• Addressed an issue where SNMP traps were not being sent after an OA reboot or Active to Standby role change occurred when the IPSWAP feature was enabled and the SNMP trap receivers are configured using their hostname.
• Limited the “ADD SNMP TRAPRECEIVE” CLI command username length to 32 characters.
• Addressed an issue which enables OA to disallow the configuration of SNMP Trap receiver names that start with a numeral.
• Addressed an issue in OA CLI command “SHOW SERVER INFO” as well as in the GUI where it did not display new processor family for Integrity i6 blade.
• Addressed an issue where OA used to send inappropriate power request denial code when there was an ekeying error, which in turn made iLO report inappropriate errorcode.
• Addressed an issue where in after OA fail over, the SNMP failover trap was not getting generated if the OA configuration contains IPV6 only network and Enclosure IP mode enabled.
• Addressed an ARP flux issue in Standby OA’s IP address, due to which Network switches connected to OA may observe potential duplicate IP address of standby OA.
• Addressed an issue in the “RESET ILO” CLI command which prevented OA from resetting Integrity Blades.

Security

     The following security issues were fixed:

• CVE-2016-2177 -Addressed a vulnerability against openSSL was incorrectly using pointer arithmetic for heap-buffer boundary checks which might allow remote attackers to cause a denial of service.
• CVE-2016-6302 - Addressed a vulnerability against openSSL integer underflow flaw leading to a buffer over-read which allows remote attackers to cause a denial of service.
• CVE-2016-6304 Addressed a vulnerability against openSSL which could lead to Denial Of Service attack through memory exhaustion which might allow remote attackers to cause a denial of service (memory consumption) via large OCSP Status Request extensions.
• CVE-2016-6306- Addressed a vulnerability against openSSL where in  some missing message length checks can result in OOB reads of up to 2 bytes beyond an allocated buffer might allow remote attackers to cause a denial of service (out-of-bounds read) via crafted certificate operations.

Issues and workarounds

Browsers

• OA GUI is not accessible in Chrome versions 43.0.2357.10 to 44.0.2383.The issue was caused by a “regression” in Chrome (or WebKit). Customers should use an alternative browser like Firefox or Internet Explorer or try a different version of Chrome.
• SSO-to-iLO connection from the OA using an iLO host name fails with Microsoft Internet Explorer11 on Windows 8.On a Windows 8 system with Internet Explorer 10 or Internet Explorer 11, if the OA web GUI session is loaded using a host name instead of an IP address, an attempt to open an iLO window using SSO from the OA web GUI might result in the iLO page loading in the OA web GUI window instead of the intended new window. This issue was determined to be a bug in Internet Explorer and is expected to be fixed in a future release or update for Internet Explorer. To work around this issue, either use an IP address to load the OA Web GUI, or turn off Protected Mode for the appropriate zone in Internet Explorer’s settings. This issue occurs only on Internet Explorer browsers.

FIPS

Certificates smaller than 2048 bits in size are not compliant with FIPS requirements as enforced by the OA firmware starting with OA 4.20. When the OA running OA firmware version 4.40 or greater is operating in FIPS Mode ON/DEBUG and is configured with a 1024-bit LDAP certificate that was installed when running a previous version of OA firmware, FIPS Mode ON/DEBUG is considered to be operating in a degraded state due to the presence of the non-compliant certificate. While operating in this FIPS-Degraded Mode operational state, attempts to set FIPS Mode OFF from the OA GUI Network Access>FIPS tab will fail and show the error message The selected FIPS mode is already enabled. When the non-compliant certificate is removed, the FIPS-Degraded operational status is cleared, FIPS Mode can then be successfully set to OFF from the GUI interface. Note that the OA CLI command SET FIPS MODE OFF can be successfully used to set FIPS Mode OFF even with non-compliant 1024-bit LDAP certificates installed in the OA.

IRC

Unable to open .net IRC console for Gen10 Blades, Gen9 Blades also have the same issue. The Java applet and Webstart however, loads but the virtual media mounting fails. The work around is to launch the IRC through IRC Application (HP Lights-Out  Stand Alone Remote Console) which is installed on terminal client.

EFM

To use EFM on Gen 10 Blades, please select options/filters “Make Bootable ISO file” and “Enclosure Firmware Management” while creating custom SPP ISO on HPSUM 8.0.0. Please refer to HPSUM 8.0.0 User guide for further details.

CAC

• In the CAC mode SSH, Telnet and XML Reply protocols will be disabled.
• Linked enclosure login will not work if the linked enclosure in CAC mode.
• If accurate Service account details are not provided, LDAP user login with certificate will fail.
• It is highly recommended to establish a recovery plan before getting started with CAC.  If something goes wrong with the OA configuration, the OA may be recovered through the serial port or Insight Display panel and USB KEY. Both methods require physical access to the OA.  However, if an LCD PIN has been configured (and forgotten) and local accounts have been disabled or CAC has been incorrectly configured then, the only way to recover is through a serial port. The two most common situations where OA recovery is needed are when LDAP has been configured incorrectly with local accounts disabled or when CAC has been configured without certificate access.

• Addressed an issue where SNMP traps were not being sent after an OA reboot or Active to Standby role change occurred when the IPSWAP feature was enabled and the SNMP trap receivers are configured using their hostname.
• Limited the “ADD SNMP TRAPRECEIVE” CLI command username length to 32 characters.
• Addressed an issue which enables OA to disallow the configuration of SNMP Trap receiver names that start with a numeral.
• Addressed an issue in OA CLI command “SHOW SERVER INFO” as well as in the GUI where it did not display new processor family for Integrity i6 blade.
• Addressed an issue where OA used to send inappropriate power request denial code when there was an ekeying error, which in turn made iLO report inappropriate errorcode.
• Addressed an issue where in after OA fail over, the SNMP failover trap was not getting generated if the OA configuration contains IPV6 only network and Enclosure IP mode enabled.
• Addressed an ARP flux issue in Standby OA’s IP address, due to which Network switches connected to OA may observe potential duplicate IP address of standby OA.
• Addressed an issue in the “RESET ILO” CLI command which prevented OA from resetting Integrity Blades.

Security

     The following security issues were fixed:

• CVE-2016-2177 -Addressed a vulnerability against openSSL was incorrectly using pointer arithmetic for heap-buffer boundary checks which might allow remote attackers to cause a denial of service.
• CVE-2016-6302 - Addressed a vulnerability against openSSL integer underflow flaw leading to a buffer over-read which allows remote attackers to cause a denial of service.
• CVE-2016-6304 Addressed a vulnerability against openSSL which could lead to Denial Of Service attack through memory exhaustion which might allow remote attackers to cause a denial of service (memory consumption) via large OCSP Status Request extensions.
• CVE-2016-6306- Addressed a vulnerability against openSSL where in  some missing message length checks can result in OOB reads of up to 2 bytes beyond an allocated buffer might allow remote attackers to cause a denial of service (out-of-bounds read) via crafted certificate operations.

Issues and workarounds

Browsers

• OA GUI is not accessible in Chrome versions 43.0.2357.10 to 44.0.2383.The issue was caused by a “regression” in Chrome (or WebKit). Customers should use an alternative browser like Firefox or Internet Explorer or try a different version of Chrome.
• SSO-to-iLO connection from the OA using an iLO host name fails with Microsoft Internet Explorer11 on Windows 8.On a Windows 8 system with Internet Explorer 10 or Internet Explorer 11, if the OA web GUI session is loaded using a host name instead of an IP address, an attempt to open an iLO window using SSO from the OA web GUI might result in the iLO page loading in the OA web GUI window instead of the intended new window. This issue was determined to be a bug in Internet Explorer and is expected to be fixed in a future release or update for Internet Explorer. To work around this issue, either use an IP address to load the OA Web GUI, or turn off Protected Mode for the appropriate zone in Internet Explorer’s settings. This issue occurs only on Internet Explorer browsers.

FIPS

Certificates smaller than 2048 bits in size are not compliant with FIPS requirements as enforced by the OA firmware starting with OA 4.20. When the OA running OA firmware version 4.40 or greater is operating in FIPS Mode ON/DEBUG and is configured with a 1024-bit LDAP certificate that was installed when running a previous version of OA firmware, FIPS Mode ON/DEBUG is considered to be operating in a degraded state due to the presence of the non-compliant certificate. While operating in this FIPS-Degraded Mode operational state, attempts to set FIPS Mode OFF from the OA GUI Network Access>FIPS tab will fail and show the error message The selected FIPS mode is already enabled. When the non-compliant certificate is removed, the FIPS-Degraded operational status is cleared, FIPS Mode can then be successfully set to OFF from the GUI interface. Note that the OA CLI command SET FIPS MODE OFF can be successfully used to set FIPS Mode OFF even with non-compliant 1024-bit LDAP certificates installed in the OA.

IRC

Unable to open .net IRC console for Gen10 Blades, Gen9 Blades also have the same issue. The Java applet and Webstart however, loads but the virtual media mounting fails. The work around is to launch the IRC through IRC Application (HP Lights-Out  Stand Alone Remote Console) which is installed on terminal client.

EFM

To use EFM on Gen 10 Blades, please select options/filters “Make Bootable ISO file” and “Enclosure Firmware Management” while creating custom SPP ISO on HPSUM 8.0.0. Please refer to HPSUM 8.0.0 User guide for further details.

CAC

• In the CAC mode SSH, Telnet and XML Reply protocols will be disabled.
• Linked enclosure login will not work if the linked enclosure in CAC mode.
• If accurate Service account details are not provided, LDAP user login with certificate will fail.
• It is highly recommended to establish a recovery plan before getting started with CAC.  If something goes wrong with the OA configuration, the OA may be recovered through the serial port or Insight Display panel and USB KEY. Both methods require physical access to the OA.  However, if an LCD PIN has been configured (and forgotten) and local accounts have been disabled or CAC has been incorrectly configured then, the only way to recover is through a serial port. The two most common situations where OA recovery is needed are when LDAP has been configured incorrectly with local accounts disabled or when CAC has been configured without certificate access.

• Addressed an issue where OA "update iLO all" command fails in an enclosure with maximum Blades.
• Addressed an issue where a Warning Alert  was wrongly sent when a fan is reseated in an enclosure
• Addressed an issue where Remote Syslog logging would fail when OA failover happened in an IPv6 only environment.
• Enhanced OA to bring the server from a power throttled state back to normal power state upon an OA reboot to circumvent an unwarranted emergency brake.
• Fixed an issue where the Active and Standby OAs can have the same IP address in some rare situations.
• Resolved an issue where a Gen9 server’s host name gets cleared when the blade is rebooted.
• Addressed an issue where server blade Power ON will be delayed in enclosures with OA Firmware Version 4.60 and managed by HPE OneView, when the OA module is reset until OneView refreshes the servers.

Security

The following security vulnerabilities were fixed:

• CVE-2016-5387– Addressed a vulnerability which might allow remote attackers to redirect an application's outbound HTTP traffic to an arbitrary proxy server via a crafted Proxy header in an HTTP request.
• CVE-2016-2183- Addressed a vulnerability against TLS ciphers with 64bit block size in which makes it easier for remote attackers to obtain cleartext data via an attack against a long-duration encrypted session
• CVE-2016-6515 - Addressed a vulnerability in OpenSSH which did not limit password lengths for password authentication, which allows remote attackers to cause a denial of service via a long string.
• CVE-2015-8215 - Addressed a vulnerability IPv6 stack which does not validate attempted changes to the MTU value, which allows context-dependent attackers to cause a denial of service.
• Addressed issue where in Onboard Administrator was vulnerable to Buffer overflow.
• Added the HSTS[HTTP strict transport security] support  in OA.
• Addressed a memory corruption vulnerability in the post-authentication sshd process.

