Host PSOD during VDFM filter detach

Products

• VMware vSphere ESXi

Issue/Introduction

• ESXi host reports PSOD with below backtrace
  

  @BlueScreen: #PF Exception 14 in world 2098206:HELPER_MISC_ IP 0x42002cb6eb4c addr 0x16
PTEs:0x0;
Code start: 0x42002ca00000 VMK uptime: ##:##:##:##.##
0x453c20f1bc38:[0x42002cb6eb4c]MCSLockWork@vmkernel#nover+0x8 stack: 0x430516609a48
0x453c20f1bc40:[0x42002d921514][email protected]#1.0.0-0+0x149 stack: 0x4305166099f0
0x453c20f1bc70:[0x42002d731209][email protected]#1.0.0-0+0x16 stack: 0x430516609a80
0x453c20f1bc90:[0x42002d731bc1][email protected]#1.0.0-0+0x2e stack: 0x0
0x453c20f1bcb0:[0x42002d7325ad][email protected]#1.0.0-0+0x7a stack: 0x431059ab9450
0x453c20f1bce0:[0x42002d0c4c68]FDS_CloseDevice@vmkernel#nover+0x2d stack: 0x0
0x453c20f1bd00:[0x42002ca52028]DevFSFileClose@vmkernel#nover+0xa1 stack: 0x5a3736352e38323a
0x453c20f1bdd0:[0x42002ca41fe0]FSSVec_CloseFile@vmkernel#nover+0x19 stack: 0x430e2e401920
0x453c20f1bde0:[0x42002d0c0fc5]FSS_DoFileClose@vmkernel#nover+0x6e stack: 0x42002cb3fc00
0x453c20f1bdf0:[0x42002ca4685a]BC_CloseFile@vmkernel#nover+0x1e7 stack: 0x431e9f21a2f0
0x453c20f1be60:[0x42002d0c126d]FSS_CloseFile@vmkernel#nover+0x62 stack: 0x430e19a8beb0
0x453c20f1beb0:[0x42002e551478]CBT_RemoveDev@(cbt)#<None>+0x8d stack: 0x430e19a8beb0
0x453c20f1bee0:[0x42002d0c5807]FDS_RemoveDev@vmkernel#nover+0x90 stack: 0xbad0020
0x453c20f1bf10:[0x42002ca52254]DevFSUnlinkObj@vmkernel#nover+0xb5 stack: 0x430e19a8be70
0x453c20f1bf40:[0x42002ca522e9]DevFSHelperUnlink@vmkernel#nover+0x2a stack: 0x430e2e4018e0
0x453c20f1bf60:[0x42002cb5ba1c]HelperQueueFunc@vmkernel#nover+0x19d stack: 0x430e2e401238
0x453c20f1bfe0:[0x42002d0dc88e]CpuSched_StartWorld@vmkernel#nover+0xbf stack: 0x0
base fs=0x0 gs=0x420040000000 Kgs=0x0

• Host comes back online during the next boot.

Environment

VMware vSphere ESXi prior to 8.0 U3j

Cause

PSOD was caused by a rare race condition during the VDFM filter detach which is directly triggered by a preceding VMX crash.

Resolution

Currently there is no workaround or resolution for this issue.

Broadcom engineering is aware the issue and will be fixed in the upcoming vSphere release.