ESXi 8.0.3 PSOD with GP Exception 13 in bnxtnet_page_alloc_and_map network polling

Products

• VMware vSphere ESXi

Issue/Introduction

• An ESXi host crashes with a Purple Screen of Death (PSOD). The crash terminates with a General Protection Fault (Exception 13) during a network polling operation on a vmnic backed by the bnxtnet driver.
• The vmkernel-zdump core dump analysis reveals the following fault string and stack trace:

  @BlueScreen: #GP Exception 13 in world 2098198:vmnic4-pollW @ 0x42002cd36971
  
  Backtrace:
  FastSlab_AllocWithTimeout
  vmk_PageSlabAlloc
  PktPageAlloc_AllocPages
  vmk_PktAllocPage
  bnxtnet_page_alloc_and_map
  bnxtnet_poll_work
  bnxtnet_poll
  NetPoll_PollPkts
  NetPollWorldCallback
  CpuSched_StartWorld

• Host details extracted from the core dump indicate the host crashed while attempting to allocate memory pages for incoming packets.

Environment

• VMware ESXi 8.0.3
• Broadcom NetXtreme NIC Driver (bnxtnet) version 233.0.156.0-10EM.800.1.0.20613240

Cause

The PSOD is caused by hardware-level memory corruption stemming from a rogue Direct Memory Access (DMA) transfer.

The physical network adapter overwrites the ESXi kernel's network packet memory allocator structures (netPktPageCache-AnyMem fastslab) instead of designated RX ring buffers due to a driver and physical adapter firmware mismatch.

This can also be caused by hardware failure of the nic card, if the driver and firmware versions are verified to be strictly VCG-compliant.

Resolution

• Validate the installed firmware on the physical Broadcom NetXtreme adapter against the VMware Compatibility Guide (VCG) for the bnxtnet driver version 233.0.156.0-10EM.800.1.0.20613240.

• Upgrade or downgrade the physical adapter firmware to match the exact certified version listed in the VCG to ensure correct DMA operations and prevent memory allocator corruption.

• Engage the hardware vendor

Additional Information

ESXi host PSOD with UD Exception 6 in bnxtnet

GP Exception 13 in world XXXXXXX:helperUpq

COMPATIBILITY GUIDE

FAQ: Recommendation for drivers/firmware for ESXi hosts