从源码的角度去学习 Go slice
点击上方“Go编程时光”，选择“加为星标”
第一时间关注Go技术干货！
作者：吴国华
原文：https://www.kevinwu0904.top/blogs/golang-slice/
slice是golang开发中最常用到的内置类型之一。与数组相比，它具有长度不固定、可动态添加元素的特性。
# 版本说明
本文涉及到的源码解析部分来源于go官方的1.15.8版本。
# 定义
slice在golang的runtime层面定义如下：
type slice struct {
array unsafe.Pointer // 数组元素
len int // 长度
cap int // 容量
}
slice包括3个字段：
array：表示当前slice存放的数组元素。
len：表示当前slice已经使用的长度。
cap：表示当前slice的总容量，即最大能够容纳的元素个数。
如图所示，slice的动态特性正是基于这样的数据结构：一方面，记录已经使用的元素个数；另一方面，元素存放在有容量上限的底层数组中。
# 创建slice
slice的创建需要借助golang builtin包中的make函数：
// The make built-in function allocates and initializes an object of type
// slice, map, or chan (only). Like new, the first argument is a type, not a
// value. Unlike new, make's return type is the same as the type of its
// argument, not a pointer to it. The specification of the result depends on
// the type:
// Slice: The size specifies the length. The capacity of the slice is
// equal to its length. A second integer argument may be provided to
// specify a different capacity; it must be no smaller than the
// length. For example, make([]int, 0, 10) allocates an underlying array
// of size 10 and returns a slice of length 0 and capacity 10 that is
// backed by this underlying array.
// Map: An empty map is allocated with enough space to hold the
// specified number of elements. The size may be omitted, in which case
// a small starting size is allocated.
// Channel: The channel's buffer is initialized with the specified
// buffer capacity. If zero, or the size is omitted, the channel is
// unbuffered.
func make(t Type, size ...IntegerType) Type
make在golang中是一个创建内置类型的通用函数，不仅仅是slice，包括chan、map都需要借助make函数来创建。slice的创建主要包括如下两种形式：
nums := make([]int, 10) // 创建长度和容量均为10的int slice
nums := make([]int, 0, 10) // 创建长度为0，容量为10的int slice
以下面的代码为例：
package main
import (
"fmt"
)
func main() {
nums := make([]int, 0, 10)
nums = append(nums, 0)
fmt.Println(nums)
}
让我们通过go的官方工具来查看汇编结果：
$ go tool compile -S main.go
"".main STEXT size=217 args=0x0 locals=0x58 funcid=0x0
0x0000 00000 (main.go:7) TEXT "".main(SB), ABIInternal, $88-0
0x0000 00000 (main.go:7) MOVQ (TLS), CX
0x0009 00009 (main.go:7) CMPQ SP, 16(CX)
0x000d 00013 (main.go:7) PCDATA $0, $-2
0x000d 00013 (main.go:7) JLS 207
0x0013 00019 (main.go:7) PCDATA $0, $-1
0x0013 00019 (main.go:7) SUBQ $88, SP
0x0017 00023 (main.go:7) MOVQ BP, 80(SP)
0x001c 00028 (main.go:7) LEAQ 80(SP), BP
0x0021 00033 (main.go:7) FUNCDATA $0, gclocals·33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0021 00033 (main.go:7) FUNCDATA $1, gclocals·f207267fbf96a0178e8758c6e3e0ce28(SB)
0x0021 00033 (main.go:7) FUNCDATA $2, "".main.stkobj(SB)
0x0021 00033 (main.go:8) LEAQ type.int(SB), AX
0x0028 00040 (main.go:8) MOVQ AX, (SP)
0x002c 00044 (main.go:8) MOVQ $0, 8(SP)
0x0035 00053 (main.go:8) MOVQ $10, 16(SP)
