参考资料:
Windows平台编译使用google breakpad
Google Breakpad 学习笔记
linux Qt中使用Google Breakpad捕获程序崩溃异常
google/breakpad: Mirror of Google Breakpad project
下载breakpad源码
git clone https://chromium.googlesource.com/breakpad/breakpad
获取gyp工具
git clone https://chromium.googlesource.com/external/gyp
与参考资料不同,获取到的gyp文件夹似乎并不需要复制到breakpad\src\tools\文件夹下,后续操作中注意各自所在路径,直接运行即可。
安装python2
用gyp打包breakpad时需要python2支持,直接官网下载即可,添加到PATH。(如同时已安装python3可能需要保证默认python执行python2,似乎后安装python2就会默认python2(存疑?)。)
获取googletest
git clone https://github.com/google/googletest.git
在breakpad\src\目录下,新建名为testing的文件夹,将获取到的googletest文件夹下的两个文件夹(googlemock和googletest)复制到testing文件夹中。
用gyp构建出VS工程文件
笔者写作时为VS2017工程文件。
gyp\gyp.bat --no-circular-check breakpad\src\client\windows\breakpad_client.gyp
测试自带示例程序
用Visual Studio打开breakpad\src\client\windows\breakpad_client.sln
直接右击build_all点击生成将出现错误,大概是:“警告被视为错误 没有生成object文件”。
将crash_generation_client和client_tests项目属性——配置属性——C/C++——常规——将警告视为错误,改为否。
重新生成build_all,与参考资料不同笔者此处可以生成成功。
测试:运行产生崩溃的程序
运行产生崩溃的程序,位于breakpad\src\client\windows\Debug\crash_generation_app.exe
点击Client——Deref Zero,程序立即崩溃。崩溃后在C:\Dumps文件夹产生名字类似于9a6851c6-a849-475a-831e-27681a273b2b.dmp的崩溃转储文件。
测试:通过崩溃转储文件进行调试
用VS打开崩溃转储文件,点击右侧:操作——使用 仅限本机 进行调试。调试效果类似于使用VS直接进行调试。图为VS2017(VS2019亦可)
应用
可以直接使用之前生成的breakpad\src\client\windows\Debug\lib中的
common.lib
crash_generation_client.lib
crash_generation_server.lib
exception_handler.lib
库文件
可以复制到当前工程中。如breakpad\lib
加入链接器——输入——附加依赖项
包含目录
可以复制breakpad\src\加入工程文件夹如breakpad\src
加入项目包含目录
运行库设置
代码生成运行库改为多线程调试(与breakpad生成时相同)
测试文件GoogleBreakpad.cpp
#include <cstdio>
#include "client/windows/handler/exception_handler.h"
bool callback(const wchar_t *dump_path, const wchar_t *id,
void *context, EXCEPTION_POINTERS *exinfo,
MDRawAssertionInfo *assertion,
bool succeeded)
{
if (succeeded) {
printf("dump guid is %ws\n", id);
}
else {
printf("dump failed\n");
}
system("pause");
return succeeded;
}
int mydiv(int x, int y)
{
int z;
z = x / y;
return z;
}
int main()
{
google_breakpad::ExceptionHandler eh(
L".", NULL, callback, NULL,
google_breakpad::ExceptionHandler::HANDLER_ALL);
printf("9/3=%d\n", mydiv(9, 3));
printf("9/0=%d\n", mydiv(9, 0)); //程序将在此崩溃
printf("8/2=%d\n", mydiv(8, 2));
system("pause");
return 0;
}
exception_handler.h
// Copyright (c) 2006, Google Inc.
