使用Windbg解析dump文件

第一章 常用的Windbg指令

①!analyze -v  ②kP                                               可以看函数的入参 ③!for_each_frame dv /t                            可以看函数中的局部变量 ④dc , db                                          产看某一内存中的值    可以直接接变量名 不过可能需要回溯栈 ⑤!threads                                         显示所有线程 ⑥~0s ， ~1s                                       进入某个线程 ⑦!frame ProcessA!FunctionA                        查看某一变量有时需要。 回溯栈   ⑧!uniqstack                                       扩展命令显示当前进程中所有线程的调用堆栈，除开重复的那些。    ⑨!teb                                             扩展以的格式化后的形式显示线程环境块(TEB)的信息。  ⑩s-sa 和 s-su                                     命令搜索未指定的 ASCII 和 Unicode 字符串。这在检查某段内存是否包含可打印字符时有用。 �7�6dds、dps 和 dqs 命令显示给定范围内存的内容。     该内存被假定为符号表中的一连串地址。相应的符号也会被显示出来。命令显示给定范围内存的内容,它们是把内存区域转储出来，并把内存中每个元素都视为一个符号对其进行解析，dds是四字节视为一个符号，dqs是每8字节视为一个符号，dps是根据当前处理器架构来选择最合适的长度 �7�7.kframes                                        命令设置堆栈回溯显示的默认长度。默认20 �7�8k, kb, kd, kp, kP, kv (Display Stack Backtrace) k*命令显示给定线程的调用堆栈，以及其他相关信息。通常要结合12)使用否则显示出来的东西很少 �7�9.reload /i xxx.dll                              忽略.pdb 文件版本不匹配的情况。

第二章 Symbol的设置方法

2.1 将远程的系统函数的PDB文件拷贝到本地「D:\mysymbol」目录下     SRV*D:\mysymbol*http://msdl.microsoft.com/download/symbols 2.2 加载设置的符号文件     .reload     可以使用菜单中的 Debug -> Modules 查看有没有加载进来

