Python Threads – Critical Section
Python Threads - Critical Section
什么是线程的"临界区"(在 Python 中)？
A thread enters the critical section
by calling the acquire() method, which
can either be blocking or
non-blocking. A thread exits the
critical section, by calling the
release() method.
- 理解 Python 中的线程，Linux Gazette
另外，锁的用途是什么？
其他人给出了非常好的定义。这是经典的例子：
import threading
account_balance = 0 # The"resource" that zenazn mentions.
account_balance_lock = threading.Lock()
def change_account_balance(delta):
global account_balance
with account_balance_lock:
# Critical section is within this block.
account_balance += deltapos = threading.Thread(target=change_account_balance, args=[100])
neg = threading.Thread(target=change_account_balance, args=[-100])
pos.start(), neg.start()
import logging # This module is thread safe.
import threading
LOCK = threading.Lock()
def run():
if LOCK.acquire(False): # Non-blocking -- return whether we got it
logging.info('Got the lock!')
LOCK.release()
else:
logging.info("Couldn't get the lock. Maybe next time")
logging.basicConfig(level=logging.INFO)
threads = [threading.Thread(target=run) for i in range(100)]
for thread in threads:
thread.start()假设 += 运算符由三个子组件组成：
- 读取当前值
- 将 RHS 添加到该值
- 将累计值写回 LHS(技术上用 Python 术语绑定它)
如果您没有 with account_balance_lock 语句并且您并行执行两个 change_account_balance 调用，您最终可能会以危险的方式交错三个子组件操作。假设您同时调用 change_account_balance(100) (AKA pos) 和 change_account_balance(-100) (AKA neg)。这可能发生：
import threading
account_balance = 0 # The"resource" that zenazn mentions.
account_balance_lock = threading.Lock()
def change_account_balance(delta):
global account_balance
with account_balance_lock:
# Critical section is within this block.
account_balance += deltapos = threading.Thread(target=change_account_balance, args=[100])
neg = threading.Thread(target=change_account_balance, args=[-100])
pos.start(), neg.start()
import logging # This module is thread safe.
import threading
LOCK = threading.Lock()
def run():
if LOCK.acquire(False): # Non-blocking -- return whether we got it
logging.info('Got the lock!')
LOCK.release()
else:
logging.info("Couldn't get the lock. Maybe next time")
logging.basicConfig(level=logging.INFO)
threads = [threading.Thread(target=run) for i in range(100)]
for thread in threads:
thread.start()- pos: 读取当前值 -> 0
- neg: 读取当前值 -> 0
- pos：将当前值添加到读取值 -> 100
- neg：将当前值添加到读取值 -> -100
- pos: 写入当前值 -> account_balance = 100
- neg: 写入当前值 -> account_balance = -100
因为您没有强制操作在离散块中发生，所以您可以有三种可能的结果(-100、0、100)。
with [lock] 语句是一个单独的、不可分割的操作，它说："让我成为执行此代码块的唯一线程。如果正在执行其他东西，那很酷——我会等待。 " 这可确保对 account_balance 的更新是"线程安全的"(并行安全)。
注意：此模式有一个警告：您必须记住每次要操作 account_balance 时都获取 account_balance_lock(通过 with)以使代码保持线程安全。有一些方法可以减少这种脆弱性，但这是另一个问题的答案。
编辑：回想起来，可能很重要的是要提到 with 语句隐式调用锁上的阻塞 acquire ——这是上面的"我将等待"部分线程对话框。相反，非阻塞获取会说，"如果我不能立即获取锁，请告诉我"，然后依靠您检查是否获得了锁。
import threading
account_balance = 0 # The"resource" that zenazn mentions.
account_balance_lock = threading.Lock()
def change_account_balance(delta):
global account_balance
with account_balance_lock:
# Critical section is within this block.
account_balance += deltapos = threading.Thread(target=change_account_balance, args=[100])
neg = threading.Thread(target=change_account_balance, args=[-100])
pos.start(), neg.start()
import logging # This module is thread safe.
import threading
LOCK = threading.Lock()
def run():
if LOCK.acquire(False): # Non-blocking -- return whether we got it
logging.info('Got the lock!')
LOCK.release()
else:
logging.info("Couldn't get the lock. Maybe next time")
logging.basicConfig(level=logging.INFO)
threads = [threading.Thread(target=run) for i in range(100)]
for thread in threads:
thread.start()我还想补充一点，锁的主要目的是保证获取的原子性(acquire 跨线程的不可分割性)，一个简单的布尔标志不能保证。原子操作的语义大概也是另一个问题的内容。
代码的关键部分是一次只能由一个线程执行的代码。以聊天服务器为例。如果每个连接(即每个最终用户)都有一个线程，则一个"关键部分"是假脱机代码(向所有客户端发送传入消息)。如果有多个线程同时尝试对一条消息进行假脱机，您将得到 BfrIToS MANTWD PIoEmesCEsaSges 交织在一起，这显然一点都不好。
锁是可以用来同步访问关键部分(或一般资源)的东西。在我们的聊天服务器示例中，锁就像一个锁着的房间，里面有一台打字机。如果一个线程在那里(输入消息)，则没有其他线程可以进入房间。一旦第一个线程完成，他解锁房间并离开。然后另一个线程可以进入房间(锁定它)。 "Aquiring" 锁只是意味着"我得到了房间。"
"临界区"是一段代码，为了正确起见，必须确保该部分中一次只能有一个控制线程。通常，您需要一个临界区来包含将值写入内存的引用，这些引用可以在多个并发进程之间共享。