PSOD with NMI IPI Timeout and Memory Controller Read Errors

Products

• VMware vSphere ESXi

Issue/Introduction

• Host experiences a Purple Screen of Death (PSOD). The panic occurs in directMapModifyMapAttr with an "NMI IPI Timeout" and an SMPWaitForPCPUs stall, indicating a deadlock during inter-processor synchronization.

Environment

• vSphere ESXi 8.x

Cause

This PSOD is a symptom of underlying hardware-level memory errors. These correctable memory errors trigger a high volume of SMIs, which degrade system performance. This slowness results in a deadlock during inter-processor synchronization (IPI), specifically causing physical CPUs (PCPUs) to stall while waiting for completion of the SMP pre-stop callback.

Resolution

• Validate Hardware: Engage the hardware vendor to investigate the memory controllers or physical DIMMs as identified. .
• Replace the faulty hardware components confirmed by the vendor diagnostics.

Additional Information

• Interpreting an ESXi host purple diagnostic screen (PSOD) 
• Contact Broadcom Support