英文:
Loop in goroutine hangs program
问题
为什么当注释掉time.Sleep(time.Nanosecond)时，以下程序会卡住？
package main
import "fmt"
import "time"
import "sync/atomic"
func main() {
var ops uint64 = 0
for i := 0; i < 50; i++ {
go func() {
for {
atomic.AddUint64(&ops, 1)
time.Sleep(time.Nanosecond)
}
}()
}
time.Sleep(time.Millisecond)
opsFinal := atomic.LoadUint64(&ops)
fmt.Println("ops:", opsFinal)
}
第二个问题，为什么在沙盒中运行这个程序会导致"process took too long"的错误？
英文:
Why does the following program hang when the time.Sleep(time.Nanosecond) is commented out?
package main
import "fmt"
import "time"
import "sync/atomic"
func main() {
var ops uint64 = 0
for i := 0; i < 50; i++ {
go func() {
for {
atomic.AddUint64(&ops, 1)
time.Sleep(time.Nanosecond)
}
}()
}
time.Sleep(time.Millisecond)
opsFinal := atomic.LoadUint64(&ops)
fmt.Println("ops:", opsFinal)
}
Second question, why does running this program in the sandbox result in "process took too long"?
答案1
得分: 10
这是要翻译的内容：
这是因为goroutine是协作式（而非完全抢占式）任务，只有在发生IO、系统调用、time.Sleep()或调用需要扩展栈的大型函数时才会发生上下文切换。
参考资料：
你的atomic.AddUint64(&ops, 1)是一个不需要扩展栈的小函数，因此不会发生上下文切换。
由于主线程也是一个goroutine，它不会进行上下文切换，并且永远处于睡眠状态。
有一个未解决的问题，即在紧密循环中使golang具有抢占性。
更多有用的参考资料：
英文:
It's because goroutine is cooperative (not fully preemptive) task, and the context switch only happen when there's some IO, system call, time.Sleep(), or calling large function which have to extend the stack.
Reference:
Your atomic.AddUint64(&ops, 1) is small function which does not have to extend stack. so the context switch never occurs.
Since main thread is also a goroutine, it does not acquire context switch, and sleeps forever.
There's an open issue to make golang preemptive in tight loop, but not yet solved.
More useful references:
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论