<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"
"http://www.w3.org/TR/REC-html40/loose.dtd">
<HTML>
<HEAD>
<TITLE>cilpp</TITLE>
<META http-equiv="Content-Type" content="text/html; charset=US-ASCII">
<META name="GENERATOR" content="hevea 1.10">
<STYLE type="text/css">
.li-itemize{margin:1ex 0ex;}
.li-enumerate{margin:1ex 0ex;}
.dd-description{margin:0ex 0ex 1ex 4ex;}
.dt-description{margin:0ex;}
.toc{list-style:none;}
.thefootnotes{text-align:left;margin:0ex;}
.dt-thefootnotes{margin:0em;}
.dd-thefootnotes{margin:0em 0em 0em 2em;}
.footnoterule{margin:1em auto 1em 0px;width:50%;}
.caption{padding-left:2ex; padding-right:2ex; margin-left:auto; margin-right:auto}
.title{margin:2ex auto;text-align:center}
.center{text-align:center;margin-left:auto;margin-right:auto;}
.flushleft{text-align:left;margin-left:0ex;margin-right:auto;}
.flushright{text-align:right;margin-left:auto;margin-right:0ex;}
DIV TABLE{margin-left:inherit;margin-right:inherit;}
PRE{text-align:left;margin-left:0ex;margin-right:auto;}
BLOCKQUOTE{margin-left:4ex;margin-right:4ex;text-align:left;}
TD P{margin:0px;}
.boxed{border:1px solid black}
.textboxed{border:1px solid black}
.vbar{border:none;width:2px;background-color:black;}
.hbar{border:none;height:2px;width:100%;background-color:black;}
.hfill{border:none;height:1px;width:200%;background-color:black;}
.vdisplay{border-collapse:separate;border-spacing:2px;width:auto; empty-cells:show; border:2px solid red;}
.vdcell{white-space:nowrap;padding:0px;width:auto; border:2px solid green;}
.display{border-collapse:separate;border-spacing:2px;width:auto; border:none;}
.dcell{white-space:nowrap;padding:0px;width:auto; border:none;}
.dcenter{margin:0ex auto;}
.vdcenter{border:solid #FF8000 2px; margin:0ex auto;}
.minipage{text-align:left; margin-left:0em; margin-right:auto;}
.marginpar{border:solid thin black; width:20%; text-align:left;}
.marginparleft{float:left; margin-left:0ex; margin-right:1ex;}
.marginparright{float:right; margin-left:1ex; margin-right:0ex;}
.theorem{text-align:left;margin:1ex auto 1ex 0ex;}
.part{margin:2ex auto;text-align:center}
</STYLE>
<base target="main">
<script language="JavaScript">
<!-- Begin
function loadTop(url) {
parent.location.href= url;
}
// -->
</script>
</HEAD>
<BODY >
<!--HEVEA command line is: hevea -exec xxdate.exe ../../cilpp -->
<!--CUT DEF section 1 --><TABLE CLASS="title"><TR><TD></TD></TR>
</TABLE><!--TOC section Introduction-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc1">1</A>  Introduction</H2><!--SEC END --><P>CIL has a Source Forge page:
<A HREF="javascript:loadTop('http://sourceforge.net/projects/cil')">http://sourceforge.net/projects/cil</A>. </P><P>CIL (<B>C</B> <B>I</B>ntermediate <B>L</B>anguage) is a high-level representation
along with a set of tools that permit easy analysis and source-to-source
transformation of C programs.</P><P>CIL is both lower-level than abstract-syntax trees, by clarifying ambiguous
constructs and removing redundant ones, and also higher-level than typical
intermediate languages designed for compilation, by maintaining types and a
close relationship with the source program. The main advantage of CIL is that
it compiles all valid C programs into a few core constructs with a very clean
semantics. Also CIL has a syntax-directed type system that makes it easy to
analyze and manipulate C programs. Furthermore, the CIL front-end is able to
process not only ANSI-C programs but also those using Microsoft C or GNU C
extensions. If you do not use CIL and want instead to use just a C parser and
analyze programs expressed as abstract-syntax trees then your analysis will
have to handle a lot of ugly corners of the language (let alone the fact that
parsing C itself is not a trivial task). See Section <A HREF="#sec-simplec">16</A> for some
examples of such extreme programs that CIL simplifies for you.</P><P>In essence, CIL is a highly-structured, “clean” subset of C. CIL features a
reduced number of syntactic and conceptual forms. For example, all looping
constructs are reduced to a single form, all function bodies are given
explicit <TT>return</TT> statements, syntactic sugar like <TT>"->"</TT> is
eliminated and function arguments with array types become pointers. (For an
extensive list of how CIL simplifies C programs, see Section <A HREF="#sec-cabs2cil">4</A>.)
This reduces the number of cases that must be considered when manipulating a C
program. CIL also separates type declarations from code and flattens scopes
within function bodies. This structures the program in a manner more amenable
to rapid analysis and transformation. CIL computes the types of all program
expressions, and makes all type promotions and casts explicit. CIL supports
all GCC and MSVC extensions except for nested functions and complex numbers.
Finally, CIL organizes C’s imperative features into expressions, instructions
and statements based on the presence and absence of side-effects and
control-flow. Every statement can be annotated with successor and predecessor
information. Thus CIL provides an integrated program representation that can
be used with routines that require an AST (e.g. type-based analyses and
pretty-printers), as well as with routines that require a CFG (e.g., dataflow
analyses). CIL also supports even lower-level representations (e.g.,
three-address code), see Section <A HREF="#sec-Extension">8</A>. </P><P>CIL comes accompanied by a number of Perl scripts that perform generally
useful operations on code:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
A <A HREF="#sec-driver">driver</A> which behaves as either the <TT>gcc</TT> or
Microsoft VC compiler and can invoke the preprocessor followed by the CIL
application. The advantage of this script is that you can easily use CIL and
the analyses written for CIL with existing make files.
</LI><LI CLASS="li-itemize">A <A HREF="#sec-merger">whole-program merger</A> that you can use as a
replacement for your compiler and it learns all the files you compile when you
make a project and merges all of the preprocessed source files into a single
one. This makes it easy to do whole-program analysis.
</LI><LI CLASS="li-itemize">A <A HREF="#sec-patcher">patcher</A> makes it easy to create modified
copies of the system include files. The CIL driver can then be told to use
these patched copies instead of the standard ones.
</LI></UL><P>CIL has been tested very extensively. It is able to process the SPECINT95
benchmarks, the Linux kernel, GIMP and other open-source projects. All of
these programs are compiled to the simple CIL and then passed to <TT>gcc</TT> and
they still run! We consider the compilation of Linux a major feat especially
since Linux contains many of the ugly GCC extensions (see Section <A HREF="#sec-ugly-gcc">16.2</A>).
This adds to about 1,000,000 lines of code that we tested it on. It is also
able to process the few Microsoft NT device drivers that we have had access
to. CIL was tested against GCC’s c-torture testsuite and (except for the tests
involving complex numbers and inner functions, which CIL does not currently
implement) CIL passes most of the tests. Specifically CIL fails 23 tests out
of the 904 c-torture tests that it should pass. GCC itself fails 19 tests. A
total of 1400 regression test cases are run automatically on each change to
the CIL sources.</P><P>CIL is relatively independent on the underlying machine and compiler. When
you build it CIL will configure itself according to the underlying compiler.
However, CIL has only been tested on Intel x86 using the gcc compiler on Linux
and cygwin and using the MS Visual C compiler. (See below for specific
versions of these compilers that we have used CIL for.)</P><P>The largest application we have used CIL for is
<A HREF="javascript:loadTop('../ccured/index.html')">CCured</A>, a compiler that compiles C code into
type-safe code by analyzing your pointer usage and inserting runtime checks in
the places that cannot be guaranteed statically to be type safe. </P><P>You can also use CIL to “compile” code that uses GCC extensions (e.g. the
Linux kernel) into standard C code.</P><P>CIL also comes accompanies by a growing library of extensions (see
Section <A HREF="#sec-Extension">8</A>). You can use these for your projects or as examples of
using CIL. </P><P><TT>PDF</TT> versions of <A HREF="CIL.pdf">this manual</A> and the
<A HREF="CIL-API.pdf">CIL API</A> are available. However, we recommend the
<TT>HTML</TT> versions because the postprocessed code examples are easier to
view. </P><P>If you use CIL in your project, we would appreciate letting us know. If you
want to cite CIL in your research writings, please refer to the paper “CIL:
Intermediate Language and Tools for Analysis and Transformation of C
Programs” by George C. Necula, Scott McPeak, S.P. Rahul and Westley Weimer,
in “Proceedings of Conference on Compilier Construction”, 2002.</P><!--TOC section Installation-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc2">2</A>  Installation</H2><!--SEC END --><P>You need the following tools to build CIL:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
A Unix-like shell environment (with bash, perl, make, mv, cp,
etc.). On Windows, you will need cygwin with those packages.
</LI><LI CLASS="li-itemize">An ocaml compiler. You will need OCaml release 3.08 or higher to build
CIL. CIL has been tested on Linux and on Windows (where it can behave as
either Microsoft Visual C or gcc). On Windows, you can build CIL both with the
cygwin version of ocaml (preferred) and with the Win32 version of ocaml.
</LI><LI CLASS="li-itemize">An underlying C compiler, which can be either gcc or Microsoft Visual C.
</LI></UL><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Get the source code.
<UL CLASS="itemize"><LI CLASS="li-itemize">
<EM>Official distribution</EM> (Recommended):
<OL CLASS="enumerate" type=a><LI CLASS="li-enumerate">
Download the CIL <A HREF="distrib">distribution</A> (latest version is
<A HREF="distrib/cil-1.3.7.tar.gz"><TT>distrib/cil-1.3.7.tar.gz</TT></A>). See the Section <A HREF="#sec-changes">20</A> for recent changes to the CIL distribution.
</LI><LI CLASS="li-enumerate">Unzip and untar the source distribution. This will create a directory
called <TT>cil</TT> whose structure is explained below.<BR>
<TT>    tar xvfz cil-1.3.7.tar.gz</TT>
</LI></OL>
</LI><LI CLASS="li-itemize"><EM>Subversion Repository</EM>:<BR>
Alternately, you can download an up to the minute version of CIL
from our Subversion repository at:
<PRE CLASS="verbatim"> svn co svn://hal.cs.berkeley.edu/home/svn/projects/trunk/cil
</PRE>However, the Subversion version may be less stable than the released
version. See the Changes section of doc/cil.tex to see what’s
changed since the last release. There may be changes that aren’t yet
documented in the .tex file or this website.<P>For those who were using the CVS server before we switched to
Subversion, revision 8603 in Subversion corresponds to the last CVS
version.</P></LI></UL>
</LI><LI CLASS="li-enumerate">Enter the <TT>cil</TT> directory and run the <TT>configure</TT> script and then
GNU make to build the distribution. If you are on Windows, at least the
<TT>configure</TT> step must be run from within <TT>bash</TT>.<BR>
    <CODE>cd cil</CODE><BR>
    <CODE>./configure</CODE><BR>
    <CODE>make</CODE><BR>
    <CODE>make quicktest</CODE><BR>
</LI><LI CLASS="li-enumerate">You should now find <TT>cilly.asm.exe</TT> in a subdirectory of <TT>obj</TT>. The
name of the subdirectory is either <TT>x86_WIN32</TT> if you are using <TT>cygwin</TT>
on Windows or <TT>x86_LINUX</TT> if you are using Linux (although you should be
using instead the Perl wrapper <TT>bin/cilly</TT>). Note that we do not have an
<TT>install</TT> make target and you should use Cil from the development directory.
</LI><LI CLASS="li-enumerate">If you decide to use CIL, <B>please</B>
<A HREF="mailto:necula@cs.berkeley.edu">send us a note</A>. This will help recharge
our batteries after a few years of development. And of course, do send us
your bug reports as well.
</LI></OL><P>The <TT>configure</TT> script tries to find appropriate defaults for your system.
You can control its actions by passing the following arguments:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>CC=foo</TT> Specifies the path for the <TT>gcc</TT> executable. By default
whichever version is in the PATH is used. If <TT>CC</TT> specifies the Microsoft
<TT>cl</TT> compiler, then that compiler will be set as the default one. Otherwise,
the <TT>gcc</TT> compiler will be the default.
</LI></UL><P>CIL requires an underlying C compiler and preprocessor. CIL depends on the
underlying compiler and machine for the sizes and alignment of types. The
installation procedure for CIL queries the underlying compiler for
architecture and compiler dependent configuration parameters, such as the size
of a pointer or the particular alignment rules for structure fields. (This
means, of course, that you should re-run <TT>./configure</TT> when you move CIL to
another machine.)</P><P>We have tested CIL on the following compilers:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
On Windows, <TT>cl</TT> compiler version 12.00.8168 (MSVC 6),
13.00.9466 (MSVC .Net), and 13.10.3077 (MSVC .Net 2003). Run <TT>cl</TT>
with no arguments to get the compiler version.
</LI><LI CLASS="li-itemize">On Windows, using <TT>cygwin</TT> and <TT>gcc</TT> version 2.95.3, 3.0,
3.2, 3.3, and 3.4.
</LI><LI CLASS="li-itemize">On Linux, using <TT>gcc</TT> version 2.95.3, 3.0, 3.2, 3.3, 4.0, and 4.1.
</LI></UL><P>Others have successfully used CIL on x86 processors with Mac OS X,
FreeBSD and OpenBSD; on amd64 processors with FreeBSD; on SPARC
processors with Solaris; and on PowerPC processors with Mac OS X. If
you make any changes to the build system in order to run CIL on your
platform, please send us a patch.</P><!--TOC subsection Building CIL on Windows with Microsoft Visual C-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc3">2.1</A>  Building CIL on Windows with Microsoft Visual C</H3><!--SEC END --><P>Some users might want to build a standalone CIL executable on Windows (an
executable that does not require cygwin.dll to run). You will need cygwin for
the build process only. Here is how we do it</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Start with a clean CIL directory
</LI><LI CLASS="li-enumerate">Start a command-line window setup with the environment variables for
Microsoft Visual Studio. You can do this by choosing Programs/Microsoft
Visual Studio/Tools/Command Prompt. Check that you can run <TT>cl</TT>.
</LI><LI CLASS="li-enumerate">Ensure that <TT>ocamlc</TT> refers to a Win32 version of ocaml. Run <TT>ocamlc
-v</TT> and look at the path to the standard library. If you have several
versions of ocaml, you must set the following variables:
<PRE CLASS="verbatim"> set OCAMLWIN=C:/Programs/ocaml-win
set OCAMLLIB=%OCAMLWIN%/lib
set PATH=%OCAMLWIN%/bin;%PATH%
set INCLUDE=%INCLUDE%;%OCAMLWIN%/inc
set LIB=%LIB%;%OCAMLWIN%/lib;obj/x86_WIN32
</PRE></LI><LI CLASS="li-enumerate">Run <TT>bash -c "./configure CC=cl"</TT>.
</LI><LI CLASS="li-enumerate">Run <TT>bash -c "make WIN32=1 quickbuild"</TT>
</LI><LI CLASS="li-enumerate">Run <TT>bash -c "make WIN32=1 NATIVECAML=1 cilly</TT>
</LI><LI CLASS="li-enumerate">Run <TT>bash -c "make WIN32=1 doc</TT>
</LI><LI CLASS="li-enumerate">Run <TT>bash -c "make WIN32=1 bindistrib-nocheck</TT>
</LI></OL><P>The above steps do not build the CIL library, but just the executable. The
last step will create a subdirectory <TT>TEMP_cil-bindistrib</TT> that contains
everything that you need to run CIL on another machine. You will have to edit
manually some of the files in the <TT>bin</TT> directory to replace <TT>CILHOME</TT>.
The resulting CIL can be run with ActiveState Perl also. </P><!--TOC section Distribution Contents-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc4">3</A>  Distribution Contents</H2><!--SEC END --><P>The file <A HREF="distrib/cil-1.3.7.tar.gz"><TT>distrib/cil-1.3.7.tar.gz</TT></A>
contains the complete source CIL distribution,
consisting of the following files:</P><P>
</P><BLOCKQUOTE CLASS="table"><DIV CLASS="center"><HR WIDTH="80%" SIZE=2></DIV><DIV CLASS="center"><TABLE CELLSPACING=6 CELLPADDING=0><TR><TD VALIGN=top ALIGN=left NOWRAP>
<EM>Filename</EM></TD><TD VALIGN=top ALIGN=left><EM>Description</EM></TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP><TT>Makefile.in</TT></TD><TD VALIGN=top ALIGN=left><TT>configure</TT> source for the
Makefile that builds CIL/</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>configure</TT></TD><TD VALIGN=top ALIGN=left>The configure script.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>configure.in</TT></TD><TD VALIGN=top ALIGN=left>The <TT>autoconf</TT> source for <TT>configure</TT>.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>config.guess</TT> </TD><TD VALIGN=top ALIGN=left>Stuff required by <TT>configure</TT>.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>config.sub</TT> </TD><TD VALIGN=top ALIGN=left>idem</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>install-sh</TT> </TD><TD VALIGN=top ALIGN=left>idem</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>doc/</TT></TD><TD VALIGN=top ALIGN=left>HTML documentation of the CIL API.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>obj/</TT></TD><TD VALIGN=top ALIGN=left>Directory that will contain the compiled
CIL modules and executables.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>bin/cilly.in</TT></TD><TD VALIGN=top ALIGN=left>The <TT>configure</TT> source for a Perl script
that can be invoked with the
same arguments as either <TT>gcc</TT> or
Microsoft Visual C and will convert the
program to CIL, perform some simple
transformations, emit it and compile it as
usual.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>lib/CompilerStub.pm</TT></TD><TD VALIGN=top ALIGN=left>A Perl class that can be used to write code
that impersonates a compiler. <TT>cilly</TT>
uses it.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>lib/Merger.pm</TT></TD><TD VALIGN=top ALIGN=left>A subclass of <TT>CompilerStub.pm</TT> that can
be used to merge source files into a single
source file.<TT>cilly</TT>
uses it.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>bin/patcher.in</TT></TD><TD VALIGN=top ALIGN=left>A Perl script that applies specified patches
to standard include files.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/check.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Checks the well-formedness of a CIL file.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/cil.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Definition of CIL abstract syntax and
utilities for manipulating it.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/clist.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Utilities for efficiently managing lists
that need to be concatenated often.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/errormsg.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Utilities for error reporting.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/ext/heapify.ml</TT></TD><TD VALIGN=top ALIGN=left>A CIL transformation that moves array local
variables from the stack to the heap.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/ext/logcalls.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>A CIL transformation that logs every
function call.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/ext/sfi.ml</TT></TD><TD VALIGN=top ALIGN=left>A CIL transformation that can log every
memory read and write.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/frontc/clexer.mll</TT></TD><TD VALIGN=top ALIGN=left>The lexer.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/frontc/cparser.mly</TT></TD><TD VALIGN=top ALIGN=left>The parser.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/frontc/cabs.ml</TT></TD><TD VALIGN=top ALIGN=left>The abstract syntax.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/frontc/cprint.ml</TT></TD><TD VALIGN=top ALIGN=left>The pretty printer for CABS.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/frontc/cabs2cil.ml</TT></TD><TD VALIGN=top ALIGN=left>The elaborator to CIL.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/main.ml</TT></TD><TD VALIGN=top ALIGN=left>The <TT>cilly</TT> application.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/pretty.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Utilities for pretty printing.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/rmtmps.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>A CIL tranformation that removes unused
types, variables and inlined functions.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/stats.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Utilities for maintaining timing statistics.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/testcil.ml</TT></TD><TD VALIGN=top ALIGN=left>A random test of CIL (against the resident
C compiler).</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/trace.ml,mli</TT></TD><TD VALIGN=top ALIGN=left>Utilities useful for printing debugging
information.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>ocamlutil/</TT></TD><TD VALIGN=top ALIGN=left>Miscellaneous libraries that are not
specific to CIL.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>ocamlutil/Makefile.ocaml</TT></TD><TD VALIGN=top ALIGN=left>A file that is included by <TT>Makefile</TT>.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>ocamlutil/perfcount.c</TT></TD><TD VALIGN=top ALIGN=left>C code that links with src/stats.ml
and reads Intel performance
counters.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>obj/@ARCHOS@/feature_config.ml</TT></TD><TD VALIGN=top ALIGN=left>File generated by the Makefile
describing which extra “features”
to compile. See Section <A HREF="#sec-cil">5</A>.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>obj/@ARCHOS@/machdep.ml</TT></TD><TD VALIGN=top ALIGN=left>File generated by the Makefile containing
information about your architecture,
such as the size of a pointer.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP>
<TT>src/machdep-ml.c</TT></TD><TD VALIGN=top ALIGN=left>C program that generates
<TT>machdep.ml</TT> files.</TD></TR>
<TR><TD VALIGN=top ALIGN=left NOWRAP> </TD></TR>
</TABLE>
</DIV><DIV CLASS="center"><HR WIDTH="80%" SIZE=2></DIV></BLOCKQUOTE><!--TOC section Compiling C to CIL-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc5">4</A>  Compiling C to CIL</H2><!--SEC END --><P><A NAME="sec-cabs2cil"></A></P><P>In this section we try to describe a few of the many transformations that are
applied to a C program to convert it to CIL. The module that implements this
conversion is about 5000 lines of OCaml code. In contrast a simple program
transformation that instruments all functions to keep a shadow stack of the
true return address (thus preventing stack smashing) is only 70 lines of code.
This example shows that the analysis is so much simpler because it has to
handle only a few simple C constructs and also because it can leverage on CIL
infrastructure such as visitors and pretty-printers.</P><P>In no particular order these are a few of the most significant ways in which
C programs are compiled into CIL:
</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
CIL will eliminate all declarations for unused entities. This means that
just because your hello world program includes <TT>stdio.h</TT> it does not mean
that your analysis has to handle all the ugly stuff from <TT>stdio.h</TT>.</LI><LI CLASS="li-enumerate">Type specifiers are interpreted and normalized:
<PRE CLASS="verbatim"><FONT COLOR=blue>int long signed x;
signed long extern x;
long static int long y;
// Some code that uses these declaration, so that CIL does not remove them
int main() { return x + y; }
</FONT></PRE>
See the <A HREF="examples/ex1.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Anonymous structure and union declarations are given a name.
<PRE CLASS="verbatim"><FONT COLOR=blue> struct { int x; } s;
</FONT></PRE>
See the <A HREF="examples/ex2.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Nested structure tag definitions are pulled apart. This means that all
structure tag definitions can be found by a simple scan of the globals.<PRE CLASS="verbatim"><FONT COLOR=blue>struct foo {
struct bar {
union baz {
int x1;
double x2;
} u1;
int y;
} s1;
int z;
} f;
</FONT></PRE><P>
See the <A HREF="examples/ex3.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">All structure, union, enumeration definitions and the type definitions
from inners scopes are moved to global scope (with appropriate renaming). This
facilitates moving around of the references to these entities.<PRE CLASS="verbatim"><FONT COLOR=blue>int main() {
struct foo {
int x; } foo;
{
struct foo {
double d;
};
return foo.x;
}
}
</FONT></PRE><P>
See the <A HREF="examples/ex4.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Prototypes are added for those functions that are called before being
defined. Furthermore, if a prototype exists but does not specify the type of
parameters that is fixed. But CIL will not be able to add prototypes for those
functions that are neither declared nor defined (but are used!).
<PRE CLASS="verbatim"><FONT COLOR=blue> int f(); // Prototype without arguments
int f(double x) {
return g(x);
}
int g(double x) {
return x;
}
</FONT></PRE>
See the <A HREF="examples/ex5.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Array lengths are computed based on the initializers or by constant
folding.
<PRE CLASS="verbatim"><FONT COLOR=blue> int a1[] = {1,2,3};
int a2[sizeof(int) >= 4 ? 8 : 16];
</FONT></PRE>
See the <A HREF="examples/ex6.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Enumeration tags are computed using constant folding:
<PRE CLASS="verbatim"><FONT COLOR=blue>int main() {
enum {
FIVE = 5,
SIX, SEVEN,
FOUR = FIVE - 1,
EIGHT = sizeof(double)
} x = FIVE;
return x;
}
</FONT></PRE>
See the <A HREF="examples/ex7.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Initializers are normalized to include specific initialization for the
missing elements:
<PRE CLASS="verbatim"><FONT COLOR=blue> int a1[5] = {1,2,3};
struct foo { int x, y; } s1 = { 4 };
</FONT></PRE>
See the <A HREF="examples/ex8.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Initializer designators are interpreted and eliminated. Subobjects are
properly marked with braces. CIL implements
the whole ISO C99 specification for initializer (neither GCC nor MSVC do) and
a few GCC extensions.
