In the Link State method, every node will possess the same topology information on the network and independently computes the routing table.
Each router will then know which subnet contain members of multicast groups and can easily construct per source multicast routing tables.
(Routing table can be found using the Dijkstra's Algorithm)
The nodes B and F will include this fact in their link state broadcast messages:
The distance matrix of the above figure is:
| A | B | C | D | E | F ----+---+---+---+---+---+--- A | - | * | 6 | 2 | 4 | * ----+---+---+---+---+---+--- B | * | - | 1 | 4 | 1 | * ----+---+---+---+---+---+--- C | 6 | 1 | - | * | * | 7 ----+---+---+---+---+---+--- D | 2 | 4 | * | - | * | 4 ----+---+---+---+---+---+--- E | 4 | 1 | * | * | - | 1 ----+---+---+---+---+---+--- F | * | * | 7 | 4 | 1 | - ----+---+---+---+---+---+--- |