Issues and workarounds

Browsers

• OA GUI is not accessible in Chrome versions 43.0.2357.10 to 44.0.2383.The issue was caused by a “regression” in Chrome (or WebKit). Customers should use an alternative browser like Firefox or Internet Explorer or try a different version of Chrome.
• SSO-to-iLO connection from the OA using an iLO host name fails with Microsoft Internet Explorer11 on Windows 8.On a Windows 8 system with Internet Explorer 10 or Internet Explorer 11, if the OA web GUI session is loaded using a host name instead of an IP address, an attempt to open an iLO window using SSO from the OA web GUI might result in the iLO page loading in the OA web GUI window instead of the intended new window. This issue was determined to be a bug in Internet Explorer and is expected to be fixed in a future release or update for Internet Explorer. To work around this issue, either use an IP address to load the OA Web GUI, or turn off Protected Mode for the appropriate zone in Internet Explorer’s settings. This issue occurs only on Internet Explorer browsers.

FIPS

Certificates smaller than 2048 bits in size are not compliant with FIPS requirements as enforced by the OA firmware starting with OA 4.20. When the OA running OA firmware version 4.40 or greater is operating in FIPS Mode ON/DEBUG and is configured with a 1024-bit LDAP certificate that was installed when running a previous version of OA firmware, FIPS Mode ON/DEBUG is considered to be operating in a degraded state due to the presence of the non-compliant certificate. While operating in this FIPS-Degraded Mode operational state, attempts to set FIPS Mode OFF from the OA GUI Network Access>FIPS tab will fail and show the error message The selected FIPS mode is already enabled. When the non-compliant certificate is removed, the FIPS-Degraded operational status is cleared, FIPS Mode can then be successfully set to OFF from the GUI interface. Note that the OA CLI command SET FIPS MODE OFF can be successfully used to set FIPS Mode OFF even with non-compliant 1024-bit LDAP certificates installed in the OA.

IRC

Unable to open .net IRC console for Gen10 Blades, Gen9 Blades also have the same issue. The Java applet and Webstart however, loads but the virtual media mounting fails. The work around is to launch the IRC through IRC Application (HP Lights-Out Stand Alone Remote Console) which is installed on terminal client.

EFM

To use EFM on Gen 10 Blades, please select options/filters “Make Bootable ISO file” and “Enclosure Firmware Management” while creating custom SPP ISO on HPSUM 8.0.0. Please refer to HPSUM 8.0.0 User guide for further details.

• Addressed an issue where OA "update iLO all" command fails in an enclosure with maximum Blades.
• Addressed an issue where a Warning Alert  was wrongly sent when a fan is reseated in an enclosure
• Addressed an issue where Remote Syslog logging would fail when OA failover happened in an IPv6 only environment.
• Enhanced OA to bring the server from a power throttled state back to normal power state upon an OA reboot to circumvent an unwarranted emergency brake.
• Fixed an issue where the Active and Standby OAs can have the same IP address in some rare situations.
• Resolved an issue where a Gen9 server’s host name gets cleared when the blade is rebooted.
• Addressed an issue where server blade Power ON will be delayed in enclosures with OA Firmware Version 4.60 and managed by HPE OneView, when the OA module is reset until OneView refreshes the servers.

Security

The following security vulnerabilities were fixed:

• CVE-2016-5387– Addressed a vulnerability which might allow remote attackers to redirect an application's outbound HTTP traffic to an arbitrary proxy server via a crafted Proxy header in an HTTP request.
• CVE-2016-2183- Addressed a vulnerability against TLS ciphers with 64bit block size in which makes it easier for remote attackers to obtain cleartext data via an attack against a long-duration encrypted session
• CVE-2016-6515 - Addressed a vulnerability in OpenSSH which did not limit password lengths for password authentication, which allows remote attackers to cause a denial of service via a long string.
• CVE-2015-8215 - Addressed a vulnerability IPv6 stack which does not validate attempted changes to the MTU value, which allows context-dependent attackers to cause a denial of service.
• Addressed issue where in Onboard Administrator was vulnerable to Buffer overflow.
• Added the HSTS[HTTP strict transport security] support  in OA.
• Addressed a memory corruption vulnerability in the post-authentication sshd process.

Issues and workarounds

Browsers

• OA GUI is not accessible in Chrome versions 43.0.2357.10 to 44.0.2383.The issue was caused by a “regression” in Chrome (or WebKit). Customers should use an alternative browser like Firefox or Internet Explorer or try a different version of Chrome.
• SSO-to-iLO connection from the OA using an iLO host name fails with Microsoft Internet Explorer11 on Windows 8.On a Windows 8 system with Internet Explorer 10 or Internet Explorer 11, if the OA web GUI session is loaded using a host name instead of an IP address, an attempt to open an iLO window using SSO from the OA web GUI might result in the iLO page loading in the OA web GUI window instead of the intended new window. This issue was determined to be a bug in Internet Explorer and is expected to be fixed in a future release or update for Internet Explorer. To work around this issue, either use an IP address to load the OA Web GUI, or turn off Protected Mode for the appropriate zone in Internet Explorer’s settings. This issue occurs only on Internet Explorer browsers.

FIPS

Certificates smaller than 2048 bits in size are not compliant with FIPS requirements as enforced by the OA firmware starting with OA 4.20. When the OA running OA firmware version 4.40 or greater is operating in FIPS Mode ON/DEBUG and is configured with a 1024-bit LDAP certificate that was installed when running a previous version of OA firmware, FIPS Mode ON/DEBUG is considered to be operating in a degraded state due to the presence of the non-compliant certificate. While operating in this FIPS-Degraded Mode operational state, attempts to set FIPS Mode OFF from the OA GUI Network Access>FIPS tab will fail and show the error message The selected FIPS mode is already enabled. When the non-compliant certificate is removed, the FIPS-Degraded operational status is cleared, FIPS Mode can then be successfully set to OFF from the GUI interface. Note that the OA CLI command SET FIPS MODE OFF can be successfully used to set FIPS Mode OFF even with non-compliant 1024-bit LDAP certificates installed in the OA.

IRC

Unable to open .net IRC console for Gen10 Blades, Gen9 Blades also have the same issue. The Java applet and Webstart however, loads but the virtual media mounting fails. The work around is to launch the IRC through IRC Application (HP Lights-Out Stand Alone Remote Console) which is installed on terminal client.

EFM

To use EFM on Gen 10 Blades, please select options/filters “Make Bootable ISO file” and “Enclosure Firmware Management” while creating custom SPP ISO on HPSUM 8.0.0. Please refer to HPSUM 8.0.0 User guide for further details.

• Firmware resolves a data mismanagement issue associated with unaligned sequential write operations.

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

Problems Fixed for HPG3 (D):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• Firmware resolves a data mismanagement issue associated with unaligned sequential write operations.

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

Problems Fixed for HPG3 (D):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• Firmware resolves a data mismanagement issue associated with unaligned sequential write operations.

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• Firmware resolves a data mismanagement issue associated with unaligned sequential write operations.

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• DDR cache could be randomly disabled after several boots
• The controller could become unresponsive while in HBA mode during the creation of a partition/file system

.

• DDR cache could be randomly disabled after several boots
• The controller could become unresponsive while in HBA mode during the creation of a partition/file system

.

• In a multi-path environment with nested expanders, the active path to a device might be incorrect following a path failure or restoration.

• In a multi-path environment with nested expanders, the active path to a device might be incorrect following a path failure or restoration.

• Improved handling of IPV6 addresses

• Updated for RHEL 7.4 support

• Supports SLES12 SP3

• Improved handling of SATA drive temperatures

• Improved handling of IPV6 addresses

• Updated for RHEL 7.4 support

• Supports SLES12 SP3

• Improved handling of SATA drive temperatures

• This firmware eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written.

• This firmware eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written.

• Normalized duplicate timeout field contexts.
• Duplicate timeout fields used in wrong contexts.
• ql_dump_buffer to avoid emitting on multiple lines.
• Allowed Simple Name Server (SNS) fabric logins to be retried.
• Callback for memory cleanup
• Relogin from being triggered too fast.

• Normalized duplicate timeout field contexts.
• Duplicate timeout fields used in wrong contexts.
• ql_dump_buffer to avoid emitting on multiple lines.
• Allowed Simple Name Server (SNS) fabric logins to be retried.
• Callback for memory cleanup
• Relogin from being triggered too fast.

• Increase the driver login retry count to 30.

• Mark Data Integrity Field (DIF) errors from target as re-tryable errors.

• Reduced excessive debug print during 27xx fwdump.

• Avoid crashing on null cred pointer in exit_creds().

• Fixed mbx pointer error in classic fwdump.

• Added check for different type of Register State Change Notification (RSCN).

• Fixed improper fcport_count accounting.

• Corrected supported Fibre Channel (FC) speed for sysfs.

• Corrected supported Fibre Channel (FC) speed for Fabric Device Management Interface (FDMI) / Remote Desktop Protocol (RDP).

• Increase the driver login retry count to 30.

• Mark Data Integrity Field (DIF) errors from target as re-tryable errors.

• Reduced excessive debug print during 27xx fwdump.

• Avoid crashing on null cred pointer in exit_creds().

• Fixed mbx pointer error in classic fwdump.

• Added check for different type of Register State Change Notification (RSCN).

• Fixed improper fcport_count accounting.

• Corrected supported Fibre Channel (FC) speed for sysfs.

• Corrected supported Fibre Channel (FC) speed for Fabric Device Management Interface (FDMI) / Remote Desktop Protocol (RDP).

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• The PortRcvPkts counter was prevented from being cleared after resetting it.
• Fixed an issue which caused a system Time Out on the configuration cycle of the VFs when more than 10 Virtual Functions performed FLR and the completion Time Out value was configured to a range of less than
  16 msec.
• Fixed an issue that caused the server to hang and result in NMI when running “mlxfwtop –d mt4103_pci_cr0” while restarting the driver in parallel (from a different thread). In this case, the downstream bridge over the device reported completion timeout error.
• Fixed an issue in flow_steering where BMC could not receive a ping over IPV6 after running bmc_reboot.
• Fixed an issue while closing the HCA, where RX packet caused bad access to resources that did not exist, and consequently caused the QPCGW or the irisc to get stuck.
• Fixed an issue where the masterSM LID and the LID was either 0 or 0xFFFF when the port was neither active nor armed.
• Fixed an issue that prevented ibdump from capturing all MADs packets.
• Fixed an issue that prevented the link to go up after reboot.
• Fixed a rare issue that cause the PCIe configuration cycle that arrived during the time of sw_reset to generate 2 completions.
• Fixed an issue that caused NC-SI not to work when adding the disable_static_steering_ini field in the ini file, due to memory allocation issue for this field in the scratchpad.