0x003e 00062 (main.go:8) PCDATA $1, $0
0x003e 00062 (main.go:8) NOP
0x0040 00064 (main.go:8) CALL runtime.makeslice(SB) # 从这一行我们可以看出slice的创建最终底层实现是runtime.makeslice函数
0x0045 00069 (main.go:8) MOVQ 24(SP), AX
0x004a 00074 (main.go:9) MOVQ $0, (AX)
0x0051 00081 (main.go:10) MOVQ AX, (SP)
0x0055 00085 (main.go:10) MOVQ $1, 8(SP)
0x005e 00094 (main.go:10) MOVQ $10, 16(SP)
0x0067 00103 (main.go:10) CALL runtime.convTslice(SB)
0x006c 00108 (main.go:10) MOVQ 24(SP), AX
0x0071 00113 (main.go:10) XORPS X0, X0
0x0074 00116 (main.go:10) MOVUPS X0, ""..autotmp_13+64(SP)
0x0079 00121 (main.go:10) LEAQ type.[]int(SB), CX
0x0080 00128 (main.go:10) MOVQ CX, ""..autotmp_13+64(SP)
0x0085 00133 (main.go:10) MOVQ AX, ""..autotmp_13+72(SP)
... 省略不相关部分
PS：请读者先不要迷失在汇编代码的细节中，目前我们只需要关注"(main.go:8)“对应底层实现是调用go的runtime.makeslice函数，了解这一点即可。
因此，我们进一步去翻阅相关的go源码：
func makeslice(et *_type, len, cap int) unsafe.Pointer { // _type是go所有类型在runtime层面的表示形式
mem, overflow := math.MulUintptr(et.size, uintptr(cap)) // 预期内存开销是：元素类型长度x元素个数，同时判断是否算术溢出
if overflow || mem > maxAlloc || len < 0 || len > cap {
// NOTE: Produce a 'len out of range' error instead of a
// 'cap out of range' error when someone does make([]T, bignumber).
// 'cap out of range' is true too, but since the cap is only being
// supplied implicitly, saying len is clearer.
// See golang.org/issue/4085.
mem, overflow := math.MulUintptr(et.size, uintptr(len))
if overflow || mem > maxAlloc || len < 0 {
panicmakeslicelen()
}
panicmakeslicecap()
}
return mallocgc(mem, et, true) // 申请内存块
}
makeslice函数计算出所需的内存开销之后，会检查是否算术溢出，紧接着通过mallocgc函数去申请内存块。PS：由于mallocgc涉及到比较多的go内存管理相关的知识点，因此不在本文展开。
至此，slice就完成了内存申请。
# 复制slice
slice的复制可以借助golang builtin包中的copy函数：
// The copy built-in function copies elements from a source slice into a
// destination slice. (As a special case, it also will copy bytes from a
// string to a slice of bytes.) The source and destination may overlap. Copy
// returns the number of elements copied, which will be the minimum of
// len(src) and len(dst).
func copy(dst, src []Type) int
以下面的代码为例：
package main
import (
"fmt"
)
func main() {
src := []int{ 1, 2, 3 }
tar := make([]int, 2)
copy(tar, src)
fmt.Println(tar)
}
让我们通过go的官方工具来查看汇编结果：
$ go tool compile -S main.go"".main STEXT size=261 args=0x0 locals=0x70 funcid=0x0 0x0000 00000 (main.go:7) TEXT "".main(SB), ABIInternal, $112-0 0x0000 00000 (main.go:7) MOVQ (TLS), CX 0x0009 00009 (main.go:7) CMPQ SP, 16(CX) 0x000d 00013 (main.go:7) PCDATA $0, $-2 0x000d 00013 (main.go:7) JLS 249 0x0013 00019 (main.go:7) PCDATA $0, $-1 0x0013 00019 (main.go:7) SUBQ $112, SP 0x0017 00023 (main.go:7) MOVQ BP, 104(SP) 0x001c 00028 (main.go:7) LEAQ 104(SP), BP 0x0021 00033 (main.go:7) FUNCDATA $0, gclocals·33cdeccccebe80329f1fdbee7f5874cb(SB) 0x0021 00033 (main.go:7) FUNCDATA $1, gclocals·f207267fbf96a0178e8758c6e3e0ce28(SB) 0x0021 00033 (main.go:7) FUNCDATA $2, "".main.stkobj(SB) 0x0021 00033 (main.go:8) MOVQ $0, ""..autotmp_12+64(SP) 0x002a 00042 (main.go:8) XORPS X0, X0 0x002d 00045 (main.go:8) MOVUPS X0, ""..autotmp_12+72(SP) 0x0032 00050 (main.go:8) MOVQ $1, ""..autotmp_12+64(SP) 0x003b 00059 (main.go:8) MOVQ $2, ""..autotmp_12+72(SP) 0x0044 00068 (main.go:8) MOVQ $3, ""..autotmp_12+80(SP) 0x004d 00077 (main.go:9) LEAQ type.int(SB), AX 0x0054 00084 (main.go:9) MOVQ AX, (SP) 0x0058 00088 (main.go:9) MOVQ $2, 8(SP) 0x0061 00097 (main.go:9) MOVQ $3, 16(SP) 0x006a 00106 (main.go:9) LEAQ ""..autotmp_12+64(SP), AX 0x006f 00111 (main.go:9) MOVQ AX, 24(SP) 0x0074 00116 (main.go:9) PCDATA $1, $0 0x0074 00116 (main.go:9) CALL runtime.makeslicecopy(SB) # 调用runtime的makeslicecopy函数 0x0079 00121 (main.go:9) MOVQ 32(SP), AX 0x007e 00126 (main.go:11) MOVQ AX, (SP) 0x0082 00130 (main.go:11) MOVQ $2, 8(SP) 0x008b 00139 (main.go:11) MOVQ $2, 16(SP) 0x0094 00148 (main.go:11) CALL runtime.convTslice(SB) 0x0099 00153 (main.go:11) MOVQ 24(SP), AX 0x009e 00158 (main.go:11) XORPS X0, X0 0x00a1 00161 (main.go:11) MOVUPS X0, ""..autotmp_17+88(SP) 0x00a6 00166 (main.go:11) LEAQ type.[]int(SB), CX 0x00ad 00173 (main.go:11) MOVQ CX, ""..autotmp_17+88(SP) 0x00b2 00178 (main.go:11) MOVQ AX, ""..autotmp_17+96(SP) ... 省略不相关部分
通过汇编代码的结果可以看出，copy函数在go的runtime层面是调用了runtime.makeslicecopy函数。因此，我们进一步查看相关源码：
// makeslicecopy allocates a slice of "tolen" elements of type "et",
// then copies "fromlen" elements of type "et" into that new allocation from "from".
func makeslicecopy(et *_type, tolen int, fromlen int, from unsafe.Pointer) unsafe.Pointer { // 参数分别是：数据类型、目标长度、来源长度、来源元素数组
/*
计算需要复制的内存块大小：
1. 目标slice长度 > 来源slice长度，以来源长度为准
2. 目标slice长度 <= 来源slice长度，以目标长度为准
*/
var tomem, copymem uintptr
if uintptr(tolen) > uintptr(fromlen) {
var overflow bool
tomem, overflow = math.MulUintptr(et.size, uintptr(tolen))
if overflow || tomem > maxAlloc || tolen < 0 {
panicmakeslicelen()
}
copymem = et.size * uintptr(fromlen)
} else {
// fromlen is a known good length providing and equal or greater than tolen,
// thereby making tolen a good slice length too as from and to slices have the
// same element width.
tomem = et.size * uintptr(tolen)
copymem = tomem
}
/*
申请内存块
*/
var to unsafe.Pointer
if et.ptrdata == 0 {
to = mallocgc(tomem, nil, false)
if copymem < tomem {
memclrNoHeapPointers(add(to, copymem), tomem-copymem)
}
} else {
// Note: can't use rawmem (which avoids zeroing of memory), because then GC can scan uninitialized memory.
to = mallocgc(tomem, et, true)
if copymem > 0 && writeBarrier.enabled {
// Only shade the pointers in old.array since we know the destination slice to
// only contains nil pointers because it has been cleared during alloc.