第三章 实例

实例1 如何调查堆被破坏问题。     错误代码：0xc0000374     错误含义：ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun 第一步、先用「!analyze -v」分析出错误的地方以及由于什么原因导致程序Dump掉的。        无非是内存溢出，访问非法地址等几种。 0:009> !analyze -v ******************************************************************************* *                                                                             * *                        Exception Analysis                                   * *                                                                             * ******************************************************************************* GetPageUrlData failed, server returned HTTP status 404 URL requested: http://watson.microsoft.com/StageOne/ProcessA_exe/1_0_0_1/5134aefd/ntdll_dll/6_1_7601_18229/51fb164a/c0000374/000c4102.htm?Retriage=1 FAULTING_IP:  ntdll!RtlReportCriticalFailure+62 00000000`777b4102 eb00            jmp     ntdll!RtlReportCriticalFailure+0x64 (00000000`777b4104) EXCEPTION_RECORD:  ffffffffffffffff -- (.exr 0xffffffffffffffff) ExceptionAddress: 00000000777b4102 (ntdll!RtlReportCriticalFailure+0x0000000000000062)    ExceptionCode: c0000374   ExceptionFlags: 00000001 NumberParameters: 1    Parameter[0]: 000000007782b4b0 PROCESS_NAME:  ProcessA.exe ERROR_CODE: (NTSTATUS) 0xc0000374 - <Unable to get error code text> EXCEPTION_CODE: (NTSTATUS) 0xc0000374 - <Unable to get error code text> EXCEPTION_PARAMETER1:  000000007782b4b0 MOD_LIST: <ANALYSIS/> NTGLOBALFLAG:  0 APPLICATION_VERIFIER_FLAGS:  0 FAULTING_THREAD:  0000000000002f8c DEFAULT_BUCKET_ID:  ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun PRIMARY_PROBLEM_CLASS:  ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun BUGCHECK_STR:  APPLICATION_FAULT_ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun LAST_CONTROL_TRANSFER:  from 00000000777b4746 to 00000000777b4102 STACK_TEXT:   00000000`0548e170 00000000`777b4746 : 00000000`00000002 00000000`00000023 00000000`00000000 00000000`00000003 : ntdll!RtlReportCriticalFailure+0x62 00000000`0548e240 00000000`777b5952 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`1c01001d : ntdll!RtlpReportHeapFailure+0x26 00000000`0548e270 00000000`777b7604 : 00000000`00c50000 00000000`00c50000 00000000`0000000a 00000000`00000000 : ntdll!RtlpHeapHandleError+0x12 00000000`0548e2a0 00000000`777b79e8 : 00000000`00c50000 00000000`00000000 00000000`00100000 00000000`00000000 : ntdll!RtlpLogHeapFailure+0xa4 00000000`0548e2d0 00000000`7774fad6 : 00000000`00c50000 00000000`00c59e50 00000000`00c50000 00000000`00000000 : ntdll!RtlpAnalyzeHeapFailure+0x3a8 00000000`0548e330 00000000`777434d8 : 00000000`00c50000 00000000`00000003 00000000`000006cc 00000000`000006e0 : ntdll!RtlpAllocateHeap+0x1d2a 00000000`0548e8d0 00000000`777247ea : 00000000`00000003 00000000`00c5ee80 00000000`00c50278 00000000`000006cc : ntdll!RtlAllocateHeap+0x16c 00000000`0548e9e0 00000000`77723ff2 : 00000000`00c50000 00000000`00000003 00000000`00c5ee90 00000000`000006cc : ntdll!RtlpReAllocateHeap+0x648 00000000`0548eca0 00000000`750c712f : 00000000`0548fbe8 00000000`00c5ee90 00000000`00000000 00000000`000005ac : ntdll!RtlReAllocateHeap+0xa2 00000000`0548edb0 00000001`40010f6f : 00000000`00000000 00000000`0548fbe8 00000000`00000000 00000000`00000661 : msvcr80!realloc+0x6f [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\realloc.c @ 332] 00000000`0548ede0 00000001`4000f63c : ffffffff`ffffffff 00000000`0548ff10 00000000`00c97fd0 00000000`0548fe48 : ProcessA!FunctionA_AnalyzeEventData+0xfcf [e:\ProcessA\FunctionA_sockserv.cpp @ 1666] 00000000`0548f8a0 00000000`774e652d : 00000000`000002a0 00000000`00000000 00000000`00000000 00000000`00000000 : ProcessA!FunctionA_SockWork+0xe1c [e:\ProcessA\FunctionA_sockserv.cpp @ 1102] 00000000`0548ff60 00000000`7771c541 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : kernel32!BaseThreadInitThunk+0xd 00000000`0548ff90 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : ntdll!RtlUserThreadStart+0x1d STACK_COMMAND:  !heap ; ~9s; .ecxr ; kb FOLLOWUP_IP:  msvcr80!realloc+6f [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\realloc.c @ 332] 00000000`750c712f 4885c0          test    rax,rax SYMBOL_STACK_INDEX:  9 SYMBOL_NAME:  msvcr80!realloc+6f FOLLOWUP_NAME:  MachineOwner MODULE_NAME: msvcr80 IMAGE_NAME:  msvcr80.dll DEBUG_FLR_IMAGE_TIMESTAMP:  4ec3407e FAILURE_BUCKET_ID:  ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun_c0000374_msvcr80.dll!realloc BUCKET_ID:  X64_APPLICATION_FAULT_ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun_msvcr80!realloc+6f WATSON_STAGEONE_URL:  http://watson.microsoft.com/StageOne/ProcessA_exe/1_0_0_1/5134aefd/ntdll_dll/6_1_7601_18229/51fb164a/c0000374/000c4102.htm?Retriage=1 Followup: MachineOwner --------- 第二步、使用「!heap」找出出错的堆。分析出错的原因。        0000000000c59c80        0000000000c59e50  ←出错的堆地址。        0000000000c59fd0 大家应该有这样的常识，在使用malloc()或者realloc()分配出来的空间的前面都有 相应的管理情报，用来记录这块分配的内存的大小以及返回的时候用的情报。 从这里很自然的猜想到，在写往0000000000c59c80里面写数据的时候写过了， 写到0000000000c59e50上去了，导致它的管理情报被覆盖了。