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
// * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
版权声明翻译(机翻,下同)
版权所有(c)2006,谷歌公司。
保留所有权利。
在满足以下条件的情况下,允许以源和二进制形式重新分发和使用,无论是否进行修改:
*源代码的重新发布必须保留上述版权声明、此条件列表和以下免责声明。
*以二进制形式重新分发必须在分发时提供的文档和/或其他材料中复制上述版权声明、本条件列表和以下免责声明。
*未经事先书面许可,不得使用谷歌公司的名称或其贡献者的名称来支持或推广从本软件衍生的产品。
该软件由版权持有人和贡献者提供,如“是”和任何明示或暗示的保证,包括但不限于,适销性和适合特定目的的默示保证被驳回。在任何情况下,版权所有人或贡献者均不对任何直接、间接、附带、特殊、惩戒性或后果性损害(包括但不限于替代货物或服务的采购;使用、数据或利润的损失;或业务中断)承担责任,无论该损害是由何种原因引起的,也不论其责任理论是否在合同中,严格责任,或因使用本软件而产生的侵权行为(包括疏忽或其他),即使被告知此类损害的可能性。
版权协议类似BSD参考资料
// ExceptionHandler can write a minidump file when an exception occurs,
// or when WriteMinidump() is called explicitly by your program.
//
// To have the exception handler write minidumps when an uncaught exception
// (crash) occurs, you should create an instance early in the execution
// of your program, and keep it around for the entire time you want to
// have crash handling active (typically, until shutdown).
//
// If you want to write minidumps without installing the exception handler,
// you can create an ExceptionHandler with install_handler set to false,
// then call WriteMinidump. You can also use this technique if you want to
// use different minidump callbacks for different call sites.
//
// In either case, a callback function is called when a minidump is written,
// which receives the unqiue id of the minidump. The caller can use this
// id to collect and write additional application state, and to launch an
// external crash-reporting application.
//
// It is important that creation and destruction of ExceptionHandler objects
// be nested cleanly, when using install_handler = true.
// Avoid the following pattern:
// ExceptionHandler *e = new ExceptionHandler(...);
// ExceptionHandler *f = new ExceptionHandler(...);
// delete e;
// This will put the exception filter stack into an inconsistent state.
ExceptionHandler可以在发生异常或程序显式调用WriteMinidump()时写入小型转储文件。
要让异常处理程序在发生未捕获的异常(崩溃)时写入小型转储,您应该在程序执行的早期创建一个实例,并在希望激活崩溃处理的整个时间(通常,直到关闭)内保持该实例。
如果要在不安装异常处理程序的情况下编写小型转储,可以创建一个将install_handler设置为false的ExceptionHandler,然后调用WriteMinidump。如果要对不同的调用站点使用不同的小型转储回调,也可以使用此技术。
在这两种情况下,当写入小型转储时调用回调函数,该函数接收小型转储的unqiue id。调用方可以使用此id收集和写入其他应用程序状态,并启动外部故障报告应用程序。
当使用install_handler=true时,创建和销毁ExceptionHandler对象的嵌套必须干净。
避免以下模式:
ExceptionHandler *e = new ExceptionHandler(…);
ExceptionHandler *f = new ExceptionHandler(…);
delete e;
这将使异常筛选器堆栈处于不一致状态。
#ifndef CLIENT_WINDOWS_HANDLER_EXCEPTION_HANDLER_H__
#define CLIENT_WINDOWS_HANDLER_EXCEPTION_HANDLER_H__
#include <stdlib.h>
#include <windows.h>
#include <dbghelp.h>
#include <rpc.h>
#pragma warning(push)
// Disable exception handler warnings.
// 禁用异常处理程序警告。
#pragma warning(disable:4530)
#include <list>
#include <string>
#include <vector>
#include "client/windows/common/ipc_protocol.h"
#include "client/windows/crash_generation/crash_generation_client.h"
#include "common/scoped_ptr.h"
#include "google_breakpad/common/minidump_format.h"
namespace google_breakpad {
using std::vector;
using std::wstring;
// These entries store a list of memory regions that the client wants included
// in the minidump.
