go1.10.3中添加键/值与mapassign方法之间的关系

Question

I am reading the source code of map in go1.10.3.It seemed there exist corresponding method about operation such as:

 makemap(t *maptype, hint int, h *hmap) *hmap ==> m = make(map[xx]yy)
 mapaccess1(t *maptype, h *hmap, key unsafe.Pointer)==> m['key']

but I cant find the correspond method for the operation which add key/value as below:

 m['xx']='yy'

there exist a method called mapassign which has some similarity with this operation.

 mapassign(t *maptype, h *hmap, key unsafe.Pointer) unsafe.Pointer

this will add a new key to the map, but as we can see, the input arguments has no value. And another question is when it has already this key, it maybe update this key.

        if !alg.equal(key, k) {
            continue
        }
        // already have a mapping for key. Update it.
        if t.needkeyupdate {//why??
            typedmemmove(t.key, k, key)
        }

since the two key is equal, why should update it?

summary:
1. the relation between add key/value operation and method mapassign?
2. why it maybe need to update the key since the insert key and the key which has already exist is equal in mapassign method?

Answer 1

1. In the operation m[k] = v, the caller copies the value v to the address returned by mapassign.

2. The comments in the function needkeyupdate explain why some types need key updates: floating point & complex -0 and 0 are equal, but different values; string might have smaller backing store.