从而程序dump掉了。 0:009> !heap ************************************************************** *                                                            * *                  HEAP ERROR DETECTED                       * *                                                            * ************************************************************** Details: Error address: 0000000000c59e50 Heap handle: 0000000000c50000 Error type heap_failure_buffer_overrun (6) Parameter 1: 000000000000000a Last known valid blocks: before - 0000000000c59c80, after -0000000000c59fd0 Stack trace:                 00000000777b79e8: ntdll!RtlpAnalyzeHeapFailure+0x00000000000003a8                 000000007774fad6: ntdll!RtlpAllocateHeap+0x0000000000001d2a                 00000000777434d8: ntdll!RtlAllocateHeap+0x000000000000016c                 00000000777247ea: ntdll!RtlpReAllocateHeap+0x0000000000000648                 0000000077723ff2: ntdll!RtlReAllocateHeap+0x00000000000000a2                 00000000750c712f: msvcr80!realloc+0x000000000000006f                 0000000140010f6f: ProcessA!FunctionA_AnalyzeEventData+0x0000000000000fcf                 000000014000f63c: ProcessA!FunctionA_SockWork+0x0000000000000e1c                 00000000774e652d: kernel32!BaseThreadInitThunk+0x000000000000000d                 000000007771c541: ntdll!RtlUserThreadStart+0x000000000000001d Index   Address  Name      Debugging options enabled   1:   001f0000                   2:   00010000                   3:   00020000                   4:   00670000                   5:   00950000                   6:   00c50000                   7:   00910000                   8:   00bc0000                   9:   010e0000                  10:   01220000                  11:   01420000                  12:   00c30000                  13:   03660000                  14:   00ba0000                  15:   037b0000                  16:   01340000                  17:   039a0000                 第三步、使用「!for_each_frame dv /t」打印出错函数的局部变量，找出元凶。        从下面的变量里面找到距离0000000000c59c80地址最近的变量，对了就是它:        char * pData_n = 0x00000000`00c59c90 "SE:Security: ???"        ※注意如果变量值指针的指针需要先用dc看一下该指针指向的地址。        之后看代码知道，程序在读取pData_n的数据的时候如果遇到是0a(Windos换行符)就自动在后面加上        0d变成0a0d。导致pData_n内存越界了。 0:009> !for_each_frame dv /t _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 12 00000000`0548edb0 00000001`40010f6f msvcr80!realloc+0x6f [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\realloc.c @ 332] void * pBlock = 0x00000000`00000000 unsigned int64 newsize = 0x548fbe8 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 13 00000000`0548ede0 00000001`4000f63c ProcessA!FunctionA_AnalyzeEventData+0xfcf [e:\ProcessA\FunctionA_sockserv.cpp @ 1666] void * cd = 0xffffffff`ffffffff struct _MpEvsHead * Head = 0x00000000`0548ff10 char * pEventData = 0x00000000`00c97fd0 "???" char ** pNewData = 0x00000000`0548fe48 char * SiteName = 0x00000000`0548fe18 "" int oval_check = 0n0 char * pszHostIp = 0x00000000`0548fbf0 "192.168.1.1" int j = 0n469 int NodeName_check = 0n0 char [2068] eventtext = char [2068] "SE:Security: ???" unsigned long err = 0 int NL_henkan = 0n1 int Evttxt_check = 0n1 char [129] nameWork = char [129] "`_???" int ret = 0n0 struct NameObject_t * pNameObj_n = 0x00000000`00c5eee8 char * pData_n = 0x00000000`00c59c90 "SE:Security: ???" long lWork = 0n9 char [257] szTrcBuff = char [257] "safely divided text.([453]bytes --> [469]bytes)" long nNameNum = 0n44 long nNewLen = 0n1740 struct NameObject_t * pNameObj_o = 0x00000000`00c98028 char * pData_o = 0x00000000`00c984c6 "SE:Security: ???" char * pt = 0x00000000`00c59e55 "[???" long i = 0n20 int IpAddr_check = 0n0 int res = 0n1 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 14 00000000`0548f8a0 00000000`774e652d ProcessA!FunctionA_SockWork+0xe1c [e:\ProcessA\FunctionA_sockserv.cpp @ 1102] void * ns = 0x00000000`000002a0 char * pRead_str = 0x00000000`00c562f0 "," int bTableRegisterd = 0n0 unsigned long err = 0 char [3] traceflg = char [3] "" int ret = 0n0 short sWork = 0n2 int oval_check = 0n0 char * pNewData = 0x00000000`00c5ee90 "???" char * wk = 0x00000000`0548f930 "192.168.1.1" char [33] SiteName = char [33] "" long lWork = 0n2032 char [257] szTrcBuff = char [257] "recv event OK" int iLastSerchedIndex = 0n0 char [256] HostIp = char [256] "192.168.1.1" int ret2 = 0n0 struct _MpEvsHead Head = struct _MpEvsHead long nDataLen = 0n3 char [257] szTrcBuff2 = char [257] "" char [20] szSendData = char [20] "OK" struct addrinfo hinst = struct addrinfo int conv_disc_set = 0n1 long lRc = 0n0 void * conv_disc = 0xffffffff`ffffffff int res = 0n1 char * pData = 0x00000000`00c97fd0 "???" long nRead = 0n3726 char [16] evttype = char [16] "Alarm.sys" char * lpszEventid = 0x00000000`00c5f180 "" long nSend = 0n12 char [256] ipTmp = char [256] "192.168.1.1" char [20] szToCode = char [20] "sjis" char [20] szFromCode = char [20] "sjis" int bWriteEvent = 0n1 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