// 这些条目存储客户端希望包含在小型转储中的内存区域列表。
struct AppMemory {
ULONG64 ptr;
ULONG length;
bool operator==(const struct AppMemory& other) const {
return ptr == other.ptr;
}
bool operator==(const void* other) const {
return ptr == reinterpret_cast<ULONG64>(other);
}
};
typedef std::list<AppMemory> AppMemoryList;
class ExceptionHandler {
public:
// A callback function to run before Breakpad performs any substantial
// processing of an exception. A FilterCallback is called before writing
// a minidump. context is the parameter supplied by the user as
// callback_context when the handler was created. exinfo points to the
// exception record, if any; assertion points to assertion information,
// if any.
// 在Breakpad执行异常的任何实质性处理之前运行的回调函数。
// 在写入小型转储之前调用FilterCallback。
// context是用户在创建处理程序时作为回调上下文提供的参数。
// exinfo指向异常记录(如果有的话);assertion指向断言信息(如果有的话)。
//
// If a FilterCallback returns true, Breakpad will continue processing,
// attempting to write a minidump. If a FilterCallback returns false,
// Breakpad will immediately report the exception as unhandled without
// writing a minidump, allowing another handler the opportunity to handle it.
// 如果FilterCallback返回true,Breakpad将继续处理,并尝试写入小型转储。
// 如果FilterCallback返回false,Breakpad将立即将异常报告为未处理,
// 而不写入小型转储,从而允许其他处理程序处理该异常。
typedef bool (*FilterCallback)(void* context, EXCEPTION_POINTERS* exinfo,
MDRawAssertionInfo* assertion);
// A callback function to run after the minidump has been written.
// minidump_id is a unique id for the dump, so the minidump
// file is <dump_path>\<minidump_id>.dmp. context is the parameter supplied
// by the user as callback_context when the handler was created. exinfo
// points to the exception record, or NULL if no exception occurred.
// succeeded indicates whether a minidump file was successfully written.
// assertion points to information about an assertion if the handler was
// invoked by an assertion.
// 写入小型转储后要运行的回调函数。minidump_id是转储的唯一id,因此minidump文件是<dump_path>\<minidump_id>.dmp。
// context是用户在创建处理程序时作为callback_context提供的参数。
// exinfo指向异常记录,如果没有异常发生,则为NULL。succeeded指示是否已成功写入小型转储文件。
// 如果断言调用了处理程序,则assertion指向有关断言的信息。
//
// If an exception occurred and the callback returns true, Breakpad will treat
// the exception as fully-handled, suppressing any other handlers from being
// notified of the exception. If the callback returns false, Breakpad will
// treat the exception as unhandled, and allow another handler to handle it.
// If there are no other handlers, Breakpad will report the exception to the
// system as unhandled, allowing a debugger or native crash dialog the
// opportunity to handle the exception. Most callback implementations
// should normally return the value of |succeeded|, or when they wish to
// not report an exception of handled, false. Callbacks will rarely want to
// return true directly (unless |succeeded| is true).
// 如果发生异常并且回调返回true,则Breakpad将把异常视为已完全处理,从而禁止任何其他处理程序收到异常通知。
// 如果回调返回false,Breakpad会将异常视为未处理,并允许另一个处理程序处理它。
// 如果没有其他处理程序,Breakpad会将异常作为未处理的异常报告给系统,允许调试器或本机崩溃对话框处理异常。
// 大多数回调实现通常应该返回| succeeded |的值,或者当它们不希望报告处理过的异常时,返回false。
// 回调很少希望直接返回true(除非| succeeded |为true)。
//
// For out-of-process dump generation, dump path and minidump ID will always
// be NULL. In case of out-of-process dump generation, the dump path and
// minidump id are controlled by the server process and are not communicated
// back to the crashing process.
