NVIDIA BlueField BSP v4.9.3 LTS

Known Issues

Ref #Issue
4445755Description: NIC firmware does not automatically update after BFB installation despite WITH_NIC_FW_UPDATE being enabled.
Workaround: Manually upgrade firmware using mlxfwmanager post-install.
Keywords: Firmware; NIC mode
Reported in version: 4.9.3-LTSU3
4515963 Description: After installing a non-Ubuntu BFB, Arm boot may hang at the grub shell.
Workaround: Reinstall the BFB.
Keywords: BFB; grub
Reported in version: 4.9.3-LTSU3
4513216 Description: BMC and CEC firmware binaries are not found during LFWP NIC firmware update run through bfb-install --lfwp/doca-installer --lfwp or OobUpdate.sh --lfwp when used with bf-fwbundle .

Workaround: Create or append the following lines to the existing bf.cfg:

Copy
Copied!
            

BMC_PATH=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/bmc BMC_IMAGE=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/bmc/bf3-bmc-fw.fwpkg CEC_PATH=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/cec CEC_IMAGE=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/cec/bf3-cec-fw.fwpkg DPU_GI_PATH=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/bmc DPU_GOLDEN_IMAGE=/root/bfb/upgrade/initramfs/usr/lib/firmware/mellanox/bmc/bf3-bmc-preboot-install.bfb

Use this bf.cfg file with bfb-install/doca-installer or OobUpdate.

Keywords: LFWP; BMC
Reported in version: 4.9.3-LTSU3
4519591Description: In NIC mode, after Arm reboot, the BMC firmware and CEC versions are intermittently missing in the host query output.
Workaround: N/A
Keywords: Host; firmware
Reported in version: 4.9.3-LTSU3
4487794Description: On BlueField-3 systems, upgrading from BFB 2.9.x to 3.0.0 may trigger a kernel crash during boot, with a paging request failure (Unable to handle kernel paging request).
Workaround: N/A
Keywords: Kernel crash; BFB upgrade; BlueField-3
Reported in version: 4.9.3-LTSU3
4342994Description: At times, BlueField may boot into grub shell after BFB installation.
Workaround: Power-cycle the BlueField and reinstall the BFB.
Keywords: BFB installation; grub shell
Reported in version: 4.9.3-LTSU3
4479477Description: On BlueField-3, when PXE boot is configured with a timeout via IPMI and the DPU is immediately power-reset, the system may fail to boot via PXE.
Workaround: N/A
Keywords: PXE boot; bootdev timeout; BlueField-3
Reported in version: 4.9.3-LTSU3
4504132Description: When the error message mlxbf_gige MLNXBF17:00: Error getting PHY irq. Use polling instead is reported, the GIGE driver is polling link status and is still fully functional. Please ignore the message.
Workaround: N/A
Keyword: mlxbf_gige
Reported in version: 4.9.2-LTSU2
4122429Description: SMMU was set to enabled by default instead of disabled.
Workaround: Disable SMMU
Keyword: SMMU, default
Reported in version: 4.9.2-LTSU2
4284756Description: Boot option re-enabled by UEFI after power reset of DPU.

Workaround:

1. Set Boot option enable attribute to false.

2. Power reset DPU.

3. Do not reboot or power reset DPU for a second time. If reboot/power reset DPU does occur, the boot option enable attribute will be recovered to true on UEFI generated Boot Options, but this issue does not affect legacy boot options.

4. Completely disable boot options getting reset by UEFI by using `SYS_DISABLE_AUTO_BOOT_REFRESH=TRUE` configuration in bf.cfg.

Keyword: Boot, power reset, UEFI
Reported in version: 4.9.2
3969477Description: At times, BlueField may boot into grub shell after BFB installation.

Workaround:

  1. Power-cycle BlueField.
  2. Repeat push BFB.
Keyword: boot, grub shell, installation, bfb
Reported in version: 4.9.1
4049034Description: On openEuler 22.03 SP3 and openEuler 20.03 SP1, it is not possible to do yum update after BFB installation.