实例2  无效参数(STATUS_INVALID_PARAMETER)。     错误代码：0xc000000d     错误含义：STATUS_INVALID_PARAMETER 第一步、先用「!analyze -v」分析出错误的地方以及由于什么原因导致程序Dump掉的。   0:000> !analyze -v ******************************************************************************* *                                                                             * *                        Exception Analysis                                   * *                                                                             * ******************************************************************************* *** ERROR: Symbol file could not be found.  Defaulted to export symbols for user32.dll -  Unable to load image C:\Windows\Odsv.dll, Win32 error 0n2 *** WARNING: Unable to verify timestamp for Odsv.dll *** ERROR: Module load completed but symbols could not be loaded for Odsv.dll GetPageUrlData failed, server returned HTTP status 404 URL requested: http://watson.microsoft.com/StageOne/ProcessB_exe/1_0_0_1/4e362265/msvcr80_dll/8_0_50727_6195/4dcdd833/c000000d/0001d5fa.htm?Retriage=1 FAULTING_IP:  msvcr80!strncpy_s+10a [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\tcsncpy_s.inl @ 62] 00000000`74e6d5fa b822000000      mov     eax,22h EXCEPTION_RECORD:  ffffffffffffffff -- (.exr 0xffffffffffffffff) ExceptionAddress: 0000000074e6d5fa (msvcr80!strncpy_s+0x000000000000010a)    ExceptionCode: c000000d   ExceptionFlags: 00000000 NumberParameters: 0 PROCESS_NAME:  ProcessB.exe ERROR_CODE: (NTSTATUS) 0xc000000d - <Unable to get error code text> EXCEPTION_CODE: (NTSTATUS) 0xc000000d - <Unable to get error code text> MOD_LIST: <ANALYSIS/> NTGLOBALFLAG:  0 APPLICATION_VERIFIER_FLAGS:  0 LAST_CONTROL_TRANSFER:  from 0000000000124250 to 0000000074e5b0ec FAULTING_THREAD:  ffffffffffffffff DEFAULT_BUCKET_ID:  STATUS_INVALID_PARAMETER PRIMARY_PROBLEM_CLASS:  STATUS_INVALID_PARAMETER BUGCHECK_STR:  APPLICATION_FAULT_STATUS_INVALID_PARAMETER IP_ON_STACK:  +2e32faf01dedf58 00000000`00124250 60              ??? FRAME_ONE_INVALID: 1 STACK_TEXT:   00000000`00124220 00000000`00124250 : 00000000`00000006 00000000`00000000 00000000`00000001 00000000`00000000 : msvcr80!_invalid_parameter+0x6c [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\invarg.c @ 88] 00000000`00124228 00000000`00000006 : 00000000`00000000 00000000`00000001 00000000`00000000 00000000`00000000 : 0x124250 00000000`00124230 00000000`00000000 : 00000000`00000001 00000000`00000000 00000000`00000000 00000000`00124260 : 0x6 STACK_COMMAND:  ~0s; .ecxr ; kb FOLLOWUP_IP:  msvcr80!strncpy_s+10a [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\tcsncpy_s.inl @ 62] 00000000`74e6d5fa b822000000      mov     eax,22h FAULTING_SOURCE_CODE:   No source found for 'f:\dd\vctools\crt_bld\self_64_amd64\crt\src\tcsncpy_s.inl' SYMBOL_STACK_INDEX:  0 SYMBOL_NAME:  msvcr80!strncpy_s+10a FOLLOWUP_NAME:  MachineOwner MODULE_NAME: msvcr80 IMAGE_NAME:  msvcr80.dll DEBUG_FLR_IMAGE_TIMESTAMP:  4dcdd833 FAILURE_BUCKET_ID:  STATUS_INVALID_PARAMETER_c000000d_msvcr80.dll!strncpy_s BUCKET_ID:  X64_APPLICATION_FAULT_STATUS_INVALID_PARAMETER_msvcr80!strncpy_s+10a WATSON_STAGEONE_URL:  http://watson.microsoft.com/StageOne/ProcessB_exe/1_0_0_1/4e362265/msvcr80_dll/8_0_50727_6195/4dcdd833/c000000d/0001d5fa.htm?Retriage=1 Followup: MachineOwner --------- 这次运气很不好，从「!analyze -v」打出来的结果来看看不出啥东西来，只知道 在调用strncpy_s的时候dmp掉了，无法定位具体是哪个函数出错的原因很多，有可能 客户采集的不是全dmp文件或者dmp文件中的栈被破坏了。    这的确很伤脑筋，就针对这个我可是花了3个星期一行行的解析栈里面的内容 才解决的。 