<PRE CLASS="verbatim"><FONT COLOR=blue> struct foo {
int x, y;
int a[5];
struct inner {
int z;
} inner;
} s = { 0, .inner.z = 3, .a[1 ... 2] = 5, 4, y : 8 };
</FONT></PRE>
See the <A HREF="examples/ex9.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">String initializers for arrays of characters are processed<PRE CLASS="verbatim"><FONT COLOR=blue>char foo[] = "foo plus bar";
</FONT></PRE><P>
See the <A HREF="examples/ex10.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">String constants are concatenated<PRE CLASS="verbatim"><FONT COLOR=blue>char *foo = "foo " " plus " " bar ";
</FONT></PRE><P>
See the <A HREF="examples/ex11.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Initializers for local variables are turned into assignments. This is in
order to separate completely the declarative part of a function body from the
statements. This has the unfortunate effect that we have to drop the <TT>const</TT>
qualifier from local variables !<PRE CLASS="verbatim"><FONT COLOR=blue> int x = 5;
struct foo { int f1, f2; } a [] = {1, 2, 3, 4, 5 };
</FONT></PRE><P>
See the <A HREF="examples/ex12.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Local variables in inner scopes are pulled to function scope (with
appropriate renaming). Local scopes thus disappear. This makes it easy to find
and operate on all local variables in a function.<PRE CLASS="verbatim"><FONT COLOR=blue> int x = 5;
int main() {
int x = 6;
{
int x = 7;
return x;
}
return x;
}
</FONT></PRE><P>
See the <A HREF="examples/ex13.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Global declarations in local scopes are moved to global scope:
<PRE CLASS="verbatim"><FONT COLOR=blue> int x = 5;
int main() {
int x = 6;
{
static int x = 7;
return x;
}
return x;
}
</FONT></PRE>
See the <A HREF="examples/ex14.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">Return statements are added for functions that are missing them. If the
return type is not a base type then a <TT>return</TT> without a value is added.
The guaranteed presence of return statements makes it easy to implement a
transformation that inserts some code to be executed immediately before
returning from a function.
<PRE CLASS="verbatim"><FONT COLOR=blue> int foo() {
int x = 5;
}
</FONT></PRE>
See the <A HREF="examples/ex15.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">One of the most significant transformations is that expressions that
contain side-effects are separated into statements. <PRE CLASS="verbatim"><FONT COLOR=blue> int x, f(int);
return (x ++ + f(x));
</FONT></PRE><P>
See the <A HREF="examples/ex16.txt">CIL output</A> for this
code fragment</P><P>Internally, the <TT>x ++</TT> statement is turned into an assignment which the
pretty-printer prints like the original. CIL has only three forms of basic
statements: assignments, function calls and inline assembly.</P></LI><LI CLASS="li-enumerate">Shortcut evaluation of boolean expressions and the <TT>?:</TT> operator are
compiled into explicit conditionals:
<PRE CLASS="verbatim"><FONT COLOR=blue> int x;
int y = x ? 2 : 4;
int z = x || y;
// Here we duplicate the return statement
if(x && y) { return 0; } else { return 1; }
// To avoid excessive duplication, CIL uses goto's for
// statement that have more than 5 instructions
if(x && y || z) { x ++; y ++; z ++; x ++; y ++; return z; }
</FONT></PRE>
See the <A HREF="examples/ex17.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">GCC’s conditional expression with missing operands are also compiled
into conditionals:
<PRE CLASS="verbatim"><FONT COLOR=blue> int f();;
return f() ? : 4;
</FONT></PRE>
See the <A HREF="examples/ex18.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">All forms of loops (<TT>while</TT>, <TT>for</TT> and <TT>do</TT>) are compiled
internally as a single <TT>while(1)</TT> looping construct with explicit <TT>break</TT>
statement for termination. For simple <TT>while</TT> loops the pretty printer is
able to print back the original:
<PRE CLASS="verbatim"><FONT COLOR=blue> int x, y;
for(int i = 0; i<5; i++) {
if(i == 5) continue;
if(i == 4) break;
i += 2;
}
while(x < 5) {
if(x == 3) continue;
x ++;
}
</FONT></PRE>
See the <A HREF="examples/ex19.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">GCC’s block expressions are compiled away. (That’s right there is an
infinite loop in this code.)<PRE CLASS="verbatim"><FONT COLOR=blue> int x = 5, y = x;
int z = ({ x++; L: y -= x; y;});
return ({ goto L; 0; });
</FONT></PRE><P>
See the <A HREF="examples/ex20.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">CIL contains support for both MSVC and GCC inline assembly (both in one
internal construct)</LI><LI CLASS="li-enumerate">CIL compiles away the GCC extension that allows many kinds of constructs
to be used as lvalues:<PRE CLASS="verbatim"><FONT COLOR=blue> int x, y, z;
return &(x ? y : z) - & (x ++, x);
</FONT></PRE><P>
See the <A HREF="examples/ex21.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">All types are computed and explicit casts are inserted for all
promotions and conversions that a compiler must insert:</LI><LI CLASS="li-enumerate">CIL will turn old-style function definition (without prototype) into
new-style definitions. This will make the compiler less forgiving when
checking function calls, and will catch for example cases when a function is
called with too few arguments. This happens in old-style code for the purpose
of implementing variable argument functions.</LI><LI CLASS="li-enumerate">Since CIL sees the source after preprocessing the code after CIL does
not contain the comments and the preprocessing directives.</LI><LI CLASS="li-enumerate">CIL will remove from the source file those type declarations, local
variables and inline functions that are not used in the file. This means that
your analysis does not have to see all the ugly stuff that comes from the
header files:
<PRE CLASS="verbatim"><FONT COLOR=blue>#include <stdio.h>
typedef int unused_type;
static char unused_static (void) { return 0; }
int main() {
int unused_local;
printf("Hello world\n"); // Only printf will be kept from stdio.h
}
</FONT></PRE>
See the <A HREF="examples/ex22.txt">CIL output</A> for this
code fragment</LI></OL><!--TOC section How to Use CIL-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc6">5</A>  How to Use CIL</H2><!--SEC END --><P><A NAME="sec-cil"></A></P><!--NAME cilly.html-->
<P>There are two predominant ways to use CIL to write a program analysis or
transformation. The first is to phrase your analysis as a module that is
called by our existing driver. The second is to use CIL as a stand-alone
library. We highly recommend that you use <TT>cilly</TT>, our driver. </P><!--TOC subsection Using <TT>cilly</TT>, the CIL driver-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc7">5.1</A>  Using <TT>cilly</TT>, the CIL driver</H3><!--SEC END --><P>The most common way to use CIL is to write an Ocaml module containing your
analysis and transformation, which you then link into our boilerplate
driver application called <TT>cilly</TT>. <TT>cilly</TT> is a Perl script that
processes and mimics <TT>GCC</TT> and <TT>MSVC</TT> command-line arguments and then
calls <TT>cilly.byte.exe</TT> or <TT>cilly.asm.exe</TT> (CIL’s Ocaml executable). </P><P>An example of such module is <TT>logwrites.ml</TT>, a transformation that is
distributed with CIL and whose purpose is to instrument code to print the
addresses of memory locations being written. (We plan to release a
C-language interface to CIL so that you can write your analyses in C
instead of Ocaml.) See Section <A HREF="#sec-Extension">8</A> for a survey of other example
modules. </P><P>Assuming that you have written <TT>/home/necula/logwrites.ml</TT>,
here is how you use it: </P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">Modify <TT>logwrites.ml</TT> so that it includes a CIL “feature
descriptor” like this:
<PRE CLASS="verbatim">let feature : featureDescr =
{ fd_name = "logwrites";
fd_enabled = ref false;
fd_description = "generation of code to log memory writes";
fd_extraopt = [];
fd_doit =
(function (f: file) ->
let lwVisitor = new logWriteVisitor in
visitCilFileSameGlobals lwVisitor f)
}
</PRE>The <TT>fd_name</TT> field names the feature and its associated
command-line arguments. The <TT>fd_enabled</TT> field is a <TT>bool ref</TT>.
“<TT>fd_doit</TT>” will be invoked if <TT>!fd_enabled</TT> is true after
argument parsing, so initialize the ref cell to true if you want
this feature to be enabled by default.<P>When the user passes the <TT>--dologwrites</TT>
command-line option to <TT>cilly</TT>, the variable associated with the
<TT>fd_enabled</TT> flag is set and the <TT>fd_doit</TT> function is called
on the <TT>Cil.file</TT> that represents the merger (see Section <A HREF="#sec-merger">13</A>) of
all C files listed as arguments. </P></LI><LI CLASS="li-enumerate">Invoke <TT>configure</TT> with the arguments
<PRE CLASS="verbatim">./configure EXTRASRCDIRS=/home/necula EXTRAFEATURES=logwrites
</PRE><P>This step works if each feature is packaged into its own ML file, and the
name of the entry point in the file is <TT>feature</TT>.</P><P>An alternative way to specify the new features is to change the build files
yourself, as explained below. You’ll need to use this method if a single
feature is split across multiple files.</P><OL CLASS="enumerate" type=a><LI CLASS="li-enumerate">
Put <TT>logwrites.ml</TT> in the <TT>src</TT> or <TT>src/ext</TT> directory. This
will make sure that <TT>make</TT> can find it. If you want to put it in some
other directory, modify <TT>Makefile.in</TT> and add to <TT>SOURCEDIRS</TT> your
directory. Alternately, you can create a symlink from <TT>src</TT> or
<TT>src/ext</TT> to your file.</LI><LI CLASS="li-enumerate">Modify the <TT>Makefile.in</TT> and add your module to the
<TT>CILLY_MODULES</TT> or
<TT>CILLY_LIBRARY_MODULES</TT> variables. The order of the modules matters. Add
your modules somewhere after <TT>cil</TT> and before <TT>main</TT>.</LI><LI CLASS="li-enumerate">If you have any helper files for your module, add those to
the makefile in the same way. e.g.:<PRE CLASS="verbatim">CILLY_MODULES = $(CILLY_LIBRARY_MODULES) \
myutilities1 myutilities2 logwrites \
main
</PRE><P>Again, order is important: <TT>myutilities2.ml</TT> will be able to refer
to Myutilities1 but not Logwrites. If you have any ocamllex or ocamlyacc
files, add them to both <TT>CILLY_MODULES</TT> and either <TT>MLLS</TT> or
<TT>MLYS</TT>.</P></LI><LI CLASS="li-enumerate">Modify <TT>main.ml</TT> so that your new feature descriptor appears in
the global list of CIL features.
<PRE CLASS="verbatim">let features : C.featureDescr list =
[ Logcalls.feature;
Oneret.feature;
Heapify.feature1;
Heapify.feature2;
makeCFGFeature;
Partial.feature;
Simplemem.feature;
Logwrites.feature; (* add this line to include the logwrites feature! *)
]
@ Feature_config.features
</PRE><P>Features are processed in the order they appear on this list. Put
your feature last on the list if you plan to run any of CIL’s
built-in features (such as makeCFGfeature) before your own.</P></LI></OL><P>Standard code in <TT>cilly</TT> takes care of adding command-line arguments,
printing the description, and calling your function automatically.
Note: do not worry about introducing new bugs into CIL by adding a single
line to the feature list. </P></LI><LI CLASS="li-enumerate">Now you can invoke the <TT>cilly</TT> application on a preprocessed file, or
instead use the <TT>cilly</TT> driver which provides a convenient compiler-like
interface to <TT>cilly</TT>. See Section <A HREF="#sec-driver">7</A> for details using <TT>cilly</TT>.
Remember to enable your analysis by passing the right argument (e.g.,
<TT>--dologwrites</TT>). </LI></OL><!--TOC subsection Using CIL as a library-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc8">5.2</A>  Using CIL as a library</H3><!--SEC END --><P>CIL can also be built as a library that is called from your stand-alone
application. Add <TT>cil/src</TT>, <TT>cil/src/frontc</TT>, <TT>cil/obj/x86_LINUX</TT>
(or <TT>cil/obj/x86_WIN32</TT>) to your Ocaml project <TT>-I</TT> include paths.
Building CIL will also build the library <TT>cil/obj/*/cil.cma</TT> (or
<TT>cil/obj/*/cil.cmxa</TT>). You can then link your application against that
library. </P><P>You can call the <TT>Frontc.parse: string -> unit -> Cil.file</TT> function with
the name of a file containing the output of the C preprocessor.
The <TT>Mergecil.merge: Cil.file list -> string -> Cil.file</TT> function merges
multiple files. You can then invoke your analysis function on the resulting
<TT>Cil.file</TT> data structure. You might want to call
<TT>Rmtmps.removeUnusedTemps</TT> first to clean up the prototypes and variables
that are not used. Then you can call the function <TT>Cil.dumpFile:
cilPrinter -> out_channel -> Cil.file -> unit</TT> to print the file to a
given output channel. A good <TT>cilPrinter</TT> to use is
<TT>defaultCilPrinter</TT>. </P><P>Check out <TT>src/main.ml</TT> and <TT>bin/cilly</TT> for other good ideas
about high-level file processing. Again, we highly recommend that you just
our <TT>cilly</TT> driver so that you can avoid spending time re-inventing the
wheel to provide drop-in support for standard <TT>makefile</TT>s. </P><P>Here is a concrete example of compiling and linking your project against
CIL. Imagine that your program analysis or transformation is contained in
the single file <TT>main.ml</TT>. </P><PRE CLASS="verbatim">$ ocamlopt -c -I $(CIL)/obj/x86_LINUX/ main.ml
$ ocamlopt -ccopt -L$(CIL)/obj/x86_LINUX/ -o main unix.cmxa str.cmxa \
$(CIL)/obj/x86_LINUX/cil.cmxa main.cmx
</PRE><P>The first line compiles your analysis, the second line links it against CIL
(as a library) and the Ocaml Unix library. For more information about
compiling and linking Ocaml programs, see the Ocaml home page
at <A HREF="javascript:loadTop('http://caml.inria.fr/ocaml/')">http://caml.inria.fr/ocaml/</A>. </P><P>In the next section we give an overview of the API that you can use
to write your analysis and transformation. </P><!--TOC section CIL API Documentation-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc9">6</A>  CIL API Documentation</H2><!--SEC END --><P><A NAME="sec-api"></A> </P><P>The CIL API is documented in the file <TT>src/cil.mli</TT>. We also have an
<A HREF="api/index.html">online documentation</A> extracted from
<TT>cil.mli</TT> and other useful modules. We
index below the main types that are used to represent C programs in CIL: </P><UL CLASS="itemize"><LI CLASS="li-itemize">
<A HREF="api/index_types.html">An index of all types</A>
</LI><LI CLASS="li-itemize"><A HREF="api/index_values.html">An index of all values</A>
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEfile">Cil.file</A> is the representation of a file.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEglobal">Cil.global</A> is the representation of a global declaration or
definitions. Values for <A HREF="api/Cil.html#VALemptyFunction">operating on globals</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEtyp">Cil.typ</A> is the representation of a type.
Values for <A HREF="api/Cil.html#VALvoidType">operating on types</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEcompinfo">Cil.compinfo</A> is the representation of a structure or a union
type
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEfieldinfo">Cil.fieldinfo</A> is the representation of a field in a structure
or a union
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEenuminfo">Cil.enuminfo</A> is the representation of an enumeration type.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEvarinfo">Cil.varinfo</A> is the representation of a variable
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEfundec">Cil.fundec</A> is the representation of a function
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPElval">Cil.lval</A> is the representation of an lvalue.
Values for <A HREF="api/Cil.html#VALmakeVarinfo">operating on lvalues</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEexp">Cil.exp</A> is the representation of an expression without
side-effects.
Values for <A HREF="api/Cil.html#VALzero">operating on expressions</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEinstr">Cil.instr</A> is the representation of an instruction (with
side-effects but without control-flow)
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEstmt">Cil.stmt</A> is the representation of a control-flow statements.
Values for <A HREF="api/Cil.html#VALmkStmt">operating on statements</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#TYPEattribute">Cil.attribute</A> is the representation of attributes.
Values for <A HREF="api/Cil.html#TYPEattributeClass">operating on attributes</A>.
</LI></UL><!--TOC subsection Using the visitor-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc10">6.1</A>  Using the visitor</H3><!--SEC END --><P><A NAME="sec-visitor"></A></P><P>One of the most useful tools exported by the CIL API is an implementation of
the visitor pattern for CIL programs. The visiting engine scans depth-first
the structure of a CIL program and at each node is queries a user-provided
visitor structure whether it should do one of the following operations:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
Ignore this node and all its descendants
</LI><LI CLASS="li-itemize">Descend into all of the children and when done rebuild the node if any
of the children have changed.
</LI><LI CLASS="li-itemize">Replace the subtree rooted at the node with another tree.
</LI><LI CLASS="li-itemize">Replace the subtree with another tree, then descend into the children
and rebuild the node if necessary and then invoke a user-specified function.
</LI><LI CLASS="li-itemize">In addition to all of the above actions then visitor can specify that
some instructions should be queued to be inserted before the current
instruction or statement being visited.
</LI></UL><P>By writing visitors you can customize the program traversal and
transformation. One major limitation of the visiting engine is that it does
not propagate information from one node to another. Each visitor must use its
own private data to achieve this effect if necessary. </P><P>Each visitor is an object that is an instance of a class of type <A HREF="api/Cil.cilVisitor.html#.">Cil.cilVisitor..</A>
The most convenient way to obtain such classes is to specialize the
<A HREF="api/Cil.nopCilVisitor.html#c">Cil.nopCilVisitor.c</A>lass (which just traverses the tree doing
nothing). Any given specialization typically overrides only a few of the
methods. Take a look for example at the visitor defined in the module
<TT>logwrites.ml</TT>. Another, more elaborate example of a visitor is the
[copyFunctionVisitor] defined in <TT>cil.ml</TT>.</P><P>Once you have defined a visitor you can invoke it with one of the functions:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<A HREF="api/Cil.html#VALvisitCilFile">Cil.visitCilFile</A> or <A HREF="api/Cil.html#VALvisitCilFileSameGlobals">Cil.visitCilFileSameGlobals</A> - visit a file
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilGlobal">Cil.visitCilGlobal</A> - visit a global
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilFunction">Cil.visitCilFunction</A> - visit a function definition
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilExp">Cil.visitCilExp</A> - visit an expression
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilLval">Cil.visitCilLval</A> - visit an lvalue
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilInstr">Cil.visitCilInstr</A> - visit an instruction
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilStmt">Cil.visitCilStmt</A> - visit a statement
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALvisitCilType">Cil.visitCilType</A> - visit a type. Note that this does not visit
the files of a composite type. use visitGlobal to visit the [GCompTag] that
defines the fields.
</LI></UL><P>Some transformations may want to use visitors to insert additional
instructions before statements and instructions. To do so, pass a list of
instructions to the <A HREF="api/Cil.html#VALqueueInstr">Cil.queueInstr</A> method of the specialized
object. The instructions will automatically be inserted before that
instruction in the transformed code. The <A HREF="api/Cil.html#VALunqueueInstr">Cil.unqueueInstr</A> method
should not normally be called by the user. </P><!--TOC subsection Interpreted Constructors and Deconstructors-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc11">6.2</A>  Interpreted Constructors and Deconstructors</H3><!--SEC END --><P>Interpreted constructors and deconstructors are a facility for constructing
and deconstructing CIL constructs using a pattern with holes that can be
filled with a variety of kinds of elements. The pattern is a string that uses
the C syntax to represent C language elements. For example, the following
code:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Formatcil.cType "void * const (*)(int x)"
</FONT></PRE><P>is an alternative way to construct the internal representation of the type of pointer to function
with an integer argument and a void * const as result:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>TPtr(TFun(TVoid [Attr("const", [])],
[ ("x", TInt(IInt, []), []) ], false, []), [])
</FONT></PRE><P>The advantage of the interpreted constructors is that you can use familiar C
syntax to construct CIL abstract-syntax trees. </P><P>You can construct this way types, lvalues, expressions, instructions and
statements. The pattern string can also contain a number of placeholders that
are replaced during construction with CIL items passed as additional argument
to the construction function. For example, the <TT>%e:id</TT> placeholder means
that the argument labeled “id” (expected to be of form <TT>Fe exp</TT>) will
supply the expression to replace the placeholder. For example, the following
code constructs an increment instruction at location <TT>loc</TT>:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Formatcil.cInstr "%v:x = %v:x + %e:something"
loc
[ ("something", Fe some_exp);
("x", Fv some_varinfo) ]
</FONT></PRE><P>An alternative way to construct the same CIL instruction is:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Set((Var some_varinfo, NoOffset),
BinOp(PlusA, Lval (Var some_varinfo, NoOffset),
some_exp, intType),
loc)
</FONT></PRE><P>See <A HREF="api/Cil.html#TYPEformatArg">Cil.formatArg</A> for a definition of the placeholders that are
understood.</P><P>A dual feature is the interpreted deconstructors. This can be used to test
whether a CIL construct has a certain form:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Formatcil.dType "void * const (*)(int x)" t
</FONT></PRE><P>will test whether the actual argument <TT>t</TT> is indeed a function pointer of
the required type. If it is then the result is <TT>Some []</TT> otherwise it is
<TT>None</TT>. Furthermore, for the purpose of the interpreted deconstructors
placeholders in patterns match anything of the right type. For example,
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Formatcil.dType "void * (*)(%F:t)" t
</FONT></PRE><P>will match any function pointer type, independent of the type and number of
the formals. If the match succeeds the result is <TT>Some [ FF forms ]</TT> where
<TT>forms</TT> is a list of names and types of the formals. Note that each member
in the resulting list corresponds positionally to a placeholder in the
pattern.</P><P>The interpreted constructors and deconstructors do not support the complete C
syntax, but only a substantial fragment chosen to simplify the parsing. The
following is the syntax that is supported:
</P><PRE CLASS="verbatim">Expressions:
E ::= %e:ID | %d:ID | %g:ID | n | L | ( E ) | Unop E | E Binop E
| sizeof E | sizeof ( T ) | alignof E | alignof ( T )
| & L | ( T ) E
Unary operators:
Unop ::= + | - | ~ | %u:ID
Binary operators:
Binop ::= + | - | * | / | << | >> | & | ``|'' | ^
| == | != | < | > | <= | >= | %b:ID
Lvalues:
L ::= %l:ID | %v:ID Offset | * E | (* E) Offset | E -> ident Offset
Offsets:
Offset ::= empty | %o:ID | . ident Offset | [ E ] Offset
Types:
T ::= Type_spec Attrs Decl
Type specifiers:
Type_spec ::= void | char | unsigned char | short | unsigned short
| int | unsigned int | long | unsigned long | %k:ID | float
| double | struct %c:ID | union %c:ID
Declarators:
Decl ::= * Attrs Decl | Direct_decl
Direct declarators:
Direct_decl ::= empty | ident | ( Attrs Decl )
| Direct_decl [ Exp_opt ]
| ( Attrs Decl )( Parameters )
Optional expressions
Exp_opt ::= empty | E | %eo:ID
Formal parameters
Parameters ::= empty | ... | %va:ID | %f:ID | T | T , Parameters
List of attributes
Attrs ::= empty | %A:ID | Attrib Attrs
Attributes
Attrib ::= const | restrict | volatile | __attribute__ ( ( GAttr ) )
GCC Attributes
GAttr ::= ident | ident ( AttrArg_List )
Lists of GCC Attribute arguments:
AttrArg_List ::= AttrArg | %P:ID | AttrArg , AttrArg_List
GCC Attribute arguments
AttrArg ::= %p:ID | ident | ident ( AttrArg_List )
Instructions
Instr ::= %i:ID ; | L = E ; | L Binop= E | Callres L ( Args )
Actual arguments
Args ::= empty | %E:ID | E | E , Args
Call destination
Callres ::= empty | L = | %lo:ID
Statements
Stmt ::= %s:ID | if ( E ) then Stmt ; | if ( E ) then Stmt else Stmt ;
| return Exp_opt | break ; | continue ; | { Stmt_list }
| while (E ) Stmt | Instr_list
Lists of statements
Stmt_list ::= empty | %S:ID | Stmt Stmt_list
| Type_spec Attrs Decl ; Stmt_list
| Type_spec Attrs Decl = E ; Stmt_list
| Type_spec Attrs Decl = L (Args) ; Stmt_list
List of instructions
Instr_list ::= Instr | %I:ID | Instr Instr_list
</PRE><P>Notes regarding the syntax:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
In the grammar description above non-terminals are written with
uppercase initial</LI><LI CLASS="li-itemize">All of the patterns consist of the <TT>%</TT> character followed by one or
two letters, followed by “:” and an indentifier. For each such
pattern there is a corresponding constructor of the <A HREF="api/Cil.html#TYPEformatArg">Cil.formatArg</A>
type, whose name is the letter ’F’ followed by the same one or two letters as
in the pattern. That constructor is used by the user code to pass a
<A HREF="api/Cil.html#TYPEformatArg">Cil.formatArg</A> actual argument to the interpreted constructor and by
the interpreted deconstructor to return what was matched for a pattern.</LI><LI CLASS="li-itemize">If the pattern name is uppercase, it designates a list of the elements
designated by the corresponding lowercase pattern. E.g. %E designated lists
of expressions (as in the actual arguments of a call).</LI><LI CLASS="li-itemize">The two-letter patterns whose second letter is “o” designate an
optional element. E.g. %eo designates an optional expression (as in the
length of an array). </LI><LI CLASS="li-itemize">Unlike in calls to <TT>printf</TT>, the pattern %g is used for strings. </LI><LI CLASS="li-itemize">The usual precedence and associativity rules as in C apply </LI><LI CLASS="li-itemize">The pattern string can contain newlines and comments, using both the
<TT>/* ... */</TT> style as well as the <TT>//</TT> one. </LI><LI CLASS="li-itemize">When matching a “cast” pattern of the form <TT>( T ) E</TT>, the
deconstructor will match even expressions that do not have the actual cast but
in that case the type is matched against the type of the expression. E.g. the
patters <TT>"(int)%e"</TT> will match any expression of type <TT>int</TT> whether it
has an explicit cast or not. </LI><LI CLASS="li-itemize">The %k pattern is used to construct and deconstruct an integer type of
any kind. </LI><LI CLASS="li-itemize">Notice that the syntax of types and declaration are the same (in order
to simplify the parser). This means that technically you can write a whole
declaration instead of a type in the cast. In this case the name that you
declare is ignored.</LI><LI CLASS="li-itemize">In lists of formal parameters and lists of attributes, an empty list in
the pattern matches any formal parameters or attributes. </LI><LI CLASS="li-itemize">When matching types, uses of named types are unrolled to expose a real
type before matching. </LI><LI CLASS="li-itemize">The order of the attributes is ignored during matching. The the pattern
for a list of attributes contains %A then the resulting <TT>formatArg</TT> will be
bound to <B>all</B> attributes in the list. For example, the pattern <TT>"const
%A"</TT> matches any list of attributes that contains <TT>const</TT> and binds the
corresponding placeholder to the entire list of attributes, including
<TT>const</TT>. </LI><LI CLASS="li-itemize">All instruction-patterns must be terminated by semicolon</LI><LI CLASS="li-itemize">The autoincrement and autodecrement instructions are not supported. Also
not supported are complex expressions, the <TT>&&</TT> and <TT>||</TT> shortcut
operators, and a number of other more complex instructions or statements. In
general, the patterns support only constructs that can be represented directly
in CIL.</LI><LI CLASS="li-itemize">The pattern argument identifiers are not used during deconstruction.
