BlueField-3 DPU (B3220) — ARM & BMC Unresponsive

Infrastructure & Networking BlueField
1

Hi, I am experiencing an Arm boot failure after configuring a BlueField-3 DPU (B3220 FHHL) with mlxconfig -d <device> s PCI_BUS00_HIERARCHY_TYPE=2. When I tried to restore the ARM using direct SCP to transfer the .bfb to the BMC rshim, I encountered an unexpected termination, and now the BMC won’t boot either.

Cause of problem:

  • After reviewing the NVIDIA documentation, I discovered that my particular SKU does not support the self-hosted/root-complex operations I was attempting.

Current symptoms:

  • Device is stuck with both BMC and Arm unresponsive;
  • Power cycles and reseating did not recover the card.

Similarity to a known case:

Requests:

  • Provide recovery guidance for the current unresponsive state.

Environment:

  • Hardware: BlueField-3 DPU B3220 Single-Slot FHHL.
  • Current status: BMC + Arm unresponsive.

Thank you for your assistance.

2

Hi,

Thanks for your question.

You can try resetting the firmware configuration from the host side:

i.e.

mlxconfig -d reset

Then perform a power cycle of the host.

If it’s still doesn’t work, please open a support case by sending an email to enterprisesupport@nvidia.com, and it will be handled according to the entitlement.

Best Regards,

Anatoly