Blame - docs/ProgrammersManual.html - fp2-dev/platform/external/llvm

blob: e2918c0438fc460d5851d9e880ea24a6618c1571 [file] [log] [blame]

Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1	<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
				2	"http://www.w3.org/TR/html4/strict.dtd">
				3	<html>
				4	<head>
				5	<title>LLVM Programmer's Manual</title>
				6	<link rel="stylesheet" href="llvm.css" type="text/css">
				7	</head>
				8	<body>
				9
				10	<div class="doc_title">
				11	LLVM Programmer's Manual
				12	</div>
				13
				14	<ol>
				15	<li><a href="#introduction">Introduction</a></li>
				16	<li><a href="#general">General Information</a>
				17	<ul>
				18	<li><a href="#stl">The C++ Standard Template Library</a></li>
				19	<!--
				20	<li>The <tt>-time-passes</tt> option</li>
				21	<li>How to use the LLVM Makefile system</li>
				22	<li>How to write a regression test</li>
				23
				24	-->
				25	</ul>
				26	</li>
				27	<li><a href="#apis">Important and useful LLVM APIs</a>
				28	<ul>
				29	<li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
				30	and <tt>dyn_cast<></tt> templates</a> </li>
				31	<li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
				32	option</a>
				33	<ul>
				34	<li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
				35	and the <tt>-debug-only</tt> option</a> </li>
				36	</ul>
				37	</li>
				38	<li><a href="#Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
				39	option</a></li>
				40	<!--
				41	<li>The <tt>InstVisitor</tt> template
				42	<li>The general graph API
				43	-->
				44	<li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
				45	</ul>
				46	</li>
				47	<li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
				48	<ul>
				49	<li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
				50	<ul>
				51	<li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
				52	<li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
				53	<li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
				54	<li><a href="#dss_vector"><vector></a></li>
				55	<li><a href="#dss_deque"><deque></a></li>
				56	<li><a href="#dss_list"><list></a></li>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame^]	57	<li><a href="#dss_ilist">llvm/ADT/ilist.h</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	58	<li><a href="#dss_other">Other Sequential Container Options</a></li>
				59	</ul></li>
				60	<li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
				61	<ul>
				62	<li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
				63	<li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
				64	<li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattner	77ab46d	2007-09-30 00:58:59 +0000	[diff] [blame]	65	<li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	66	<li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
				67	<li><a href="#dss_set"><set></a></li>
				68	<li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
				69	<li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
				70	<li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
				71	</ul></li>
				72	<li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
				73	<ul>
				74	<li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
				75	<li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
				76	<li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
				77	<li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
				78	<li><a href="#dss_map"><map></a></li>
				79	<li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
				80	</ul></li>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	81	<li><a href="#ds_bit">BitVector-like containers</a>
				82	<ul>
				83	<li><a href="#dss_bitvector">A dense bitvector</a></li>
				84	<li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
				85	</ul></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	86	</ul>
				87	</li>
				88	<li><a href="#common">Helpful Hints for Common Operations</a>
				89	<ul>
				90	<li><a href="#inspection">Basic Inspection and Traversal Routines</a>
				91	<ul>
				92	<li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
				93	in a <tt>Function</tt></a> </li>
				94	<li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
				95	in a <tt>BasicBlock</tt></a> </li>
				96	<li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
				97	in a <tt>Function</tt></a> </li>
				98	<li><a href="#iterate_convert">Turning an iterator into a
				99	class pointer</a> </li>
				100	<li><a href="#iterate_complex">Finding call sites: a more
				101	complex example</a> </li>
				102	<li><a href="#calls_and_invokes">Treating calls and invokes
				103	the same way</a> </li>
				104	<li><a href="#iterate_chains">Iterating over def-use &
				105	use-def chains</a> </li>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	106	<li><a href="#iterate_preds">Iterating over predecessors &
				107	successors of blocks</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	108	</ul>
				109	</li>
				110	<li><a href="#simplechanges">Making simple changes</a>
				111	<ul>
				112	<li><a href="#schanges_creating">Creating and inserting new
				113	<tt>Instruction</tt>s</a> </li>
				114	<li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
				115	<li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
				116	with another <tt>Value</tt></a> </li>
				117	<li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
				118	</ul>
				119	</li>
				120	<!--
				121	<li>Working with the Control Flow Graph
				122	<ul>
				123	<li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
				124	<li>
				125	<li>
				126	</ul>
				127	-->
				128	</ul>
				129	</li>
				130
				131	<li><a href="#advanced">Advanced Topics</a>
				132	<ul>
				133	<li><a href="#TypeResolve">LLVM Type Resolution</a>
				134	<ul>
				135	<li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
				136	<li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
				137	<li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
				138	<li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
				139	</ul></li>
				140
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	141	<li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes</a></li>
				142	<li><a href="#UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	143	</ul></li>
				144
				145	<li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
				146	<ul>
				147	<li><a href="#Type">The <tt>Type</tt> class</a> </li>
				148	<li><a href="#Module">The <tt>Module</tt> class</a></li>
				149	<li><a href="#Value">The <tt>Value</tt> class</a>
				150	<ul>
				151	<li><a href="#User">The <tt>User</tt> class</a>
				152	<ul>
				153	<li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
				154	<li><a href="#Constant">The <tt>Constant</tt> class</a>
				155	<ul>
				156	<li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
				157	<ul>
				158	<li><a href="#Function">The <tt>Function</tt> class</a></li>
				159	<li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
				160	</ul>
				161	</li>
				162	</ul>
				163	</li>
				164	</ul>
				165	</li>
				166	<li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
				167	<li><a href="#Argument">The <tt>Argument</tt> class</a></li>
				168	</ul>
				169	</li>
				170	</ul>
				171	</li>
				172	</ol>
				173
				174	<div class="doc_author">
				175	<p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
				176	<a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	177	<a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
				178	<a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a> and
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	179	<a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
				180	</div>
				181
				182	<!-- *********************************************************************** -->
				183	<div class="doc_section">
				184	<a name="introduction">Introduction </a>
				185	</div>
				186	<!-- *********************************************************************** -->
				187
				188	<div class="doc_text">
				189
				190	<p>This document is meant to highlight some of the important classes and
				191	interfaces available in the LLVM source-base. This manual is not
				192	intended to explain what LLVM is, how it works, and what LLVM code looks
				193	like. It assumes that you know the basics of LLVM and are interested
				194	in writing transformations or otherwise analyzing or manipulating the
				195	code.</p>
				196
				197	<p>This document should get you oriented so that you can find your
				198	way in the continuously growing source code that makes up the LLVM
				199	infrastructure. Note that this manual is not intended to serve as a
				200	replacement for reading the source code, so if you think there should be
				201	a method in one of these classes to do something, but it's not listed,
				202	check the source. Links to the <a href="/doxygen/">doxygen</a> sources
				203	are provided to make this as easy as possible.</p>
				204
				205	<p>The first section of this document describes general information that is
				206	useful to know when working in the LLVM infrastructure, and the second describes
				207	the Core LLVM classes. In the future this manual will be extended with
				208	information describing how to use extension libraries, such as dominator
				209	information, CFG traversal routines, and useful utilities like the <tt><a
				210	href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
				211
				212	</div>
				213
				214	<!-- *********************************************************************** -->
				215	<div class="doc_section">
				216	<a name="general">General Information</a>
				217	</div>
				218	<!-- *********************************************************************** -->
				219
				220	<div class="doc_text">
				221
				222	<p>This section contains general information that is useful if you are working
				223	in the LLVM source-base, but that isn't specific to any particular API.</p>
				224
				225	</div>
				226
				227	<!-- ======================================================================= -->
				228	<div class="doc_subsection">
				229	<a name="stl">The C++ Standard Template Library</a>
				230	</div>
				231
				232	<div class="doc_text">
				233
				234	<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
				235	perhaps much more than you are used to, or have seen before. Because of
				236	this, you might want to do a little background reading in the
				237	techniques used and capabilities of the library. There are many good
				238	pages that discuss the STL, and several books on the subject that you
				239	can get, so it will not be discussed in this document.</p>
				240
				241	<p>Here are some useful links:</p>
				242
				243	<ol>
				244
				245	<li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
				246	reference</a> - an excellent reference for the STL and other parts of the
				247	standard C++ library.</li>
				248
				249	<li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
				250	O'Reilly book in the making. It has a decent
				251	Standard Library
				252	Reference that rivals Dinkumware's, and is unfortunately no longer free since the book has been
				253	published.</li>
				254
				255	<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
				256	Questions</a></li>
				257
				258	<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
				259	Contains a useful <a
				260	href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
				261	STL</a>.</li>
				262
				263	<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
				264	Page</a></li>
				265
				266	<li><a href="http://64.78.49.204/">
				267	Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
				268	the book).</a></li>
				269
				270	</ol>
				271
				272	<p>You are also encouraged to take a look at the <a
				273	href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
				274	to write maintainable code more than where to put your curly braces.</p>
				275
				276	</div>
				277
				278	<!-- ======================================================================= -->
				279	<div class="doc_subsection">
				280	<a name="stl">Other useful references</a>
				281	</div>
				282
				283	<div class="doc_text">
				284
				285	<ol>
				286	<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
				287	Branch and Tag Primer</a></li>
				288	<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
				289	static and shared libraries across platforms</a></li>
				290	</ol>
				291
				292	</div>
				293
				294	<!-- *********************************************************************** -->
				295	<div class="doc_section">
				296	<a name="apis">Important and useful LLVM APIs</a>
				297	</div>
				298	<!-- *********************************************************************** -->
				299
				300	<div class="doc_text">
				301
				302	<p>Here we highlight some LLVM APIs that are generally useful and good to
				303	know about when writing transformations.</p>
				304
				305	</div>
				306
				307	<!-- ======================================================================= -->
				308	<div class="doc_subsection">
				309	<a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
				310	<tt>dyn_cast<></tt> templates</a>
				311	</div>
				312
				313	<div class="doc_text">
				314
				315	<p>The LLVM source-base makes extensive use of a custom form of RTTI.
				316	These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
				317	operator, but they don't have some drawbacks (primarily stemming from
				318	the fact that <tt>dynamic_cast<></tt> only works on classes that
				319	have a v-table). Because they are used so often, you must know what they
				320	do and how they work. All of these templates are defined in the <a
				321	href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
				322	file (note that you very rarely have to include this file directly).</p>
				323
				324	<dl>
				325	<dt><tt>isa<></tt>: </dt>
				326
				327	<dd><p>The <tt>isa<></tt> operator works exactly like the Java
				328	"<tt>instanceof</tt>" operator. It returns true or false depending on whether
				329	a reference or pointer points to an instance of the specified class. This can
				330	be very useful for constraint checking of various sorts (example below).</p>
				331	</dd>
				332
				333	<dt><tt>cast<></tt>: </dt>
				334
				335	<dd><p>The <tt>cast<></tt> operator is a "checked cast" operation. It
Chris Lattner	1d5610a	2008-06-20 05:03:17 +0000	[diff] [blame]	336	converts a pointer or reference from a base class to a derived class, causing
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	337	an assertion failure if it is not really an instance of the right type. This
				338	should be used in cases where you have some information that makes you believe
				339	that something is of the right type. An example of the <tt>isa<></tt>
				340	and <tt>cast<></tt> template is:</p>
				341
				342	<div class="doc_code">
				343	<pre>
				344	static bool isLoopInvariant(const <a href="#Value">Value</a> V, const Loop L) {
				345	if (isa<<a href="#Constant">Constant</a>>(V) \|\| isa<<a href="#Argument">Argument</a>>(V) \|\| isa<<a href="#GlobalValue">GlobalValue</a>>(V))
				346	return true;
				347
				348	// <i>Otherwise, it must be an instruction...</i>
				349	return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
				350	}
				351	</pre>
				352	</div>
				353
				354	<p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
				355	by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
				356	operator.</p>
				357
				358	</dd>
				359
				360	<dt><tt>dyn_cast<></tt>:</dt>
				361
				362	<dd><p>The <tt>dyn_cast<></tt> operator is a "checking cast" operation.
