIDIA HGX PCIe Switch versions are not being displayed on HPE ProLiant Compute XD685 with NVIDIA H200 GPU trays.

• Fixed issues with auxiliary power reset not working properly on AMD Instinct platforms on HPE ProLiant Compute XD685.

• Improved firmware update progress percentage when flashing NVIDIA HGX GPU or AMD Instinct GPU firmware packages on HPE ProLiant Compute XD685.

• Fixed issues with GPU management failing to work after firmware update on HPE ProLiant Compute XD685 with AMD Instinct MI355X GPUs.

• Fixed issues with AMD Pensando Pollara 400 AI NIC firmware version getting partially truncated on iLO Firmware Inventory page.

• OFED installation caused CIFS to break in RHEL 8.4 and above. A dummy module was added so that CIFS will be disabled after OFED installation in RHEL 8.4 and above.
• Performance degradation on older kernel versions using RX cache, particularly on slower ARM CPUs with larger RX buffers. The issue was caused by the driver attempting to allocate new RX pages too quickly, leading to head-of-line blocking in the RX cache. The fix improves RX cache usage by triggering page allocation for a bulk of at least 2 WQEs, allowing the application more time to process packets and return buffers to the RX cache, thereby reducing blocking and enhancing performance.
• VF behavior was not aligned with /sys/class/net/<interface-name>/device/sriov_numvfs (silently ignore attempts to set the same number of VFs, and prevent changing the number of VFs until existing VFs are removed).
• On BlueField-3 devices running linux-bluefield kernel versions 5.15.0-1050 or 5.15.0-1060, a kernel crash occurred due to a NULL pointer dereference in the cls_api network scheduler.

• Buffer initialization became a performance bottleneck during the allocation of large buffers, typically when using a high number of QPs with large message sizes. The root cause was the inefficient use of rand(). This has been resolved by replacing it with a faster pseudo-random algorithm.
• A potential deadlock that could occur during the handling of peer memory registration failures.
• The following error occurred when switching to SwitchDev mode: "mlx5_core 0000:03:00.0: mlx5_cmd_out_err:835:(pid 24362): CREATE_FLOW_GROUP(0x933) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x4065f0), err(-22); mlx5_core 0000:03:00.0: E-Switch: Failed to create peer miss flow group err(-22)"
• A sysfs issue that occurred when accessing hardware counters from within a namespace.
• "ibstat" would fail and crash when encountering a non-RoCE/IB device, preventing it from displaying information for the remaining valid RoCE/IB devices.
• Bandwidth degradation in cases of high number of QPs, where polling only 16 CQEs per iteration may not be sufficient to process all completions in time.
• A race condition between firmware syndrome report and driver initialization during boot.
• The driver failed to load when a firmware syndrome was detected during boot.
• The mlnx_tune -l command did not list several operating systems that were in fact supported.
• The kernel did not define TCA_TUNNEL_KEY_ENC_SRC_PORT. To align offload behavior with non-offload, the OVS community introduced a commit ( netdev-offload-tc: Fix offload of tunnel key tp_src) that causes offload to fail if this tunnel attribute is used. Now, any rule with a tunnel set action that includes a tunnel source port can no longer be offloaded.
• LLDP traffic from VFs or BF host PFs was not reaching the representor kernel interfaces.
• Flows where both the inner and outer destination (dst) and/or source (src) MAC addresses were rewritten to the same value, the outer MAC address rewrite was ignored, leading to an outer MAC address of 00:00:00:00:00:00.
• Enabling sFlow with OVN caused OVS to crash.
• OVS crashed unexpectedly after DPUs repeatedly broadcast the error message “packet with own source address.”
• Changing the hw-offload setting from true to false while ports were configured led to errors reported in the OVS log.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where reverting an Opal drive to factory default using PSID after previously selecting it in the "Encrypted Devices" menu caused an error when navigating back to "Encrypted Devices."

Addressed an issue where, after an AC cycle with Secure Boot enabled, the SecureBootCurrentBoot setting became disabled, causing the iLO web interface to incorrectly display Secure Boot as disabled.

Addressed an issue where boot time was not consistently reported to boot progress in all states.

Addressed an issue where changes made in the RBSU were not persisting when using OpenBMC.

Addressed an issue where the RBSU menu displayed an option to log in using iLO credentials even when OpenBMC was detected.

Addressed an issue where the virtual USB device name was incorrectly shown as "iLO Virtual USB" instead of "BMC Virtual USB" when OpenBMC was detected.

