Bug Fixes in this Firmware Version | ConnectX-8 Firmware RN

Internal Ref.	Issue
5000069	Description: Fixed an issue where repeatedly creating and destroying virtual machines with ConnectX-8 virtual functions could cause a hypervisor panic.
	Keywords: VM, VF
	Detected in version: 40.46.1006
	Fixed in Release: 40.50.1002
5159911 / NVbug 6427609	Description: Fixed an issue where a full NIC flap could occur due to a failed I2C expander read transaction. A retry mechanism was added for failed expander read transactions.
	Keywords: NIC flap
	Detected in version: 40.46.1006
	Fixed in Release: 40.50.1002
4988895	Description: Added 25G support in the INI configuration.
	Keywords: Cable speed
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4994872	Description: Fixed link-up issues by correcting marker lane and virtual lane calculations across split configurations, centralizing Ethernet interleave detection logic, passing FEC mode into PLU Ethernet configuration, and improving debug logs for marker mapping.
	Keywords: Link-up, FEC
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5028793	Description: Resolved issues in the TLP-SACK handling flow, including PSN calculation, SACK packet handling, and capability enablement for ConnectX-8 and later devices.
	Keywords: TLP-SACK handling flow
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5232646	Description: Resolved an issue where PLDM update from the auxiliary card side could fail on devices in multi-host configurations.
	Keywords: PLDM
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5073671	Description: Resolved an issue where a firmware assert could be raised due to a GPIO expander write timeout.
	Keywords: GPIO
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5130705 / 5131942 / NVbug 6376684	Description: Resolved an issue that could cause CoRIM FLM test failures during SBIOS and SMA validation on MGX Vera platform/system configurations.
	Keywords: CoRIM
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5139055 / 5072582 / NVbug 6427860	Description: Resolved an issue where firmware did not validate all port health threshold slots. Firmware now rejects unsupported threshold values instead of silently accepting them.
	Keywords: Firmware threshold
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5173627	Description: Resolved an issue where critical messages could be dropped due to queue overflow.
	Keywords: Queue overflow
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5041711	Description: Resolved an issue where, after toggling a port, the peer could incorrectly remain in the PHY-up state.
	Keywords: Port toggle
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4971613 / 4852107	Description: Resolved an issue where, after setting FORCE on a specific speed and rebooting, mlxlink could incorrectly show all default speeds as enabled instead of showing only the forced speed.
	Keywords: mlxlink
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4936278	Description: Added support for 10G and 25G using limited LLU request messages via the disable_b2b flag, with clause49_en enabled and the numerator/denominator configured to a 1:1 ratio.
	Keywords: LLU, 10G/25G speed
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
5042147 / 5041441	Description: Updated the Best Known Values (BKV) to version 1.1, improving the overall stability and robustness of the link-training reset flow during boot-up.
	Keywords: BKV
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
5012336 / NVbug 6140150	Description: Resolved a potential issue that could result in lower-than-expected bidirectional RDMA bandwidth in PCIe switch topologies.
	Keywords: RDMA bandwidth
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4851684	Description: Resolved an issue where RDMA workloads using null memory regions with MKEY_BY_NAME enabled could experience up to a 30% throughput drop. The internal null/dump-fill key was incorrectly routed through the slower signature datapath. Workloads not using MKEY_BY_NAME or null memory regions were unaffected.
	Keywords: RDMA
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4851684	Description: Resolved an issue where, under high-stress LRO traffic, an internal FIFO could enter a hardware deadlock state, preventing the NIC from receiving additional traffic. The fix detects and releases the deadlock if it occurs, restoring the NIC to a functional state.
	Keywords: LRO traffic
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5174504 / 5140787	Description: Resolved an issue where intensive parallel QP INIT2RTR / DESTROY operations could incur high operation latency and reduce connection establishment rates when PCC was enabled.
	Keywords: PCC, INIT2RTR, DOCA PCC
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5044357	Description: Resolved an issue where a RISC could become stuck in a fw_memcpy loop called by the multi-ASIC I2C mailbox function. This could occur if firmware read invalid parameters from I2C HW fields. Parameter validation was added before calling fw_memcpy, along with an error indication when the parameters are invalid.
