线程同步技术:
解决多个线程争抢同一个资源的情况,线程协作工作。一份数据同一时刻只能有一个线程处理。
解决线程同步的几种方法:
Lock、RLock、Condition、Barrier、semaphore
1)Lock 锁
锁,一旦线程获得锁,其它试图获取锁的线程将被阻塞。
当用阻塞参数设置为 False 时, 不要阻止。如果将阻塞设置为 True 的调用将阻止, 则立即返回 False;否则, 将锁定设置为锁定并返回 True。
Lock的方法:
acquire(blocking=True,timeout=-1) 加锁。默认True阻塞,阻塞可以设置超时时间。非阻塞时成功获取锁返回True,否则返回False。
当blocking设置为False时,不阻塞,同一个锁对象,其它线程可以重用,但最后都必须释放。
如果设置为True(默认True),其它试图调用锁的线程将阻塞,并立即返回False。阻塞可以设置超时时间。
release() 释放锁。可以从任何线程调用释放。已上锁的锁,会被重置为unlocked,对未上锁的锁调用,会抛RuntimeError异常: cannot release un-acquired lock。
不使用Lock的例子:
#不使用Lock锁的例子 import logging import threading,time logging.basicConfig(level=logging.INFO) # 10 -> 100cups cups = [] lock = threading.Lock() def worker(lock:threading.Lock,task=100): while True: count = len(cups) time.sleep(0.1) if count >= task: break logging.info(count) cups.append(1) logging.info("{} make 1........ ".format(threading.current_thread().name)) logging.info("{} ending=======".format(len(cups))) for x in range(10): threading.Thread(target=worker,args=(lock,100)).start() 运行结果: INFO:root:Thread-7 make 1........ INFO:root:93 INFO:root:Thread-5 make 1........ INFO:root:95 INFO:root:Thread-6 make 1........ INFO:root:92 INFO:root:Thread-2 make 1........ INFO:root:94 INFO:root:Thread-8 make 1........ INFO:root:97 INFO:root:Thread-10 make 1........ INFO:root:96 INFO:root:Thread-4 make 1........ INFO:root:98 INFO:root:Thread-1 make 1........ INFO:root:99 INFO:root:Thread-9 make 1........ INFO:root:109 ending======= INFO:root:109 ending======= INFO:root:109 ending=======
还是使用前面的10个工人生产100杯子的例子, 当做到99个杯子时,10个工人都发现还少一个,都去做了一个,一共做了109个,超出了100个,就发生了不可预期的结果。
临界线判断失误,多生产了杯子。
解决方法就可以用锁,来解决资源争抢。当一个人看杯子数量时,就上锁,其它人只能等着,看完杯子后发现少一个就把这最后一个做出来,然后数量加一,解锁,其他人再看到已经有100个杯子时,就可以停止工作。
加锁的时机非常重要:看杯子数量时加锁,增加数量后释放锁。
使用Lock的例子:
#Lock import logging import threading import time logging.basicConfig(level=logging.INFO) # 10 -> 100cups cups = [] lock = threading.Lock() def worker(lock:threading.Lock,task=100): while True: if lock.acquire(False): count = len(cups) time.sleep(0.1) if count >= task: lock.release() break logging.info(count) cups.append(1) lock.release() logging.info("{} make 1........ ".format(threading.current_thread().name)) logging.info("{} ending=======".format(len(cups))) for x in range(10): threading.Thread(target=worker,args=(lock,100)).start() 运行结果: INFO:root:0 INFO:root:Thread-1 make 1........ INFO:root:1 INFO:root:Thread-5 make 1........ INFO:root:2 INFO:root:Thread-6 make 1........ .... INFO:root:Thread-3 make 1........ INFO:root:97 INFO:root:Thread-3 make 1........ INFO:root:98 INFO:root:Thread-4 make 1........ INFO:root:99 INFO:root:Thread-3 make 1........ INFO:root:100 ending======= INFO:root:100 ending======= INFO:root:100 ending======= .....