第二步、先用「!teb」看一下这个程序的栈是从哪里到哪里的。 0:000>!teb TEB at 000007ffffeee000     ExceptionList:        0000000000000000     StackBase:            0000000008d50000     StackLimit:           0000000008d4d000     SubSystemTib:         0000000000000000     FiberData:            0000000000001e00     ArbitraryUserPointer: 0000000000000000     Self:                 000007ffffeee000     EnvironmentPointer:   0000000000000000     ClientId:             0000000000001bdc . 0000000000001868     RpcHandle:            0000000000000000     Tls Storage:          000007ffffeee058     PEB Address:          000007fffffd6000     LastErrorValue:       87     LastStatusValue:      c000000d     Count Owned Locks:    0     HardErrorMode:        0 第三步、先用「dps」看一下这个程序的栈中的内存的内容。 下面截取其中比较重要的一段。 ------------------------------------------------------------------------------------------------------------------------------- 00000000`001247d8  00000000`74e6d5fa msvcr80!strncpy_s+0x10a [f:\dd\vctools\crt_bld\self_64_amd64\crt\src\tcsncpy_s.inl @ 62] 00000000`001247e0  00000000`009c01e0 00000000`001247e8  00000000`030f5810 00000000`001247f0  00000000`0057e310 ProcessB2!work　　　 ★「ProcessB2!work」的内容本应该是像这样的数据「DNxxxxxxxx_150_109」 但是现在「ProcessB2!work」中的内容却是「VIP_rtcrx00184-004a/b-y3b-d」这个。 00000000`001247f8  00000000`005782c0 ProcessB2!trcData　 ▲「ProcessB2!trcData」的内容是「Function:testB call」。  函数List::testB　の　trace("testB", __FILE__, __LINE__, TRCLV_3); 00000000`00124800  00000000`00000000 00000000`00124808  00000000`00000000 00000000`00124810  00000000`004a3150 ProcessB2!`string'　 ▲「 ProcessB2!`string'」的内容是「e:\ProcessB\FunctionB.cpp　　__FILE__」。 00000000`00124818  00000000`00455b65 ProcessB2!List::testB+0x55 [e:\ProcessB\Listset.cpp @ 719] 00000000`00124820  00000000`009c01e0 00000000`00124828  00000000`030f5810 00000000`00124830  00000000`0057e310 ProcessB2!work 00000000`00124838  00000000`001249e0 00000000`00124840  32322e35`322e3000 00000000`00124848  30614031`33312e34 00000000`00124850  7097fb8e`bc923730 00000000`00124858  5049565f`5753334c 00000000`00124860  00000000`0000125f 00000000`00124868  000082bd`b1200d5e 00000000`00124870  00000000`009c01e0 00000000`00124878  00000000`00467bda ProcessB2!FunctionB+0x73a [e:\ProcessB\FunctionB.cpp @ 181]    ------------------------------------------------------------------------------------------------------------------------------- 这里终于定位到是哪个函数出问题。搞清楚这些函数的功能，然后打印出所有可能打印的内容，发现 函数传递了一个不合法的数据。在这里要说一下为啥传的数据不合法就会Dmp掉。 首先strncpy 这个函数在使用的时候只要有个宏定义(默认是有的)在编译的时候就会使用strncpy_s这个安全函数。 详情可以参考下面微软的说明文档。 http://msdn.microsoft.com/zh-cn/LIBRARY/ms175759(v=vs.80) 其次说明一下为什么会dmp掉。strncpy在使用的时候如果转化成strncpy_s的时候是这样一种形式。 char dst[5]; strncpy(dst, "a long string", 5);    ---->  strncpy_s(dst, 5, "a long string", 5); 而这样就会到时报STATUS_INVALID_PARAMETER这个错误这是strncpy_s的特性。具体使用方法可以参考下面的文档。 http://msdn.microsoft.com/zh-cn/library/5dae5d43(v=vs.90).aspx 节选: char dst[5]; strncpy_s(dst, 5, "a long string", 5); means that we are asking strncpy_s to copy five characters into a buffer five bytes long; this would leave no space for the null terminator, hence strncpy_s zeroes out the string and calls the invalid parameter handler. If truncation behavior is needed, use _TRUNCATE or (size – 1): strncpy_s(dst, 5, "a long string", _TRUNCATE); strncpy_s(dst, 5, "a long string", 4);

详细的ACTIONABLE_HEAP_CORRUPTION_heap_failure_buffer_overrun方法还可以参考以下的例子:

http://blogs.msdn.com/b/jiangyue/archive/2010/03/16/windows-heap-overrun-monitoring.aspx