VMs enter reboot loop after upgrade to VCF 9.1 due to legacy CPUID masks - VMware vCenter Server

Products

• VMware vSphere ESXi

Issue/Introduction

After upgrading to VCF 9.1 / ESXi 9.1, virtual machines experience continuous reboots or Blue Screen of Death (BSOD) errors immediately upon power-on.

Environment

VCF 9.1

ESX 9.1

 

Cause

VCF 9.1 introduces stricter enforcement of CPU features via the User Level Monitor (ULM). Virtual machines that contain legacy configuration lines in their .vmx file—specifically CPUID or RDTSCP masks (e.g., featMask.vm.cpuid.RDTSCP)—may fail to boot if the Guest Operating System expects those specific features that are now being masked or incorrectly handled during the upgrade. These masks were often legacy workarounds from older versions (vSphere 5.x/6.x) that were never cleaned up.

Resolution

To resolve this issue, the legacy masking lines must be removed from the affected virtual machine configuration:

1. Identify affected VMs: Perform an audit of the .vmx files for active VMs in the environment to locate lines containing featMask.vm.cpuid.
2. Power off the VM: The VM must be powered off to modify the configuration file.
3. Edit the .vmx file:
   • Navigate to the VM directory on the datastore.
   • Remove lines starting with featMask.vm.cpuid. or other custom CPUID modifications that are no longer supported or required in the current environment.
4. Reload the VM: Reload the configuration (or re-register the VM) and power it on.
5. Verify: Confirm the VM boots successfully into the Guest Operating System.

Note: It is recommended to perform this remediation on a subset of VMs first to verify compatibility before applying changes across the entire environment.