ESXi may PSOD (purple screen of death) with "PFrame_IsBackedByLPage" due to a rare race condition

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

ESXi hosts may experience a Purple Screen of Death with a backtrace similar to:

[0x42001ab7c298]PFrame_IsBackedByLPage@vmkernel#nover+0x0
[0x42001abb988f]VmMemCowHintUpdatePost@vmkernel#nover+0x94
[0x42001abbb201]VmMemCowPostToHintWorldIfMatched@vmkernel#nover+0x79e
[0x42001abbc136]VmMemCowPShareFn@vmkernel#nover+0xedf
[0x42001ab14432]VmAssistantProcessTasks@vmkernel#nover+0x14b
[0x42001adb33e1]CpuSched_StartWorld@vmkernel#nover+0x86
 

This may happen in rare instances under the following conditions:

• ESXi version 7.0.3 or 8.0 GA
• Inter-VM Transparent Page Sharing (TPS) enabled
  • Mem.ShareForceSalting set to 0 or 1
• A Fast Suspend Resume (FSR) operation on a VM with inter-VM shared pages
  • E.g. Hot-Add (Device, CPU, Memory), svMotion or en/disabling vSphere Fault Tolerance

Environment

VMware vSphere ESXi 8.0.0
VMware vSphere ESXi 7.0.3

Cause

Recent performance optimizations for FSRs of large virtual machines have introduced a rare race condition that may lead to a PSOD of the host.

Resolution

This issue is resolved in VMware ESXi 7.0 Update 3l (build number 21424296)
This issue is resolved in VMware ESXi 8.0b (build number 21203435)

Workaround:
You may disable inter-VM TPS before powering on or migrating any virtual machines to the host, alternatively avoid FSRs by powering off the VM before adding devices or migrating it to different storage.