Pinger.Stop() is unsafe when used with a timeout or count about ping HOT 4 CLOSED

I see what's going on there. The time.AfterFunc() could be an interrupt handler, for example, which might race if the user hits Ctrl+C at or near the configured pinger.Timeout.

I wonder whether pinger.Stop() could be made to return an error in cases when it is called more than once and/or whether this is actually useful to anyone?

from ping.

It looks like #83 by @voidint already fixes this issue with a sync.Once

from ping.

Unable to reproduce, I'm afraid. If you can provide a concrete example that fails when using the latest commit on master then we can reopen.

from ping.

The issue is that the internal timeout and packet counting mechanisms close the p.done channel, which Stop() also closes. You can reproduce with a contrived example like:

https://gist.github.com/ks07/4f9a57af233027b194801323017b0141

2020/10/17 17:33:45 24 bytes from 8.8.8.8: icmp_seq=0 time=31.0067ms
panic: close of closed channel

goroutine 9 [running]:
github.com/go-ping/ping.(*Pinger).Stop(...)
	C:/Users/George/go/pkg/mod/github.com/go-ping/[email protected]/ping.go:381
main.main.func2()
	C:/Users/George/go/src/pingerExample/main.go:30 +0x5d
created by time.goFunc
	C:/Programs/Go/src/time/sleep.go:167 +0x4b

Process finished with exit code 2

In a real world use case the time.AfterFunc() call would be some other cancellation method handled in a separate goroutine.

from ping.