大师兄的Python源码学习笔记(三十九）: Python的多线程机制(一）
大师兄的Python源码学习笔记(四十一）: Python的多线程机制(三）

二、关于_thread包和threading

• Python提供的最基础的多线程机制接口是_thread，用C实现。
• threading包提供更高层的多线程机制接口，用Python实现。
• 我们先从更底层的_thread接口开始：

Modules\_threadmodule.c

static PyMethodDef thread_methods[] = {
    {"start_new_thread",        (PyCFunction)thread_PyThread_start_new_thread,
     METH_VARARGS, start_new_doc},
    {"start_new",               (PyCFunction)thread_PyThread_start_new_thread,
     METH_VARARGS, start_new_doc},
    {"allocate_lock",           (PyCFunction)thread_PyThread_allocate_lock,
     METH_NOARGS, allocate_doc},
    {"allocate",                (PyCFunction)thread_PyThread_allocate_lock,
     METH_NOARGS, allocate_doc},
    {"exit_thread",             (PyCFunction)thread_PyThread_exit_thread,
     METH_NOARGS, exit_doc},
    {"exit",                    (PyCFunction)thread_PyThread_exit_thread,
     METH_NOARGS, exit_doc},
    {"interrupt_main",          (PyCFunction)thread_PyThread_interrupt_main,
     METH_NOARGS, interrupt_doc},
    {"get_ident",               (PyCFunction)thread_get_ident,
     METH_NOARGS, get_ident_doc},
    {"_count",                  (PyCFunction)thread__count,
     METH_NOARGS, _count_doc},
    {"stack_size",              (PyCFunction)thread_stack_size,
     METH_VARARGS, stack_size_doc},
    {"_set_sentinel",           (PyCFunction)thread__set_sentinel,
     METH_NOARGS, _set_sentinel_doc},
    {NULL,                      NULL}           /* sentinel */
};

• 可以发现，threadmodule中有的接口以不同形式出现了两次，比如allocate_lock和allocate，但实际在他们都对应的是thread_PyThread_allocate_lock函数。

三、Python线程的创建

• 观察thread_methods，可以发现其中创建线程的接口对应的是thread_PyThread_start_new_thread函数：

Modules\_threadmodule.c

static PyObject *
thread_PyThread_start_new_thread(PyObject *self, PyObject *fargs)
{
    PyObject *func, *args, *keyw = NULL;
    struct bootstate *boot;
    unsigned long ident;

    ... ...
    boot = PyMem_NEW(struct bootstate, 1);
    if (boot == NULL)
        return PyErr_NoMemory();
    boot->interp = PyThreadState_GET()->interp;
    boot->func = func;
    boot->args = args;
    boot->keyw = keyw;
    boot->tstate = _PyThreadState_Prealloc(boot->interp);
    if (boot->tstate == NULL) {
        PyMem_DEL(boot);
        return PyErr_NoMemory();
    }
    Py_INCREF(func);
    Py_INCREF(args);
    Py_XINCREF(keyw);
    PyEval_InitThreads(); /* Start the interpreter's thread-awareness */
    ident = PyThread_start_new_thread(t_bootstrap, (void*) boot);
    ...
    return PyLong_FromUnsignedLong(ident);
}

• 在thread_PyThread_start_new_thread中，虚拟机通过三个主要动作来完成线程的创建：

1. 创建并初始化boot，其中保存着线程的所有信息：

   boot->interp = PyThreadState_GET()->interp;
   boot->func = func;
   boot->args = args;
   boot->keyw = keyw;
   boot->tstate = _PyThreadState_Prealloc(boot->interp);

2. 初始化Python的多线程环境：

   PyEval_InitThreads(); /* Start the interpreter's thread-awareness */

3. 以boot为参数，创建操作系统的原生线程：

   ident = PyThread_start_new_thread(t_bootstrap, (void*) boot);
   ...
   return PyLong_FromUnsignedLong(ident);

• 可以看到boot->interp中保存了Python的PyInterpreterState对象，这个对象中携带了module pool这样的全局信息，所有的线程会共享这些全局信息。

1. 建立多线程环境

• 建立多线程环境的核心是建立GIL，GIL是一个_gil_runtime_state结构体：

struct _gil_runtime_state {
    /* microseconds (the Python API uses seconds, though) */
    unsigned long interval;
    /* Last PyThreadState holding / having held the GIL. This helps us
       know whether anyone else was scheduled after we dropped the GIL. */
    _Py_atomic_address last_holder;
    /* Whether the GIL is already taken (-1 if uninitialized). This is
       atomic because it can be read without any lock taken in ceval.c. */
    _Py_atomic_int locked;
    /* Number of GIL switches since the beginning. */
    unsigned long switch_number;
    /* This condition variable allows one or several threads to wait
       until the GIL is released. In addition, the mutex also protects
       the above variables. */
    PyCOND_T cond;
    PyMUTEX_T mutex;
#ifdef FORCE_SWITCHING
    /* This condition variable helps the GIL-releasing thread wait for
       a GIL-awaiting thread to be scheduled and take the GIL. */
    PyCOND_T switch_cond;
    PyMUTEX_T switch_mutex;
#endif
};

