crashpad/test/mac/exception_swallower.cc

194 lines
6.9 KiB
C++
Raw Normal View History

mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
// Copyright 2017 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.
#include "test/mac/exception_swallower.h"
#include <errno.h>
#include <stdlib.h>
#include <unistd.h>
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
#include <string>
#include "base/logging.h"
#include "base/mac/scoped_mach_port.h"
#include "base/strings/stringprintf.h"
#include "handler/mac/exception_handler_server.h"
#include "util/mach/exc_server_variants.h"
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
#include "util/mach/exception_ports.h"
#include "util/mach/mach_extensions.h"
#include "util/misc/random_string.h"
#include "util/thread/thread.h"
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
namespace crashpad {
namespace test {
namespace {
constexpr char kServiceEnvironmentVariable[] =
"CRASHPAD_EXCEPTION_SWALLOWER_SERVICE";
ExceptionSwallower* g_exception_swallower;
// Like getenv(), but fails a CHECK() if the underlying function fails. Its not
// considered a failure for |name| to be unset in the environment. In that case,
// nullptr is returned.
const char* CheckedGetenv(const char* name) {
errno = 0;
const char* value;
PCHECK((value = getenv(name)) || errno == 0) << "getenv";
return value;
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
} // namespace
class ExceptionSwallower::ExceptionSwallowerThread
: public Thread,
public UniversalMachExcServer::Interface {
public:
explicit ExceptionSwallowerThread(
base::mac::ScopedMachReceiveRight receive_right)
: Thread(),
UniversalMachExcServer::Interface(),
exception_handler_server_(std::move(receive_right), true),
pid_(getpid()) {
Start();
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
~ExceptionSwallowerThread() override {}
void Stop() { exception_handler_server_.Stop(); }
// Returns the process ID that the thread is running in. This is used to
// detect misuses that place the exception swallower server thread and code
// that wants its exceptions swallowed in the same process.
pid_t ProcessID() const { return pid_; }
private:
// Thread:
void ThreadMain() override { exception_handler_server_.Run(this); }
// UniversalMachExcServer::Interface:
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
kern_return_t CatchMachException(exception_behavior_t behavior,
exception_handler_t exception_port,
thread_t thread,
task_t task,
exception_type_t exception,
const mach_exception_data_type_t* code,
mach_msg_type_number_t code_count,
thread_state_flavor_t* flavor,
ConstThreadState old_state,
mach_msg_type_number_t old_state_count,
thread_state_t new_state,
mach_msg_type_number_t* new_state_count,
const mach_msg_trailer_t* trailer,
bool* destroy_complex_request) override {
*destroy_complex_request = true;
// Swallow.
ExcServerCopyState(
behavior, old_state, old_state_count, new_state, new_state_count);
return ExcServerSuccessfulReturnValue(exception, behavior, false);
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
ExceptionHandlerServer exception_handler_server_;
pid_t pid_;
DISALLOW_COPY_AND_ASSIGN(ExceptionSwallowerThread);
};
ExceptionSwallower::ExceptionSwallower() : exception_swallower_thread_() {
CHECK(!g_exception_swallower);
g_exception_swallower = this;
if (CheckedGetenv(kServiceEnvironmentVariable)) {
// The environment variable is already set, so just proceed with the
// existing service. This normally happens when the gtest “threadsafe” death
// test style is chosen, because the test child process will re-execute code
// already run in the test parent process. See
// https://github.com/google/googletest/blob/master/googletest/docs/AdvancedGuide.md#death-test-styles.