// 对于进程外转储生成,dump_path和minidump_id将始终为NULL。
// 在进程外转储生成的情况下,dump_path和minidump_id由服务器进程控制,不会与崩溃进程通信。
typedef bool (*MinidumpCallback)(const wchar_t* dump_path,
const wchar_t* minidump_id,
void* context,
EXCEPTION_POINTERS* exinfo,
MDRawAssertionInfo* assertion,
bool succeeded);
// HandlerType specifies which types of handlers should be installed, if
// any. Use HANDLER_NONE for an ExceptionHandler that remains idle,
// without catching any failures on its own. This type of handler may
// still be triggered by calling WriteMinidump. Otherwise, use a
// combination of the other HANDLER_ values, or HANDLER_ALL to install
// all handlers.
// HandlerType指定应设置哪些类型的处理程序(如果有)。
// 对保持空闲的异常处理程序使用HANDLER_NONE,而不会单独捕获任何失败。
// 这种类型的处理程序仍然可以通过调用WriteMinidump来触发。
// 否则,请使用其他处理程序值或处理程序全部的组合来设置所有处理程序。
enum HandlerType {
HANDLER_NONE = 0,
HANDLER_EXCEPTION = 1 << 0, // SetUnhandledExceptionFilter
HANDLER_INVALID_PARAMETER = 1 << 1, // _set_invalid_parameter_handler
HANDLER_PURECALL = 1 << 2, // _set_purecall_handler
HANDLER_ALL = HANDLER_EXCEPTION |
HANDLER_INVALID_PARAMETER |
HANDLER_PURECALL
};
// Creates a new ExceptionHandler instance to handle writing minidumps.
// Before writing a minidump, the optional filter callback will be called.
// Its return value determines whether or not Breakpad should write a
// minidump. Minidump files will be written to dump_path, and the optional
// callback is called after writing the dump file, as described above.
// handler_types specifies the types of handlers that should be installed.
// 创建一个新的ExceptionHandler实例来处理写入小型转储。
// 在编写小型转储之前,将调用可选的筛选器回调filter。
// 它的返回值决定Breakpad是否应该写一个小型转储。
// 小型转储文件将写入dump_path,并在写入转储文件后调用可选回调callback,如上所述。
// handler_types指定应设置的处理程序类型。
ExceptionHandler(const wstring& dump_path,
FilterCallback filter,
MinidumpCallback callback,
void* callback_context,
int handler_types);
// Creates a new ExceptionHandler instance that can attempt to perform
// out-of-process dump generation if pipe_name is not NULL. If pipe_name is
// NULL, or if out-of-process dump generation registration step fails,
// in-process dump generation will be used. This also allows specifying
// the dump type to generate.
ExceptionHandler(const wstring& dump_path,
FilterCallback filter,
MinidumpCallback callback,
void* callback_context,
int handler_types,
MINIDUMP_TYPE dump_type,
const wchar_t* pipe_name,
const CustomClientInfo* custom_info);
// As above, creates a new ExceptionHandler instance to perform
// out-of-process dump generation if the given pipe_handle is not NULL.
ExceptionHandler(const wstring& dump_path,
FilterCallback filter,
MinidumpCallback callback,
void* callback_context,
int handler_types,
MINIDUMP_TYPE dump_type,
HANDLE pipe_handle,
const CustomClientInfo* custom_info);
// ExceptionHandler that ENSURES out-of-process dump generation. Expects a
// crash generation client that is already registered with a crash generation
// server. Takes ownership of the passed-in crash_generation_client.
//
// Usage example:
// crash_generation_client = new CrashGenerationClient(..);
// if (crash_generation_client->Register()) {
// // Registration with the crash generation server succeeded.
// // Out-of-process dump generation is guaranteed.
// g_handler = new ExceptionHandler(.., crash_generation_client, ..);
// return true;
// }
ExceptionHandler(const wstring& dump_path,
FilterCallback filter,
MinidumpCallback callback,
void* callback_context,
int handler_types,
CrashGenerationClient* crash_generation_client);
~ExceptionHandler();
// Get and set the minidump path.
wstring dump_path() const { return dump_path_; }
void set_dump_path(const wstring &dump_path) {
dump_path_ = dump_path;
dump_path_c_ = dump_path_.c_str();
UpdateNextID(); // Necessary to put dump_path_ in next_minidump_path_.
