Go1.19.3 map原理简析

map简析

map是一个集数组与链表(美貌与智慧)特性于一身的数据结构，其增删改查时间复杂度都非常优秀，在很多场景下用其替代树结构。关于map的特性请自行学习。
Go语言在语言层面就支持了map，而非其他语言(如Java)通过外置类库的方式实现。
使用 map[KeyType]ValueType 方式定义map类型变量。如 var mp map[string]int。
在使用map之前，必须先初始化map，使用 mp = make(map[string]int,cap)等形式。
一个未经初始化的map类型变量是可以读的，但读出来的内容皆为ValueType类型的零值。
map类型不支持并发的读写，但可以在非写入的状态并发的读取其中内容。
map的初始化同样是编译时与运行时共同努力的成果。

常用的位运算操作

移位左移1位相当于原数乘以2，右移1位相当于原数除以2

B = 5
1 << B = 00000001 << 5 = 00100000 = 32

2的B次幂 - 1 相当于将2的B-1次幂 + 2的B-2次幂 +…+ 2的0次幂

1 << B - 1 = 00100000 - 00000001 = 00011111 = 31

位清空M &^= N

01010101 &^ 00001111 = 01010000

位异或 M ^ N

01010101 ^ 10101010 = 11111111
11111111 ^ 10101010 = 01010101

Go map整体结构图

go map源码里的介绍

// This file contains the implementation of Go's map type.
//
// A map is just a hash table. The data is arranged
// into an array of buckets. Each bucket contains up to
// 8 key/elem pairs. The low-order bits of the hash are
// used to select a bucket. Each bucket contains a few
// high-order bits of each hash to distinguish the entries
// within a single bucket.
//
// If more than 8 keys hash to a bucket, we chain on
// extra buckets.
//
// When the hashtable grows, we allocate a new array
// of buckets twice as big. Buckets are incrementally
// copied from the old bucket array to the new bucket array.
//
// Map iterators walk through the array of buckets and
// return the keys in walk order (bucket #, then overflow
// chain order, then bucket index).  To maintain iteration
// semantics, we never move keys within their bucket (if
// we did, keys might be returned 0 or 2 times).  When
// growing the table, iterators remain iterating through the
// old table and must check the new table if the bucket
// they are iterating through has been moved ("evacuated")
// to the new table.

// Picking loadFactor: too large and we have lots of overflow
// buckets, too small and we waste a lot of space. I wrote
// a simple program to check some stats for different loads:
// (64-bit, 8 byte keys and elems)
//  loadFactor    %overflow  bytes/entry     hitprobe    missprobe
//        4.00         2.13        20.77         3.00         4.00
//        4.50         4.05        17.30         3.25         4.50
//        5.00         6.85        14.77         3.50         5.00
//        5.50        10.55        12.94         3.75         5.50
//        6.00        15.27        11.67         4.00         6.00
//        6.50        20.90        10.79         4.25         6.50
//        7.00        27.14        10.15         4.50         7.00
//        7.50        34.03         9.73         4.75         7.50
//        8.00        41.10         9.40         5.00         8.00
//
// %overflow   = percentage of buckets which have an overflow bucket
// bytes/entry = overhead bytes used per key/elem pair
// hitprobe    = # of entries to check when looking up a present key
// missprobe   = # of entries to check when looking up an absent key
//
// Keep in mind this data is for maximally loaded tables, i.e. just
// before the table grows. Typical tables will be somewhat less loaded.

哈希冲突解决方法

开放寻址法：线性探查、二次方探查、双散列等
拉链法 (Go语言中的map 使用了该方法解决哈希冲突）

负载因子 OverLoadFactor

map中负载因子 = 元素的数量 / 桶的数量， Go语言中map的负载因子为6.5
负载因子超过既定的值会引发扩容操作

map运行时的结构

以下代码若未标注出处，则均来自src/runtime/map.go文件

// A header for a Go map.
type hmap struct {
	// Note: the format of the hmap is also encoded in cmd/compile/internal/reflectdata/reflect.go.
	// Make sure this stays in sync with the compiler's definition.
	count     int // # live cells == size of map.  Must be first (used by len() builtin)
	flags     uint8
	B         uint8  // log_2 of # of buckets (can hold up to loadFactor * 2^B items)
	noverflow uint16 // approximate number of overflow buckets; see incrnoverflow for details
	hash0     uint32 // hash seed

	buckets    unsafe.Pointer // array of 2^B Buckets. may be nil if count==0.
	oldbuckets unsafe.Pointer // previous bucket array of half the size, non-nil only when growing
	nevacuate  uintptr        // progress counter for evacuation (buckets less than this have been evacuated)

	extra *mapextra // optional fields
}

count map中元素的数量，用内置函数len()获取的map数量就是该数量
flags 标记map状态的字段，如正在写入，正在迭代等
B 2的指数，map中一共有1 << B 个桶
noverflow 溢出桶的个数
hash0 哈希种子，对key做哈希时，会加入该种子，确保map的安全性
buckets 桶数组的指针
oldbuckets 旧桶数组的指针
nevacuate 该字段表示迁移进度，或当前应迁移桶的下标，即小于该字段值的桶都已迁移完成。
extra 溢出桶的指针

type mapextra struct {
	// If both key and elem do not contain pointers and are inline, then we mark bucket
	// type as containing no pointers. This avoids scanning such maps.
	// However, bmap.overflow is a pointer. In order to keep overflow buckets
	// alive, we store pointers to all overflow buckets in hmap.extra.overflow and hmap.extra.oldoverflow.
	// overflow and oldoverflow are only used if key and elem do not contain pointers.
	// overflow contains overflow buckets for hmap.buckets.
	// oldoverflow contains overflow buckets for hmap.oldbuckets.
	// The indirection allows to store a pointer to the slice in hiter.
	overflow    *[]*bmap
	oldoverflow *[]*bmap

	// nextOverflow holds a pointer to a free overflow bucket.
	nextOverflow *bmap
}

overflow 溢出桶数组指针，仅当key和elem非指针时才使用
oldoverflow 旧的溢出桶数组指针，仅当key和elem非指针时才使用
nextOverflow 下一个可用的溢出桶地址

// A bucket for a Go map.
type bmap struct {
	// tophash generally contains the top byte of the hash value
	// for each key in this bucket. If tophash[0] < minTopHash,
	// tophash[0] is a bucket evacuation state instead.
	tophash [bucketCnt]uint8
	// Followed by bucketCnt keys and then bucketCnt elems.
	// NOTE: packing all the keys together and then all the elems together makes the
	// code a bit more complicated than alternating key/elem/key/elem/... but it allows
	// us to eliminate padding which would be needed for, e.g., map[int64]int8.
	// Followed by an overflow pointer.
}

tophash 存储哈希值的高八位，加快增删改查寻址效率
key [bucketCnt]KeyType 存储key的数组
elem [bucketCnt]ElementType 存储value的数组
…
overflow *bmap 溢出桶指针

bmap是Go语言中map的桶类型，在源码中，bmap类型只有一个tophash字段。但在编译时期，Go编译器会根据用户代码自动注入相应的key，value等结构。bmap中最后一个字段是当前桶的溢出桶指针，一个桶可能会有多个溢出桶，他们是以链表的形式连接的。

map中需要的常量

const (
	// Maximum number of key/elem pairs a bucket can hold.
	bucketCntBits = 3   // bucketCount 左移 的位数 
	bucketCnt     = 1 << bucketCntBits  // 每个bucket中存储八组键值对

	// Maximum average load of a bucket that triggers growth is 6.5.
	// Represent as loadFactorNum/loadFactorDen, to allow integer math.
	loadFactorNum = 13    // 负载因子分子
	loadFactorDen = 2     // 负载因子分母

	// Maximum key or elem size to keep inline (instead of mallocing per element).
	// Must fit in a uint8.
	// Fast versions cannot handle big elems - the cutoff size for
	// fast versions in cmd/compile/internal/gc/walk.go must be at most this elem.
	maxKeySize  = 128    //Key 最大的size
	maxElemSize = 128    //Value(Elem)的最大size

	// data offset should be the size of the bmap struct, but needs to be
	// aligned correctly. For amd64p32 this means 64-bit alignment
	// even though pointers are 32 bit.
	dataOffset = unsafe.Offsetof(struct {     //key在内存中的偏移量
		b bmap
		v int64
	}{}.v)