Instead, the result contains a sequence of values in the same order as the
appearance of pattern arguments in the pattern.</LI><LI CLASS="li-itemize">You can mix statements with declarations. For each declaration a new
temporary will be constructed (using a function you provive). You can then
refer to that temporary by name in the rest of the pattern.</LI><LI CLASS="li-itemize">The <TT>%v:</TT> pattern specifier is optional.
</LI></UL><P>The following function are defined in the <TT>Formatcil</TT> module for
constructing and deconstructing:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<A HREF="api/Formatcil.html#VALcExp">Formatcil.cExp</A> constructs <A HREF="api/Cil.html#TYPEexp">Cil.exp</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALcType">Formatcil.cType</A> constructs <A HREF="api/Cil.html#TYPEtyp">Cil.typ</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALcLval">Formatcil.cLval</A> constructs <A HREF="api/Cil.html#TYPElval">Cil.lval</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALcInstr">Formatcil.cInstr</A> constructs <A HREF="api/Cil.html#TYPEinstr">Cil.instr</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALcStmt">Formatcil.cStmt</A> and <A HREF="api/Formatcil.html#VALcStmts">Formatcil.cStmts</A> construct <A HREF="api/Cil.html#TYPEstmt">Cil.stmt</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALdExp">Formatcil.dExp</A> deconstructs <A HREF="api/Cil.html#TYPEexp">Cil.exp</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALdType">Formatcil.dType</A> deconstructs <A HREF="api/Cil.html#TYPEtyp">Cil.typ</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALdLval">Formatcil.dLval</A> deconstructs <A HREF="api/Cil.html#TYPElval">Cil.lval</A>.
</LI><LI CLASS="li-itemize"><A HREF="api/Formatcil.html#VALdInstr">Formatcil.dInstr</A> deconstructs <A HREF="api/Cil.html#TYPElval">Cil.lval</A>.
</LI></UL><P>Below is an example using interpreted constructors. This example generates
the CIL representation of code that scans an array backwards and initializes
every even-index element with an expression:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>Formatcil.cStmts
loc
"int idx = sizeof(array) / sizeof(array[0]) - 1;
while(idx >= 0) {
// Some statements to be run for all the elements of the array
%S:init
if(! (idx & 1))
array[idx] = %e:init_even;
/* Do not forget to decrement the index variable */
idx = idx - 1;
}"
(fun n t -> makeTempVar myfunc ~name:n t)
[ ("array", Fv myarray);
("init", FS [stmt1; stmt2; stmt3]);
("init_even", Fe init_expr_for_even_elements) ]
</FONT></PRE><P>To write the same CIL statement directly in CIL would take much more effort.
Note that the pattern is parsed only once and the result (a function that
takes the arguments and constructs the statement) is memoized. </P><!--TOC subsubsection Performance considerations for interpreted constructors-->
<H4 CLASS="subsubsection"><!--SEC ANCHOR --><A NAME="htoc12">6.2.1</A>  Performance considerations for interpreted constructors</H4><!--SEC END --><P>Parsing the patterns is done with a LALR parser and it takes some time. To
improve performance the constructors and deconstructors memoize the parsed
patterns and will only compile a pattern once. Also all construction and
deconstruction functions can be applied partially to the pattern string to
produce a function that can be later used directly to construct or
deconstruct. This function appears to be about two times slower than if the
construction is done using the CIL constructors (without memoization the
process would be one order of magnitude slower.) However, the convenience of
interpreted constructor might make them a viable choice in many situations
when performance is not paramount (e.g. prototyping).</P><!--TOC subsection Printing and Debugging support-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc13">6.3</A>  Printing and Debugging support</H3><!--SEC END --><P>The Modules <A HREF="api/Pretty.html">Pretty</A> and <A HREF="api/Errormsg.html">Errormsg</A> contain respectively
utilities for pretty printing and reporting errors and provide a convenient
<TT>printf</TT>-like interface. </P><P>Additionally, CIL defines for each major type a pretty-printing function that
you can use in conjunction with the <A HREF="api/Pretty.html">Pretty</A> interface. The
following are some of the pretty-printing functions:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<A HREF="api/Cil.html#VALd_exp">Cil.d_exp</A> - print an expression
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_type">Cil.d_type</A> - print a type
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_lval">Cil.d_lval</A> - print an lvalue
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_global">Cil.d_global</A> - print a global
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_stmt">Cil.d_stmt</A> - print a statment
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_instr">Cil.d_instr</A> - print an instruction
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_init">Cil.d_init</A> - print an initializer
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_attr">Cil.d_attr</A> - print an attribute
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_attrlist">Cil.d_attrlist</A> - print a set of attributes
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_loc">Cil.d_loc</A> - print a location
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_ikind">Cil.d_ikind</A> - print an integer kind
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_fkind">Cil.d_fkind</A> - print a floating point kind
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_const">Cil.d_const</A> - print a constant
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALd_storage">Cil.d_storage</A> - print a storage specifier
</LI></UL><P>You can even customize the pretty-printer by creating instances of
<A HREF="api/Cil.cilPrinter.html#.">Cil.cilPrinter..</A> Typically such an instance extends
<A HREF="api/Cil.html#VALdefaultCilPrinter">Cil.defaultCilPrinter</A>. Once you have a customized pretty-printer you
can use the following printing functions:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<A HREF="api/Cil.html#VALprintExp">Cil.printExp</A> - print an expression
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintType">Cil.printType</A> - print a type
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintLval">Cil.printLval</A> - print an lvalue
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintGlobal">Cil.printGlobal</A> - print a global
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintStmt">Cil.printStmt</A> - print a statment
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintInstr">Cil.printInstr</A> - print an instruction
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintInit">Cil.printInit</A> - print an initializer
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintAttr">Cil.printAttr</A> - print an attribute
</LI><LI CLASS="li-itemize"><A HREF="api/Cil.html#VALprintAttrs">Cil.printAttrs</A> - print a set of attributes
</LI></UL><P>CIL has certain internal consistency invariants. For example, all references
to a global variable must point to the same <TT>varinfo</TT> structure. This
ensures that one can rename the variable by changing the name in the
<TT>varinfo</TT>. These constraints are mentioned in the API documentation. There
is also a consistency checker in file <TT>src/check.ml</TT>. If you suspect that
your transformation is breaking these constraints then you can pass the
<TT>--check</TT> option to cilly and this will ensure that the consistency checker
is run after each transformation. </P><!--TOC subsection Attributes-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc14">6.4</A>  Attributes</H3><!--SEC END --><P><A NAME="sec-attrib"></A></P><!--NAME attributes.html-->
<P>In CIL you can attach attributes to types and to names (variables, functions
and fields). Attributes are represented using the type <A HREF="api/Cil.html#TYPEattribute">Cil.attribute</A>.
An attribute consists of a name and a number of arguments (represented using
the type <A HREF="api/Cil.html#TYPEattrparam">Cil.attrparam</A>). Almost any expression can be used as an
attribute argument. Attributes are stored in lists sorted by the name of the
attribute. To maintain list ordering, use the functions
<A HREF="api/Cil.html#VALtypeAttrs">Cil.typeAttrs</A> to retrieve the attributes of a type and the functions
<A HREF="api/Cil.html#VALaddAttribute">Cil.addAttribute</A> and <A HREF="api/Cil.html#VALaddAttributes">Cil.addAttributes</A> to add attributes.
Alternatively you can use <A HREF="api/Cil.html#VALtypeAddAttributes">Cil.typeAddAttributes</A> to add an attribute to
a type (and return the new type).</P><P>GCC already has extensive support for attributes, and CIL extends this
support to user-defined attributes. A GCC attribute has the syntax:</P><PRE CLASS="verbatim"> gccattribute ::= __attribute__((attribute)) (Note the double parentheses)
</PRE><P>Since GCC and MSVC both support various flavors of each attribute (with or
without leading or trailing _) we first strip ALL leading and trailing _
from the attribute name (but not the identified in [ACons] parameters in
<A HREF="api/Cil.html#TYPEattrparam">Cil.attrparam</A>). When we print attributes, for GCC we add two leading
and two trailing _; for MSVC we add just two leading _.</P><P>There is support in CIL so that you can control the printing of attributes
(see <A HREF="api/Cil.html#VALsetCustomPrintAttribute">Cil.setCustomPrintAttribute</A> and
<A HREF="api/Cil.html#VALsetCustomPrintAttributeScope">Cil.setCustomPrintAttributeScope</A>). This custom-printing support is now
used to print the "const" qualifier as "<TT>const</TT>" and not as
"<TT>__attribute__((const))</TT>".</P><P>The attributes are specified in declarations. This is unfortunate since the C
syntax for declarations is already quite complicated and after writing the
parser and elaborator for declarations I am convinced that few C programmers
understand it completely. Anyway, this seems to be the easiest way to support
attributes. </P><P>Name attributes must be specified at the very end of the declaration, just
before the <TT>=</TT> for the initializer or before the <TT>,</TT> that separates a
declaration in a group of declarations or just before the <TT>;</TT> that
terminates the declaration. A name attribute for a function being defined can
be specified just before the brace that starts the function body.</P><P>For example (in the following examples <TT>A1</TT>,...,<TT>An</TT> are type attributes
and <TT>N</TT> is a name attribute (each of these uses the <TT>__attribute__</TT> syntax):</P><PRE CLASS="verbatim"><FONT COLOR=blue> int x N;
int x N, * y N = 0, z[] N;
extern void exit() N;
int fact(int x) N { ... }
</FONT></PRE><P>Type attributes can be specified along with the type using the following
rules:
</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
The type attributes for a base type (int, float, named type, reference
to struct or union or enum) must be specified immediately following the
type (actually it is Ok to mix attributes with the specification of the
type, in between unsigned and int for example).<P>For example:
</P><PRE CLASS="verbatim"><FONT COLOR=blue> int A1 x N; /* A1 applies to the type int. An example is an attribute
"even" restricting the type int to even values. */
struct foo A1 A2 x; // Both A1 and A2 apply to the struct foo type
</FONT></PRE></LI><LI CLASS="li-enumerate">The type attributes for a pointer type must be specified immediately
after the * symbol.
<PRE CLASS="verbatim"><FONT COLOR=blue> /* A pointer (A1) to an int (A2) */
int A2 * A1 x;
/* A pointer (A1) to a pointer (A2) to a float (A3) */
float A3 * A2 * A1 x;
</FONT></PRE><P>Note: The attributes for base types and for pointer types are a strict
extension of the ANSI C type qualifiers (const, volatile and restrict). In
fact CIL treats these qualifiers as attributes. </P></LI><LI CLASS="li-enumerate">The attributes for a function type or for an array type can be
specified using parenthesized declarators.<P>For example:
</P><PRE CLASS="verbatim"><FONT COLOR=blue> /* A function (A1) from int (A2) to float (A3) */
float A3 (A1 f)(int A2);
/* A pointer (A1) to a function (A2) that returns an int (A3) */
int A3 (A2 * A1 pfun)(void);
/* An array (A1) of int (A2) */
int A2 (A1 x0)[]
/* Array (A1) of pointers (A2) to functions (A3) that take an int (A4) and
* return a pointer (A5) to int (A6) */
int A6 * A5 (A3 * A2 (A1 x1)[5])(int A4);
/* A function (A4) that takes a float (A5) and returns a pointer (A6) to an
* int (A7) */
extern int A7 * A6 (A4 x2)(float A5 x);
/* A function (A1) that takes a int (A2) and that returns a pointer (A3) to
* a function (A4) that takes a float (A5) and returns a pointer (A6) to an
* int (A7) */
int A7 * A6 (A4 * A3 (A1 x3)(int A2 x))(float A5) {
return & x2;
}
</FONT></PRE></LI></OL><P>Note: ANSI C does not allow the specification of type qualifiers for function
and array types, although it allows for the parenthesized declarator. With
just a bit of thought (looking at the first few examples above) I hope that
the placement of attributes for function and array types will seem intuitive.</P><P>This extension is not without problems however. If you want to refer just to
a type (in a cast for example) then you leave the name out. But this leads to
strange conflicts due to the parentheses that we introduce to scope the
attributes. Take for example the type of x0 from above. It should be written
as: </P><PRE CLASS="verbatim"><FONT COLOR=blue> int A2 (A1 )[]
</FONT></PRE><P>But this will lead most C parsers into deep confusion because the parentheses
around A1 will be confused for parentheses of a function designator. To push
this problem around (I don’t know a solution) whenever we are about to print a
parenthesized declarator with no name but with attributes, we comment out the
attributes so you can see them (for whatever is worth) without confusing the
compiler. For example, here is how we would print the above type:</P><PRE CLASS="verbatim"><FONT COLOR=blue> int A2 /*(A1 )*/[]
</FONT></PRE><!--TOC paragraph Handling of predefined GCC attributes-->
<H5 CLASS="paragraph"><!--SEC ANCHOR -->Handling of predefined GCC attributes</H5><!--SEC END --><P>GCC already supports attributes in a lot of places in declarations. The only
place where we support attributes and GCC does not is right before the { that
starts a function body. </P><P>GCC classifies its attributes in attributes for functions, for variables and
for types, although the latter category is only usable in definition of struct
or union types and is not nearly as powerful as the CIL type attributes. We
have made an effort to reclassify GCC attributes as name and type attributes
(they only apply for function types). Here is what we came up with:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
GCC name attributes:<P>section, constructor, destructor, unused, weak, no_instrument_function,
noreturn, alias, no_check_memory_usage, dllinport, dllexport, exception,
model</P><P>Note: the "noreturn" attribute would be more appropriately qualified as a
function type attribute. But we classify it as a name attribute to make
it easier to support a similarly named MSVC attribute. </P></LI><LI CLASS="li-itemize">GCC function type attributes:<P>fconst (printed as "const"), format, regparm, stdcall,
cdecl, longcall</P><P>I was not able to completely decipher the position in which these attributes
must go. So, the CIL elaborator knows these names and applies the following
rules:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
All of the name attributes that appear in the specifier part (i.e. at
the beginning) of a declaration are associated with all declared names. </LI><LI CLASS="li-itemize">All of the name attributes that appear at the end of a declarator are
associated with the particular name being declared.</LI><LI CLASS="li-itemize">More complicated is the handling of the function type attributes, since
there can be more than one function in a single declaration (a function
returning a pointer to a function). Lacking any real understanding of how
GCC handles this, I attach the function type attribute to the "nearest"
function. This means that if a pointer to a function is "nearby" the
attribute will be correctly associated with the function. In truth I pray
that nobody uses declarations as that of x3 above.
</LI></UL></LI></UL><!--TOC paragraph Handling of predefined MSVC attributes-->
<H5 CLASS="paragraph"><!--SEC ANCHOR -->Handling of predefined MSVC attributes</H5><!--SEC END --><P>MSVC has two kinds of attributes, declaration modifiers to be printed before
the storage specifier using the notation "<TT>__declspec(...)</TT>" and a few
function type attributes, printed almost as our CIL function type
attributes. </P><P>The following are the name attributes that are printed using
<TT>__declspec</TT> right before the storage designator of the declaration:
thread, naked, dllimport, dllexport, noreturn</P><P>The following are the function type attributes supported by MSVC:
fastcall, cdecl, stdcall</P><P>It is not worth going into the obscure details of where MSVC accepts these
type attributes. The parser thinks it knows these details and it pulls
these attributes from wherever they might be placed. The important thing
is that MSVC will accept if we print them according to the rules of the CIL
attributes ! </P><!--TOC section The CIL Driver-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc15">7</A>  The CIL Driver</H2><!--SEC END --><P><A NAME="sec-driver"></A></P><P>We have packaged CIL as an application <TT>cilly</TT> that contains certain
example modules, such as <TT>logwrites.ml</TT> (a module
that instruments code to print the addresses of memory locations being
written). Normally, you write another module like that, add command-line
options and an invocation of your module in <TT>src/main.ml</TT>. Once you compile
CIL you will obtain the file <TT>obj/cilly.asm.exe</TT>. </P><P>We wrote a driver for this executable that makes it easy to invoke your
analysis on existing C code with very little manual intervention. This driver
is <TT>bin/cilly</TT> and is quite powerful. Note that the <TT>cilly</TT> script
is configured during installation with the path where CIL resides. This means
that you can move it to any place you want. </P><P>A simple use of the driver is:
</P><PRE CLASS="verbatim">bin/cilly --save-temps -D HAPPY_MOOD -I myincludes hello.c -o hello
</PRE><P><FONT COLOR=blue>--save-temps</FONT> tells CIL to save the resulting output files in the
current directory. Otherwise, they’ll be put in <TT>/tmp</TT> and deleted
automatically. Not that this is the only CIL-specific flag in the
list – the other flags use <TT>gcc</TT>’s syntax.</P><P>This performs the following actions:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
preprocessing using the -D and -I arguments with the resulting
file left in <TT>hello.i</TT>,
</LI><LI CLASS="li-itemize">the invocation of the <TT>cilly.asm</TT> application which parses <TT>hello.i</TT>
converts it to CIL and the pretty-prints it to <TT>hello.cil.c</TT>
</LI><LI CLASS="li-itemize">another round of preprocessing with the result placed in <TT>hello.cil.i</TT>
</LI><LI CLASS="li-itemize">the true compilation with the result in <TT>hello.cil.o</TT>
</LI><LI CLASS="li-itemize">a linking phase with the result in <TT>hello</TT>
</LI></UL><P>Note that <TT>cilly</TT> behaves like the <TT>gcc</TT> compiler. This makes it
easy to use it with existing <TT>Makefiles</TT>:
</P><PRE CLASS="verbatim">make CC="bin/cilly" LD="bin/cilly"
</PRE><P><TT>cilly</TT> can also behave as the Microsoft Visual C compiler, if the first
argument is <TT>--mode=MSVC</TT>:
</P><PRE CLASS="verbatim">bin/cilly --mode=MSVC /D HAPPY_MOOD /I myincludes hello.c /Fe hello.exe
</PRE><P>(This in turn will pass a <TT>--MSVC</TT> flag to the underlying <TT>cilly.asm</TT>
process which will make it understand the Microsoft Visual C extensions)</P><P><TT>cilly</TT> can also behave as the archiver <TT>ar</TT>, if it is passed an
argument <TT>--mode=AR</TT>. Note that only the <TT>cr</TT> mode is supported
(create a new archive and replace all files in there). Therefore the
previous version of the archive is lost. You will also need to remove
any other commands that operate on the generated library
(e.g. <TT>ranlib</TT>, <TT>lorder</TT>), as the .a file is no longer an actual
binary library.</P><P>Furthermore, <TT>cilly</TT> allows you to pass some arguments on to the
underlying <TT>cilly.asm</TT> process. As a general rule all arguments that start
with <TT>--</TT> and that <TT>cilly</TT> itself does not process, are passed on. For
example,
</P><PRE CLASS="verbatim">bin/cilly --dologwrites -D HAPPY_MOOD -I myincludes hello.c -o hello.exe
</PRE><P>will produce a file <TT>hello.cil.c</TT> that prints all the memory addresses
written by the application. </P><P>The most powerful feature of <TT>cilly</TT> is that it can collect all the
sources in your project, merge them into one file and then apply CIL. This
makes it a breeze to do whole-program analysis and transformation. All you
have to do is to pass the <TT>--merge</TT> flag to <TT>cilly</TT>:
</P><PRE CLASS="verbatim">make CC="bin/cilly --save-temps --dologwrites --merge"
</PRE><P>You can even leave some files untouched:
</P><PRE CLASS="verbatim">make CC="bin/cilly --save-temps --dologwrites --merge --leavealone=foo --leavealone=bar"
</PRE><P>This will merge all the files except those with the basename <TT>foo</TT> and
<TT>bar</TT>. Those files will be compiled as usual and then linked in at the very
end. </P><P>The sequence of actions performed by <TT>cilly</TT> depends on whether merging
is turned on or not:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
If merging is off
<OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
For every file <TT>file.c</TT> to compile
<OL CLASS="enumerate" type=a><LI CLASS="li-enumerate">
Preprocess the file with the given arguments to
produce <TT>file.i</TT>
</LI><LI CLASS="li-enumerate">Invoke <TT>cilly.asm</TT> to produce a <TT>file.cil.c</TT>
</LI><LI CLASS="li-enumerate">Preprocess to <TT>file.cil.i</TT>
</LI><LI CLASS="li-enumerate">Invoke the underlying compiler to produce <TT>file.cil.o</TT>
</LI></OL>
</LI><LI CLASS="li-enumerate">Link the resulting objects
</LI></OL>
</LI><LI CLASS="li-itemize">If merging is on
<OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
For every file <TT>file.c</TT> to compile
<OL CLASS="enumerate" type=a><LI CLASS="li-enumerate">
Preprocess the file with the given arguments to
produce <TT>file.i</TT>
</LI><LI CLASS="li-enumerate">Save the preprocessed source as <TT>file.o</TT>
</LI></OL>
</LI><LI CLASS="li-enumerate">When linking executable <TT>hello.exe</TT>, look at every object
file that must be linked and see if it actually
contains preprocessed source. Pass all those files to a
special merging application (described in
Section <A HREF="#sec-merger">13</A>) to produce <TT>hello.exe_comb.c</TT>
</LI><LI CLASS="li-enumerate">Invoke <TT>cilly.asm</TT> to produce a <TT>hello.exe_comb.cil.c</TT>
</LI><LI CLASS="li-enumerate">Preprocess to <TT>hello.exe_comb.cil.i</TT>
</LI><LI CLASS="li-enumerate">Invoke the underlying compiler to produce <TT>hello.exe_comb.cil.o</TT>
</LI><LI CLASS="li-enumerate">Invoke the actual linker to produce <TT>hello.exe</TT>
</LI></OL>
</LI></UL><P>Note that files that you specify with <TT>--leavealone</TT> are not merged and
never presented to CIL. They are compiled as usual and then are linked in at
the end. </P><P>And a final feature of <TT>cilly</TT> is that it can substitute copies of the
system’s include files:</P><PRE CLASS="verbatim">make CC="bin/cilly --includedir=myinclude"
</PRE><P>This will force the preprocessor to use the file <TT>myinclude/xxx/stdio.h</TT>
(if it exists) whenever it encounters <TT>#include <stdio.h></TT>. The <TT>xxx</TT> is
a string that identifies the compiler version you are using. This modified
include files should be produced with the patcher script (see
Section <A HREF="#sec-patcher">14</A>).</P><!--TOC subsection <TT>cilly</TT> Options-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc16">7.1</A>  <TT>cilly</TT> Options</H3><!--SEC END --><P>Among the options for the <TT>cilly</TT> you can put anything that can normally
go in the command line of the compiler that <TT>cilly</TT> is impersonating.
