\section{Architecture of MLRISC}

\subsection{Core Components}

  The core components of MLRISC allow the client to quickly construct 
an backend for various architectures.  These components include:
\begin{itemize}
  \item The \href{mltree.html}{MLTREE} language, 
       which is a RTL-like intermediate language
       that is used by the client
       to communicate to the MLRISC system.  A client is
       responsible for writing the module that generates MLTREE from
       the source program representation.
  \item \href{instrsel.html}{Instruction selection modules}, 
      which generates target machine 
       instructions from MLTREE.
  \item The \href{ra.html}{Register Allocator},
       which performs register allocation.
  \item \href{asm.html}{Assemblers}, which emits assembly code.
\end{itemize}

For systems that require direct machine code generation, the following
modules are included:
\begin{itemize}
  \item \href{span-dep.html}{Span dependency resolution} 
       modules, which compute addresses    
       from symbolic addresses,
       fill delay slots, and expand instructions that are 
       \newdef{span dependent}
  \item \href{mc.html}{Machine code emitters}, 
        which emit executable machine code into a binary stream.
\end{itemize}

\subsection{Optimization Modules}

In addition, MLRISC has been enhanced to support various types of
machine level optimizations.  These include:

\begin{itemize}
  \item Core optimizations, which includes
       various types of control flow transformation, 
       and architectural specific peephole optimizations. 
  \item SSA based scalar optimizations
  \item ILP optimizations for superscalars
  \item ILP optimizations for VLIW/EPIC architectures
  \item GC safety analysis
\end{itemize}

\subsection{Basic Concepts}

  Basic concepts in MLRISC are: 
\begin{itemize}
    \item \href{instructions.html}{Instructions} --
    the instruction set of the target architecture.
    \item \href{cells.html}{Cells} -- which describes registers,
memory and other mutable resources in the machine.
    \item \href{regions.html}{Regions} -- a client defined
   abstract type used to represent aliasing information available from
the front-end.
    \item \href{constants.html}{Constants} -- a client defined
   place holder used to represent constants whose values are unknown 
   in the front-end. 
    \item \href{pseudo-ops.html}{Pseudo Ops} -- a client defined
      
    \item \href{annotations.html}{Annotations} -- this is
   a generic mechinism for propagating information in the MLRISC sstem.
   The client may attach arbitrary annotation of various granularity 
   to MLRISC's program representation,
   which can then be propagated to later phases.
   These can be information related to profiling frequency, dependence, 
   comments, and/or types.
   The same mechanism is also used to propagate 
   analysis information one optimization phase to 
   another.
    \item \href{streams.html}{Instruction Streams} -- an abstraction
   for describing a stream of instructions.  Instruction streams are
   used to connect modules such as instruction selection,  assembler, 
   machine code emitter, and 
   control flow graph builder.
   \item \href{regmap.html}{Regmap} -- a mapping between registers
     names.  MLRISC register allocators represent the result of register
   allocation as a regmap.
   \item \href{labels.html}{Labels} -- a type representing
symbolic labels.
   \item \href{labelexp.html}{Label Expressions} -- a type representing
     constant expressions
    involving symbolic labels.
\end{itemize}

\subsection{How Things Are Fit Together}

  MLRISC uses two different program representations, clusters and MLRISC IR.
\begin{itemize}
  \item \href{cluster.html}{Cluster} is light-weight representation
that is used when only the most basic optimizations are required.
  \item \href{mlrisc-ir.html}{MLRISC IR} is more heavy-weight
   representation that is built from the 
    \href{graphs.html}{MLRISC graph library} and the
    \href{compiler-graphs.html}{MLRISC compiler graph library}.
   MLRISC IR allows more complex transformations and analysis of the
   program graph.
\end{itemize}
Conversion modules between the two representations are provided.

In general MLRISC optimization phases are transformations applied on one
of these representations.  Optimizations may be chained together to form
a compiler backend.  For example, a minimal backend consists of
\begin{itemize}
  \item the instruction selection module, which translates 
\href{mltree.html}{MLTree} into target instructions,
  \item the flowgraph builder, which conversts a stream of target instructions
   into a cluster,
  \item the register allocator, which performs register allocation, and
  \item the assembly code emitter, which generates assembly output
\end{itemize}