go unsafe常见应用

1.强制类型转换

2.string与[]byte的高效转换

3.高效动态替换

go 本身是类似java一样，底层运行虚拟机，整个程序处于托管状态(GC等)，要想和底层打交道需要用到unsafe模块，unsafe也常用于性能提升和程序灵活处理场景，本文介绍unsafe常用的几个场景。

常见unsafe用法如下

1.强制类型转换

要求必须内存布局一致，如下

// 正确转换必须保证内存布局一致
func TestConvert(t *testing.T) {var num int = 2numf := *(*float64)(unsafe.Pointer(&num))t.Logf("Convert int %v to float64 %v", num, numf)type MyInt intnums := []int{1, 2, 3}mynums := *(*[]MyInt)(unsafe.Pointer(&nums))t.Logf("Convert int array %v to myint array %v", nums, mynums)
}

结果如下, float64和int布局不一致所有不能正确转换

unsafe_test.go:15: Convert int 2 to float64 1e-323
unsafe_test.go:20: Convert int array [1 2 3] to myint array [1 2 3]

2.string与[]byte的高效转换

go的默认实现中，go和byte的转换需要通过拷贝，性能较低，如下测试

func BenchmarkByteString1(b *testing.B) {str := "this is a string"var bs []byteb.ResetTimer()for i := 0; i < b.N; i++ {bs = []byte(str)str = string(bs)}b.StopTimer()
}

结果如下

cpu: Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz
BenchmarkByteString1
BenchmarkByteString1-12 25468410 42.08 ns/op
PASS

实际上，如下，因为string []byte底层结构很像，

// []byte其实就是byte类型的切片，对应的底层结构体定义如下（在runtime/slice.go文件中）
type slice struct {array unsafe.Pointerlen   intcap   int
}// string对应的底层结构体定义如下（在runtime/string.go文件中）
type stringStruct struct {str unsafe.Pointerlen int
}

因此，可以如下直接转换

// string转ytes
func Str2sbyte(s string) (b []byte) {*(*string)(unsafe.Pointer(&b)) = s    // 把s的地址付给b*(*int)(unsafe.Pointer(uintptr(unsafe.Pointer(&b)) + 2*unsafe.Sizeof(&b))) = len(s)    // 修改容量为长度return
}// []byte转string
func Sbyte2str(b []byte) string {return *(*string)(unsafe.Pointer(&b))
}func BenchmarkByteString2(b *testing.B) {str := "this is a string"var bs []byteb.ResetTimer()for i := 0; i < b.N; i++ {bs = Str2sbyte(str)str = Sbyte2str(bs)}b.StopTimer()
}

结果如下，大大提升

cpu: Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz
BenchmarkByteString2
BenchmarkByteString2-12 1000000000 0.3190 ns/op
PASS

实际上，go本身的实现也用到这个技巧，参考runtime/string.go中slicebytetostringtmp和stringtoslicebytetmp两个函数，如下

func slicebytetostringtmp(b []byte) string {// Return a "string" referring to the actual []byte bytes.// This is only for use by internal compiler optimizations// that know that the string form will be discarded before// the calling goroutine could possibly modify the original// slice or synchronize with another goroutine.// First such case is a m[string(k)] lookup where// m is a string-keyed map and k is a []byte.// Second such case is "<"+string(b)+">" concatenation where b is []byte.// Third such case is string(b)=="foo" comparison where b is []byte.if raceenabled && len(b) > 0 {racereadrangepc(unsafe.Pointer(&b[0]),uintptr(len(b)),getcallerpc(unsafe.Pointer(&b)),funcPC(slicebytetostringtmp))}return *(*string)(unsafe.Pointer(&b))
}func stringtoslicebytetmp(s string) []byte {// Return a slice referring to the actual string bytes.// This is only for use by internal compiler optimizations// that know that the slice won't be mutated.// The only such case today is:// for i, c := range []byte(str)str := (*stringStruct)(unsafe.Pointer(&s))ret := slice{array: unsafe.Pointer(str.str), len: str.len, cap: str.len}return *(*[]byte)(unsafe.Pointer(&ret))
}

引用文章Go语言中[]byte和string类型相互转换时的性能分析和优化，描述如下

stringtoslicebytetmp调用的前提是保证返回的[]byte之后不会被修改，只用于编译器内部优化，目前唯一的场景是在for loop中将string转换成[]byte做遍历操作时，比如 for i, c := range []byte(str)

slicebytetostringtmp调用的前提其实也是类似，保证返回的string在生命周期结束之前，[]byte不会被修改，也是只用于编译器内部优化，目前有三种场景：

假设有一个key为string的map遍历m，你想使用[]byte类型的变量k做查找操作，比如 m[string(k)]

做字符串拼接操作时，比如 <"+string(b)+">，其中b是[]byte类型

[]byte类型和常量字符串做比较操作，比如 string(b)=="foo"

3.高效动态替换

一个典型的场景是，一块共享的buffer，很多程序在同时读写，如何高性能的处理共享竞争问题？一个可行的方案是，写入时先写一个copy，然后原子替换老的内容，指针保持不变，如下

// 原子替换内存地址
func TestBuffer(t *testing.T) {var buffer unsafe.Pointervar wg sync.WaitGroupvar writeFn = func(index int) {b := make([]int, 0)b = append(b, index)b = append(b, index)b = append(b, index)atomic.StorePointer(&buffer, unsafe.Pointer(&b))}var readFn = func() {b := atomic.LoadPointer(&buffer)data := *(*[]int)(b)t.Log(b, data)}// 初始化writeFn(0)// 写入for i := 0; i < 10; i++ {wg.Add(1)go func(index int) {writeFn(i)time.Sleep(time.Millisecond * 100)wg.Done()}(i)}// 读取for i := 0; i < 3; i++ {wg.Add(1)go func() {readFn()time.Sleep(time.Millisecond * 100)wg.Done()}()}wg.Wait()
}

原创，转载请注明来自

博客https://blog.csdn.net/wenzhou1219
个人网站http://jimwen.net/