面试题:
- 为什么map不能并发读写?
- map 并发读写会panic吗?
- 为什么sync.Map 没有len方法?
- map + lock 和 sync.Map 差别在哪里?
map
结构图
sync .Map
源码分析
package sync
import (
"sync/atomic"
"unsafe"
)
/*
大概思想:
readMap 相当于 是 dirtyMap的缓存
每次增加新的key都是加锁然后存到dirtyMap,
每次读取数据的时候每次都是去无锁读readMap,如果readMap没有读到,miss+1,
miss达到一个数字后,就把dirtyMap设置为readMap,
接下来存新的数据的时候,又会遍历readMap把当前有效的数据,存到dirtyMap,然后把新的数据存到dirtyMap,如此往复
*/
type Map struct {
// 锁(对dirtyMap操作的时候会使用)
mu Mutex
// read是个原子变量,在read里面的值是能够并发修改的 ,其实read相当于是dirtyMap的缓存
read atomic.Value // readOnly
// 加锁进行操作,和read构成冗余,misses达到len(dirty)后升级为read
dirty map[interface{}]*entry
// 查询没有命中read的次数
misses int
}
// readOnly is an immutable struct stored atomically in the Map.read field.
type readOnly struct {
m map[interface{}]*entry
amended bool // 是否有新数据写入dirty
}
// 来代替被删除的占位符
var expunged = unsafe.Pointer(new(interface{}))
// entry: 用于保存value的interface指针,通过atomic进行原子操作
type entry struct {
// p points to the interface{} value stored for the entry.
//
// 这里注释感觉有问题,如果p == nil 的话 不能说明m.dirty == nil
// If p == nil, the entry has been deleted and m.dirty == nil.
//
// If p == expunged, the entry has been deleted, m.dirty != nil, and the entry
// is missing from m.dirty.
//
// Otherwise, the entry is valid and recorded in m.read.m[key] and, if m.dirty
// != nil, in m.dirty[key].
//
// An entry can be deleted by atomic replacement with nil: when m.dirty is
// next created, it will atomically replace nil with expunged and leave
// m.dirty[key] unset.
//
// An entry's associated value can be updated by atomic replacement, provided
// p != expunged. If p == expunged, an entry's associated value can be updated
// only after first setting m.dirty[key] = e so that lookups using the dirty
// map find the entry.
// 这就是interface{}的指针,能够被原子操作
// dirtyMap 和 readMap 的值都指向一个指针,这样修改了什么,另一个也会跟着变
p unsafe.Pointer // *interface{}
}
func newEntry(i interface{}) *entry {
return &entry{p: unsafe.Pointer(&i)}
}
// 从map里面获得数据
func (m *Map) Load(key interface{}) (value interface{}, ok bool) {
// 先把read 转成map
read, _ := m.read.Load().(readOnly)
// 先判断 read 该key是否存在
e, ok := read.m[key]
// 如果不存在 且 amended 为true(表示dirtyMap里面有readMap不存在的值)就加锁,去dirtyMap看看值存不存在
if !ok && read.amended {
// 加锁
m.mu.Lock()
// 加锁成功后,还要判断一下 read 数据 ,因为可能在加锁过程中,dirty 已经升级成read了
read, _ = m.read.Load().(readOnly)
e, ok = read.m[key]
if !ok && read.amended {
// 尝试从dirtyMap里面获得数据
e, ok = m.dirty[key]
// 增加未命中次数,达到一个数字就把 dirty 升级 为read
m.missLocked()
}
m.mu.Unlock()
}
if !ok {
return nil, false
}
// 把指针转化为interface的值返回
return e.load()
}
func (e *entry) load() (value interface{}, ok bool) {
p := atomic.LoadPointer(&e.p)
if p == nil || p == expunged {
return nil, false
}
return *(*interface{})(p), true
}
// Store sets the value for a key.