Fixes submitted in version 14.21.2800:

• Fixed an issue that caused QP connection timeout due to firmware not being able to handle duplicate packets with AckReq bit set. The fix stopped ignoring duplicate AckReq packets to avoid timeout on the sender side.
• MPFS load balance (DUP_MAC_ACTION==LOAD_BALANCE(1)) is not working as long as IB_ROUTING_MODE/SRIOV_IB_ROUTING_MODE is configured to LID.
• Fixed an issue that caused a SX engine deadlock (the SX engine handles software port/priority changes for a specific Send Queue) when more than a single SX engine handled the "prio diff" flow simultaneously, thus caused the hardware to get stuck.
  The issue happened as the firmware releases the SX engine and waits for various operation to complete. However, due to a race that allows a different transaction to get into the SX engine and cause the lock to be taken by it, the SX engine release is prevented.
• Fixed an issue that caused the system to hang while changing QPTS/QPDPM/QPDP parameters during traffic.
• Fixed unfairness between senders in RoCE LAG while ECN is configured.
• Fixed an issue which displayed diagnostic counters only on the adapter that was initialized first, which is the counters’ owner. The owner received correct values, while the other adapter only received zeros.
• Fixed an issue that caused the rdma_cm traffic to fail on the 2nd port when more than 32 VFs were configured, when the RoCE Dual Port vHCA (a.k.a Multi-Port vHCA: MPV) feature was enabled.

• The PortRcvPkts counter was prevented from being cleared after resetting it.
• Fixed an issue which caused a system Time Out on the configuration cycle of the VFs when more than 10 Virtual Functions performed FLR and the completion Time Out value was configured to a range of less than
  16 msec.
• Fixed an issue that caused the server to hang and result in NMI when running “mlxfwtop –d mt4103_pci_cr0” while restarting the driver in parallel (from a different thread). In this case, the downstream bridge over the device reported completion timeout error.
• Fixed an issue in flow_steering where BMC could not receive a ping over IPV6 after running bmc_reboot.
• Fixed an issue while closing the HCA, where RX packet caused bad access to resources that did not exist, and consequently caused the QPCGW or the irisc to get stuck.
• Fixed an issue where the masterSM LID and the LID was either 0 or 0xFFFF when the port was neither active nor armed.
• Fixed an issue that prevented ibdump from capturing all MADs packets.
• Fixed an issue that prevented the link to go up after reboot.
• Fixed a rare issue that cause the PCIe configuration cycle that arrived during the time of sw_reset to generate 2 completions.
• Fixed an issue that caused NC-SI not to work when adding the disable_static_steering_ini field in the ini file, due to memory allocation issue for this field in the scratchpad.

Fixes submitted in version 14.21.2800:

• Fixed an issue that caused QP connection timeout due to firmware not being able to handle duplicate packets with AckReq bit set. The fix stopped ignoring duplicate AckReq packets to avoid timeout on the sender side.
• MPFS load balance (DUP_MAC_ACTION==LOAD_BALANCE(1)) is not working as long as IB_ROUTING_MODE/SRIOV_IB_ROUTING_MODE is configured to LID.
• Fixed an issue that caused a SX engine deadlock (the SX engine handles software port/priority changes for a specific Send Queue) when more than a single SX engine handled the "prio diff" flow simultaneously, thus caused the hardware to get stuck.
  The issue happened as the firmware releases the SX engine and waits for various operation to complete. However, due to a race that allows a different transaction to get into the SX engine and cause the lock to be taken by it, the SX engine release is prevented.
• Fixed an issue that caused the system to hang while changing QPTS/QPDPM/QPDP parameters during traffic.
• Fixed unfairness between senders in RoCE LAG while ECN is configured.
• Fixed an issue which displayed diagnostic counters only on the adapter that was initialized first, which is the counters’ owner. The owner received correct values, while the other adapter only received zeros.
• Fixed an issue that caused the rdma_cm traffic to fail on the 2nd port when more than 32 VFs were configured, when the RoCE Dual Port vHCA (a.k.a Multi-Port vHCA: MPV) feature was enabled.

• Fixed a race between the firmware and the hardware during driver start which blocked outbound completions.
• Fixed an issue which caused the firmware not to send link_down event to the driver when running the close_port command.
• Fixed an issue where in rare cases the Auto Sense failed to detect the right protocol.
• Fixed signal integrity issue when connecting a WCS ConnectX4 mezz card to Pikes peak FPGA..
• Added the option to transmit corrupted DME pages for a very short period of time at the beginning of the Auto-Negotiation flow.
• Fixed an incorrect report of the PortRcvDataVLExtended / PortXmitDataVLExtended counters by the firmware.
• Fixed a rare issue which caused firmware’s packet injector to cut off packets when the TX was congested.
• Fixed an issue that caused the response to TX requests to take up to 10 mili-seconds in IEEE clause 72 Link Training.
• Fixed a race between 2 iriscs which caused a QP to get stuck in burst control limit state.
• When a QP was in error state, the firmware generated too many err CQEs at once, thus causing the cmdif responsiveness to be too slow. To prevent the above, the number of err CQEs was limited to 16 at a time.
• Fixed an issue that caused the MAC address that was set from the OS using ifconfig to be not reflected in the OCBB buffer.
• Fixed an issue where the ibdump got broken when running with loopback traffic.
• Fixed an issue where the firmware took QP to firmware ownership and then released it to the hardware ownership without checking if another firmware flow owns the same QP.
• Fixed an issue which occurred after disconnecting cable which showed the link type as IB even if the link type of the port is ETH.
• Fixed an issue related to the HCA PoerXmitWait counter on port 2 (connected to port 2 on Switch-IB) where it started counting and reached 0xFF's regardless of connection to switch.
• Fixed a completion error issue when ECN was enabled. The ECN usage caused ordering errors in completion queues (CQ).
• Fixed the length calculation of UDP. The incorrect UDP length in the CNP packet caused miss-calculation of the ICRC.
• Fixed a wrong returned status in cable info MAD when the cable was not connected
• Fixed failure instances when initiating FLR in the Physical Function.
• Disabled High Rate Steering mode in the INI to enable its compatibility with NC-SI over VLAN.
• Fixed a default hardware configuration issue which caused RDP over IPv4 traffic to be dropped.
• MLNX_OEM command GET_TEMP returned a wrong value in the max_temp field.
• Fixed an issue which caused TX traffic to stop when the message MTU size was larger than QP.mtu.
• Fixed an issue which caused NVCONFIG to fail when the number of sector was set to 1 and the sector was zeroed.
• Fixed a race in handling a duplicated “read request from middle”.
• Fixed an issue which caused lack of IB traffic on SR-IOV VPI.
• Fixed an issue which caused NVRAM to get stuck when it filled non-valid information in TLV.
• Fixed an issue which caused an internal firmware error when APM changed the QPs port mapping. 
• Fixed an issue which caused a firmware internal error when handling QP alternative context. 
• Fixed an issue which caused packet transmission to get stuck when the software tried to send pause frames with dmac equal to one of the device's MAC addresses.
• Fixed a wrong reporting of section 5 event 8 - LSO support.
• Fixed a mistakenly dropped ETH packet with ethertype 0x600 by the NIC.
• Fixed a case preventing broadcast traffic from arriving to their destination after detaching high priority broadcast rule on a port where NC-SI was enabled.
• Fixed an issue where the port raised as SDR vs. InfiniScale IV QDR Switch
• Fixed a rare case of completion Error with Bad Opcode sequence status which occurred when retransmitting read requests
• Fixed a case where the actual bandwidth did not match the user settings in VM QoS.
• Fixed a case where on rare cases, communication to BMC was lost during driver initialization.
• Fixed an issue with cable reading, which caused the link not to raise.
• Set the maximum EQN number to 1024.
• Fixed a rare issue with VPD init flow which caused read failures.
• Fixed an issue with RX size counter not being reported.
• Fixed promiscuous mode compatibility with A0-DMFS steering.
• Fixed promiscuous mode compatibility when NC-SI is enabled and configured.
• Fixed sending/receiving OEM temp commands (set/get) with channel ID 0x1f failure.
• Fixed an issue which caused packets to drop on a port when changing the interface state of the other port.
• Fixed long management communication loss and SOL hang during reboot cycles.
• Fixed wrong processing of inbound traffic towards BMC which caused communication loss.
• Fixed management link loss upon closing port interface through the driver.
• Fixed a false indication in firmware of an expander presence causing delay in EEPROM reading.
• Fixed an issue which caused a link down on a port when the cable was removed from the other port.
• Fixed a rare case where packet with length zero got stuck in hardware queues.
• Fixed an issue which caused InfiniBand congestion control packet (CNP) to hang in hardware.
• Fixed an issue which caused AEN to be sent after channel reset.
• Fixed an issue which prevented the restoring of QoS setting to its default consequently causing bandwidth degradation.
• Fixed an occasional long link up time with 10GbE based devices.
• Fixed an issue preventing cable readings from i2c slave address 0x51
• Fixed a wrong parity bit calculation when transmitting PCIe TS1 packets.
• Fixed a possible deadlock in PM turnoff request transmission and ack acceptance flow.
• Fixed a rare case with alignments state machines which caused occasional width degradation.
• Fixed an issue where the transmit queues hanged while congestion control was enabled and operational (EQC/QCN)
• Fixed an unexpected work completion syndrome with vendor syndrome 0x77 received when running RDMA SEN/WRITE traffic with retransmissions.
• Fixed an issue which caused SetPortInfo to return a good status when receiving invalid LinkSpeedEnabled value.
• Fixed an issue which caused dual port SFPP module cards to be automatically mapped with expander.
• Fixed an issue where firmware overrides the steering mode that was chosen by the driver.
• Fixed invalid return sensing results occurred when the link was up.
• Fixed an issue causing the sensing result to be delayed when cable was unplugged.
• Fixed an issue causing the link type to be displayed as ETH when set to AUTO.
• Fixed an issue causing ARP not to reply when connected to Hyper-V vSwitch.