	Keywords: RISC
	Detected in version: 40.47.1026
	Fixed in Release: 40.50.1002
5166648	Description: Resolved an issue involving an uninitialized stack read and an inverted EtherType check in the RX parser.
	Keywords: RX parser
	Detected in version: 40.47.1026
	Fixed in Release: 40.50.1002
4937640	Description: Resolved an issue where a PPCC mlxreg read operation could succeed when executed on a PF that is not the port owner, which could have security implications.
	Keywords: PCC, PPCC
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5161601	Description: Resolved an issue where multiplane emulation mode conditioning used the raw num_pfs value before invalid values were reduced during FMT initialization. The fix reuses the PF reduction calculation after logical-port discovery, without updating PCI settings before FMT.
	Keywords: PF, FMT initialization
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
4670019	Description: Resolved an issue where GPIO expander write operations could take more than 200 ms.
	Keywords: GPIO expander
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5154717 / 5152076 / 5154806 / NVbug 6422398	Description: Resolved an issue where, when a transceiver module held its I2C management bus line stuck (SDA or SCL), the firmware detected the stuck bus and reset the module. After the reset, the module remained in low-power mode and its datapath was not re-established, preventing the link from coming up. On a shared module, all associated ports could drop simultaneously.
	Keywords: Cables
	Detected in version: 40.45.3034
	Fixed in Release: 40.50.1002
4923806 / 4867111 / 4927875	Description: Resolved an issue that could cause RoCE communication failures between specific NIC pairs after repeated queue pair (QP) create/destroy cycles. In the affected cases, traffic could enter a bad state and fail with transport retry counter exceeded errors. This fix prevents the NIC from entering that condition.
	Keywords: RDMA; connection failure
	Detected in version: 40.46.3036
	Fixed in Release: 40.50.1002
5155445	Description: Resolved an issue with SyncE frequency adjustment overflow when using a third-party secondary vendor.
	Keywords: SyncE frequency
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5125089	Description: Resolved an issue where an incorrect Flex Parser configuration could trigger an assert. A syndrome was added to notify HWS when an invalid configuration is provided.
	Keywords: Flex Parser configuration
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5093695	Description: Resolved an issue where using mlxconfig to configure all modules with the 0xff module split value could reset the modules to the default configuration.
	Keywords: mlxconfig
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
5090639	Description: Resolved an issue in single-domain mode where an FLR on a GVMI did not clear the per-domain usage bits in `SP_core_diag_multi_domain->pcie_link_latency_max_read` and `SP_core_diag_multi_domain->pcie_link_latency_min_read`. As a result, when the next GVMI took ownership and attempted to configure these counters, it could receive `DIAG_DATA_LIST_WRITE_STATUS_NO_RESOURCES`.
	Keywords: PCIe Link Latency Arch counters
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002
5089348	Description: Resolved an issue where the MFT database was not updated to configure profile 12 for the Flex Parser.
	Keywords: MFT, Flex Parser
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
4950259	Description: Unblocked 4-port, single‑LAG configurations by updating the INI setting nv_config.global.power_cap.max_num_port=4, and fixed the Queue Affinity error seen in LAG mode.
	Keywords: LAG, Queue Affinity
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
5077326 / 4902171	Description: Resolved an issue where a slow module response to the NCSI GetModuleSerialData command could incorrectly trigger a dead IRISC watchdog assert. The firmware now handles this waiting state correctly, so slow module responses no longer cause a false watchdog hang indication.
	Keywords: NCSI
	Detected in version: 40.44.1036
	Fixed in Release: 40.50.1002
5025688	Description: Resolved an incorrect INI settings on one downstream link of CX8_Rithmax_CX8aas_BP_4P. All DSLs now operate in Gen6.
	Keywords: INI settings
	Detected in version: 40.48.1000
	Fixed in Release: 40.50.1002
4947006 / 4980956 / 4982511	Description: Resolved two rare race conditions in the PCC steering extension when the feature is enabled through path migration, empty RTT, or a custom header. Under rare timing conditions during QP creation or destruction, these issues could cause traffic disruption or resource leaks.
	Keywords: PCC steering extension
	Detected in version: 40.49.1014
	Fixed in Release: 40.50.1002

Last updated: September 02, 2026