Workaround: Attempt the following:

  1. To update DOCA only, disable ALL repos:

    Copy
    Copied!
                

    dnf --disablerepo='*' --enablerepo='kubernetes,doca' -y update

  2. To update OS components, exclude rdma-core using one of the following options:

    • Option 1 – Disable rdma-corem from dnf.conf using excludepkgs=rdma-core*oe2203sp3*
    • Option 2 – Disable rdma-core from openEuler.repo file under OS and everything using exclude=rdma-core*
    • Option 3 – yum update -x rdma-core
Keyword: openEuler
Reported in version: 4.9.0
4119852Description: Redfish communication may be skipped upon reboot immediately following the installation of DOCA BFB-bundle.
Workaround: Reboot (SW_RESET) the BlueField Arm to fix Redfish communication.
Keyword: RF; installation; connection
Reported in version: 4.9.0
4098782Description: BlueField does not support the fru print format because it uses ipmitool version 1.8.18.
Workaround: Read the information from the host using ipmitool version 1.8.19.
Keyword: IPMI
Reported in version: 4.9.0
4129715Description: Compiling Rocky 9.2 may fail when using GCC with the native arch flag.
Workaround: Upgrade to toolset 13 (gcc 13).
Keyword: Linux; GCC
Reported in version: 4.9.0
3859084Description: Running bfb-install in remote mode may fail if the remote server lacks an up-to-date nc software that supports TCP server functionality. This issue is commonly observed with servers such as BlueField BMC.
Workaround: N/A
Keyword: bfb-install; TCP
Reported in version: 4.8.0
3876636Description: RDMA devices are missing after installing BFB in NIC mode.
Workaround: An additional BlueField reboot is required
Keyword: RDMA; reboot; NIC mode
Reported in version: 4.8.0
3960613Description: When configuring NVMe-oF target offload on Arm, harmless CMA allocation errors appear at the kernel log.

Workaround: Increase the CMA limit or cancel its use using the kernel's CMD line parameters:

  • Add the parameter cma=256M to increase the CMA limit to 256MB; or
  • Add the parameter cma=0 to disable the use of CMA
Keyword: NVMe-oF target offload; CMA
Reported in version: 4.8.0
3969477Description: At times, BlueField may boot into grub shell after BFB installation.
Workaround: Power-cycle the BlueField.
Keyword: BFB installation; grub shell
Reported in version: 4.8.0
3837649Description: Possible stack trace in mlxbf_i2c can occur intermittently when booting BFB.
Workaround: Reboot the BlueField.
Keyword: Boot
Reported in version: 4.8.0
3914629Description: When "Force PXE" is set right before installing a BFB image via BMC RShim, BlueField PXE boot fails to boot from NET-OOB-IPV4.
Workaround: N/A
Keyword: PXE boot
Reported in version: 4.7.0
3239320Description: Resetting hugepage size to 0 on Rocky Linux 8.6 using the sysctl tool fails.

Workaround: Use the following command instead:

Copy
Copied!
            

echo 0 > /sys/kernel/mm/hugepages/hugepages-<Size>/nr_hugepages

Keyword: Hugepage; sysctl
Reported in version: 4.7.0
3859113Description: Reloading MLNX_OFED drivers with the command /etc/init.d/openibd restart fails when the NVMe driver is installed and in use.
Workaround: Reboot the machine to load all the MLNX_OFED drivers.
Keyword: NVMe; driver
Reported in version: 4.7.0
3748649Description: With the numbering of CPUs in an 8-core configuration, the kernel is expected to assign virtual CPU ID numbers from 0-7, where N is the number of cores enabled. With CTyunOS, however, the numbering is unexpected.
Workaround: N/A
Keyword: CTyunOS; CPU numbering
Reported in version: 4.7.0
3844066

Description: On CentOS 7.6 with kernel 4.19, bringing up OVS bridge interface causes call traces:

Copy
Copied!
            

WARNING: CPU: 5 PID: 14339 at kernel/rcu/tree_plugin.h:342 rcu_note_context_switch+0x48/0x538

Workaround: Do not bring UP OVS bridge interfaces.
Keyword: CentOS; kernel; rcu_note_context_switch
Reported in version: 4.7.0
3844705Description: In OpenEuler 20.03, the Linux Kernel version 4.19.90 is affected by an issue that impacts the discard/trim functionality for the DPU eMMC device which may cause degraded performance of the DPU eMMC over time.
Workaround: Upgrade to Linux Kernel version 5.10 or later.
Keyword: eMMC discard; trim functionality
Reported in version: 4.7.0
3855702Description: Trying to jump from a steering level in the hardware to a lower level using software steering is not supported on rdma-core lower than 48.x.
Workaround: N/A
Keyword: RDMA; SWS
Reported in version: 4.7.0
3743879Description: mlxfwreset could timeout on servers where the RShim driver is running and INTx is not supported. The following error message is printed: BF reset flow encountered a failure due to a reset state error of negotiation timeout.

