PSOD in world "CAM taskq" when CDROM is attached to ESXi host with old BMC versions (approx. 4.40 and below)

Products

• VMware vSphere ESXi

Issue/Introduction

ESXi host with old BMC version has a PSOD with the following signature and backtrace:

 

(gdb+) vmklog | grep BlueScreen

2025-01-01T00:00:00Z cpuX:XXX)@BlueScreen: #PF Exception 14 in world XX:CAM taskq IP 0x0 addr 0x0

(gdb+) bt

#0 0x0000000000000000 in ?? ()

#1 0x0000000000000000 in taskqueue_run_locked (queue=queue@entry=0x0000000000000000) at vmkdrivers/native/BSD/vmkbsd/kern/subr_taskqueue.c:366

#2 0x0000000000000000 in taskqueue_thread_loop (arg=arg@entry=0x0000000000000000) at vmkdrivers/native/BSD/vmkbsd/kern/subr_taskqueue.c:598

#3 0x0000000000000000 in kthread_func (data=data@entry=0x0000000000000000) at vmkdrivers/native/BSD/vmkbsd/compat/util_freebsd_compat.c:456

#4 0x0000000000000000 in vmkWorldFunc (data=<optimized out>) at bora/vmkernel/main/vmkapi_world.c:590

#5 0x0000000000000000 in CpuSched_StartWorld (destWorld=<optimized out>, previous=<optimized out>) at bora/vmkernel/sched/cpusched.c:15324

#6 0x0000000000000000 in ?? () at bora/vmkernel/main/debug.c:4125

#7 0x0000000000000000 in ?? ()

Environment

ESXi 7.x, 8.x, 9.x

Issue noticed BMC versions 3.68 and 4.40, but not 6.10. This can be checked on ESXi with the command - esxcfg-info -a | grep "BMC Version"

 

Cause

 

• The PSOD was caused by a race condition in the taskqueue_run_locked thread.

• This condition was triggered because the thread attempted to handle a corrupted task resulting from the BMC's Virtual CD-ROM device attaching and detaching repeatedly over a short period.

 

Resolution

Contact your hardware vendor to update the BMC version.