Host down with PSOD with a backtrace "UnmapUpdateOnChange"

Products

• VMware vSphere ESXi

Issue/Introduction

In the core dump you will see backtrace similar to this:

cpu13:1234567)Backtrace for current CPU #13, worldID=1234567, fp=0x1
cpu13:1234567)0x453aa151b5e0:[0x420037ff4ed7]UnmapUpdateOnChange@esx#nover+0x11f stack: 0x43168585da30, 0x43168585dbc8, 0x85c0f0, 0x420037fbfd9e, 0x4316859c83dc
cpu13:1234567)0x453aa151b650:[0x420037ff52d6]UnmapRCUpdateCB@esx#nover+0x63 stack: 0x431687285c00, 0x420037fb8f3a, 0x431687286200, 0x36984df3, 0x0
cpu13:1234567)0x453aa151b670:[0x420037fb8f39]Res6UnmapRCUpdate@esx#nover+0x3e stack: 0x0, 0x420037f1f1d3, 0x431687285c00, 0xc474119ad8, 0x431687285c00
cpu13:1234567)0x453aa151b690:[0x420037f1f1d2]J3PerformPrePostCommitActionsInt@esx#nover+0x77 stack: 0x431687285c00, 0x1, 0x0, 0x420037f235b4, 0x8c1d00008
cpu13:1234567)0x453aa151b6c0:[0x420037f235b3]J3_CommitTransactionWithErr@esx#nover+0x784 stack: 0x43168585bc70, 0x43168585c0f0, 0x431687076200, 0x431687145070, 0x43168585bc70
cpu13:1234567)0x453aa151b7b0:[0x420037f24408]J3_CommitTransaction@esx#nover+0x49 stack: 0x430dbc60a150, 0x420037fa0b19, 0x453aa151b8a0, 0x0, 0x430dbc60a150
cpu13:1234567)0x453aa151b7d0:[0x420037fa0b18]Fil3_RenameFileVMFS6@esx#nover+0xbbd stack: 0x430dbc60a150, 0x431aee224870, 0x43168585bc70, 0x453aa151bb20, 0x43168711c120
cpu13:1234567)0x453aa151bc80:[0x420037f133a5]Fil3_RenameFileWithRetry@esx#nover+0xc2 stack: 0xa0014000a0014, 0x32000a, 0x0, 0x43168585bc70, 0x0
cpu13:1234567)0x453aa151bd20:[0x420036842623]FSSVec_RenameFile@vmkernel#nover+0x1c stack: 0x430dbc60a150, 0x430dbc60a150, 0x1cf533ee223f90, 0x63fe73ff00000003, 0x7ab430e13c21d9fe
cpu13:1234567)0x453aa151bd30:[0x420036ebd1bf]FSS_RenameFile@vmkernel#nover+0x160 stack: 0x1cf533ee223f90, 0x63fe73ff00000003, 0x7ab430e13c21d9fe, 0xc00b447adcf0f1, 0x2596f
cpu13:1234567)0x453aa151be10:[0x420036dc58f6]LinuxFileDesc_RenameAt@vmkernel#nover+0x24b stack: 0x431aee2237c0, 0x431aee223f90, 0x45ca8bea87c0, 0x0, 0x45ca8be6c8f8
cpu13:1234567)0x453aa151bf10:[0x420036dcfddc]User_LinuxSyscallHandler@vmkernel#nover+0xd9 stack: 0x3, 0x0, 0x0, 0x420036e9b0c7, 0x103
cpu13:1234567)0x453aa151bf40:[0x420036e9b0c6]gate_entry@vmkernel#nover+0xa7 stack: 0x0, 0x52, 0xe53eb98317, 0x0, 0x34652fa9760
cpu13:1234567)VMware ESXi 8.0.3 [Releasebuild-24280767 x86_64]

Environment

7.x

8.x

Cause

This issue is caused due to a bug in the ESXi software due to an  unmap feature. It is seen only when the unmap feature is enabled. It is caused by a race condition and is extremely rare to occur during regular workflow.

Resolution

Broadcom is working on coming up with a fix for this issue. There is no known solution at this point.