	// Possible tophash values. We reserve a few possibilities for special marks.
	// Each bucket (including its overflow buckets, if any) will have either all or none of its
	// entries in the evacuated* states (except during the evacuate() method, which only happens
	// during map writes and thus no one else can observe the map during that time).
	emptyRest      = 0 // this cell is empty, and there are no more non-empty cells at higher indexes or overflows.  // tophash 数组中 emptyRest 代表其后单元格均为空，包括其后的所有溢出桶
	emptyOne       = 1 // this cell is empty // tophash 数组中 代表当前单元格为空
	evacuatedX     = 2 // key/elem is valid.  Entry has been evacuated to first half of larger table.  // 迁移后的元素落入的新桶的index 与 旧桶一致  位于 旧桶的 tophash 数组中
	evacuatedY     = 3 // same as above, but evacuated to second half of larger table. // 迁移后的元素落入的新桶的index 是 该元素在旧桶中的index的二倍  位于 旧桶的 tophash 数组中
	evacuatedEmpty = 4 // cell is empty, bucket is evacuated.  // 迁移后的旧桶中无元素，代表迁移完成 位于 旧桶的 tophash 数组中
	minTopHash     = 5 // minimum tophash for a normal filled cell.  // hash 计算完成后在若小于该值则 加上该值 保证 0 ~ 4 是预留的tophash数组中的标记状态

	// flags
	iterator     = 1 // there may be an iterator using buckets  // 一个迭代器在使用 buckets
	oldIterator  = 2 // there may be an iterator using oldbuckets // 同上，oldbuckets
	hashWriting  = 4 // a goroutine is writing to the map         // 一个协程正在向哈希表写内容
	sameSizeGrow = 8 // the current map growth is to a new map of the same size // 相等大小扩容

	// sentinel bucket ID for iterator checks
	noCheck = 1<<(8*goarch.PtrSize) - 1   // 哨兵 bucket ID 用于 迭代器检查， 表示无需检查的buckket
)

const maxZero = 1024 // must match value in reflect/value.go:maxZero cmd/compile/internal/gc/walk.go:zeroValSize
var zeroVal [maxZero]byte    // 1024字节大小的空内存区域，作为返回的零值使用
`

检测 tophash 代表的单元格是否为空

// isEmpty reports whether the given tophash array entry represents an empty bucket entry.
func isEmpty(x uint8) bool {
	return x <= emptyOne  // <= 1  1 or 0 -> empty
}

bucketShift 1 << b 代表桶的数量即 2 ^ b

// bucketShift returns 1<<b, optimized for code generation.
func bucketShift(b uint8) uintptr {
	// Masking the shift amount allows overflow checks to be elided.
	return uintptr(1) << (b & (goarch.PtrSize*8 - 1))
}

bucketMask 1 << b - 1 用hash的低B位与该值相与，即可得到hash对应的桶号

// bucketMask returns 1<<b - 1, optimized for code generation.
func bucketMask(b uint8) uintptr {
	return bucketShift(b) - 1
}

tophash hash的高八位用于确定桶中存放数据的具体单元格位置，若该值 < minTopHash则加之，为单元格预留表示状态的值

// tophash calculates the tophash value for hash.
func tophash(hash uintptr) uint8 {
	top := uint8(hash >> (goarch.PtrSize*8 - 8))
	if top < minTopHash {
		top += minTopHash
	}
	return top
}

evacuated 检查传入的桶是否已经迁移完

func evacuated(b *bmap) bool {
	h := b.tophash[0]
	return h > emptyOne && h < minTopHash  // 若迁移完成，则tophash[0]的状态会被修正，从下往上扫描，见后续， 此处检查状态即可
}

overflow 获取b的溢出桶指针，位于桶的低goarch.PtrSize位，长度为goarch.PtrSize

func (b *bmap) overflow(t *maptype) *bmap {
	return *(**bmap)(add(unsafe.Pointer(b), uintptr(t.bucketsize)-goarch.PtrSize))
}

setoverflow 设置b的溢出桶，将桶的溢出桶指针字段设置成ovf

func (b *bmap) setoverflow(t *maptype, ovf *bmap) {
	*(**bmap)(add(unsafe.Pointer(b), uintptr(t.bucketsize)-goarch.PtrSize)) = ovf
}

keys 根据偏移量拿到key的指针

func (b *bmap) keys() unsafe.Pointer {
	return add(unsafe.Pointer(b), dataOffset)
}

incrnoverflow 增加溢出桶个数，若溢出桶小于 2 ^ 16个则直接 + 1，否则有1/(1<<(h.B-15))的概率 + 1

// incrnoverflow increments h.noverflow.
// noverflow counts the number of overflow buckets.
// This is used to trigger same-size map growth.
// See also tooManyOverflowBuckets.
// To keep hmap small, noverflow is a uint16.
// When there are few buckets, noverflow is an exact count.
// When there are many buckets, noverflow is an approximate count.
func (h *hmap) incrnoverflow() {
	// We trigger same-size map growth if there are
	// as many overflow buckets as buckets.
	// We need to be able to count to 1<<h.B.
	if h.B < 16 {
		h.noverflow++
		return
	}
	// Increment with probability 1/(1<<(h.B-15)).
	// When we reach 1<<15 - 1, we will have approximately
	// as many overflow buckets as buckets.
	mask := uint32(1)<<(h.B-15) - 1
	// Example: if h.B == 18, then mask == 7,
	// and fastrand & 7 == 0 with probability 1/8.
	if fastrand()&mask == 0 {
		h.noverflow++
	}
}

newoverflow 创建一个溢出桶，如果是bucketArray中最后一个可用的溢出桶，该桶的溢出桶指针置空，h.extra.nextOverflow 置空，表示无可用溢出桶

func (h *hmap) newoverflow(t *maptype, b *bmap) *bmap {
	var ovf *bmap
	if h.extra != nil && h.extra.nextOverflow != nil { // hmap中溢出桶字段非空
		// We have preallocated overflow buckets available.
		// See makeBucketArray for more details.
		ovf = h.extra.nextOverflow
		if ovf.overflow(t) == nil { //溢出桶的溢出桶指针为空，代表有可用的溢出桶，在makebucketArray的时候，将最后一个溢出桶的溢出桶指针指向了bucket的头部位置，用该状态代表已经是最后一个溢出桶。
			// We're not at the end of the preallocated overflow buckets. Bump the pointer.
			h.extra.nextOverflow = (*bmap)(add(unsafe.Pointer(ovf), uintptr(t.bucketsize))) //下移可用的溢出桶指针
		} else { //代表已经是最后一个溢出桶了
			// This is the last preallocated overflow bucket.
			// Reset the overflow pointer on this bucket,
			// which was set to a non-nil sentinel value.
			ovf.setoverflow(t, nil) //其指针置空
			h.extra.nextOverflow = nil // 可用溢出桶指针置空
		}
	} else {  // hmap中溢出桶字段为空
		ovf = (*bmap)(newobject(t.bucket))  // 创建一个1个单位的新的溢出桶
	}
	h.incrnoverflow() //增长溢出桶个数
	if t.bucket.ptrdata == 0 { //若 桶中的元素非指针类型
		h.createOverflow() // 为hmap创建新的溢出桶，保证溢出桶字段非空
		*h.extra.overflow = append(*h.extra.overflow, ovf) // 将溢出桶指针加入overflow数组
	}
	b.setoverflow(t, ovf) // 为b设置溢出桶指针为ovf
	return ovf  // 返回ovf
}

createOverflow 为hmap创建新的溢出桶，原字段为空则创建

func (h *hmap) createOverflow() {
	if h.extra == nil {  // 为空则创建
		h.extra = new(mapextra)
	}
	if h.extra.overflow == nil {
		h.extra.overflow = new([]*bmap)
	}
}

makemap64 & makemap_small 对makemap的封装，使用场景不一

func makemap64(t *maptype, hint int64, h *hmap) *hmap {
	if int64(int(hint)) != hint { // 32位机器？
		hint = 0
	}
	return makemap(t, int(hint), h)
}

// makemap_small implements Go map creation for make(map[k]v) and
// make(map[k]v, hint) when hint is known to be at most bucketCnt
// at compile time and the map needs to be allocated on the heap.
func makemap_small() *hmap { // 桶的size 最多为bucketcount时...
	h := new(hmap)
	h.hash0 = fastrand()
	return h
}

makemap 创建一个map，内置函数make的封装

// makemap implements Go map creation for make(map[k]v, hint).
// If the compiler has determined that the map or the first bucket
// can be created on the stack, h and/or bucket may be non-nil.
// If h != nil, the map can be created directly in h.
// If h.buckets != nil, bucket pointed to can be used as the first bucket.
func makemap(t *maptype, hint int, h *hmap) *hmap {
	mem, overflow := math.MulUintptr(uintptr(hint), t.bucket.size) //根据桶数量和桶大小计算所需开辟内存，及其是否溢出
	if overflow || mem > maxAlloc {
		hint = 0
	}