				363	It checks to see if the operand is of the specified type, and if so, returns a
				364	pointer to it (this operator does not work with references). If the operand is
				365	not of the correct type, a null pointer is returned. Thus, this works very
				366	much like the <tt>dynamic_cast<></tt> operator in C++, and should be
				367	used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
				368	operator is used in an <tt>if</tt> statement or some other flow control
				369	statement like this:</p>
				370
				371	<div class="doc_code">
				372	<pre>
				373	if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
				374	// <i>...</i>
				375	}
				376	</pre>
				377	</div>
				378
				379	<p>This form of the <tt>if</tt> statement effectively combines together a call
				380	to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
				381	statement, which is very convenient.</p>
				382
				383	<p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
				384	<tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
				385	abused. In particular, you should not use big chained <tt>if/then/else</tt>
				386	blocks to check for lots of different variants of classes. If you find
				387	yourself wanting to do this, it is much cleaner and more efficient to use the
				388	<tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
				389
				390	</dd>
				391
				392	<dt><tt>cast_or_null<></tt>: </dt>
				393
				394	<dd><p>The <tt>cast_or_null<></tt> operator works just like the
				395	<tt>cast<></tt> operator, except that it allows for a null pointer as an
				396	argument (which it then propagates). This can sometimes be useful, allowing
				397	you to combine several null checks into one.</p></dd>
				398
				399	<dt><tt>dyn_cast_or_null<></tt>: </dt>
				400
				401	<dd><p>The <tt>dyn_cast_or_null<></tt> operator works just like the
				402	<tt>dyn_cast<></tt> operator, except that it allows for a null pointer
				403	as an argument (which it then propagates). This can sometimes be useful,
				404	allowing you to combine several null checks into one.</p></dd>
				405
				406	</dl>
				407
				408	<p>These five templates can be used with any classes, whether they have a
				409	v-table or not. To add support for these templates, you simply need to add
				410	<tt>classof</tt> static methods to the class you are interested casting
				411	to. Describing this is currently outside the scope of this document, but there
				412	are lots of examples in the LLVM source base.</p>
				413
				414	</div>
				415
				416	<!-- ======================================================================= -->
				417	<div class="doc_subsection">
				418	<a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
				419	</div>
				420
				421	<div class="doc_text">
				422
				423	<p>Often when working on your pass you will put a bunch of debugging printouts
				424	and other code into your pass. After you get it working, you want to remove
				425	it, but you may need it again in the future (to work out new bugs that you run
				426	across).</p>
				427
				428	<p> Naturally, because of this, you don't want to delete the debug printouts,
				429	but you don't want them to always be noisy. A standard compromise is to comment
				430	them out, allowing you to enable them if you need them in the future.</p>
				431
				432	<p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
				433	file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
				434	this problem. Basically, you can put arbitrary code into the argument of the
				435	<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
				436	tool) is run with the '<tt>-debug</tt>' command line argument:</p>
				437
				438	<div class="doc_code">
				439	<pre>
				440	DOUT << "I am here!\n";
				441	</pre>
				442	</div>
				443
				444	<p>Then you can run your pass like this:</p>
				445
				446	<div class="doc_code">
				447	<pre>
				448	$ opt < a.bc > /dev/null -mypass
				449	<i><no output></i>
				450	$ opt < a.bc > /dev/null -mypass -debug
				451	I am here!
				452	</pre>
				453	</div>
				454
				455	<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
				456	to not have to create "yet another" command line option for the debug output for
				457	your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
				458	so they do not cause a performance impact at all (for the same reason, they
				459	should also not contain side-effects!).</p>
				460
				461	<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
				462	enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
				463	"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
				464	program hasn't been started yet, you can always just run it with
				465	<tt>-debug</tt>.</p>
				466
				467	</div>
				468
				469	<!-- _______________________________________________________________________ -->
				470	<div class="doc_subsubsection">
				471	<a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
				472	the <tt>-debug-only</tt> option</a>
				473	</div>
				474
				475	<div class="doc_text">
				476
				477	<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
				478	just turns on <b>too much</b> information (such as when working on the code
				479	generator). If you want to enable debug information with more fine-grained
				480	control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
				481	option as follows:</p>
				482
				483	<div class="doc_code">
				484	<pre>
				485	DOUT << "No debug type\n";
				486	#undef DEBUG_TYPE
				487	#define DEBUG_TYPE "foo"
				488	DOUT << "'foo' debug type\n";
				489	#undef DEBUG_TYPE
				490	#define DEBUG_TYPE "bar"
				491	DOUT << "'bar' debug type\n";
				492	#undef DEBUG_TYPE
				493	#define DEBUG_TYPE ""
				494	DOUT << "No debug type (2)\n";
				495	</pre>
				496	</div>
				497
				498	<p>Then you can run your pass like this:</p>
				499
				500	<div class="doc_code">
				501	<pre>
				502	$ opt < a.bc > /dev/null -mypass
				503	<i><no output></i>
				504	$ opt < a.bc > /dev/null -mypass -debug
				505	No debug type
				506	'foo' debug type
				507	'bar' debug type
				508	No debug type (2)
				509	$ opt < a.bc > /dev/null -mypass -debug-only=foo
				510	'foo' debug type
				511	$ opt < a.bc > /dev/null -mypass -debug-only=bar
				512	'bar' debug type
				513	</pre>
				514	</div>
				515
				516	<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
				517	a file, to specify the debug type for the entire module (if you do this before
				518	you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
				519	<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
				520	"bar", because there is no system in place to ensure that names do not
				521	conflict. If two different modules use the same string, they will all be turned
				522	on when the name is specified. This allows, for example, all debug information
				523	for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
				524	even if the source lives in multiple files.</p>
				525
				526	</div>
				527
				528	<!-- ======================================================================= -->
				529	<div class="doc_subsection">
				530	<a name="Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
				531	option</a>
				532	</div>
				533
				534	<div class="doc_text">
				535
				536	<p>The "<tt><a
				537	href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
				538	provides a class named <tt>Statistic</tt> that is used as a unified way to
				539	keep track of what the LLVM compiler is doing and how effective various
				540	optimizations are. It is useful to see what optimizations are contributing to
				541	making a particular program run faster.</p>
				542
				543	<p>Often you may run your pass on some big program, and you're interested to see
				544	how many times it makes a certain transformation. Although you can do this with
				545	hand inspection, or some ad-hoc method, this is a real pain and not very useful
				546	for big programs. Using the <tt>Statistic</tt> class makes it very easy to
				547	keep track of this information, and the calculated information is presented in a
				548	uniform manner with the rest of the passes being executed.</p>
				549
				550	<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
				551	it are as follows:</p>
				552
				553	<ol>
				554	<li><p>Define your statistic like this:</p>
				555
				556	<div class="doc_code">
				557	<pre>
				558	#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
				559	STATISTIC(NumXForms, "The # of times I did stuff");
				560	</pre>
				561	</div>
				562
				563	<p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
				564	specified by the first argument. The pass name is taken from the DEBUG_TYPE
				565	macro, and the description is taken from the second argument. The variable
				566	defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
				567
				568	<li><p>Whenever you make a transformation, bump the counter:</p>
				569
				570	<div class="doc_code">
				571	<pre>
				572	++NumXForms; // <i>I did stuff!</i>
				573	</pre>
				574	</div>
				575
				576	</li>
				577	</ol>
				578
				579	<p>That's all you have to do. To get '<tt>opt</tt>' to print out the
				580	statistics gathered, use the '<tt>-stats</tt>' option:</p>
				581
				582	<div class="doc_code">
				583	<pre>
				584	$ opt -stats -mypassname < program.bc > /dev/null
				585	<i>... statistics output ...</i>
				586	</pre>
				587	</div>
				588
				589	<p> When running <tt>opt</tt> on a C file from the SPEC benchmark
				590	suite, it gives a report that looks like this:</p>
				591
				592	<div class="doc_code">
				593	<pre>
				594	7646 bitcodewriter - Number of normal instructions
				595	725 bitcodewriter - Number of oversized instructions
				596	129996 bitcodewriter - Number of bitcode bytes written
				597	2817 raise - Number of insts DCEd or constprop'd
				598	3213 raise - Number of cast-of-self removed
				599	5046 raise - Number of expression trees converted
				600	75 raise - Number of other getelementptr's formed
				601	138 raise - Number of load/store peepholes
				602	42 deadtypeelim - Number of unused typenames removed from symtab
				603	392 funcresolve - Number of varargs functions resolved
				604	27 globaldce - Number of global variables removed
				605	2 adce - Number of basic blocks removed
				606	134 cee - Number of branches revectored
				607	49 cee - Number of setcc instruction eliminated
				608	532 gcse - Number of loads removed
				609	2919 gcse - Number of instructions removed
				610	86 indvars - Number of canonical indvars added
				611	87 indvars - Number of aux indvars removed
				612	25 instcombine - Number of dead inst eliminate
				613	434 instcombine - Number of insts combined
				614	248 licm - Number of load insts hoisted
				615	1298 licm - Number of insts hoisted to a loop pre-header
				616	3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
				617	75 mem2reg - Number of alloca's promoted
				618	1444 cfgsimplify - Number of blocks simplified
				619	</pre>
				620	</div>
				621
				622	<p>Obviously, with so many optimizations, having a unified framework for this
				623	stuff is very nice. Making your pass fit well into the framework makes it more
				624	maintainable and useful.</p>
				625
				626	</div>
				627
				628	<!-- ======================================================================= -->
				629	<div class="doc_subsection">
				630	<a name="ViewGraph">Viewing graphs while debugging code</a>
				631	</div>
				632
				633	<div class="doc_text">
				634
				635	<p>Several of the important data structures in LLVM are graphs: for example
				636	CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
				637	LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
				638	<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
				639	DAGs</a>. In many cases, while debugging various parts of the compiler, it is
				640	nice to instantly visualize these graphs.</p>
				641
				642	<p>LLVM provides several callbacks that are available in a debug build to do
				643	exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
				644	the current LLVM tool will pop up a window containing the CFG for the function
				645	where each basic block is a node in the graph, and each node contains the
				646	instructions in the block. Similarly, there also exists
				647	<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
				648	<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
				649	and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
				650	you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
				651	up a window. Alternatively, you can sprinkle calls to these functions in your
				652	code in places you want to debug.</p>
				653
				654	<p>Getting this to work requires a small amount of configuration. On Unix
				655	systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
				656	toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
				657	Mac OS/X, download and install the Mac OS/X <a
				658	href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
				659	<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
				660	it) to your path. Once in your system and path are set up, rerun the LLVM
				661	configure script and rebuild LLVM to enable this functionality.</p>
				662
				663	<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
				664	<i>interesting</i> nodes in large complex graphs. From gdb, if you
				665	<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
				666	next <tt>call DAG.viewGraph()</tt> would highlight the node in the
				667	specified color (choices of colors can be found at <a
				668	href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
				669	complex node attributes can be provided with <tt>call
				670	DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
				671	found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
				672	Attributes</a>.) If you want to restart and clear all the current graph
				673	attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
				674
				675	</div>
				676
				677	<!-- *********************************************************************** -->
				678	<div class="doc_section">
				679	<a name="datastructure">Picking the Right Data Structure for a Task</a>
				680	</div>
				681	<!-- *********************************************************************** -->
				682
				683	<div class="doc_text">
				684
				685	<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
				686	and we commonly use STL data structures. This section describes the trade-offs
				687	you should consider when you pick one.</p>
				688
				689	<p>
				690	The first step is a choose your own adventure: do you want a sequential
				691	container, a set-like container, or a map-like container? The most important
				692	thing when choosing a container is the algorithmic properties of how you plan to
				693	access the container. Based on that, you should use:</p>
				694
				695	<ul>
				696	<li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
				697	of an value based on another value. Map-like containers also support
				698	efficient queries for containment (whether a key is in the map). Map-like
				699	containers generally do not support efficient reverse mapping (values to
				700	keys). If you need that, use two maps. Some map-like containers also
				701	support efficient iteration through the keys in sorted order. Map-like
				702	containers are the most expensive sort, only use them if you need one of
				703	these capabilities.</li>
				704
				705	<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
				706	stuff into a container that automatically eliminates duplicates. Some
				707	set-like containers support efficient iteration through the elements in
				708	sorted order. Set-like containers are more expensive than sequential
				709	containers.
				710	</li>
				711
				712	<li>a <a href="#ds_sequential">sequential</a> container provides
				713	the most efficient way to add elements and keeps track of the order they are
				714	added to the collection. They permit duplicates and support efficient
				715	iteration, but do not support efficient look-up based on a key.
				716	</li>
				717
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	718	<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
				719	perform set operations on sets of numeric id's, while automatically
				720	eliminating duplicates. Bit containers require a maximum of 1 bit for each
				721	identifier you want to store.
				722	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	723	</ul>
				724
				725	<p>
				726	Once the proper category of container is determined, you can fine tune the
				727	memory use, constant factors, and cache behaviors of access by intelligently
				728	picking a member of the category. Note that constant factors and cache behavior
				729	can be a big deal. If you have a vector that usually only contains a few
				730	elements (but could contain many), for example, it's much better to use
				731	<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
				732	. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
				733	cost of adding the elements to the container. </p>
				734
				735	</div>
				736
				737	<!-- ======================================================================= -->
				738	<div class="doc_subsection">
				739	<a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
				740	</div>
				741
				742	<div class="doc_text">
				743	There are a variety of sequential containers available for you, based on your
				744	needs. Pick the first in this section that will do what you want.
				745	</div>
				746
				747	<!-- _______________________________________________________________________ -->
				748	<div class="doc_subsubsection">
				749	<a name="dss_fixedarrays">Fixed Size Arrays</a>
				750	</div>
				751
				752	<div class="doc_text">
				753	<p>Fixed size arrays are very simple and very fast. They are good if you know
				754	exactly how many elements you have, or you have a (low) upper bound on how many
				755	you have.</p>
				756	</div>
				757
				758	<!-- _______________________________________________________________________ -->
				759	<div class="doc_subsubsection">
				760	<a name="dss_heaparrays">Heap Allocated Arrays</a>
				761	</div>
				762
				763	<div class="doc_text">
				764	<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
				765	the number of elements is variable, if you know how many elements you will need
				766	before the array is allocated, and if the array is usually large (if not,
				767	consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
				768	allocated array is the cost of the new/delete (aka malloc/free). Also note that
				769	if you are allocating an array of a type with a constructor, the constructor and
				770	destructors will be run for every element in the array (re-sizable vectors only
				771	construct those elements actually used).</p>
				772	</div>
				773
				774	<!-- _______________________________________________________________________ -->
				775	<div class="doc_subsubsection">
				776	<a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
				777	</div>
				778
				779	<div class="doc_text">
				780	<p><tt>SmallVector<Type, N></tt> is a simple class that looks and smells
				781	just like <tt>vector<Type></tt>:
				782	it supports efficient iteration, lays out elements in memory order (so you can
				783	do pointer arithmetic between elements), supports efficient push_back/pop_back
				784	operations, supports efficient random access to its elements, etc.</p>
				785
				786	<p>The advantage of SmallVector is that it allocates space for
				787	some number of elements (N) <b>in the object itself</b>. Because of this, if
				788	the SmallVector is dynamically smaller than N, no malloc is performed. This can
				789	be a big win in cases where the malloc/free call is far more expensive than the
				790	code that fiddles around with the elements.</p>
				791
				792	<p>This is good for vectors that are "usually small" (e.g. the number of
				793	predecessors/successors of a block is usually less than 8). On the other hand,
				794	this makes the size of the SmallVector itself large, so you don't want to
				795	allocate lots of them (doing so will waste a lot of space). As such,
				796	SmallVectors are most useful when on the stack.</p>
				797
				798	<p>SmallVector also provides a nice portable and efficient replacement for
				799	<tt>alloca</tt>.</p>
				800
				801	</div>
				802
				803	<!-- _______________________________________________________________________ -->
				804	<div class="doc_subsubsection">
				805	<a name="dss_vector"><vector></a>
				806	</div>
				807
				808	<div class="doc_text">
				809	<p>
				810	std::vector is well loved and respected. It is useful when SmallVector isn't:
				811	when the size of the vector is often large (thus the small optimization will
				812	rarely be a benefit) or if you will be allocating many instances of the vector
				813	itself (which would waste space for elements that aren't in the container).