<TT>cilly</TT> will do its best to pass those options along to the appropriate
subprocess. In addition, the following options are supported (a complete and
up-to-date list can always be obtained by running <TT>cilly --help</TT>):</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>--gcc=command</TT> Tell <TT>cilly</TT> to use <TT>command</TT> to invoke
gcc, e.g. <TT>--gcc=arm-elf-gcc</TT> to use a cross-compiler. See also
the <TT>--envmachine</TT> option below that tells CIL to assume a
different machine model.
</LI><LI CLASS="li-itemize"><TT>--mode=mode</TT> This must be the first argument if present. It makes
<TT>cilly</TT> behave as a given compiled. The following modes are recognized:
<UL CLASS="itemize"><LI CLASS="li-itemize">
GNUCC - the GNU C Compiler. This is the default.
</LI><LI CLASS="li-itemize">MSVC - the Microsoft Visual C compiler. Of course, you should
pass only MSVC valid options in this case.
</LI><LI CLASS="li-itemize">AR - the archiver <TT>ar</TT>. Only the mode <TT>cr</TT> is supported and
the original version of the archive is lost.
</LI></UL>
</LI><LI CLASS="li-itemize"><TT>--help</TT> Prints a list of the options supported.
</LI><LI CLASS="li-itemize"><TT>--verbose</TT> Prints lots of messages about what is going on.
</LI><LI CLASS="li-itemize"><TT>--stages</TT> Less than <TT>--verbose</TT> but lets you see what <TT>cilly</TT>
is doing.
</LI><LI CLASS="li-itemize"><TT>--merge</TT> This tells <TT>cilly</TT> to first attempt to collect into one
source file all of the sources that make your application, and then to apply
<TT>cilly.asm</TT> on the resulting source. The sequence of actions in this case is
described above and the merger itself is described in Section <A HREF="#sec-merger">13</A>.</LI><LI CLASS="li-itemize"><TT>--leavealone=xxx</TT>. Do not merge and do not present to CIL the files
whose basename is "xxx". These files are compiled as usual and linked in at
the end.
</LI><LI CLASS="li-itemize"><TT>--includedir=xxx</TT>. Override the include files with those in the given
directory. The given directory is the same name that was given an an argument
to the patcher (see Section <A HREF="#sec-patcher">14</A>). In particular this means that
that directory contains subdirectories named based on the current compiler
version. The patcher creates those directories.
</LI><LI CLASS="li-itemize"><TT>--usecabs</TT>. Do not CIL, but instead just parse the source and print
its AST out. This should looked like the preprocessed file. This is useful
when you suspect that the conversion to CIL phase changes the meaning of the
program.
</LI><LI CLASS="li-itemize"><TT>--save-temps=xxx</TT>. Temporary files are preserved in the xxx
directory. For example, the output of CIL will be put in a file
named <TT>*.cil.c</TT>.
</LI><LI CLASS="li-itemize"><TT>--save-temps</TT>. Temporay files are preserved in the current directory.
</LI></UL><!--TOC subsection <TT>cilly.asm</TT> Options-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc17">7.2</A>  <TT>cilly.asm</TT> Options</H3><!--SEC END --><P>
<A NAME="sec-cilly-asm-options"></A></P><P>All of the options that start with <TT>--</TT> and are not understood by
<TT>cilly</TT> are passed on to <TT>cilly.asm</TT>. <TT>cilly</TT> also passes along to
<TT>cilly.asm</TT> flags such as <TT>--MSVC</TT> that both need to know
about. The following options are supported. Many of these flags also
have corresponding “<TT>--no</TT>*” versions if you need to go back to
the default, as in “<TT>--nowarnall</TT>”.</P><P>       <B>General Options:</B>
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>--version</TT> output version information and exit
</LI><LI CLASS="li-itemize"><TT>--verbose</TT> Print lots of random stuff. This is passed on from cilly
</LI><LI CLASS="li-itemize"><TT>--warnall</TT> Show all warnings.
</LI><LI CLASS="li-itemize"><TT>--debug=xxx</TT> turns on debugging flag xxx
</LI><LI CLASS="li-itemize"><TT>--nodebug=xxx</TT> turns off debugging flag xxx
</LI><LI CLASS="li-itemize"><TT>--flush</TT> Flush the output streams often (aids debugging).
</LI><LI CLASS="li-itemize"><TT>--check</TT> Run a consistency check over the CIL after every operation.
</LI><LI CLASS="li-itemize"><TT>--strictcheck</TT> Same as <TT>--check</TT>, but it treats
consistency problems as errors instead of warnings.
</LI><LI CLASS="li-itemize"><TT>--nocheck</TT> turns off consistency checking of CIL.
</LI><LI CLASS="li-itemize"><TT>--noPrintLn</TT> Don’t output #line directives in the output.
</LI><LI CLASS="li-itemize"><TT>--commPrintLn</TT> Print #line directives in the output, but
put them in comments.
</LI><LI CLASS="li-itemize"><TT>--commPrintLnSparse</TT> Like <TT>--commPrintLn</TT> but print only new
line numbers.
</LI><LI CLASS="li-itemize"><TT>--log=xxx</TT> Set the name of the log file. By default stderr is used
</LI><LI CLASS="li-itemize"><TT>--MSVC</TT> Enable MSVC compatibility. Default is GNU.
</LI><LI CLASS="li-itemize"><TT>--ignore-merge-conflicts</TT> ignore merging conflicts.
</LI><LI CLASS="li-itemize"><TT>--extrafiles=filename</TT>: the name of a file that contains
a list of additional files to process, separated by whitespace.
</LI><LI CLASS="li-itemize"><TT>--stats</TT> Print statistics about the running time of the
parser, conversion to CIL, etc. Also prints memory-usage
statistics. You can time parts of your own code as well. Calling
(<TT>Stats.time ‘‘label’’ func arg</TT>) will evaluate <TT>(func arg)</TT>
and remember how long this takes. If you call <TT>Stats.time</TT>
repeatedly with the same label, CIL will report the aggregate
time.<P>If available, CIL uses the x86 performance counters for these
stats. This is very precise, but results in “wall-clock time.”
To report only user-mode time, find the call to <TT>Stats.reset</TT> in
<TT>main.ml</TT>, and change it to <TT>Stats.reset Stats.SoftwareTimer</TT>.</P></LI><LI CLASS="li-itemize"><TT>--envmachine</TT>. Use machine model specified in
<TT>CIL_MACHINE</TT> environment variable, rather than the one
compiled into CIL. Note that you should not pass gcc’s 32/64-bit
<TT>-m32</TT> and <TT>-m64</TT> options to <TT>cilly</TT> if you use
<TT>--envmachine</TT> (they use <TT>--envmachine</TT> under the hood).
See Section <A HREF="#sec-cilmachine">7.4</A> for a description of <TT>CIL_MACHINE</TT>’s
format.<P><B>Lowering Options</B>
</P></LI><LI CLASS="li-itemize"><TT>--noLowerConstants</TT> do not lower constant expressions.
</LI><LI CLASS="li-itemize"><TT>--noInsertImplicitCasts</TT> do not insert implicit casts.
</LI><LI CLASS="li-itemize"><TT>--forceRLArgEval</TT> Forces right to left evaluation of function arguments.
</LI><LI CLASS="li-itemize"><TT>--disallowDuplication</TT> Prevent small chunks of code from being duplicated.
</LI><LI CLASS="li-itemize"><TT>--keepunused</TT> Do not remove the unused variables and types.
</LI><LI CLASS="li-itemize"><TT>--rmUnusedInlines</TT> Delete any unused inline functions. This is the default in MSVC mode.<P><B>Output Options:</B>
</P></LI><LI CLASS="li-itemize"><TT>--printCilAsIs</TT> Do not try to simplify the CIL when
printing. Without this flag, CIL will attempt to produce prettier
output by e.g. changing <TT>while(1)</TT> into more meaningful loops.
</LI><LI CLASS="li-itemize"><TT>--noWrap</TT> do not wrap long lines when printing
</LI><LI CLASS="li-itemize"><TT>--out=xxx</TT> the name of the output CIL file. <TT>cilly</TT>
sets this for you.
</LI><LI CLASS="li-itemize"><TT>--mergedout=xxx</TT> specify the name of the merged file
</LI><LI CLASS="li-itemize"><TT>--cabsonly=xxx</TT> CABS output file name
<P><B>Selected features.</B> See Section <A HREF="#sec-Extension">8</A> for more information.
</P></LI><LI CLASS="li-itemize"><TT>--dologcalls</TT>. Insert code in the processed source to print the name of
functions as are called. Implemented in <TT>src/ext/logcalls.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--dologwrites</TT>. Insert code in the processed source to print the
address of all memory writes. Implemented in <TT>src/ext/logwrites.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--dooneRet</TT>. Make each function have at most one ’return’.
Implemented in <TT>src/ext/oneret.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--dostackGuard</TT>. Instrument function calls and returns to
maintain a separate stack for return addresses. Implemeted in
<TT>src/ext/heapify.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--domakeCFG</TT>. Make the program look more like a CFG. Implemented
in <TT>src/cil.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--dopartial</TT>. Do interprocedural partial evaluation and
constant folding. Implemented in <TT>src/ext/partial.ml</TT>.
</LI><LI CLASS="li-itemize"><TT>--dosimpleMem</TT>. Simplify all memory expressions. Implemented in
<TT>src/ext/simplemem.ml</TT>. <P>For an up-to-date list of available options, run <TT>cilly.asm --help</TT>. </P></LI></UL><!--TOC subsection Internal Options-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc18">7.3</A>  Internal Options</H3><!--SEC END --><P>
<A NAME="sec-cilly-internal-options"></A></P><P>All of the <TT>cilly.asm</TT> options described above can be set
programmatically – see <TT>src/ciloptions.ml</TT> or the individual
extensions to see how. Some options should be set before parsing to
be effective. </P><P>Additionally, a few CIL options have no command-line flag and can only
be set programmatically. These options may be useful for certain
analyses:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>Cabs2cil.doCollapseCallCast</TT>:This is false by default. Set
to true to replicate the behavior of CIL 1.3.5 and earlier.<P>When false, all casts in the program are made explicit using the
<TT>CastE</TT> expression. Accordingly, the destination of a Call
instruction will always have the same type as the function’s return
type.</P><P>If true, the destination type of a Call may differ from the return type, so
there is an implicit cast. This is useful for analyses involving
<TT>malloc</TT>. Without this option, CIL converts “<TT>T* x = malloc(n);</TT>”
into “<TT>void* tmp = malloc(n); T* x = (T*)tmp;</TT>”. If you don’t
need all casts to be made explicit, you can set
<TT>Cabs2cil.doCollapseCallCast</TT> to true so that CIL won’t insert a
temporary and you can more easily determine the allocation type from
calls to <TT>malloc</TT>.</P></LI></UL><!--TOC subsection Specifying a machine model-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc19">7.4</A>  Specifying a machine model</H3><!--SEC END --><P>
<A NAME="sec-cilmachine"></A></P><P>The <TT>--envmachine</TT> option tells CIL to get its machine model from
the <TT>CIL_MACHINE</TT> environment variable, rather than use the model
compiled in to CIL itself. This is necessary when using CIL as part of
a cross-compilation setup, and to handle gcc’s <TT>-m32</TT> and <TT>-m64</TT>
which select between a 32-bit and 64-bit machine model.</P><P><TT>CIL_MACHINE</TT> is a space-separated list of key=value pairs. Unknown
keys are ignored. The following keys must be defined:
</P><DIV CLASS="center">
<TABLE BORDER=1 CELLSPACING=0 CELLPADDING=1><TR><TD ALIGN=left NOWRAP> Key</TD><TD ALIGN=left NOWRAP>Value</TD></TR>
<TR><TD ALIGN=left NOWRAP> bool</TD><TD ALIGN=left NOWRAP>sizeof(_Bool),alignof(_Bool)</TD></TR>
<TR><TD ALIGN=left NOWRAP> short</TD><TD ALIGN=left NOWRAP>sizeof(short),alignof(short)</TD></TR>
<TR><TD ALIGN=left NOWRAP> int</TD><TD ALIGN=left NOWRAP>sizeof(int),alignof(int)</TD></TR>
<TR><TD ALIGN=left NOWRAP> long</TD><TD ALIGN=left NOWRAP>sizeof(long),alignof(long)</TD></TR>
<TR><TD ALIGN=left NOWRAP> long_long</TD><TD ALIGN=left NOWRAP>sizeof(long long),alignof(long long)</TD></TR>
<TR><TD ALIGN=left NOWRAP> float</TD><TD ALIGN=left NOWRAP>sizeof(float),alignof(float)</TD></TR>
<TR><TD ALIGN=left NOWRAP> double</TD><TD ALIGN=left NOWRAP>sizeof(double),alignof(double)</TD></TR>
<TR><TD ALIGN=left NOWRAP> long_double</TD><TD ALIGN=left NOWRAP>sizeof(long double),alignof(long double)</TD></TR>
<TR><TD ALIGN=left NOWRAP> pointer</TD><TD ALIGN=left NOWRAP>sizeof(all pointers),alignof(all pointers)</TD></TR>
<TR><TD ALIGN=left NOWRAP> enum</TD><TD ALIGN=left NOWRAP>sizeof(all enums),alignof(all enums)</TD></TR>
<TR><TD ALIGN=left NOWRAP> fun</TD><TD ALIGN=left NOWRAP>sizeof(all fn ptrs),alignof(all fn ptrs)</TD></TR>
<TR><TD ALIGN=left NOWRAP> alignof_string</TD><TD ALIGN=left NOWRAP>alignof(all string constants)</TD></TR>
<TR><TD ALIGN=left NOWRAP> max_alignment</TD><TD ALIGN=left NOWRAP>maximum alignment for any type</TD></TR>
<TR><TD ALIGN=left NOWRAP> size_t</TD><TD ALIGN=left NOWRAP>the definition of size_t</TD></TR>
<TR><TD ALIGN=left NOWRAP> wchar_t</TD><TD ALIGN=left NOWRAP>the definition of wchar_t</TD></TR>
<TR><TD ALIGN=left NOWRAP> char_signed</TD><TD ALIGN=left NOWRAP>true if char is signed</TD></TR>
<TR><TD ALIGN=left NOWRAP> big_endian</TD><TD ALIGN=left NOWRAP>true for big-endian machine models</TD></TR>
<TR><TD ALIGN=left NOWRAP> const_string_literals</TD><TD ALIGN=left NOWRAP>true if string constants are const</TD></TR>
<TR><TD ALIGN=left NOWRAP> __thread_is_keyword</TD><TD ALIGN=left NOWRAP>true if <TT>__thread</TT> is a keyword</TD></TR>
<TR><TD ALIGN=left NOWRAP> __builtin_va_list</TD><TD ALIGN=left NOWRAP>true if <TT>__builtin_va_list</TT> is a builtin type</TD></TR>
<TR><TD ALIGN=left NOWRAP> underscore_name</TD><TD ALIGN=left NOWRAP>true if generated symbols preceded by _</TD></TR>
</TABLE>
</DIV><P>Some notes:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
Values cannot contain spaces as spaces separate key/value pairs.
</LI><LI CLASS="li-itemize">The value of size_t is the text for the type defined as
size_t, e.g. <TT>unsigned long</TT>. As spaces are not allowed in values,
CIL will replace underscores by spaces: <TT>size_t=unsigned_long</TT>.
</LI><LI CLASS="li-itemize">sizeof(t) and alignof(t) are respectively the size and alignment of
type t.
</LI><LI CLASS="li-itemize">The boolean-valued keys expect <TT>true</TT> for true, and <TT>false</TT>
for false.
</LI><LI CLASS="li-itemize">The <TT>src/machdep-ml.c</TT> program will print a <TT>CIL_MACHINE</TT> value
when run with the <TT>--env</TT> option.
</LI></UL><P>As an example, here’s the <TT>CIL_MACHINE</TT> value for 64-bit targets on
an x86-based Mac OS X machine:
</P><PRE CLASS="verbatim"> short=2,2 int=4,4 long=8,8 long_long=8,8 pointer=8,8 enum=4,4
float=4,4 double=8,8 long_double=16,16 void=1 bool=1,1 fun=1,1
alignof_string=1 max_alignment=16 size_t=unsigned_long
wchar_t=int char_signed=true const_string_literals=true
big_endian=false __thread_is_keyword=true
__builtin_va_list=true underscore_name=true
</PRE><!--TOC section Library of CIL Modules-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc20">8</A>  Library of CIL Modules</H2><!--SEC END --><P> <A NAME="sec-Extension"></A></P><!--NAME ext.html-->
<P>We are developing a suite of modules that use CIL for program analyses and
transformations that we have found useful. You can use these modules directly
on your code, or generally as inspiration for writing similar modules. A
particularly big and complex application written on top of CIL is CCured
(<A HREF="../ccured/index.html"><TT>../ccured/index.html</TT></A>).</P><!--TOC subsection Control-Flow Graphs-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc21">8.1</A>  Control-Flow Graphs</H3><!--SEC END --><P> <A NAME="sec-cfg"></A></P><P>The <A HREF="api/Cil.html#TYPEstmt">Cil.stmt</A> datatype includes fields for intraprocedural
control-flow information: the predecessor and successor statements of
the current statement. This information is not computed by default.
If you want to use the control-flow graph, or any of the extensions in
this section that require it, you have to explicitly ask CIL to
compute the CFG using one of these two methods:</P><!--TOC subsubsection The CFG module (new in CIL 1.3.5)-->
<H4 CLASS="subsubsection"><!--SEC ANCHOR --><A NAME="htoc22">8.1.1</A>  The CFG module (new in CIL 1.3.5)</H4><!--SEC END --><P>The best way to compute the CFG is with the CFG module. Just invoke
<A HREF="api/Cfg.html#VALcomputeFileCFG">Cfg.computeFileCFG</A> on your file. The <A HREF="api/Cfg.html">Cfg</A> API
describes the rest of actions you can take with this module, including
computing the CFG for one function at a time, or printing the CFG in
<TT>dot</TT> form.</P><!--TOC subsubsection Simplified control flow-->
<H4 CLASS="subsubsection"><!--SEC ANCHOR --><A NAME="htoc23">8.1.2</A>  Simplified control flow</H4><!--SEC END --><P>CIL can reduce high-level C control-flow constructs like <TT>switch</TT> and
<TT>continue</TT> to lower-level <TT>goto</TT>s. This completely eliminates some
possible classes of statements from the program and may make the result
easier to analyze (e.g., it simplifies data-flow analysis).</P><P>You can invoke this transformation on the command line with
<TT>--domakeCFG</TT> or programatically with <A HREF="api/Cil.html#VALprepareCFG">Cil.prepareCFG</A>.
After calling Cil.prepareCFG, you can use <A HREF="api/Cil.html#VALcomputeCFGInfo">Cil.computeCFGInfo</A>
to compute the CFG information and find the successor and predecessor
of each statement.</P><P>For a concrete example, you can see how <TT>cilly --domakeCFG</TT>
transforms the following code (note the fall-through in case 1): </P><PRE CLASS="verbatim"><FONT COLOR=blue> int foo (int predicate) {
int x = 0;
switch (predicate) {
case 0: return 111;
case 1: x = x + 1;
case 2: return (x+3);
case 3: break;
default: return 222;
}
return 333;
}
</FONT></PRE><P>
See the <A HREF="examples/ex23.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Data flow analysis framework-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc24">8.2</A>  Data flow analysis framework</H3><!--SEC END --><P>The <A HREF="api/Dataflow.html">Dataflow</A> module (click for the ocamldoc) contains a
parameterized framework for forward and backward data flow
analyses. You provide the transfer functions and this module does the
analysis. You must compute control-flow information (Section <A HREF="#sec-cfg">8.1</A>)
before invoking the Dataflow module.</P><!--TOC subsection Inliner-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc25">8.3</A>  Inliner</H3><!--SEC END --><P>The file ext/inliner.ml contains a function inliner. </P><!--TOC subsection Dominators-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc26">8.4</A>  Dominators</H3><!--SEC END --><P>The module <A HREF="api/Dominators.html">Dominators</A> contains the computation of immediate
dominators. It uses the <A HREF="api/Dataflow.html">Dataflow</A> module. </P><!--TOC subsection Points-to Analysis-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc27">8.5</A>  Points-to Analysis</H3><!--SEC END --><P>The module <TT>ptranal.ml</TT> contains two interprocedural points-to
analyses for CIL: <TT>Olf</TT> and <TT>Golf</TT>. <TT>Olf</TT> is the default.
(Switching from <TT>olf.ml</TT> to <TT>golf.ml</TT> requires a change in
<TT>Ptranal</TT> and a recompiling <TT>cilly</TT>.)</P><P>The analyses have the following characteristics:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
Not based on C types (inferred pointer relationships are sound
despite most kinds of C casts)
</LI><LI CLASS="li-itemize">One level of subtyping
</LI><LI CLASS="li-itemize">One level of context sensitivity (Golf only)
</LI><LI CLASS="li-itemize">Monomorphic type structures
</LI><LI CLASS="li-itemize">Field insensitive (fields of structs are conflated)
</LI><LI CLASS="li-itemize">Demand-driven (points-to queries are solved on demand)
</LI><LI CLASS="li-itemize">Handle function pointers
</LI></UL><P>The analysis itself is factored into two components: <TT>Ptranal</TT>,
which walks over the CIL file and generates constraints, and <TT>Olf</TT>
or <TT>Golf</TT>, which solve the constraints. The analysis is invoked
with the function <TT>Ptranal.analyze_file: Cil.file ->
unit</TT>. This function builds the points-to graph for the CIL file
and stores it internally. There is currently no facility for clearing
internal state, so <TT>Ptranal.analyze_file</TT> should only be called
once.</P><P>The constructed points-to graph supports several kinds of queries,
including alias queries (may two expressions be aliased?) and
points-to queries (to what set of locations may an expression point?).</P><P>The main interface with the alias analysis is as follows:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>Ptranal.may_alias: Cil.exp -> Cil.exp -> bool</TT>. If
<TT>true</TT>, the two expressions may have the same value.
</LI><LI CLASS="li-itemize"><TT>Ptranal.resolve_lval: Cil.lval -> (Cil.varinfo
list)</TT>. Returns the list of variables to which the given
left-hand value may point.
</LI><LI CLASS="li-itemize"><TT>Ptranal.resolve_exp: Cil.exp -> (Cil.varinfo list)</TT>.
Returns the list of variables to which the given expression may
point.
</LI><LI CLASS="li-itemize"><TT>Ptranal.resolve_funptr: Cil.exp -> (Cil.fundec
list)</TT>. Returns the list of functions to which the given
expression may point.
</LI></UL><P>The precision of the analysis can be customized by changing the values
of several flags:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>Ptranal.no_sub: bool ref</TT>.
If <TT>true</TT>, subtyping is disabled. Associated commandline option:
<B>--ptr_unify</B>.
</LI><LI CLASS="li-itemize"><TT>Ptranal.analyze_mono: bool ref</TT>.
(Golf only) If <TT>true</TT>, context sensitivity is disabled and the
analysis is effectively monomorphic. Commandline option:
<B>--ptr_mono</B>.
</LI><LI CLASS="li-itemize"><TT>Ptranal.smart_aliases: bool ref</TT>.
(Golf only) If <TT>true</TT>, “smart” disambiguation of aliases is
enabled. Otherwise, aliases are computed by intersecting points-to
sets. This is an experimental feature.
</LI><LI CLASS="li-itemize"><TT>Ptranal.model_strings: bool ref</TT>.
Make the alias analysis model string constants by treating them as
pointers to chars. Commandline option: <B>--ptr_model_strings</B>
</LI><LI CLASS="li-itemize"><TT>Ptranal.conservative_undefineds: bool ref</TT>.
