dpxo13079 2017-04-15 00:38
浏览 16
已采纳

Go语言中潜在的种族条件

I am not sure why the following code has the race condition, can someone give me a hint? I think there is no potential race condition. Thank you in advance.

type PossiblySafeCounter struct {
    mu sync.Mutex
    sum int
}

func (c *PossiblySafeCounter) inc() {
   c.mu.Lock();
   defer c.mu.Unlock();
   go func() {
       c.sum++
   }() 
}
func (c *PossiblySafeCounter) read() int {
    c.mu.Lock();
    defer c.mu.Unlock();
    return c.sum
 }
  • 写回答

1条回答 默认 最新

  • duanjiuhong5843 2017-04-15 00:41
    关注

    The c.sum++ is in a goroutine that is scheduled independently of the execution of the inc() method. When the inc() method exits the defered unlock of the mutex will happen and will very likely happen at the wrong time, leading to a race condition.

    As @Flimzy suggests using atomic.AddInt32 would remove the need for a mutex at all.

    two mutex based solutions are either to not increment in a goroutine:

    func (c *PossiblySafeCounter) inc() {
       c.mu.Lock();
       defer c.mu.Unlock();
       c.sum++
    }
    

    or do the locking and unlocking in the goroutine:

    func (c *PossiblySafeCounter) inc() {
       go func() {
           c.mu.Lock();
           defer c.mu.Unlock();
           c.sum++
       }() 
    }
    

    but honestly, doing any kind of goroutine doesn't make sense in this example. Why do you need the increment to be in a goroutine?

    本回答被题主选为最佳回答 , 对您是否有帮助呢?
    评论

报告相同问题?

悬赏问题

  • ¥15 Android STD快速启动
  • ¥15 如何使用simulink建立一个永磁同步直线电机模型?
  • ¥30 天体光谱图的的绘制并得到星表
  • ¥15 PointNet++的onnx模型只能使用一次
  • ¥20 西南科技大学数字信号处理
  • ¥15 有两个非常“自以为是”烦人的问题急期待大家解决!
  • ¥30 STM32 INMP441无法读取数据
  • ¥15 R语言绘制密度图,一个密度曲线内fill不同颜色如何实现
  • ¥100 求汇川机器人IRCB300控制器和示教器同版本升级固件文件升级包
  • ¥15 用visualstudio2022创建vue项目后无法启动