				814	vector is also useful when interfacing with code that expects vectors :).
				815	</p>
				816
				817	<p>One worthwhile note about std::vector: avoid code like this:</p>
				818
				819	<div class="doc_code">
				820	<pre>
				821	for ( ... ) {
				822	std::vector<foo> V;
				823	use V;
				824	}
				825	</pre>
				826	</div>
				827
				828	<p>Instead, write this as:</p>
				829
				830	<div class="doc_code">
				831	<pre>
				832	std::vector<foo> V;
				833	for ( ... ) {
				834	use V;
				835	V.clear();
				836	}
				837	</pre>
				838	</div>
				839
				840	<p>Doing so will save (at least) one heap allocation and free per iteration of
				841	the loop.</p>
				842
				843	</div>
				844
				845	<!-- _______________________________________________________________________ -->
				846	<div class="doc_subsubsection">
				847	<a name="dss_deque"><deque></a>
				848	</div>
				849
				850	<div class="doc_text">
				851	<p>std::deque is, in some senses, a generalized version of std::vector. Like
				852	std::vector, it provides constant time random access and other similar
				853	properties, but it also provides efficient access to the front of the list. It
				854	does not guarantee continuity of elements within memory.</p>
				855
				856	<p>In exchange for this extra flexibility, std::deque has significantly higher
				857	constant factor costs than std::vector. If possible, use std::vector or
				858	something cheaper.</p>
				859	</div>
				860
				861	<!-- _______________________________________________________________________ -->
				862	<div class="doc_subsubsection">
				863	<a name="dss_list"><list></a>
				864	</div>
				865
				866	<div class="doc_text">
				867	<p>std::list is an extremely inefficient class that is rarely useful.
				868	It performs a heap allocation for every element inserted into it, thus having an
				869	extremely high constant factor, particularly for small data types. std::list
				870	also only supports bidirectional iteration, not random access iteration.</p>
				871
				872	<p>In exchange for this high cost, std::list supports efficient access to both
				873	ends of the list (like std::deque, but unlike std::vector or SmallVector). In
				874	addition, the iterator invalidation characteristics of std::list are stronger
				875	than that of a vector class: inserting or removing an element into the list does
				876	not invalidate iterator or pointers to other elements in the list.</p>
				877	</div>
				878
				879	<!-- _______________________________________________________________________ -->
				880	<div class="doc_subsubsection">
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame^]	881	<a name="dss_ilist">llvm/ADT/ilist.h</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	882	</div>
				883
				884	<div class="doc_text">
				885	<p><tt>ilist<T></tt> implements an 'intrusive' doubly-linked list. It is
				886	intrusive, because it requires the element to store and provide access to the
				887	prev/next pointers for the list.</p>
				888
				889	<p>ilist has the same drawbacks as std::list, and additionally requires an
				890	ilist_traits implementation for the element type, but it provides some novel
				891	characteristics. In particular, it can efficiently store polymorphic objects,
				892	the traits class is informed when an element is inserted or removed from the
				893	list, and ilists are guaranteed to support a constant-time splice operation.
				894	</p>
				895
				896	<p>These properties are exactly what we want for things like Instructions and
				897	basic blocks, which is why these are implemented with ilists.</p>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame^]	898
				899	Related classes of interest are explained in the following subsections:
				900	<ul>
				901	<li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
				902	</ul>
				903	</div>
				904
				905	<!-- _______________________________________________________________________ -->
				906	<div class="doc_subsubsection">
				907	<a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
				908	</div>
				909
				910	<div class="doc_text">
				911	<p><tt>ilist_node<T></tt> implements a the forward and backward links
				912	that are expected by the <tt>ilist<T></tt> (and analogous containers)
				913	in the default manner.</p>
				914
				915	<p><tt>ilist_node<T></tt>s are meant to be embedded in the node type
				916	<tt>T</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	917	</div>
				918
				919	<!-- _______________________________________________________________________ -->
				920	<div class="doc_subsubsection">
				921	<a name="dss_other">Other Sequential Container options</a>
				922	</div>
				923
				924	<div class="doc_text">
				925	<p>Other STL containers are available, such as std::string.</p>
				926
				927	<p>There are also various STL adapter classes such as std::queue,
				928	std::priority_queue, std::stack, etc. These provide simplified access to an
				929	underlying container but don't affect the cost of the container itself.</p>
				930
				931	</div>
				932
				933
				934	<!-- ======================================================================= -->
				935	<div class="doc_subsection">
				936	<a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
				937	</div>
				938
				939	<div class="doc_text">
				940
				941	<p>Set-like containers are useful when you need to canonicalize multiple values
				942	into a single representation. There are several different choices for how to do
				943	this, providing various trade-offs.</p>
				944
				945	</div>
				946
				947
				948	<!-- _______________________________________________________________________ -->
				949	<div class="doc_subsubsection">
				950	<a name="dss_sortedvectorset">A sorted 'vector'</a>
				951	</div>
				952
				953	<div class="doc_text">
				954
				955	<p>If you intend to insert a lot of elements, then do a lot of queries, a
				956	great approach is to use a vector (or other sequential container) with
				957	std::sort+std::unique to remove duplicates. This approach works really well if
				958	your usage pattern has these two distinct phases (insert then query), and can be
				959	coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
				960	</p>
				961
				962	<p>
				963	This combination provides the several nice properties: the result data is
				964	contiguous in memory (good for cache locality), has few allocations, is easy to
				965	address (iterators in the final vector are just indices or pointers), and can be
				966	efficiently queried with a standard binary or radix search.</p>
				967
				968	</div>
				969
				970	<!-- _______________________________________________________________________ -->
				971	<div class="doc_subsubsection">
				972	<a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
				973	</div>
				974
				975	<div class="doc_text">
				976
				977	<p>If you have a set-like data structure that is usually small and whose elements
				978	are reasonably small, a <tt>SmallSet<Type, N></tt> is a good choice. This set
				979	has space for N elements in place (thus, if the set is dynamically smaller than
				980	N, no malloc traffic is required) and accesses them with a simple linear search.
				981	When the set grows beyond 'N' elements, it allocates a more expensive representation that
				982	guarantees efficient access (for most types, it falls back to std::set, but for
				983	pointers it uses something far better, <a
				984	href="#dss_smallptrset">SmallPtrSet</a>).</p>
				985
				986	<p>The magic of this class is that it handles small sets extremely efficiently,
				987	but gracefully handles extremely large sets without loss of efficiency. The
				988	drawback is that the interface is quite small: it supports insertion, queries
				989	and erasing, but does not support iteration.</p>
				990
				991	</div>
				992
				993	<!-- _______________________________________________________________________ -->
				994	<div class="doc_subsubsection">
				995	<a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
				996	</div>
				997
				998	<div class="doc_text">
				999
				1000	<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
				1001	transparently implemented with a SmallPtrSet), but also supports iterators. If
				1002	more than 'N' insertions are performed, a single quadratically
				1003	probed hash table is allocated and grows as needed, providing extremely
				1004	efficient access (constant time insertion/deleting/queries with low constant
				1005	factors) and is very stingy with malloc traffic.</p>
				1006
				1007	<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
				1008	whenever an insertion occurs. Also, the values visited by the iterators are not
				1009	visited in sorted order.</p>
				1010
				1011	</div>
				1012
				1013	<!-- _______________________________________________________________________ -->
				1014	<div class="doc_subsubsection">
Chris Lattner	77ab46d	2007-09-30 00:58:59 +0000	[diff] [blame]	1015	<a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
				1016	</div>
				1017
				1018	<div class="doc_text">
				1019
				1020	<p>
				1021	DenseSet is a simple quadratically probed hash table. It excels at supporting
				1022	small values: it uses a single allocation to hold all of the pairs that
				1023	are currently inserted in the set. DenseSet is a great way to unique small
				1024	values that are not simple pointers (use <a
				1025	href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
				1026	the same requirements for the value type that <a
				1027	href="#dss_densemap">DenseMap</a> has.
				1028	</p>
				1029
				1030	</div>
				1031
				1032	<!-- _______________________________________________________________________ -->
				1033	<div class="doc_subsubsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1034	<a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
				1035	</div>
				1036
				1037	<div class="doc_text">
				1038
				1039	<p>
				1040	FoldingSet is an aggregate class that is really good at uniquing
				1041	expensive-to-create or polymorphic objects. It is a combination of a chained
				1042	hash table with intrusive links (uniqued objects are required to inherit from
				1043	FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
				1044	its ID process.</p>
				1045
				1046	<p>Consider a case where you want to implement a "getOrCreateFoo" method for
				1047	a complex object (for example, a node in the code generator). The client has a
				1048	description of what it wants to generate (it knows the opcode and all the
				1049	operands), but we don't want to 'new' a node, then try inserting it into a set
				1050	only to find out it already exists, at which point we would have to delete it
				1051	and return the node that already exists.
				1052	</p>
				1053
				1054	<p>To support this style of client, FoldingSet perform a query with a
				1055	FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
				1056	element that we want to query for. The query either returns the element
				1057	matching the ID or it returns an opaque ID that indicates where insertion should
				1058	take place. Construction of the ID usually does not require heap traffic.</p>
				1059
				1060	<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
				1061	in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
				1062	Because the elements are individually allocated, pointers to the elements are
				1063	stable: inserting or removing elements does not invalidate any pointers to other
				1064	elements.
				1065	</p>
				1066
				1067	</div>
				1068
				1069	<!-- _______________________________________________________________________ -->
				1070	<div class="doc_subsubsection">
				1071	<a name="dss_set"><set></a>
				1072	</div>
				1073
				1074	<div class="doc_text">
				1075
				1076	<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
				1077	many things but great at nothing. std::set allocates memory for each element
				1078	inserted (thus it is very malloc intensive) and typically stores three pointers
				1079	per element in the set (thus adding a large amount of per-element space
				1080	overhead). It offers guaranteed log(n) performance, which is not particularly
				1081	fast from a complexity standpoint (particularly if the elements of the set are
				1082	expensive to compare, like strings), and has extremely high constant factors for
				1083	lookup, insertion and removal.</p>
				1084
				1085	<p>The advantages of std::set are that its iterators are stable (deleting or
				1086	inserting an element from the set does not affect iterators or pointers to other
				1087	elements) and that iteration over the set is guaranteed to be in sorted order.
				1088	If the elements in the set are large, then the relative overhead of the pointers
				1089	and malloc traffic is not a big deal, but if the elements of the set are small,
				1090	std::set is almost never a good choice.</p>
				1091
				1092	</div>
				1093
				1094	<!-- _______________________________________________________________________ -->
				1095	<div class="doc_subsubsection">
				1096	<a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
				1097	</div>
				1098
				1099	<div class="doc_text">
				1100	<p>LLVM's SetVector<Type> is an adapter class that combines your choice of
				1101	a set-like container along with a <a href="#ds_sequential">Sequential
				1102	Container</a>. The important property
				1103	that this provides is efficient insertion with uniquing (duplicate elements are
				1104	ignored) with iteration support. It implements this by inserting elements into
				1105	both a set-like container and the sequential container, using the set-like
				1106	container for uniquing and the sequential container for iteration.
				1107	</p>
				1108
				1109	<p>The difference between SetVector and other sets is that the order of
				1110	iteration is guaranteed to match the order of insertion into the SetVector.
				1111	This property is really important for things like sets of pointers. Because
				1112	pointer values are non-deterministic (e.g. vary across runs of the program on
				1113	different machines), iterating over the pointers in the set will
				1114	not be in a well-defined order.</p>
				1115
				1116	<p>
				1117	The drawback of SetVector is that it requires twice as much space as a normal
				1118	set and has the sum of constant factors from the set-like container and the
				1119	sequential container that it uses. Use it only if you need to iterate over
				1120	the elements in a deterministic order. SetVector is also expensive to delete
				1121	elements out of (linear time), unless you use it's "pop_back" method, which is
				1122	faster.