Make the most pessimistic assumptions about globals if an undefined
function is present. Such a function can write to every global
variable. Commandline option: <B>--ptr_conservative</B>
</LI></UL><P>In practice, the best precision/efficiency tradeoff is achieved by
setting
</P><PRE CLASS="verbatim"> Ptranal.no_sub = false
Ptranal.analyze_mono = true
Ptranal.smart_aliases = false
</PRE><P>These are the
default values of the flags.</P><P>There are also a few flags that can be used to inspect or serialize
the results of the analysis.
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>Ptranal.debug_may_aliases</TT>.
Print the may-alias relationship of each pair of expressions in the
program. Commandline option: <B>--ptr_may_aliases</B>.
</LI><LI CLASS="li-itemize"><TT>Ptranal.print_constraints: bool ref</TT>.
If <TT>true</TT>, the analysis will print each constraint as it is
generated.
</LI><LI CLASS="li-itemize"><TT>Ptranal.print_types: bool ref</TT>.
If <TT>true</TT>, the analysis will print the inferred type of each
variable in the program.<P>If <TT>Ptranal.analyze_mono</TT> and <TT>Ptranal.no_sub</TT> are both
<TT>true</TT>, this output is sufficient to reconstruct the points-to
graph. One nice feature is that there is a pretty printer for
recursive types, so the print routine does not loop.
</P></LI><LI CLASS="li-itemize"><TT>Ptranal.compute_results: bool ref</TT>.
If <TT>true</TT>, the analysis will print out the points-to set of each
variable in the program. This will essentially serialize the
points-to graph.
</LI></UL><!--TOC subsection StackGuard-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc28">8.6</A>  StackGuard</H3><!--SEC END --><P>The module <TT>heapify.ml</TT> contains a transformation similar to the one
described in “StackGuard: Automatic Adaptive Detection and Prevention of
Buffer-Overflow Attacks”, <EM>Proceedings of the 7th USENIX Security
Conference</EM>. In essence it modifies the program to maintain a separate
stack for return addresses. Even if a buffer overrun attack occurs the
actual correct return address will be taken from the special stack. </P><P>Although it does work, this CIL module is provided mainly as an example of
how to perform a simple source-to-source program analysis and
transformation. As an optimization only functions that contain a dangerous
local array make use of the special return address stack. </P><P>For a concrete example, you can see how <TT>cilly --dostackGuard</TT>
transforms the following dangerous code: </P><PRE CLASS="verbatim"><FONT COLOR=blue> int dangerous() {
char array[10];
scanf("%s",array); // possible buffer overrun!
}
int main () {
return dangerous();
}
</FONT></PRE><P>
See the <A HREF="examples/ex24.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Heapify-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc29">8.7</A>  Heapify</H3><!--SEC END --><P>The module <TT>heapify.ml</TT> also contains a transformation that moves all
dangerous local arrays to the heap. This also prevents a number of buffer
overruns. </P><P>For a concrete example, you can see how <TT>cilly --doheapify</TT>
transforms the following dangerous code: </P><PRE CLASS="verbatim"><FONT COLOR=blue> int dangerous() {
char array[10];
scanf("%s",array); // possible buffer overrun!
}
int main () {
return dangerous();
}
</FONT></PRE><P>
See the <A HREF="examples/ex25.txt">CIL output</A> for this
code fragment</P><!--TOC subsection One Return-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc30">8.8</A>  One Return</H3><!--SEC END --><P>The module <TT>oneret.ml</TT> contains a transformation the ensures that all
function bodies have at most one return statement. This simplifies a number
of analyses by providing a canonical exit-point. </P><P>For a concrete example, you can see how <TT>cilly --dooneRet</TT>
transforms the following code: </P><PRE CLASS="verbatim"><FONT COLOR=blue> int foo (int predicate) {
if (predicate <= 0) {
return 1;
} else {
if (predicate > 5)
return 2;
return 3;
}
}
</FONT></PRE><P>
See the <A HREF="examples/ex26.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Partial Evaluation and Constant Folding-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc31">8.9</A>  Partial Evaluation and Constant Folding</H3><!--SEC END --><P>The <TT>partial.ml</TT> module provides a simple interprocedural partial
evaluation and constant folding data-flow analysis and transformation.
This transformation always requires the <TT>--domakeCFG</TT> option. It
performs:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
Constant folding even of compiler-dependent constants as, for
example <TT>sizeof(T)</TT>.
</LI><LI CLASS="li-itemize"><TT>if</TT>-statement simplification for conditional expressions that
evaluate to a constant. The <TT>if</TT>-statement gets replaced with the
taken branch.
</LI><LI CLASS="li-itemize">Call elimination for
<OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
<A NAME="enum:partial-empty-proc"></A> empty functions and
</LI><LI CLASS="li-enumerate"><A NAME="enum:partial-const-func"></A> functions that return a
constant.
</LI></OL>
In case <A HREF="#enum:partial-empty-proc">1</A> the call disappears completely
and in case <A HREF="#enum:partial-const-func">2</A> it is replaced by the
constant the function returns.
</LI></UL><P>Several commandline options control the behavior of the feature.
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>--partial_no_global_const</TT>:
Treat global constants as unknown values. This is the default.
</LI><LI CLASS="li-itemize"><TT>--partial_global_const</TT>:
Treat global constants as initialized. Let global constants
participate in the partial evaluation.
</LI><LI CLASS="li-itemize"><TT>--partial_root_function</TT> ifunction-name:
Name of the function where the simplification starts. Default:
<TT>main</TT>.
</LI><LI CLASS="li-itemize"><TT>--partial_use_easy_alias</TT>
Use Partial’s built-in easy alias to analyze pointers. This is the
default.
</LI><LI CLASS="li-itemize"><TT>--partial_use_ptranal_alias</TT>
Use feature Ptranal to analyze pointers. Setting this option
requires <TT>--doptranal</TT>.
</LI></UL><P>For a concrete example, you can see how <TT>cilly --domakeCFG --dopartial</TT>
transforms the following code (note the eliminated <TT>if</TT>-branch and the
partial optimization of <TT>foo</TT>):</P><PRE CLASS="verbatim"><FONT COLOR=blue> int foo(int x, int y) {
int unknown;
if (unknown)
return y + 2;
return x + 3;
}
int bar(void) {
return -1;
}
int main(void) {
int a, b, c;
a = foo(5, 7) + foo(6, 7) + bar();
b = 4;
c = b * b;
if (b > c)
return b - c;
else
return b + c;
}
</FONT></PRE><P>
See the <A HREF="examples/ex27.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Reaching Definitions-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc32">8.10</A>  Reaching Definitions</H3><!--SEC END --><P>The <TT>reachingdefs.ml</TT> module uses the dataflow framework and CFG
information to calculate the definitions that reach each
statement. After computing the CFG (Section <A HREF="#sec-cfg">8.1</A>) and calling
<TT>computeRDs</TT> on a
function declaration, <TT>ReachingDef.stmtStartData</TT> will contain a
mapping from statement IDs to data about which definitions reach each
statement. In particular, it is a mapping from statement IDs to a
triple the first two members of which are used internally. The third
member is a mapping from variable IDs to Sets of integer options. If
the set contains <TT>Some(i)</TT>, then the definition of that variable
with ID <TT>i</TT> reaches that statement. If the set contains <TT>None</TT>,
then there is a path to that statement on which there is no definition
of that variable. Also, if the variable ID is unmapped at a
statement, then no definition of that variable reaches that statement.</P><P>To summarize, reachingdefs.ml has the following interface:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>computeRDs</TT> – Computes reaching definitions. Requires that
CFG information has already been computed for each statement.
</LI><LI CLASS="li-itemize"><TT>ReachingDef.stmtStartData</TT> – contains reaching
definition data after <TT>computeRDs</TT> is called.
</LI><LI CLASS="li-itemize"><TT>ReachingDef.defIdStmtHash</TT> – Contains a mapping
from definition IDs to the ID of the statement in which
the definition occurs.
</LI><LI CLASS="li-itemize"><TT>getRDs</TT> – Takes a statement ID and returns
reaching definition data for that statement.
</LI><LI CLASS="li-itemize"><TT>instrRDs</TT> – Takes a list of instructions and the
definitions that reach the first instruction, and for
each instruction calculates the definitions that reach
either into or out of that instruction.
</LI><LI CLASS="li-itemize"><TT>rdVisitorClass</TT> – A subclass of nopCilVisitor that
can be extended such that the current reaching definition
data is available when expressions are visited through
the <TT>get_cur_iosh</TT> method of the class.
</LI></UL><!--TOC subsection Available Expressions-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc33">8.11</A>  Available Expressions</H3><!--SEC END --><P>The <TT>availexps.ml</TT> module uses the dataflow framework and CFG
information to calculate something similar to a traditional available
expressions analysis. After <TT>computeAEs</TT> is called following a CFG
calculation (Section <A HREF="#sec-cfg">8.1</A>), <TT>AvailableExps.stmtStartData</TT> will
contain a mapping
from statement IDs to data about what expressions are available at
that statement. The data for each statement is a mapping for each
variable ID to the whole expression available at that point(in the
traditional sense) which the variable was last defined to be. So,
this differs from a traditional available expressions analysis in that
only whole expressions from a variable definition are considered rather
than all expressions.</P><P>The interface is as follows:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>computeAEs</TT> – Computes available expressions. Requires
that CFG information has already been comptued for each statement.
</LI><LI CLASS="li-itemize"><TT>AvailableExps.stmtStartData</TT> – Contains available
expressions data for each statement after <TT>computeAEs</TT> has been
called.
</LI><LI CLASS="li-itemize"><TT>getAEs</TT> – Takes a statement ID and returns
available expression data for that statement.
</LI><LI CLASS="li-itemize"><TT>instrAEs</TT> – Takes a list of instructions and
the availalbe expressions at the first instruction, and
for each instruction calculates the expressions available
on entering or exiting each instruction.
</LI><LI CLASS="li-itemize"><TT>aeVisitorClass</TT> – A subclass of nopCilVisitor that
can be extended such that the current available expressions
data is available when expressions are visited through the
<TT>get_cur_eh</TT> method of the class.
</LI></UL><!--TOC subsection Liveness Analysis-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc34">8.12</A>  Liveness Analysis</H3><!--SEC END --><P>The <TT>liveness.ml</TT> module uses the dataflow framework and
CFG information to calculate which variables are live at
each program point. After <TT>computeLiveness</TT> is called
following a CFG calculation (Section <A HREF="#sec-cfg">8.1</A>), <TT>LiveFlow.stmtStartData</TT> will
contain a mapping for each statement ID to a set of <TT>varinfo</TT>s
for varialbes live at that program point.</P><P>The interface is as follows:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>computeLiveness</TT> – Computes live variables. Requires
that CFG information has already been computed for each statement.
</LI><LI CLASS="li-itemize"><TT>LiveFlow.stmtStartData</TT> – Contains live variable data
for each statement after <TT>computeLiveness</TT> has been called.
</LI></UL><P>Also included in this module is a command line interface that
will cause liveness data to be printed to standard out for
a particular function or label.</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>--doliveness</TT> – Instructs cilly to comptue liveness
information and to print on standard out the variables live
at the points specified by <TT>--live_func</TT> and <TT>live_label</TT>.
If both are ommitted, then nothing is printed.
</LI><LI CLASS="li-itemize"><TT>--live_func</TT> – The name of the function whose
liveness data is of interest. If <TT>--live_label</TT> is ommitted,
then data for each statement is printed.
</LI><LI CLASS="li-itemize"><TT>--live_label</TT> – The name of the label at which
the liveness data will be printed.
</LI></UL><!--TOC subsection Dead Code Elimination-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc35">8.13</A>  Dead Code Elimination</H3><!--SEC END --><P>The module <TT>deadcodeelim.ml</TT> uses the reaching definitions
analysis to eliminate assignment instructions whose results
are not used. The interface is as follows:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>elim_dead_code</TT> – Performs dead code elimination
on a function. Requires that CFG information has already
been computed (Section <A HREF="#sec-cfg">8.1</A>).
</LI><LI CLASS="li-itemize"><TT>dce</TT> – Performs dead code elimination on an
entire file. Requires that CFG information has already
been computed.
</LI></UL><!--TOC subsection Simple Memory Operations-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc36">8.14</A>  Simple Memory Operations</H3><!--SEC END --><P>The <TT>simplemem.ml</TT> module allows CIL lvalues that contain memory
accesses to be even futher simplified via the introduction of
well-typed temporaries. After this transformation all lvalues involve
at most one memory reference.</P><P>For a concrete example, you can see how <TT>cilly --dosimpleMem</TT>
transforms the following code:</P><PRE CLASS="verbatim"><FONT COLOR=blue> int main () {
int ***three;
int **two;
***three = **two;
}
</FONT></PRE><P>
See the <A HREF="examples/ex28.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Simple Three-Address Code-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc37">8.15</A>  Simple Three-Address Code</H3><!--SEC END --><P>The <TT>simplify.ml</TT> module further reduces the complexity of program
expressions and gives you a form of three-address code. After this
transformation all expressions will adhere to the following grammar:
</P><PRE CLASS="verbatim"> basic::=
Const _
Addrof(Var v, NoOffset)
StartOf(Var v, NoOffset)
Lval(Var v, off), where v is a variable whose address is not taken
and off contains only "basic"
exp::=
basic
Lval(Mem basic, NoOffset)
BinOp(bop, basic, basic)
UnOp(uop, basic)
CastE(t, basic)
lval ::=
Mem basic, NoOffset
Var v, off, where v is a variable whose address is not taken and off
contains only "basic"
</PRE><P>In addition, all <TT>sizeof</TT> and <TT>alignof</TT> forms are turned into
constants. Accesses to arrays and variables whose address is taken are
turned into "Mem" accesses. All field and index computations are turned
into address arithmetic.</P><P>For a concrete example, you can see how <TT>cilly --dosimplify</TT>
transforms the following code:</P><PRE CLASS="verbatim"><FONT COLOR=blue> int main() {
struct mystruct {
int a;
int b;
} m;
int local;
int arr[3];
int *ptr;
ptr = &local;
m.a = local + sizeof(m) + arr[2];
return m.a;
}
</FONT></PRE><P>
See the <A HREF="examples/ex29.txt">CIL output</A> for this
code fragment</P><!--TOC subsection Converting C to C++-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc38">8.16</A>  Converting C to C++</H3><!--SEC END --><P>The module canonicalize.ml performs several transformations to correct
differences between C and C++, so that the output is (hopefully) valid
C++ code. This may be incomplete — certain fixes which are necessary
for some programs are not yet implemented.</P><P>Using the <TT>--doCanonicalize</TT> option with CIL will perform the
following changes to your program:</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Any variables that use C++ keywords as identifiers are renamed.
</LI><LI CLASS="li-enumerate">C allows global variables to have multiple declarations and
multiple (equivalent) definitions. This transformation removes
all but one declaration and all but one definition.
</LI><LI CLASS="li-enumerate"><TT>__inline</TT> is #defined to <TT>inline</TT>, and <TT>__restrict</TT>
is #defined to nothing.
</LI><LI CLASS="li-enumerate">C allows function pointers with no specified arguments to be used on
any argument list. To make C++ accept this code, we insert a cast
from the function pointer to a type that matches the arguments. Of
course, this does nothing to guarantee that the pointer actually has
that type.
</LI><LI CLASS="li-enumerate">Makes casts from int to enum types explicit. (CIL changes enum
constants to int constants, but doesn’t use a cast.)
</LI></OL><!--TOC subsection Generating LLVM code (new in 1.3.7)-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc39">8.17</A>  Generating LLVM code (new in 1.3.7)</H3><!--SEC END --><P>The llvm.ml module generates LLVM (“Low Level Virtual Machine”,
<A HREF="javascript:loadTop('http://llvm.org')">http://llvm.org</A>) code from a CIL file. The
current version only targets 32-bit x86 code (as of version 2.5,
LLVM’s 64-bit x86 support is still incomplete), and has a few significant
limitations:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
No support for bitfields.
</LI><LI CLASS="li-itemize">No support for inline assembly.
</LI><LI CLASS="li-itemize">Ignores gcc pragmas and attributes (except those explicitly handled by CIL).
</LI><LI CLASS="li-itemize">No support for variable-sized types.
</LI></UL><P>LLVM support is enabled by running <TT>configure</TT> with the
<TT>--with-llvm</TT> option before compiling CIL. You can then convert C
code to LLVM assembly with <TT>cilly --dollvm somefile.c</TT>. Or you can
call <TT>Llvm.generate</TT> on a CIL file to get LLVM assembly as a <TT>doc</TT>
string.</P><!--TOC section Controlling CIL-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc40">9</A>  Controlling CIL</H2><!--SEC END --><P>In the process of converting a C file to CIL we drop the unused prototypes
and even inline function definitions. This results in much smaller files. If
you do not want this behavior then you must pass the <TT>--keepunused</TT> argument
to the CIL application. </P><P>Alternatively you can put the following pragma in the code (instructing CIL
to specifically keep the declarations and definitions of the function
<TT>func1</TT> and variable <TT>var2</TT>, the definition of type <TT>foo</TT> and of
structure <TT>bar</TT>):
</P><PRE CLASS="verbatim"><FONT COLOR=blue>#pragma cilnoremove("func1", "var2", "type foo", "struct bar")
</FONT></PRE><!--TOC section GCC Extensions-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc41">10</A>  GCC Extensions</H2><!--SEC END --><P>The CIL parser handles most of the <TT>gcc</TT>
<A HREF="javascript:loadTop('http://gcc.gnu.org/onlinedocs/gcc-3.0.2/gcc_5.html#SEC67')">extensions</A>
and compiles them to CIL. The following extensions are not handled (note that
we are able to compile a large number of programs, including the Linux kernel,
without encountering these):</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Nested function definitions.
</LI><LI CLASS="li-enumerate">Constructing function calls.
</LI><LI CLASS="li-enumerate">Naming an expression’s type.
</LI><LI CLASS="li-enumerate">Complex numbers
</LI><LI CLASS="li-enumerate">Hex floats
</LI><LI CLASS="li-enumerate">Subscripts on non-lvalue arrays.
</LI><LI CLASS="li-enumerate">Forward function parameter declarations
</LI></OL><P>The following extensions are handled, typically by compiling them away:
</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Attributes for functions, variables and types. In fact, we have a clear
specification (see Section <A HREF="#sec-attrib">6.4</A>) of how attributes are interpreted. The
specification extends that of <TT>gcc</TT>.
</LI><LI CLASS="li-enumerate">Old-style function definitions and prototypes. These are translated to
new-style.
</LI><LI CLASS="li-enumerate">Locally-declared labels. As part of the translation to CIL, we generate
new labels as needed.
</LI><LI CLASS="li-enumerate">Labels as values and computed goto. This allows a program to take the
address of a label and to manipulate it as any value and also to perform a
computed goto. We compile this by assigning each label whose address is taken
a small integer that acts as its address. Every computed <TT>goto</TT> in the body
of the function is replaced with a <TT>switch</TT> statement. If you want to invoke
the label from another function, you are on your own (the <TT>gcc</TT>
documentation says the same.)
</LI><LI CLASS="li-enumerate">Generalized lvalues. You can write code like <TT>(a, b) += 5</TT> and it gets
translated to CIL.
</LI><LI CLASS="li-enumerate">Conditionals with omitted operands. Things like <TT>x ? : y</TT> are
translated to CIL.
</LI><LI CLASS="li-enumerate">Double word integers. The type <TT>long long</TT> and the <TT>LL</TT> suffix on
constants is understood. This is currently interpreted as 64-bit integers.
</LI><LI CLASS="li-enumerate">Local arrays of variable length. These are converted to uses of
<TT>alloca</TT>, the array variable is replaced with a pointer to the allocated
array and the instances of <TT>sizeof(a)</TT> are adjusted to return the size of
the array and not the size of the pointer.
</LI><LI CLASS="li-enumerate">Non-constant local initializers. Like all local initializers these are
compiled into assignments.
</LI><LI CLASS="li-enumerate">Compound literals. These are also turned into assignments.
</LI><LI CLASS="li-enumerate">Designated initializers. The CIL parser actually supports the full ISO
syntax for initializers, which is more than both <TT>gcc</TT> and <TT>MSVC</TT>. I
(George) think that this is the most complicated part of the C language and
whoever designed it should be banned from ever designing languages again.
</LI><LI CLASS="li-enumerate">Case ranges. These are compiled into separate cases. There is no code
duplication, just a larger number of <TT>case</TT> statements.
</LI><LI CLASS="li-enumerate">Transparent unions. This is a strange feature that allows you to define
a function whose formal argument has a (tranparent) union type, but the
argument is called as if it were the first element of the union. This is
compiled away by saying that the type of the formal argument is that of the
first field, and the first thing in the function body we copy the formal into
a union. </LI><LI CLASS="li-enumerate">Inline assembly-language. The full syntax is supported and it is carried
as such in CIL.</LI><LI CLASS="li-enumerate">Function names as strings. The identifiers <TT>__FUNCTION__</TT> and
<TT>__PRETTY_FUNCTION__</TT> are replaced with string literals. </LI><LI CLASS="li-enumerate">Keywords <TT>typeof</TT>, <TT>alignof</TT>, <TT>inline</TT> are supported.
</LI></OL><!--TOC section CIL Limitations-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc42">11</A>  CIL Limitations</H2><!--SEC END --><P>There are several implementation details of CIL that might make it unusable
or less than ideal for certain tasks:</P><UL CLASS="itemize"><LI CLASS="li-itemize">
CIL operates after preprocessing. If you need to see comments, for
example, you cannot use CIL. But you can use attributes and pragmas instead.
And there is some support to help you patch the include files before they are
seen by the preprocessor. For example, this is how we turn some
<TT>#define</TT>s that we don’t like into function calls. </LI><LI CLASS="li-itemize">CIL does transform the code in a non-trivial way. This is done in order
to make most analyses easier. But if you want to see the code <TT>e1, e2++</TT>
exactly as it appears in the code, then you should not use CIL. </LI><LI CLASS="li-itemize">CIL removes all local scopes and moves all variables to function
scope. It also separates a declaration with an initializer into a declaration
plus an assignment. The unfortunate effect of this transformation is that
local variables cannot have the <TT>const</TT> qualifier.</LI></UL><!--TOC section Known Bugs and Limitations-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc43">12</A>  Known Bugs and Limitations</H2><!--SEC END --><!--TOC subsection Code that CIL won’t compile-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc44">12.1</A>  Code that CIL won’t compile</H3><!--SEC END --><UL CLASS="itemize"><LI CLASS="li-itemize">
We do not support tri-graph sequences (ISO 5.2.1.1).</LI><LI CLASS="li-itemize">CIL cannot parse arbitrary <TT>#pragma</TT> directives. Their
syntax must follow gcc’s attribute syntax to be understood. If you
need a pragma that does not follow gcc syntax, add that pragma’s name
to <TT>no_parse_pragma</TT> in <TT>src/frontc/clexer.mll</TT> to indicate that
CIL should treat that pragma as a monolithic string rather than try
to parse its arguments.<P>CIL cannot parse a line containing an empty <TT>#pragma</TT>.</P></LI><LI CLASS="li-itemize">CIL only parses <TT>#pragma</TT> directives at the "top level", this is,
outside of any enum, structure, union, or function definitions.<P>If your compiler uses pragmas in places other than the top-level,
you may have to preprocess the sources in a special way (sed, perl,
etc.) to remove pragmas from these locations.</P></LI><LI CLASS="li-itemize">CIL cannot parse the following code (fixing this problem would require
extensive hacking of the LALR grammar):
<PRE CLASS="verbatim"><FONT COLOR=blue>int bar(int ()); // This prototype cannot be parsed
int bar(int x()); // If you add a name to the function, it works
int bar(int (*)()); // This also works (and it is more appropriate)
</FONT></PRE></LI><LI CLASS="li-itemize">CIL also cannot parse certain K&R old-style prototypes with missing
return type:
<PRE CLASS="verbatim"><FONT COLOR=blue>g(); // This cannot be parsed
int g(); // This is Ok
</FONT></PRE></LI><LI CLASS="li-itemize">CIL does not understand some obscure combinations of type
specifiers (“signed” and “unsigned” applied to typedefs that
themselves contain a sign specification; you could argue that this
should not be allowed anyway):
<PRE CLASS="verbatim"><FONT COLOR=blue>typedef signed char __s8;
__s8 unsigned uchartest; // This is unsigned char for gcc
</FONT></PRE></LI><LI CLASS="li-itemize">CIL does not support constant-folding of floating-point values,
because it is difficult to simulate the behavior of various
C floating-point implementations in Ocaml. Therefore, code such as
this will not compile:
<PRE CLASS="verbatim"><FONT COLOR=blue>int globalArray[(1.0 < 2.0) ? 5 : 50]
</FONT></PRE></LI><LI CLASS="li-itemize">CIL uses Ocaml ints to represent the size of an object.