Fixes in 14.18.2030:

• On rare occasions during UEFI boot cycles system got stuck while WinPE is loaded. (OS WinPE, system DL160).
• Single FTE that catches both untagged and prio-tagged packets (by giving an FTE with match_value.vlan_tag = 0 and match_value.vid = 0) is currently not supported.
• Flashing the firmware requires server reboot. Firmware cannot be flashed twice without server reboot after first flashing
• If the vport state is DOWN and a packet is sent in local loopback, the sx_sniffer tool will not function.
• Fixed an issue causing bubbles to appear as symbol errors when link raised FDR 1x.
• When Clause 74 Fire-Code FEC is active, and there are FC corrected errors, both the FC_correctable counter and the FC_uncoreectable counter are increment.
• Some Port Control Register do not return to the default value after the last port owner host restarts the driver.
• Fixed an issue which caused RX to hang when the UDP packet had a reserved UDP destination port.
• Fixed DMAC reporting mapping per host.
• Fixed an EEH error from PCI which caused firmware to hang.
• Fixed the default value of the PCIe target_link_speed to Gen3 in link control2.
• Fixed an issue which prevented LEDs from blinking when the traffic was less than 0.1% of the link speed.
• Fixed an issue which caused the mlxconfig configuration of VF_LOG_BAR_SIZE to be ignored and to be set to 5 (32MB).
• A server getting into a Standby mode while Packet-Pacing is enabled might cause firmware to hang and driver call-trace.
• Fixed an issue which caused unexpected QoS functionality in case of multiple sources to single destination traffic transmission.
• Fixed an issue which occasionally caused the RX traffic to hang in DC when received a PCI error on WQE fetch.
• Fixed OOB connection issue during Intel's ITP inject errors test.
• Fixed an issue which prevented MAC address changes by to driver to be reflected in the OBCC and NC-SI interfaces.
• Added protection from IOPX thermal diode destabilization to prevent UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GE cards.
• Fixed an issue which caused a link down in Port 2 when unplugging the cable from Port 1.
• In some cases, a Bit Error Rate is not optimal on 10G/40G links.
• Instability of Link Training Flow occurs during 100G Auto-Negotiation.
• Fixed a rare issue which caused the command to hang when moved the QP to RESET and back to RTS.
• Improved RDMA READ bandwidth under packet lost scenario.
• Added support for pnat = 1 in HCA access_reg command as required by the ibdiagnet tool.
• Fixed the LLDP OCBB response: return value is now ascii.
• Fixed a very rare NMI issue during PXE cycles.
• Increased the steering hash tables static size from 128 to a maximum of 32K entries.
• Prevented miscalculation of module temperature when using 100Gb/s cables (OPN: MFA1A00-Cxxx for 100GbE).
• Reduced one hop for Unicast RX steering, steering pipes balancing.
• Non-volatile configuration of Port Type TLV more than 50 times might cause system hang.
• Enabled RoCE IPv4 Multicast. This prevents MCG command from failing when an IPv4 is mapped to an IPv6 address.
• If the PF driver or the tool (e.g. ethtool) use PAOS DOWN command (e.g. by ifconfig down or ip link set down), loopback traffic is blocked for all functions on this port (PF<->VFs / VF<->VF) In Multihost loopback, the traffic will be blocked once the firmware receives the PAOS down command from all PFs. However, the loopback traffic will not be blocked when the port is down  due to the physical link (for example: cable plugged out, switch port down).
• Fixed a 25G and 50G link issue when Clause 91 RS FEC was active.
• Added a missing invalidation of eSwitch cache upon FLR which caused the upcoming driver load to either fail or not to be able to transmit.
• Fixed a UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GbE OCP card.
• Fixed an issue which prevented Vport counters from counting local loopback packets. Packets now are filter by the self-loopback prevention.
• Reported INTx as unsupported to allow PFs Passthrough on PowerKVM.
• SR-IOV Ethernet supports up to 18 VFs per port only.
• Fixed and incident what allowed local (internal) loopbacked packets to be counted by the Vport counters, although Vport counters should count only traffic that crosses the Vport.
• Fixed an issue preventing driver loading or TX traffic sending upon reboot, after ungraceful driver unload.
• Fixed casting of BMC MAC before steering API.
• Fixed the PCI write flow to take into consideration the PCI MTU. This fix eliminates the need for NOPs in the flow, which resulted from PPC larger PCI MTU.
  The single queue limitation for READ is due to a hardware limitation of the number of READ request in a given time.
• Fixed a case that caused FlexBoot to not work as expected with systems that run with "large bar" enabled (Above 4G Decoding) over Connect-IB or ConnectX-4 HCAs.
• Fixed an issue which prevented link creation when connected to IXIA 25G.

• Fixed a race between the firmware and the hardware during driver start which blocked outbound completions.
• Fixed an issue which caused the firmware not to send link_down event to the driver when running the close_port command.
• Fixed an issue where in rare cases the Auto Sense failed to detect the right protocol.
• Fixed signal integrity issue when connecting a WCS ConnectX4 mezz card to Pikes peak FPGA..
• Added the option to transmit corrupted DME pages for a very short period of time at the beginning of the Auto-Negotiation flow.
• Fixed an incorrect report of the PortRcvDataVLExtended / PortXmitDataVLExtended counters by the firmware.
• Fixed a rare issue which caused firmware’s packet injector to cut off packets when the TX was congested.
• Fixed an issue that caused the response to TX requests to take up to 10 mili-seconds in IEEE clause 72 Link Training.
• Fixed a race between 2 iriscs which caused a QP to get stuck in burst control limit state.
• When a QP was in error state, the firmware generated too many err CQEs at once, thus causing the cmdif responsiveness to be too slow. To prevent the above, the number of err CQEs was limited to 16 at a time.
• Fixed an issue that caused the MAC address that was set from the OS using ifconfig to be not reflected in the OCBB buffer.
• Fixed an issue where the ibdump got broken when running with loopback traffic.
• Fixed an issue where the firmware took QP to firmware ownership and then released it to the hardware ownership without checking if another firmware flow owns the same QP.
• Fixed an issue which occurred after disconnecting cable which showed the link type as IB even if the link type of the port is ETH.
• Fixed an issue related to the HCA PoerXmitWait counter on port 2 (connected to port 2 on Switch-IB) where it started counting and reached 0xFF's regardless of connection to switch.
• Fixed a completion error issue when ECN was enabled. The ECN usage caused ordering errors in completion queues (CQ).
• Fixed the length calculation of UDP. The incorrect UDP length in the CNP packet caused miss-calculation of the ICRC.
• Fixed a wrong returned status in cable info MAD when the cable was not connected
• Fixed failure instances when initiating FLR in the Physical Function.
• Disabled High Rate Steering mode in the INI to enable its compatibility with NC-SI over VLAN.
• Fixed a default hardware configuration issue which caused RDP over IPv4 traffic to be dropped.
• MLNX_OEM command GET_TEMP returned a wrong value in the max_temp field.
• Fixed an issue which caused TX traffic to stop when the message MTU size was larger than QP.mtu.
• Fixed an issue which caused NVCONFIG to fail when the number of sector was set to 1 and the sector was zeroed.
• Fixed a race in handling a duplicated “read request from middle”.
• Fixed an issue which caused lack of IB traffic on SR-IOV VPI.
• Fixed an issue which caused NVRAM to get stuck when it filled non-valid information in TLV.
• Fixed an issue which caused an internal firmware error when APM changed the QPs port mapping. 
• Fixed an issue which caused a firmware internal error when handling QP alternative context. 
• Fixed an issue which caused packet transmission to get stuck when the software tried to send pause frames with dmac equal to one of the device's MAC addresses.
• Fixed a wrong reporting of section 5 event 8 - LSO support.
• Fixed a mistakenly dropped ETH packet with ethertype 0x600 by the NIC.
• Fixed a case preventing broadcast traffic from arriving to their destination after detaching high priority broadcast rule on a port where NC-SI was enabled.
• Fixed an issue where the port raised as SDR vs. InfiniScale IV QDR Switch
• Fixed a rare case of completion Error with Bad Opcode sequence status which occurred when retransmitting read requests
• Fixed a case where the actual bandwidth did not match the user settings in VM QoS.
• Fixed a case where on rare cases, communication to BMC was lost during driver initialization.
• Fixed an issue with cable reading, which caused the link not to raise.
• Set the maximum EQN number to 1024.
• Fixed a rare issue with VPD init flow which caused read failures.
• Fixed an issue with RX size counter not being reported.
• Fixed promiscuous mode compatibility with A0-DMFS steering.
• Fixed promiscuous mode compatibility when NC-SI is enabled and configured.
• Fixed sending/receiving OEM temp commands (set/get) with channel ID 0x1f failure.
• Fixed an issue which caused packets to drop on a port when changing the interface state of the other port.
• Fixed long management communication loss and SOL hang during reboot cycles.
• Fixed wrong processing of inbound traffic towards BMC which caused communication loss.
• Fixed management link loss upon closing port interface through the driver.
• Fixed a false indication in firmware of an expander presence causing delay in EEPROM reading.
• Fixed an issue which caused a link down on a port when the cable was removed from the other port.
• Fixed a rare case where packet with length zero got stuck in hardware queues.
• Fixed an issue which caused InfiniBand congestion control packet (CNP) to hang in hardware.
• Fixed an issue which caused AEN to be sent after channel reset.
• Fixed an issue which prevented the restoring of QoS setting to its default consequently causing bandwidth degradation.
• Fixed an occasional long link up time with 10GbE based devices.
• Fixed an issue preventing cable readings from i2c slave address 0x51
• Fixed a wrong parity bit calculation when transmitting PCIe TS1 packets.
• Fixed a possible deadlock in PM turnoff request transmission and ack acceptance flow.
• Fixed a rare case with alignments state machines which caused occasional width degradation.
• Fixed an issue where the transmit queues hanged while congestion control was enabled and operational (EQC/QCN)
• Fixed an unexpected work completion syndrome with vendor syndrome 0x77 received when running RDMA SEN/WRITE traffic with retransmissions.
• Fixed an issue which caused SetPortInfo to return a good status when receiving invalid LinkSpeedEnabled value.
• Fixed an issue which caused dual port SFPP module cards to be automatically mapped with expander.
• Fixed an issue where firmware overrides the steering mode that was chosen by the driver.
• Fixed invalid return sensing results occurred when the link was up.
• Fixed an issue causing the sensing result to be delayed when cable was unplugged.
• Fixed an issue causing the link type to be displayed as ETH when set to AUTO.
• Fixed an issue causing ARP not to reply when connected to Hyper-V vSwitch.

Fixes in 14.18.2030:

