ESXi host running virtual machines with GPU hardware acceleration enabled experiences a PSOD (purple screen of death)
search cancel

ESXi host running virtual machines with GPU hardware acceleration enabled experiences a PSOD (purple screen of death)

book

Article ID: 334453

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

You see these issues on hosts running virtual machines with GPU hardware acceleration enabled:

  • The ESXi host fails with a purple diagnostic screen
  • You see a backtrace similar to:

YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: VM: nv_alloc_system_pages: failed to allocate memory
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: map pages vmk_Map failure: Failure
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: failed to map pages!
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: kernel mapping failed: Failure
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: VM: nv_alloc_system_pages: failed to allocate memory
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: kernel mapping failed: Failure
YYYY-MM-DDTHH:MM:SS.Z cpu##:15####)NVRM: VM: nv_alloc_system_pages: failed to allocate memory
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)World: 9###: PRDA 0x4########ss 0x0 ds 0x10b es 0x10b fs 0x10b gs 0x0
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)World: 9###: TR 0x4### GDT 0x439####### (0x402f) IDT 0x4###### (0xfff)
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)World: 9###: CR0 0x8###### CR3 0x10####### CR4 0x4###
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)Backtrace for current CPU ###, worldID=33###, rbp=0xd
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)0x4######:[0x4#######]PageCacheRemoveFirstPageLocked@vmkernel# nover+0x2f stack: 0x43#####
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)0x4######:[0x4#######]PageCacheAdjustSize @vmkernel#nover+0x260 stack: 0x0, 0x3c#######
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)0x4######:[0x4######]CpuSched_StartWorld@vmkernel#nover+0xa2 stack: 0x0, 0x0, 0x0, 0x0, 0
YYYY-MM-DDTHH:MM:SS.Z cpu##:33###)^[[45m^[[33;1mVMware ESXi 6.0.0 [Releasebuild-6921384 x86_64]^[[0m
#GP Exception 13 in world #####:memMap-13 @ 0x41########

 
Note: The preceding log excerpts are only examples. Date, time, and environmental variables may vary depending on your environment.
 

Environment

VMware vSphere ESXi 6.0

Cause

This issue occurs due to exhausted xmap and xmap allocation failure resulting in ESXi host PSOD

Resolution

This is a known issue affecting ESXi 6.0 and resolved in ESXi 6.0 P07 (ESXi600-201807001) patch release.