Workaround: Set PCIE_HAS_VFIO=0 and PCIE_HAS_UIO=0 in /etc/rshim.conf and restart the RShim driver. Then re-run the mlxfwreset command.

If host Linux kernel lockdown is enabled, then manually unbind the RShim driver before mlxfwreset and bind it back after mlxfwreset:

Copy
Copied!
            

echo "DROP_MODE 1" > /dev/rshim0/misc mlxfwreset <arguments> echo "DROP_MODE 0" > /dev/rshim0/misc

Keyword: Timeout; mlxfwreset; INTx
Reported in version: 4.7.0
3746866Description: The error message IANA PEN registry open failed: No such file or directory may appear when using ipmitool version 1.8.19-7. This message can be safely ignored.
Workaround: N/A
Keywords: IPMI; Debian
Reported in version: 4.6.0
3755143

Description: UEFI synchronous exception is observed at address 0x479B7xxxx where the UEFI module names are not printed. See the following example:

Copy
Copied!
            

ERR[UEFI]: PC=0x479B78480(B4000040 3900001F A94153F3 F94013F5) ERR[UEFI]: PC=0x479B78480 ERR[UEFI]: PC=0x479B7E684 ERR[UEFI]: PC=0x47A0E93F4 ERR[UEFI]: PC=0x47A0E9608

Workaround: Run software reset or reinstall the BFB.
Keywords: UEFI synchronous exception
Reported in version: 4.6.0
3767580

Description: On Debian 12, the first boot after BFB installation may fail with the following kernel panic:

Copy
Copied!
            

[ end Kernel panic - not syncing: Attempted to kill init! exitcode=0x00000100 ]

Workaround: Reset the DPU using the RShim interface:

Copy
Copied!
            

echo "SW_RESET 1" > /dev/rshim0/misc

Keywords: Debian 12; Kernel panic; kill init
Reported in version: 4.6.0
3678069Description: If using DPUs with NVMe and mmcbld and configured to boot from mmcblk, users must create a bf.cfg file with device=/dev/mmcblk0 before installing the *.bfb.
Workaround: N/A
Keyword: NVMe
Reported in version: 4.6.0
3747285Description: The ipmitool command to force PXE in BMC modifies both the IPMI and Redfish request settings. When Redfish is enabled in UEFI, Redfish takes priority, so all PXE boot entries are attempted and before regular boot continues.
Workaround: Redfish must be disabled if IPMI force PXE retry behavior is expected.
Keyword: PXE; retry; fail
Reported in version: 4.6.0
3666574Description: Running systemctl restart openibd on the DPU can result in openvswitch service crash.
Workaround: Run /etc/init.d/openvswitch-switch start.
Keyword: OVS fail; openibd
Reported in version: 4.6.0
3880194Description: mlxbf-bootctl command failed to install default.bfb.

Workaround: The following are possible options –

  • Boot the BFB file doca_2.5.0_bsp_4.5.0_ubuntu_22.04-1.23-10.prod.bfb to update the platform software
  • Download, compile, and install latest mlxbf-bootctl command from GitHub
  • Edit default.bfb by using the mlx-mkbfb command to incorporate the platform-specific images and filtering out unused images. Example for a BlueField-2 device:

    Copy
    Copied!
                

    $ mlx-mkbfb -x default.bfb $ mlx-mkbfb \ --bl2r-v1=dump-bl2r-v1 \ --bl2r-cert-v1=dump-bl2r-cert-v1 \ --bl2-v1=dump-bl2-v1 \ --bl2-cert-v1=dump-bl2-cert-v1 \ --bl31-v1=dump-bl31-v1 \ --bl31-cert-v1=dump-bl31-cert-v1 \ --bl31-key-cert-v1=dump-bl31-key-cert-v1 \ --bl33-v0=dump-bl33-v0 \ --bl33-cert-v1=dump-bl33-cert-v1 \ --bl33-key-cert-v1=dump-bl33-key-cert-v1 \ --boot-acpi-v0=dump-boot-acpi-v0 \ --boot-args-v0=dump-boot-args-v0 \ --boot-desc-v0=dump-boot-desc-v0 \ --boot-path-v0=dump-boot-path-v0 \ --ddr_ini-v1=dump-ddr_ini-v1 \ --ddr-cert-v1=dump-ddr-cert-v1 \ --ddr_ate_imem-v1=dump-ddr_ate_imem-v1 \ --ddr_ate_dmem-v1=dump-ddr_ate_dmem-v1 \ --snps_images-v1=dump-snps_images-v1 \ --trusted-key-cert-v1=dump-trusted-key-cert-v1 \ default_min.bfb