	// initialize Hmap
	if h == nil {
		h = new(hmap)
	}
	h.hash0 = fastrand() // 初始化哈希种子，该算法使用了wangyi_fudan的库，是国内大佬的

	// Find the size parameter B which will hold the requested # of elements.
	// For hint < 0 overLoadFactor returns false since hint < bucketCnt.
	B := uint8(0)
	for overLoadFactor(hint, B) {  //根据负载因子公式计算B的个数，保证其不大不小，刚刚好
		B++
	}
	h.B = B

	// allocate initial hash table
	// if B == 0, the buckets field is allocated lazily later (in mapassign)
	// If hint is large zeroing this memory could take a while.
	if h.B != 0 {
		var nextOverflow *bmap
		h.buckets, nextOverflow = makeBucketArray(t, h.B, nil) // 创建buckets，将其与第一个可用的溢出桶指针一同返回
		if nextOverflow != nil { // hmap的溢出桶赋值
			h.extra = new(mapextra)
			h.extra.nextOverflow = nextOverflow
		}
	}

	return h
}

overLoadFactor 返回count是否符合负载因子

// overLoadFactor reports whether count items placed in 1<<B buckets is over loadFactor.
func overLoadFactor(count int, B uint8) bool {
	return count > bucketCnt && uintptr(count) > loadFactorNum*(bucketShift(B)/loadFactorDen)
}

makeBucketArray 创建buckets，并返回其后的第一个可用的溢出桶指针

// makeBucketArray initializes a backing array for map buckets.
// 1<<b is the minimum number of buckets to allocate.
// dirtyalloc should either be nil or a bucket array previously
// allocated by makeBucketArray with the same t and b parameters.
// If dirtyalloc is nil a new backing array will be alloced and
// otherwise dirtyalloc will be cleared and reused as backing array.
func makeBucketArray(t *maptype, b uint8, dirtyalloc unsafe.Pointer) (buckets unsafe.Pointer, nextOverflow *bmap) {
	base := bucketShift(b)
	nbuckets := base
	// For small b, overflow buckets are unlikely.
	// Avoid the overhead of the calculation.
	if b >= 4 {
		// Add on the estimated number of overflow buckets
		// required to insert the median number of elements
		// used with this value of b.
		nbuckets += bucketShift(b - 4)
		sz := t.bucket.size * nbuckets
		up := roundupsize(sz)  // 向上取整
		if up != sz {
			nbuckets = up / t.bucket.size   // 最终所有桶的个数
		}
	}

	if dirtyalloc == nil { // 脏的内存为空
		buckets = newarray(t.bucket, int(nbuckets)) // 直接分配
	} else { // 复用内存
		// dirtyalloc was previously generated by
		// the above newarray(t.bucket, int(nbuckets))
		// but may not be empty.
		buckets = dirtyalloc
		size := t.bucket.size * nbuckets
		if t.bucket.ptrdata != 0 { // 桶内保存的是key value的指针
			memclrHasPointers(buckets, size)  // 清理内存
		} else {
			memclrNoHeapPointers(buckets, size)
		}
	}

	if base != nbuckets { // 证明创建过了溢出桶
		// We preallocated some overflow buckets.
		// To keep the overhead of tracking these overflow buckets to a minimum,
		// we use the convention that if a preallocated overflow bucket's overflow
		// pointer is nil, then there are more available by bumping the pointer.
		// We need a safe non-nil pointer for the last overflow bucket; just use buckets.
		nextOverflow = (*bmap)(add(buckets, base*uintptr(t.bucketsize))) // 第一个可用的溢出桶指针
		last := (*bmap)(add(buckets, (nbuckets-1)*uintptr(t.bucketsize))) // 最后一个溢出桶的指针
		last.setoverflow(t, (*bmap)(buckets)) // 最后一个溢出桶的指针，指向桶数组开头，用该状态代表溢出桶用尽，该桶为最后一个溢出桶。
	}
	return buckets, nextOverflow
}

若B > 4 则创建 B - 4 个溢出桶
溢出桶与桶使用的是同一片连续的内存区域
最终所有桶的个数：t.bucket.size * (桶的个数 + 溢出桶的个数 ) 向上取整后除以t.bucket.size。

mapaccess1 相当于 v := Map[key] 即返回key 对应的 v

// mapaccess1 returns a pointer to h[key].  Never returns nil, instead
// it will return a reference to the zero object for the elem type if
// the key is not in the map.
// NOTE: The returned pointer may keep the whole map live, so don't
// hold onto it for very long.
func mapaccess1(t *maptype, h *hmap, key unsafe.Pointer) unsafe.Pointer {
	if raceenabled && h != nil {
		callerpc := getcallerpc()
		pc := abi.FuncPCABIInternal(mapaccess1)
		racereadpc(unsafe.Pointer(h), callerpc, pc)
		raceReadObjectPC(t.key, key, callerpc, pc)
	}
	if msanenabled && h != nil {
		msanread(key, t.key.size)
	}
	if asanenabled && h != nil {
		asanread(key, t.key.size)
	}
	if h == nil || h.count == 0 {   // map 是空的
		if t.hashMightPanic() {
			t.hasher(key, 0) // see issue 23734
		}
		return unsafe.Pointer(&zeroVal[0]) // 返回一个零值
	}
	if h.flags&hashWriting != 0 {  // 若map有其他协程写入，则panic
		fatal("concurrent map read and map write")
	}
	hash := t.hasher(key, uintptr(h.hash0))   // 使用key 和 哈希种子 计算哈希 算法位于 src/runtime/alg.go 文件中
	m := bucketMask(h.B) // 计算桶的掩码， 若B为10 则m为 1 << 10 - 1 = 0b111111111 = 1023
	b := (*bmap)(add(h.buckets, (hash&m)*uintptr(t.bucketsize))) // hash & m 计算哈希应落入哪个桶 再 * t.bucketsize 即为第N个桶的偏移地址 加上h.buckets起始地址，b为选中桶的起始地址 
	if c := h.oldbuckets; c != nil { // 若旧桶不为空
		if !h.sameSizeGrow() { // 非等量扩容
			// There used to be half as many buckets; mask down one more power of two.
			m >>= 1   // 旧桶的下标
		}
		oldb := (*bmap)(add(c, (hash&m)*uintptr(t.bucketsize))) // 旧桶的地址
		if !evacuated(oldb) { // 如果未迁移完成
			b = oldb
		}
	}
	top := tophash(hash) // 计算高八位哈希
bucketloop:
	for ; b != nil; b = b.overflow(t) { //遍历 桶 和 溢出桶
		for i := uintptr(0); i < bucketCnt; i++ { // 遍历桶中八组单元
			if b.tophash[i] != top {  // tophash 不相等的情况下检查当前存储单元标记是否为emtyRest
				if b.tophash[i] == emptyRest { // i 后无有效值
					break bucketloop // 退出
				}
				continue
			}
			k := add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize)) // 根据偏移量计算当前单元key的地址
			if t.indirectkey() { // 如果key是间接引用 即指针
				k = *((*unsafe.Pointer)(k)) // 解引用
			}
			if t.key.equal(key, k) { // 算法位于src/runtime/alg.go文件中，比较key是否相等
				e := add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize)) // 计算value / elem 的地址
				if t.indirectelem() { // 解引用value / elem
					e = *((*unsafe.Pointer)(e))
				}
				return e  // 返回value的地址
			}
		}
	}
	return unsafe.Pointer(&zeroVal[0]) // 返回一片空的内存区域的起始地址
}

访问map原理

若hmap为空或hmap的元素数量等于0，则返回空的内存区域代表零值，即空的map也可以访问，但返回的是“零值”
调用hash := t.hasher(key, uintptr(h.hash0)) 计算哈希值
使用m := bucketMask(h.B)计算桶位置的掩码
使用hash&m计算桶的位置，相当于取余操作hash % m
b := (*bmap)(add(h.buckets, (hash&m)*uintptr(t.bucketsize))) 桶起始地址 + 偏移地址 = 目标桶的起始地址
如果hmap有旧桶，若是(非等量扩容)2倍扩容，则m除以2，得到所在原桶中索引
若旧桶未迁移完成，并且查询的元素所在的桶号位于旧桶，则去旧桶中查询 b = oldb
top := tophash(hash) 得到高八位哈希
查询的步骤是两层循环，外层循环遍历桶及溢出桶，链式结构。内存循环遍历桶内的八组单元，对比tophash，若某个tophash 等于 emptyRest，则证明后续单元格及溢出桶均为空，则break掉两层循环。否则计算k单元格的偏移量，取出k 和传入的key做比较，若相同计算elem的偏移量，并返回。
src/runtime/alg.go 中定义了具体类型的hash等算法

