为什么在golang中写入全局地图时恢复不起作用

I have a global map, and I use many goroutines to write map concurrently without without limit. Then this of course will cause panic. So I add recover method in order to deal with panic. But it seemed they make no difference.
my code as below:

var m = make(map[int]int)

func main(){
    defer func(){
        if err:=recover();err!=nil{
            fmt.Printf("=====recover in main:%s
",err)
        }
    }()
    count := 1000
    for i:=0;i<count;i++{
        go AddEle()
    }
}

func AddEle(){
    defer func(){
        if err:=recover();err!=nil{
            fmt.Printf("====recover in child goroutines:%s",err)
        }
    }()
    for i:=1;i<1000;i++{
        m[i] = i
    }
}

and the output as below:

 fatal error: concurrent map writesc
 goroutine 5 [running]:
    runtime.throw(0x666514, 0x15)
/usr/local/go/src/runtime/panic.go:605 +0x95 fp=0xc42002ff48 
sp=0xc42002ff28 pc=0x42b2a5
runtime.mapassign_fast64(0x61c360, 0xc42007cbd0, 0x1, 0x0)
    /usr/local/go/src/runtime/hashmap_fast.go:519 +0x3d2 
fp=0xc42002ffa8 sp=0xc42002ff48 pc=0x40dbb2
    main.AddEle()
    /home/geek/go/work/src/web/main.go:34 +0x6a fp=0xc42002ffe0 
sp=0xc42002ffa8 pc=0x5e257a
runtime.goexit()
    /usr/local/go/src/runtime/asm_amd64.s:2337 +0x1 fp=0xc42002ffe8 
sp=0xc42002ffe0 pc=0x458821
created by main.main
    c/home/geek/go/work/src/web/main.go:22 +0x5at.

so I want to know why recover dont make effect.

The recover() function handles panics.

When the runtime detects a concurrent map access, the runtime exits the process with a call to fatalthrow.

The runtime does not panic in this situation, therefore the recover handler is not invoked.