				1123	</p>
				1124
				1125	<p>SetVector is an adapter class that defaults to using std::vector and std::set
				1126	for the underlying containers, so it is quite expensive. However,
				1127	<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
				1128	defaults to using a SmallVector and SmallSet of a specified size. If you use
				1129	this, and if your sets are dynamically smaller than N, you will save a lot of
				1130	heap traffic.</p>
				1131
				1132	</div>
				1133
				1134	<!-- _______________________________________________________________________ -->
				1135	<div class="doc_subsubsection">
				1136	<a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
				1137	</div>
				1138
				1139	<div class="doc_text">
				1140
				1141	<p>
				1142	UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
				1143	retains a unique ID for each element inserted into the set. It internally
				1144	contains a map and a vector, and it assigns a unique ID for each value inserted
				1145	into the set.</p>
				1146
				1147	<p>UniqueVector is very expensive: its cost is the sum of the cost of
				1148	maintaining both the map and vector, it has high complexity, high constant
				1149	factors, and produces a lot of malloc traffic. It should be avoided.</p>
				1150
				1151	</div>
				1152
				1153
				1154	<!-- _______________________________________________________________________ -->
				1155	<div class="doc_subsubsection">
				1156	<a name="dss_otherset">Other Set-Like Container Options</a>
				1157	</div>
				1158
				1159	<div class="doc_text">
				1160
				1161	<p>
				1162	The STL provides several other options, such as std::multiset and the various
				1163	"hash_set" like containers (whether from C++ TR1 or from the SGI library).</p>
				1164
				1165	<p>std::multiset is useful if you're not interested in elimination of
				1166	duplicates, but has all the drawbacks of std::set. A sorted vector (where you
				1167	don't delete duplicate entries) or some other approach is almost always
				1168	better.</p>
				1169
				1170	<p>The various hash_set implementations (exposed portably by
				1171	"llvm/ADT/hash_set") is a simple chained hashtable. This algorithm is as malloc
				1172	intensive as std::set (performing an allocation for each element inserted,
				1173	thus having really high constant factors) but (usually) provides O(1)
				1174	insertion/deletion of elements. This can be useful if your elements are large
				1175	(thus making the constant-factor cost relatively low) or if comparisons are
				1176	expensive. Element iteration does not visit elements in a useful order.</p>
				1177
				1178	</div>
				1179
				1180	<!-- ======================================================================= -->
				1181	<div class="doc_subsection">
				1182	<a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
				1183	</div>
				1184
				1185	<div class="doc_text">
				1186	Map-like containers are useful when you want to associate data to a key. As
				1187	usual, there are a lot of different ways to do this. :)
				1188	</div>
				1189
				1190	<!-- _______________________________________________________________________ -->
				1191	<div class="doc_subsubsection">
				1192	<a name="dss_sortedvectormap">A sorted 'vector'</a>
				1193	</div>
				1194
				1195	<div class="doc_text">
				1196
				1197	<p>
				1198	If your usage pattern follows a strict insert-then-query approach, you can
				1199	trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
				1200	for set-like containers</a>. The only difference is that your query function
				1201	(which uses std::lower_bound to get efficient log(n) lookup) should only compare
				1202	the key, not both the key and value. This yields the same advantages as sorted
				1203	vectors for sets.
				1204	</p>
				1205	</div>
				1206
				1207	<!-- _______________________________________________________________________ -->
				1208	<div class="doc_subsubsection">
				1209	<a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
				1210	</div>
				1211
				1212	<div class="doc_text">
				1213
				1214	<p>
				1215	Strings are commonly used as keys in maps, and they are difficult to support
				1216	efficiently: they are variable length, inefficient to hash and compare when
				1217	long, expensive to copy, etc. StringMap is a specialized container designed to
				1218	cope with these issues. It supports mapping an arbitrary range of bytes to an
				1219	arbitrary other object.</p>
				1220
				1221	<p>The StringMap implementation uses a quadratically-probed hash table, where
				1222	the buckets store a pointer to the heap allocated entries (and some other
				1223	stuff). The entries in the map must be heap allocated because the strings are
				1224	variable length. The string data (key) and the element object (value) are
				1225	stored in the same allocation with the string data immediately after the element
				1226	object. This container guarantees the "<tt>(char*)(&Value+1)</tt>" points
				1227	to the key string for a value.</p>
				1228
				1229	<p>The StringMap is very fast for several reasons: quadratic probing is very
				1230	cache efficient for lookups, the hash value of strings in buckets is not
				1231	recomputed when lookup up an element, StringMap rarely has to touch the
				1232	memory for unrelated objects when looking up a value (even when hash collisions
				1233	happen), hash table growth does not recompute the hash values for strings
				1234	already in the table, and each pair in the map is store in a single allocation
				1235	(the string data is stored in the same allocation as the Value of a pair).</p>
				1236
				1237	<p>StringMap also provides query methods that take byte ranges, so it only ever
				1238	copies a string if a value is inserted into the table.</p>
				1239	</div>
				1240
				1241	<!-- _______________________________________________________________________ -->
				1242	<div class="doc_subsubsection">
				1243	<a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
				1244	</div>
				1245
				1246	<div class="doc_text">
				1247	<p>
				1248	IndexedMap is a specialized container for mapping small dense integers (or
				1249	values that can be mapped to small dense integers) to some other type. It is
				1250	internally implemented as a vector with a mapping function that maps the keys to
				1251	the dense integer range.
				1252	</p>
				1253
				1254	<p>
				1255	This is useful for cases like virtual registers in the LLVM code generator: they
				1256	have a dense mapping that is offset by a compile-time constant (the first
				1257	virtual register ID).</p>
				1258
				1259	</div>
				1260
				1261	<!-- _______________________________________________________________________ -->
				1262	<div class="doc_subsubsection">
				1263	<a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
				1264	</div>
				1265
				1266	<div class="doc_text">
				1267
				1268	<p>
				1269	DenseMap is a simple quadratically probed hash table. It excels at supporting
				1270	small keys and values: it uses a single allocation to hold all of the pairs that
				1271	are currently inserted in the map. DenseMap is a great way to map pointers to
				1272	pointers, or map other small types to each other.
				1273	</p>
				1274
				1275	<p>
				1276	There are several aspects of DenseMap that you should be aware of, however. The
				1277	iterators in a densemap are invalidated whenever an insertion occurs, unlike
				1278	map. Also, because DenseMap allocates space for a large number of key/value
				1279	pairs (it starts with 64 by default), it will waste a lot of space if your keys
				1280	or values are large. Finally, you must implement a partial specialization of
Chris Lattner	92eea07	2007-09-17 18:34:04 +0000	[diff] [blame]	1281	DenseMapInfo for the key that you want, if it isn't already supported. This
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1282	is required to tell DenseMap about two special marker values (which can never be
				1283	inserted into the map) that it needs internally.</p>
				1284
				1285	</div>
				1286
				1287	<!-- _______________________________________________________________________ -->
				1288	<div class="doc_subsubsection">
				1289	<a name="dss_map"><map></a>
				1290	</div>
				1291
				1292	<div class="doc_text">
				1293
				1294	<p>
				1295	std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
				1296	a single allocation per pair inserted into the map, it offers log(n) lookup with
				1297	an extremely large constant factor, imposes a space penalty of 3 pointers per
				1298	pair in the map, etc.</p>
				1299
				1300	<p>std::map is most useful when your keys or values are very large, if you need
				1301	to iterate over the collection in sorted order, or if you need stable iterators
				1302	into the map (i.e. they don't get invalidated if an insertion or deletion of
				1303	another element takes place).</p>
				1304
				1305	</div>
				1306
				1307	<!-- _______________________________________________________________________ -->
				1308	<div class="doc_subsubsection">
				1309	<a name="dss_othermap">Other Map-Like Container Options</a>
				1310	</div>
				1311
				1312	<div class="doc_text">
				1313
				1314	<p>
				1315	The STL provides several other options, such as std::multimap and the various
				1316	"hash_map" like containers (whether from C++ TR1 or from the SGI library).</p>
				1317
				1318	<p>std::multimap is useful if you want to map a key to multiple values, but has
				1319	all the drawbacks of std::map. A sorted vector or some other approach is almost
				1320	always better.</p>
				1321
				1322	<p>The various hash_map implementations (exposed portably by
				1323	"llvm/ADT/hash_map") are simple chained hash tables. This algorithm is as
				1324	malloc intensive as std::map (performing an allocation for each element
				1325	inserted, thus having really high constant factors) but (usually) provides O(1)
				1326	insertion/deletion of elements. This can be useful if your elements are large
				1327	(thus making the constant-factor cost relatively low) or if comparisons are
				1328	expensive. Element iteration does not visit elements in a useful order.</p>
				1329
				1330	</div>
				1331
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1332	<!-- ======================================================================= -->
				1333	<div class="doc_subsection">
				1334	<a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
				1335	</div>
				1336
				1337	<div class="doc_text">
Chris Lattner	62a4eae	2007-09-25 22:37:50 +0000	[diff] [blame]	1338	<p>Unlike the other containers, there are only two bit storage containers, and
				1339	choosing when to use each is relatively straightforward.</p>
				1340
				1341	<p>One additional option is
				1342	<tt>std::vector<bool></tt>: we discourage its use for two reasons 1) the
				1343	implementation in many common compilers (e.g. commonly available versions of
				1344	GCC) is extremely inefficient and 2) the C++ standards committee is likely to
				1345	deprecate this container and/or change it significantly somehow. In any case,
				1346	please don't use it.</p>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1347	</div>
				1348
				1349	<!-- _______________________________________________________________________ -->
				1350	<div class="doc_subsubsection">
				1351	<a name="dss_bitvector">BitVector</a>
				1352	</div>
				1353
				1354	<div class="doc_text">
				1355	<p> The BitVector container provides a fixed size set of bits for manipulation.
				1356	It supports individual bit setting/testing, as well as set operations. The set
				1357	operations take time O(size of bitvector), but operations are performed one word
				1358	at a time, instead of one bit at a time. This makes the BitVector very fast for
				1359	set operations compared to other containers. Use the BitVector when you expect
				1360	the number of set bits to be high (IE a dense set).
				1361	</p>
				1362	</div>
				1363
				1364	<!-- _______________________________________________________________________ -->
				1365	<div class="doc_subsubsection">
				1366	<a name="dss_sparsebitvector">SparseBitVector</a>
				1367	</div>
				1368
				1369	<div class="doc_text">
				1370	<p> The SparseBitVector container is much like BitVector, with one major
				1371	difference: Only the bits that are set, are stored. This makes the
				1372	SparseBitVector much more space efficient than BitVector when the set is sparse,
				1373	as well as making set operations O(number of set bits) instead of O(size of
				1374	universe). The downside to the SparseBitVector is that setting and testing of random bits is O(N), and on large SparseBitVectors, this can be slower than BitVector. In our implementation, setting or testing bits in sorted order
				1375	(either forwards or reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends on size) of the current bit is also O(1). As a general statement, testing/setting bits in a SparseBitVector is O(distance away from last set bit).
				1376	</p>
				1377	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1378
				1379	<!-- *********************************************************************** -->
				1380	<div class="doc_section">
				1381	<a name="common">Helpful Hints for Common Operations</a>
				1382	</div>
				1383	<!-- *********************************************************************** -->
				1384
				1385	<div class="doc_text">
				1386
				1387	<p>This section describes how to perform some very simple transformations of
				1388	LLVM code. This is meant to give examples of common idioms used, showing the
				1389	practical side of LLVM transformations. <p> Because this is a "how-to" section,
				1390	you should also read about the main classes that you will be working with. The
				1391	<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
				1392	and descriptions of the main classes that you should know about.</p>
				1393
				1394	</div>
				1395
				1396	<!-- NOTE: this section should be heavy on example code -->
				1397	<!-- ======================================================================= -->
				1398	<div class="doc_subsection">
				1399	<a name="inspection">Basic Inspection and Traversal Routines</a>
				1400	</div>
				1401
				1402	<div class="doc_text">
				1403
				1404	<p>The LLVM compiler infrastructure have many different data structures that may
				1405	be traversed. Following the example of the C++ standard template library, the
				1406	techniques used to traverse these various data structures are all basically the
				1407	same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
				1408	method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
				1409	function returns an iterator pointing to one past the last valid element of the
				1410	sequence, and there is some <tt>XXXiterator</tt> data type that is common
				1411	between the two operations.</p>
				1412
				1413	<p>Because the pattern for iteration is common across many different aspects of
				1414	the program representation, the standard template library algorithms may be used
				1415	on them, and it is easier to remember how to iterate. First we show a few common
				1416	examples of the data structures that need to be traversed. Other data
				1417	structures are traversed in very similar ways.</p>
				1418
				1419	</div>
				1420
				1421	<!-- _______________________________________________________________________ -->
				1422	<div class="doc_subsubsection">
				1423	<a name="iterate_function">Iterating over the </a><a
				1424	href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
				1425	href="#Function"><tt>Function</tt></a>
				1426	</div>
				1427
				1428	<div class="doc_text">
				1429
				1430	<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
				1431	transform in some way; in particular, you'd like to manipulate its
				1432	<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
				1433	the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
				1434	an example that prints the name of a <tt>BasicBlock</tt> and the number of
				1435	<tt>Instruction</tt>s it contains:</p>
				1436
				1437	<div class="doc_code">
				1438	<pre>
				1439	// <i>func is a pointer to a Function instance</i>
				1440	for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i)
				1441	// <i>Print out the name of the basic block if it has one, and then the</i>
				1442	// <i>number of instructions that it contains</i>
				1443	llvm::cerr << "Basic block (name=" << i->getName() << ") has "
				1444	<< i->size() << " instructions.\n";
				1445	</pre>
				1446	</div>
				1447
				1448	<p>Note that i can be used as if it were a pointer for the purposes of
				1449	invoking member functions of the <tt>Instruction</tt> class. This is
				1450	because the indirection operator is overloaded for the iterator
				1451	classes. In the above code, the expression <tt>i->size()</tt> is
				1452	exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
				1453
				1454	</div>
				1455
				1456	<!-- _______________________________________________________________________ -->
				1457	<div class="doc_subsubsection">
				1458	<a name="iterate_basicblock">Iterating over the </a><a
				1459	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1460	href="#BasicBlock"><tt>BasicBlock</tt></a>
				1461	</div>
				1462
				1463	<div class="doc_text">
				1464
				1465	<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
				1466	easy to iterate over the individual instructions that make up
				1467	<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
				1468	a <tt>BasicBlock</tt>:</p>
				1469
				1470	<div class="doc_code">
				1471	<pre>
				1472	// <i>blk is a pointer to a BasicBlock instance</i>
				1473	for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
				1474	// <i>The next statement works since operator<<(ostream&,...)</i>
				1475	// <i>is overloaded for Instruction&</i>
				1476	llvm::cerr << *i << "\n";
				1477	</pre>
				1478	</div>
				1479
				1480	<p>However, this isn't really the best way to print out the contents of a
				1481	<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
				1482	anything you'll care about, you could have just invoked the print routine on the
				1483	basic block itself: <tt>llvm::cerr << *blk << "\n";</tt>.</p>
				1484
				1485	</div>
				1486
				1487	<!-- _______________________________________________________________________ -->
				1488	<div class="doc_subsubsection">
				1489	<a name="iterate_institer">Iterating over the </a><a
				1490	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1491	href="#Function"><tt>Function</tt></a>
				1492	</div>
				1493
				1494	<div class="doc_text">
				1495
				1496	<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
				1497	<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
				1498	<tt>InstIterator</tt> should be used instead. You'll need to include <a
				1499	href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
				1500	and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
				1501	small example that shows how to dump all instructions in a function to the standard error stream:<p>
				1502
				1503	<div class="doc_code">
				1504	<pre>
				1505	#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
				1506
				1507	// <i>F is a pointer to a Function instance</i>
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1508	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
				1509	llvm::cerr << *I << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1510	</pre>
				1511	</div>
				1512
				1513	<p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
				1514	work list with its initial contents. For example, if you wanted to
				1515	initialize a work list to contain all instructions in a <tt>Function</tt>
				1516	F, all you would need to do is something like:</p>
				1517
				1518	<div class="doc_code">
				1519	<pre>
				1520	std::set<Instruction*> worklist;
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1521	// or better yet, SmallPtrSet<Instruction*, 64> worklist;
				1522
				1523	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
				1524	worklist.insert(&*I);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1525	</pre>
				1526	</div>
				1527
				1528	<p>The STL set <tt>worklist</tt> would now contain all instructions in the
				1529	<tt>Function</tt> pointed to by F.</p>
				1530
				1531	</div>
				1532
				1533	<!-- _______________________________________________________________________ -->
				1534	<div class="doc_subsubsection">
				1535	<a name="iterate_convert">Turning an iterator into a class pointer (and
				1536	vice-versa)</a>
				1537	</div>
				1538
				1539	<div class="doc_text">
				1540
				1541	<p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
				1542	instance when all you've got at hand is an iterator. Well, extracting
				1543	a reference or a pointer from an iterator is very straight-forward.
