Calling this a hash is very misleading:
1. This function gives an object identifier (ID), which uniquely identifies the object for its whole lifetime. This is similar to the address of an object in C or the id() function in Python. I'm sure other languages have similar constructs.
2. This is not a hash and has nothing to do with it. A hash takes data and algorithmically reduces that data to some kind of scalar value. The only guarantee is that two equal inputs provide the same output, but not that two different inputs provide different outputs (hint: hash collisions). spl_object_hash() guarantees different outputs for non-identical objects though.
3. As someone mentioned already, this does not look at the content of the object. If you consider the difference between equality and identity, it only allows determining identity. If you serialize and unserialize an object, it will not be identical to its former self, but it will be equal, just to give an example. If you want a key to use in a response cache, using this function on the request is not only useless, because equal requests have different IDs, but possibly even harmful, because when a request object is garbage collected, its ID can be reused.