Known Issues:

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system might experience Keep Alive Timeout when establishing NVMe-oF connection.

Known Issues:

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

This version of the System ROM contains updates aligned with Intel Birch Stream UPLR2 and UPLR2.5 BKC updates.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system configured with Mirrored Memory in Advanced ECC mode could become unresponsive after performing Memory Address Translation via Redfish.

Addressed an issue where the system secure boot status could appear inconsistent between RBSU and iLO.

Addressed an issue where the BootProgress property’s OSRunning status in Redfish could be report incorrectly.

Addressed an issue where an error message might pop up when accessing Encrypted Device Page (System Utility > System Configurations > RBSU> Server Security > Device Encryption Option > Device Encryptions Settings > Encrypted Device)

Addressed an issue where the system could encounter insufficient resources when GPUs were configured in Shared Direct mode.

Addressed an issue where the system might encounter a Red Screen of Death (RSOD) after installing a SMART Modular CXL card.

Known Issues:

• Configuring a small MTU led to fragmentation of packets critical for the PXE boot process. As a result, the PXE boot filters mistakenly discarded these packets, causing the PXE boot to fail.

• Configuring PHY_RATE_MASK for 10G in NV settings incorrectly disabled 10G capabilities.
• RTT packets with any destination MAC address were incorrectly treated as having a valid destination MAC. The new firmware now discards RTT packets if their destination MAC does not match the port's MAC.

•     corner case performance enhance.
•     Servo change that resolves issue with early heat application from Idle sweep seek terminated that may cause read errors

• The ZTR_RTTCC algorithm parameters AI and HAI did not support a sufficient range.
• Coalescing regular SX events with SX RTT events under ZTR_RTTCC would keep improper event fields, which would impact congestion control behavior.
• Issue in the ZTR_RTTCC algorithm where probe-abortion handling would behave improperly under high-stress network conditions, affecting proper congestion control and stable traffic performance.
• An assertion failure that would occur with the E-Switch uplink in specific configurations where the e-switch was disabled and Path Migration was active or GVMIs were using SRQ loopback in SQs. The issue occurred because the firmware attempted to perform cleanup operations when the uplink configuration lacked sufficient capacity. Now, when the E-Switch is disabled and no actions are available in the uplink STE, the firmware connects to the uplink STE instead of copying it.
• MCTP SMBus configuration issue which affected proper initialization and reliable communication between firmware components using the SMBus transport.
• During failover or restart, the SM sending a PortInfo MAD to the HCA firmware triggered reinitialization of port buffers, momentarily halting ingress traffic and causing packet drops. The firmware now avoids reconfiguring port buffers when the new configuration matches the current one.
• Under the ZTR_RTTCC algorithm, a flow that reached its minimum rate due to heavy congestion would not recover its rate once the congestion cleared.
• Destroying or modifying a DPA partition from a non-owner VHCA was incorrectly allowed, such actions are now properly disallowed.
• PTP was not supported when the port speed was configured to 1G.

• The adapter card would drop NC-SI over MCTP commands when padding bytes were present after the NC-SI checksum.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Address an issue where the system reports “266 – Non-Volatile Memory Corruption Detected in the IML logs.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Address an issue where the system logs correctable errors as Uncorrectable Machine Check Exceptions in the IML and AHS.

Address an issue where the system reports incorrect slot information for NVMe devices when the PCIe link cannot train to its maximum supported speed.

Address an issue where the system cannot enter the RBSU menu after Intel VMD/VROC is enabled.

Address an issue where the system reports an “Unsupported DIMM Configuration error when Granite Rapids XCC processors are installed with a 2-DIMM configuration.

Address an issue where certain devices are incorrectly configured with 10-bit TLP tags even though they do not support this feature.

Address an issue where incorrect OCP E610 NIC information is displayed in the RBSU one-time boot menu and PCIe device information (DL380a).

Known Issues:

This version of the System ROM contains updates aligned with Intel Birch Stream UPLR2 and UPLR2.5 BKC updates.

To streamline content and avoid duplication across documents, starting the Gen12 platform, the UEFI Workload Based Profiles and Tuning Guide has been deprecated, and the contents of this document are now available as part of the current UEFI System Utilities User Guide for HPE Compute Gen12 servers based on iLO7 (https://hpe.com/support/UEFIG12-iLO7-UG-en).

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system configured with Mirrored Memory in Advanced ECC mode could become unresponsive after performing Memory Address Translation via Redfish.

Addressed an issue where the system secure boot status could appear inconsistent between RBSU and iLO.