func (m *Map) Store(key, value interface{}) {
// 把read转化为map
read, _ := m.read.Load().(readOnly)
// 如果read里面已经存在该key 直接尝试设置值 , 这步并不需要加锁 ,这就很nice
if e, ok := read.m[key]; ok && e.tryStore(&value) {
return
}
// 如果readMap 并没有找到该key,或者说该readMap的值不存在或者被删除的占位符标记了,就需要加锁操作 dirtyMap了
m.mu.Lock()
read, _ = m.read.Load().(readOnly)
// double checking,防止在加锁的时候 dirty Map 升级为readMap
if e, ok := read.m[key]; ok {
// key在read中被标记为expunge删除的话,相当于dirtyMap没有这个key的指针 (而且 dirty 肯定不为空只有dirtyMap被同步了才会有expunged状态,
// 需要插入dirty,并且修改该指针的值
if e.unexpungeLocked() {
m.dirty[key] = e
}
// 使用原子操作存储值
e.storeLocked(&value)
} else if e, ok := m.dirty[key]; ok { // 看看key值是否已经在dirty里面了
e.storeLocked(&value)
} else {
// amended 若为false,则表示dirty未被初始化过
if !read.amended {
// 初始化dirty,将read中未被删除的有效的数据全都复制到dirty中,read中指向nil的数据会被标记为expunged(并且不被同步到dirtyMap)
m.dirtyLocked()
// 将amended改为true
m.read.Store(readOnly{m: read.m, amended: true})
}
// 将值存入dirty
// 只有dirtyMap 存入了readMap不存在的值,amended才会为true
m.dirty[key] = newEntry(value)
}
m.mu.Unlock()
}
// tryStore stores a value if the entry has not been expunged.
//
// If the entry is expunged, tryStore returns false and leaves the entry
// unchanged.
// 因为存数据的时候,必须确保 dirtyMap 也要存入, 如果 p 为 expunged ,说明dirtMap 没有该key的指针,所以不能直接存,所以返回false
func (e *entry) tryStore(i *interface{}) bool {
for {
p := atomic.LoadPointer(&e.p)
if p == expunged {
return false
}
if atomic.CompareAndSwapPointer(&e.p, p, unsafe.Pointer(i)) {
return true
}
}
}
// unexpungeLocked ensures that the entry is not marked as expunged.
//
// If the entry was previously expunged, it must be added to the dirty map
// before m.mu is unlocked.
func (e *entry) unexpungeLocked() (wasExpunged bool) {
return atomic.CompareAndSwapPointer(&e.p, expunged, nil)
}
// storeLocked unconditionally stores a value to the entry.
//
// The entry must be known not to be expunged.
func (e *entry) storeLocked(i *interface{}) {
atomic.StorePointer(&e.p, unsafe.Pointer(i))
}
// LoadOrStore returns the existing value for the key if present.
// Otherwise, it stores and returns the given value.
// The loaded result is true if the value was loaded, false if stored.
func (m *Map) LoadOrStore(key, value interface{}) (actual interface{}, loaded bool) {
// Avoid locking if it's a clean hit.
read, _ := m.read.Load().(readOnly)
if e, ok := read.m[key]; ok {
actual, loaded, ok := e.tryLoadOrStore(value)
if ok {
return actual, loaded
}
}
m.mu.Lock()
read, _ = m.read.Load().(readOnly)
if e, ok := read.m[key]; ok {
if e.unexpungeLocked() {
m.dirty[key] = e
}
actual, loaded, _ = e.tryLoadOrStore(value)
} else if e, ok := m.dirty[key]; ok {
actual, loaded, _ = e.tryLoadOrStore(value)
m.missLocked()
} else {
if !read.amended {
// We're adding the first new key to the dirty map.
// Make sure it is allocated and mark the read-only map as incomplete.
m.dirtyLocked()
m.read.Store(readOnly{m: read.m, amended: true})
}
m.dirty[key] = newEntry(value)
actual, loaded = value, false
}
m.mu.Unlock()
return actual, loaded
}
// tryLoadOrStore atomically loads or stores a value if the entry is not
// expunged.
//
// If the entry is expunged, tryLoadOrStore leaves the entry unchanged and
// returns with ok==false.
func (e *entry) tryLoadOrStore(i interface{}) (actual interface{}, loaded, ok bool) {
p := atomic.LoadPointer(&e.p)
if p == expunged {
return nil, false, false
}
if p != nil {
return *(*interface{})(p), true, true
}
// Copy the interface after the first load to make this method more amenable
// to escape analysis: if we hit the "load" path or the entry is expunged, we
// shouldn't bother heap-allocating.
ic := i
for {
if atomic.CompareAndSwapPointer(&e.p, nil, unsafe.Pointer(&ic)) {
return i, false, true
}
p = atomic.LoadPointer(&e.p)
if p == expunged {
return nil, false, false
}
if p != nil {
return *(*interface{})(p), true, true
}
}
}
// Delete deletes the value for a key.
