ESXi 8.0u3i PSOD: PF Exception 14 in world <ID>:vmnic#-pollW IP
search cancel

ESXi 8.0u3i PSOD: PF Exception 14 in world <ID>:vmnic#-pollW IP

book

Article ID: 434303

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

ESXi 8.0u3i hosts may experience a Purple Screen of Death (PSOD) during Virtual Machine operations such as vMotion or Snapshot creation/deletion.

The PSOD error message typically includes: #PF Exception 14 in world <ID>:vmnic#-pollW IP 0x4200055dde0c addr 0x432acd353000

The backtrace will contain calls similar to the following:

0x453b1761b620:[0x4200055dde0c]PhysMemFireTracesSlow@vmkernel#nover+0x5c stack: 0x453b4ab1f000, 0x43080693e580, 0x453b1761b8f8, 0x2d, 0x430927f0b100
0x453b1761b660:[0x4200055ddeb5]PhysMem_FireTraces@vmkernel#nover+0x36 stack: 0x453b0000001b, 0x453b00000000, 0x45bb8d828b40, 0x430927f07040, 0xffff8282187f3cc0
0x453b1761b670:[0x4200056b5baf]Vmxnet3VMKDevDeliverPktToQueue@vmkernel#nover+0x894 stack: 0x45bb8d828b40, 0x430927f07040, 0xffff8282187f3cc0, 0xffff8282187f3cc0, 0x43090000017a
0x453b1761b790:[0x4200056b8af0]Vmxnet3VMKDevRxWithLock@vmkernel#nover+0xdc1 stack: 0xfa24457ba100, 0x43080693e580, 0x0, 0x420000000000, 0xfffffffffffffcc0
0x453b1761ba10:[0x4200056bb659]Vmxnet3VMKDevRxFunc@vmkernel#nover+0x476 stack: 0x43080693e870, 0x453b1761bae0, 0x430927f07300, 0x43080693e870, 0x0
0x453b1761bad0:[0x42000566d364]PortOutput@vmkernel#nover+0x31d stack: 0x1, 0x45bb8d828b40, 0x45bb8d828b40, 0x12d00000001, 0x1
0x453b1761bb50:[0x420005766f73]vmk_PortOutput@vmkernel#nover+0x11c stack: 0x45bb67fa6048, 0x0, 0x45bb67fa6378, 0x0, 0x0
0x453b1761bb90:[0x4200063d868f][email protected]#1.0.0+0x9c stack: 0x453b1761bc60, 0x8400000a00000049, 0x0, 0x40000498d828e80, 0x453b1761be40
0x453b1761bc00:[0x4200063e1bf6][email protected]#1.0.0+0x74f stack: 0x0, 0x420000000000, 0x0, 0x0, 0x0
0x453b1761be50:[0x42000566d5be]Port_InputResume@vmkernel#nover+0x1bf stack: 0x42000566d5ad, 0x0, 0x430806804bf0, 0x430806804bf0, 0x4302d62a5f80
0x453b1761bea0:[0x4200056cf1e2]Net_AcceptRxList@vmkernel#nover+0x25b stack: 0x4522c4e2a000, 0x420006823852, 0x0, 0x0, 0x202
0x453b1761bf70:[0x42000569ec03]NetPollWorldCallback@vmkernel#nover+0xb8 stack: 0x0, 0x0, 0x0, 0x0, 0x0
0x453b1761bfe0:[0x420005adde22]CpuSched_StartWorld@vmkernel#nover+0xbf stack: 0x0, 0x420005545570, 0x0, 0x0, 0x0
0x453b1761c000:[0x42000554556f]Debug_IsInitialized@vmkernel#nover+0xc stack: 0x0, 0x0, 0x0, 0x0, 0x0

Environment

ESXi 8.0u3i

Cause

The system panic is triggered by a conflict within the vIOMMU (Virtual Input/Output Memory Management Unit) device. This specific feature was introduced in ESXi 8.0u3i. Under certain conditions, the vIOMMU device interferes with the vmnic polling process, leading to a page fault (Exception 14).

Resolution

Use one of the following methods to mitigate the risk.

Step 1: Identify Impacted Virtual Machines

Before applying workarounds, identify which VMs have vIOMMU (VVTD) enabled using PowerCLI:

Get-VM | Where-Object {$_.ExtensionData.Config.Flags.VvtdEnabled -eq $true} | Select-Object Name, @{N="vIOMMU_Enabled"; E={$_.ExtensionData.Config.Flags.VvtdEnabled}} | Format-Table -AutoSize

Option 1: Disable vIOMMU (Recommended)

This is the most effective way to prevent the PSOD while maintaining operational flexibility.

  1. Power off the affected Virtual Machine.

  2. Right-click the VM in the vSphere Client and select Edit Settings.

  3. Go to Virtual Hardware > CPU.

  4. Uncheck the box for I/O MMU.

  5. Click OK and power the VM back on.

Option 2: Restrict Operations (Alternative)

If vIOMMU is strictly required for the guest OS workload, you must prevent the triggers (vMotion and Snapshots):

To prevent vMotion:

  • Configure VM-Host Affinity Rules (specifically "Must run on hosts in group") to pin the VM to a specific host.

  • Alternatively, set the DRS automation level for the specific VM to Disabled or Manual.

To prevent Snapshots:

  1. Power off the VM.

  2. Right-click the VM > Edit Settings > Advanced Parameters.

  3. Add a new row:

    • Attribute: snapshot.maxSnapshots

    • Value: 0

  4. Click OK and power the VM back on.

Additional Information

This is a known issue. VMware Engineering is currently working on a permanent fix, which is scheduled for inclusion in ESXi 8.0 P09.

To receive updates on this article, please click Subscribe in the Actions column.

Powered by Wolken Software