• On rare occasions during UEFI boot cycles system got stuck while WinPE is loaded. (OS WinPE, system DL160).
• Single FTE that catches both untagged and prio-tagged packets (by giving an FTE with match_value.vlan_tag = 0 and match_value.vid = 0) is currently not supported.
• Flashing the firmware requires server reboot. Firmware cannot be flashed twice without server reboot after first flashing
• If the vport state is DOWN and a packet is sent in local loopback, the sx_sniffer tool will not function.
• Fixed an issue causing bubbles to appear as symbol errors when link raised FDR 1x.
• When Clause 74 Fire-Code FEC is active, and there are FC corrected errors, both the FC_correctable counter and the FC_uncoreectable counter are increment.
• Some Port Control Register do not return to the default value after the last port owner host restarts the driver.
• Fixed an issue which caused RX to hang when the UDP packet had a reserved UDP destination port.
• Fixed DMAC reporting mapping per host.
• Fixed an EEH error from PCI which caused firmware to hang.
• Fixed the default value of the PCIe target_link_speed to Gen3 in link control2.
• Fixed an issue which prevented LEDs from blinking when the traffic was less than 0.1% of the link speed.
• Fixed an issue which caused the mlxconfig configuration of VF_LOG_BAR_SIZE to be ignored and to be set to 5 (32MB).
• A server getting into a Standby mode while Packet-Pacing is enabled might cause firmware to hang and driver call-trace.
• Fixed an issue which caused unexpected QoS functionality in case of multiple sources to single destination traffic transmission.
• Fixed an issue which occasionally caused the RX traffic to hang in DC when received a PCI error on WQE fetch.
• Fixed OOB connection issue during Intel's ITP inject errors test.
• Fixed an issue which prevented MAC address changes by to driver to be reflected in the OBCC and NC-SI interfaces.
• Added protection from IOPX thermal diode destabilization to prevent UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GE cards.
• Fixed an issue which caused a link down in Port 2 when unplugging the cable from Port 1.
• In some cases, a Bit Error Rate is not optimal on 10G/40G links.
• Instability of Link Training Flow occurs during 100G Auto-Negotiation.
• Fixed a rare issue which caused the command to hang when moved the QP to RESET and back to RTS.
• Improved RDMA READ bandwidth under packet lost scenario.
• Added support for pnat = 1 in HCA access_reg command as required by the ibdiagnet tool.
• Fixed the LLDP OCBB response: return value is now ascii.
• Fixed a very rare NMI issue during PXE cycles.
• Increased the steering hash tables static size from 128 to a maximum of 32K entries.
• Prevented miscalculation of module temperature when using 100Gb/s cables (OPN: MFA1A00-Cxxx for 100GbE).
• Reduced one hop for Unicast RX steering, steering pipes balancing.
• Non-volatile configuration of Port Type TLV more than 50 times might cause system hang.
• Enabled RoCE IPv4 Multicast. This prevents MCG command from failing when an IPv4 is mapped to an IPv6 address.
• If the PF driver or the tool (e.g. ethtool) use PAOS DOWN command (e.g. by ifconfig down or ip link set down), loopback traffic is blocked for all functions on this port (PF<->VFs / VF<->VF) In Multihost loopback, the traffic will be blocked once the firmware receives the PAOS down command from all PFs. However, the loopback traffic will not be blocked when the port is down  due to the physical link (for example: cable plugged out, switch port down).
• Fixed a 25G and 50G link issue when Clause 91 RS FEC was active.
• Added a missing invalidation of eSwitch cache upon FLR which caused the upcoming driver load to either fail or not to be able to transmit.
• Fixed a UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GbE OCP card.
• Fixed an issue which prevented Vport counters from counting local loopback packets. Packets now are filter by the self-loopback prevention.
• Reported INTx as unsupported to allow PFs Passthrough on PowerKVM.
• SR-IOV Ethernet supports up to 18 VFs per port only.
• Fixed and incident what allowed local (internal) loopbacked packets to be counted by the Vport counters, although Vport counters should count only traffic that crosses the Vport.
• Fixed an issue preventing driver loading or TX traffic sending upon reboot, after ungraceful driver unload.
• Fixed casting of BMC MAC before steering API.
• Fixed the PCI write flow to take into consideration the PCI MTU. This fix eliminates the need for NOPs in the flow, which resulted from PPC larger PCI MTU.
  The single queue limitation for READ is due to a hardware limitation of the number of READ request in a given time.
• Fixed a case that caused FlexBoot to not work as expected with systems that run with "large bar" enabled (Above 4G Decoding) over Connect-IB or ConnectX-4 HCAs.
• Fixed an issue which prevented link creation when connected to IXIA 25G.

• HPG6 firmware improves drive reliability where disk drives are exposed to long periods of host inactivity which exceed 1 second.

• HPG6 firmware improves drive reliability where disk drives are exposed to long periods of host inactivity which exceed 1 second.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPG4.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPG4.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• Reliability enhancement for applications that write data to a narrow range of tracks.

• Reliability enhancement for applications that write data to a narrow range of tracks.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

• Traffic loss of large packets occured after MTU change.

• Traffic loss of large packets occured after MTU change.

• In ESXi5.5 – when using a server with many ConnectX-4 / ConnectX-4 Lx ports, some of the interfaces do not show up in the esxcfg-nics -l list. This can occur if no MSI-X resources are available to enable loading all interfaces.
• When SR-IOV is enabled and the max_vfs is not equal to 0, new filters are not applied.
• Added a new module parameter to nmlx5_core "supported_ num_ports" to support up to "8" ConnectX-4 Lx ports.
• Fixed an issue which prevented the driver from loading on machines with 64 or more CPU cores.

• In ESXi5.5 – when using a server with many ConnectX-4 / ConnectX-4 Lx ports, some of the interfaces do not show up in the esxcfg-nics -l list. This can occur if no MSI-X resources are available to enable loading all interfaces.
• When SR-IOV is enabled and the max_vfs is not equal to 0, new filters are not applied.
• Added a new module parameter to nmlx5_core "supported_ num_ports" to support up to "8" ConnectX-4 Lx ports.
• Fixed an issue which prevented the driver from loading on machines with 64 or more CPU cores.

• After long term use of the HDD, a rare condition might occur following a power cycle where the drive heads may land on areas of the disk containing data, which could potentially cause data loss or mechanical damage. Firmware version HPGD prevents this condition from occurring.

Problems Fixed for HPGD(C):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPGD (D):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• After long term use of the HDD, a rare condition might occur following a power cycle where the drive heads may land on areas of the disk containing data, which could potentially cause data loss or mechanical damage. Firmware version HPGD prevents this condition from occurring.

Problems Fixed for HPGD(C):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPGD (D):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• After long term use of the HDD, a rare condition might occur following a power cycle where the drive heads may land on areas of the disk containing data, which could potentially cause data loss or mechanical damage. Firmware version HPGD prevents this condition from occurring.

Problems Fixed for HPGD(C):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPGD (D):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• After long term use of the HDD, a rare condition might occur following a power cycle where the drive heads may land on areas of the disk containing data, which could potentially cause data loss or mechanical damage. Firmware version HPGD prevents this condition from occurring.

Problems Fixed for HPGD(C):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPGD (D):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

Problems Fixed for HPS4 (B):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

Problems Fixed for HPS4 (B):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Ver. 1.0.9 (F) contains support for new server products.  It is functionally equivalent to ver. 1.0.9.  It is not necessary to upgrade with Revision F if a previous component revision was used to upgrade the firmware to ver. 1.0.9.

Firmware Dependencies:

None

Problems Fixed:

Addresses an issue in which the minimum power capping value was incorrectly being calculated on certain systems. This fix increases the accuracy of the minimum capping value set during POST.

Known Issues:

None

Ver. 1.0.9 (F) contains support for new server products.  It is functionally equivalent to ver. 1.0.9.  It is not necessary to upgrade with Revision F if a previous component revision was used to upgrade the firmware to ver. 1.0.9.

Firmware Dependencies:

None

Problems Fixed:

Addresses an issue in which the minimum power capping value was incorrectly being calculated on certain systems. This fix increases the accuracy of the minimum capping value set during POST.

Known Issues:

None

Ver. 1.0.9(D) contains updates to the component packaging and is functionally equivalent to ver. 1.0.9.  It is not necessary to upgrade with Revision D if a previous component Revision was used to upgrade the firmware to version 1.0.9.

Firmware Dependencies:

None

Problems Fixed:

Addresses an issue in which the minimum power capping value was incorrectly being calculated on certain systems. This fix increases the accuracy of the minimum capping value set during POST.

Known Issues:

None

Ver. 1.0.9(D) contains updates to the component packaging and is functionally equivalent to ver. 1.0.9.  It is not necessary to upgrade with Revision D if a previous component Revision was used to upgrade the firmware to version 1.0.9.

Firmware Dependencies:

None

Problems Fixed:

Addresses an issue in which the minimum power capping value was incorrectly being calculated on certain systems. This fix increases the accuracy of the minimum capping value set during POST.

Known Issues:

None

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD7. 

• This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware prevents asserts on a sequential write which can result in devices being lost from the configuration. It also preserves persistent reservation through downloads which prevents loss of performance.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD7. 

• This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware prevents asserts on a sequential write which can result in devices being lost from the configuration. It also preserves persistent reservation through downloads which prevents loss of performance.

• Version HPD4 prevents the potential for incorrect data to be written to the drive under extremely rare circumstances when the drive experiences a hard reset. This issue has only been observed in a rigorous test environment and has NOT been reported in a customer production environment.

Problems Fixed for HPD4 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• Version HPD4 prevents the potential for incorrect data to be written to the drive under extremely rare circumstances when the drive experiences a hard reset. This issue has only been observed in a rigorous test environment and has NOT been reported in a customer production environment.

Problems Fixed for HPD4 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

Changed the settings in the SAS Expander to support disk discovery when 12Gb SAS HDDs are installed in the enclosure 

Changed the settings in the SAS Expander to support disk discovery when 12Gb SAS HDDs are installed in the enclosure 

• Under rarely occurring conditions where a host error occurred during a write with 4K unaligned sectors, the write operation might result in stale data. Firmware HPD3 prevents the possibility of stale data occurring under these conditions.

• Under rarely occurring conditions where a host error occurred during a write with 4K unaligned sectors, the write operation might result in stale data. Firmware HPD3 prevents the possibility of stale data occurring under these conditions.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD2.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD2.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• This firmware contains several low level maintenance fixes including debug data retrieval through read buffer.  The fixes addressed events observed in a specialized lab testing environment, and are not expected to be experienced in customer use case scenarios.

• RDMA_CM driver was causing the Queue Pair (QP) Rate Limit to be activated unexpectedly and reduce the bandwidth significantly on this QP.
• Low throughput when Eplicit Congestion Notification (ECN) was enabled in a many-to-one scenario.
• A rare issue that caused the firmware to hang.
• The maximum temperature in a setup was incorrectly reported as the current temperature regardless of the actual temperature.
• Virtual Addressing Capability incorrect report in NC-SI (Sideband).
• Enhanced RX sensitivity per Vcore.
• Disabled the option to write to the protected modules to avoid receiving NACK upon module initialization.
• Enabled cold_flicks reset when sending invalidate_all command to invalidate all the ECs.
• SR-IOV min & max rate limiter can only support up to 64 VFs per port.
• Explicit Congestion Notification (ECN) did not function as expected when the number of Queue Pairs (QPs) > ~500 per host.
• When using Unreliable Datagram (UD) RDMA over Converged Ethernet (RoCE) multicast traffic over SR-IOV, packets were scattered to all the attached QPs in the e-sw (PF and its VFs) and not only on the vport that was specified in the e-se FDB.

• RDMA_CM driver was causing the Queue Pair (QP) Rate Limit to be activated unexpectedly and reduce the bandwidth significantly on this QP.
• Low throughput when Eplicit Congestion Notification (ECN) was enabled in a many-to-one scenario.
• A rare issue that caused the firmware to hang.
• The maximum temperature in a setup was incorrectly reported as the current temperature regardless of the actual temperature.
• Virtual Addressing Capability incorrect report in NC-SI (Sideband).
• Enhanced RX sensitivity per Vcore.
• Disabled the option to write to the protected modules to avoid receiving NACK upon module initialization.
• Enabled cold_flicks reset when sending invalidate_all command to invalidate all the ECs.
• SR-IOV min & max rate limiter can only support up to 64 VFs per port.
• Explicit Congestion Notification (ECN) did not function as expected when the number of Queue Pairs (QPs) > ~500 per host.
• When using Unreliable Datagram (UD) RDMA over Converged Ethernet (RoCE) multicast traffic over SR-IOV, packets were scattered to all the attached QPs in the e-sw (PF and its VFs) and not only on the vport that was specified in the e-se FDB.

