mlite / mlitecc Goto Github PK

                        The Mlite C Compiler

This directory contains the source code for Mlite C Compiler.  It is written in
Ocaml and C. It borrows the C Parser from CIL and does its own C construct
simplification. Unlike CIL, it uses a sequence of small step translations to
simplify C construct. Each step of translation can be tested by using gcc to
compile the emitted C code. There are 12 translation steps. Each translation
results in a simpler AST. The ASTs are named as Ast_xx_yyyy. Where xx are two
alphabatic characters and yyyy describes what construct is simplified in this
AST.

Ast_aa_gram is the initial syntax tree constructed from Cabs of CIL. The AST is
close to the grammar of C99 specification with some GCC extension.

Ast_ba_stmt is the syntax tree in which the "for" statement is simplified.

Ast_ca_expr is the syntax tree in which the "c_expression" non-terminal in C99
specification is simplificed.

Ast_da_type is the syntax tree in which declaration and declarator are combined
such that each declaration only declare on identifier. Type declaration and
object declarations are seperated, e.g., struct type0 { ... } v0, v1 [] = ...;"
is translated to "struct type0 {...}; struct typ0 v0; struct type0 v1 [] = ...."

In the implementation, Identifier in the C programming language which denotes
data or code is a closure entity. Identifer representing a build-in type, or a
typedef type is a type entity.

Ast_ea_expr is the syntax tree in which all identifiers are tokenized as either
Closure entity or Type entity and expression are further simplified. If an
identifier represents an enumeration constant, a variable or a function, it is
tokenized as a closure entity and represented as one entry in
Cent.ml. Otherwise, the identifier denotes a type, it is tokenized as a Type
entity and represented as one entry in Tent.ml. The storage size of a type is
also computed.

Ast_eb_expr is the syntax tree in which all arithmatic operations and type
convertions are augmented with the storage sizes.

Ast_ec_expr converts Local_obj_decl_init to Obj_decl_init

Ast_fa_stmt is the syntax tree in which high level control constructs are
simplified as GOTO or JMP. Nested and paired BEGIN_DECL and END_DECL are
inserted for variable declarations to keep their local scopes.

Ast_ga_code is the syntax tree in which all instructions are stored in an code
array, and the target labels of GOTO or JMP are tokenized as indices in the code
array.

Ast_ha_graf is the syntax tree in which control flow graph is explicitly
reprsented and dominance relation is computed.

Ast_ia_init is the syntax tree in which static or dynamic variable
initialization are simplified.

Cmm_ast is the syntax tree of C--.