Keywords: Software; upgrade
Discovered in version: 4.5.1
3204153Description: On BlueField-2, the OOB may not get an IP address due to the interface being down.
Workaround: restart auto-negotiation using the command ethtool -r oob_net0.
Keyword: OOB; IP
Reported in version: 4.5.0
3601491

Description: Symmetric pause must be enabled in the DHCP server for the OOB to be able to reliably get an IP address assigned.

Workaround: N/A
Keyword: OOB; IP
Reported in version: 4.5.0
3673330

Description: On Debian 12, Arm ports remain in Legacy mode after multiple Arm reboot iterations. The following error message appears in /var/log/syslog:

Copy
Copied!
            

mlnx_bf_configure[2601]: ERR: Failed to configure switchdev mode for 0000:03:00.0 after 61 retries

Workaround: Run:

Copy
Copied!
            

$ echo SET_MODE_RETRY_NUM=300 >> /etc/mellanox/mlnx-bf.conf $ reboot

Keyword: Debian; Arm
Reported in version: 4.5.0
3695543Description: PXE boot may fail after a firmware upgrade from 32.36.xxxx, 32.37.xxxx, to 32.38.xxxx and above.

Workaround: Create /etc/bf.cfg with the following lines, then run bfcfg to recreate the PXE boot entries:

Copy
Copied!
            

BOOT0=DISK BOOT1=NET-NIC_P0-IPV4 BOOT2=NET-NIC_P0-IPV6 BOOT3=NET-NIC_P1-IPV4 BOOT4=NET-NIC_P1-IPV6 BOOT5=NET-OOB-IPV4 BOOT6=NET-OOB-IPV6

Keyword: MAC allocation; PXE boot
Reported in version: 4.5.0
3647476Description: Debian 12 OS does not support CT tunnel offload.
Workaround: Recompile the kernel with CONFIG_NET_TC_SKB_EXT set.
Keyword: Connection tracking; Linux
Reported in version: 4.5.0
3007696Description: When configuring a static IP address for tmfifo_net0 interface in /etc/network/interfaces, the IP address is lost after restarting the RShim driver on Debian Linux.

Workaround: Use netplan configuration. For example

Copy
Copied!
            

# cat /etc/netplan/tmfifo_net0.yaml network: version: 2 renderer: networkd ethernets: tmfifo_net0: addresses: - 192.168.100.1/30 dhcp4: false

Then run "netplan apply".