return;
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
std::string service_name =
base::StringPrintf("org.chromium.crashpad.test.exception_swallower.%d.%s",
getpid(),
RandomString().c_str());
base::mac::ScopedMachReceiveRight receive_right(
BootstrapCheckIn(service_name));
CHECK(receive_right.is_valid());
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
exception_swallower_thread_.reset(
new ExceptionSwallowerThread(std::move(receive_right)));
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
PCHECK(setenv(kServiceEnvironmentVariable, service_name.c_str(), 1) == 0)
<< "setenv";
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
ExceptionSwallower::~ExceptionSwallower() {
PCHECK(unsetenv(kServiceEnvironmentVariable) == 0) << "unsetenv";
exception_swallower_thread_->Stop();
exception_swallower_thread_->Join();
CHECK_EQ(g_exception_swallower, this);
g_exception_swallower = nullptr;
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
}
// static
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
void ExceptionSwallower::SwallowExceptions() {
// The exception swallower thread cant be in this process, because the
// EXC_CRASH or EXC_CORPSE_NOTIFY exceptions that it needs to swallow will be
// delivered after a crash has occurred and none of its threads will be
// scheduled to run.
CHECK(!g_exception_swallower ||
!g_exception_swallower->exception_swallower_thread_ ||
g_exception_swallower->exception_swallower_thread_->ProcessID() !=
getpid());
const char* service_name = CheckedGetenv(kServiceEnvironmentVariable);
CHECK(service_name);
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
base::mac::ScopedMachSendRight exception_swallower_port(
BootstrapLookUp(service_name));
mac: Tests that crash intentionally shouldn’t go to ReportCrash Crashpad has many tests that crash intentionally. Some of these are gtest death tests, and others arrange for intentional crashes to test Crashpad’s own crash-catching logic. On macOS, all of the gtest death tests and some of the other intentional crashes were being logged by ReportCrash, the system’s crash reporter. Since these reports corresponded to intentional crashes, they were never useful, and served only to clutter ~/Library/Logs/DiagnosticReports. Since Crashpad is adept at handling exceptions on its own, this introduces the “exception swallowing server”, crashpad_exception_swallower, which is a Mach exception server that implements a no-op exception handler routine for all exceptions received. The exception swallowing server is established as the task handler for EXC_CRASH and EXC_CORPSE_NOTIFY exceptions during gtest death tests invoked by {ASSERT,EXPECT}_DEATH_{CHECK,CRASH}, and for all child processes invoked by the Multiprocess test infrastructure. The exception swallowing server is not in effect at other times, so unexpected crashes in test code can still be handled by ReportCrash or another crash reporter. With this change in place, no new reports are generated in the user-level ~/Library/Logs/DiagnosticReports or the system’s /Library/Logs/DiagnosticReports during a run of Crashpad’s full test suite on macOS. Bug: crashpad:33 Change-Id: I13891853a7e25accc30da21fa7ea8bd7d1f3bd2f Reviewed-on: https://chromium-review.googlesource.com/777859 Commit-Queue: Mark Mentovai <mark@chromium.org> Reviewed-by: Robert Sesek <rsesek@chromium.org>
2017-11-20 13:32:26 -05:00
CHECK(exception_swallower_port.is_valid());
ExceptionPorts task_exception_ports(ExceptionPorts::kTargetTypeTask,
TASK_NULL);
// The mask is similar to the one used by CrashpadClient::UseHandler(), but
// EXC_CORPSE_NOTIFY is added. This is done for the benefit of tests that
// crash intentionally with their own custom exception port set for EXC_CRASH.
// In that case, depending on the actions taken by the EXC_CRASH handler, the
// exception may be transformed by the kernel into an EXC_CORPSE_NOTIFY, which
// would be sent to an EXC_CORPSE_NOTIFY handler, normally the systems crash
// reporter at the task or host level. See 10.13.0
// xnu-4570.1.46/bsd/kern/kern_exit.c proc_prepareexit(). Swallowing
// EXC_CORPSE_NOTIFY at the task level prevents such exceptions from reaching
// the systems crash reporter.
CHECK(task_exception_ports.SetExceptionPort(
(EXC_MASK_CRASH |
EXC_MASK_RESOURCE |
EXC_MASK_GUARD |
EXC_MASK_CORPSE_NOTIFY) & ExcMaskValid(),
exception_swallower_port.get(),
EXCEPTION_DEFAULT,
THREAD_STATE_NONE));
}
} // namespace test
} // namespace crashpad