如何理解 WmiPrvSE 进程中明显的内存泄漏？

Question

我的应用程序运行有很多 WMI 查询，这些查询是通过打开与 root\cimv2 的连接、执行查询，然后关闭该连接来完成的。

现在 WmiPrvSE.exe 进程似乎正在发生内存泄漏。

heap_stat.py 内存泄漏调查（heap_stat.py，如 this URL 中所述），显示以下对象是此泄漏的来源：

            Type name      Count     New count
   fastprox!CClassPart      2305          2409
fastprox!CInstancePart      1719          1705
  cimwin32!CRefPtrLite      1131          1205
  fastprox!CWbemObject      1363          1303
fastprox!CWbemInstance      1347          1443
                   ...
  provthrd!WmiTreeNode       734           882
    combase!CNdrStream       166           300

在 Internet 上，有很多关于 WMI 相关内存泄漏的修补程序（example 1, example 2，...），但这些似乎只与某些情况相关。我怎么知道我的情况是否属于这些案件中的任何一个以及属于哪个案件？（我不能告诉我的客户只在他的系统上应用一个修补程序而不解释为什么这个修补程序可以解决他的问题）

我的 Windows 版本（Winver.exe 结果）是：

Windows Server 2016
Microsoft Windows Server
Version 1607 (OS Build 14393.1770)
Copyright 2016 Microsoft Corporation. All rights reserved.
...

与此同时，我也了解到我们可以使用一些工具来监控 WMI 状态，例如 described here。我们如何知道发生了什么以及我们如何（patch/solution）解决它？

根据MagicAndre1981的评论编辑

在提到的超级用户post中，提到Windbg的!Analyze -v命令可能会泄露一些信息，因此我启动了这个命令，特此给出结果（我有看看它，但我不知道这一切意味着什么）。供您参考：我在两个不同的 WmiPrvSE.exe 转储上启动了这个命令，结果相似：

0:000> !analyze -v
*******************************************************************************
*                                                                             *
*                        Exception Analysis                                   *
*                                                                             *
*******************************************************************************

GetUrlPageData2 (WinHttp) failed: 12030.

DUMP_CLASS: 2

DUMP_QUALIFIER: 400

FAULTING_IP: 
+0
00000000`00000000 ??              ???

EXCEPTION_RECORD:  (.exr -1)
ExceptionAddress: 0000000000000000
   ExceptionCode: 80000003 (Break instruction exception)
  ExceptionFlags: 00000000
NumberParameters: 0

FAULTING_THREAD:  00005e64

BUGCHECK_STR:  BREAKPOINT

DEFAULT_BUCKET_ID:  BREAKPOINT

PROCESS_NAME:  WmiPrvSE.exe

ERROR_CODE: (NTSTATUS) 0x80000003 - {EXCEPTION}  Breakpoint  A breakpoint has been reached.

EXCEPTION_CODE: (HRESULT) 0x80000003 (2147483651) - One or more arguments are invalid

EXCEPTION_CODE_STR:  80000003

WATSON_BKT_PROCSTAMP:  57899ab2

WATSON_BKT_PROCVER:  6.2.14393.0

PROCESS_VER_PRODUCT:  Microsoft® Windows® Operating System

WATSON_BKT_MODULE:  unknown

WATSON_BKT_MODVER:  0.0.0.0

WATSON_BKT_MODOFFSET:  0

WATSON_BKT_MODSTAMP:  bbbbbbb4

BUILD_VERSION_STRING:  10.0.14393.1198 (rs1_release_sec.170427-1353)

MODLIST_WITH_TSCHKSUM_HASH:  89fd758871dd996e76ac11caaaa9667af30618db

MODLIST_SHA1_HASH:  f52f927737ff9b80664faa9d7561eb8997ba5a98

COMMENT:  
*** procdump  -ma 32476
*** Manual dump

NTGLOBALFLAG:  0

PROCESS_BAM_CURRENT_THROTTLED: 0

PROCESS_BAM_PREVIOUS_THROTTLED: 0

APPLICATION_VERIFIER_FLAGS:  0

PRODUCT_TYPE:  3

SUITE_MASK:  272

DUMP_FLAGS:  8000c07

DUMP_TYPE:  3

ANALYSIS_SESSION_HOST:  DOMINIQUEDS

ANALYSIS_SESSION_TIME:  03-28-2019 16:54:42.0605

ANALYSIS_VERSION: 10.0.16299.15 x86fre

THREAD_ATTRIBUTES: 
OS_LOCALE:  FRB

PROBLEM_CLASSES: 

    ID:     [0n309]
    Type:   [@APPLICATION_FAULT_STRING]
    Class:  Primary
    Scope:  DEFAULT_BUCKET_ID (Failure Bucket ID prefix)
            BUCKET_ID
    Name:   Omit
    Data:   Add
            String: [BREAKPOINT]
    PID:    [Unspecified]
    TID:    [Unspecified]
    Frame:  [0]