在使用了锁以后,虽然保证了结果的准确性,但是性能下降了很多。
一般来说加锁以后还要有一些功能实现,在释放之前还有可能抛异常,一旦抛出异常,锁是无法释放,但是当前线程可能因为这个异常被终止了,这就产生了死锁。
死锁解决办法:
1、使用 try..except..finally 语句处理异常、保证锁的释放
2、with 语句上下文管理,锁对象支持上下文管理。只要实现了__enter__和__exit__魔术方法的对象都支持上下文管理。
锁的应用场景:
独占锁: 锁适用于访问和修改同一个共享资源的时候,即读写同一个资源的时候。
共享锁: 如果共享资源是不可变的值时,所有线程每一次读取它都是同一样的值,这样的情况就不需要锁。
使用锁的注意事项:
- 少用锁,必要时用锁。使用了锁,多线程访问被锁的资源时,就变成了串行,要么排队执行,要么争抢执行。
- 加锁时间越短越好,不需要就立即释放锁。
- 一定要避免死锁。
不使用锁时,有了效率,但是结果是错的。
使用了锁,变成了串行,效率地下,但是结果是对的。
import threading import time lock = threading.Lock() def work(): print('working..') time.sleep(0.2) lock.release() # 1解锁 lock.acquire() # 1上锁 print("get locker 1") threading.Thread(target=work).start() time.sleep(1) lock.acquire() # 2上锁 print("get locker 2") threading.Thread(target=work).start() print("release locker") 运行结果: get locker 1 working.. get locker 2 working.. release locker
同一个锁对象在释放后可以再次使用。
但是如果同一把锁加锁后,又被别人拿了,自己就阻塞了:
import threading import time lock = threading.Lock() def work(): print('working..') time.sleep(0.2) lock.release() # 1解锁 lock.acquire() # 1上锁 print("get locker 1") lock.acquire() # 2上锁 print("get locker 2") threading.Thread(target=work).start() threading.Thread(target=work).start() print("release locker") 运行结果: get locker 1 阻塞状态....
阻塞锁:
#阻塞锁 import threading,time lock = threading.Lock() def foo(): ret = lock.acquire() print("{} Locked. {}".format(ret,threading.current_thread())) time.sleep(10) threading.Thread(target=foo).start() threading.Thread(target=foo).start() 运行结果: True Locked. <Thread(Thread-1, started 123145559191552)>
lock.acquire()默认设置blocking=True,两个线程使用同一个Lock锁对象,只要Thread-1线程不释放,第二个线程就无法获取锁,且会使Thread-1线程阻塞。
如果想让多个线程同时都可以使用一个锁对象,就必须使用非阻塞锁,或者第一个线程使用完锁之后立刻释放,然后第二个线程再使用。
非阻塞锁:
#非阻塞锁 import threading,time lock = threading.Lock() def foo(): ret = lock.acquire(False) print("{} Locked. {}".format(ret,threading.current_thread())) time.sleep(10) threading.Thread(target=foo).start() threading.Thread(target=foo).start() 运行结果: True Locked. <Thread(Thread-1, started 123145516146688)> False Locked. <Thread(Thread-2, started 123145521401856)> Process finished with exit code 0
lock.acquire(False)设置blocking=False表示不阻塞,使用同一个Lock锁对象时,第二个线程仍可以使用锁,且第一个锁不会被阻塞。
非阻塞锁2:
#非阻塞锁 import threading,logging,time FORMAT = '%(asctime)s [%(threadName)s,%(thread)d] %(message)s' logging.basicConfig(level=logging.INFO,format=FORMAT) def worker(tasks): for task in tasks: time.sleep(0.01) if task.lock.acquire(False): #False非阻塞 logging.info('{} {} begin to start'.format(threading.current_thread().name,task.name)) else: logging.info('{} {} is working'.format(threading.current_thread().name,task.name)) class Task: def __init__(self,name): self.name = name self.lock = threading.Lock() tasks = [Task('task={}'.format(t)) for t in range(5)] for i in range(3): t = threading.Thread(target=worker,name='worker-{}'.format(i),args=(tasks,)) t.start() 运行结果: 2017-12-19 16:37:49,556 [worker-2,123145390018560] worker-2 task=0 begin to start 2017-12-19 16:37:49,556 [worker-1,123145384763392] worker-1 task=0 is working 2017-12-19 16:37:49,557 [worker-0,123145379508224] worker-0 task=0 is working 2017-12-19 16:37:49,567 [worker-2,123145390018560] worker-2 task=1 begin to start 2017-12-19 16:37:49,567 [worker-1,123145384763392] worker-1 task=1 is working 2017-12-19 16:37:49,568 [worker-0,123145379508224] worker-0 task=1 is working 2017-12-19 16:37:49,580 [worker-1,123145384763392] worker-1 task=2 begin to start 2017-12-19 16:37:49,580 [worker-2,123145390018560] worker-2 task=2 is working 2017-12-19 16:37:49,580 [worker-0,123145379508224] worker-0 task=2 is working 2017-12-19 16:37:49,591 [worker-1,123145384763392] worker-1 task=3 begin to start 2017-12-19 16:37:49,592 [worker-2,123145390018560] worker-2 task=3 is working 2017-12-19 16:37:49,592 [worker-0,123145379508224] worker-0 task=3 is working 2017-12-19 16:37:49,604 [worker-1,123145384763392] worker-1 task=4 begin to start 2017-12-19 16:37:49,604 [worker-2,123145390018560] worker-2 task=4 is working 2017-12-19 16:37:49,604 [worker-0,123145379508224] worker-0 task=4 is working