ESXi host fails with PSOD (purple screen of death) in NSX Environment
Article ID: 306803
Updated On:
Products
Issue/Introduction
PSOD (purple screen of death) is experienced with the following backtrace in NSX environment:
2016-10-15T02:44:04.401Z cpu24:33537)World: 9729: PRDA 0x418046000000 ss 0x0 ds 0x4018 es 0x4018 fs 0x0 gs 0x0
2016-10-15T02:44:04.401Z cpu24:33537)World: 9731: TR 0x4000 GDT 0xfffffffffc60a000 (0xffff) IDT 0xfffffffffc608000 (0xffff)
2016-10-15T02:44:04.401Z cpu24:33537)World: 9732: CR0 0x80050033 CR3 0x80ed16f000 CR4 0x52668
2016-10-15T02:44:04.438Z cpu24:33537)Backtrace for current CPU #24, worldID=33537, rbp=0x43a641cc47d0
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b4a0:[0x41800a325397][email protected]#v2_3_0_0+0x3b stack:
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b4d0:[0x41800994e998]IOChain_Resume@vmkernel#nover+0x210 stack: 0x4303e66bf658, 0x4391980
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b570:[0x41800993226e]PortOutput@vmkernel#nover+0xae stack: 0x417fc9a7f280, 0x43a641cc47d0
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b5b0:[0x418009fff5ec]EtherswitchForwardLeafPorts@<None>#<None>+0x244 stack: 0x54a33c00, 0
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b620:[0x41800a000778]EtherswitchPortDispatch@<None>#<None>+0x604 stack: 0x4303e6659f00, 0
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b810:[0x4180099324d3]Port_InputResume@vmkernel#nover+0x17b stack: 0x41800a000174, 0x41800
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b870:[0x4180099dd0f0]vmk_PortInput@vmkernel#nover+0x80 stack: 0x0, 0x4303e6648200, 0x4391
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b8a0:[0x41800a545a4f]VdrTxDispatchQueueDrain@<None>#<None>+0xbf stack: 0x0, 0x80000080000
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809b950:[0x41800a53589d]VdrProcessPackets@<None>#<None>+0x161 stack: 0x0, 0x418000000000, 0x
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bb80:[0x41800994e981]IOChain_Resume@vmkernel#nover+0x1f9 stack: 0x4303e6648798, 0x4391980
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bc20:[0x41800993226e]PortOutput@vmkernel#nover+0xae stack: 0x43a6549dc0c0, 0x43a641d30828
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bc60:[0x418009fff2f2]EtherswitchForwardLeafPortsQuick@<None>#<None>+0x136 stack: 0x439198
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bcb0:[0x41800a000760]EtherswitchPortDispatch@<None>#<None>+0x5ec stack: 0x4303e6648200, 0
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bea0:[0x4180099324d3]Port_InputResume@vmkernel#nover+0x17b stack: 0x41800a000174, 0x41800
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bf00:[0x41800991e7e5]Net_AcceptRxList@vmkernel#nover+0x115 stack: 0x4303e66586c0, 0x1, 0x
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bf60:[0x4180099d9db7]NetPollWorldCallback@vmkernel#nover+0x12f stack: 0x14, 0x418009fc06d
2016-10-15T02:44:04.438Z cpu24:33537)0x43919809bfd0:[0x418009a14a3e]CpuSched_StartWorld@vmkernel#nover+0xa2 stack: 0x0, 0x0, 0x0, 0x0, 0
2016-10-15T02:44:04.456Z cpu24:33537)^[[45m^[[33;1mVMware ESXi 6.0.0 [Releasebuild-3620759 x86_64]^[[0m
#PF Exception 14 in world 33537:vmnic0-pollW IP 0x41800a325397 addr 0x43070000005a
PTEs:0x8000000100009023;0x800000016a328063;0x0;
PSOD can occur during dvfilter usage. When the IOChain callback for dvfilter need to be installed during PortDoEnable, there is a possibility that the filter is not ready, it's like a race condition. The function which inserts the Iochain already catches this race, but one of the callers doesn't verify if the filter is ready.
Environment
Resolution
This is a known issue in ESXi 6.0 Update 2 (Build 3620759) and ESXi 6.0 Patch 3 (Build 4192238).
To resolve this issue, upgrade to ESXi 6.0 Express Patch 7 (Build 4510822) or later versions.
Feedback