• Python在初始化解释器时，会创建一把未初始化的GIL：

ceval.c

void
_PyEval_Initialize(struct _ceval_runtime_state *state)
{
    state->recursion_limit = Py_DEFAULT_RECURSION_LIMIT;
    _Py_CheckRecursionLimit = Py_DEFAULT_RECURSION_LIMIT;
    _gil_initialize(&state->gil);
}

Python\ceval_gil.h

#define DEFAULT_INTERVAL 5000

static void _gil_initialize(struct _gil_runtime_state *state)
{
    _Py_atomic_int uninitialized = {-1};
    state->locked = uninitialized;
    state->interval = DEFAULT_INTERVAL;
}

• 在激活多线程机制前，除了上面提到的未初始化的GIL，Python实际是不支持多线程的，这是因为Python选择了让用户激活多线程机制的策略，一旦用户调用thread.start_new_thread，才会开始实际初始化多线程环境。
• 而我们知道，start_new_thread会对应调用thread_PyThread_start_new_thread函数,并在其中调用PyEval_InitThreads初始化多线程环境：

ceval.c

void
PyEval_InitThreads(void)
{
    if (gil_created())
        return;
    create_gil();
    take_gil(PyThreadState_GET());
    _PyRuntime.ceval.pending.main_thread = PyThread_get_thread_ident();
    if (!_PyRuntime.ceval.pending.lock)
        _PyRuntime.ceval.pending.lock = PyThread_allocate_lock();
}

• PyEval_InitThreads会通过gil_created以原子读操作检查GIL在线程中是否已经初始化：

ceval.c

static int gil_created(void)
{
    return (_Py_atomic_load_explicit(&_PyRuntime.ceval.gil.locked,
                                     _Py_memory_order_acquire)
            ) >= 0;
}

• 如果没有初始化，则通过create_gil初始化GIL：

ceval.c

static void create_gil(void)
{
    MUTEX_INIT(_PyRuntime.ceval.gil.mutex);
#ifdef FORCE_SWITCHING
    MUTEX_INIT(_PyRuntime.ceval.gil.switch_mutex);
#endif
    COND_INIT(_PyRuntime.ceval.gil.cond);
#ifdef FORCE_SWITCHING
    COND_INIT(_PyRuntime.ceval.gil.switch_cond);
#endif
    _Py_atomic_store_relaxed(&_PyRuntime.ceval.gil.last_holder, 0);
    _Py_ANNOTATE_RWLOCK_CREATE(&_PyRuntime.ceval.gil.locked);
    _Py_atomic_store_explicit(&_PyRuntime.ceval.gil.locked, 0,
                              _Py_memory_order_release);
}

• 在初始化GIL后，会通过take_gil函数获取GIL:

ceval.c

static void take_gil(PyThreadState *tstate)
{
    int err;
    if (tstate == NULL)
        Py_FatalError("take_gil: NULL tstate");

    err = errno;
    MUTEX_LOCK(_PyRuntime.ceval.gil.mutex);

    if (!_Py_atomic_load_relaxed(&_PyRuntime.ceval.gil.locked))
        goto _ready;

    while (_Py_atomic_load_relaxed(&_PyRuntime.ceval.gil.locked)) {
        int timed_out = 0;
        unsigned long saved_switchnum;

        saved_switchnum = _PyRuntime.ceval.gil.switch_number;
        COND_TIMED_WAIT(_PyRuntime.ceval.gil.cond, _PyRuntime.ceval.gil.mutex,
                        INTERVAL, timed_out);
        /* If we timed out and no switch occurred in the meantime, it is time
           to ask the GIL-holding thread to drop it. */
        if (timed_out &&
            _Py_atomic_load_relaxed(&_PyRuntime.ceval.gil.locked) &&
            _PyRuntime.ceval.gil.switch_number == saved_switchnum) {
            SET_GIL_DROP_REQUEST();
        }
    }
_ready:
#ifdef FORCE_SWITCHING
    /* This mutex must be taken before modifying
       _PyRuntime.ceval.gil.last_holder (see drop_gil()). */
    MUTEX_LOCK(_PyRuntime.ceval.gil.switch_mutex);
#endif
    /* We now hold the GIL */
    _Py_atomic_store_relaxed(&_PyRuntime.ceval.gil.locked, 1);
    _Py_ANNOTATE_RWLOCK_ACQUIRED(&_PyRuntime.ceval.gil.locked, /*is_write=*/1);

    if (tstate != (PyThreadState*)_Py_atomic_load_relaxed(
                    &_PyRuntime.ceval.gil.last_holder))
    {
        _Py_atomic_store_relaxed(&_PyRuntime.ceval.gil.last_holder,
                                 (uintptr_t)tstate);
        ++_PyRuntime.ceval.gil.switch_number;
    }