• Fixed an issue which caused bi-directional traffic 10% BW degradation in Multihost.
• Increased the CQE zipping aggressive mode timer to 9000.
• Moving IPoIB enhanced QP to ERR or RST state results in the corruption of the service_type and pm_state in the QPC.
• Attaching RoCE IPv4 QPs to MCG when the vport state is set to toggle (DOWN/UP), prevents the QPs that are listed on that MCG from receiving any traffic.
• When arming SRQ for limit event, the device might issue an event with context_index=0.
• Occasionally, when moving UD QP from error state to RTS, the QP re-enters the error state
• When performing Pkey check for IPoIB enhanced traffic, the Pkey membership bit is ignored.
• Stopping the Rate Limiter while traffic is being transmitted might cause the adapter card to hang.
• Privileged Vport egress traffic is not blocked when Vport is not active
• PF direct pass-through is not supported in InfiniBand (since PF FLR is not supported)
• Missing invalidation upon Set().pkey leads to bad Pkey checks.
• Fixed an issue which caused the HCA mad response to contain the incoming packet Pkey and not the matched Pkey.
• Modified PCIe Tx configuration.
• Fixed an issue that prevented the software to set ECN parameters (min_rate, max_rate, rate_to_set_on_first_cnp) to values >32768.
• Fixed an issue which caused the link speed to raise as DDR when connected with certain copper cables to devices supporting up to QDR speed.
• Fixed an issue which prevented physical counters from reseting. Now the physical counters are reset on first driver start.
• Fixed possible negotiation issues with 3rd parties.
• Fixed a rare issue which caused 56GbE link to raise with errors.
• Fixed an issue which caused scheduling_context.element_type to be taken into consideration with performing verifications, when running the modify_scheduling_context command, although the field is reserved.
• Fixed an issue which caused the eSwitch max_average_bw ref counter to decrement in TEARDOWN_HCA/ FLR VF regardless of the max_average_bw value set, although the ref counter design was to increment on every max_average_bw != 0 (limited).
• On rare occasions during UEFI boot cycles system got stuck while WinPE is loaded. (OS WinPE, system DL160).
• Single FTE that catches both untagged and prio-tagged packets (by giving an FTE with match_value.vlan_tag = 0 and match_value.vid = 0) is currently not supported.
• Flashing the firmware requires server reboot. Firmware cannot be flashed twice without server reboot after first flashing.
• If the vport state is DOWN and a packet is sent in local loopback, the sx_sniffer tool will not function.
• Fixed an issue causing bubbles to appear as symbol errors when link raised FDR 1x.
• When Clause 74 Fire-Code FEC is active, and there are FC corrected errors, both the FC_correctable counter and the FC_uncoreectable counter are increment.
• Some Port Control Register do not return to the default value after the last port owner host restarts the driver.
• Fixed an issue which caused RX to hang when the UDP packet had a reserved UDP destination port.
• Fixed DMAC reporting mapping per host.
• Fixed an EEH error from PCI which caused firmware to hang.
• Fixed the default value of the PCIe target_link_speed to Gen3 in link control2.
• Fixed an issue which prevented LEDs from blinking when the traffic was less than 0.1% of the link speed.
• Fixed an issue which caused the mlxconfig configuration of VF_LOG_BAR_SIZE to be ignored and to be set to 5 (32MB).
• Fixed an issues which occasionally caused the driver to hang during unload on some VLs when configuring the SM with a VL weight 0 and running traffic on it.
• A server getting into a Standby mode while Packet-Pacing is enabled might cause firmware to hang and driver call-trace.
• Fixed an issue which caused unexpected QoS functionality in case of multiple sources to single destination traffic transmission.
• Fixed an issue which occasionally caused the RX traffic to hang in DC when received a PCI error on WQE fetch.
• Fixed OOB connection issue during Intel's ITP inject errors test.
• Fixed an issue in IEEE Auto-Negotiation where 25G FireCode FEC and 25G Reed-Solomon FEC bits were reversed.
• Fixed an issue which caused RX to hang when a UDP packet with destination port of RoCE v2 arrived and the data matched the DC transport service.
• Added protection from IOPX thermal diode destabilization to prevent UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GE OCP card.
• Fixed an issue causing single port devices to query and write Physical Port TLVs to Port 2.
• Enabled mlxfwreset to work using the PCIe Secondary Bus Reset.
• Fixed an issue causing link flapping as a result, incorrect link settings.
• Fixed an issue causing wrong alignment markers to be used when running 50G with Clause91 FEC enabled.
• Reduced the default BAR size for VF (SR-IOV) from 5 (32 MB) to 1 (2MB).
• Added legacy interrupts support in FlexBoot.
• Modified the TX configuration to support EMI crossing margins in 16Ghz.
• In some cases, a Bit Error Rate is not optimal on 10G/40G links.
• Instability of Link Training Flow occurs during 100G Auto-Negotiation.
• Fixed a rare issue which caused the command to hang when moved the QP to RESET and back to RTS.
• Improved RDMA READ bandwidth under packet lost scenario.
• Added support for pnat = 1 in HCA access_reg command as required by the ibdiagnet tool.
• Fixed the LLDP OCBB response: return value is now ascii.
• Fixed a very rare NMI issue during PXE cycles.
• Increased the steering hash tables static size from 128 to a maximum of 32K entries.
• Prevented miscalculation of module temperature when using 100Gb/s cables (OPN: MFA1A00-Cxxx for 100GbE and MFA1A00-Exxx for IB EDR).
• Fixed an issue which caused the device to hang when resetting qkey/pkey violation counter via port_info mad.
• Changed the MTU size in OCBB report. Now MTU size does not include the packet headers.
• Reduced one hop for Unicast RX steering, steering pipes balancing.
• Non-volatile configuration of Port Type TLV more than 50 times might cause system hang.
• Fixed a 25G and 50G link issue when Clause 91 RS FEC was active.
• Added a missing invalidation of SwitchX cache upon FLR which caused the upcoming driver load to either fail or not to be able to transmit.
• Fixed an issue which prevented Vport counters from counting local loopback packets. Packets now are filter by the self-loopback prevent
• Reported INTx as unsupported to allow PFs Passthrough on PowerKVM.
• SR-IOV Ethernet supports up to 18 VFs per port only.
• Fixed and incident what allowed local (internal) loopbacked packets to be counted by the Vport counters, although Vport counters should count only traffic that crosses the Vport.
• Fixed an issue preventing driver loading or TX traffic sending upon reboot, after ungraceful driver unload.
• Fixed casting of BMC MAC before steering API.
• Fixed the PCI write flow to take into consideration the PCI MTU. This fix eliminates the need for NOPs in the flow, which resulted from PPC larger PCI MTU.
  The single queue limitation for READ is due to a hardware limitation of the number of READ request in a given time.
• Fixed a case that caused FlexBoot to not work as expected with systems that run with "large bar" enabled (Above 4G Decoding) over Connect-IB or ConnectX-4 HCAs.
• Fixed PLDM support for single port NICs. Currently port’s relevant sensors/states are reported only for a single port.

• Fixed an issue which caused bi-directional traffic 10% BW degradation in Multihost.
• Increased the CQE zipping aggressive mode timer to 9000.
• Moving IPoIB enhanced QP to ERR or RST state results in the corruption of the service_type and pm_state in the QPC.
• Attaching RoCE IPv4 QPs to MCG when the vport state is set to toggle (DOWN/UP), prevents the QPs that are listed on that MCG from receiving any traffic.
• When arming SRQ for limit event, the device might issue an event with context_index=0.
• Occasionally, when moving UD QP from error state to RTS, the QP re-enters the error state
• When performing Pkey check for IPoIB enhanced traffic, the Pkey membership bit is ignored.
• Stopping the Rate Limiter while traffic is being transmitted might cause the adapter card to hang.
• Privileged Vport egress traffic is not blocked when Vport is not active
• PF direct pass-through is not supported in InfiniBand (since PF FLR is not supported)
• Missing invalidation upon Set().pkey leads to bad Pkey checks.
• Fixed an issue which caused the HCA mad response to contain the incoming packet Pkey and not the matched Pkey.
• Modified PCIe Tx configuration.
• Fixed an issue that prevented the software to set ECN parameters (min_rate, max_rate, rate_to_set_on_first_cnp) to values >32768.
• Fixed an issue which caused the link speed to raise as DDR when connected with certain copper cables to devices supporting up to QDR speed.
• Fixed an issue which prevented physical counters from reseting. Now the physical counters are reset on first driver start.
• Fixed possible negotiation issues with 3rd parties.
• Fixed a rare issue which caused 56GbE link to raise with errors.
• Fixed an issue which caused scheduling_context.element_type to be taken into consideration with performing verifications, when running the modify_scheduling_context command, although the field is reserved.
• Fixed an issue which caused the eSwitch max_average_bw ref counter to decrement in TEARDOWN_HCA/ FLR VF regardless of the max_average_bw value set, although the ref counter design was to increment on every max_average_bw != 0 (limited).
• On rare occasions during UEFI boot cycles system got stuck while WinPE is loaded. (OS WinPE, system DL160).
• Single FTE that catches both untagged and prio-tagged packets (by giving an FTE with match_value.vlan_tag = 0 and match_value.vid = 0) is currently not supported.
• Flashing the firmware requires server reboot. Firmware cannot be flashed twice without server reboot after first flashing.
• If the vport state is DOWN and a packet is sent in local loopback, the sx_sniffer tool will not function.
• Fixed an issue causing bubbles to appear as symbol errors when link raised FDR 1x.
• When Clause 74 Fire-Code FEC is active, and there are FC corrected errors, both the FC_correctable counter and the FC_uncoreectable counter are increment.
• Some Port Control Register do not return to the default value after the last port owner host restarts the driver.
• Fixed an issue which caused RX to hang when the UDP packet had a reserved UDP destination port.
• Fixed DMAC reporting mapping per host.
• Fixed an EEH error from PCI which caused firmware to hang.
• Fixed the default value of the PCIe target_link_speed to Gen3 in link control2.
• Fixed an issue which prevented LEDs from blinking when the traffic was less than 0.1% of the link speed.
• Fixed an issue which caused the mlxconfig configuration of VF_LOG_BAR_SIZE to be ignored and to be set to 5 (32MB).
• Fixed an issues which occasionally caused the driver to hang during unload on some VLs when configuring the SM with a VL weight 0 and running traffic on it.
• A server getting into a Standby mode while Packet-Pacing is enabled might cause firmware to hang and driver call-trace.
• Fixed an issue which caused unexpected QoS functionality in case of multiple sources to single destination traffic transmission.
• Fixed an issue which occasionally caused the RX traffic to hang in DC when received a PCI error on WQE fetch.
• Fixed OOB connection issue during Intel's ITP inject errors test.
• Fixed an issue in IEEE Auto-Negotiation where 25G FireCode FEC and 25G Reed-Solomon FEC bits were reversed.
• Fixed an issue which caused RX to hang when a UDP packet with destination port of RoCE v2 arrived and the data matched the DC transport service.
• Added protection from IOPX thermal diode destabilization to prevent UEFI IPv6 PXE boot failure on ConnectX-4 Lx 25GE OCP card.
• Fixed an issue causing single port devices to query and write Physical Port TLVs to Port 2.
• Enabled mlxfwreset to work using the PCIe Secondary Bus Reset.
• Fixed an issue causing link flapping as a result, incorrect link settings.
• Fixed an issue causing wrong alignment markers to be used when running 50G with Clause91 FEC enabled.
• Reduced the default BAR size for VF (SR-IOV) from 5 (32 MB) to 1 (2MB).
• Added legacy interrupts support in FlexBoot.
• Modified the TX configuration to support EMI crossing margins in 16Ghz.
• In some cases, a Bit Error Rate is not optimal on 10G/40G links.
• Instability of Link Training Flow occurs during 100G Auto-Negotiation.
• Fixed a rare issue which caused the command to hang when moved the QP to RESET and back to RTS.
• Improved RDMA READ bandwidth under packet lost scenario.
• Added support for pnat = 1 in HCA access_reg command as required by the ibdiagnet tool.
• Fixed the LLDP OCBB response: return value is now ascii.
• Fixed a very rare NMI issue during PXE cycles.
• Increased the steering hash tables static size from 128 to a maximum of 32K entries.
• Prevented miscalculation of module temperature when using 100Gb/s cables (OPN: MFA1A00-Cxxx for 100GbE and MFA1A00-Exxx for IB EDR).
• Fixed an issue which caused the device to hang when resetting qkey/pkey violation counter via port_info mad.
• Changed the MTU size in OCBB report. Now MTU size does not include the packet headers.
• Reduced one hop for Unicast RX steering, steering pipes balancing.
• Non-volatile configuration of Port Type TLV more than 50 times might cause system hang.
• Fixed a 25G and 50G link issue when Clause 91 RS FEC was active.
• Added a missing invalidation of SwitchX cache upon FLR which caused the upcoming driver load to either fail or not to be able to transmit.
• Fixed an issue which prevented Vport counters from counting local loopback packets. Packets now are filter by the self-loopback prevent
• Reported INTx as unsupported to allow PFs Passthrough on PowerKVM.
• SR-IOV Ethernet supports up to 18 VFs per port only.
• Fixed and incident what allowed local (internal) loopbacked packets to be counted by the Vport counters, although Vport counters should count only traffic that crosses the Vport.
• Fixed an issue preventing driver loading or TX traffic sending upon reboot, after ungraceful driver unload.
• Fixed casting of BMC MAC before steering API.
• Fixed the PCI write flow to take into consideration the PCI MTU. This fix eliminates the need for NOPs in the flow, which resulted from PPC larger PCI MTU.
  The single queue limitation for READ is due to a hardware limitation of the number of READ request in a given time.
• Fixed a case that caused FlexBoot to not work as expected with systems that run with "large bar" enabled (Above 4G Decoding) over Connect-IB or ConnectX-4 HCAs.
• Fixed PLDM support for single port NICs. Currently port’s relevant sensors/states are reported only for a single port.