				1544	Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
				1545	is a <tt>BasicBlock::const_iterator</tt>:</p>
				1546
				1547	<div class="doc_code">
				1548	<pre>
				1549	Instruction& inst = *i; // <i>Grab reference to instruction reference</i>
				1550	Instruction* pinst = &*i; // <i>Grab pointer to instruction reference</i>
				1551	const Instruction& inst = *j;
				1552	</pre>
				1553	</div>
				1554
				1555	<p>However, the iterators you'll be working with in the LLVM framework are
				1556	special: they will automatically convert to a ptr-to-instance type whenever they
				1557	need to. Instead of dereferencing the iterator and then taking the address of
				1558	the result, you can simply assign the iterator to the proper pointer type and
				1559	you get the dereference and address-of operation as a result of the assignment
				1560	(behind the scenes, this is a result of overloading casting mechanisms). Thus
				1561	the last line of the last example,</p>
				1562
				1563	<div class="doc_code">
				1564	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1565	Instruction pinst = &i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1566	</pre>
				1567	</div>
				1568
				1569	<p>is semantically equivalent to</p>
				1570
				1571	<div class="doc_code">
				1572	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1573	Instruction *pinst = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1574	</pre>
				1575	</div>
				1576
				1577	<p>It's also possible to turn a class pointer into the corresponding iterator,
				1578	and this is a constant time operation (very efficient). The following code
				1579	snippet illustrates use of the conversion constructors provided by LLVM
				1580	iterators. By using these, you can explicitly grab the iterator of something
				1581	without actually obtaining it via iteration over some structure:</p>
				1582
				1583	<div class="doc_code">
				1584	<pre>
				1585	void printNextInstruction(Instruction* inst) {
				1586	BasicBlock::iterator it(inst);
				1587	++it; // <i>After this line, it refers to the instruction after *inst</i>
				1588	if (it != inst->getParent()->end()) llvm::cerr << *it << "\n";
				1589	}
				1590	</pre>
				1591	</div>
				1592
				1593	</div>
				1594
				1595	<!--_______________________________________________________________________-->
				1596	<div class="doc_subsubsection">
				1597	<a name="iterate_complex">Finding call sites: a slightly more complex
				1598	example</a>
				1599	</div>
				1600
				1601	<div class="doc_text">
				1602
				1603	<p>Say that you're writing a FunctionPass and would like to count all the
				1604	locations in the entire module (that is, across every <tt>Function</tt>) where a
				1605	certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
				1606	learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
				1607	much more straight-forward manner, but this example will allow us to explore how
				1608	you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
				1609	is what we want to do:</p>
				1610
				1611	<div class="doc_code">
				1612	<pre>
				1613	initialize callCounter to zero
				1614	for each Function f in the Module
				1615	for each BasicBlock b in f
				1616	for each Instruction i in b
				1617	if (i is a CallInst and calls the given function)
				1618	increment callCounter
				1619	</pre>
				1620	</div>
				1621
				1622	<p>And the actual code is (remember, because we're writing a
				1623	<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
				1624	override the <tt>runOnFunction</tt> method):</p>
				1625
				1626	<div class="doc_code">
				1627	<pre>
				1628	Function* targetFunc = ...;
				1629
				1630	class OurFunctionPass : public FunctionPass {
				1631	public:
				1632	OurFunctionPass(): callCounter(0) { }
				1633
				1634	virtual runOnFunction(Function& F) {
				1635	for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
Eric Christopher	5fbecf7	2008-11-08 08:20:49 +0000	[diff] [blame]	1636	for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) {
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1637	if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
				1638	href="#CallInst">CallInst</a>>(&*i)) {
				1639	// <i>We know we've encountered a call instruction, so we</i>
				1640	// <i>need to determine if it's a call to the</i>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1641	// <i>function pointed to by m_func or not.</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1642	if (callInst->getCalledFunction() == targetFunc)
				1643	++callCounter;
				1644	}
				1645	}
				1646	}
				1647	}
				1648
				1649	private:
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1650	unsigned callCounter;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1651	};
				1652	</pre>
				1653	</div>
				1654
				1655	</div>
				1656
				1657	<!--_______________________________________________________________________-->
				1658	<div class="doc_subsubsection">
				1659	<a name="calls_and_invokes">Treating calls and invokes the same way</a>
				1660	</div>
				1661
				1662	<div class="doc_text">
				1663
				1664	<p>You may have noticed that the previous example was a bit oversimplified in
				1665	that it did not deal with call sites generated by 'invoke' instructions. In
				1666	this, and in other situations, you may find that you want to treat
				1667	<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
				1668	most-specific common base class is <tt>Instruction</tt>, which includes lots of
				1669	less closely-related things. For these cases, LLVM provides a handy wrapper
				1670	class called <a
				1671	href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
				1672	It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
				1673	methods that provide functionality common to <tt>CallInst</tt>s and
				1674	<tt>InvokeInst</tt>s.</p>
				1675
				1676	<p>This class has "value semantics": it should be passed by value, not by
				1677	reference and it should not be dynamically allocated or deallocated using
				1678	<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
				1679	assignable and constructable, with costs equivalents to that of a bare pointer.
				1680	If you look at its definition, it has only a single pointer member.</p>
				1681
				1682	</div>
				1683
				1684	<!--_______________________________________________________________________-->
				1685	<div class="doc_subsubsection">
				1686	<a name="iterate_chains">Iterating over def-use & use-def chains</a>
				1687	</div>
				1688
				1689	<div class="doc_text">
				1690
				1691	<p>Frequently, we might have an instance of the <a
				1692	href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
				1693	determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
				1694	<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
				1695	For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
				1696	particular function <tt>foo</tt>. Finding all of the instructions that
				1697	<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
				1698	of <tt>F</tt>:</p>
				1699
				1700	<div class="doc_code">
				1701	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1702	Function *F = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1703
				1704	for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i)
				1705	if (Instruction Inst = dyn_cast<Instruction>(i)) {
				1706	llvm::cerr << "F is used in instruction:\n";
				1707	llvm::cerr << *Inst << "\n";
				1708	}
				1709	</pre>
				1710	</div>
				1711
				1712	<p>Alternately, it's common to have an instance of the <a
				1713	href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
				1714	<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
				1715	<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
				1716	<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
				1717	all of the values that a particular instruction uses (that is, the operands of
				1718	the particular <tt>Instruction</tt>):</p>
				1719
				1720	<div class="doc_code">
				1721	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1722	Instruction *pi = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1723
				1724	for (User::op_iterator i = pi->op_begin(), e = pi->op_end(); i != e; ++i) {
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1725	Value v = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1726	// <i>...</i>
				1727	}
				1728	</pre>
				1729	</div>
				1730
				1731	<!--
				1732	def-use chains ("finding all users of"): Value::use_begin/use_end
				1733	use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
				1734	-->
				1735
				1736	</div>
				1737
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1738	<!--_______________________________________________________________________-->
				1739	<div class="doc_subsubsection">
				1740	<a name="iterate_preds">Iterating over predecessors &
				1741	successors of blocks</a>
				1742	</div>
				1743
				1744	<div class="doc_text">
				1745
				1746	<p>Iterating over the predecessors and successors of a block is quite easy
				1747	with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
				1748	this to iterate over all predecessors of BB:</p>
				1749
				1750	<div class="doc_code">
				1751	<pre>
				1752	#include "llvm/Support/CFG.h"
				1753	BasicBlock *BB = ...;
				1754
				1755	for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
				1756	BasicBlock Pred = PI;
				1757	// <i>...</i>
				1758	}
				1759	</pre>
				1760	</div>
				1761
				1762	<p>Similarly, to iterate over successors use
				1763	succ_iterator/succ_begin/succ_end.</p>
				1764
				1765	</div>
				1766
				1767
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1768	<!-- ======================================================================= -->
				1769	<div class="doc_subsection">
				1770	<a name="simplechanges">Making simple changes</a>
				1771	</div>
				1772
				1773	<div class="doc_text">
				1774
				1775	<p>There are some primitive transformation operations present in the LLVM
				1776	infrastructure that are worth knowing about. When performing
				1777	transformations, it's fairly common to manipulate the contents of basic
				1778	blocks. This section describes some of the common methods for doing so
				1779	and gives example code.</p>
				1780
				1781	</div>
				1782
				1783	<!--_______________________________________________________________________-->
				1784	<div class="doc_subsubsection">
				1785	<a name="schanges_creating">Creating and inserting new
				1786	<tt>Instruction</tt>s</a>
				1787	</div>
				1788
				1789	<div class="doc_text">
				1790
				1791	<p><i>Instantiating Instructions</i></p>
				1792
				1793	<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
				1794	constructor for the kind of instruction to instantiate and provide the necessary
				1795	parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
				1796	(const-ptr-to) <tt>Type</tt>. Thus:</p>
				1797
				1798	<div class="doc_code">
				1799	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	1800	AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1801	</pre>
				1802	</div>
				1803
				1804	<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
				1805	one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
				1806	subclass is likely to have varying default parameters which change the semantics
				1807	of the instruction, so refer to the <a
				1808	href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
				1809	Instruction</a> that you're interested in instantiating.</p>
				1810
				1811	<p><i>Naming values</i></p>
				1812
				1813	<p>It is very useful to name the values of instructions when you're able to, as
				1814	this facilitates the debugging of your transformations. If you end up looking
				1815	at generated LLVM machine code, you definitely want to have logical names
				1816	associated with the results of instructions! By supplying a value for the
				1817	<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
				1818	associate a logical name with the result of the instruction's execution at
				1819	run time. For example, say that I'm writing a transformation that dynamically
				1820	allocates space for an integer on the stack, and that integer is going to be
				1821	used as some kind of index by some other code. To accomplish this, I place an
				1822	<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
				1823	<tt>Function</tt>, and I'm intending to use it within the same
				1824	<tt>Function</tt>. I might do:</p>
				1825
				1826	<div class="doc_code">
				1827	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	1828	AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1829	</pre>
				1830	</div>
				1831
				1832	<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
				1833	execution value, which is a pointer to an integer on the run time stack.</p>
				1834
				1835	<p><i>Inserting instructions</i></p>
				1836
				1837	<p>There are essentially two ways to insert an <tt>Instruction</tt>
				1838	into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
				1839
				1840	<ul>
				1841	<li>Insertion into an explicit instruction list
				1842
				1843	<p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
				1844	<tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
				1845	before <tt>*pi</tt>, we do the following: </p>
				1846
				1847	<div class="doc_code">
				1848	<pre>
				1849	BasicBlock *pb = ...;
				1850	Instruction *pi = ...;
				1851	Instruction *newInst = new Instruction(...);
				1852
				1853	pb->getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
				1854	</pre>
				1855	</div>
				1856
				1857	<p>Appending to the end of a <tt>BasicBlock</tt> is so common that
				1858	the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
				1859	classes provide constructors which take a pointer to a
				1860	<tt>BasicBlock</tt> to be appended to. For example code that
				1861	looked like: </p>
				1862
				1863	<div class="doc_code">
				1864	<pre>
				1865	BasicBlock *pb = ...;
				1866	Instruction *newInst = new Instruction(...);
				1867
				1868	pb->getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
				1869	</pre>
				1870	</div>
				1871
				1872	<p>becomes: </p>
				1873
				1874	<div class="doc_code">
				1875	<pre>
				1876	BasicBlock *pb = ...;
				1877	Instruction *newInst = new Instruction(..., pb);
				1878	</pre>
				1879	</div>
				1880
				1881	<p>which is much cleaner, especially if you are creating
				1882	long instruction streams.</p></li>
				1883
				1884	<li>Insertion into an implicit instruction list
				1885
				1886	<p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
				1887	are implicitly associated with an existing instruction list: the instruction
				1888	list of the enclosing basic block. Thus, we could have accomplished the same
				1889	thing as the above code without being given a <tt>BasicBlock</tt> by doing:
				1890	</p>
				1891
				1892	<div class="doc_code">
				1893	<pre>
				1894	Instruction *pi = ...;
