crashpad/util/win/registration_protocol_win.h

131 lines
4.5 KiB
C
Raw Normal View History

win: Crash handler server This replaces the registration server, and adds dispatch to a delegate on crash requests. (As you are already aware) we went around in circles on trying to come up with a slightly-too-fancy threading design. All of them seemed to have problems when it comes to out of order events, and orderly shutdown, so I've gone back to something not-too-fancy. Two named pipe instances (that clients connect to) are created. These are used only for registration (which should take <1ms), so 2 should be sufficient to avoid any waits. When a client registers, we duplicate an event to it, which is used to signal when it wants a dump taken. The server registers threadpool waits on that event, and also on the process handle (which will be signalled when the client process exits). These requests (in particular the taking of the dump) are serviced on the threadpool, which avoids us needing to manage those threads, but still allows parallelism in taking dumps. On process termination, we use an IO Completion Port to post a message back to the main thread to request cleanup. This complexity is necessary so that we can unregister the threadpool waits without being on the threadpool, which we need to do synchronously so that we can be sure that no further callbacks will execute (and expect to have the client data around still). In a followup, I will readd support for DumpWithoutCrashing -- I don't think it will be too difficult now that we have an orderly way to clean up client records in the server. R=cpu@chromium.org, mark@chromium.org, jschuh@chromium.org BUG=crashpad:1,crashpad:45 Review URL: https://codereview.chromium.org/1301853002 .
2015-09-03 11:06:17 -07:00
// Copyright 2015 The Crashpad Authors. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#ifndef CRASHPAD_UTIL_WIN_REGISTRATION_PROTOCOL_WIN_H_
#define CRASHPAD_UTIL_WIN_REGISTRATION_PROTOCOL_WIN_H_
#include <windows.h>
#include <stdint.h>
#include "base/strings/string16.h"
#include "util/win/address_types.h"
namespace crashpad {
#pragma pack(push, 1)
//! \brief Structure read out of the client process by the crash handler when an
//! exception occurs.
struct ExceptionInformation {
//! \brief The address of an EXCEPTION_POINTERS structure in the client
//! process that describes the exception.
WinVMAddress exception_pointers;
//! \brief The thread on which the exception happened.
DWORD thread_id;
};
//! \brief A client registration request.
struct RegistrationRequest {
//! \brief The expected value of `version`. This should be changed whenever
//! the messages or ExceptionInformation are modified incompatibly.
enum { kMessageVersion = 1 };
//! \brief Version field to detect skew between client and server. Should be
//! set to kMessageVersion.
int version;
win: Crash handler server This replaces the registration server, and adds dispatch to a delegate on crash requests. (As you are already aware) we went around in circles on trying to come up with a slightly-too-fancy threading design. All of them seemed to have problems when it comes to out of order events, and orderly shutdown, so I've gone back to something not-too-fancy. Two named pipe instances (that clients connect to) are created. These are used only for registration (which should take <1ms), so 2 should be sufficient to avoid any waits. When a client registers, we duplicate an event to it, which is used to signal when it wants a dump taken. The server registers threadpool waits on that event, and also on the process handle (which will be signalled when the client process exits). These requests (in particular the taking of the dump) are serviced on the threadpool, which avoids us needing to manage those threads, but still allows parallelism in taking dumps. On process termination, we use an IO Completion Port to post a message back to the main thread to request cleanup. This complexity is necessary so that we can unregister the threadpool waits without being on the threadpool, which we need to do synchronously so that we can be sure that no further callbacks will execute (and expect to have the client data around still). In a followup, I will readd support for DumpWithoutCrashing -- I don't think it will be too difficult now that we have an orderly way to clean up client records in the server. R=cpu@chromium.org, mark@chromium.org, jschuh@chromium.org BUG=crashpad:1,crashpad:45 Review URL: https://codereview.chromium.org/1301853002 .
2015-09-03 11:06:17 -07:00
//! \brief The PID of the client process.
DWORD client_process_id;
//! \brief The address, in the client process address space, of an
//! ExceptionInformation structure, used when handling a crash dump
//! request.
WinVMAddress crash_exception_information;
//! \brief The address, in the client process address space, of an
//! ExceptionInformation structure, used when handling a non-crashing dump
//! request.