Therefore, it can’t compute the size of any object that is larger
than 2<SUP>30</SUP> bits (134 MB) on 32-bit computers, or 2<SUP>62</SUP> bits on
64-bit computers.</LI></UL><!--TOC subsection Code that behaves differently under CIL-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc45">12.2</A>  Code that behaves differently under CIL</H3><!--SEC END --><UL CLASS="itemize"><LI CLASS="li-itemize">
GCC has a strange feature called “extern inline”. Such a function can
be defined twice: first with the “extern inline” specifier and the second
time without it. If optimizations are turned off then the “extern inline”
definition is considered a prototype (its body is ignored). <P>With old versions of gcc, if optimizations
are turned on then the extern inline function is inlined at all of its
occurrences from the point of its definition all the way to the point where the
(optional) second definition appears. No body is generated for an extern
inline function. A body is generated for the real definition and that one is
used in the rest of the file. </P><P>With new versions of gcc, the extern inline function is used only if there is no
actual (non-extern) second definition (i.e. the second definition is used in the
whole file, even for calls between the extern inline definition and the second
definition).</P><P>By default, CIL follows the current gcc behavior for extern
inline. However, if you set <TT>oldstyleExternInline</TT> to true, you will
get an emulation of gcc’s old behaviour (under the assumption that
optimizations are enabled): CIL will rename your extern inline
function (and its uses) with the suffix <TT>__extinline</TT>. This means
that if you have two such definition, that do different things and the
optimizations are not on, then the CIL version might compute a
different answer! Also, if you have multiple extern inline
declarations then CIL will ignore but the first one. This is not so
bad because GCC itself would not like it.</P></LI><LI CLASS="li-itemize">The implementation of <TT>bitsSizeOf</TT> does not take into account the
packing pragmas. However it was tested to be accurate on cygwin/gcc-2.95.3,
Linux/gcc-2.95.3 and on Windows/MSVC.</LI><LI CLASS="li-itemize"><TT>-malign-double</TT> is ignored.</LI><LI CLASS="li-itemize">The statement <TT>x = 3 + x ++</TT> will perform the increment of <TT>x</TT>
before the assignment, while <TT>gcc</TT> delays the increment after the
assignment. It turned out that this behavior is much easier to implement
than gcc’s one, and either way is correct (since the behavior is unspecified
in this case). Similarly, if you write <TT>x = x ++;</TT> then CIL will perform
the increment before the assignment, whereas GCC and MSVC will perform it
after the assignment. </LI><LI CLASS="li-itemize">Because CIL uses 64-bit floating point numbers in its internal
representation of floating point numbers, <TT>long double</TT> constants
are parsed as if they were <TT>double</TT> constants.</LI></UL><!--TOC subsection Effects of the CIL translation-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc46">12.3</A>  Effects of the CIL translation</H3><!--SEC END --><UL CLASS="itemize"><LI CLASS="li-itemize">
CIL cleans up C code in various ways that may suppress compiler
warnings. For example, CIL will add casts where they are needed
while <TT>gcc</TT> might print a warning for the missing cast. It is
not a goal of CIL to emit such warnings — we support several
versions of several different compilers, and mimicking the warnings
of each is simply not possible. If you want to see compiler
warnings, compile your program with your favorite compiler before
using CIL.</LI><LI CLASS="li-itemize">When you use variable-length arrays, CIL turns them into calls
to <TT>alloca</TT>. This means that they are deallocated when the
function returns and not when the local scope ends.<P>Variable-length arrays are not supported as fields of a struct or union.</P></LI><LI CLASS="li-itemize">In the new versions of <TT>glibc</TT> there is a function
<TT>__builtin_va_arg</TT> that takes a type as its second argument. CIL
handles that through a slight trick. As it parses the function it changes a
call like:
<PRE CLASS="verbatim"> mytype x = __builtin_va_arg(marker, mytype)
</PRE>into
<PRE CLASS="verbatim"> mytype x;
__builtin_va_arg(marker, sizeof(mytype), &x);
</PRE><P>The latter form is used internally in CIL. However, the CIL pretty printer
will try to emit the original code. </P><P>Similarly, <TT>__builtin_types_compatible_p(t1, t2)</TT>, which takes
types as arguments, is represented internally as
<TT>__builtin_types_compatible_p(sizeof t1, sizeof t2)</TT>, but the
sizeofs are removed when printing.</P></LI></UL><!--TOC section Using the merger-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc47">13</A>  Using the merger</H2><!--SEC END --><P><A NAME="sec-merger"></A></P><!--NAME merger.html-->
<P>There are many program analyses that are more effective when
done on the whole program.</P><P>The merger is a tool that combines all of the C source files in a project
into a single C file. There are two tasks that a merger must perform:
</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Detect what are all the sources that make a project and with what
compiler arguments they are compiled.</LI><LI CLASS="li-enumerate">Merge all of the source files into a single file.
</LI></OL><P>For the first task the merger impersonates a compiler and a linker (both a
GCC and a Microsoft Visual C mode are supported) and it expects to be invoked
(from a build script or a Makefile) on all sources of the project. When
invoked to compile a source the merger just preprocesses the source and saves
the result using the name of the requested object file. By preprocessing at
this time the merger is able to take into account variations in the command
line arguments that affect preprocessing of different source files.</P><P>When the merger is invoked to link a number of object files it collects the
preprocessed sources that were stored with the names of the object files, and
invokes the merger proper. Note that arguments that affect the compilation or
linking must be the same for all source files.</P><P>For the second task, the merger essentially concatenates the preprocessed
sources with care to rename conflicting file-local declarations (we call this
process alpha-conversion of a file). The merger also attempts to remove
duplicate global declarations and definitions. Specifically the following
actions are taken: </P><UL CLASS="itemize"><LI CLASS="li-itemize">
File-scope names (<TT>static</TT> globals, names of types defined with
<TT>typedef</TT>, and structure/union/enumeration tags) are given new names if they
conflict with declarations from previously processed sources. The new name is
formed by appending the suffix <TT>___n</TT>, where <TT>n</TT> is a unique integer
identifier. Then the new names are applied to their occurrences in the file. </LI><LI CLASS="li-itemize">Non-static declarations and definitions of globals are never renamed.
But we try to remove duplicate ones. Equality of globals is detected by
comparing the printed form of the global (ignoring the line number directives)
after the body has been alpha-converted. This process is intended to remove
those declarations (e.g. function prototypes) that originate from the same
include file. Similarly, we try to eliminate duplicate definitions of
<TT>inline</TT> functions, since these occasionally appear in include files.</LI><LI CLASS="li-itemize">The types of all global declarations with the same name from all files
are compared for type isomorphism. During this process, the merger detects all
those isomorphisms between structures and type definitions that are <B>required</B> for the merged program to be legal. Such structure tags and
typenames are coalesced and given the same name. </LI><LI CLASS="li-itemize">Besides the structure tags and type names that are required to be
isomorphic, the merger also tries to coalesce definitions of structures and
types with the same name from different file. However, in this case the merger
will not give an error if such definitions are not isomorphic; it will just
use different names for them. </LI><LI CLASS="li-itemize">In rare situations, it can happen that a file-local global in
encountered first and it is not renamed, only to discover later when
processing another file that there is an external symbol with the same name.
In this case, a second pass is made over the merged file to rename the
file-local symbol.
</LI></UL><P>Here is an example of using the merger:</P><P>The contents of <TT>file1.c</TT> is:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>struct foo; // Forward declaration
extern struct foo *global;
</FONT></PRE><P>The contents of <TT>file2.c</TT> is:</P><PRE CLASS="verbatim"><FONT COLOR=blue>struct bar {
int x;
struct bar *next;
};
extern struct bar *global;
struct foo {
int y;
};
extern struct foo another;
void main() {
}
</FONT></PRE><P>There are several ways in which one might create an executable from these
files:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<PRE CLASS="verbatim">gcc file1.c file2.c -o a.out
</PRE></LI><LI CLASS="li-itemize"><PRE CLASS="verbatim">gcc -c file1.c -o file1.o
gcc -c file2.c -o file2.o
ld file1.o file2.o -o a.out
</PRE></LI><LI CLASS="li-itemize"><PRE CLASS="verbatim">gcc -c file1.c -o file1.o
gcc -c file2.c -o file2.o
ar r libfile2.a file2.o
gcc file1.o libfile2.a -o a.out
</PRE></LI><LI CLASS="li-itemize"><PRE CLASS="verbatim">gcc -c file1.c -o file1.o
gcc -c file2.c -o file2.o
ar r libfile2.a file2.o
gcc file1.o -lfile2 -o a.out
</PRE></LI></UL><P>In each of the cases above you must replace all occurrences of <TT>gcc</TT> and
<TT>ld</TT> with <TT>cilly --merge</TT>, and all occurrences of <TT>ar</TT> with <TT>cilly
--merge --mode=AR</TT>. It is very important that the <TT>--merge</TT> flag be used
throughout the build process. If you want to see the merged source file you
must also pass the <TT>--keepmerged</TT> flag to the linking phase. </P><P>The result of merging file1.c and file2.c is:
</P><PRE CLASS="verbatim"><FONT COLOR=blue>// from file1.c
struct foo; // Forward declaration
extern struct foo *global;
// from file2.c
struct foo {
int x;
struct foo *next;
};
struct foo___1 {
int y;
};
extern struct foo___1 another;
</FONT></PRE><!--TOC section Using the patcher-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc48">14</A>  Using the patcher</H2><!--SEC END --><P><A NAME="sec-patcher"></A></P><!--NAME patcher.html-->
<P>Occasionally we have needed to modify slightly the standard include files.
So, we developed a simple mechanism that allows us to create modified copies
of the include files and use them instead of the standard ones. For this
purpose we specify a patch file and we run a program caller Patcher which
makes modified copies of include files and applies the patch.</P><P>The patcher is invoked as follows:
</P><PRE CLASS="verbatim">bin/patcher [options]
Options:
--help Prints this help message
--verbose Prints a lot of information about what is being done
--mode=xxx What tool to emulate:
GNUCC - GNU CC
MSVC - MS VC cl compiler
--dest=xxx The destination directory. Will make one if it does not exist
--patch=xxx Patch file (can be specified multiple times)
--ppargs=xxx An argument to be passed to the preprocessor (can be specified
multiple times)
--ufile=xxx A user-include file to be patched (treated as \#include "xxx")
--sfile=xxx A system-include file to be patched (treated as \#include <xxx>)
--clean Remove all files in the destination directory
--dumpversion Print the version name used for the current compiler
All of the other arguments are passed to the preprocessor. You should pass
enough arguments (e.g., include directories) so that the patcher can find the
right include files to be patched.
</PRE><P>Based on the given <TT>mode</TT> and the current version of the compiler (which
the patcher can print when given the <TT>dumpversion</TT> argument) the patcher
will create a subdirectory of the <TT>dest</TT> directory, such as:
</P><PRE CLASS="verbatim"> /usr/home/necula/cil/include/gcc_2.95.3-5
</PRE><P>In that file the patcher will copy the modified versions of the include files
specified with the <TT>ufile</TT> and <TT>sfile</TT> options. Each of these options can
be specified multiple times. </P><P>The patch file (specified with the <TT>patch</TT> option) has a format inspired by
the Unix <TT>patch</TT> tool. The file has the following grammar:</P><PRE CLASS="verbatim"> <<< flags
patterns
===
replacement
>>>
</PRE><P>The flags are a comma separated, case-sensitive, sequence of keywords or
keyword = value. The following flags are supported:
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>file=foo.h</TT> - will only apply the patch on files whose name is
<TT>foo.h</TT>.
</LI><LI CLASS="li-itemize"><TT>optional</TT> - this means that it is Ok if the current patch does not
match any of the processed files.
</LI><LI CLASS="li-itemize"><TT>group=foo</TT> - will add this patch to the named group. If this is not
specified then a unique group is created to contain just the current patch.
When all files specified in the command line have been patched, an error
message is generated for all groups for whom no member patch was used. We use
this mechanism to receive notice when the patch triggers are out-dated with
respect to the new include files.
</LI><LI CLASS="li-itemize"><TT>system=sysname</TT> - will only consider this pattern on a given
operating system. The “sysname” is reported by the “$Ô” variable in
Perl, except that Windows is always considered to have sysname
“cygwin.” For Linux use “linux” (capitalization matters).
</LI><LI CLASS="li-itemize"><TT>ateof</TT> - In this case the patterns are ignored and the replacement
text is placed at the end of the patched file. Use the <TT>file</TT> flag if you
want to restrict the files in which this replacement is performed.
</LI><LI CLASS="li-itemize"><TT>atsof</TT> - The patterns are ignored and the replacement text is placed
at the start of the patched file. Uf the <TT>file</TT> flag to restrict the
application of this patch to a certain file.
</LI><LI CLASS="li-itemize"><TT>disabled</TT> - Use this flag if you want to disable the pattern.
</LI></UL><P>The patterns can consist of several groups of lines separated by the <TT>|||</TT>
marker. Each of these group of lines is a multi-line pattern that if found in
the file will be replaced with the text given at the end of the block. </P><P>The matching is space-insensitive.</P><P>All of the markers <TT><<<</TT>, <TT>|||</TT>, <TT>===</TT> and <TT>>>></TT> must appear at the
beginning of a line but they can be followed by arbitrary text (which is
ignored).</P><P>The replacement text can contain the special keyword <TT>@__pattern__@</TT>,
which is substituted with the pattern that matched. </P><!--TOC section Debugging support-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc49">15</A>  Debugging support</H2><!--SEC END --><P><A NAME="sec-debugger"></A></P><P>Most of the time we debug our code using the Errormsg module along with the
pretty printer. But if you want to use the Ocaml debugger here is an easy way
to do it. Say that you want to debug the invocation of cilly that arises out
of the following command:
</P><PRE CLASS="verbatim">cilly -c hello.c
</PRE><P>You must follow the installation <A HREF="../ccured/setup.html">instructions</A>
to install the Elist support files for ocaml and to extend your .emacs
appropriately. Then from within Emacs you do
</P><PRE CLASS="verbatim">ALT-X my-camldebug
</PRE><P>This will ask you for the command to use for running the Ocaml debugger
(initially the default will be “ocamldebug” or the last command you
introduced). You use the following command:
</P><PRE CLASS="verbatim">cilly --ocamldebug -c hello.c
</PRE><P>This will run <TT>cilly</TT> as usual and invoke the Ocaml debugger when the cilly
engine starts. The advantage of this way of invoking the debugger is that the
directory search paths are set automatically and the right set or arguments is
passed to the debugger. </P><!--TOC section Who Says C is Simple?-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc50">16</A>  Who Says C is Simple?</H2><!--SEC END --><P><A NAME="sec-simplec"></A></P><P>When I (George) started to write CIL I thought it was going to take two weeks.
Exactly a year has passed since then and I am still fixing bugs in it. This
gross underestimate was due to the fact that I thought parsing and making
sense of C is simple. You probably think the same. What I did not expect was
how many dark corners this language has, especially if you want to parse
real-world programs such as those written for GCC or if you are more ambitious
and you want to parse the Linux or Windows NT sources (both of these were
written without any respect for the standard and with the expectation that
compilers will be changed to accommodate the program). </P><P>The following examples were actually encountered either in real programs or
are taken from the ISO C99 standard or from the GCC’s testcases. My first
reaction when I saw these was: <EM>Is this C?</EM>. The second one was : <EM>What the hell does it mean?</EM>. </P><P>If you are contemplating doing program analysis for C on abstract-syntax
trees then your analysis ought to be able to handle these things. Or, you can
use CIL and let CIL translate them into clean C code. </P><!--TOC subsection Standard C-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc51">16.1</A>  Standard C</H3><!--SEC END --><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">Why does the following code return 0 for most values of <TT>x</TT>? (This
should be easy.)<PRE CLASS="verbatim"><FONT COLOR=blue> int x;
return x == (1 && x);
</FONT></PRE><P>
See the <A HREF="examples/ex30.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Why does the following code return 0 and not -1? (Answer: because
<TT>sizeof</TT> is unsigned, thus the result of the subtraction is unsigned, thus
the shift is logical.)<PRE CLASS="verbatim"><FONT COLOR=blue> return ((1 - sizeof(int)) >> 32);
</FONT></PRE><P>
See the <A HREF="examples/ex31.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Scoping rules can be tricky. This function returns 5.<PRE CLASS="verbatim"><FONT COLOR=blue>int x = 5;
int f() {
int x = 3;
{
extern int x;
return x;
}
}
</FONT></PRE><P>
See the <A HREF="examples/ex32.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Functions and function pointers are implicitly converted to each other. <PRE CLASS="verbatim"><FONT COLOR=blue>int (*pf)(void);
int f(void) {
pf = &f; // This looks ok
pf = ***f; // Dereference a function?
pf(); // Invoke a function pointer?
(****pf)(); // Looks strange but Ok
(***************f)(); // Also Ok
}
</FONT></PRE><P>
See the <A HREF="examples/ex33.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Initializer with designators are one of the hardest parts about ISO C.
Neither MSVC or GCC implement them fully. GCC comes close though. What is the
final value of <TT>i.nested.y</TT> and <TT>i.nested.z</TT>? (Answer: 2 and respectively
6). <PRE CLASS="verbatim"><FONT COLOR=blue>struct {
int x;
struct {
int y, z;
} nested;
} i = { .nested.y = 5, 6, .x = 1, 2 };
</FONT></PRE><P>
See the <A HREF="examples/ex34.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">This is from c-torture. This function returns 1.<PRE CLASS="verbatim"><FONT COLOR=blue>typedef struct
{
char *key;
char *value;
} T1;
typedef struct
{
long type;
char *value;
} T3;
T1 a[] =
{
{
"",
((char *)&((T3) {1, (char *) 1}))
}
};
int main() {
T3 *pt3 = (T3*)a[0].value;
return pt3->value;
}
</FONT></PRE><P>
See the <A HREF="examples/ex35.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Another one with constructed literals. This one is legal according to
the GCC documentation but somehow GCC chokes on (it works in CIL though). This
code returns 2.<PRE CLASS="verbatim"><FONT COLOR=blue> return ((int []){1,2,3,4})[1];
</FONT></PRE><P>
See the <A HREF="examples/ex36.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">In the example below there is one copy of “bar” and two copies of
“pbar” (static prototypes at block scope have file scope, while for all
other types they have block scope). <PRE CLASS="verbatim"><FONT COLOR=blue> int foo() {
static bar();
static (*pbar)() = bar;
}
static bar() {
return 1;
}
static (*pbar)() = 0;
</FONT></PRE><P>
See the <A HREF="examples/ex37.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Two years after heavy use of CIL, by us and others, I discovered a bug
in the parser. The return value of the following function depends on what
precedence you give to casts and unary minus:
<PRE CLASS="verbatim"><FONT COLOR=blue> unsigned long foo() {
return (unsigned long) - 1 / 8;
}
</FONT></PRE>
See the <A HREF="examples/ex38.txt">CIL output</A> for this
code fragment<P>The correct interpretation is <TT>((unsigned long) - 1) / 8</TT>, which is a
relatively large number, as opposed to <TT>(unsigned long) (- 1 / 8)</TT>, which
is 0. </P></LI></OL><!--TOC subsection GCC ugliness-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc52">16.2</A>  GCC ugliness</H3><!--SEC END --><P><A NAME="sec-ugly-gcc"></A></P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">GCC has generalized lvalues. You can take the address of a lot of
strange things:<PRE CLASS="verbatim"><FONT COLOR=blue> int x, y, z;
return &(x ? y : z) - & (x++, x);
</FONT></PRE><P>
See the <A HREF="examples/ex39.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">GCC lets you omit the second component of a conditional expression.<PRE CLASS="verbatim"><FONT COLOR=blue> extern int f();
return f() ? : -1; // Returns the result of f unless it is 0
</FONT></PRE><P>
See the <A HREF="examples/ex40.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">Computed jumps can be tricky. CIL compiles them away in a fairly clean
way but you are on your own if you try to jump into another function this way.<PRE CLASS="verbatim"><FONT COLOR=blue>static void *jtab[2]; // A jump table
static int doit(int x){
static int jtab_init = 0;
if(!jtab_init) { // Initialize the jump table
jtab[0] = &&lbl1;
jtab[1] = &&lbl2;
jtab_init = 1;
}
goto *jtab[x]; // Jump through the table
lbl1:
return 0;
lbl2:
return 1;
}
int main(void){
if (doit(0) != 0) exit(1);
if (doit(1) != 1) exit(1);
exit(0);
}
</FONT></PRE><P>
See the <A HREF="examples/ex41.txt">CIL output</A> for this
code fragment</P></LI><LI CLASS="li-enumerate">A cute little example that we made up. What is the returned value?
(Answer: 1);
<PRE CLASS="verbatim"><FONT COLOR=blue> return ({goto L; 0;}) && ({L: 5;});
</FONT></PRE>
See the <A HREF="examples/ex42.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate"><TT>extern inline</TT> is a strange feature of GNU C. Can you guess what the
following code computes?<PRE CLASS="verbatim"><FONT COLOR=blue>extern inline foo(void) { return 1; }
int firstuse(void) { return foo(); }
// A second, incompatible definition of foo
int foo(void) { return 2; }
int main() {
return foo() + firstuse();
}
</FONT></PRE><P>
See the <A HREF="examples/ex43.txt">CIL output</A> for this
code fragment</P><P>The answer depends on whether the optimizations are turned on. If they are
then the answer is 3 (the first definition is inlined at all occurrences until
the second definition). If the optimizations are off, then the first
definition is ignore (treated like a prototype) and the answer is 4. </P><P>CIL will misbehave on this example, if the optimizations are turned off (it
always returns 3).</P></LI><LI CLASS="li-enumerate">GCC allows you to cast an object of a type T into a union as long as the
union has a field of that type:
<PRE CLASS="verbatim"><FONT COLOR=blue>union u {
int i;
struct s {
int i1, i2;
} s;
};
union u x = (union u)6;
int main() {
struct s y = {1, 2};
union u z = (union u)y;
}
</FONT></PRE>
See the <A HREF="examples/ex44.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">GCC allows you to use the <TT>__mode__</TT> attribute to specify the size
of the integer instead of the standard <TT>char</TT>, <TT>short</TT> and so on:
<PRE CLASS="verbatim"><FONT COLOR=blue>int __attribute__ ((__mode__ ( __QI__ ))) i8;
int __attribute__ ((__mode__ ( __HI__ ))) i16;
int __attribute__ ((__mode__ ( __SI__ ))) i32;
int __attribute__ ((__mode__ ( __DI__ ))) i64;
</FONT></PRE>
See the <A HREF="examples/ex45.txt">CIL output</A> for this
code fragment</LI><LI CLASS="li-enumerate">The “alias” attribute on a function declaration tells the
linker to treat this declaration as another name for the specified
function. CIL will replace the declaration with a trampoline
function pointing to the specified target.