Addressed an issue where the BootProgress property’s OSRunning status in Redfish could be report incorrectly.

Addressed an issue where an error message might pop up when accessing Encrypted Device Page (System Utility > System Configurations > RBSU> Server Security > Device Encryption Option > Device Encryptions Settings > Encrypted Device)

Addressed an issue where the system could encounter insufficient resources when GPUs were configured in Shared Direct mode.

Addressed an issue where the system might encounter a Red Screen of Death (RSOD) after installing a SMART Modular CXL card.

Known Issues:

• OFED installation caused CIFS to break in RHEL 8.4 and above. A dummy module was added so that CIFS will be disabled after OFED installation in RHEL 8.4 and above.
• Performance degradation on older kernel versions using RX cache, particularly on slower ARM CPUs with larger RX buffers. The issue was caused by the driver attempting to allocate new RX pages too quickly, leading to head-of-line blocking in the RX cache. The fix improves RX cache usage by triggering page allocation for a bulk of at least 2 WQEs, allowing the application more time to process packets and return buffers to the RX cache, thereby reducing blocking and enhancing performance.
• VF behavior was not aligned with /sys/class/net/<interface-name>/device/sriov_numvfs (silently ignore attempts to set the same number of VFs, and prevent changing the number of VFs until existing VFs are removed).
• On BlueField-3 devices running linux-bluefield kernel versions 5.15.0-1050 or 5.15.0-1060, a kernel crash occurred due to a NULL pointer dereference in the cls_api network scheduler.

• Buffer initialization became a performance bottleneck during the allocation of large buffers, typically when using a high number of QPs with large message sizes. The root cause was the inefficient use of rand(). This has been resolved by replacing it with a faster pseudo-random algorithm.
• A potential deadlock that could occur during the handling of peer memory registration failures.
• The following error occurred when switching to SwitchDev mode: "mlx5_core 0000:03:00.0: mlx5_cmd_out_err:835:(pid 24362): CREATE_FLOW_GROUP(0x933) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x4065f0), err(-22); mlx5_core 0000:03:00.0: E-Switch: Failed to create peer miss flow group err(-22)"
• A sysfs issue that occurred when accessing hardware counters from within a namespace.
• "ibstat" would fail and crash when encountering a non-RoCE/IB device, preventing it from displaying information for the remaining valid RoCE/IB devices.
• Bandwidth degradation in cases of high number of QPs, where polling only 16 CQEs per iteration may not be sufficient to process all completions in time.
• A race condition between firmware syndrome report and driver initialization during boot.
• The driver failed to load when a firmware syndrome was detected during boot.
• The mlnx_tune -l command did not list several operating systems that were in fact supported.
• The kernel did not define TCA_TUNNEL_KEY_ENC_SRC_PORT. To align offload behavior with non-offload, the OVS community introduced a commit ( netdev-offload-tc: Fix offload of tunnel key tp_src) that causes offload to fail if this tunnel attribute is used. Now, any rule with a tunnel set action that includes a tunnel source port can no longer be offloaded.
• LLDP traffic from VFs or BF host PFs was not reaching the representor kernel interfaces.
• Flows where both the inner and outer destination (dst) and/or source (src) MAC addresses were rewritten to the same value, the outer MAC address rewrite was ignored, leading to an outer MAC address of 00:00:00:00:00:00.
• Enabling sFlow with OVN caused OVS to crash.
• OVS crashed unexpectedly after DPUs repeatedly broadcast the error message “packet with own source address.”
• Changing the hw-offload setting from true to false while ports were configured led to errors reported in the OVS log.

• The ZTR_RTTCC algorithm parameters AI and HAI did not support a sufficient range.
• Coalescing regular SX events with SX RTT events under ZTR_RTTCC would keep improper event fields, which would impact congestion control behavior.
• Issue in the ZTR_RTTCC algorithm where probe-abortion handling would behave improperly under high-stress network conditions, affecting proper congestion control and stable traffic performance.
• An assertion failure that would occur with the E-Switch uplink in specific configurations where the e-switch was disabled and Path Migration was active or GVMIs were using SRQ loopback in SQs. The issue occurred because the firmware attempted to perform cleanup operations when the uplink configuration lacked sufficient capacity. Now, when the E-Switch is disabled and no actions are available in the uplink STE, the firmware connects to the uplink STE instead of copying it.
• MCTP SMBus configuration issue which affected proper initialization and reliable communication between firmware components using the SMBus transport.
• During failover or restart, the SM sending a PortInfo MAD to the HCA firmware triggered reinitialization of port buffers, momentarily halting ingress traffic and causing packet drops. The firmware now avoids reconfiguring port buffers when the new configuration matches the current one.
• Under the ZTR_RTTCC algorithm, a flow that reached its minimum rate due to heavy congestion would not recover its rate once the congestion cleared.
• Destroying or modifying a DPA partition from a non-owner VHCA was incorrectly allowed, such actions are now properly disallowed.
• PTP was not supported when the port speed was configured to 1G.