				1895	Instruction *newInst = new Instruction(...);
				1896
				1897	pi->getParent()->getInstList().insert(pi, newInst);
				1898	</pre>
				1899	</div>
				1900
				1901	<p>In fact, this sequence of steps occurs so frequently that the
				1902	<tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
				1903	constructors which take (as a default parameter) a pointer to an
				1904	<tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
				1905	precede. That is, <tt>Instruction</tt> constructors are capable of
				1906	inserting the newly-created instance into the <tt>BasicBlock</tt> of a
				1907	provided instruction, immediately before that instruction. Using an
				1908	<tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
				1909	parameter, the above code becomes:</p>
				1910
				1911	<div class="doc_code">
				1912	<pre>
				1913	Instruction* pi = ...;
				1914	Instruction* newInst = new Instruction(..., pi);
				1915	</pre>
				1916	</div>
				1917
				1918	<p>which is much cleaner, especially if you're creating a lot of
				1919	instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
				1920	</ul>
				1921
				1922	</div>
				1923
				1924	<!--_______________________________________________________________________-->
				1925	<div class="doc_subsubsection">
				1926	<a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
				1927	</div>
				1928
				1929	<div class="doc_text">
				1930
				1931	<p>Deleting an instruction from an existing sequence of instructions that form a
				1932	<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
				1933	you must have a pointer to the instruction that you wish to delete. Second, you
				1934	need to obtain the pointer to that instruction's basic block. You use the
				1935	pointer to the basic block to get its list of instructions and then use the
				1936	erase function to remove your instruction. For example:</p>
				1937
				1938	<div class="doc_code">
				1939	<pre>
				1940	<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner	3db8f77	2008-02-15 22:57:17 +0000	[diff] [blame]	1941	I->eraseFromParent();
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1942	</pre>
				1943	</div>
				1944
				1945	</div>
				1946
				1947	<!--_______________________________________________________________________-->
				1948	<div class="doc_subsubsection">
				1949	<a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
				1950	<tt>Value</tt></a>
				1951	</div>
				1952
				1953	<div class="doc_text">
				1954
				1955	<p><i>Replacing individual instructions</i></p>
				1956
				1957	<p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
				1958	permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
				1959	and <tt>ReplaceInstWithInst</tt>.</p>
				1960
				1961	<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
				1962
				1963	<ul>
				1964	<li><tt>ReplaceInstWithValue</tt>
				1965
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	1966	<p>This function replaces all uses of a given instruction with a value,
				1967	and then removes the original instruction. The following example
				1968	illustrates the replacement of the result of a particular
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1969	<tt>AllocaInst</tt> that allocates memory for a single integer with a null
				1970	pointer to an integer.</p>
				1971
				1972	<div class="doc_code">
				1973	<pre>
				1974	AllocaInst* instToReplace = ...;
				1975	BasicBlock::iterator ii(instToReplace);
				1976
				1977	ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	1978	Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1979	</pre></div></li>
				1980
				1981	<li><tt>ReplaceInstWithInst</tt>
				1982
				1983	<p>This function replaces a particular instruction with another
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	1984	instruction, inserting the new instruction into the basic block at the
				1985	location where the old instruction was, and replacing any uses of the old
				1986	instruction with the new instruction. The following example illustrates
				1987	the replacement of one <tt>AllocaInst</tt> with another.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1988
				1989	<div class="doc_code">
				1990	<pre>
				1991	AllocaInst* instToReplace = ...;
				1992	BasicBlock::iterator ii(instToReplace);
				1993
				1994	ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	1995	new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1996	</pre></div></li>
				1997	</ul>
				1998
				1999	<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
				2000
				2001	<p>You can use <tt>Value::replaceAllUsesWith</tt> and
				2002	<tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
				2003	doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
				2004	and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
				2005	information.</p>
				2006
				2007	<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
				2008	include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
				2009	ReplaceInstWithValue, ReplaceInstWithInst -->
				2010
				2011	</div>
				2012
				2013	<!--_______________________________________________________________________-->
				2014	<div class="doc_subsubsection">
				2015	<a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
				2016	</div>
				2017
				2018	<div class="doc_text">
				2019
				2020	<p>Deleting a global variable from a module is just as easy as deleting an
				2021	Instruction. First, you must have a pointer to the global variable that you wish
				2022	to delete. You use this pointer to erase it from its parent, the module.
				2023	For example:</p>
				2024
				2025	<div class="doc_code">
				2026	<pre>
				2027	<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
				2028
				2029	GV->eraseFromParent();
				2030	</pre>
				2031	</div>
				2032
				2033	</div>
				2034
				2035	<!-- *********************************************************************** -->
				2036	<div class="doc_section">
				2037	<a name="advanced">Advanced Topics</a>
				2038	</div>
				2039	<!-- *********************************************************************** -->
				2040
				2041	<div class="doc_text">
				2042	<p>
				2043	This section describes some of the advanced or obscure API's that most clients
				2044	do not need to be aware of. These API's tend manage the inner workings of the
				2045	LLVM system, and only need to be accessed in unusual circumstances.
				2046	</p>
				2047	</div>
				2048
				2049	<!-- ======================================================================= -->
				2050	<div class="doc_subsection">
				2051	<a name="TypeResolve">LLVM Type Resolution</a>
				2052	</div>
				2053
				2054	<div class="doc_text">
				2055
				2056	<p>
				2057	The LLVM type system has a very simple goal: allow clients to compare types for
				2058	structural equality with a simple pointer comparison (aka a shallow compare).
				2059	This goal makes clients much simpler and faster, and is used throughout the LLVM
				2060	system.
				2061	</p>
				2062
				2063	<p>
				2064	Unfortunately achieving this goal is not a simple matter. In particular,
				2065	recursive types and late resolution of opaque types makes the situation very
				2066	difficult to handle. Fortunately, for the most part, our implementation makes
				2067	most clients able to be completely unaware of the nasty internal details. The
				2068	primary case where clients are exposed to the inner workings of it are when
				2069	building a recursive type. In addition to this case, the LLVM bitcode reader,
				2070	assembly parser, and linker also have to be aware of the inner workings of this
				2071	system.
				2072	</p>
				2073
				2074	<p>
				2075	For our purposes below, we need three concepts. First, an "Opaque Type" is
				2076	exactly as defined in the <a href="LangRef.html#t_opaque">language
				2077	reference</a>. Second an "Abstract Type" is any type which includes an
				2078	opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
				2079	Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
				2080	float }</tt>").
				2081	</p>
				2082
				2083	</div>
				2084
				2085	<!-- ______________________________________________________________________ -->
				2086	<div class="doc_subsubsection">
				2087	<a name="BuildRecType">Basic Recursive Type Construction</a>
				2088	</div>
				2089
				2090	<div class="doc_text">
				2091
				2092	<p>
				2093	Because the most common question is "how do I build a recursive type with LLVM",
				2094	we answer it now and explain it as we go. Here we include enough to cause this
				2095	to be emitted to an output .ll file:
				2096	</p>
				2097
				2098	<div class="doc_code">
				2099	<pre>
				2100	%mylist = type { %mylist*, i32 }
				2101	</pre>
				2102	</div>
				2103
				2104	<p>
				2105	To build this, use the following LLVM APIs:
				2106	</p>
				2107
				2108	<div class="doc_code">
				2109	<pre>
				2110	// <i>Create the initial outer struct</i>
				2111	<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
				2112	std::vector<const Type*> Elts;
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	2113	Elts.push_back(PointerType::getUnqual(StructTy));
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2114	Elts.push_back(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2115	StructType *NewSTy = StructType::get(Elts);
				2116
				2117	// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
				2118	// <i>the struct and the opaque type are actually the same.</i>
				2119	cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
				2120
				2121	// <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
				2122	// <i>kept up-to-date</i>
				2123	NewSTy = cast<StructType>(StructTy.get());
				2124
				2125	// <i>Add a name for the type to the module symbol table (optional)</i>
				2126	MyModule->addTypeName("mylist", NewSTy);
				2127	</pre>
				2128	</div>
				2129
				2130	<p>
				2131	This code shows the basic approach used to build recursive types: build a
				2132	non-recursive type using 'opaque', then use type unification to close the cycle.
				2133	The type unification step is performed by the <tt><a
				2134	href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
				2135	described next. After that, we describe the <a
				2136	href="#PATypeHolder">PATypeHolder class</a>.
				2137	</p>
				2138
				2139	</div>
				2140
				2141	<!-- ______________________________________________________________________ -->
				2142	<div class="doc_subsubsection">
				2143	<a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
				2144	</div>
				2145
				2146	<div class="doc_text">
				2147	<p>
				2148	The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
				2149	While this method is actually a member of the DerivedType class, it is most
				2150	often used on OpaqueType instances. Type unification is actually a recursive
				2151	process. After unification, types can become structurally isomorphic to
				2152	existing types, and all duplicates are deleted (to preserve pointer equality).
				2153	</p>
				2154
				2155	<p>
				2156	In the example above, the OpaqueType object is definitely deleted.
				2157	Additionally, if there is an "{ \2*, i32}" type already created in the system,
				2158	the pointer and struct type created are <b>also</b> deleted. Obviously whenever
				2159	a type is deleted, any "Type*" pointers in the program are invalidated. As
				2160	such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
				2161	live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
				2162	types can never move or be deleted). To deal with this, the <a
				2163	href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
				2164	reference to a possibly refined type, and the <a
				2165	href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
				2166	complex datastructures.
				2167	</p>
				2168
				2169	</div>
				2170
				2171	<!-- ______________________________________________________________________ -->
				2172	<div class="doc_subsubsection">
				2173	<a name="PATypeHolder">The PATypeHolder Class</a>
				2174	</div>
				2175
				2176	<div class="doc_text">
				2177	<p>
				2178	PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
				2179	happily goes about nuking types that become isomorphic to existing types, it
				2180	automatically updates all PATypeHolder objects to point to the new type. In the
				2181	example above, this allows the code to maintain a pointer to the resultant
				2182	resolved recursive type, even though the Type*'s are potentially invalidated.
				2183	</p>
				2184
				2185	<p>
				2186	PATypeHolder is an extremely light-weight object that uses a lazy union-find
				2187	implementation to update pointers. For example the pointer from a Value to its
				2188	Type is maintained by PATypeHolder objects.
				2189	</p>
				2190
				2191	</div>
				2192
				2193	<!-- ______________________________________________________________________ -->
				2194	<div class="doc_subsubsection">
				2195	<a name="AbstractTypeUser">The AbstractTypeUser Class</a>
				2196	</div>
				2197
				2198	<div class="doc_text">
				2199
				2200	<p>
				2201	Some data structures need more to perform more complex updates when types get
				2202	resolved. To support this, a class can derive from the AbstractTypeUser class.
				2203	This class
				2204	allows it to get callbacks when certain types are resolved. To register to get
				2205	callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
				2206	methods can be called on a type. Note that these methods only work for <i>
				2207	abstract</i> types. Concrete types (those that do not include any opaque
				2208	objects) can never be refined.
				2209	</p>
				2210	</div>
				2211
				2212
				2213	<!-- ======================================================================= -->
				2214	<div class="doc_subsection">
				2215	<a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
				2216	<tt>TypeSymbolTable</tt> classes</a>
				2217	</div>
				2218
				2219	<div class="doc_text">
				2220	<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
				2221	ValueSymbolTable</a></tt> class provides a symbol table that the <a
				2222	href="#Function"><tt>Function</tt></a> and <a href="#Module">
				2223	<tt>Module</tt></a> classes use for naming value definitions. The symbol table
				2224	can provide a name for any <a href="#Value"><tt>Value</tt></a>.
				2225	The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
				2226	TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
				2227	names for types.</p>
				2228
				2229	<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
				2230	by most clients. It should only be used when iteration over the symbol table
				2231	names themselves are required, which is very special purpose. Note that not
				2232	all LLVM
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2233	<tt><a href="#Value">Value</a></tt>s have names, and those without names (i.e. they have
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2234	an empty name) do not exist in the symbol table.