<PRE CLASS="verbatim"><FONT COLOR=blue> static int bar(int x, char y) {
return x + y;
}
//foo is considered another name for bar.
int foo(int x, char y) __attribute__((alias("bar")));
</FONT></PRE>
See the <A HREF="examples/ex46.txt">CIL output</A> for this
code fragment</LI></OL><!--TOC subsection Microsoft VC ugliness-->
<H3 CLASS="subsection"><!--SEC ANCHOR --><A NAME="htoc53">16.3</A>  Microsoft VC ugliness</H3><!--SEC END --><P>This compiler has few extensions, so there is not much to say here.</P><OL CLASS="enumerate" type=1><LI CLASS="li-enumerate">
Why does the following code return 0 and not -1? (Answer: because of a
bug in Microsoft Visual C. It thinks that the shift is unsigned just because
the second operator is unsigned. CIL reproduces this bug when in MSVC mode.)<PRE CLASS="verbatim"><FONT COLOR=blue> return -3 >> (8 * sizeof(int));
</FONT></PRE></LI><LI CLASS="li-enumerate">Unnamed fields in a structure seem really strange at first. It seems
that Microsoft Visual C introduced this extension, then GCC picked it up (but
in the process implemented it wrongly: in GCC the field <TT>y</TT> overlaps with
<TT>x</TT>!).<PRE CLASS="verbatim"><FONT COLOR=blue>struct {
int x;
struct {
int y, z;
struct {
int u, v;
};
};
} a;
return a.x + a.y + a.z + a.u + a.v;
</FONT></PRE><P>
See the <A HREF="examples/ex47.txt">CIL output</A> for this
code fragment</P></LI></OL><!--TOC section Authors-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc54">17</A>  Authors</H2><!--SEC END --><P>The CIL parser was developed starting from Hugues Casse’s <TT>frontc</TT>
front-end for C although all the files from the <TT>frontc</TT> distribution have
been changed very extensively. The intermediate language and the elaboration
stage are all written from scratch. The main author is
<A HREF="mailto:necula@cs.berkeley.edu">George Necula</A>, with significant
contributions from <A HREF="mailto:smcpeak@cs.berkeley.edu">Scott McPeak</A>,
<A HREF="mailto:weimer@cs.berkeley.edu">Westley Weimer</A>,
<A HREF="mailto:liblit@cs.wisc.edu">Ben Liblit</A>,
<A HREF="javascript:loadTop('http://www.cs.berkeley.edu/~matth/')">Matt Harren</A>,
Raymond To and Aman Bhargava.</P><P>This work is based upon work supported in part by the National Science
Foundation under Grants No. 9875171, 0085949 and 0081588, and gifts from
Microsoft Research. Any opinions, findings, and conclusions or recommendations
expressed in this material are those of the author(s) and do not necessarily
reflect the views of the National Science Foundation or the other sponsors.</P><!--TOC section License-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc55">18</A>  License</H2><!--SEC END --><P>Copyright (c) 2001-2007,
</P><UL CLASS="itemize"><LI CLASS="li-itemize">
George C. Necula <necula@cs.berkeley.edu>
</LI><LI CLASS="li-itemize">Scott McPeak <smcpeak@cs.berkeley.edu>
</LI><LI CLASS="li-itemize">Wes Weimer <weimer@cs.berkeley.edu>
</LI><LI CLASS="li-itemize">Ben Liblit <liblit@cs.wisc.edu>
</LI><LI CLASS="li-itemize">Matt Harren <matth@cs.berkeley.edu>
</LI></UL><P>
All rights reserved.</P><P>Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:</P><P>1. Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.</P><P>2. Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.</P><P>3. The names of the contributors may not be used to endorse or promote
products derived from this software without specific prior written
permission.</P><P>THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.</P><!--TOC section Bug reports-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc56">19</A>  Bug reports</H2><!--SEC END --><P>We are certain that there are still some remaining bugs in CIL. If you find
one please file a bug report in our Source Forge space
<A HREF="javascript:loadTop('http://sourceforge.net/projects/cil')">http://sourceforge.net/projects/cil</A>. </P><P>You can find there the latest announcements, a source distribution,
bug report submission instructions and a mailing list: cil-users[at
sign]lists.sourceforge.net. Please use this list to ask questions about CIL,
as it will ensure your message is viewed by a broad audience. </P><!--TOC section Changes-->
<H2 CLASS="section"><!--SEC ANCHOR --><A NAME="htoc57">20</A>  Changes</H2><!--SEC END --><P><A NAME="sec-changes"></A></P><!--NAME changes.html-->
<UL CLASS="itemize"><LI CLASS="li-itemize">
<B>April 24, 2009</B>: <B>Released version 1.3.7</B> (Subversion
revision 10727). Includes change below and other miscellaneous bug fixes.
</LI><LI CLASS="li-itemize"><B>April 22, 2009</B>: <TT>ocamlbuild</TT> make target added (thanks
to Gabriel Kerneis).
</LI><LI CLASS="li-itemize"><B>April 21, 2009</B>: <TT>__builtin_va_arg_pack</TT> support:
calls to this builtin need to remain as the last argument to a
function call. So to prevent CIL’s usual rewriting, we internally
represent calls to <TT>__builtin_va_arg_pack</TT> as
<TT>sizeof(__builtin_va_arg_pack)</TT>…</LI><LI CLASS="li-itemize"><B>April 21, 2009</B>: <TT>_Bool</TT> support, thanks to patches
from Christopher Ian Stern (Sourceforge bug #1497763).<P>This leads to the addition of a new integer kind, <TT>IBool</TT>
which you might have to handle in some of your patterns. Note also, for
those not familiar with <TT>_Bool</TT>, that casts/conversion to <TT>_Bool</TT>
give the value 1 for non-zero values, and 0 otherwise (i.e.
<TT>(_Bool)x</TT> behaves like <TT>x </TT><TT>=</TT><TT> 0</TT>).</P></LI><LI CLASS="li-itemize"><B>April 20, 2009</B>: Update builtin support, including
__builtin_strlen (Sourceforge bug #1873374), __builtin_choose_expr,
__builtin_types_compatible_p (Sourceforge bug #1852730).
For the latter two, we evaluate them in CIL (in line with the earlier handling
of __builtin_constant_p).<P>Note that as part of these fixes, mkAddrOf no longer replaces
<TT>&a[0]</TT> with <TT>a</TT>, as this can break <TT>typeof</TT> (Sourceforge bug
#1852730 involves a use of <TT>typeof(&a[0])</TT>).</P><P>Typo fixes from Gabriel Kerneis.</P></LI><LI CLASS="li-itemize"><B>April 17, 2009</B>: Update extern inline handling to match
current gcc versions (Sourceforge bug #1689503): the actual
definition is always preferred over the extern inline one. CIL
handles this by suppressing the extern inline definition (replacing
it by a declaration) if it sees a real definition. However, if no
real definition exists, you will get a CIL varinfo for an inline
function with storage class extern…<P>You can revert to CIL’s previous handling of extern inline by setting
<TT>oldstyleExternInline</TT> to true.
</P></LI><LI CLASS="li-itemize"><B>April 16, 2009</B>: Add allStmts function to cfg.ml, and
remove public visibility of nodeList and numNodes (Sourceforge bug
#1819746).
</LI><LI CLASS="li-itemize"><B>April 15, 2009</B>: Fix Sourceforge bugs #2265867 (bad constant
folding of unary minus), #2134504 (truncation warning), #1811676 (typo).
</LI><LI CLASS="li-itemize"><B>April 13, 2009</B>: Recognize gcc’s -m32 and -m64 flags that select
between 32 and 64 bit targets.
</LI><LI CLASS="li-itemize"><B>April 8, 2009</B>: Rename libcil.a to libcil.o (ocaml 3.11 forces
us to use the correct extension).
</LI><LI CLASS="li-itemize"><B>April 7, 2009</B>: Add LLVM bitcode generator (partial C support, target
32-bit x86 only).
</LI><LI CLASS="li-itemize"><B>April 7, 2009</B>: Support enums over greater-than-int types (gcc
extension).
</LI><LI CLASS="li-itemize"><B>September 19, 2008</B>: Remove excess newlines in warnings and errors.
</LI><LI CLASS="li-itemize"><B>August 30, 2008</B>: Added an install-findlib Make target for
ocamlfind. Thanks to ploc for the patch.
</LI><LI CLASS="li-itemize"><B>May 28, 2008</B>: Cross-compilation support: ability to
select a machine model specified in an environment variable (CIL_MACHINE).
See Section <A HREF="#sec-cilmachine">7.4</A> for details.
</LI><LI CLASS="li-itemize"><B>July 2, 2008</B>: Add a <TT>--gcc</TT> option to cilly to tell it to use
a particular version of gcc (useful for cross-compilation environments).
</LI><LI CLASS="li-itemize"><B>February 14, 2008</B>: Fixed a bug in temporary file
creation. Thanks to J. Aaron Pendergrass for the patch.
</LI><LI CLASS="li-itemize"><B>November 30, 2007</B>: Fixed a bug in assignment to lvalues that
depend on themselves.
</LI><LI CLASS="li-itemize"><B>April 4, 2007</B>: Benjamin Monate fixed a bug in
<A HREF="api/Cfg.html">Cfg</A> for empty loop bodies.
</LI><LI CLASS="li-itemize"><B>March 29, 2007</B>: Polyvios Pratikakis fixed a bug in
<TT>src/ext/pta/uref.ml</TT>.
</LI><LI CLASS="li-itemize"><B>March 15, 2007</B>: Added support for
<TT>__attribute__((aligned))</TT> and
<TT>__attribute__((packed))</TT>.
</LI><LI CLASS="li-itemize"><B>March 7, 2007</B>: <TT>typeOf(StartOf _)</TT> now preserves the
attributes of the array.
</LI><LI CLASS="li-itemize"><B>February 22, 2007</B>: Added an inliner (ext/inliner.ml)
</LI><LI CLASS="li-itemize"><B>February 21, 2007</B>: We now constant-fold bitfield
expressions. Thanks to Virgile Prevosto for the patch.
</LI><LI CLASS="li-itemize"><B>February 13, 2007</B>: gcc preprocessor arguments passed
using <TT>-Wp</TT> are now used only during initial preproccessing, not
for the preprocessing after CIL. This fixes problems in the Linux
makefiles with dependency generation.
</LI><LI CLASS="li-itemize"><B>February 6, 2007</B>: Fixed <A HREF="api/Cil.html#VALparseInt">Cil.parseInt</A> for non-32
bit architectures.
</LI><LI CLASS="li-itemize"><B>February 5, 2007</B>: <B>Released version 1.3.6</B> (Subversion
revision 9211)
</LI><LI CLASS="li-itemize"><B>February 2, 2007</B>: Improved the way CIL gets configured for the
actual definitions of <TT>size_t</TT> and <TT>wchar_t</TT>.
</LI><LI CLASS="li-itemize"><B>February 1, 2007</B>: Fixed the parser to support the unused
attribute on labels. For now, we just drop this attribute since
Rmtmps will remove unused labels anyways. Thanks to Peter Hawkins
for the patch.
</LI><LI CLASS="li-itemize"><B>January 18, 2007</B>: Require the destination of a <TT>Call</TT> to
have the same type as the function’s return type, even if it means
inserting a temporary. To get the old behavior, set
<TT>Cabs2cil.doCollapseCallCast</TT> to true as described in
Section <A HREF="#sec-cilly-internal-options">7.3</A>.
</LI><LI CLASS="li-itemize"><B>January 17, 2007</B>: Fix for <TT>__builtin_offsetof</TT> when
the field name is also a typedef name.
</LI><LI CLASS="li-itemize"><B>January 17, 2007</B>: Fixed <A HREF="api/Cil.html#VALloadBinaryFile">Cil.loadBinaryFile</A>
(Sourceforge bug #1548894).
You should only use loadBinaryFile if no other code has been loaded
or generated in the current CIL process, since loadBinaryFile needs
to load some global state.
</LI><LI CLASS="li-itemize"><B>December 18, 2006</B>: The <TT>--stats</TT> flag now gets the CPU
speed at runtime rather than configure-time, so binary executables
can be moved to different computers.
</LI><LI CLASS="li-itemize"><B>December 14, 2006</B>: Fixed various warnings and errors on
64-bit architectures.
</LI><LI CLASS="li-itemize"><B>November 26, 2006</B>: Christoph Spiel added “<TT>--no</TT>”
options to many of CIL’s command-line flags.
</LI><LI CLASS="li-itemize"><B>November 21, 2006</B>: Merged <TT>gccBuiltins</TT> and
<TT>msvcBuiltins</TT> into a single table <A HREF="api/Cil.html#VALbuiltinFunctions">Cil.builtinFunctions</A>
that is initialized by <A HREF="api/Cil.html#VALinitCIL">Cil.initCIL</A>.
</LI><LI CLASS="li-itemize"><B>October 28, 2006</B>: Added the field <TT>vdescr</TT> to the
<A HREF="api/Cil.html#TYPEvarinfo">Cil.varinfo</A> struct to remember what value is stored in
certain CIL-introduced temporary variables. For example, if CIL
adds a temporary to store the result of <TT>foo(a,b)</TT>, then the
description will be “foo(a,b)”. The new printer
<A HREF="api/Cil.html#VALdescriptiveCilPrinter">Cil.descriptiveCilPrinter</A> will substitute descriptions for
the names of temporaries. The result is not necessarily valid C,
but it may let you produce more helpful error messages in your
analysis tools: “The value foo(a,b) may be tainted” vs. “The
value __cil_tmp29 may be tainted.”
</LI><LI CLASS="li-itemize"><B>October 27, 2006</B>: Fixed a bug with duplicate entries in
the statement list of Switch nodes, and forbade duplicate
<TT>default</TT> cases.
</LI><LI CLASS="li-itemize"><B>October 12, 2006</B>: Added a new function <A HREF="api/Cil.html#VALexpToAttrParam">Cil.expToAttrParam</A>
that attempts to convert an expression into a attribute parameter.
</LI><LI CLASS="li-itemize"><B>October 12, 2006</B>: Added an attribute with the length of the array,
when array types of formal arguments are converted to pointer types.
</LI><LI CLASS="li-itemize"><B>September 29, 2006</B>: Benjamin Monate fixed a bug in compound
local initializers that was causing duplicate code to be added.
</LI><LI CLASS="li-itemize"><B>August 9, 2006</B>: Changed the patcher to print “<TT>#line
nnn</TT>” directives instead of “<TT># nnn</TT>”.
</LI><LI CLASS="li-itemize"><B>August 6, 2006</B>: Joseph Koshy patched <TT>./configure</TT> for
FreeBSD on amd64.
</LI><LI CLASS="li-itemize"><B>July 27, 2006</B>: CIL files now include the prototypes of
builtin functions (such as <TT>__builtin_va_arg</TT>). This
preserves the invariant that every function call has a corresponding
function or function prototype in the file. However, the prototypes
of builtins are not printed in the output files.
</LI><LI CLASS="li-itemize"><B>July 23, 2006</B>: Incorporated some fixes for the constant folding
for lvalues, and fixed grammatical errors. Thanks to Christian Stork.
</LI><LI CLASS="li-itemize"><B>July 23, 2006</B>: Changed the way ./configure works. We now generate
the file Makefile.features to record the configuration features. This is
because autoconf does not work properly with multiline substitutions.
</LI><LI CLASS="li-itemize"><B>July 21, 2006</B>: Cleaned up the printing of some Lvals. Things that
were printed as “(*i)” before are now printed simply as “*i” (no
parentheses). However, this means that when you use pLval to print lvalues
inside expressions, you must take care about parentheses yourself. Thanks to
Benjamin Monate for pointing this out.
</LI><LI CLASS="li-itemize"><B>July 21, 2006</B>: Added new hooks to the Usedef and
Dataflow.BackwardsTransfer APIs. Code that uses these will need to
be changed slightly. Also, updated the <A HREF="api/Cfg.html">Cfg</A> code to
handle noreturn functions.
</LI><LI CLASS="li-itemize"><B>July 17, 2006</B>: Fix parsing of attributes on bitfields and
empty attribute lists. Thanks to Peter Hawkins.
</LI><LI CLASS="li-itemize"><B>July 10, 2006</B>: Fix Makefile problem for FreeBSD. Thanks
to Joseph Koshy for the patch.
</LI><LI CLASS="li-itemize"><B>June 25, 2006</B>: Extended the inline assembly to support named
arguments, as added in gcc 3.0. This changes the types of the input
and output lists from “<TT>(string * lval) list</TT>” to
“<TT>(string option * string * lval) list</TT>”. Some existing code
will need to be modified accordingly.
</LI><LI CLASS="li-itemize"><B>June 11, 2006</B>: Removed the function <TT>Cil.foldLeftCompoundAll</TT>.
Use instead <A HREF="api/Cil.html#VALfoldLeftCompound">Cil.foldLeftCompound</A> with <TT> implicit:true</TT>.
</LI><LI CLASS="li-itemize"><B>June 9, 2006</B>: Extended the definition of the cilVisitor for
initializers to pass more information around. This might result in backward
incompatibilities with code that uses the visitor for initializers.
</LI><LI CLASS="li-itemize"><B>June 2, 2006</B>: Added <TT>--commPrintLnSparse</TT> flag.
</LI><LI CLASS="li-itemize"><B>June 1, 2006</B>: Christian Stork provided some fixes for the handling
of variable argument functions.
</LI><LI CLASS="li-itemize"><B>June 1, 2006</B>: Added support for x86 performance counters
on 64-bit processors. Thanks to tbergen for the patch.
</LI><LI CLASS="li-itemize"><B>May 23, 2006</B>: Benjamin Monate fixed a lexer bug when a
preprocessed file is missing a final newline.
</LI><LI CLASS="li-itemize"><B>May 23, 2006</B>: Fix for <TT>typeof(</TT><TT><I>e</I></TT><TT>)</TT> when <I>e</I> has type
<TT>void</TT>.
</LI><LI CLASS="li-itemize"><B>May 20, 2006</B>: <B>Released version 1.3.5</B> (Subversion
revision 8093)
</LI><LI CLASS="li-itemize"><B>May 19, 2006</B>: <TT>Makefile.cil.in</TT>/<TT>Makefile.cil</TT> have
been renamed <TT>Makefile.in</TT>/<TT>Makefile</TT>. And <TT>maincil.ml</TT> has
been renamed <TT>main.ml</TT>.
</LI><LI CLASS="li-itemize"><B>May 18, 2006</B>: Added a new module <A HREF="api/Cfg.html">Cfg</A> to compute the
control-flow graph. Unlike the older <A HREF="api/Cil.html#VALcomputeCFGInfo">Cil.computeCFGInfo</A>,
the new version does not modify the code.
</LI><LI CLASS="li-itemize"><B>May 18, 2006</B>: Added several new analyses: reaching
definitions, available expressions, liveness analysis, and dead code
elimination. See Section <A HREF="#sec-Extension">8</A>.
</LI><LI CLASS="li-itemize"><B>May 2, 2006</B>: Added a flag <TT>--noInsertImplicitCasts</TT>.
When this flag is used, CIL code will only include casts inserted by
the programmer. Implicit coercions are not changed to explicit casts.
</LI><LI CLASS="li-itemize"><B>April 16, 2006</B>: Minor improvements to the <TT>--stats</TT>
flag (Section <A HREF="#sec-cilly-asm-options">7.2</A>). We now use Pentium performance
counters by default, if your processor supports them.
</LI><LI CLASS="li-itemize"><B>April 10, 2006</B>: Extended <TT>machdep.c</TT> to support
microcontroller compilers where the struct alignment of integer
types does not match the size of the type. Thanks to Nathan
Cooprider for the patch.
</LI><LI CLASS="li-itemize"><B>April 6, 2006</B>: Fix for global initializers of unions when
the union field being initialized is not the first one, and for
missing initializers of unions when the first field is not the
largest field.
</LI><LI CLASS="li-itemize"><B>April 6, 2006</B>: Fix for bitfields in the SFI module.
</LI><LI CLASS="li-itemize"><B>April 6, 2006</B>: Various fixes for gcc attributes.
<TT>packed</TT>, <TT>section</TT>, and <TT>always_inline</TT> attributes are now
parsed correctly. Also fixed printing of attributes on enum types.
</LI><LI CLASS="li-itemize"><B>March 30, 2006</B>: Fix for <TT>rmtemps.ml</TT>, which deletes
unused inline functions. When in <TT>gcc</TT> mode CIL now leaves all
inline functions in place, since <TT>gcc</TT> treats these as externally
visible.
</LI><LI CLASS="li-itemize"><B>March 3, 2006</B>: Assume inline assembly instructions can
fall through for the purposes of adding return statements. Thanks to
Nathan Cooprider for the patch.
</LI><LI CLASS="li-itemize"><B>February 27, 2006</B>: Fix for extern inline functions when
the output of CIL is fed back into CIL.
</LI><LI CLASS="li-itemize"><B>January 30, 2006</B>: Fix parsing of <TT>switch</TT> without braces.
</LI><LI CLASS="li-itemize"><B>January 30, 2006</B>: Allow ‘$’ to appear in identifiers.
</LI><LI CLASS="li-itemize"><B>January 13, 2006</B>: Added support for gcc’s alias attribute
on functions. See Section <A HREF="#sec-ugly-gcc">16.2</A>, item 8.
</LI><LI CLASS="li-itemize"><B>December 9, 2005</B>: Christoph Spiel fixed the Golf and
Olf modules so that Golf can be used with the points-to analysis.
He also added performance fixes and cleaned up the documentation.
</LI><LI CLASS="li-itemize"><B>December 1, 2005</B>: Major rewrite of the ext/callgraph module.
</LI><LI CLASS="li-itemize"><B>December 1, 2005</B>: Preserve enumeration constants in CIL. Default
is the old behavior to replace them with integers.
</LI><LI CLASS="li-itemize"><B>November 30, 2005</B>: Added support for many GCC <TT>__builtin</TT>
functions.
</LI><LI CLASS="li-itemize"><B>November 30, 2005</B>: Added the EXTRAFEATURES configure
option, making it easier to add Features to the build process.
</LI><LI CLASS="li-itemize"><B>November 23, 2005</B>: In MSVC mode do not remove any locals whose name
appears as a substring in an inline assembly.
</LI><LI CLASS="li-itemize"><B>November 23, 2005</B>: Do not add a return to functions that have the
noreturn attribute.
</LI><LI CLASS="li-itemize"><B>November 22, 2005</B>: <B>Released version 1.3.4</B>
</LI><LI CLASS="li-itemize"><B>November 21, 2005</B>: Performance and correctness fixes for
the Points-to Analysis module. Thanks to Christoph Spiel for the
patches.
</LI><LI CLASS="li-itemize"><B>October 5, 2005</B>: CIL now builds on SPARC/Solaris. Thanks
to Nick Petroni and Remco van Engelen for the patches.
</LI><LI CLASS="li-itemize"><B>September 26, 2005</B>: CIL no longer uses the ‘<TT>-I-</TT>’ flag
by default when preprocessing with gcc.
</LI><LI CLASS="li-itemize"><B>August 24, 2005</B>: Added a command-line option
“--forceRLArgEval” that forces function arguments to be evaluated
right-to-left. This is the default behavior in unoptimized gcc and
MSVC, but the order of evaluation is undefined when using
optimizations, unless you apply this CIL transformation. This flag
does not affect the order of evaluation of e.g. binary operators,
which remains undefined. Thanks to Nathan Cooprider for the patch.
</LI><LI CLASS="li-itemize"><B>August 9, 2005</B>: Fixed merging when there are more than 20
input files.
</LI><LI CLASS="li-itemize"><B>August 3, 2005</B>: When merging, it is now an error to
declare the same global variable twice with different initializers.
</LI><LI CLASS="li-itemize"><B>July 27, 2005</B>: Fixed bug in transparent unions.
</LI><LI CLASS="li-itemize"><B>July 27, 2005</B>: Fixed bug in collectInitializer. Thanks to
Benjamin Monate for the patch.
</LI><LI CLASS="li-itemize"><B>July 26, 2005</B>: Better support for extended inline assembly
in gcc.
</LI><LI CLASS="li-itemize"><B>July 26, 2005</B>: Added many more gcc __builtin* functions
to CIL. Most are treated as Call instructions, but a few are
translated into expressions so that they can be used in global
initializers. For example, “<TT>__builtin_offsetof(t, field)</TT>” is
rewritten as “<TT>&((t*)0)->field</TT>”, the traditional way of calculating
an offset.
</LI><LI CLASS="li-itemize"><B>July 18, 2005</B>: Fixed bug in the constant folding of shifts
when the second argument was negative or too large.
</LI><LI CLASS="li-itemize"><B>July 18, 2005</B>: Fixed bug where casts were not always
inserted in function calls.
</LI><LI CLASS="li-itemize"><B>June 10, 2005</B>: Fixed bug in the code that makes implicit
returns explicit. We weren’t handling switch blocks correctly.
</LI><LI CLASS="li-itemize"><B>June 1, 2005</B>: <B>Released version 1.3.3</B>
</LI><LI CLASS="li-itemize"><B>May 31, 2005</B>: Fixed handling of noreturn attribute for function
pointers.
</LI><LI CLASS="li-itemize"><B>May 30, 2005</B>: Fixed bugs in the handling of constructors in gcc.
</LI><LI CLASS="li-itemize"><B>May 30, 2005</B>: Fixed bugs in the generation of global variable IDs.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Reimplemented the translation of function calls so
that we can intercept some builtins. This is important for the uses of
__builtin_constant_p in constants.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Export the plainCilPrinter, for debugging.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Fixed bug with printing of const attribute for
arrays.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Fixed bug in generation of type signatures. Now they
should not contain expressions anymore, so you can use structural equality.
This used to lead to Out_of_Memory exceptions.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Fixed bug in type comparisons using
TBuiltin_va_list.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Improved the constant folding in array lengths and
case expressions.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Added the <TT>__builtin_frame_address</TT> to the set
of gcc builtins.
</LI><LI CLASS="li-itemize"><B>May 27, 2005</B>: Added the CIL project to SourceForge.
</LI><LI CLASS="li-itemize"><B>April 23, 2005</B>: The cattr field was not visited.
</LI><LI CLASS="li-itemize"><B>March 6, 2005</B>: Debian packaging support
</LI><LI CLASS="li-itemize"><B>February 16, 2005</B>: Merger fixes.
</LI><LI CLASS="li-itemize"><B>February 11, 2005</B>: Fixed a bug in <TT>--dopartial</TT>. Thanks to
Nathan Cooprider for this fix.