PRIMARY_PROBLEM_CLASS:  BREAKPOINT

LAST_CONTROL_TRANSFER:  from 00007ffba40b4856 to 00007ffba2fe1164

STACK_TEXT:  
00000044`9451f6f8 00007ffb`a40b4856 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : win32u!NtUserGetMessage+0x14
00000044`9451f700 00007ff6`656c5d7e : 00000044`9451f780 00000000`00000000 00000000`00000000 00000000`00000000 : user32!GetMessageW+0x26
00000044`9451f730 00007ff6`656c192b : 00000000`00000000 000001b1`ffffffff 000001b1`ae01a6c0 000001b1`adff0c40 : WmiPrvSE!Process+0x4ee
00000044`9451f860 00007ff6`656d9257 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : WmiPrvSE!WinMain+0x21b
00000044`9451f8e0 00007ffb`a3fd8364 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : WmiPrvSE!WinMainCRTStartup+0x1b7
00000044`9451f9a0 00007ffb`a66a7091 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : kernel32!BaseThreadInitThunk+0x14
00000044`9451f9d0 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : ntdll!RtlUserThreadStart+0x21


STACK_COMMAND:  ~0s; .ecxr ; kb

THREAD_SHA1_HASH_MOD_FUNC:  69047869935fe9e3124f9ea8ff8b6da09a09db5f

THREAD_SHA1_HASH_MOD_FUNC_OFFSET:  23d47332d72414f95439acd0d8334dbbce9ac40b

THREAD_SHA1_HASH_MOD:  07201fdab54c758a75c51b7668701b4fab031f6d

FOLLOWUP_IP: 
win32u!NtUserGetMessage+14
00007ffb`a2fe1164 c3              ret

FAULT_INSTR_CODE:  c32ecdc3

SYMBOL_STACK_INDEX:  0

SYMBOL_NAME:  win32u!NtUserGetMessage+14

FOLLOWUP_NAME:  MachineOwner

MODULE_NAME: win32u

IMAGE_NAME:  win32u.dll

DEBUG_FLR_IMAGE_TIMESTAMP:  57a05800

BUCKET_ID:  BREAKPOINT_win32u!NtUserGetMessage+14

FAILURE_EXCEPTION_CODE:  80000003

FAILURE_IMAGE_NAME:  win32u.dll

BUCKET_ID_IMAGE_STR:  win32u.dll

FAILURE_MODULE_NAME:  win32u

BUCKET_ID_MODULE_STR:  win32u

FAILURE_FUNCTION_NAME:  NtUserGetMessage

BUCKET_ID_FUNCTION_STR:  NtUserGetMessage

BUCKET_ID_OFFSET:  14

BUCKET_ID_MODTIMEDATESTAMP:  57a05800

BUCKET_ID_MODCHECKSUM:  22f84

BUCKET_ID_MODVER_STR:  6.2.14393.51

BUCKET_ID_PREFIX_STR:  BREAKPOINT_

FAILURE_PROBLEM_CLASS:  BREAKPOINT

FAILURE_SYMBOL_NAME:  win32u.dll!NtUserGetMessage

FAILURE_BUCKET_ID:  BREAKPOINT_80000003_win32u.dll!NtUserGetMessage

WATSON_STAGEONE_URL:  http://watson.microsoft.com/StageOne/WmiPrvSE.exe/6.2.14393.0/57899ab2/unknown/0.0.0.0/bbbbbbb4/80000003/00000000.htm?Retriage=1

TARGET_TIME:  2019-03-20T13:39:18.000Z

OSBUILD:  14393

OSSERVICEPACK:  1198

SERVICEPACK_NUMBER: 0

OS_REVISION: 0

OSPLATFORM_TYPE:  x64

OSNAME:  Windows 10

OSEDITION:  Windows 10 Server TerminalServer SingleUserTS

USER_LCID:  0

OSBUILD_TIMESTAMP:  2017-04-28 01:48:56

BUILDDATESTAMP_STR:  170427-1353

BUILDLAB_STR:  rs1_release_sec

BUILDOSVER_STR:  10.0.14393.1198

ANALYSIS_SESSION_ELAPSED_TIME:  378d

ANALYSIS_SOURCE:  UM

FAILURE_ID_HASH_STRING:  um:breakpoint_80000003_win32u.dll!ntusergetmessage

FAILURE_ID_HASH:  {3112b5eb-303b-e877-0655-90bdfa336126}

Followup:     MachineOwner
---------

对问题有更多了解后编辑

同时我查看了有关 WMI 的事件日志 activity:

Application and Services Logs, Microsoft, Windows, WMI-Activity.

