Python multiprocessing中独立进程的mp.Value变量异常被篡改问题问询

这个问题的根源在Windows系统下multiprocessing的spawn启动机制，以及mp.Value的命名共享内存实现方式上，具体原因和解决办法如下：

为什么会出现“跨进程修改”的诡异现象？ #

在Windows中，multiprocessing默认使用spawn模式创建子进程——这种模式下，每个子进程会重新加载整个脚本，并通过pickle序列化传递参数。而mp.Value在Windows下是通过命名共享内存实现的：

• 如果你没有给mp.Value指定唯一的name参数，系统会自动生成一个临时名称。
• 当你快速连续创建两个mp.Value时（比如间隔0.03秒），系统可能会复用之前的共享内存名称，导致两个进程的mp.Value意外指向了同一块共享内存区域！

这就解释了你看到的现象：第二个Worker进程初始化mp.Value为-1时，直接覆盖了第一个进程已经改成1的共享内存值，触发了第一个进程的非法变更检测。

解决方案 #

1. 给每个mp.Value指定唯一名称 #

显式给每个共享变量设置独一无二的name参数，避免系统自动生成的名称冲突：

import multiprocessing as mp
import time

class Worker:
    def __init__(self, tag, service_state) -> None:
        self.tag = tag
        self.local_state = int(service_state.value)
        self.state = service_state
        self.run_work_loop()

    def run_work_loop(self) -> None:
        print(f"[{self.tag}] Running... {self.state.value} {self.local_state}")
        while True:
            if self.state.value != self.local_state:
                print(f"[{self.tag}] Illegal change. Shared state: {self.state.value} Local State: {self.local_state}")
                break
            elif self.state.value == -1:
                self.state.value = self.local_state = 1
                print(f"[{self.tag}] Set Shared State: {self.state.value} Local State: {self.local_state}.")
            time.sleep(0.01)  # 加sleep避免CPU跑满

if __name__ == "__main__":
    # 给每个Value指定唯一name
    mp.Process(target=Worker, args=("A", mp.Value('i', -1, name="worker_A"))).start()
    time.sleep(.03)
    mp.Process(target=Worker, args=("B", mp.Value('i', -1, name="worker_B"))).start()

2. 使用mp.Manager创建共享变量（更推荐） #

mp.Manager会统一管理共享资源，自动处理命名和隔离问题，比直接用mp.Value更安全：

import multiprocessing as mp
import time

class Worker:
    def __init__(self, tag, service_state) -> None:
        self.tag = tag
        self.local_state = service_state.value
        self.state = service_state
        self.run_work_loop()

    def run_work_loop(self) -> None:
        print(f"[{self.tag}] Running... {self.state.value} {self.local_state}")
        while True:
            if self.state.value != self.local_state:
                print(f"[{self.tag}] Illegal change. Shared state: {self.state.value} Local State: {self.local_state}")
                break
            elif self.state.value == -1:
                self.state.value = self.local_state = 1
                print(f"[{self.tag}] Set Shared State: {self.state.value} Local State: {self.local_state}.")
            time.sleep(0.01)

if __name__ == "__main__":
    with mp.Manager() as manager:
        # 通过Manager创建共享Value
        val_a = manager.Value('i', -1)
        val_b = manager.Value('i', -1)
        mp.Process(target=Worker, args=("A", val_a)).start()
        time.sleep(.03)
        mp.Process(target=Worker, args=("B", val_b)).start()

3. 规范multiprocessing的使用方式 #

更符合Python多进程规范的做法是让Worker继承mp.Process，重写run方法，而不是在__init__里直接启动循环：

import multiprocessing as mp
import time

class Worker(mp.Process):
    def __init__(self, tag, service_state) -> None:
        super().__init__()
        self.tag = tag
        self.local_state = int(service_state.value)
        self.state = service_state

    def run(self) -> None:
        print(f"[{self.tag}] Running... {self.state.value} {self.local_state}")
        while True:
            if self.state.value != self.local_state:
                print(f"[{self.tag}] Illegal change. Shared state: {self.state.value} Local State: {self.local_state}")
                break
            elif self.state.value == -1:
                self.state.value = self.local_state = 1
                print(f"[{self.tag}] Set Shared State: {self.state.value} Local State: {self.local_state}.")
            time.sleep(0.01)

if __name__ == "__main__":
    with mp.Manager() as manager:
        val_a = manager.Value('i', -1)
        val_b = manager.Value('i', -1)
        worker_a = Worker("A", val_a)
        worker_a.start()
        time.sleep(.03)
        worker_b = Worker("B", val_b)
        worker_b.start()

额外提示 #

• 工作循环里一定要加time.sleep()，否则进程会占用100% CPU，不仅影响系统性能，还可能导致一些难以排查的调度问题。
• Linux/macOS下使用fork模式创建进程，不会出现这个问题（因为fork是复制父进程内存，mp.Value用匿名共享内存），但Windows下必须注意spawn模式的特性。

内容的提问来源于stack exchange，提问作者Alan