ESXi host PSOD in CpuSchedUnitsToBaseShares due to invalid memory access

Products

• VMware vSphere ESXi

Issue/Introduction

• An ESXi host may experience a Purple Screen of Death (PSOD) when a thread loads an incorrect value from schedTree.nodeRoot, leading to an invalid memory access. This discrepancy typically indicates an underlying platform hardware issue, such as a malfunctioning processor.

• The backtrace of the crash looks like below.

  0x45398e51bf50:[0x42000946586d]CpuSchedSnapshotNodes@vmkernel#nover+0x2ed stack: 0x420001889a5b
0x45398e51bfb0:[0x420009468265]CpuSched_Reallocate@vmkernel#nover+0x4e stack: 0x45398e71f 100
0x45398e51bfd0:[0x4200096df3ba]CpuSchedReallocLoop@vmkernel#nover+0xb stack: 0x0
0x45398e51bfe0:[0x4200096dc88e]CpuSched_StartWor ld@vmkernel#nover+0xbf stack: 0x0
0x45398e51c000:[0x420009144faf ]Debug_IsInitialized@vmkernel#nover+0xc stack: 0x0

Environment

VMware vSphere ESXi

Cause

A thread loaded an incorrect value from schedTree.nodeRoot, leading to an invalid memory access. This is a platform hardware issue, typically involving the physical CPU or memory controller.

Resolution

1. Analyze the backtrace: Review the PSOD log for the specific backtrace: 0x45398e51bf50:[0x42000946586d]CpuSchedSnapshotNodes@vmkernel#nover+0x2ed stack: 0x420001889a5b 0x45398e51bfb0:[0x4200096df3ba]CpuSchedReallocLoop@vmkernel#nover+0xb stack: 0x0
2. Update Firmware/BIOS: Update the server BIOS and firmware to the latest available versions provided by your hardware vendor.
   • For instructions on downloading patches, see Download Broadcom Products and Software.
3. Contact Hardware Vendor: If the issue persists after BIOS/firmware updates, contact your hardware vendor to investigate potential CPU or platform hardware failure.
4. Support Contact: If you require further assistance from Broadcom, see Contact Broadcom Support.

Additional Information

For information on how to retrieve logs, see Collecting Diagnostic Information.