The various parts of pg_buffercache include a library written in C and some SQL that handles viewing the information. While the buffer cache itself is shared among every database in the cluster, pg_buffercache can only show you really useful information related to the database you're currently connected to, and its SQL component needs to be installed in each database you want to monitor (but not the template ones) as well.
In order to install this utility or to use it, you will need to be connected to the database as its superuser. Here's a sample that installs pg_buffercache into a new database, presuming the RedHat Linux standard directory installation tree; substitute your own PostgreSQL installation ...