// in asm_*.s
func memhash(p unsafe.Pointer, h, s uintptr) uintptr
func memhash32(p unsafe.Pointer, h uintptr) uintptr
func memhash64(p unsafe.Pointer, h uintptr) uintptr
func strhash(p unsafe.Pointer, h uintptr) uintptr

func strhashFallback(a unsafe.Pointer, h uintptr) uintptr {
	x := (*stringStruct)(a)
	return memhashFallback(x.str, h, uintptr(x.len))
}

mapaccess2 相当于 v, ok := Map[key] 即返回key 对应的 v, 及key是否在map中存在，原理同mapaccess1 一致，只不过返回值中多了一个bool变量

func mapaccess2(t *maptype, h *hmap, key unsafe.Pointer) (unsafe.Pointer, bool) {
	if raceenabled && h != nil {
		callerpc := getcallerpc()
		pc := abi.FuncPCABIInternal(mapaccess2)
		racereadpc(unsafe.Pointer(h), callerpc, pc)
		raceReadObjectPC(t.key, key, callerpc, pc)
	}
	if msanenabled && h != nil {
		msanread(key, t.key.size)
	}
	if asanenabled && h != nil {
		asanread(key, t.key.size)
	}
	if h == nil || h.count == 0 {
		if t.hashMightPanic() {
			t.hasher(key, 0) // see issue 23734
		}
		return unsafe.Pointer(&zeroVal[0]), false
	}
	if h.flags&hashWriting != 0 {
		fatal("concurrent map read and map write")
	}
	hash := t.hasher(key, uintptr(h.hash0))
	m := bucketMask(h.B)
	b := (*bmap)(add(h.buckets, (hash&m)*uintptr(t.bucketsize)))
	if c := h.oldbuckets; c != nil {
		if !h.sameSizeGrow() {
			// There used to be half as many buckets; mask down one more power of two.
			m >>= 1
		}
		oldb := (*bmap)(add(c, (hash&m)*uintptr(t.bucketsize)))
		if !evacuated(oldb) {
			b = oldb
		}
	}
	top := tophash(hash)
bucketloop:
	for ; b != nil; b = b.overflow(t) {
		for i := uintptr(0); i < bucketCnt; i++ {
			if b.tophash[i] != top {
				if b.tophash[i] == emptyRest {
					break bucketloop
				}
				continue
			}
			k := add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize))
			if t.indirectkey() {
				k = *((*unsafe.Pointer)(k))
			}
			if t.key.equal(key, k) {
				e := add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize))
				if t.indirectelem() {
					e = *((*unsafe.Pointer)(e))
				}
				return e, true  // 存在
			}
		}
	}
	return unsafe.Pointer(&zeroVal[0]), false // 不存在
}

mapaccessK 同时返回key 和 elem 原理同上

// returns both key and elem. Used by map iterator
func mapaccessK(t *maptype, h *hmap, key unsafe.Pointer) (unsafe.Pointer, unsafe.Pointer) {
	if h == nil || h.count == 0 {
		return nil, nil
	}
	hash := t.hasher(key, uintptr(h.hash0))
	m := bucketMask(h.B)
	b := (*bmap)(add(h.buckets, (hash&m)*uintptr(t.bucketsize)))
	if c := h.oldbuckets; c != nil {
		if !h.sameSizeGrow() {
			// There used to be half as many buckets; mask down one more power of two.
			m >>= 1
		}
		oldb := (*bmap)(add(c, (hash&m)*uintptr(t.bucketsize)))
		if !evacuated(oldb) {
			b = oldb
		}
	}
	top := tophash(hash)
bucketloop:
	for ; b != nil; b = b.overflow(t) {
		for i := uintptr(0); i < bucketCnt; i++ {
			if b.tophash[i] != top {
				if b.tophash[i] == emptyRest {
					break bucketloop
				}
				continue
			}
			k := add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize))
			if t.indirectkey() {
				k = *((*unsafe.Pointer)(k))
			}
			if t.key.equal(key, k) {
				e := add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize))
				if t.indirectelem() {
					e = *((*unsafe.Pointer)(e))
				}
				return k, e
			}
		}
	}
	return nil, nil
}

mapaccess1_fat, mapaccess2_fat 对 mapaccess1 的封装

func mapaccess1_fat(t *maptype, h *hmap, key, zero unsafe.Pointer) unsafe.Pointer {
	e := mapaccess1(t, h, key)
	if e == unsafe.Pointer(&zeroVal[0]) {
		return zero
	}
	return e
}

func mapaccess2_fat(t *maptype, h *hmap, key, zero unsafe.Pointer) (unsafe.Pointer, bool) {
	e := mapaccess1(t, h, key)
	if e == unsafe.Pointer(&zeroVal[0]) {
		return zero, false
	}
	return e, true
}

mapassign map[key] = value

// Like mapaccess, but allocates a slot for the key if it is not present in the map.
func mapassign(t *maptype, h *hmap, key unsafe.Pointer) unsafe.Pointer {
	if h == nil {  // 向nil的map中写值 触发panic
		panic(plainError("assignment to entry in nil map"))
	}
	if raceenabled {
		callerpc := getcallerpc()
		pc := abi.FuncPCABIInternal(mapassign)
		racewritepc(unsafe.Pointer(h), callerpc, pc)
		raceReadObjectPC(t.key, key, callerpc, pc)
	}
	if msanenabled {
		msanread(key, t.key.size)
	}
	if asanenabled {
		asanread(key, t.key.size)
	}
	if h.flags&hashWriting != 0 { // 多协程写入 panic
		fatal("concurrent map writes")
	}
	hash := t.hasher(key, uintptr(h.hash0)) // 计算哈希

	// Set hashWriting after calling t.hasher, since t.hasher may panic,
	// in which case we have not actually done a write.
	h.flags ^= hashWriting // 标记为写入状态

	if h.buckets == nil { // 如果没有桶则new一个1个单位的桶
		h.buckets = newobject(t.bucket) // newarray(t.bucket, 1)
	}

again:
	bucket := hash & bucketMask(h.B) // 计算桶的下标
	if h.growing() { // 桶正在扩容
		growWork(t, h, bucket) // 优先迁移当前桶
	}
	b := (*bmap)(add(h.buckets, bucket*uintptr(t.bucketsize))) // 计算桶地址
	top := tophash(hash)  // 获取tophash

	var inserti *uint8  // 指向将要插入tophash[i]的地址
	var insertk unsafe.Pointer // 将要插入的key的地址
	var elem unsafe.Pointer    // 将要插入的elem的地址
bucketloop:
	for {
		for i := uintptr(0); i < bucketCnt; i++ {
			if b.tophash[i] != top {
				if isEmpty(b.tophash[i]) && inserti == nil {  // tohash[i] 为空 并且 inserti 还没有指向，则找到一个合适的单元格，计算key 和 elem的地址
					inserti = &b.tophash[i]
					insertk = add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize))
					elem = add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize))
				}
				if b.tophash[i] == emptyRest { // 如果单元格其后均为空，包括溢出桶，则返回即可
					break bucketloop
				}
				continue  // 否则找下一个位置
			}
			k := add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize))
			if t.indirectkey() {
				k = *((*unsafe.Pointer)(k))
			}
			if !t.key.equal(key, k) {  // 已经存在的 k 与 key 相同
				continue
			}
			// already have a mapping for key. Update it.
			if t.needkeyupdate() { // 更新k
				typedmemmove(t.key, k, key)
			}
			elem = add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize)) // 计算elem的地址
			goto done // 跳转，准备返回
		}
		ovf := b.overflow(t) // 上边的桶没有位置了，去溢出桶里存储
		if ovf == nil {  // 如果溢出桶为空，则证明无可用溢出桶，需要添加溢出桶，或扩容
			break
		}
		b = ovf // 去溢出桶里继续找合适的位置插入或更新
	}

	// Did not find mapping for key. Allocate new cell & add entry.