}
// Requests that a previously reported crash be uploaded.
bool RequestUpload(DWORD crash_id);
// Writes a minidump immediately. This can be used to capture the
// execution state independently of a crash. Returns true on success.
bool WriteMinidump();
// Writes a minidump immediately, with the user-supplied exception
// information.
bool WriteMinidumpForException(EXCEPTION_POINTERS* exinfo);
// Convenience form of WriteMinidump which does not require an
// ExceptionHandler instance.
static bool WriteMinidump(const wstring &dump_path,
MinidumpCallback callback, void* callback_context,
MINIDUMP_TYPE dump_type = MiniDumpNormal);
// Write a minidump of |child| immediately. This can be used to
// capture the execution state of |child| independently of a crash.
// Pass a meaningful |child_blamed_thread| to make that thread in
// the child process the one from which a crash signature is
// extracted.
static bool WriteMinidumpForChild(HANDLE child,
DWORD child_blamed_thread,
const wstring& dump_path,
MinidumpCallback callback,
void* callback_context,
MINIDUMP_TYPE dump_type = MiniDumpNormal);
// Get the thread ID of the thread requesting the dump (either the exception
// thread or any other thread that called WriteMinidump directly). This
// may be useful if you want to include additional thread state in your
// dumps.
DWORD get_requesting_thread_id() const { return requesting_thread_id_; }
// Controls behavior of EXCEPTION_BREAKPOINT and EXCEPTION_SINGLE_STEP.
bool get_handle_debug_exceptions() const { return handle_debug_exceptions_; }
void set_handle_debug_exceptions(bool handle_debug_exceptions) {
handle_debug_exceptions_ = handle_debug_exceptions;
}
// Controls behavior of EXCEPTION_INVALID_HANDLE.
bool get_consume_invalid_handle_exceptions() const {
return consume_invalid_handle_exceptions_;
}
void set_consume_invalid_handle_exceptions(
bool consume_invalid_handle_exceptions) {
consume_invalid_handle_exceptions_ = consume_invalid_handle_exceptions;
}
// Returns whether out-of-process dump generation is used or not.
bool IsOutOfProcess() const { return crash_generation_client_.get() != NULL; }
// Calling RegisterAppMemory(p, len) causes len bytes starting
// at address p to be copied to the minidump when a crash happens.
void RegisterAppMemory(void* ptr, size_t length);
void UnregisterAppMemory(void* ptr);
private:
friend class AutoExceptionHandler;
// Initializes the instance with given values.
void Initialize(const wstring& dump_path,
FilterCallback filter,
MinidumpCallback callback,
void* callback_context,
int handler_types,
MINIDUMP_TYPE dump_type,
const wchar_t* pipe_name,
HANDLE pipe_handle,
CrashGenerationClient* crash_generation_client,
const CustomClientInfo* custom_info);
// Function pointer type for MiniDumpWriteDump, which is looked up
// dynamically.
typedef BOOL (WINAPI *MiniDumpWriteDump_type)(
HANDLE hProcess,
DWORD dwPid,
HANDLE hFile,
MINIDUMP_TYPE DumpType,
CONST PMINIDUMP_EXCEPTION_INFORMATION ExceptionParam,
CONST PMINIDUMP_USER_STREAM_INFORMATION UserStreamParam,
CONST PMINIDUMP_CALLBACK_INFORMATION CallbackParam);
// Function pointer type for UuidCreate, which is looked up dynamically.
typedef RPC_STATUS (RPC_ENTRY *UuidCreate_type)(UUID* Uuid);
// Runs the main loop for the exception handler thread.
static DWORD WINAPI ExceptionHandlerThreadMain(void* lpParameter);
// Called on the exception thread when an unhandled exception occurs.