				2235	</p>
				2236
				2237	<p>These symbol tables support iteration over the values/types in the symbol
				2238	table with <tt>begin/end/iterator</tt> and supports querying to see if a
				2239	specific name is in the symbol table (with <tt>lookup</tt>). The
				2240	<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
				2241	simply call <tt>setName</tt> on a value, which will autoinsert it into the
				2242	appropriate symbol table. For types, use the Module::addTypeName method to
				2243	insert entries into the symbol table.</p>
				2244
				2245	</div>
				2246
				2247
				2248
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2249	<!-- ======================================================================= -->
				2250	<div class="doc_subsection">
				2251	<a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
				2252	</div>
				2253
				2254	<div class="doc_text">
				2255	<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1User.html">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2256	User</a></tt> class provides a basis for expressing the ownership of <tt>User</tt>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2257	towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
				2258	Value</a></tt>s. The <tt><a href="http://llvm.org/doxygen/classllvm_1_1Use.html">
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2259	Use</a></tt> helper class is employed to do the bookkeeping and to facilitate <i>O(1)</i>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2260	addition and removal.</p>
				2261
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2262	<!-- ______________________________________________________________________ -->
				2263	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2264	<a name="Use2User">Interaction and relationship between <tt>User</tt> and <tt>Use</tt> objects</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2265	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2266
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2267	<div class="doc_text">
				2268	<p>
				2269	A subclass of <tt>User</tt> can choose between incorporating its <tt>Use</tt> objects
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2270	or refer to them out-of-line by means of a pointer. A mixed variant
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2271	(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
				2272	that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
				2273	</p>
				2274	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2275
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2276	<p>
				2277	We have 2 different layouts in the <tt>User</tt> (sub)classes:
				2278	<ul>
				2279	<li><p>Layout a)
				2280	The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
				2281	object and there are a fixed number of them.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2282
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2283	<li><p>Layout b)
				2284	The <tt>Use</tt> object(s) are referenced by a pointer to an
				2285	array from the <tt>User</tt> object and there may be a variable
				2286	number of them.</p>
				2287	</ul>
				2288	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2289	As of v2.4 each layout still possesses a direct pointer to the
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2290	start of the array of <tt>Use</tt>s. Though not mandatory for layout a),
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2291	we stick to this redundancy for the sake of simplicity.
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2292	The <tt>User</tt> object also stores the number of <tt>Use</tt> objects it
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2293	has. (Theoretically this information can also be calculated
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2294	given the scheme presented below.)</p>
				2295	<p>
				2296	Special forms of allocation operators (<tt>operator new</tt>)
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2297	enforce the following memory layouts:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2298
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2299	<ul>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2300	<li><p>Layout a) is modelled by prepending the <tt>User</tt> object by the <tt>Use[]</tt> array.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2301
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2302	<pre>
				2303	...---.---.---.---.-------...
				2304	\| P \| P \| P \| P \| User
				2305	'''---'---'---'---'-------'''
				2306	</pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2307
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2308	<li><p>Layout b) is modelled by pointing at the <tt>Use[]</tt> array.</p>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2309	<pre>
				2310	.-------...
				2311	\| User
				2312	'-------'''
				2313	\|
				2314	v
				2315	.---.---.---.---...
				2316	\| P \| P \| P \| P \|
				2317	'---'---'---'---'''
				2318	</pre>
				2319	</ul>
				2320	<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
				2321	is stored in each <tt>Use</tt> object in the member <tt>Use::Prev</tt>)</i>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2322
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2323	<!-- ______________________________________________________________________ -->
				2324	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2325	<a name="Waymarking">The waymarking algorithm</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2326	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2327
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2328	<div class="doc_text">
				2329	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2330	Since the <tt>Use</tt> objects are deprived of the direct (back)pointer to
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2331	their <tt>User</tt> objects, there must be a fast and exact method to
				2332	recover it. This is accomplished by the following scheme:</p>
				2333	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2334
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2335	A bit-encoding in the 2 LSBits (least significant bits) of the <tt>Use::Prev</tt> allows to find the
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2336	start of the <tt>User</tt> object:
				2337	<ul>
				2338	<li><tt>00</tt> —> binary digit 0</li>
				2339	<li><tt>01</tt> —> binary digit 1</li>
				2340	<li><tt>10</tt> —> stop and calculate (<tt>s</tt>)</li>
				2341	<li><tt>11</tt> —> full stop (<tt>S</tt>)</li>
				2342	</ul>
				2343	<p>
				2344	Given a <tt>Use*</tt>, all we have to do is to walk till we get
				2345	a stop and we either have a <tt>User</tt> immediately behind or
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2346	we have to walk to the next stop picking up digits
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2347	and calculating the offset:</p>
				2348	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2349	.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
				2350	\| 1 \| s \| 1 \| 0 \| 1 \| 0 \| s \| 1 \| 1 \| 0 \| s \| 1 \| 1 \| s \| 1 \| S \| User (or User*)
				2351	'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
				2352	\|+15 \|+10 \|+6 \|+3 \|+1
				2353	\| \| \| \| \|__>
				2354	\| \| \| \|__________>
				2355	\| \| \|______________________>
				2356	\| \|______________________________________>
				2357	\|__________________________________________________________>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2358	</pre>
				2359	<p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2360	Only the significant number of bits need to be stored between the
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2361	stops, so that the <i>worst case is 20 memory accesses</i> when there are
				2362	1000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2363
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2364	<!-- ______________________________________________________________________ -->
				2365	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2366	<a name="ReferenceImpl">Reference implementation</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2367	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2368
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2369	<div class="doc_text">
				2370	<p>
				2371	The following literate Haskell fragment demonstrates the concept:</p>
				2372	</div>
				2373
				2374	<div class="doc_code">
				2375	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2376	> import Test.QuickCheck
				2377	>
				2378	> digits :: Int -> [Char] -> [Char]
				2379	> digits 0 acc = '0' : acc
				2380	> digits 1 acc = '1' : acc
				2381	> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
				2382	>
				2383	> dist :: Int -> [Char] -> [Char]
				2384	> dist 0 [] = ['S']
				2385	> dist 0 acc = acc
				2386	> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
				2387	> dist n acc = dist (n - 1) $ dist 1 acc
				2388	>
				2389	> takeLast n ss = reverse $ take n $ reverse ss
				2390	>
				2391	> test = takeLast 40 $ dist 20 []
				2392	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2393	</pre>
				2394	</div>
				2395	<p>
				2396	Printing <test> gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
				2397	<p>
				2398	The reverse algorithm computes the length of the string just by examining
				2399	a certain prefix:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2400
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2401	<div class="doc_code">
				2402	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2403	> pref :: [Char] -> Int
				2404	> pref "S" = 1
				2405	> pref ('s':'1':rest) = decode 2 1 rest
				2406	> pref (_:rest) = 1 + pref rest
				2407	>
				2408	> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
				2409	> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
				2410	> decode walk acc _ = walk + acc
				2411	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2412	</pre>
				2413	</div>
				2414	<p>
				2415	Now, as expected, printing <pref test> gives <tt>40</tt>.</p>
				2416	<p>
				2417	We can <i>quickCheck</i> this with following property:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2418
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2419	<div class="doc_code">
				2420	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2421	> testcase = dist 2000 []
				2422	> testcaseLength = length testcase
				2423	>
				2424	> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
				2425	> where arr = takeLast n testcase
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2426	>
				2427	</pre>
				2428	</div>
				2429	<p>
				2430	As expected <quickCheck identityProp> gives:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2431
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2432	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2433	*Main> quickCheck identityProp
				2434	OK, passed 100 tests.
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2435	</pre>
				2436	<p>
				2437	Let's be a bit more exhaustive:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2438
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2439	<div class="doc_code">
				2440	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2441	>
				2442	> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
				2443	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2444	</pre>
				2445	</div>
				2446	<p>
				2447	And here is the result of <deepCheck identityProp>:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2448
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2449	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2450	*Main> deepCheck identityProp
				2451	OK, passed 500 tests.
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2452	</pre>
				2453
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2454	<!-- ______________________________________________________________________ -->
				2455	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2456	<a name="Tagging">Tagging considerations</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2457	</div>
				2458
				2459	<p>
				2460	To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
				2461	never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
				2462	new <tt>Use**</tt> on every modification. Accordingly getters must strip the
				2463	tag bits.</p>
				2464	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2465	For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
				2466	Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
				2467	that the first bytes of <tt>User</tt> (if interpreted as a pointer) never has
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2468	the LSBit set. (Portability is relying on the fact that all known compilers place the
				2469	<tt>vptr</tt> in the first word of the instances.)</p>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2470
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2471	</div>
				2472
				2473	<!-- *********************************************************************** -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2474	<div class="doc_section">
				2475	<a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
				2476	</div>
				2477	<!-- *********************************************************************** -->
				2478
				2479	<div class="doc_text">
				2480	<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
				2481	<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
				2482
				2483	<p>The Core LLVM classes are the primary means of representing the program
				2484	being inspected or transformed. The core LLVM classes are defined in
				2485	header files in the <tt>include/llvm/</tt> directory, and implemented in
				2486	the <tt>lib/VMCore</tt> directory.</p>
				2487
				2488	</div>
				2489
				2490	<!-- ======================================================================= -->
				2491	<div class="doc_subsection">
				2492	<a name="Type">The <tt>Type</tt> class and Derived Types</a>
				2493	</div>
				2494
				2495	<div class="doc_text">
				2496
				2497	<p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
				2498	a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
				2499	through its subclasses. Certain primitive types (<tt>VoidType</tt>,
				2500	<tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
				2501	subclasses. They are hidden because they offer no useful functionality beyond
				2502	what the <tt>Type</tt> class offers except to distinguish themselves from
				2503	other subclasses of <tt>Type</tt>.</p>
				2504	<p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
				2505	named, but this is not a requirement. There exists exactly
				2506	one instance of a given shape at any one time. This allows type equality to
				2507	be performed with address equality of the Type Instance. That is, given two
				2508	<tt>Type*</tt> values, the types are identical if the pointers are identical.
				2509	</p>
				2510	</div>
				2511
				2512	<!-- _______________________________________________________________________ -->
				2513	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2514	<a name="m_Type">Important Public Methods</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2515	</div>
				2516
				2517	<div class="doc_text">
				2518
				2519	<ul>
				2520	<li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
				2521
				2522	<li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
				2523	floating point types.</li>
				2524
				2525	<li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
				2526	an OpaqueType anywhere in its definition).</li>
				2527
				2528	<li><tt>bool isSized()</tt>: Return true if the type has known size. Things
				2529	that don't have a size are abstract types, labels and void.</li>
				2530
				2531	</ul>
				2532	</div>
				2533
				2534	<!-- _______________________________________________________________________ -->
				2535	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2536	<a name="derivedtypes">Important Derived Types</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2537	</div>
				2538	<div class="doc_text">
				2539	<dl>
				2540	<dt><tt>IntegerType</tt></dt>
				2541	<dd>Subclass of DerivedType that represents integer types of any bit width.
				2542	Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
				2543	<tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
				2544	<ul>
				2545	<li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
				2546	type of a specific bit width.</li>
				2547	<li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
				2548	type.</li>
				2549	</ul>
				2550	</dd>
				2551	<dt><tt>SequentialType</tt></dt>
				2552	<dd>This is subclassed by ArrayType and PointerType
				2553	<ul>
				2554	<li><tt>const Type * getElementType() const</tt>: Returns the type of each
				2555	of the elements in the sequential type. </li>
				2556	</ul>
				2557	</dd>
				2558	<dt><tt>ArrayType</tt></dt>
				2559	<dd>This is a subclass of SequentialType and defines the interface for array
				2560	types.
				2561	<ul>
				2562	<li><tt>unsigned getNumElements() const</tt>: Returns the number of
				2563	elements in the array. </li>
				2564	</ul>
				2565	</dd>
				2566	<dt><tt>PointerType</tt></dt>
				2567	<dd>Subclass of SequentialType for pointer types.</dd>
				2568	<dt><tt>VectorType</tt></dt>
				2569	<dd>Subclass of SequentialType for vector types. A
				2570	vector type is similar to an ArrayType but is distinguished because it is
				2571	a first class type wherease ArrayType is not. Vector types are used for
				2572	vector operations and are usually small vectors of of an integer or floating
				2573	point type.</dd>
				2574	<dt><tt>StructType</tt></dt>
				2575	<dd>Subclass of DerivedTypes for struct types.</dd>
				2576	<dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
				2577	<dd>Subclass of DerivedTypes for function types.