	// If we hit the max load factor or we have too many overflow buckets,
	// and we're not already in the middle of growing, start growing.
	if !h.growing() && (overLoadFactor(h.count+1, h.B) || tooManyOverflowBuckets(h.noverflow, h.B)) { // 非扩容阶段，并且 负载因子被超过，或 有太多的溢出桶
		hashGrow(t, h)  // 扩容
		goto again // Growing the table invalidates everything, so try again  //从头来一遍
	}

	if inserti == nil { // 当前桶的所有溢出桶都满了，
		// The current bucket and all the overflow buckets connected to it are full, allocate a new one.
		newb := h.newoverflow(t, b) // 创建一个溢出桶 挂在 b的溢出桶后边
		inserti = &newb.tophash[0] // 重新计算i,k, e地址
		insertk = add(unsafe.Pointer(newb), dataOffset)
		elem = add(insertk, bucketCnt*uintptr(t.keysize))
	}

	// store new key/elem at insert position
	if t.indirectkey() { // 间接引用 ,开辟并挂载新空间
		kmem := newobject(t.key) 
		*(*unsafe.Pointer)(insertk) = kmem
		insertk = kmem
	}
	if t.indirectelem() { // 同上
		vmem := newobject(t.elem)
		*(*unsafe.Pointer)(elem) = vmem
	}
	typedmemmove(t.key, insertk, key)
	*inserti = top    // 赋tophash值
	h.count++  // 元素个数+1

done:
	if h.flags&hashWriting == 0 { //如果写状态被修改，则证明多个协程同时写，则panic
		fatal("concurrent map writes")
	}
	h.flags &^= hashWriting // 清除写状态标记
	if t.indirectelem() { // 解引用
		elem = *((*unsafe.Pointer)(elem))
	}
	return elem // 返回的elem 可以被赋value的值了，value是在该方法外部赋值的
}

mapdelete delete(Map,key)

func mapdelete(t *maptype, h *hmap, key unsafe.Pointer) {
	if raceenabled && h != nil {
		callerpc := getcallerpc()
		pc := abi.FuncPCABIInternal(mapdelete)
		racewritepc(unsafe.Pointer(h), callerpc, pc)
		raceReadObjectPC(t.key, key, callerpc, pc)
	}
	if msanenabled && h != nil {
		msanread(key, t.key.size)
	}
	if asanenabled && h != nil {
		asanread(key, t.key.size)
	}
	if h == nil || h.count == 0 {
		if t.hashMightPanic() {
			t.hasher(key, 0) // see issue 23734
		}
		return
	}
	if h.flags&hashWriting != 0 {  // 删除也置写状态标记，若已有其他协程操作map写入，则panic
		fatal("concurrent map writes")
	}

	hash := t.hasher(key, uintptr(h.hash0)) // 计算hash

	// Set hashWriting after calling t.hasher, since t.hasher may panic,
	// in which case we have not actually done a write (delete).
	h.flags ^= hashWriting   //置写状态标记

	bucket := hash & bucketMask(h.B)  //计算桶索引
	if h.growing() {  // 若正在扩容
		growWork(t, h, bucket) //优先迁移当前要操作的旧桶
	}
	b := (*bmap)(add(h.buckets, bucket*uintptr(t.bucketsize))) // 计算桶位置
	bOrig := b  // 原桶 与操作桶 做对比
	top := tophash(hash) // hash高八位
search:
	for ; b != nil; b = b.overflow(t) { // 外层遍历桶及溢出桶
		for i := uintptr(0); i < bucketCnt; i++ { // 内层遍历桶内的八组单元
			if b.tophash[i] != top {
				if b.tophash[i] == emptyRest {  // 桶内没有该key
					break search
				}
				continue
			}
			// top 相等的情况
			k := add(unsafe.Pointer(b), dataOffset+i*uintptr(t.keysize)) // 计算key的地址
			k2 := k
			if t.indirectkey() { // 若间接引用则解引用
				k2 = *((*unsafe.Pointer)(k2))
			}
			if !t.key.equal(key, k2) { // 传入key 和 已有k 不等
				continue
			}
			// Only clear key if there are pointers in it.
			if t.indirectkey() {  // 置空key的内存
				*(*unsafe.Pointer)(k) = nil
			} else if t.key.ptrdata != 0 {
				memclrHasPointers(k, t.key.size)
			}
			e := add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+i*uintptr(t.elemsize)) // 计算e的内存地址
			if t.indirectelem() { //置空e的内存
				*(*unsafe.Pointer)(e) = nil
			} else if t.elem.ptrdata != 0 {
				memclrHasPointers(e, t.elem.size)
			} else {
				memclrNoHeapPointers(e, t.elem.size)
			}
			b.tophash[i] = emptyOne //置空当前tophash单元格
			// If the bucket now ends in a bunch of emptyOne states,
			// change those to emptyRest states.
			// It would be nice to make this a separate function, but
			// for loops are not currently inlineable.
			if i == bucketCnt-1 { //最后一个单元格
				if b.overflow(t) != nil && b.overflow(t).tophash[0] != emptyRest {
					goto notLast //非最后一个桶(溢出桶)
				}
			} else {
				if b.tophash[i+1] != emptyRest { //下一个单元非emptyRest状态
					goto notLast
				}
			}
			for {   //为桶及溢出桶，合理的位置，置emptyRest
				b.tophash[i] = emptyRest // 当前位置置空
				if i == 0 { // 到达第1个单元
					if b == bOrig { // 当前桶和原桶一致， 置空完毕
						break // beginning of initial bucket, we're done.
					}
					// Find previous bucket, continue at its last entry.
					c := b
					for b = bOrig; b.overflow(t) != c; b = b.overflow(t) { // 寻找b的上一个桶
					}
					i = bucketCnt - 1 //从最后一个单元向上扫描
				} else {
					i-- //向上扫描
				}
				if b.tophash[i] != emptyOne { // 当前格子的状态非emptyOne 则返回
					break
				}
			}
		notLast:
			h.count-- //map中元素个数-1
			// Reset the hash seed to make it more difficult for attackers to
			// repeatedly trigger hash collisions. See issue 25237.
			if h.count == 0 { // map元素个数空了，重新计算hash种子
				h.hash0 = fastrand()
			}
			break search
		}
	}

	if h.flags&hashWriting == 0 { // 判定是否多协程操作
		fatal("concurrent map writes")
	}
	h.flags &^= hashWriting //清除写状态
}

mapclear 从一个map中删除所有key

// mapclear deletes all keys from a map.
func mapclear(t *maptype, h *hmap) {
	if raceenabled && h != nil {
		callerpc := getcallerpc()
		pc := abi.FuncPCABIInternal(mapclear)
		racewritepc(unsafe.Pointer(h), callerpc, pc)
	}

	if h == nil || h.count == 0 {
		return
	}

	if h.flags&hashWriting != 0 {
		fatal("concurrent map writes")
	}

	h.flags ^= hashWriting    // 写状态

	h.flags &^= sameSizeGrow //各种清空
	h.oldbuckets = nil
	h.nevacuate = 0
	h.noverflow = 0
	h.count = 0

	// Reset the hash seed to make it more difficult for attackers to
	// repeatedly trigger hash collisions. See issue 25237.
	h.hash0 = fastrand() //重置hash种子

	// Keep the mapextra allocation but clear any extra information.
	if h.extra != nil { // 溢出桶字段清空
		*h.extra = mapextra{}
	}

	// makeBucketArray clears the memory pointed to by h.buckets
	// and recovers any overflow buckets by generating them
	// as if h.buckets was newly alloced.
	_, nextOverflow := makeBucketArray(t, h.B, h.buckets) //重置nextOverflow
	if nextOverflow != nil {
		// If overflow buckets are created then h.extra
		// will have been allocated during initial bucket creation.
		h.extra.nextOverflow = nextOverflow
	}

	if h.flags&hashWriting == 0 {
		fatal("concurrent map writes")
	}
	h.flags &^= hashWriting //清空写状态
}

hashGrow 哈希扩容，触发桶的迁移

func hashGrow(t *maptype, h *hmap) {
	// If we've hit the load factor, get bigger.
	// Otherwise, there are too many overflow buckets,
	// so keep the same number of buckets and "grow" laterally.
	bigger := uint8(1)
	if !overLoadFactor(h.count+1, h.B) {  // 等量扩容
		bigger = 0
		h.flags |= sameSizeGrow
	}
	oldbuckets := h.buckets   
	newbuckets, nextOverflow := makeBucketArray(t, h.B+bigger, nil) //重新规划新桶和溢出桶

	flags := h.flags &^ (iterator | oldIterator) // 新flag 是 清空 迭代器和旧迭代器标记 的值
	if h.flags&iterator != 0 { // 正在被迭代
		flags |= oldIterator   // 加上旧迭代器的标记
	}
	// commit the grow (atomic wrt gc)
	h.B += bigger  // B += 1 or 0  
	h.flags = flags 
	h.oldbuckets = oldbuckets // buckets 迁移给 oldbuckets
	h.buckets = newbuckets    // 新开辟的空间给buckets
	h.nevacuate = 0           // 迁移进度置0，表示当前应迁移0号旧桶
	h.noverflow = 0           // 溢出桶个数置0

	if h.extra != nil && h.extra.overflow != nil { //转移旧的溢出桶指针数组
		// Promote current overflow buckets to the old generation.
		if h.extra.oldoverflow != nil {
			throw("oldoverflow is not nil")
		}
		h.extra.oldoverflow = h.extra.overflow
		h.extra.overflow = nil
	}
	if nextOverflow != nil { //重置nextOVerflow 可用溢出桶指针
		if h.extra == nil {
			h.extra = new(mapextra)
		}
		h.extra.nextOverflow = nextOverflow
	}