// Signals the exception handler thread to handle the exception.
static LONG WINAPI HandleException(EXCEPTION_POINTERS* exinfo);
#if _MSC_VER >= 1400 // MSVC 2005/8
// This function will be called by some CRT functions when they detect
// that they were passed an invalid parameter. Note that in _DEBUG builds,
// the CRT may display an assertion dialog before calling this function,
// and the function will not be called unless the assertion dialog is
// dismissed by clicking "Ignore."
static void HandleInvalidParameter(const wchar_t* expression,
const wchar_t* function,
const wchar_t* file,
unsigned int line,
uintptr_t reserved);
#endif // _MSC_VER >= 1400
// This function will be called by the CRT when a pure virtual
// function is called.
static void HandlePureVirtualCall();
// This is called on the exception thread or on another thread that
// the user wishes to produce a dump from. It calls
// WriteMinidumpWithException on the handler thread, avoiding stack
// overflows and inconsistent dumps due to writing the dump from
// the exception thread. If the dump is requested as a result of an
// exception, exinfo contains exception information, otherwise, it
// is NULL. If the dump is requested as a result of an assertion
// (such as an invalid parameter being passed to a CRT function),
// assertion contains data about the assertion, otherwise, it is NULL.
bool WriteMinidumpOnHandlerThread(EXCEPTION_POINTERS* exinfo,
MDRawAssertionInfo* assertion);
// This function is called on the handler thread. It calls into
// WriteMinidumpWithExceptionForProcess() with a handle to the
// current process. requesting_thread_id is the ID of the thread
// that requested the dump. If the dump is requested as a result of
// an exception, exinfo contains exception information, otherwise,
// it is NULL.
bool WriteMinidumpWithException(DWORD requesting_thread_id,
EXCEPTION_POINTERS* exinfo,
MDRawAssertionInfo* assertion);
// This function is used as a callback when calling MinidumpWriteDump,
// in order to add additional memory regions to the dump.
static BOOL CALLBACK MinidumpWriteDumpCallback(
PVOID context,
const PMINIDUMP_CALLBACK_INPUT callback_input,
PMINIDUMP_CALLBACK_OUTPUT callback_output);
// This function does the actual writing of a minidump. It is
// called on the handler thread. requesting_thread_id is the ID of
// the thread that requested the dump, if that information is
// meaningful. If the dump is requested as a result of an
// exception, exinfo contains exception information, otherwise, it
// is NULL. process is the one that will be dumped. If
// requesting_thread_id is meaningful and should be added to the
// minidump, write_requester_stream is |true|.
bool WriteMinidumpWithExceptionForProcess(DWORD requesting_thread_id,
EXCEPTION_POINTERS* exinfo,
MDRawAssertionInfo* assertion,
HANDLE process,
bool write_requester_stream);
// Generates a new ID and stores it in next_minidump_id_, and stores the
// path of the next minidump to be written in next_minidump_path_.
void UpdateNextID();
FilterCallback filter_;
MinidumpCallback callback_;
void* callback_context_;
scoped_ptr<CrashGenerationClient> crash_generation_client_;
// The directory in which a minidump will be written, set by the dump_path
// argument to the constructor, or set_dump_path.
wstring dump_path_;
// The basename of the next minidump to be written, without the extension.
wstring next_minidump_id_;
// The full pathname of the next minidump to be written, including the file
// extension.
wstring next_minidump_path_;
// Pointers to C-string representations of the above. These are set when
// the above wstring versions are set in order to avoid calling c_str during
// an exception, as c_str may attempt to allocate heap memory. These
// pointers are not owned by the ExceptionHandler object, but their lifetimes
// should be equivalent to the lifetimes of the associated wstring, provided
// that the wstrings are not altered.
const wchar_t* dump_path_c_;
const wchar_t* next_minidump_id_c_;
const wchar_t* next_minidump_path_c_;
HMODULE dbghelp_module_;
MiniDumpWriteDump_type minidump_write_dump_;
MINIDUMP_TYPE dump_type_;
HMODULE rpcrt4_module_;
UuidCreate_type uuid_create_;
// Tracks the handler types that were installed according to the
// handler_types constructor argument.
int handler_types_;
// When installed_handler_ is true, previous_filter_ is the unhandled
// exception filter that was set prior to installing ExceptionHandler as
// the unhandled exception filter and pointing it to |this|. NULL indicates
// that there is no previous unhandled exception filter.