bulkBarrierPreWriteSrcOnly(uintptr(to), uintptr(from), copymem)
}
}
/*
数据竞争检测和支持与内存清理程序的互操作
*/
if raceenabled {
callerpc := getcallerpc()
pc := funcPC(makeslicecopy)
racereadrangepc(from, copymem, callerpc, pc)
}
if msanenabled {
msanread(from, copymem)
}
// 复制内存块
memmove(to, from, copymem)
return to
}
小结一下，通过copy函数的底层实现，我们能够知道：
slice的复制是基于复制内存块来完成的。
slice的复制会考虑目标slice的长度，仅会复制不大于目标slice长度的元素。
# 新增元素
给slice新增元素需要借助golang builtin包中的append函数：
// The append built-in function appends elements to the end of a slice. If// it has sufficient capacity, the destination is resliced to accommodate the// new elements. If it does not, a new underlying array will be allocated.// Append returns the updated slice. It is therefore necessary to store the// result of append, often in the variable holding the slice itself:// slice = append(slice, elem1, elem2)// slice = append(slice, anotherSlice...)// As a special case, it is legal to append a string to a byte slice, like this:// slice = append([]byte("hello "), "world"...)func append(slice []Type, elems ...Type) []Type
日常使用示例如下：
nums := make([]int, 0, 10)
nums = append(nums, 1)
nums = append(nums, 5)
nums = append(nums, 6)
# 删除元素
不同于新增元素，golang中并未内置删除slice指定元素的函数，需要开发者自己去实现。就笔者所了解的删除元素方法主要有如下几种方式：
方法一：创建新的slice切片
func removeElem(src []int, toDel int) []int {
tar := make([]int, 0, len(src))
for _, num := range src {
if num != toDel {
tar = append(tar, num)
}
}
return tar
}
方法二：复用原有的slice切片
func removeElem(src []int, toDel int) []int {
tar := src[:0]
for _, num := range src {
if num != toDel {
tar = append(tar, num)
}
}
return tar
}
方法三：复用原有的slice切片，平移剩下的元素
func removeElem(src []int, toDel int) []int {
for i := 0; i < len(src); i++ {
if src[i] == toDel {
src = append(src[:i], src[i+1:]...)
i--
}
}
return src
}
小结一下，方法一的主要优点是不修改原有slice的数据，仅当不允许修改源slice数据的前提下推荐方法一。其它情况下，请使用方法二或者方法三，性能方面和开销方面都比较优秀。
# 扩容slice
动态扩容是slice最大的特点。我们日常使用slice的过程中，并不会感受到slice的真实容量变化情况，golang底层会帮助我们完成对应的slice扩容。
以下面的代码为例：
package main
import (
"fmt"
)
func main() {
nums := make([]int, 0, 2) // 初始设置为2
nums = append(nums, 1)
nums = append(nums, 2)
nums = append(nums, 3) // 触发扩容
nums = append(nums, 4)
nums = append(nums, 5)
fmt.Println(nums)
}
让我们通过go的官方工具来查看汇编结果：
$ go tool compile -S main.go
"".main STEXT size=490 args=0x0 locals=0x60 funcid=0x0
0x0000 00000 (main.go:7) TEXT "".main(SB), ABIInternal, $96-0
0x0000 00000 (main.go:7) MOVQ (TLS), CX
0x0009 00009 (main.go:7) CMPQ SP, 16(CX)
0x000d 00013 (main.go:7) PCDATA $0, $-2
0x000d 00013 (main.go:7) JLS 480
0x0013 00019 (main.go:7) PCDATA $0, $-1
0x0013 00019 (main.go:7) SUBQ $96, SP
0x0017 00023 (main.go:7) MOVQ BP, 88(SP)
0x001c 00028 (main.go:7) LEAQ 88(SP), BP
0x0021 00033 (main.go:7) FUNCDATA $0, gclocals·33cdeccccebe80329f1fdbee7f5874cb(SB)
0x0021 00033 (main.go:7) FUNCDATA $1, gclocals·f207267fbf96a0178e8758c6e3e0ce28(SB)
0x0021 00033 (main.go:7) FUNCDATA $2, "".main.stkobj(SB)