Keyword: IP address; tmfifo_net0; host
Reported in version: 4.5.0
3670628Description: When NIC subsystem is in recovery mode, the interface towards to NVMe is not accessible. Thus, the SSD boot device would not be available.
Workaround: The admin must configure the Arm subsystem boot device to boot from the eMMC, for example.
Keyword: mlxfwreset; RShim
Reported in version: 4.5.0
3702393Description: On rare occasions, the boot process part of SWRESET (via RShim) or FWRESET (via mlxfwreset) may result in a device hanging on the boot flow or cause the host server to reboot.
Workaround: Perform graceful shutdown and then a power cycle.
Keyword: mlxfwreset; RShim
Reported in version: 4.5.0
3665724Description: If the UEFI password is an empty string (""), then it cannot be changed via Redfish.
Workaround: UEFI; password; Redfish
Keyword: UEFI; password; Redfish
Reported in version: 4.5.0
3671185Description: XFRM rules must be deleted before driver restart or warm reboot are performed.
Workaround: N/A
Keyword: IPsec
Reported in version: 4.5.0
3666160Description: Installing BFB using bfb-install when mlxconfig PF_TOTAL_SF>1700, triggers server reboot immediately.
Workaround: Change PF_TOTAL_SF to 0, perform graceful shutdown, then power cycle, and then install the BFB.
Keyword: SF; PF_TOTAL_SF; BFB installation
Reported in version: 4.2.2
3605254Description: Following a system power cycle, both the DPU and BMC boot independently which may lead to the DPU's UEFI boot process to complete before the BMC's. As a result, when attempting to establish Redfish communication, the BMC may not yet be prepared to respond.
Workaround: Wait until the BMC is done booting before issuing a reset command to the DPU.
Keyword: Power cycle; Redfish; boot
Reported in version: 4.2.1
3602044Description: When the public key is deleted while Redfish is enabled, UEFI secure boot is disabled and UEFI reverts to Setup Mode (i.e., the SecureBootEnable Redfish property is reset to false). If later, the public key is re-enrolled, the platform does not implement UEFI secure boot until the SecureBootEnable Redfish property is explicitly changed to true.
Workaround: Set SecureBootEnable to true using the Redfish API.
Keyword: Redfish; UEFI secure boot
Reported in version: 4.2.1
3592080Description: When using UEK8 on the host in DPU mode, creating a VF on the host consumes about 100MB memory on the DPU.
Workaround: N/A
Keyword: UEK; VF
Reported in version: 4.2.1
3568341Description: Downgrading BSP software from 4.2.0 fails if UEFI secure boot is enabled.
Workaround: Disable UEFI secure boot before downgrading.
Keyword: Software; downgrade
Reported in version: 4.2.0
3566042Description: Virtio hotplug is not supported in GPU-HOST mode on the NVIDIA Converged Accelerator.
Workaround: N/A
Keyword: Virtio; Converged Accelerator
Reported in version: 4.2.0
3546474Description: PXE boot over ConnectX interface might not work due to an invalid MAC address in the UEFI boot entry.
Workaround: On the DPU, create /etc/bf.cfg file with the relevant PXE boot entries, then run the command bfcfg.
Keyword: PXE; boot; MAC
Reported in version: 4.2.0
3546202

Description: After rebooting a BlueField-3 DPU running Rocky Linux 8.6 BFB, the kernel log shows the following error:

Copy
Copied!
            

[ 3.787135] mlxbf_gige MLNXBF17:00: Error getting PHY irq. Use polling instead

This message indicates that the Ethernet driver will function normally in all aspects, except that PHY polling is enabled.

Workaround: N/A
Keyword: Linux; PHY; kernel
Reported in version: 4.2.0
3306489Description: When performing longevity tests (e.g., mlxfwreset, DPU reboot, burning of new BFBs), a host running an Intel CPU may observer errors related to "CPU 0: Machine Check Exception".
Workaround: Add intel_idle.max_cstate=1 entry to the kernel command line.
Keyword: Longevity; mlxfwreset; DPU reboot
Reported in version: 4.2.0
3538486Description: When removing LAG configuration from the DPU, a kernel warning for uverbs_destroy_ufile_hw is observed if virtio-net-controller is still running.
Workaround: Stop virtio-net-controller service before cleaning up bond configuration.
Keyword: Virtio-net; LAG
Reported in version: 4.2.0
3462630

Description: When trying to perform a PXE installation when UEFI Secure Boot is enabled, the following error messages may be observed:

Copy
Copied!
            

error: shim_lock protocol not found. error: you need to load the kernel first.

Workaround: Download a Grub EFI binary from the Ubuntu website. For further information on Ubuntu UEFI Secure Boot PXE Boot, please visit Ubuntu's official website.
Keyword: PXE; UEFI Secure Boot
Reported in version: 4.0.2
3412847Description: Socket-Direct is currently not supported on BlueField-3 devices.
Workaround: N/A
Keyword: Socket-Direct; support
Reported in version: 4.0.2
3448841Description: While running CentOS 8.2, switchdev Ethernet DPU runs in "shared" RDMA net namespace mode instead of "exclusive".

Workaround: Use ib_core module parameter netns_mode=0. For example:

Copy
Copied!
            