	// the actual copying of the hash table data is done incrementally
	// by growWork() and evacuate().
}

tooManyOverflowBuckets 溢出桶个数超过既定值，触发等量扩容 noverflow > 1 << 15

// tooManyOverflowBuckets reports whether noverflow buckets is too many for a map with 1<<B buckets.
// Note that most of these overflow buckets must be in sparse use;
// if use was dense, then we'd have already triggered regular map growth.
func tooManyOverflowBuckets(noverflow uint16, B uint8) bool {
	// If the threshold is too low, we do extraneous work.
	// If the threshold is too high, maps that grow and shrink can hold on to lots of unused memory.
	// "too many" means (approximately) as many overflow buckets as regular buckets.
	// See incrnoverflow for more details.
	if B > 15 {
		B = 15
	}
	// The compiler doesn't see here that B < 16; mask B to generate shorter shift code.
	return noverflow >= uint16(1)<<(B&15) // noverflow > 1 << 15
}

growing 是否正在扩容检查旧桶是否为空即可

// growing reports whether h is growing. The growth may be to the same size or bigger.
func (h *hmap) growing() bool {
	return h.oldbuckets != nil
}

sameSizeGrow 检查是否等量扩容

// sameSizeGrow reports whether the current growth is to a map of the same size.
func (h *hmap) sameSizeGrow() bool {
	return h.flags&sameSizeGrow != 0
}

noldbuckets 旧桶个数，非等量扩容则是新桶的一半，等量扩容则与新桶一致

// noldbuckets calculates the number of buckets prior to the current map growth.
func (h *hmap) noldbuckets() uintptr {
	oldB := h.B
	if !h.sameSizeGrow() {
		oldB--   // 指数-1 相当于 个数/2
	}
	return bucketShift(oldB) // 1 << oldB 
}

oldbucketmask 旧桶的掩码

// oldbucketmask provides a mask that can be applied to calculate n % noldbuckets().
func (h *hmap) oldbucketmask() uintptr {
	return h.noldbuckets() - 1
}

growWork

数据的迁移用到了copy on write的思想，即cow写时复制，在插入和删除时均使用了cow

func growWork(t *maptype, h *hmap, bucket uintptr) {
	// make sure we evacuate the oldbucket corresponding
	// to the bucket we're about to use
	evacuate(t, h, bucket&h.oldbucketmask()) //先迁移将用到的桶

	// evacuate one more oldbucket to make progress on growing
	if h.growing() { // 再迁移进度中的一个桶
		evacuate(t, h, h.nevacuate)
	}
}

bucketEvacuated 检查桶是否迁移完毕

func bucketEvacuated(t *maptype, h *hmap, bucket uintptr) bool {
	b := (*bmap)(add(h.oldbuckets, bucket*uintptr(t.bucketsize)))
	return evacuated(b)
}

evacuate 桶迁移过程

// evacDst is an evacuation destination.
type evacDst struct {          //迁移目的地结构体        
	b *bmap          // current destination bucket      //当前目标桶
	i int            // key/elem index into b           //正在迁移桶中的第i个单元
	k unsafe.Pointer // pointer to current key storage  // 当前的key指针
	e unsafe.Pointer // pointer to current elem storage // 当前的 elem指针
}

func evacuated(b *bmap) bool { // 检测是否迁移过
	h := b.tophash[0]
	return h > emptyOne && h < minTopHash
}

func evacuate(t *maptype, h *hmap, oldbucket uintptr) {
	b := (*bmap)(add(h.oldbuckets, oldbucket*uintptr(t.bucketsize))) //计算要迁移的旧桶的地址
	newbit := h.noldbuckets() //计算旧桶个数
	if !evacuated(b) { // 未迁移过
		// TODO: reuse overflow buckets instead of using new ones, if there
		// is no iterator using the old buckets.  (If !oldIterator.)

		// xy contains the x and y (low and high) evacuation destinations.
		var xy [2]evacDst  // 等量扩容时，只用到了x (low 目标地址)，非等量扩容时，用到了x,y(low,high) 2个目标地址，即一个旧桶中的值会被分配到两个新桶中，根据tophhash的oldB位的值，0 -> x  1 -> y 当遇见NaN的情况则取低位的 0 或 1 
		x := &xy[0] //x 目标地址是一定会被使用到的
		x.b = (*bmap)(add(h.buckets, oldbucket*uintptr(t.bucketsize))) //新桶索引与旧桶索引相同
		x.k = add(unsafe.Pointer(x.b), dataOffset) //   k地址
		x.e = add(x.k, bucketCnt*uintptr(t.keysize)) // e地址

		if !h.sameSizeGrow() { // 非等量扩容，会用到y 目标地址
			// Only calculate y pointers if we're growing bigger.
			// Otherwise GC can see bad pointers.
			y := &xy[1]
			y.b = (*bmap)(add(h.buckets, (oldbucket+newbit)*uintptr(t.bucketsize))) // 旧桶索引 + 旧桶个数的偏移
			y.k = add(unsafe.Pointer(y.b), dataOffset)
			y.e = add(y.k, bucketCnt*uintptr(t.keysize))
		}

		for ; b != nil; b = b.overflow(t) { //外层遍历桶及其溢出桶
			k := add(unsafe.Pointer(b), dataOffset))  // k偏移地址
			e := add(k, bucketCnt*uintptr(t.keysize)) // e偏移地址
			for i := 0; i < bucketCnt; i, k, e = i+1, add(k, uintptr(t.keysize)), add(e, uintptr(t.elemsize)) { // 内层遍历桶内的八组单元
				top := b.tophash[i]
				if isEmpty(top) {
					b.tophash[i] = evacuatedEmpty // top为空赋 evacuatedEmpty 表示 迁移过的空单元
					continue
				}
				if top < minTopHash {
					throw("bad map state")
				}
				k2 := k
				if t.indirectkey() {
					k2 = *((*unsafe.Pointer)(k2))
				}
				var useY uint8  // 初始值为 0
				if !h.sameSizeGrow() { // 非等量扩容 即2倍扩容
					// Compute hash to make our evacuation decision (whether we need
					// to send this key/elem to bucket x or bucket y).
					hash := t.hasher(k2, uintptr(h.hash0)) //重新计算hash 因为k2可能是解引用后的值
					if h.flags&iterator != 0 && !t.reflexivekey() && !t.key.equal(k2, k2) { // 处理浮点数 NaN问题
						// If key != key (NaNs), then the hash could be (and probably
						// will be) entirely different from the old hash. Moreover,
						// it isn't reproducible. Reproducibility is required in the
						// presence of iterators, as our evacuation decision must
						// match whatever decision the iterator made.
						// Fortunately, we have the freedom to send these keys either
						// way. Also, tophash is meaningless for these kinds of keys.
						// We let the low bit of tophash drive the evacuation decision.
						// We recompute a new random tophash for the next level so
						// these keys will get evenly distributed across all buckets
						// after multiple grows.
						useY = top & 1 // 取低位
						top = tophash(hash) //重新计算top
					} else {
						if hash&newbit != 0 { // 取旧hash的
							useY = 1
						}
					}
				}

				if evacuatedX+1 != evacuatedY || evacuatedX^1 != evacuatedY {
					throw("bad evacuatedN")
				}

				b.tophash[i] = evacuatedX + useY // evacuatedX + 1 == evacuatedY  //标记原桶迁移的目的地
				dst := &xy[useY]                 // evacuation destination //向迁移目标地址迁移