0x0021 00033 (main.go:8) LEAQ type.int(SB), AX
0x0028 00040 (main.go:8) MOVQ AX, (SP)
0x002c 00044 (main.go:8) MOVQ $0, 8(SP)
0x0035 00053 (main.go:8) MOVQ $2, 16(SP)
0x003e 00062 (main.go:8) PCDATA $1, $0
0x003e 00062 (main.go:8) NOP
0x0040 00064 (main.go:8) CALL runtime.makeslice(SB)
0x0045 00069 (main.go:8) MOVQ 24(SP), AX
0x004a 00074 (main.go:9) MOVQ $1, (AX)
0x0051 00081 (main.go:10) MOVQ $2, 8(AX)
0x0059 00089 (main.go:11) LEAQ type.int(SB), CX
0x0060 00096 (main.go:11) MOVQ CX, (SP)
0x0064 00100 (main.go:11) MOVQ AX, 8(SP)
0x0069 00105 (main.go:11) MOVQ $2, 16(SP)
0x0072 00114 (main.go:11) MOVQ $2, 24(SP)
0x007b 00123 (main.go:11) MOVQ $3, 32(SP)
0x0084 00132 (main.go:11) CALL runtime.growslice(SB) # 调用runtime的growslice函数
0x0089 00137 (main.go:11) MOVQ 40(SP), AX
0x008e 00142 (main.go:11) MOVQ 56(SP), CX
0x0093 00147 (main.go:11) MOVQ 48(SP), DX
0x0098 00152 (main.go:11) MOVQ $3, 16(AX)
0x00a0 00160 (main.go:11) INCQ DX
0x00a3 00163 (main.go:12) LEAQ 1(DX), BX
0x00a7 00167 (main.go:12) CMPQ CX, BX
0x00aa 00170 (main.go:12) JCS 403
0x00b0 00176 (main.go:12) MOVQ $4, (AX)(DX*8)
0x00b8 00184 (main.go:13) LEAQ 1(BX), DX
0x00bc 00188 (main.go:13) NOP
0x00c0 00192 (main.go:13) CMPQ CX, DX
0x00c3 00195 (main.go:13) JCS 325
0x00c9 00201 (main.go:13) MOVQ $5, (AX)(BX*8)
0x00d1 00209 (main.go:14) MOVQ AX, (SP)
0x00d5 00213 (main.go:14) MOVQ DX, 8(SP)
0x00da 00218 (main.go:14) MOVQ CX, 16(SP)
0x00df 00223 (main.go:14) NOP
0x00e0 00224 (main.go:14) CALL runtime.convTslice(SB)
0x00e5 00229 (main.go:14) MOVQ 24(SP), AX
... 省略不相关部分
因此，我们进一步查看相关源码：
func growslice(et *_type, old slice, cap int) slice {
if raceenabled {
callerpc := getcallerpc()
racereadrangepc(old.array, uintptr(old.len*int(et.size)), callerpc, funcPC(growslice))
}
if msanenabled {
msanread(old.array, uintptr(old.len*int(et.size)))
}
if cap < old.cap {
panic(errorString("growslice: cap out of range"))
}
if et.size == 0 {
// append should not create a slice with nil pointer but non-zero len.
// We assume that append doesn't need to preserve old.array in this case.
return slice{unsafe.Pointer(&zerobase), old.len, cap}
}
/*
扩容规则：
1. 长度 < 1024的情况下，扩容2倍
2. 长度 > 1024的情况下，扩容1.25倍
*/
newcap := old.cap
doublecap := newcap + newcap
if cap > doublecap {
newcap = cap
} else {
if old.len < 1024 {
newcap = doublecap
} else {
// Check 0 < newcap to detect overflow
// and prevent an infinite loop.
for 0 < newcap && newcap < cap {
newcap += newcap / 4
}
// Set newcap to the requested cap when
// the newcap calculation overflowed.
if newcap <= 0 {
newcap = cap
}
}
}
/*
内存对齐操作，对齐之后的容量相比于上述规则来说会有所偏差！
*/
var overflow bool
var lenmem, newlenmem, capmem uintptr
// Specialize for common values of et.size.
// For 1 we don't need any division/multiplication.
// For sys.PtrSize, compiler will optimize division/multiplication into a shift by a constant.