func (m *Map) Delete(key interface{}) {
// 把read转成结构体
read, _ := m.read.Load().(readOnly)
e, ok := read.m[key]
//不在read中,且dirty中有新数据
if !ok && read.amended {
m.mu.Lock()
// 第二次尝试读,防止加锁中发生了变化, 比如dirty map 升级为 read map了
read, _ = m.read.Load().(readOnly)
e, ok = read.m[key]
if !ok && read.amended {
// 如果read里面还是没有,而且dirty还有别的数据,就对dirty进行删除一下(不管dirtyMap里面有没有这个值,反正删就完了)
delete(m.dirty, key)
}
m.mu.Unlock()
}
if ok {
// read中存在key,将这个key标记为删除状态,但并不删除数据
e.delete()
}
}
func (e *entry) delete() (hadValue bool) {
for {
p := atomic.LoadPointer(&e.p)
if p == nil || p == expunged {
return false
}
if atomic.CompareAndSwapPointer(&e.p, p, nil) {
return true
}
}
}
// Range calls f sequentially for each key and value present in the map.
// If f returns false, range stops the iteration.
//
// Range does not necessarily correspond to any consistent snapshot of the Map's
// contents: no key will be visited more than once, but if the value for any key
// is stored or deleted concurrently, Range may reflect any mapping for that key
// from any point during the Range call.
//
// Range may be O(N) with the number of elements in the map even if f returns
// false after a constant number of calls.
func (m *Map) Range(f func(key, value interface{}) bool) {
// We need to be able to iterate over all of the keys that were already
// present at the start of the call to Range.
// If read.amended is false, then read.m satisfies that property without
// requiring us to hold m.mu for a long time.
read, _ := m.read.Load().(readOnly)
if read.amended {
// m.dirty contains keys not in read.m. Fortunately, Range is already O(N)
// (assuming the caller does not break out early), so a call to Range
// amortizes an entire copy of the map: we can promote the dirty copy
// immediately!
m.mu.Lock()
read, _ = m.read.Load().(readOnly)
if read.amended {
read = readOnly{m: m.dirty}
m.read.Store(read)
m.dirty = nil
m.misses = 0
}
m.mu.Unlock()
}
for k, e := range read.m {
v, ok := e.load()
if !ok {
continue
}
if !f(k, v) {
break
}
}
}
// 如果没有命中一定次数就把 dirtyMap 升级为readMap 然后把原来的 dirtyMap 设置为空
func (m *Map) missLocked() {
m.misses++
if m.misses < len(m.dirty) {
return
}
// 这里可以看到dirty刚升级到read,amended是false的 ,amended 只有在 dirtyMap 包含readMap不存在的key的时候才会为true
m.read.Store(readOnly{m: m.dirty})
m.dirty = nil
m.misses = 0
}
// 把readMap的有效的数据同步到dirtyMap中
func (m *Map) dirtyLocked() {
if m.dirty != nil {
return
}
read, _ := m.read.Load().(readOnly)
m.dirty = make(map[interface{}]*entry, len(read.m))
// 遍历readMap的所有的值,然后把readMap中有效值赋值到dirtyMap。readMap值为nil的对象改成expunged,
// 这相当于是一个标记,如果readMap中有值为expunged,就说明,当前dirtyMap 已经被初始化了,并且dirtyMap不存在这个key
for k, e := range read.m {
if !e.tryExpungeLocked() {
m.dirty[k] = e
}
}
}
// 把nil的改为expunged
func (e *entry) tryExpungeLocked() (isExpunged bool) {
p := atomic.LoadPointer(&e.p)
for p == nil {
if atomic.CompareAndSwapPointer(&e.p, nil, expunged) {
return true
}
p = atomic.LoadPointer(&e.p)
}
return p == expunged
}
存储过程图解
结论:
map 之所以不能并发存储,是因为在并发存新的key的时候,可能会hash到同一个槽上,导致 原来 key1 : val1 , 和 key2 : val2 变成 key2:val1 的这种非预期场景,而且在 map 增在bucket的时候可能造成bucket链表出错。
那么如果在 不增加新key的情况下,map 并发读写是否是安全的呢? 答案是安全的,所以 sync.Map 就是利用了这个特性,减少了 map 加锁的粒度。更大的提高了map的性能。