ninja代码解析02-如何执行一条command

ninja如何执行一条编译指令，这个主要在builder.cc里的Build(string* err)里实现

这里主要关注一条edge的执行

bool Builder::StartEdge(Edge* edge, string* err) {
  // ... previous code
  // start command computing and run it
  if (!command_runner_->StartCommand(edge)) {
    err->assign("command '" + edge->EvaluateCommand() + "' failed.");
    return false;
  }
  return true;
}

这里会开始一条edge的build，具体调用是在RealCommandRunner里的StartCommand

bool RealCommandRunner::StartCommand(Edge* edge) {
  string command = edge->EvaluateCommand();
  Subprocess* subproc = subprocs_.Add(command, edge->use_console());
  if (!subproc)
    return false;
  subproc_to_edge_.insert(make_pair(subproc, edge));

  return true;
}

这里一条edge的编译command会被算出来，然后起一个subprocess去run，跟py里的subprocess run查不多，之后会起一个Subprocess加到ProcessSet里

Subprocess *SubprocessSet::Add(const string& command, bool use_console) {
  Subprocess *subprocess = new Subprocess(use_console);
  if (!subprocess->Start(this, command)) {
    delete subprocess;
    return 0;
  }
  running_.push_back(subprocess);
  return subprocess;
}

这里可以开始看subprocess的头文件了, 这里的subprocess设计如此，需要用户去关注fd的可读事件来标志process的finish，ninja里为了通用，在unix下都是用的ppoll或者pselect来实现的。理论上讲，这里的并发数受到cpu core的限制，不会有太大的问题。

/// Subprocess wraps a single async subprocess.  It is entirely
/// passive: it expects the caller to notify it when its fds are ready
/// for reading, as well as call Finish() to reap the child once done()
/// is true.
struct Subprocess {
  ~Subprocess();

  /// Returns ExitSuccess on successful process exit, ExitInterrupted if
  /// the process was interrupted, ExitFailure if it otherwise failed.
  ExitStatus Finish();

  bool Done() const;

  const std::string& GetOutput() const;

 private:
  Subprocess(bool use_console);
  bool Start(struct SubprocessSet* set, const std::string& command);
  void OnPipeReady();

  std::string buf_;

#ifdef _WIN32
  /// Set up pipe_ as the parent-side pipe of the subprocess; return the
  /// other end of the pipe, usable in the child process.
  HANDLE SetupPipe(HANDLE ioport);

  HANDLE child_;
  HANDLE pipe_;
  OVERLAPPED overlapped_;
  char overlapped_buf_[4 << 10];
  bool is_reading_;
#else
  int fd_;
  pid_t pid_;
#endif
  bool use_console_;

  friend struct SubprocessSet;
};

随后具体看Start(struct SubprocessSet* set, const std::string& command)的这个调用，这里设置process的action + attribute, posix_spawn来创建一个新的process 输出会被写道pipe[0]里，此时fd_被记录成pipe[0], 这个fd_在结束后是可读的

bool Subprocess::Start(SubprocessSet* set, const string& command) {
  int output_pipe[2];
  if (pipe(output_pipe) < 0)
    Fatal("pipe: %s", strerror(errno));
  fd_ = output_pipe[0];
#if !defined(USE_PPOLL)
  // If available, we use ppoll in DoWork(); otherwise we use pselect
  // and so must avoid overly-large FDs.
  if (fd_ >= static_cast<int>(FD_SETSIZE))
    Fatal("pipe: %s", strerror(EMFILE));
#endif  // !USE_PPOLL
  // exec前关闭fd，防止fd泄漏
  SetCloseOnExec(fd_);

  posix_spawn_file_actions_t action;
  int err = posix_spawn_file_actions_init(&action);
  if (err != 0)
    Fatal("posix_spawn_file_actions_init: %s", strerror(err));

  // output[0]是读，子进程不需要读，关闭
  err = posix_spawn_file_actions_addclose(&action, output_pipe[0]);
  if (err != 0)
    Fatal("posix_spawn_file_actions_addclose: %s", strerror(err));

  posix_spawnattr_t attr;
  err = posix_spawnattr_init(&attr);
  if (err != 0)
    Fatal("posix_spawnattr_init: %s", strerror(err));

  short flags = 0;
  // 这里是设置process的sig掩码
  flags |= POSIX_SPAWN_SETSIGMASK;
  err = posix_spawnattr_setsigmask(&attr, &set->old_mask_);
  if (err != 0)
    Fatal("posix_spawnattr_setsigmask: %s", strerror(err));
  // Signals which are set to be caught in the calling process image are set to
  // default action in the new process image, so no explicit
  // POSIX_SPAWN_SETSIGDEF parameter is needed.

  if (!use_console_) {
    // Put the child in its own process group, so ctrl-c won't reach it.
    flags |= POSIX_SPAWN_SETPGROUP;
    // No need to posix_spawnattr_setpgroup(&attr, 0), it's the default.