</LI><LI CLASS="li-itemize"><B>January 31, 2005</B>: Make sure the input file is closed even if a
parsing error is encountered.
</LI><LI CLASS="li-itemize"><B>January 11, 2005</B>: <B>Released version 1.3.2</B>
</LI><LI CLASS="li-itemize"><B>January 11, 2005</B>: Fixed printing of integer constants whose
integer kind is shorter than an int.
</LI><LI CLASS="li-itemize"><B>January 11, 2005</B>: Added checks for negative size arrays and arrays
too big.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Added support for GCC attribute “volatile” for
tunctions (as a synonim for noreturn).
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Improved the comparison of array sizes when
comparing array types.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Fixed handling of shell metacharacters in the
cilly command lione.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Fixed dropping of cast in initialization of
local variable with the result of a function call.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Fixed some structural comparisons that were
broken in the Ocaml 3.08.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Fixed the <TT>unrollType</TT> function to not forget
attributes.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Better keeping track of locations of function
prototypes and definitions.
</LI><LI CLASS="li-itemize"><B>January 10, 2005</B>: Fixed bug with the expansion of enumeration
constants in attributes.
</LI><LI CLASS="li-itemize"><B>October 18, 2004</B>: Fixed a bug in cabsvisit.ml. CIl would wrap a
BLOCK around a single atom unnecessarily.
</LI><LI CLASS="li-itemize"><B>August 7, 2004</B>: <B>Released version 1.3.1</B>
</LI><LI CLASS="li-itemize"><B>August 4, 2004</B>: Fixed a bug in splitting of structs using
<TT>--dosimplify</TT>
</LI><LI CLASS="li-itemize"><B>July 29, 2004</B>: Minor changes to the type typeSig (type signatures)
to ensure that they do not contain types, so that you can do structural
comparison without danger of nontermination.
</LI><LI CLASS="li-itemize"><B>July 28, 2004</B>: Ocaml version 3.08 is required. Numerous small
changes while porting to Ocaml 3.08.
</LI><LI CLASS="li-itemize"><B>July 7, 2004</B>: <B>Released version 1.2.6</B>
</LI><LI CLASS="li-itemize"><B>July 2, 2004</B>: Character constants such as <TT>’c’</TT> should
have type <TT>int</TT>, not <TT>char</TT>. Added a utility function
<TT>Cil.charConstToInt</TT> that sign-extends chars greater than 128, if needed.
</LI><LI CLASS="li-itemize"><B>July 2, 2004</B>: Fixed a bug that was casting values to int
before applying the logical negation operator !. This caused
problems for floats, and for integer types bigger than <TT>int</TT>.
</LI><LI CLASS="li-itemize"><B>June 13, 2004</B>: Added the field <TT>sallstmts</TT> to a function
description, to hold all statements in the function.
</LI><LI CLASS="li-itemize"><B>June 13, 2004</B>: Added new extensions for data flow analyses, and
for computing dominators.
</LI><LI CLASS="li-itemize"><B>June 10, 2004</B>: Force initialization of CIL at the start of
Cabs2cil.
</LI><LI CLASS="li-itemize"><B>June 9, 2004</B>: Added support for GCC <TT>__attribute_used__</TT>
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: <B>Released version 1.2.5</B>
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: Allow now to run ./configure CC=cl and set the MSVC
compiler to be the default. The MSVC driver will now select the default name
of the .exe file like the CL compiler.
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: Fixed a bug in the driver. The temporary files are
deleted by the Perl script before the CL compiler gets to them?
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: Added the - form of arguments to the MSVC driver.
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: Added a few more GCC-specific string escapes, (, [,
{, %, E.
</LI><LI CLASS="li-itemize"><B>April 7, 2004</B>: Fixed bug with continuation lines in MSVC.
</LI><LI CLASS="li-itemize"><B>April 6, 2004</B>: Fixed embarassing bug in the parser: the precedence
of casts and unary operators was switched.
</LI><LI CLASS="li-itemize"><B>April 5, 2004</B>: Fixed a bug involving statements mixed between
declarations containing initializers. Now we make sure that the initializers
are run in the proper order with respect to the statements.
</LI><LI CLASS="li-itemize"><B>April 5, 2004</B>: Fixed a bug in the merger. The merger was keeping
separate alpha renaming talbes (namespaces) for variables and types. This
means that it might end up with a type and a variable named the same way, if
they come from different files, which breaks an important CIL invariant.
</LI><LI CLASS="li-itemize"><B>March 11, 2004</B> : Fixed a bug in the Cil.copyFunction function. The
new local variables were not getting fresh IDs.
</LI><LI CLASS="li-itemize"><B>March 5, 2004</B>: Fixed a bug in the handling of static function
prototypes in a block scope. They used to be renamed. Now we just consider
them global.
</LI><LI CLASS="li-itemize"><B>February 20, 2004</B>: <B>Released version 1.2.4</B>
</LI><LI CLASS="li-itemize"><B>February 15, 2004</B>: Changed the parser to allow extra semicolons
after field declarations.
</LI><LI CLASS="li-itemize"><B>February 14, 2004</B>: Changed the Errormsg functions: error, unimp,
bug to not raise an exception. Instead they just set Errormsg.hadErrors.
</LI><LI CLASS="li-itemize"><B>February 13, 2004</B>: Change the parsing of attributes to recognize
enumeration constants.
</LI><LI CLASS="li-itemize"><B>February 10, 2004</B>: In some versions of <TT>gcc</TT> the identifier
_{thread is an identifier and in others it is a keyword. Added code
during configuration to detect which is the case.
</LI><LI CLASS="li-itemize"><B>January 7, 2004</B>: <B>Released version 1.2.3</B>
</LI><LI CLASS="li-itemize"><B>January 7, 2004</B>: Changed the alpha renamer to be less
conservative. It will remember all versions of a name that were seen and will
only create a new name if we have not seen one.
</LI><LI CLASS="li-itemize"><B>December 30, 2003</B> : Extended the <TT>cilly</TT> command to understand
better linker command options <TT>-lfoo</TT>.
</LI><LI CLASS="li-itemize"><B>December 5, 2003</B>: Added markup commands to the pretty-printer
module. Also, changed the “@<” left-flush command into “@'’.
</LI><LI CLASS="li-itemize"><B>December 4, 2003</B>: Wide string literals are now handled
directly by Cil (rather than being exploded into arrays). This is
apparently handy for Microsoft Device Driver APIs that use intrinsic
functions that require literal constant wide-string arguments.
</LI><LI CLASS="li-itemize"><B>December 3, 2003</B>: Added support for structured exception handling
extensions for the Microsoft compilers.
</LI><LI CLASS="li-itemize"><B>December 1, 2003</B>: Fixed a Makefile bug in the generation of the
Cil library (e.g., <TT>cil.cma</TT>) that was causing it to be unusable. Thanks
to KEvin Millikin for pointing out this bug.
</LI><LI CLASS="li-itemize"><B>November 26, 2003</B>: Added support for linkage specifications
(extern “C”).
</LI><LI CLASS="li-itemize"><B>November 26, 2003</B>: Added the ocamlutil directory to contain some
utilities shared with other projects.
</LI><LI CLASS="li-itemize"><B>November 25, 2003</B>: <B>Released version 1.2.2</B>
</LI><LI CLASS="li-itemize"><B>November 24, 2003</B>: Fixed a bug that allowed a static local to
conflict with a global with the same name that is declared later in the
file.
</LI><LI CLASS="li-itemize"><B>November 24, 2003</B>: Removed the <TT>--keep</TT> option of the <TT>cilly</TT>
driver and replaced it with <TT>--save-temps</TT>.
</LI><LI CLASS="li-itemize"><B>November 24, 2003</B>: Added printing of what CIL features are being
run.
</LI><LI CLASS="li-itemize"><B>November 24, 2003</B>: Fixed a bug that resulted in attributes being
dropped for integer types.
</LI><LI CLASS="li-itemize"><B>November 11, 2003</B>: Fixed a bug in the visitor for enumeration
definitions.
</LI><LI CLASS="li-itemize"><B>October 24, 2003</B>: Fixed a problem in the configuration script. It
was not recognizing the Ocaml version number for beta versions.
</LI><LI CLASS="li-itemize"><B>October 15, 2003</B>: Fixed a problem in version 1.2.1 that was
preventing compilation on OCaml 3.04.
</LI><LI CLASS="li-itemize"><B>September 17, 2003: Released version 1.2.1.</B>
</LI><LI CLASS="li-itemize"><B>September 7, 2003</B>: Redesigned the interface for choosing
<TT>#line</TT> directive printing styles. Cil.printLn and
Cil.printLnComment have been merged into Cil.lineDirectiveStyle.
</LI><LI CLASS="li-itemize"><B>August 8, 2003</B>: Do not silently pad out functions calls with
arguments to match the prototype.
</LI><LI CLASS="li-itemize"><B>August 1, 2003</B>: A variety of fixes suggested by Steve Chamberlain:
initializers for externs, prohibit float literals in enum, initializers for
unsized arrays were not working always, an overflow problem in Ocaml, changed
the processing of attributes before struct specifiers</LI><LI CLASS="li-itemize"><B>July 14, 2003</B>: Add basic support for GCC’s "__thread" storage
qualifier. If given, it will appear as a "thread" attribute at the top of the
type of the declared object. Treatment is very similar to "__declspec(...)"
in MSVC</LI><LI CLASS="li-itemize"><B>July 8, 2003</B>: Fixed some of the __alignof computations. Fixed
bug in the designated initializers for arrays (Array.get error).
</LI><LI CLASS="li-itemize"><B>July 8, 2003</B>: Fixed infinite loop bug (Stack Overflow) in the
visitor for __alignof.
</LI><LI CLASS="li-itemize"><B>July 8, 2003</B>: Fixed bug in the conversion to CIL. A function or
array argument of
the GCC __typeof() was being converted to pointer type. Instead, it should
be left alone, just like for sizeof.
</LI><LI CLASS="li-itemize"><B>July 7, 2003</B>: New Escape module provides utility functions
for escaping characters and strings in accordance with C lexical
rules.</LI><LI CLASS="li-itemize"><B>July 2, 2003</B>: Relax CIL’s rules for when two enumeration types are
considered compatible. Previously CIL considered two enums to be compatible if
they were the same enum. Now we follow the C99 standard.</LI><LI CLASS="li-itemize"><B>June 28, 2003</B>: In the Formatparse module, Eric Haugh found and
fixed a bug in the handling of lvalues of the form “lv->field.more”.</LI><LI CLASS="li-itemize"><B>June 28, 2003</B>: Extended the handling of gcc command lines
arguments in the Perl scripts. </LI><LI CLASS="li-itemize"><B>June 23, 2003</B>: In Rmtmps module, simplified the API for
customizing the root set. Clients may supply a predicate that
returns true for each root global. Modifying various
“<TT>referenced</TT>” fields directly is no longer supported.</LI><LI CLASS="li-itemize"><B>June 17, 2003</B>: Reimplement internal utility routine
<TT>Cil.escape_char</TT>. Faster and better. </LI><LI CLASS="li-itemize"><B>June 14, 2003</B>: Implemented support for <TT>__attribute__s</TT>
appearing between "struct" and the struct tag name (also for unions and
enums), since gcc supports this as documented in section 4.30 of the gcc
(2.95.3) manual</LI><LI CLASS="li-itemize"><B>May 30, 2003</B>: Released the regression tests.
</LI><LI CLASS="li-itemize"><B>May 28, 2003</B>: <B>Released version 1.1.2</B>
</LI><LI CLASS="li-itemize"><B>May 26, 2003</B>: Add the <TT>simplify</TT> module that compiles CIL
expressions into simpler expressions, similar to those that appear in a
3-address intermediate language.
</LI><LI CLASS="li-itemize"><B>May 26, 2003</B>: Various fixes and improvements to the pointer
analysis modules.
</LI><LI CLASS="li-itemize"><B>May 26, 2003</B>: Added optional consistency checking for
transformations.
</LI><LI CLASS="li-itemize"><B>May 25, 2003</B>: Added configuration support for big endian machines.
Now <A HREF="api/Cil.html#VALlittle_endian">Cil.little_endian</A> can be used to test whether the machine is
little endian or not.
</LI><LI CLASS="li-itemize"><B>May 22, 2003</B>: Fixed a bug in the handling of inline functions. The
CIL merger used to turn these functions into “static”, which is incorrect.
</LI><LI CLASS="li-itemize"><B>May 22, 2003</B>: Expanded the CIL consistency checker to verify
undesired sharing relationships between data structures.
</LI><LI CLASS="li-itemize"><B>May 22, 2003</B>: Fixed bug in the <TT>oneret</TT> CIL module: it was
mishandling certain labeled return statements.
</LI><LI CLASS="li-itemize"><B>May 5, 2003</B>: <B>Released version 1.0.11</B>
</LI><LI CLASS="li-itemize"><B>May 5, 2003</B>: OS X (powerpc/darwin) support for CIL. Special
thanks to Jeff Foster, Andy Begel and Tim Leek.
</LI><LI CLASS="li-itemize"><B>April 30, 2003</B>: Better description of how to use CIL for your
analysis.
</LI><LI CLASS="li-itemize"><B>April 28, 2003</B>: Fixed a bug with <TT>--dooneRet</TT> and
<TT>--doheapify</TT>. Thanks, Manos Renieris.
</LI><LI CLASS="li-itemize"><B>April 16, 2003</B>: Reworked management of
temporary/intermediate output files in Perl driver scripts. Default
behavior is now to remove all such files. To keep intermediate
files, use one of the following existing flags:
<UL CLASS="itemize"><LI CLASS="li-itemize">
<TT>--keepmerged</TT> for the single-file merge of all sources
</LI><LI CLASS="li-itemize"><TT>--keep=<</TT><TT><I>dir</I></TT><TT>></TT> for various other CIL and
CCured output files
</LI><LI CLASS="li-itemize"><TT>--save-temps</TT> for various gcc intermediate files; MSVC
has no equivalent option
</LI></UL>
As part of this change, some intermediate files have changed their
names slightly so that new suffixes are always preceded by a
period. For example, CCured output that used to appear in
“<TT>foocured.c</TT>” now appears in “<TT>foo.cured.c</TT>”.
</LI><LI CLASS="li-itemize"><B>April 7, 2003</B>: Changed the representation of the <A HREF="api/Cil.html#VALGVar">Cil.GVar</A>
global constructor. Now it is possible to update the initializer without
reconstructing the global (which in turn it would require reconstructing the
list of globals that make up a program). We did this because it is often
tempting to use <A HREF="api/Cil.html#VALvisitCilFileSameGlobals">Cil.visitCilFileSameGlobals</A> and the <A HREF="api/Cil.html#VALGVar">Cil.GVar</A>
was the only global that could not be updated in place.
</LI><LI CLASS="li-itemize"><B>April 6, 2003</B>: Reimplemented parts of the cilly.pl script to make
it more robust in the presence of complex compiler arguments.
</LI><LI CLASS="li-itemize"><B>March 10, 2003</B>: <B>Released version 1.0.9</B>
</LI><LI CLASS="li-itemize"><B>March 10, 2003</B>: Unified and documented a large number of CIL
Library Modules: oneret, simplemem, makecfg, heapify, stackguard, partial.
Also documented the main client interface for the pointer analysis.
</LI><LI CLASS="li-itemize"><B>February 18, 2003</B>: Fixed a bug in logwrites that was causing it
to produce invalid C code on writes to bitfields. Thanks, David Park.
</LI><LI CLASS="li-itemize"><B>February 15, 2003</B>: <B>Released version 1.0.8</B>
</LI><LI CLASS="li-itemize"><B>February 15, 2003</B>: PDF versions of the manual and API are
available for those who would like to print them out.
</LI><LI CLASS="li-itemize"><B>February 14, 2003</B>: CIL now comes bundled with alias analyses.
</LI><LI CLASS="li-itemize"><B>February 11, 2003</B>: Added support for adding/removing options from
<TT>./configure</TT>.
</LI><LI CLASS="li-itemize"><B>February 3, 2003</B>: <B>Released version 1.0.7</B>
</LI><LI CLASS="li-itemize"><B>February 1, 2003</B>: Some bug fixes in the handling of variable
argument functions in new versions of <TT>gcc</TT> And <TT>glibc</TT>.
</LI><LI CLASS="li-itemize"><B>January 29, 2003</B>: Added the logical AND and OR operators.
Exapanded the translation to CIL to handle more complicated initializers
(including those that contain logical operators).
</LI><LI CLASS="li-itemize"><B>January 28, 2003</B>: <B>Released version 1.0.6</B>
</LI><LI CLASS="li-itemize"><B>January 28, 2003</B>: Added support for the new handling of
variable-argument functions in new versions of <TT>glibc</TT>.
</LI><LI CLASS="li-itemize"><B>January 19, 2003</B>: Added support for declarations in interpreted
constructors. Relaxed the semantics of the patterns for variables.
</LI><LI CLASS="li-itemize"><B>January 17, 2003</B>: Added built-in prototypes for the gcc built-in
functions. Changed the <TT>pGlobal</TT> method in the printers to print the
carriage return as well.
</LI><LI CLASS="li-itemize"><B>January 9, 2003</B>: Reworked lexer and parser’s strategy for
tracking source file names and line numbers to more closely match
typical native compiler behavior. The visible CIL interface is
unchanged.
</LI><LI CLASS="li-itemize"><B>January 9, 2003</B>: Changed the interface to the alpha convertor. Now
you can pass a list where it will record undo information that you can use to
revert the changes that it makes to the scope tables.
</LI><LI CLASS="li-itemize"><B>January 6, 2003</B>: <B>Released version 1.0.5</B>
</LI><LI CLASS="li-itemize"><B>January 4, 2003</B>: Changed the interface for the Formatcil module.
Now the placeholders in the pattern have names. Also expanded the
documentation of the Formatcil module.
Now the placeholders in the pattern have names.
</LI><LI CLASS="li-itemize"><B>January 3, 2003</B>: Extended the <TT>rmtmps</TT> module to also remove
unused labels that are generated in the conversion to CIL. This reduces the
number of warnings that you get from <TT>cgcc</TT> afterwards.
</LI><LI CLASS="li-itemize"><B>December 17, 2002</B>: Fixed a few bugs in CIL related to the
representation of string literals. The standard says that a string literal
is an array. In CIL, a string literal has type pointer to character. This is
Ok, except as an argument of sizeof. To support this exception, we have
added to CIL the expression constructor SizeOfStr. This allowed us to fix
bugs with computing <TT>sizeof("foo bar")</TT> and <TT>sizeof((char*)"foo bar")</TT>
(the former is 8 and the latter is 4).</LI><LI CLASS="li-itemize"><B>December 8, 2002</B>: Fixed a few bugs in the lexer and parser
relating to hex and octal escapes in string literals. Also fixed
the dependencies between the lexer and parser.
</LI><LI CLASS="li-itemize"><B>December 5, 2002</B>: Fixed visitor bugs that were causing
some attributes not to be visited and some queued instructions to be
dropped.
</LI><LI CLASS="li-itemize"><B>December 3, 2002</B>: Added a transformation to catch stack
overflows. Fixed the heapify transformation.
</LI><LI CLASS="li-itemize"><B>October 14, 2002</B>: CIL is now available under the BSD license
(see the License section or the file LICENSE). <B>Released version 1.0.4</B>
</LI><LI CLASS="li-itemize"><B>October 9, 2002</B>: More FreeBSD configuration changes, support
for the GCC-ims <TT>__signed</TT> and <TT>__volatile</TT>. Thanks to Axel
Simon for pointing out these problems. <B>Released version 1.0.3</B>
</LI><LI CLASS="li-itemize"><B>October 8, 2002</B>: FreeBSD configuration and porting fixes.
Thanks to Axel Simon for pointing out these problems.
</LI><LI CLASS="li-itemize"><B>September 10, 2002</B>: Fixed bug in conversion to CIL. Now we drop
all “const” qualifiers from the types of locals, even from the fields of
local structures or elements of arrays.
</LI><LI CLASS="li-itemize"><B>September 7, 2002</B>: Extended visitor interface to distinguish visitng
offsets inside lvalues from offsets inside initializer lists.
</LI><LI CLASS="li-itemize"><B>September 7, 2002</B>: <B>Released version 1.0.1</B>
</LI><LI CLASS="li-itemize"><B>September 6, 2002</B>: Extended the patcher with the <TT>ateof</TT> flag.
</LI><LI CLASS="li-itemize"><B>September 4, 2002</B>: Fixed bug in the elaboration to CIL. In some
cases constant folding of <TT>||</TT> and <TT>&&</TT> was computed wrong.
</LI><LI CLASS="li-itemize"><B>September 3, 2002</B>: Fixed the merger documentation.
</LI><LI CLASS="li-itemize"><B>August 29, 2002</B>: <B>Released version 1.0.0.</B>
</LI><LI CLASS="li-itemize"><B>August 29, 2002</B>: Started numbering versions with a major nubmer,
minor and revisions. Released version 1.0.0.
</LI><LI CLASS="li-itemize"><B>August 25, 2002</B>: Fixed the implementation of the unique
identifiers for global variables and composites. Now those identifiers are
globally unique.
</LI><LI CLASS="li-itemize"><B>August 24, 2002</B>: Added to the machine-dependent configuration the
<TT>sizeof</TT><TT>void</TT>. It is 1 on gcc and 0 on MSVC. Extended the implementation of
<TT>Cil.bitsSizeOf</TT> to handle this (it was previously returning an error when
trying to compute the size of <TT>void</TT>).
</LI><LI CLASS="li-itemize"><B>August 24, 2002</B>: Changed the representation of structure and
unions to distinguish between undefined structures and those that are defined
to be empty (allowed on gcc). The sizeof operator is undefined for the former
and returns 0 for the latter.
</LI><LI CLASS="li-itemize"><B>August 22, 2002</B>: Apply a patch from Richard H. Y. to support
FreeBSD installations. Thanks, Richard!
</LI><LI CLASS="li-itemize"><B>August 12, 2002</B>: Fixed a bug in the translation of wide-character
strings. Now this translation matches that of the underlying compiler. Changed
the implementation of the compiler dependencies.
</LI><LI CLASS="li-itemize"><B>May 25, 2002</B>: Added interpreted constructors and destructors.
</LI><LI CLASS="li-itemize"><B>May 17, 2002</B>: Changed the representation of functions to move the
“inline” information to the varinfo. This way we can print the “inline”
even in declarations which is what gcc does.
</LI><LI CLASS="li-itemize"><B>May 15, 2002</B>: Changed the visitor for initializers to make two
tail-recursive passes (the second is a <TT>List.rev</TT> and only done if one of
the initializers change). This prevents <TT>Stack_Overflow</TT> for large
initializers. Also improved the processing of initializers when converting to
CIL.
</LI><LI CLASS="li-itemize"><B>May 15, 2002</B>: Changed the front-end to allow the use of <TT>MSVC</TT>
mode even on machines that do not have MSVC. The machine-dependent parameters
for GCC will be used in that case.
</LI><LI CLASS="li-itemize"><B>May 11, 2002</B>: Changed the representation of formals in function
types. Now the function type is purely functional.
</LI><LI CLASS="li-itemize"><B>May 4, 2002</B>: Added the function
<A HREF="api/Cil.html#VALvisitCilFileSameGlobals">Cil.visitCilFileSameGlobals</A> and changed <A HREF="api/Cil.html#VALvisitCilFile">Cil.visitCilFile</A> to be
tail recursive. This prevents stack overflow on huge files.
</LI><LI CLASS="li-itemize"><B>February 28, 2002</B>: Changed the significance of the
<TT>CompoundInit</TT> in <A HREF="api/Cil.html#TYPEinit">Cil.init</A> to allow for missing initializers at the
end of an array initializer. Added the API function
<A HREF="api/Cil.html#VALfoldLeftCompoundAll">Cil.foldLeftCompoundAll</A>.
</LI></UL><!--CUT END -->
<!--HTMLFOOT-->
<!--ENDHTML-->
<!--FOOTER-->
<HR SIZE=2><BLOCKQUOTE CLASS="quote"><EM>This document was translated from L<sup>A</sup>T<sub>E</sub>X by
</EM><A HREF="http://hevea.inria.fr/index.html"><EM>H</EM><EM><FONT SIZE=2><sup>E</sup></FONT></EM><EM>V</EM><EM><FONT SIZE=2><sup>E</sup></FONT></EM><EM>A</EM></A><EM>.</EM></BLOCKQUOTE></BODY>
</HTML>
distrib
>
Fedora
>
13
>
i386
>
media
>
os
>
by-pkgid
>
82156e51e136cf1c3e3b0cfb61ec2f5a
>
files
>
70