• Reliability enhancement for applications that write data to a narrow range of tracks.

• Reliability enhancement for applications that write data to a narrow range of tracks.

• Reliability enhancement for applications that write data to a narrow range of tracks.

• Reliability enhancement for applications that write data to a narrow range of tracks.

• Fixes a data integrity risk which could occur during 4k or greater unaligned writes while the device incurs a smart trip event. During these conditions there is a potential for data intended to be written directly to disk to fail to be written.
• Eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written. 

• Fixes a data integrity risk which could occur during 4k or greater unaligned writes while the device incurs a smart trip event. During these conditions there is a potential for data intended to be written directly to disk to fail to be written.
• Eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written. 

• Fixes a rare but potential data integrity error during low 5v drive voltage and specific sequential data streaming conditions, which could result in data written to incorrect sectors.

Problems Fixed for HPG5 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

Problems Fixed for HPG5 (E):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• Fixes a rare but potential data integrity error during low 5v drive voltage and specific sequential data streaming conditions, which could result in data written to incorrect sectors.

Problems Fixed for HPG5 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

Problems Fixed for HPG5 (E):

• When attempting to update drive firmware in a VMware vSphere 6.5 environment, the update would fail and the event was logged as a segmentation fault error.

• Fixes a rare but potential data integrity error during low 5v drive voltage and specific sequential data streaming conditions, which could result in data written to incorrect sectors.

Problems Fixed for HPG5 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• Fixes a rare but potential data integrity error during low 5v drive voltage and specific sequential data streaming conditions, which could result in data written to incorrect sectors.

Problems Fixed for HPG5 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

• Fixes a data integrity risk which could occur during 4k or greater unaligned writes while the device incurs a smart trip event. During these conditions there is a potential for data intended to be written directly to disk to fail to be written.
• Eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written. 

• Fixes a data integrity risk which could occur during 4k or greater unaligned writes while the device incurs a smart trip event. During these conditions there is a potential for data intended to be written directly to disk to fail to be written.
• Eliminates a data integrity risk when an unaligned Write and Verify command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written. 

• This firmware corrects a potential issue where the data in the reserved tracks is not properly updated, eliminating the risk of a drive not finishing the boot process on power up.
• Other maintenance fixes and updates are also included with the new firmware.

• This firmware corrects a potential issue where the data in the reserved tracks is not properly updated, eliminating the risk of a drive not finishing the boot process on power up.
• Other maintenance fixes and updates are also included with the new firmware.

• This firmware corrects a potential issue where the data in the reserved tracks is not properly updated, eliminating the risk of a drive not finishing the boot process on power up.
• Other maintenance fixes and updates are also included with the new firmware.

• This firmware corrects a potential issue where the data in the reserved tracks is not properly updated, eliminating the risk of a drive not finishing the boot process on power up.
• Other maintenance fixes and updates are also included with the new firmware.

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

8Gb Standup & 8Gb Mezzanine

BIOS

•  CTRL-Q cosmetic unexpected behavior

UEFI

•  Simple Control Transmission (SCT) 2.5 Extended Small Computer System Interface (SCSI) Pass Thru Protocol test incompleteness. PassThru_Func test no longer goes incomplete.
•  Operating System (OS) boot incomplete behavior when Read Only Memory (ROM) Based Setup-Utility (RBSU) Power On Self -Test (POST) Discovery Mode was set to Force Fast Discovery
•  Human Interface Infrastructure (HII) save unexpected behavior on Superdome servers. In the World Wide Name (WWN) Database menu, changes to Drive0 and Drive1 are now saved to Non  Volatile Random Access Memory (NVRAM)
•  Unexpected behavior which caused adapter to not appear on the Health Summary page.
•  Operating System (OS) boot incomplete behavior when Read Only Memory (ROM) Based Setup-Utility (RBSU) Power On Self -Test (POST) Discovery Mode was set to Auto
•  World Login unexpected behavior when Power On Self -Test (POST) Discovery Mode was set to Auto

16Gb Standup & 16Gb Mezzanine

UEFI

•  Simple Control Transmission (SCT) 2.5 Extended Small Computer System Interface (SCSI) Pass Thru Protocol test incompleteness. PassThru_Func test no longer goes incomplete

16Gb/32Gb Standup

UEFI

•  Human Interface Infrastructure (HII) save unexpected behavior. In the World Wide Name (WWN) Database menu, changes to Drive0 and Drive1 are now saved to Non Volatile Random Access Memory (NVRAM).
•  Corrected adapter name format in Read Only Memory (ROM) Based Setup-Utility (RBSU) menus.
•  Unexpected behavior that caused Non Volatile Random Access Memory (NVRAM) overwritten after Human Interface Infrastructure (HII) changes
•  Behavior that caused Human Interface Infrastructure (HII) fields to not revert back to original settings when ESC key was used to discard changes

8Gb Standup & 8Gb Mezzanine

BIOS

•  CTRL-Q cosmetic unexpected behavior

UEFI

•  Simple Control Transmission (SCT) 2.5 Extended Small Computer System Interface (SCSI) Pass Thru Protocol test incompleteness. PassThru_Func test no longer goes incomplete.
•  Operating System (OS) boot incomplete behavior when Read Only Memory (ROM) Based Setup-Utility (RBSU) Power On Self -Test (POST) Discovery Mode was set to Force Fast Discovery
•  Human Interface Infrastructure (HII) save unexpected behavior on Superdome servers. In the World Wide Name (WWN) Database menu, changes to Drive0 and Drive1 are now saved to Non  Volatile Random Access Memory (NVRAM)
•  Unexpected behavior which caused adapter to not appear on the Health Summary page.
•  Operating System (OS) boot incomplete behavior when Read Only Memory (ROM) Based Setup-Utility (RBSU) Power On Self -Test (POST) Discovery Mode was set to Auto
•  World Login unexpected behavior when Power On Self -Test (POST) Discovery Mode was set to Auto

16Gb Standup & 16Gb Mezzanine

UEFI

•  Simple Control Transmission (SCT) 2.5 Extended Small Computer System Interface (SCSI) Pass Thru Protocol test incompleteness. PassThru_Func test no longer goes incomplete

16Gb/32Gb Standup

UEFI

•  Human Interface Infrastructure (HII) save unexpected behavior. In the World Wide Name (WWN) Database menu, changes to Drive0 and Drive1 are now saved to Non Volatile Random Access Memory (NVRAM).
•  Corrected adapter name format in Read Only Memory (ROM) Based Setup-Utility (RBSU) menus.
•  Unexpected behavior that caused Non Volatile Random Access Memory (NVRAM) overwritten after Human Interface Infrastructure (HII) changes
•  Behavior that caused Human Interface Infrastructure (HII) fields to not revert back to original settings when ESC key was used to discard changes

BIOS

• Incorrect processing of Command Line Processor (CLP)  ‘set netport’
• Improper display of Command Line Processor (CLP) Logical Unit Number (LUN)

UEFI

• Changed Operation Mode Human Interface Infrastructure (HII) strings. The new strings match the FastUTIL strings.
• Self-certification Test (SCT) 2.4B BlockIO2 Protocol bug.
• No response during Self-certification Test (SCT) 2.4B BlockIO2 Protocol test.
• Self-certification Test (SCT) 2.4B AdapterInfo Protocol bugs.

16Gb Standup & 16Gb Mezzanine

UEFI

• Adapter not appearing in Health Summary page.
• Invalid Health message when link was down or driver was disabled.
• Adapter connection mode to change (OS install related) when the Human Interface Infrastructure (HII) Link Speed field changed.

32Gb Standup & 16Gb Mezzanine

UEFI

• Adapter not appearing in Health Summary page.
• Invalid Health message when link was down or driver was disabled.
• Added code to start operational firmware early.

BIOS

• Incorrect processing of Command Line Processor (CLP)  ‘set netport’
• Improper display of Command Line Processor (CLP) Logical Unit Number (LUN)

UEFI

• Changed Operation Mode Human Interface Infrastructure (HII) strings. The new strings match the FastUTIL strings.
• Self-certification Test (SCT) 2.4B BlockIO2 Protocol bug.
• No response during Self-certification Test (SCT) 2.4B BlockIO2 Protocol test.
• Self-certification Test (SCT) 2.4B AdapterInfo Protocol bugs.

16Gb Standup & 16Gb Mezzanine

UEFI

• Adapter not appearing in Health Summary page.
• Invalid Health message when link was down or driver was disabled.
• Adapter connection mode to change (OS install related) when the Human Interface Infrastructure (HII) Link Speed field changed.

32Gb Standup & 16Gb Mezzanine

UEFI

• Adapter not appearing in Health Summary page.
• Invalid Health message when link was down or driver was disabled.
• Added code to start operational firmware early.

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

• Reliability enhancement for applications that write data to a narrow range of tracks.

Known Issues:

• Firmware cannot be downgraded to HPG3 after updating to HPG4.

• Reliability enhancement for applications that write data to a narrow range of tracks.

Known Issues:

• Firmware cannot be downgraded to HPG3 after updating to HPG4.

• Reliability enhancement for applications that write data to a narrow range of tracks.