WinVMAddress non_crash_exception_information;
win: Crash handler server This replaces the registration server, and adds dispatch to a delegate on crash requests. (As you are already aware) we went around in circles on trying to come up with a slightly-too-fancy threading design. All of them seemed to have problems when it comes to out of order events, and orderly shutdown, so I've gone back to something not-too-fancy. Two named pipe instances (that clients connect to) are created. These are used only for registration (which should take <1ms), so 2 should be sufficient to avoid any waits. When a client registers, we duplicate an event to it, which is used to signal when it wants a dump taken. The server registers threadpool waits on that event, and also on the process handle (which will be signalled when the client process exits). These requests (in particular the taking of the dump) are serviced on the threadpool, which avoids us needing to manage those threads, but still allows parallelism in taking dumps. On process termination, we use an IO Completion Port to post a message back to the main thread to request cleanup. This complexity is necessary so that we can unregister the threadpool waits without being on the threadpool, which we need to do synchronously so that we can be sure that no further callbacks will execute (and expect to have the client data around still). In a followup, I will readd support for DumpWithoutCrashing -- I don't think it will be too difficult now that we have an orderly way to clean up client records in the server. R=cpu@chromium.org, mark@chromium.org, jschuh@chromium.org BUG=crashpad:1,crashpad:45 Review URL: https://codereview.chromium.org/1301853002 .
2015-09-03 11:06:17 -07:00
};
//! \brief A message only sent to the server by itself to trigger shutdown.
struct ShutdownRequest {
//! \brief A randomly generated token used to validate the the shutdown
//! request was not sent from another process.
uint64_t token;
};
//! \brief The message passed from client to server by
//! SendToCrashHandlerServer().
struct ClientToServerMessage {
//! \brief Indicates which field of the union is in use.
enum Type : uint32_t {
//! \brief For RegistrationRequest.
kRegister,
//! \brief For ShutdownRequest.
kShutdown,
} type;
union {
RegistrationRequest registration;
ShutdownRequest shutdown;
};
};
//! \brief A client registration response.
//!
//! See <a
//! href="https://msdn.microsoft.com/en-us/library/windows/desktop/aa384203">Interprocess
//! Communication Between 32-bit and 64-bit Applications</a> for details on
//! communicating handle values between processes of varying bitness.
struct RegistrationResponse {
//! \brief An event `HANDLE`, valid in the client process, that should be
//! signaled to request a crash report. 64-bit clients should convert the
//! value to a `HANDLE` using sign-extension.
uint32_t request_crash_dump_event;
//! \brief An event `HANDLE`, valid in the client process, that should be
//! signaled to request a non-crashing dump be taken. 64-bit clients
//! should convert the value to `HANDLEEE` using sign-extension.
uint32_t request_non_crash_dump_event;
//! \brief An event `HANDLE`, valid in the client process, that will be
//! signaled by the server when the non-crashing dump is complete. 64-bit
//! clients should convert the value to `HANDLEEE` using sign-extension.
uint32_t non_crash_dump_completed_event;
win: Crash handler server This replaces the registration server, and adds dispatch to a delegate on crash requests. (As you are already aware) we went around in circles on trying to come up with a slightly-too-fancy threading design. All of them seemed to have problems when it comes to out of order events, and orderly shutdown, so I've gone back to something not-too-fancy. Two named pipe instances (that clients connect to) are created. These are used only for registration (which should take <1ms), so 2 should be sufficient to avoid any waits. When a client registers, we duplicate an event to it, which is used to signal when it wants a dump taken. The server registers threadpool waits on that event, and also on the process handle (which will be signalled when the client process exits). These requests (in particular the taking of the dump) are serviced on the threadpool, which avoids us needing to manage those threads, but still allows parallelism in taking dumps. On process termination, we use an IO Completion Port to post a message back to the main thread to request cleanup. This complexity is necessary so that we can unregister the threadpool waits without being on the threadpool, which we need to do synchronously so that we can be sure that no further callbacks will execute (and expect to have the client data around still). In a followup, I will readd support for DumpWithoutCrashing -- I don't think it will be too difficult now that we have an orderly way to clean up client records in the server. R=cpu@chromium.org, mark@chromium.org, jschuh@chromium.org BUG=crashpad:1,crashpad:45 Review URL: https://codereview.chromium.org/1301853002 .
2015-09-03 11:06:17 -07:00
};
//! \brief The response sent back to the client via SendToCrashHandlerServer().
union ServerToClientMessage {
RegistrationResponse registration;
};
#pragma pack(pop)
//! \brief Connect over the given \a pipe_name, passing \a message to the
//! server, storing the server's reply into \a response.
//!
//! Typically clients will not use this directly, instead using
//! CrashpadClient::SetHandler().
//!
//! \sa CrashpadClient::SetHandler()
bool SendToCrashHandlerServer(const base::string16& pipe_name,
const ClientToServerMessage& message,
ServerToClientMessage* response);
} // namespace crashpad
#endif // CRASHPAD_UTIL_WIN_REGISTRATION_PROTOCOL_WIN_H_