这包含很多错误（事件 ID 5858），包含以下信息（出于安全原因混淆了客户信息、计算机名称和 GUID）：

Log Name:      Microsoft-Windows-WMI-Activity/Operational
Source:        Microsoft-Windows-WMI-Activity
Date:          29/03/2019 11:44:54
Event ID:      5858
Task Category: None
Level:         Error
Keywords:      
User:          SYSTEM
Computer:      Computer_Name.customer_name.intra
Description:
Id = {00000000-0000-0000-0000-000000000000}; ClientMachine = Computer_Name; User = NT AUTHORITY\SYSTEM; ClientProcessId = 1220; Component = Unknown; Operation = Start IWbemServices::DeleteInstance - Root\Rsop\Computer : RSOP_ExtensionStatus.extensionGuid="{........-....-....-....-............}"; ResultCode = 0x80041002; PossibleCause = Unknown
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
  <System>
    <Provider Name="Microsoft-Windows-WMI-Activity" Guid="{1418ef04-b0b4-4623-bf7e-d74ab47bbdaa}" />
    <EventID>5858</EventID>
    <Version>0</Version>
    <Level>2</Level>
    <Task>0</Task>
    <Opcode>0</Opcode>
    <Keywords>0x4000000000000000</Keywords>
    <TimeCreated SystemTime="2019-03-29T10:44:54.842915300Z" />
    <EventRecordID>564437</EventRecordID>
    <Correlation ActivityID="{........-....-....-....-............}" />
    <Execution ProcessID="1736" ThreadID="3860" />
    <Channel>Microsoft-Windows-WMI-Activity/Operational</Channel>
    <Computer>Computer_Name.customer_name.intra</Computer>
    <Security UserID="S-1-5-18" />
  </System>
  <UserData>
    <Operation_ClientFailure xmlns="http://manifests.microsoft.com/win/2006/windows/WMI">
      <Id>{00000000-0000-0000-0000-000000000000}</Id>
      <ClientMachine>Computer_Name</ClientMachine>
      <User>NT AUTHORITY\SYSTEM</User>
      <ClientProcessId>1220</ClientProcessId>
      <Component>Unknown</Component>
      <Operation>Start IWbemServices::DeleteInstance - Root\Rsop\Computer : RSOP_ExtensionStatus.extensionGuid="{........-....-....-....-............}"</Operation>
      <ResultCode>0x80041002</ResultCode>
      <PossibleCause>Unknown</PossibleCause>
    </Operation_ClientFailure>
  </UserData>
</Event>

这些事件是否阐明了系统的 WMI 处理可能存在的问题？

上述PID（1220）对应的服务如下（完整列表）：

Windows Push Notifications System Service
User Profile Service
User Manager
Themes
Task Scheduler
System Event Notificiation Service
Shell hardware Detection
Remote Desktop Configuration
Network Setup Service
IP Helper
IKE and AuthIP IPsec Keying Modules
Group Policy Client
Geolocation Service
Certificate Propagation
Application Information

Answer 1

要跟踪 WmiPrvSE.exe，您需要使用 ETW 或通过事件查看器捕获 Microsoft-Windows-WMI-Activity 事件（单击显示分析和调试日志。在应用程序和服务日志下找到 WMI 的跟踪通道日志 | Microsoft | Windows | WMI Activity)

我更喜欢 xperf/ETW way，因为您可以将跟踪复制到不同的系统并仍然获得所有数据。

xperf -on PROC_THREAD+LOADER+PROFILE+INTERRUPT+DPC+DISPATCHER -stackwalk profile -BufferSize 1024 -MaxFile 256 -FileMode Circular -f Kernel.etl
xperf -start WMILogger -on Microsoft-Windows-WMI-Activity::0xff -BufferSize 1024 -f WMI.etl

echo Please capture about 30s of the WMI activity.

pause

xperf -stop
xperf -stop WMILogger
xperf -merge WMI.etl kernel.etl WMItracing.etl

del WMI.etl
del kernel.etl

打开 WPA.exe 中生成的 WMItracing.etl，然后将 "Generic Events" 图表从左侧拖放到分析窗格中。

现在仅过滤到 Microsoft-Windows-WMI-Activity 事件，并查找 WMI 操作和 ClientProcessId。

此客户端进程 ID 显示执行 WMI 操作的进程。

在我的示例中，此 ClientProcessId 属于名为 Veeam ONE Monitor Server 的工具。

检查您的系统是否有 WMI 调用以及哪个 ClientProcessId 属于 WMI 调用。

如何理解 WmiPrvSE 进程中明显的内存泄漏？

How to understand apparent memory leak in WmiPrvSE process?

wmi

memory-leaks

windows-server-2016