// For powers of 2, use a variable shift.
switch {
case et.size == 1:
lenmem = uintptr(old.len)
newlenmem = uintptr(cap)
capmem = roundupsize(uintptr(newcap))
overflow = uintptr(newcap) > maxAlloc
newcap = int(capmem)
case et.size == sys.PtrSize:
lenmem = uintptr(old.len) * sys.PtrSize
newlenmem = uintptr(cap) * sys.PtrSize
capmem = roundupsize(uintptr(newcap) * sys.PtrSize)
overflow = uintptr(newcap) > maxAlloc/sys.PtrSize
newcap = int(capmem / sys.PtrSize)
case isPowerOfTwo(et.size):
var shift uintptr
if sys.PtrSize == 8 {
// Mask shift for better code generation.
shift = uintptr(sys.Ctz64(uint64(et.size))) & 63
} else {
shift = uintptr(sys.Ctz32(uint32(et.size))) & 31
}
lenmem = uintptr(old.len) << shift
newlenmem = uintptr(cap) << shift
capmem = roundupsize(uintptr(newcap) << shift)
overflow = uintptr(newcap) > (maxAlloc >> shift)
newcap = int(capmem >> shift)
default:
lenmem = uintptr(old.len) * et.size
newlenmem = uintptr(cap) * et.size
capmem, overflow = math.MulUintptr(et.size, uintptr(newcap))
capmem = roundupsize(capmem)
newcap = int(capmem / et.size)
}
// The check of overflow in addition to capmem > maxAlloc is needed
// to prevent an overflow which can be used to trigger a segfault
// on 32bit architectures with this example program:
//
// type T [1<<27 + 1]int64
//
// var d T
// var s []T
//
// func main() {
// s = append(s, d, d, d, d)
// print(len(s), "\n")
// }
if overflow || capmem > maxAlloc {
panic(errorString("growslice: cap out of range"))
}
/*
分配内存块
*/
var p unsafe.Pointer
if et.ptrdata == 0 {
p = mallocgc(capmem, nil, false)
// The append() that calls growslice is going to overwrite from old.len to cap (which will be the new length).
// Only clear the part that will not be overwritten.
memclrNoHeapPointers(add(p, newlenmem), capmem-newlenmem)
} else {
// Note: can't use rawmem (which avoids zeroing of memory), because then GC can scan uninitialized memory.
p = mallocgc(capmem, et, true)
if lenmem > 0 && writeBarrier.enabled {
// Only shade the pointers in old.array since we know the destination slice p
// only contains nil pointers because it has been cleared during alloc.
bulkBarrierPreWriteSrcOnly(uintptr(p), uintptr(old.array), lenmem-et.size+et.ptrdata)
}
}
/*
把旧slice的内容复制到新slice中
*/
memmove(p, old.array, lenmem)
return slice{p, old.len, newcap}
}
小结一下，通过growslice函数的底层实现，我们能够知道：
append操作在golang编译结果中会转化成runtime.growslice函数，对上层透明，因而开发者无需关心slice底层容量大小的问题。
slice触发扩容操作会创建新的容量数组，将旧的slice内容复制过去，开发者需要关注有多个变量引用同一个slice且发生扩容的情况。
slice的扩容原则是：对于长度 < 1024的情况下，2倍增长；对于长度 > 1024的情况下，1.25倍增长。但由于存在内存对齐，slice的容量在扩容结束后会有所偏差！
# 总结
本文详细介绍了Go的内置类型slice，以常见的代码为例，通过go官方工具的汇编结果解析了底层实现。另一方面，本文从源码角度得出一些开发者较为容易迷惑的语法问题，例如slice的复制长度问题、slice扩容之后的底层数组更替问题、slice扩容容量大小问题等等。
总的来说，源码学习是一条漫漫长路。笔者认为开发者既不应该无视源码而只靠背诵一些结论来学习语言，也不应该过分关注源码细节导致无从入手。本文从slice的源码中，也牵扯出一些未能深入解析的概念：Go的runtime、Plan 9汇编、内存对齐等，关于这些概念的进一步解析则会随着笔者的深入学习而逐步完善，在此也希望能与读者共勉！
⬇⬇⬇
评论