Known Issues:

• Firmware cannot be downgraded to HPG3 after updating to HPG4.

• Reliability enhancement for applications that write data to a narrow range of tracks.

Known Issues:

• Firmware cannot be downgraded to HPG3 after updating to HPG4.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPG5.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPG5.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• After HDD power cycle, the Background Media Scan (BGMS) pointer would invoke a second BGMS routine starting at LBA 0, which could affect  performance until the second BGMS has completed.
• When writing to the HDD after it entered or returned from a standby state, the potential existed for a Non Volatile Cache issue to occur..

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• After HDD power cycle, the Background Media Scan (BGMS) pointer would invoke a second BGMS routine starting at LBA 0, which could affect  performance until the second BGMS has completed.
• When writing to the HDD after it entered or returned from a standby state, the potential existed for a Non Volatile Cache issue to occur..

Problems Fixed for HPG3 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• Normalized duplicate timeout field contexts.
• Duplicate timeout fields used in wrong contexts.
• ql_dump_buffer to avoid emitting on multiple lines.
• Allowed Simple Name Server (SNS) fabric logins to be retried.
• Callback for memory cleanup
• Relogin from being triggered too fast.

• Normalized duplicate timeout field contexts.
• Duplicate timeout fields used in wrong contexts.
• ql_dump_buffer to avoid emitting on multiple lines.
• Allowed Simple Name Server (SNS) fabric logins to be retried.
• Callback for memory cleanup
• Relogin from being triggered too fast.

• Snoop Inquiry response spanning multiple scatter/gather elements.

• Snoop Inquiry response spanning multiple scatter/gather elements.

• Under rarely occurring conditions where a host error occurred during a write with 4K unaligned sectors, the write operation might result in stale data. Firmware HPD3 prevents the possibility of stale data occurring under these conditions.

• Under rarely occurring conditions where a host error occurred during a write with 4K unaligned sectors, the write operation might result in stale data. Firmware HPD3 prevents the possibility of stale data occurring under these conditions.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD2.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD2.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• Improved handling of IPV6 addresses

• Updated for RHEL 7.4 support

• Supports SLES12 SP3

• Improved handling of SATA drive temperatures

• Improved handling of IPV6 addresses

• Updated for RHEL 7.4 support

• Supports SLES12 SP3

• Improved handling of SATA drive temperatures

• This firmware corrects possible unrecovered errors caused by the track refresh algorithm not working properly.

Problems Fixed for HPG7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware corrects possible unrecovered errors caused by the track refresh algorithm not working properly.

Problems Fixed for HPG7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware corrects possible unrecovered errors caused by the track refresh algorithm not working properly.

Problems Fixed for HPG7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

• This firmware corrects possible unrecovered errors caused by the track refresh algorithm not working properly.

Problems Fixed for HPG7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component fails to update the firmware for drives connected behind an HP Smart Array P431 controller. Component reports the firmware update was successful. However, after a power cycle, the firmware version remains unchanged.

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

• HPS4 firmware release resolved a firmware timing issue which occurred during drive long self-test and resulted in a timeout condition that caused the drive to become unrecognized by the system.

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version

Known Issues:

None

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

None

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

• Enclosure in Unexpected status: Non-critical after restoring the Encl ID.

• Enclosure in Unexpected status: Non-critical after restoring the Encl ID.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• This firmware fixes a potential incorrect data issue in write-cached enabled multi-initiator unaligned write environments, where reservation commands are used.

Problems Fixed for HPD7 (B):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.

• This firmware contains a change to prevent occasional command completion times in the 4-5 second window when command is received just as the drive is transitioning from active to Idle A.

• This firmware contains a change to prevent occasional command completion times in the 4-5 second window when command is received just as the drive is transitioning from active to Idle A.

• Removed the very verbose command output for hfi1_eprom

Fixes in version 1.5.2:

• Fix in the usage string
• Fix for listing the PCI domain name

Fixes in version 1.4.2:

• Initial version of the Intel OPA firmware component.

• Removed the very verbose command output for hfi1_eprom

Fixes in version 1.5.2:

• Fix in the usage string
• Fix for listing the PCI domain name

Fixes in version 1.4.2:

• Initial version of the Intel OPA firmware component.

• This firmware contains a change to prevent a drive reset issue, which may affect performance.

• This firmware contains a change to prevent a drive reset issue, which may affect performance.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD5.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD5.  This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

This revision of the System ROM includes the latest revision of the Intel microcode which, in combination with operating system updates, provides mitigation for Variant 2 of the Side Channel Analysis vulnerability, also known as Spectre. The revision of the microcode included in this System ROM addresses issues with more frequent reboots and unpredictable system behavior which impacted the previous Intel microcode which was part of the Spectre Variant 2 mitigation. Additional information is available from Intel’s Security Exploit Newsroom, https://newsroom.intel.com/press-kits/security-exploits-intel-products/.

Firmware Dependencies:

None

Problems Fixed:

Updated the Intel processor microcode to the latest version.

Known Issues:

None

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

Firmware Dependencies:

Problems Fixed:

Addressed an issue where the system may experience an erroneous boot-time message and Integrated Management Log (IML) entry stating that Core Boost Technology is disabled due to a lack of an iLO Advanced License (531 - Core Boost Technology missing required iLO License) when using the Version 1.20 System ROM. This issue does NOT occur with the Version 1.00 or Version 1.22 System ROM. Note that this IML entry is only valid for servers using the Intel Xeon Scalable 6143 processor. If the message and IML entry are received with any other processor model installed, the message is erroneous and can be ignored without any negative impact to the system.

Known Issues:

• This firmware resolves several drive issues including internal resets, firmware download functionality, and performance issues.  This firmware could improve the robustness of the system by eliminating these possible issues within the drive.

• This firmware resolves several drive issues including internal resets, firmware download functionality, and performance issues.  This firmware could improve the robustness of the system by eliminating these possible issues within the drive.

• Incorrect data reads might occur when accessing unaligned 4k reads and the data read has zero content preceding customer data. (Additional information can be obtained for this issue in Customer Advisory c04650586)

Problems Fixed for HPD1 (B):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPD1 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• Incorrect data reads might occur when accessing unaligned 4k reads and the data read has zero content preceding customer data. (Additional information can be obtained for this issue in Customer Advisory c04650586)

Problems Fixed for HPD1 (B):

• The component would fail to flash drive firmware on a server with a Trusted Platform Module (TPM) enabled when using the /tpmbypass switch.

Problems Fixed for HPD1 (C):

• Component would fail to install drive firmware for drives present in a system configured with two or more external drive enclosures attached to an HP Host Bus Adapter H22x. The following message would be reported in the component log file - “Device appears more than once in tree”. The drive firmware installation failure was not observed in configurations having only one external drive enclosure attached to an HP Host Bus Adapter H22x.
• Component would cause an exception error when deployed on a computer with a fully qualified domain name greater than 40 characters in length.

• NVMe drive details are not displayed in the iLO web interface for HPE ProLiant XL190r Gen10 servers.
• The system BIOS version is missing in AlertMail messages.
• The IPMI Get Chassis Capabilities and Get Chassis Status commands do not reflect the Chassis Intrusion state.
• Virtual Media devices are inaccessible when two virtual devices are mounted simultaneously.
• Patched configuration file system.
• The IPMI pre-watchdog-timeout interrupt does not occur when no watchdog action is selected.
• Intermittent loss of network interface card (NIC) and drive detection (P408e-m controller) by BIOS on system boot up of HPE ProLiant BL460c servers.

• NVMe drive details are not displayed in the iLO web interface for HPE ProLiant XL190r Gen10 servers.
• The system BIOS version is missing in AlertMail messages.
• The IPMI Get Chassis Capabilities and Get Chassis Status commands do not reflect the Chassis Intrusion state.
• Virtual Media devices are inaccessible when two virtual devices are mounted simultaneously.
• Patched configuration file system.
• The IPMI pre-watchdog-timeout interrupt does not occur when no watchdog action is selected.
• Intermittent loss of network interface card (NIC) and drive detection (P408e-m controller) by BIOS on system boot up of HPE ProLiant BL460c servers.

  • SSH does not work correctly when iLO is configured to use the SuiteB security state.  
  • Importing a certificate does not work correctly after the generation of a SuiteB-compliant CSR.

  • SSH does not work correctly when iLO is configured to use the SuiteB security state.  
  • Importing a certificate does not work correctly after the generation of a SuiteB-compliant CSR.

• Improves the drive idle time behavior by reducing long periods of track dwell time, which if not implemented, could increase media lube depletion and reduce drive reliability.  

• Improves the drive idle time behavior by reducing long periods of track dwell time, which if not implemented, could increase media lube depletion and reduce drive reliability.  

• Improves the drive idle time behavior by reducing long periods of track dwell time, which if not implemented, could increase media lube depletion and reduce drive reliability.  

• Improves the drive idle time behavior by reducing long periods of track dwell time, which if not implemented, could increase media lube depletion and reduce drive reliability.  

http://h20564.www2.hpe.com/portal/site/hpsc/public/kb/docDisplay/?docId=emr_na-a00025801en_us

http://h20564.www2.hpe.com/portal/site/hpsc/public/kb/docDisplay/?docId=emr_na-a00025801en_us

• Enabling ExpressLane on LUNs caused the LUN to become inaccessible

• Enabling ExpressLane on LUNs caused the LUN to become inaccessible

Fixed the following:

• Link down encountered during high traffic on 81E, 82E, 84E, LPe1205a

UEFI:

• Boot Device Order menu not displaying targets on HPE servers

16G Firmware:

Fixed the following:

• Incorrect adapter showed up in PCI slot 1 in Dashboard and Static Networking

16G/32G Firmware:

Fixed the following:

• Incorrect adapter showed up in PCI slot 1 in Dashboard and Static Networking

Fixed the following:

• Link down encountered during high traffic on 81E, 82E, 84E, LPe1205a

UEFI:

• Boot Device Order menu not displaying targets on HPE servers

16G Firmware:

Fixed the following:

• Incorrect adapter showed up in PCI slot 1 in Dashboard and Static Networking

16G/32G Firmware:

Fixed the following:

• Incorrect adapter showed up in PCI slot 1 in Dashboard and Static Networking

• This firmware update eliminates a data integrity risk when an unaligned WRITE and VERIFY command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written.

• This firmware update eliminates a data integrity risk when an unaligned WRITE and VERIFY command is sent to a bad sector. During these conditions there is a potential for data intended to be written to disk to fail to be written.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD6. 

• This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware contains a change to prevent an incorrect sense code from being posted when a Stop command is received during power-on sequence.

• This firmware improves HDD reliability when HDDs are exposed to long periods of host inactivity that exceed 1 second. HDDs may become unresponsive when using HDD firmware prior to version HPD6. 

• This firmware also contains a change which prevents an incorrect reassign status of a repaired sector from being logged.

• This firmware contains a change to prevent an incorrect sense code from being posted when a Stop command is received during power-on sequence.

• This firmware contains a change to prevent an incorrect sense code from being posted when a Stop command is received during power-on sequence.

• This firmware contains a change to prevent an incorrect sense code from being posted when a Stop command is received during power-on sequence.

• This firmware implements the Firmware Security Log Page.

• This firmware implements the Firmware Security Log Page.