#ifdef FORCE_SWITCHING
    COND_SIGNAL(_PyRuntime.ceval.gil.switch_cond);
    MUTEX_UNLOCK(_PyRuntime.ceval.gil.switch_mutex);
#endif
    if (_Py_atomic_load_relaxed(&_PyRuntime.ceval.gil_drop_request)) {
        RESET_GIL_DROP_REQUEST();
    }
    if (tstate->async_exc != NULL) {
        _PyEval_SignalAsyncExc();
    }

    MUTEX_UNLOCK(_PyRuntime.ceval.gil.mutex);
    errno = err;
}

• 在take_gil时，首先获取互斥锁(mutex)，并检查是否有线程持有GIL。
• 如果GIL被占用，则设置信号量(cond)等待，并将线程挂起，并释放互斥锁(mutex)。
• 当信号量(cond)被唤醒时，互斥锁(mutex)会被自动加锁。
• 大多数情况下通过条件满足唤醒，while循环用于避免意外唤醒时条件不足。
• 如果等待超时并且期间没有发生线程切换，则通过SET_GIL_DROP_REQUEST请求last_holder释放GIL。
• 如果GIL没有被占用，则将locked设置为1占用GIL，并将当前线程状态对象保存到last_holder，最后将切换次数(switch_number)加1。

• 获取GIL后，会通过PyThread_get_thread_ident函数获取线程id，PyThread_get_thread_ident会根据编译器和平台来确认初始化动作：

Include\pythread.h

unsigned long
PyThread_get_thread_ident(void)
{
    if (!initialized)
        PyThread_init_thread();

    return GetCurrentThreadId();
}

Python\thread_pthread.h

unsigned long
PyThread_get_thread_ident(void)
{
    volatile pthread_t threadid;
    if (!initialized)
        PyThread_init_thread();
    threadid = pthread_self();
    return (unsigned long) threadid;
}

• 在获取主线程id之前，会先检查 initialized，也就是底层平台所提供的原生线程，如果原生线程没有建立，则通过PyThread_init_thread创建。

Python\thread.c

void
PyThread_init_thread(void)
{
#ifdef Py_DEBUG
   const char *p = Py_GETENV("PYTHONTHREADDEBUG");

   if (p) {
       if (*p)
           thread_debug = atoi(p);
       else
           thread_debug = 1;
   }
#endif /* Py_DEBUG */
   if (initialized)
       return;
   initialized = 1;
   dprintf(("PyThread_init_thread called\n"));
   PyThread__init_thread();
}

• 最后，创建GIL对应的互斥锁的工作由PyThread_allocate_lock完成，PyThread_allocate_lock会根据平台来确认动作：

Python\thread_nt.h

PyThread_type_lock
PyThread_allocate_lock(void)
{
    PNRMUTEX aLock;

    dprintf(("PyThread_allocate_lock called\n"));
    if (!initialized)
        PyThread_init_thread();

    aLock = AllocNonRecursiveMutex() ;

    dprintf(("%lu: PyThread_allocate_lock() -> %p\n", PyThread_get_thread_ident(), aLock));

    return (PyThread_type_lock) aLock;
}

Python\thread_pthread.h

PyThread_type_lock
PyThread_allocate_lock(void)
{
    pthread_lock *lock;
    int status, error = 0;

    dprintf(("PyThread_allocate_lock called\n"));
    if (!initialized)
        PyThread_init_thread();

    lock = (pthread_lock *) PyMem_RawMalloc(sizeof(pthread_lock));
    if (lock) {
        memset((void *)lock, '\0', sizeof(pthread_lock));
        lock->locked = 0;

        status = pthread_mutex_init(&lock->mut,
                                    pthread_mutexattr_default);
        CHECK_STATUS_PTHREAD("pthread_mutex_init");
        /* Mark the pthread mutex underlying a Python mutex as
           pure happens-before.  We can't simply mark the
           Python-level mutex as a mutex because it can be
           acquired and released in different threads, which
           will cause errors. */
        _Py_ANNOTATE_PURE_HAPPENS_BEFORE_MUTEX(&lock->mut);

        status = pthread_cond_init(&lock->lock_released,
                                   pthread_condattr_default);
        CHECK_STATUS_PTHREAD("pthread_cond_init");

        if (error) {
            PyMem_RawFree((void *)lock);
            lock = 0;
        }
    }

    dprintf(("PyThread_allocate_lock() -> %p\n", lock));
    return (PyThread_type_lock) lock;
}

• 创建的互斥锁结构体也根据平台并不相同：

Python\thread_nt.h

typedef struct _NRMUTEX
{
   PyMUTEX_T cs;
   PyCOND_T cv;
   int locked;
} NRMUTEX;
typedef NRMUTEX *PNRMUTEX;

Python\thread_pthread.h

typedef struct {
   char             locked; /* 0=unlocked, 1=locked */
   /* a <cond, mutex> pair to handle an acquire of a locked lock */
   pthread_cond_t   lock_released;
   pthread_mutex_t  mut;
} pthread_lock;