This question has the answer.
For the first question when an SPV client starts up and sets a bloom filter it will not receive any special messages from connected bitcoin nodes. The bloom filter will result in transaction broadcasts to the SPV being filtered.
In answer to the second question an SPV client must still request ALL block headers and send a getdata request for every block in order to determine whether it contains any relevant transactions. The purpose of the bloom filter is to restrict the number of transactions returned by the merkleblock response to the getdata request.
An SPV client that's relying on bitcoin peer nodes for transaction information will have a slow initial start up as it needs to request all the block headers.