    // Open /dev/null over stdin.
    err = posix_spawn_file_actions_addopen(&action, 0, "/dev/null", O_RDONLY,
          0);
    if (err != 0) {
      Fatal("posix_spawn_file_actions_addopen: %s", strerror(err));
    }
    // 写端复制到标准输出和错误，然后关闭原始输出写断
    err = posix_spawn_file_actions_adddup2(&action, output_pipe[1], 1);
    if (err != 0)
      Fatal("posix_spawn_file_actions_adddup2: %s", strerror(err));
    err = posix_spawn_file_actions_adddup2(&action, output_pipe[1], 2);
    if (err != 0)
      Fatal("posix_spawn_file_actions_adddup2: %s", strerror(err));
    err = posix_spawn_file_actions_addclose(&action, output_pipe[1]);
    if (err != 0)
      Fatal("posix_spawn_file_actions_addclose: %s", strerror(err));
    // In the console case, output_pipe is still inherited by the child and
    // closed when the subprocess finishes, which then notifies ninja.
  }
  // 
#ifdef POSIX_SPAWN_USEVFORK
  // POSIX_SPAWN_USEVFORK是一个标志位，值为0x40
  // 它指示系统在创建子进程时使用vfork()而非fork()
  // vfork()比fork()更高效，因为它不会复制父进程的内存页面
  // 而是共享父进程的地址空间，直到exec调用
  flags |= POSIX_SPAWN_USEVFORK;
#endif

  err = posix_spawnattr_setflags(&attr, flags);
  if (err != 0)
    Fatal("posix_spawnattr_setflags: %s", strerror(err));

  const char* spawned_args[] = { "/bin/sh", "-c", command.c_str(), NULL };
  err = posix_spawn(&pid_, "/bin/sh", &action, &attr,
        const_cast<char**>(spawned_args), environ);
  if (err != 0)
    Fatal("posix_spawn: %s", strerror(err));

  err = posix_spawnattr_destroy(&attr);
  if (err != 0)
    Fatal("posix_spawnattr_destroy: %s", strerror(err));
  err = posix_spawn_file_actions_destroy(&action);
  if (err != 0)
    Fatal("posix_spawn_file_actions_destroy: %s", strerror(err));

  close(output_pipe[1]);
  return true;
}

随后要自己处理SubprocessSet::DoWork()，走一个event loop去处理完成的process，配合之前说的被动处理fd。把runing queue的fd放进finished queue

bool SubprocessSet::DoWork() {
  vector<pollfd> fds;
  nfds_t nfds = 0;

  for (vector<Subprocess*>::iterator i = running_.begin();
       i != running_.end(); ++i) {
    int fd = (*i)->fd_;
    if (fd < 0)
      continue;
    pollfd pfd = { fd, POLLIN | POLLPRI, 0 };
    fds.push_back(pfd);
    ++nfds;
  }

  interrupted_ = 0;
  int ret = ppoll(&fds.front(), nfds, NULL, &old_mask_);
  if (ret == -1) {
    if (errno != EINTR) {
      perror("ninja: ppoll");
      return false;
    }
    return IsInterrupted();
  }

  HandlePendingInterruption();
  if (IsInterrupted())
    return true;

  nfds_t cur_nfd = 0;
  for (vector<Subprocess*>::iterator i = running_.begin();
       i != running_.end(); ) {
    int fd = (*i)->fd_;
    if (fd < 0)
      continue;
    assert(fd == fds[cur_nfd].fd);
    if (fds[cur_nfd++].revents) {
      (*i)->OnPipeReady();
      if ((*i)->Done()) {
        finished_.push(*i);
        i = running_.erase(i);
        continue;
      }
    }
    ++i;
  }

  return IsInterrupted();
}

Subprocess + SubprocessSet还有很多信号的处理，可以具体看看