				if dst.i == bucketCnt { //当前桶可用单元用尽
					dst.b = h.newoverflow(t, dst.b) // 挂一个溢出桶
					dst.i = 0 //从溢出桶0位置开始写
					dst.k = add(unsafe.Pointer(dst.b), dataOffset) //取k的地址
					dst.e = add(dst.k, bucketCnt*uintptr(t.keysize)) // 取e的地址
				}
				dst.b.tophash[dst.i&(bucketCnt-1)] = top // mask dst.i as an optimization, to avoid a bounds check //赋topahsh 为top
				if t.indirectkey() { //间接引用解引用
					*(*unsafe.Pointer)(dst.k) = k2 // copy pointer  //复制指针
				} else {
					typedmemmove(t.key, dst.k, k) // copy elem      //复制内容
				}
				if t.indirectelem() { //同上
					*(*unsafe.Pointer)(dst.e) = *(*unsafe.Pointer)(e)
				} else {
					typedmemmove(t.elem, dst.e, e)
				}
				dst.i++  //迁移的单元索引+1
				// These updates might push these pointers past the end of the
				// key or elem arrays.  That's ok, as we have the overflow pointer
				// at the end of the bucket to protect against pointing past the
				// end of the bucket.
				dst.k = add(dst.k, uintptr(t.keysize)) //dst 中下一轮的k地址
				dst.e = add(dst.e, uintptr(t.elemsize))//dst 中下一轮的e地址
			}
		}
		// Unlink the overflow buckets & clear key/elem to help GC.
		if h.flags&oldIterator == 0 && t.bucket.ptrdata != 0 {   //迁移完成 对旧桶 手动GC
			b := add(h.oldbuckets, oldbucket*uintptr(t.bucketsize))
			// Preserve b.tophash because the evacuation
			// state is maintained there.
			ptr := add(b, dataOffset)
			n := uintptr(t.bucketsize) - dataOffset
			memclrHasPointers(ptr, n)
		}
	}

	if oldbucket == h.nevacuate {  //如果刚好，当前桶索引 等于 要迁移的桶
		advanceEvacuationMark(h, t, newbit) // 修改迁移进度
	}
}

advanceEvacuationMark 修改迁移进度，释放旧桶资源

func advanceEvacuationMark(h *hmap, t *maptype, newbit uintptr) {
	h.nevacuate++ //迁移进度+1
	// Experiments suggest that 1024 is overkill by at least an order of magnitude.
	// Put it in there as a safeguard anyway, to ensure O(1) behavior.
	stop := h.nevacuate + 1024
	if stop > newbit {
		stop = newbit
	}
	for h.nevacuate != stop && bucketEvacuated(t, h, h.nevacuate) {
		h.nevacuate++
	}
	if h.nevacuate == newbit { // newbit == # of oldbuckets //迁移完成了
		// Growing is all done. Free old main bucket array.
		h.oldbuckets = nil // 旧桶置空
		// Can discard old overflow buckets as well.
		// If they are still referenced by an iterator,
		// then the iterator holds a pointers to the slice.
		if h.extra != nil {
			h.extra.oldoverflow = nil //旧桶的溢出桶置空
		}
		h.flags &^= sameSizeGrow //清空等量扩容标记
	}
}

hiter 迭代器结构体

// A hash iteration structure.
// If you modify hiter, also change cmd/compile/internal/reflectdata/reflect.go
// and reflect/value.go to match the layout of this structure.
type hiter struct {
	key         unsafe.Pointer // Must be in first position.  Write nil to indicate iteration end (see cmd/compile/internal/walk/range.go). //当前key
	elem        unsafe.Pointer // Must be in second position (see cmd/compile/internal/walk/range.go). //当前elem
	t           *maptype //map类型
	h           *hmap //hmap指针，指向将遍历的hmap
	buckets     unsafe.Pointer // bucket ptr at hash_iter initialization time //指向hmap.buckets将要遍历的桶
	bptr        *bmap          // current bucket //当前的桶指针
	overflow    *[]*bmap       // keeps overflow buckets of hmap.buckets alive //溢出桶指针
	oldoverflow *[]*bmap       // keeps overflow buckets of hmap.oldbuckets alive //旧溢出桶指针
	startBucket uintptr        // bucket iteration started at //随机桶起始位置
	offset      uint8          // intra-bucket offset to start from during iteration (should be big enough to hold bucketCnt-1) //随机桶内单元偏移位置
	wrapped     bool           // already wrapped around from end of bucket array to beginning //标识bucket遍历到末尾由回到了开头
	B           uint8   // hmap中的B
	i           uint8   // 这个i有点任性，当前桶内单元的索引
	bucket      uintptr // 起始的随机桶号，当前遍历的桶
	checkBucket uintptr // 需检查的bucket，有扩容操作时使用
}

mapiterinit map迭代器初始化函数，保证遍历的随机性，随机的桶及桶中单元起始位置

// mapiterinit initializes the hiter struct used for ranging over maps.
// The hiter struct pointed to by 'it' is allocated on the stack
// by the compilers order pass or on the heap by reflect_mapiterinit.
// Both need to have zeroed hiter since the struct contains pointers.
func mapiterinit(t *maptype, h *hmap, it *hiter) {
	if raceenabled && h != nil {
		callerpc := getcallerpc()
		racereadpc(unsafe.Pointer(h), callerpc, abi.FuncPCABIInternal(mapiterinit))
	}

	it.t = t
	if h == nil || h.count == 0 {
		return
	}

	if unsafe.Sizeof(hiter{})/goarch.PtrSize != 12 {
		throw("hash_iter size incorrect") // see cmd/compile/internal/reflectdata/reflect.go
	}
	it.h = h

	// grab snapshot of bucket state
	it.B = h.B
	it.buckets = h.buckets
	if t.bucket.ptrdata == 0 {
		// Allocate the current slice and remember pointers to both current and old.
		// This preserves all relevant overflow buckets alive even if
		// the table grows and/or overflow buckets are added to the table
		// while we are iterating.
		h.createOverflow()
		it.overflow = h.extra.overflow
		it.oldoverflow = h.extra.oldoverflow
	}

	// decide where to start
	var r uintptr
	if h.B > 31-bucketCntBits {
		r = uintptr(fastrand64())
	} else {
		r = uintptr(fastrand())
	}
	it.startBucket = r & bucketMask(h.B)  //随机桶号，向后遍历后从前向后遍历，做到闭环
	it.offset = uint8(r >> h.B & (bucketCnt - 1)) // 随机桶内单元，向后遍历后从前向后遍历，做到闭环

	// iterator state
	it.bucket = it.startBucket

	// Remember we have an iterator.
	// Can run concurrently with another mapiterinit().
	if old := h.flags; old&(iterator|oldIterator) != iterator|oldIterator { //并发读取
		atomic.Or8(&h.flags, iterator|oldIterator)
	}

	mapiternext(it)
}

mapiternext 咱就把迭代器不断的指向下一个位置

func mapiternext(it *hiter) {
	h := it.h
	if raceenabled {
		callerpc := getcallerpc()
		racereadpc(unsafe.Pointer(h), callerpc, abi.FuncPCABIInternal(mapiternext))
	}
	if h.flags&hashWriting != 0 {
		fatal("concurrent map iteration and map write")
	}
	t := it.t
	bucket := it.bucket
	b := it.bptr
	i := it.i
	checkBucket := it.checkBucket