echo "options ib_core netns_mode=0" >> /etc/modprobe.d/mlnx-bf.conf

Keywords: RDMA; isolation; Net NS
Reported in version: 4.0.2
3413938Description: Using mlnx-sf script, creating and deleting an SF with same ID number in a stressful manner may cause the setup to hang due to a race between create and delete commands.
Workaround: N/A
Keywords: Hang; mlnx-sf
Reported in version: 4.0.2
3452740Description: Ovs-pki is not working due to two versions of OpenSSL being installed, causing the PKA engine to not load properly.
Workaround: N/A
Keywords: PKA; OpenSSL
Reported in version: 4.0.2
3273435Description: Changing the mode of operation between NIC and DPU modes results in different capabilities for the host driver which might cause unexpected behavior.
Workaround: Reload the host driver or reboot the host.
Keywords: Modes of operation; driver
Reported in version: 4.0.2
2706803Description: When an NVMe controller, SoC management controller, and DMA controller are configured, the maximum number of VFs is limited to 124.
Workaround: N/A
Keywords: VF; limitation
Reported in version: 4.0.2
3264224Description: When trying to change boot order using efibootmgr, BlueField fails to attempt PXE boot from p0 even though efibootmgr returns a successful result.
Workaround: Drop into the UEFI menu and regenerate all the EFI entries.
Keywords: PXE; efibootmgr
Reported in version: 3.9.3.1
3188415Description: An Arm firmware update to the same version that is installed will fail and is not supported.
Workaround: N/A
Keywords: Arm; firmware; update
Reported in version: 3.9.2
N/ADescription: The BootOptionEnabled attribute changes back to true after DPU-force reset.
Workaround: N/A
Keywords: Redfish; BootOptionEnabled
Reported in version: 3.9.2
3012182 Description: The command ethtool -I --show-fec is not supported by the DPU with kernel 5.4.
Workaround: N/A
Keywords: Kernel; show-fec
Reported in version: 3.9.0
2855986Description: After disabling SR-IOV VF on a virtio device, removing virtio-net/PCIe driver from guest OS may render the virtio controller unusable .
Workaround: Restart the virtio-net controller to recover it. To avoid this issue, m onitor the log from controller and make sure VF resources are destroyed before unloading virtio-net/PCIe drivers.
Keywords: Virtio-net; VF
Reported in version: 3.9.0
2863456Description: SA limit by packet count (hard and soft) are supported only on traffic originated from the ECPF. Trying to configure them on VF traffic removes the SA when hard limit is hit. However, traffic could still pass as plain text due to the tunnel offload used in such configuration.
Workaround: N/A
Keywords: ASAP2; IPsec Full Offload
Reported in version: 3.9.0
2982184Description: When multiple BlueField resets are issued within 10 seconds of each other, EEPROM error messages are displayed on the console and, as a result, the BlueField may not boot from the eMMC and may halt at the UEFI menu.
Workaround: Power-cycle the BlueField to fix the EEPROM issue. Manual recovery of the boot options and/or SW installation may be needed.
Keywords: Reset; EEPROM
Reported in version: 3.9.0
2853408Description: Some pre-OS environments may fail when sensing a hot plug operation during their boot stage.
Workaround: Run "mlxconfig -d <mst dev> set PF_LOG_BAR_SIZE=0".
Keywords: BIOS; hot-plug; Virtio-net
Reported in version: 3.9.0
2934833Description: Running I/O traffic and toggling both physical ports status in a stressful manner on the receiving-end machine may cause traffic loss.
Workaround: N/A
Keywords: MLNX_OFED; RDMA; port toggle
Reported in version: 3.8.5
2911425Description: ProLiant DL385 Gen10 Plus server with BIOS version 1.3 hangs when large number of SFs (PF_TOTAL_SF=252) are configured.
Workaround: Update the BIOS version to 2.4 which should correctly detect the PCIe device with the bigger BAR size.
Keywords: Scalable functions; BIOS
Reported in version: 3.8.5
N/ADescription: Only QP queues are supported for GGA accelerators from this version onward.
Workaround: N/A
Keywords: Firmware; SQ; QP
Reported in version: 3.8.0
2846108 Description: Setting VHCA_TRUST_LEVEL does not work when there are active SFs or VFs.
Workaround: N/A
Keywords: Firmware; SF; VF
Reported in version: 3.8.0
2750499Description: Some devlink commands are only supported by mlnx devlink (/opt/mellanox/iproute2/sbin/devlink). The default devlink from the OS may produce failure (e.g., devlink port show -j).
Workaround: N/A
Keywords: Devlink
Reported in version: 3.7.1
2730157Description: Kernel upgrade is not currently supported on BlueField as there are out of tree kernel modules (e.g., ConnectX drivers that will stop working after kernel upgrade).
Workaround: Kernel can be upgraded if there is a matching DOCA repository that includes all the drivers compiled with the new kernel or as a part of the new BFB package.
Keywords: Kernel; upgrade
Reported in version: 3.7.0
2706710Description: Call traces are seen on the host when recreating VFs before the controller side finishes the deletion procedure.
Workaround: N/A
Keywords: Virtio-net controller
Reported in version: 3.7.0
2685478Description: 3rd party (netkvm.sys) Virtio-net drivers for Windows do not support SR-IOV.
Workaround: N/A
Keywords: Virtio-net; SR-IOV; WinOF-2
Reported in version: 3.7.0
2684501Description: Once the contiguous memory pool, a limited resource, is exhausted, fallback allocation to other methods occurs. This process triggers cma_alloc failures in the dmesg log.
Workaround: N/A
Keywords: Log; cma_alloc; memory
Reported in version: 3.7.0
2590016Description: ibdev2netdev tool is not supported for PCIe PF operating in switchdev mode or on SFs.
Workaround: N/A
Keywords: ibdev2netdev
Reported in version: 3.6.0.11699
2590016Description: A "double free" error is seen when using the "curl" utility. This error is from libcrypto.so library which is part of the OpenSSL package. This happens only when OpenSSL is configured to use a dynamic engine (e.g. Bluefield PKA engine).

