VM Panic 'Failed to run VCPU: error: Bad parameter' after VM migration

Products

• VMware vSphere ESXi
• VMware Cloud Foundation

Issue/Introduction

• Following a virtual machine (VM) migration, the VM may register an ungraceful power-off without user intervention or Guest OS shutdown initiation.

• vmware.log for the VM reports the following entries:
  
  YYYY-MM-DD-INFO ... [esx@#### threadName="vcpu-0"] CPT: vmstartYYYY-MM-DD-CRITICAL... [esx@#### threadName="vcpu-#"] PANIC: Failed to run VCPU: error: Bad parameterYYYY-MM-DD-CRITICAL... [esx@#### threadName="vcpu-#"] Panic loop

• The corresponding /var/run/log/vmkernel.log on the target ESXi host reports a panic handler signal:
  
  YYYY-MM-DD -INFO vmkernel - [esx@####] cpu##:####)UserDump: 3158: vmx-vcpu-2:RCCVidyoRouter: Dumping cartel #### (from world #### ) to file .../vmx-zdump.000 ...
YYYY-MM-DD -INFO vmkernel - [esx@####] cpu##:####)User: ####: vmx-vmPanic: set panic handler to ##### with signal 10

Environment

VMware Cloud Foundation 9.x
VMware vSphere ESXi 9.x

Cause

Virtual Machine HWv8 supports XSAVE, the hypervisor needs to manage XSAVE state registers (like XCR0).
However, when handling checkpoints created under legacy migration mask logic, older code paths failed to initialize registers properly, causing the system panic.

Resolution

To resolve this issue, apply one of the following options:

1. Upgrade the VM's virtual hardware version to version 9 or higher.
   
   
2. Upgrade the ESXi host version to 9.1.1.

Note: This issue is addressed in VCF release 9.1.1.x, not in VCF release 9.1.0.x.

Additional Information

This issue typically manifests during or immediately following a vMotion/migration event where the host environment and VM virtual hardware version conflict regarding VCPU parameter handling.