• The adapter card would drop NC-SI over MCTP commands when padding bytes were present after the NC-SI checksum.

• OFED installation caused CIFS to break in RHEL 8.4 and above. A dummy module was added so that CIFS will be disabled after OFED installation in RHEL 8.4 and above.
• Performance degradation on older kernel versions using RX cache, particularly on slower ARM CPUs with larger RX buffers. The issue was caused by the driver attempting to allocate new RX pages too quickly, leading to head-of-line blocking in the RX cache. The fix improves RX cache usage by triggering page allocation for a bulk of at least 2 WQEs, allowing the application more time to process packets and return buffers to the RX cache, thereby reducing blocking and enhancing performance.
• VF behavior was not aligned with /sys/class/net/<interface-name>/device/sriov_numvfs (silently ignore attempts to set the same number of VFs, and prevent changing the number of VFs until existing VFs are removed).
• On BlueField-3 devices running linux-bluefield kernel versions 5.15.0-1050 or 5.15.0-1060, a kernel crash occurred due to a NULL pointer dereference in the cls_api network scheduler.

• Buffer initialization became a performance bottleneck during the allocation of large buffers, typically when using a high number of QPs with large message sizes. The root cause was the inefficient use of rand(). This has been resolved by replacing it with a faster pseudo-random algorithm.
• A potential deadlock that could occur during the handling of peer memory registration failures.
• The following error occurred when switching to SwitchDev mode: "mlx5_core 0000:03:00.0: mlx5_cmd_out_err:835:(pid 24362): CREATE_FLOW_GROUP(0x933) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x4065f0), err(-22); mlx5_core 0000:03:00.0: E-Switch: Failed to create peer miss flow group err(-22)"
• A sysfs issue that occurred when accessing hardware counters from within a namespace.
• "ibstat" would fail and crash when encountering a non-RoCE/IB device, preventing it from displaying information for the remaining valid RoCE/IB devices.
• Bandwidth degradation in cases of high number of QPs, where polling only 16 CQEs per iteration may not be sufficient to process all completions in time.
• A race condition between firmware syndrome report and driver initialization during boot.
• The driver failed to load when a firmware syndrome was detected during boot.
• The mlnx_tune -l command did not list several operating systems that were in fact supported.
• The kernel did not define TCA_TUNNEL_KEY_ENC_SRC_PORT. To align offload behavior with non-offload, the OVS community introduced a commit ( netdev-offload-tc: Fix offload of tunnel key tp_src) that causes offload to fail if this tunnel attribute is used. Now, any rule with a tunnel set action that includes a tunnel source port can no longer be offloaded.
• LLDP traffic from VFs or BF host PFs was not reaching the representor kernel interfaces.
• Flows where both the inner and outer destination (dst) and/or source (src) MAC addresses were rewritten to the same value, the outer MAC address rewrite was ignored, leading to an outer MAC address of 00:00:00:00:00:00.
• Enabling sFlow with OVN caused OVS to crash.
• OVS crashed unexpectedly after DPUs repeatedly broadcast the error message “packet with own source address.”
• Changing the hw-offload setting from true to false while ports were configured led to errors reported in the OVS log.

SFP Parameters value decimal part is truncated

This version of the System ROM contains updates aligned with Intel Kaseyville Granite Rapids-D XCC PV & HCC PLR4 BKC update.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

Support for server Asset Tag length up to 63 characters. This requires a compatible iLO 7 v1.20.00 and above to be installed.

Problems Fixed:

Addressed an issue where OS driver may have intermittent instability issue when system is using GNSS module via UART interface.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Known Issues:

None

This version of the System ROM contains updates aligned with Intel Kaseyville Granite Rapids-D XCC PV & HCC PLR4 BKC update.

Details about reported security vulnerabilities and their mitigation can be found at the following link Security Bulletin Library | HPE Support.

Firmware Dependencies:

None

Problems Fixed:

Addressed an issue where OS driver may have intermittent instability issue when system is using GNSS module via UART interface.

Address an issue where the system Asset Tag can be modified even when Asset Tag Protection is set to Locked.

Known Issues:

None