Workaround: Set OPENSSL_CONF=/etc/ssl/openssl.cnf.orig before using the curl utility.

For example:

Copy
Copied!
            

# OPENSSL_CONF=/etc/ssl/openssl.cnf.orig curl -O https://tpo.pe/pathogen.vim

Note

OPENSSL_CONF is aimed at using a custom config file for applications. In this case, it is used to point to a config file where dynamic engine (PKA engine) is not enabled.

Keywords: OpenSSL; curl
Reported in version: 3.6.0.11699
2407897

Description: The host may crash when the number of PCIe devices overflows the PCIe device address. According to the PCIe spec, the device address space is 8 bits in total—device (5 bits) and function (3 bits)—which means that the total number of devices cannot be more than 256.

The second PF maximum number of VFs is limited by the total number of additional PCIe devices that precedes it. By default, the preceding PCIe devices are 2 PFs + RShim DMA + 127 VFs of the first PF. This means that the maximum valid number of VFs for the second port will be 126.

Workaround: Use the maximum allowed VFs on the 2nd PCIe PF of BlueField instead of the maximum of 127 VFs.
Keywords: Emulated devices; VirtIO-net; VirtIO-blk; VFs; RShim
Reported in version: 3.6.0.11699
2445289Description: If secure boot is enabled, MFT cannot be installed on the BlueField DPU independently from BlueField drivers (MLNX_OFED).
Workaround: N/A
Keywords: MFT; secure boot
Reported in version: 3.5.1.11601
2377021Description: Executing sudo poweroff on the Arm side causes the system to hang.
Workaround: Perform graceful shutdown, then reboot your BlueField device or power cycle the server.
Keywords: Hang; reboot
Reported in version: 3.5.0.11563
2350132Description: Boot process hangs at BIOS (version 1.2.11) stage when power cycling a server (model Dell PowerEdge R7525) after configuring "PCI_SWITCH_EMULATION_NUM_PORT" > 27​​​​​​​.
Workaround: N/A
Keywords: Server; hang; power cycle
Reported in version: 3.5.0.11563
2581408Description: On a BlueField device operating in Embedded CPU mode, PXE driver will fail to boot if the Arm side is not fully loaded and the OVS bridge is not configured.
Workaround: Run warm reboot on the host side and boot again via the device when Arm is up and the OVS bridge is configured.
Keywords: Embedded CPU; PXE; UEFI; Arm
Reported in version: 2.5.0.11176
1859322Description: On some setups, DPU does not power on following server cold boot when UART cable is attached to the same server.
Workaround: As long as the RShim driver is loaded on the server and the RShim interface is visible, the RShim driver will detect this and auto-reset the card into normal state.
Keywords: DPU; Arm; Cold Boot
Reported in version: 2.4.0.11082
1899921Description: Driver restart fails when SNAP service is running.
Workaround: Stop the SNAP services nvme_sf and nvme_snap@nvme0, then restart the driver. After the driver loads restart the services.
Keywords: SNAP
Reported in version: 2.2.0.11000
1911618Description: Defining namespaces with certain Micron disks (Micron_9300_MTFDHAL3T8TDP) using consecutive attach-ns commands can cause errors.
Workaround: Add delay between attach-ns commands.
Keywords: Micron; disk; namespace; attach-ns
Reported in version: 2.2.0.11000

© Copyright 2025, NVIDIA. Last updated on Aug 4, 2025.