LPTOP_LEVEL_EXCEPTION_FILTER previous_filter_;
#if _MSC_VER >= 1400 // MSVC 2005/8
// Beginning in VC 8, the CRT provides an invalid parameter handler that will
// be called when some CRT functions are passed invalid parameters. In
// earlier CRTs, the same conditions would cause unexpected behavior or
// crashes.
_invalid_parameter_handler previous_iph_;
#endif // _MSC_VER >= 1400
// The CRT allows you to override the default handler for pure
// virtual function calls.
_purecall_handler previous_pch_;
// The exception handler thread.
HANDLE handler_thread_;
// True if the exception handler is being destroyed.
// Starting with MSVC 2005, Visual C has stronger guarantees on volatile vars.
// It has release semantics on write and acquire semantics on reads.
// See the msdn documentation.
volatile bool is_shutdown_;
// The critical section enforcing the requirement that only one exception be
// handled by a handler at a time.
CRITICAL_SECTION handler_critical_section_;
// Semaphores used to move exception handling between the exception thread
// and the handler thread. handler_start_semaphore_ is signalled by the
// exception thread to wake up the handler thread when an exception occurs.
// handler_finish_semaphore_ is signalled by the handler thread to wake up
// the exception thread when handling is complete.
HANDLE handler_start_semaphore_;
HANDLE handler_finish_semaphore_;
// The next 2 fields contain data passed from the requesting thread to
// the handler thread.
// The thread ID of the thread requesting the dump (either the exception
// thread or any other thread that called WriteMinidump directly).
DWORD requesting_thread_id_;
// The exception info passed to the exception handler on the exception
// thread, if an exception occurred. NULL for user-requested dumps.
EXCEPTION_POINTERS* exception_info_;
// If the handler is invoked due to an assertion, this will contain a
// pointer to the assertion information. It is NULL at other times.
MDRawAssertionInfo* assertion_;
// The return value of the handler, passed from the handler thread back to
// the requesting thread.
bool handler_return_value_;
// If true, the handler will intercept EXCEPTION_BREAKPOINT and
// EXCEPTION_SINGLE_STEP exceptions. Leave this false (the default)
// to not interfere with debuggers.
bool handle_debug_exceptions_;
// If true, the handler will consume any EXCEPTION_INVALID_HANDLE exceptions.
// Leave this false (the default) to handle these exceptions as normal.
bool consume_invalid_handle_exceptions_;
// Callers can request additional memory regions to be included in
// the dump.
AppMemoryList app_memory_info_;
// A stack of ExceptionHandler objects that have installed unhandled
// exception filters. This vector is used by HandleException to determine
// which ExceptionHandler object to route an exception to. When an
// ExceptionHandler is created with install_handler true, it will append
// itself to this list.
static vector<ExceptionHandler*>* handler_stack_;
// The index of the ExceptionHandler in handler_stack_ that will handle the
// next exception. Note that 0 means the last entry in handler_stack_, 1
// means the next-to-last entry, and so on. This is used by HandleException
// to support multiple stacked Breakpad handlers.
static LONG handler_stack_index_;
// handler_stack_critical_section_ guards operations on handler_stack_ and
// handler_stack_index_. The critical section is initialized by the
// first instance of the class and destroyed by the last instance of it.
static CRITICAL_SECTION handler_stack_critical_section_;
// The number of instances of this class.
static volatile LONG instance_count_;
// disallow copy ctor and operator=
explicit ExceptionHandler(const ExceptionHandler &);
void operator=(const ExceptionHandler &);
};
} // namespace google_breakpad
#pragma warning(pop)
#endif // CLIENT_WINDOWS_HANDLER_EXCEPTION_HANDLER_H__