next:
	if b == nil {
		if bucket == it.startBucket && it.wrapped { // 遍历完一圈了
			// end of iteration
			it.key = nil
			it.elem = nil
			return
		}
		if h.growing() && it.B == h.B { //若正在扩容
			// Iterator was started in the middle of a grow, and the grow isn't done yet.
			// If the bucket we're looking at hasn't been filled in yet (i.e. the old
			// bucket hasn't been evacuated) then we need to iterate through the old
			// bucket and only return the ones that will be migrated to this bucket.
			oldbucket := bucket & it.h.oldbucketmask() // 旧桶的既定的 偏移（随机偏移）值
			b = (*bmap)(add(h.oldbuckets, oldbucket*uintptr(t.bucketsize))) //相应旧桶的地址
			if !evacuated(b) { //若未迁移完成
				checkBucket = bucket // 需要检查的桶 为 bucket
			} else { //迁移完成了，就去新桶相应的位置迭代
				b = (*bmap)(add(it.buckets, bucket*uintptr(t.bucketsize)))
				checkBucket = noCheck // 无需checkBucket
			}
		} else { //迁移完成了，就去新桶相应的位置迭代
			b = (*bmap)(add(it.buckets, bucket*uintptr(t.bucketsize)))
			checkBucket = noCheck // 无需checkBucket
		}
		bucket++ //向后迭代，下一个桶
		if bucket == bucketShift(it.B) { //迭代到最后一个桶
			bucket = 0 //下一次访问第个桶，形成闭环，苦海无涯，回头是岸
			it.wrapped = true // 标记置true 代表已经从桶数组的结尾转到开头
		}
		i = 0  // i 置零
	}
	for ; i < bucketCnt; i++ {
		offi := (i + it.offset) & (bucketCnt - 1) //(i + it.offset) & (bucketCnt - 1) 在 0 ~ 7 之间形成了环，所有单元都会遍历到
		if isEmpty(b.tophash[offi]) || b.tophash[offi] == evacuatedEmpty { //越过绵绵的高山。呸，越过空单元
			// TODO: emptyRest is hard to use here, as we start iterating
			// in the middle of a bucket. It's feasible, just tricky.
			continue
		}
		k := add(unsafe.Pointer(b), dataOffset+uintptr(offi)*uintptr(t.keysize)) // 取k地址
		if t.indirectkey() {
			k = *((*unsafe.Pointer)(k))
		}
		e := add(unsafe.Pointer(b), dataOffset+bucketCnt*uintptr(t.keysize)+uintptr(offi)*uintptr(t.elemsize)) //取e地址
		if checkBucket != noCheck && !h.sameSizeGrow() { //倦了，直接机器翻译吧。。。特殊情况：迭代器在增长到更大的大小时启动//而且生长还没有完成。我们正在研究一个桶//旧桶还没有被疏散。或者至少不是//当我们启动水桶时撤离。所以我们在迭代//通过旧桶，跳过任何将要删除的键//到另一个新存储桶（每个旧存储桶扩展为两个//生长期间的桶
			// Special case: iterator was started during a grow to a larger size
			// and the grow is not done yet. We're working on a bucket whose
			// oldbucket has not been evacuated yet. Or at least, it wasn't
			// evacuated when we started the bucket. So we're iterating
			// through the oldbucket, skipping any keys that will go
			// to the other new bucket (each oldbucket expands to two
			// buckets during a grow).
			if t.reflexivekey() || t.key.equal(k, k) { 如果旧存储桶中的项目未指定用于//迭代中的当前新桶，跳过它。
				// If the item in the oldbucket is not destined for
				// the current new bucket in the iteration, skip it.
				hash := t.hasher(k, uintptr(h.hash0))
				if hash&bucketMask(it.B) != checkBucket {
					continue
				}
			} else {
				// Hash isn't repeatable if k != k (NaNs).  We need a
				// repeatable and randomish choice of which direction
				// to send NaNs during evacuation. We'll use the low
				// bit of tophash to decide which way NaNs go.
				// NOTE: this case is why we need two evacuate tophash
				// values, evacuatedX and evacuatedY, that differ in
				// their low bit.
				if checkBucket>>(it.B-1) != uintptr(b.tophash[offi]&1) { //处理NaN问题
					continue
				}
			}
		}
		if (b.tophash[offi] != evacuatedX && b.tophash[offi] != evacuatedY) ||
			!(t.reflexivekey() || t.key.equal(k, k)) { //赋值it.key 和 it.elem 
			// This is the golden data, we can return it.
			// OR
			// key!=key, so the entry can't be deleted or updated, so we can just return it.
			// That's lucky for us because when key!=key we can't look it up successfully.
			it.key = k
			if t.indirectelem() {
				e = *((*unsafe.Pointer)(e))
			}
			it.elem = e
		} else {
			// The hash table has grown since the iterator was started.
			// The golden data for this key is now somewhere else.
			// Check the current hash table for the data.
			// This code handles the case where the key
			// has been deleted, updated, or deleted and reinserted.
			// NOTE: we need to regrab the key as it has potentially been
			// updated to an equal() but not identical key (e.g. +0.0 vs -0.0).
			rk, re := mapaccessK(t, h, k) //通过key 取的key 和 elem
			if rk == nil {
				continue // key has been deleted
			}
			it.key = rk
			it.elem = re
		}
		it.bucket = bucket //下一个迭代的bucket
		if it.bptr != b { // avoid unnecessary write barrier; see issue 14921
			it.bptr = b
		}
		it.i = i + 1 //下一个单元
		it.checkBucket = checkBucket //状态转移
		return
	}
	b = b.overflow(t) //获取溢出桶
	i = 0 // i 置零
	goto next // 跳转，代替了递归
}

reflect 相关的map操作，就是对以上操作的封装，不做介绍

// Reflect stubs. Called from ../reflect/asm_*.s

//go:linkname reflect_makemap reflect.makemap
func reflect_makemap(t *maptype, cap int) *hmap {
	// Check invariants and reflects math.
	if t.key.equal == nil {
		throw("runtime.reflect_makemap: unsupported map key type")
	}
	if t.key.size > maxKeySize && (!t.indirectkey() || t.keysize != uint8(goarch.PtrSize)) ||
		t.key.size <= maxKeySize && (t.indirectkey() || t.keysize != uint8(t.key.size)) {
		throw("key size wrong")
	}
	if t.elem.size > maxElemSize && (!t.indirectelem() || t.elemsize != uint8(goarch.PtrSize)) ||
		t.elem.size <= maxElemSize && (t.indirectelem() || t.elemsize != uint8(t.elem.size)) {
		throw("elem size wrong")
	}
	if t.key.align > bucketCnt {
		throw("key align too big")
	}
	if t.elem.align > bucketCnt {
		throw("elem align too big")
	}
	if t.key.size%uintptr(t.key.align) != 0 {
		throw("key size not a multiple of key align")
	}
	if t.elem.size%uintptr(t.elem.align) != 0 {
		throw("elem size not a multiple of elem align")
	}
	if bucketCnt < 8 {
		throw("bucketsize too small for proper alignment")
	}
	if dataOffset%uintptr(t.key.align) != 0 {
		throw("need padding in bucket (key)")
	}
	if dataOffset%uintptr(t.elem.align) != 0 {
		throw("need padding in bucket (elem)")
	}

	return makemap(t, cap, nil)
}

//go:linkname reflect_mapaccess reflect.mapaccess
func reflect_mapaccess(t *maptype, h *hmap, key unsafe.Pointer) unsafe.Pointer {
	elem, ok := mapaccess2(t, h, key)
	if !ok {
		// reflect wants nil for a missing element
		elem = nil
	}
	return elem
}

//go:linkname reflect_mapaccess_faststr reflect.mapaccess_faststr
func reflect_mapaccess_faststr(t *maptype, h *hmap, key string) unsafe.Pointer {
	elem, ok := mapaccess2_faststr(t, h, key)
	if !ok {
		// reflect wants nil for a missing element
		elem = nil
	}
	return elem
}

//go:linkname reflect_mapassign reflect.mapassign
func reflect_mapassign(t *maptype, h *hmap, key unsafe.Pointer, elem unsafe.Pointer) {
	p := mapassign(t, h, key)
	typedmemmove(t.elem, p, elem)
}

//go:linkname reflect_mapassign_faststr reflect.mapassign_faststr
func reflect_mapassign_faststr(t *maptype, h *hmap, key string, elem unsafe.Pointer) {
	p := mapassign_faststr(t, h, key)
	typedmemmove(t.elem, p, elem)
}

//go:linkname reflect_mapdelete reflect.mapdelete
func reflect_mapdelete(t *maptype, h *hmap, key unsafe.Pointer) {
	mapdelete(t, h, key)
}

//go:linkname reflect_mapdelete_faststr reflect.mapdelete_faststr
func reflect_mapdelete_faststr(t *maptype, h *hmap, key string) {
	mapdelete_faststr(t, h, key)
}

//go:linkname reflect_mapiterinit reflect.mapiterinit
func reflect_mapiterinit(t *maptype, h *hmap, it *hiter) {
	mapiterinit(t, h, it)
}

//go:linkname reflect_mapiternext reflect.mapiternext
func reflect_mapiternext(it *hiter) {
	mapiternext(it)
}

//go:linkname reflect_mapiterkey reflect.mapiterkey
func reflect_mapiterkey(it *hiter) unsafe.Pointer {
	return it.key
}

//go:linkname reflect_mapiterelem reflect.mapiterelem
func reflect_mapiterelem(it *hiter) unsafe.Pointer {
	return it.elem
}

//go:linkname reflect_maplen reflect.maplen
func reflect_maplen(h *hmap) int {
	if h == nil {
		return 0
	}
	if raceenabled {
		callerpc := getcallerpc()
		racereadpc(unsafe.Pointer(h), callerpc, abi.FuncPCABIInternal(reflect_maplen))
	}
	return h.count
}

//go:linkname reflectlite_maplen internal/reflectlite.maplen
func reflectlite_maplen(h *hmap) int {
	if h == nil {
		return 0
	}
	if raceenabled {
		callerpc := getcallerpc()
		racereadpc(unsafe.Pointer(h), callerpc, abi.FuncPCABIInternal(reflect_maplen))
	}
	return h.count
}