crashpad/util/posix/spawn_subprocess.h

70 lines
3.5 KiB
C
Raw Normal View History

// Copyright 2017 The Crashpad Authors
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
//
// 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_POSIX_SPAWN_SUBPROCESS_H_
#define CRASHPAD_UTIL_POSIX_SPAWN_SUBPROCESS_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 <vector>
namespace crashpad {
//! \brief Spawns a subprocess.
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
//!
//! A grandchild process will be started through the
//! `fork()`-and-`posix_spawn()` pattern where supported, and
//! double-`fork()`-and-`execv()` pattern elsewhere. This allows the grandchild
//! to fully disassociate from the parent. The grandchild will not be a member
//! of the parents process group or session and will not have a controlling
//! terminal, providing isolation from signals not intended for it. The
//! grandchilds parent process, in terms of the process tree hierarchy, will be
//! the process with process ID 1, relieving any other process of the
//! responsibility to reap it via `waitpid()`. Aside from the three file
//! descriptors associated with the standard input/output streams and any file
//! descriptor passed in \a preserve_fd, the grandchild will not inherit any
//! file descriptors from the parent process.
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
//!
//! \param[in] argv The argument vector to start the grandchild process with.
//! `argv[0]` is used as the path to the executable.
//! \param[in] envp A vector of environment variables of the form `var=value` to
//! be passed to the spawned process. If this value is `nullptr`, the
//! current environment is used.
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
//! \param[in] preserve_fd A file descriptor to be inherited by the grandchild
//! process. This file descriptor is inherited in addition to the three file
//! descriptors associated with the standard input/output streams. Use `-1`
//! if no additional file descriptors are to be inherited.
//! \param[in] use_path Whether to consult the `PATH` environment variable when
//! requested to start an executable at a non-absolute path.
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
//! \param[in] child_function If not `nullptr`, this function will be called in
//! the intermediate child process. Take note that this function will run in
//! the context of a forked process, and must be safe for that purpose.
Revert "posix: Replace DoubleForkAndExec() with ForkAndSpawn()" This reverts commit 460943dd9a71dc76f68182a8ede766d5543e5341. Reason for revert: This fails to compile in Chromium Android. posix_spawn and posix_spawnp are available in Android NDK 28, but Chromium is building with version 23. https://ci.chromium.org/ui/p/chromium/builders/try/android_compile_dbg/1179765/overview Original change's description: > posix: Replace DoubleForkAndExec() with ForkAndSpawn() > > The DoubleForkAndExec() function was taking over 622 milliseconds to run > on macOS 11 (BigSur) on Intel i5-1038NG7. I did some debugging by adding > some custom traces and found that the fork() syscall is the bottleneck > here, i.e., the first fork() takes around 359 milliseconds and the > nested fork() takes around 263 milliseconds. Replacing the nested fork() > and exec() with posix_spawn() reduces the time consumption to 257 > milliseconds! > > See https://github.com/libuv/libuv/pull/3064 to know why fork() is so > slow on macOS and why posix_spawn() is a better replacement. > > Another point to note is that even base::LaunchProcess() from Chromium > calls posix_spawnp() on macOS - > https://source.chromium.org/chromium/chromium/src/+/8f8d82dea0fa8f11f57c74dbb65126f8daba58f7:base/process/launch_mac.cc;l=295-296 > > Change-Id: I25c6ee9629a1ae5d0c32b361b56a1ce0b4b0fd26 > Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3641386 > Reviewed-by: Mark Mentovai <mark@chromium.org> > Commit-Queue: Mark Mentovai <mark@chromium.org> Change-Id: I7f6161bc4734c50308438cdde1e193023ee9bfb8 Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3719439 Reviewed-by: Mark Mentovai <mark@chromium.org> Commit-Queue: Justin Cohen <justincohen@chromium.org>
2022-06-23 03:12:41 +00:00
//!
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
//! \return `true` on success, and `false` on failure with a message logged.
//! Only failures that occur in the parent process that indicate a definite
//! failure to start the the grandchild are reported in the return value.
//! Failures in the intermediate child or grandchild processes cannot be
//! reported in the return value, and are addressed by logging a message and
//! terminating. The caller assumes the responsibility for detecting such
//! failures, for example, by observing a failure to perform a successful
//! handshake with the grandchild process.
bool SpawnSubprocess(const std::vector<std::string>& argv,
const std::vector<std::string>* envp,
int preserve_fd,
bool use_path,
void (*child_function)());
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
#endif // CRASHPAD_UTIL_POSIX_SPAWN_SUBPROCESS_H_