When the parser parses an identifier, it generates a RdrName. A RdrName is pretty much just a string, or a pair of strings, for a qualified name, such as M.x. Here's the data type declaration, from compiler/GHC/Types/Name/Reader.hs:
data RdrName = Unqual OccName -- Used for ordinary, unqualified occurrences | Qual ModuleName OccName -- A qualified name written by the user in -- *source* code. The module isn't necessarily -- the module where the thing is defined; -- just the one from which it is imported | Orig Module OccName -- An original name; the module is the *defining* module. -- This is used when GHC generates code that will be fed -- into the renamer (e.g. from deriving clauses), but where -- we want to say "Use Prelude.map dammit". | Exact Name -- We know exactly the Name. This is used -- (a) when the parser parses built-in syntax like "" -- and "(,)", but wants a RdrName from it -- (b) by Template Haskell, when TH has generated a unique name
User-written code never gets translated into the last two alternatives. They are used only internally by the compiler.
For example, code generated by deriving might use an Orig to refer to Prelude.map, ignoring whatever map might happen to be in scope (dammit).
You'll notice that a QualRdrName contains a ModuleName; which module is referred to depends on the import declarations in that module. In contrast, a OrigRdrName refers to a unique Module.
The OccName type
An OccName is more-or-less just a string, like "foo" or "Tree", giving the (unqualified) name of an entity.
Well, not quite just a string, because in Haskell a name like "C" could mean a type constructor or data constructor, depending on context. So GHC defines a type OccName that is a pair of a FastString and a NameSpace indicating which name space the name is drawn from. The data type is defined (abstractly) in compiler/GHC/Types/Name/Occurrence.hs:
data NameSpace = VarName -- Variables, including "source" data constructors | DataName -- "Real" data constructors | TvName -- Type variables | TcClsName -- Type constructors and classes; Haskell has them -- in the same name space for now.
Attaching the names to their name spaces makes it very convenient to build mappings from names to things; where such a mapping might contain two strings that are identical, they can be distinguished by the name space, so when mapping OccNames, a single map suffices.