ESXi Host Experiences PSOD with MCSUnlockIRQWork() messages

Products

• VMware vSphere ESXi

Issue/Introduction

ESXi 8.0 Update 3 host unexpectedly fails with a Purple Screen of Death (PSOD).

• The host may reboot automatically or via manual intervention (e.g., iDRAC).
• Virtual Machines (VMs) on the affected host are restarted on other hosts via VMware HA.
• The panic message typically includes: @BlueScreen: NMI IPI: Panic requested by another PCPU.

The backtrace often shows the following sequence:

NMI_Interrupt@vmkernel#nover+0x516 stack: 0x0, 0x0, 0x0, 0x0, 0x0
IDTNMIWork@vmkernel#nover+0x95 stack: 0x0, 0x42001da#####, 0x0, 0x42001da#####, 0x103
Int2_NMI@vmkernel#nover+0x9 stack: 0x103, 0x0, 0x0, 0x0, 0x45dc2b8#####
gate_entry@vmkernel#nover+0xa7 stack: 0x0, 0x0, 0x45dc2b8#####, 0x2b8#####, 0x0
MCSUnlockIRQWork@vmkernel#nover+0x26 stack: 0x453bf86#####, 0x430cf96#####, 0x42001eb#####, 0x430486e#####, 0x1d8#####
EventQ_WakeupCount@vmkernel#nover+0x24 stack: 0x430cf96#####, 0x42001eb#####, 0x430486e#####, 0x1d8#####, 0x45dc44a#####
Fil6_NoTxnIOSM@esx#nover+0x9b stack: 0x45dc44a#####, 0x42001ebf3248, 0x135, 0x1, 0x45dc44a11300
Fil6_IOCompleteNoTxn@esx#nover+0xb8 stack: 0x45dc44a#####, 0x45dc006#####, 0x1, 0x42001da#####, 0x0
AsyncPopCallbackFrameInt@vmkernel#nover+0xe3 stack: 0x45dc0b1#####, 0x45dc44a#####, 0x0, 0x42001d8#####, 0xc19704000#####

Environment

ESXi 8.x

Cause

Further analysis reveals that a thread on a PCPU entered an infinite loop within the MCSUnlockIRQWork() function

 

 

Resolution

This issue is resolved in ESXi 8.0 Update 3i (Patch 08) and later versions.