				2578	<ul>
				2579	<li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
				2580	function</li>
				2581	<li><tt> const Type * getReturnType() const</tt>: Returns the
				2582	return type of the function.</li>
				2583	<li><tt>const Type * getParamType (unsigned i)</tt>: Returns
				2584	the type of the ith parameter.</li>
				2585	<li><tt> const unsigned getNumParams() const</tt>: Returns the
				2586	number of formal parameters.</li>
				2587	</ul>
				2588	</dd>
				2589	<dt><tt>OpaqueType</tt></dt>
				2590	<dd>Sublcass of DerivedType for abstract types. This class
				2591	defines no content and is used as a placeholder for some other type. Note
				2592	that OpaqueType is used (temporarily) during type resolution for forward
				2593	references of types. Once the referenced type is resolved, the OpaqueType
				2594	is replaced with the actual type. OpaqueType can also be used for data
				2595	abstraction. At link time opaque types can be resolved to actual types
				2596	of the same name.</dd>
				2597	</dl>
				2598	</div>
				2599
				2600
				2601
				2602	<!-- ======================================================================= -->
				2603	<div class="doc_subsection">
				2604	<a name="Module">The <tt>Module</tt> class</a>
				2605	</div>
				2606
				2607	<div class="doc_text">
				2608
				2609	<p><tt>#include "<a
				2610	href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
				2611	<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
				2612
				2613	<p>The <tt>Module</tt> class represents the top level structure present in LLVM
				2614	programs. An LLVM module is effectively either a translation unit of the
				2615	original program or a combination of several translation units merged by the
				2616	linker. The <tt>Module</tt> class keeps track of a list of <a
				2617	href="#Function"><tt>Function</tt></a>s, a list of <a
				2618	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
				2619	href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
				2620	helpful member functions that try to make common operations easy.</p>
				2621
				2622	</div>
				2623
				2624	<!-- _______________________________________________________________________ -->
				2625	<div class="doc_subsubsection">
				2626	<a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
				2627	</div>
				2628
				2629	<div class="doc_text">
				2630
				2631	<ul>
				2632	<li><tt>Module::Module(std::string name = "")</tt></li>
				2633	</ul>
				2634
				2635	<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
				2636	provide a name for it (probably based on the name of the translation unit).</p>
				2637
				2638	<ul>
				2639	<li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
				2640	<tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
				2641
				2642	<tt>begin()</tt>, <tt>end()</tt>
				2643	<tt>size()</tt>, <tt>empty()</tt>
				2644
				2645	<p>These are forwarding methods that make it easy to access the contents of
				2646	a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
				2647	list.</p></li>
				2648
				2649	<li><tt>Module::FunctionListType &getFunctionList()</tt>
				2650
				2651	<p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
				2652	necessary to use when you need to update the list or perform a complex
				2653	action that doesn't have a forwarding method.</p>
				2654
				2655	<p><!-- Global Variable --></p></li>
				2656	</ul>
				2657
				2658	<hr>
				2659
				2660	<ul>
				2661	<li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
				2662
				2663	<tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
				2664
				2665	<tt>global_begin()</tt>, <tt>global_end()</tt>
				2666	<tt>global_size()</tt>, <tt>global_empty()</tt>
				2667
				2668	<p> These are forwarding methods that make it easy to access the contents of
				2669	a <tt>Module</tt> object's <a
				2670	href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
				2671
				2672	<li><tt>Module::GlobalListType &getGlobalList()</tt>
				2673
				2674	<p>Returns the list of <a
				2675	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
				2676	use when you need to update the list or perform a complex action that
				2677	doesn't have a forwarding method.</p>
				2678
				2679	<p><!-- Symbol table stuff --> </p></li>
				2680	</ul>
				2681
				2682	<hr>
				2683
				2684	<ul>
				2685	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				2686
				2687	<p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				2688	for this <tt>Module</tt>.</p>
				2689
				2690	<p><!-- Convenience methods --></p></li>
				2691	</ul>
				2692
				2693	<hr>
				2694
				2695	<ul>
				2696	<li><tt><a href="#Function">Function</a> *getFunction(const std::string
				2697	&Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
				2698
				2699	<p>Look up the specified function in the <tt>Module</tt> <a
				2700	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
				2701	<tt>null</tt>.</p></li>
				2702
				2703	<li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
				2704	std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
				2705
				2706	<p>Look up the specified function in the <tt>Module</tt> <a
				2707	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
				2708	external declaration for the function and return it.</p></li>
				2709
				2710	<li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
				2711
				2712	<p>If there is at least one entry in the <a
				2713	href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
				2714	href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
				2715	string.</p></li>
				2716
				2717	<li><tt>bool addTypeName(const std::string &Name, const <a
				2718	href="#Type">Type</a> *Ty)</tt>
				2719
				2720	<p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				2721	mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
				2722	name, true is returned and the <a
				2723	href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
				2724	</ul>
				2725
				2726	</div>
				2727
				2728
				2729	<!-- ======================================================================= -->
				2730	<div class="doc_subsection">
				2731	<a name="Value">The <tt>Value</tt> class</a>
				2732	</div>
				2733
				2734	<div class="doc_text">
				2735
				2736	<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
				2737	<br>
				2738	doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
				2739
				2740	<p>The <tt>Value</tt> class is the most important class in the LLVM Source
				2741	base. It represents a typed value that may be used (among other things) as an
				2742	operand to an instruction. There are many different types of <tt>Value</tt>s,
				2743	such as <a href="#Constant"><tt>Constant</tt></a>s,<a
				2744	href="#Argument"><tt>Argument</tt></a>s. Even <a
				2745	href="#Instruction"><tt>Instruction</tt></a>s and <a
				2746	href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
				2747
				2748	<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
				2749	for a program. For example, an incoming argument to a function (represented
				2750	with an instance of the <a href="#Argument">Argument</a> class) is "used" by
				2751	every instruction in the function that references the argument. To keep track
				2752	of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
				2753	href="#User"><tt>User</tt></a>s that is using it (the <a
				2754	href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
				2755	graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
				2756	def-use information in the program, and is accessible through the <tt>use_</tt>*
				2757	methods, shown below.</p>
				2758
				2759	<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
				2760	and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
				2761	method. In addition, all LLVM values can be named. The "name" of the
				2762	<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
				2763
				2764	<div class="doc_code">
				2765	<pre>
				2766	%<b>foo</b> = add i32 1, 2
				2767	</pre>
				2768	</div>
				2769
				2770	<p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
				2771	that the name of any value may be missing (an empty string), so names should
				2772	<b>ONLY</b> be used for debugging (making the source code easier to read,
				2773	debugging printouts), they should not be used to keep track of values or map
				2774	between them. For this purpose, use a <tt>std::map</tt> of pointers to the
				2775	<tt>Value</tt> itself instead.</p>
				2776
				2777	<p>One important aspect of LLVM is that there is no distinction between an SSA
				2778	variable and the operation that produces it. Because of this, any reference to
				2779	the value produced by an instruction (or the value available as an incoming
				2780	argument, for example) is represented as a direct pointer to the instance of
				2781	the class that
				2782	represents this value. Although this may take some getting used to, it
				2783	simplifies the representation and makes it easier to manipulate.</p>
				2784
				2785	</div>
				2786
				2787	<!-- _______________________________________________________________________ -->
				2788	<div class="doc_subsubsection">
				2789	<a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
				2790	</div>
				2791
				2792	<div class="doc_text">
				2793
				2794	<ul>
				2795	<li><tt>Value::use_iterator</tt> - Typedef for iterator over the
				2796	use-list<br>
				2797	<tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
				2798	the use-list<br>
				2799	<tt>unsigned use_size()</tt> - Returns the number of users of the
				2800	value.<br>
				2801	<tt>bool use_empty()</tt> - Returns true if there are no users.<br>
				2802	<tt>use_iterator use_begin()</tt> - Get an iterator to the start of
				2803	the use-list.<br>
				2804	<tt>use_iterator use_end()</tt> - Get an iterator to the end of the
				2805	use-list.<br>
				2806	<tt><a href="#User">User</a> *use_back()</tt> - Returns the last
				2807	element in the list.
				2808	<p> These methods are the interface to access the def-use
				2809	information in LLVM. As with all other iterators in LLVM, the naming
				2810	conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
				2811	</li>
				2812	<li><tt><a href="#Type">Type</a> *getType() const</tt>
				2813	<p>This method returns the Type of the Value.</p>
				2814	</li>
				2815	<li><tt>bool hasName() const</tt><br>
				2816	<tt>std::string getName() const</tt><br>
				2817	<tt>void setName(const std::string &Name)</tt>
				2818	<p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
				2819	be aware of the <a href="#nameWarning">precaution above</a>.</p>
				2820	</li>
				2821	<li><tt>void replaceAllUsesWith(Value *V)</tt>
				2822
				2823	<p>This method traverses the use list of a <tt>Value</tt> changing all <a
				2824	href="#User"><tt>User</tt>s</a> of the current value to refer to
				2825	"<tt>V</tt>" instead. For example, if you detect that an instruction always
				2826	produces a constant value (for example through constant folding), you can
				2827	replace all uses of the instruction with the constant like this:</p>
				2828
				2829	<div class="doc_code">
				2830	<pre>
				2831	Inst->replaceAllUsesWith(ConstVal);
				2832	</pre>
				2833	</div>
				2834
				2835	</ul>
				2836
				2837	</div>
				2838
				2839	<!-- ======================================================================= -->
				2840	<div class="doc_subsection">
				2841	<a name="User">The <tt>User</tt> class</a>
				2842	</div>
				2843
				2844	<div class="doc_text">
				2845
				2846	<p>
				2847	<tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
				2848	doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
				2849	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				2850
				2851	<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
				2852	refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
				2853	that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
				2854	referring to. The <tt>User</tt> class itself is a subclass of
				2855	<tt>Value</tt>.</p>
				2856
				2857	<p>The operands of a <tt>User</tt> point directly to the LLVM <a
				2858	href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
				2859	Single Assignment (SSA) form, there can only be one definition referred to,
				2860	allowing this direct connection. This connection provides the use-def
				2861	information in LLVM.</p>
				2862
				2863	</div>
				2864
				2865	<!-- _______________________________________________________________________ -->
				2866	<div class="doc_subsubsection">
				2867	<a name="m_User">Important Public Members of the <tt>User</tt> class</a>
				2868	</div>
				2869
				2870	<div class="doc_text">
				2871
				2872	<p>The <tt>User</tt> class exposes the operand list in two ways: through
				2873	an index access interface and through an iterator based interface.</p>
				2874
				2875	<ul>
				2876	<li><tt>Value *getOperand(unsigned i)</tt><br>
				2877	<tt>unsigned getNumOperands()</tt>
				2878	<p> These two methods expose the operands of the <tt>User</tt> in a
				2879	convenient form for direct access.</p></li>
				2880
				2881	<li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
				2882	list<br>
				2883	<tt>op_iterator op_begin()</tt> - Get an iterator to the start of
				2884	the operand list.<br>
				2885	<tt>op_iterator op_end()</tt> - Get an iterator to the end of the
				2886	operand list.
				2887	<p> Together, these methods make up the iterator based interface to
				2888	the operands of a <tt>User</tt>.</p></li>
				2889	</ul>
				2890
				2891	</div>
				2892
				2893	<!-- ======================================================================= -->
				2894	<div class="doc_subsection">
				2895	<a name="Instruction">The <tt>Instruction</tt> class</a>
				2896	</div>
				2897
				2898	<div class="doc_text">
				2899
				2900	<p><tt>#include "</tt><tt><a
				2901	href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
				2902	doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
				2903	Superclasses: <a href="#User"><tt>User</tt></a>, <a
				2904	href="#Value"><tt>Value</tt></a></p>
				2905
				2906	<p>The <tt>Instruction</tt> class is the common base class for all LLVM
				2907	instructions. It provides only a few methods, but is a very commonly used
				2908	class. The primary data tracked by the <tt>Instruction</tt> class itself is the
				2909	opcode (instruction type) and the parent <a
				2910	href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
				2911	into. To represent a specific type of instruction, one of many subclasses of
				2912	<tt>Instruction</tt> are used.</p>
				2913
				2914	<p> Because the <tt>Instruction</tt> class subclasses the <a
				2915	href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
				2916	way as for other <a href="#User"><tt>User</tt></a>s (with the
				2917	<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
				2918	<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
				2919	the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
				2920	file contains some meta-data about the various different types of instructions
				2921	in LLVM. It describes the enum values that are used as opcodes (for example
				2922	<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
				2923	concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
				2924	example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
				2925	href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
				2926	this file confuses doxygen, so these enum values don't show up correctly in the
				2927	<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
				2928
				2929	</div>
				2930
				2931	<!-- _______________________________________________________________________ -->
				2932	<div class="doc_subsubsection">
				2933	<a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
				2934	class</a>
				2935	</div>
				2936	<div class="doc_text">
				2937	<ul>
				2938	<li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
				2939	<p>This subclasses represents all two operand instructions whose operands
				2940	must be the same type, except for the comparison instructions.</p></li>
				2941	<li><tt><a name="CastInst">CastInst</a></tt>
				2942	<p>This subclass is the parent of the 12 casting instructions. It provides
				2943	common operations on cast instructions.</p>
				2944	<li><tt><a name="CmpInst">CmpInst</a></tt>
				2945	<p>This subclass respresents the two comparison instructions,
				2946	<a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
				2947	<a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
				2948	<li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
				2949	<p>This subclass is the parent of all terminator instructions (those which
				2950	can terminate a block).</p>
				2951	</ul>
				2952	</div>
				2953
				2954	<!-- _______________________________________________________________________ -->
				2955	<div class="doc_subsubsection">
				2956	<a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
				2957	class</a>
				2958	</div>
				2959
				2960	<div class="doc_text">
				2961
				2962	<ul>
				2963	<li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
				2964	<p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
				2965	this <tt>Instruction</tt> is embedded into.</p></li>
				2966	<li><tt>bool mayWriteToMemory()</tt>
				2967	<p>Returns true if the instruction writes to memory, i.e. it is a
				2968	<tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
				2969	<li><tt>unsigned getOpcode()</tt>
				2970	<p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
				2971	<li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
				2972	<p>Returns another instance of the specified instruction, identical
				2973	in all ways to the original except that the instruction has no parent
				2974	(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
				2975	and it has no name</p></li>
				2976	</ul>
				2977
				2978	</div>
				2979
				2980	<!-- ======================================================================= -->
				2981	<div class="doc_subsection">
				2982	<a name="Constant">The <tt>Constant</tt> class and subclasses</a>
				2983	</div>
				2984
				2985	<div class="doc_text">
				2986
				2987	<p>Constant represents a base class for different types of constants. It
				2988	is subclassed by ConstantInt, ConstantArray, etc. for representing
				2989	the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
				2990	a subclass, which represents the address of a global variable or function.
				2991	</p>
				2992
				2993	</div>
				2994
				2995	<!-- _______________________________________________________________________ -->
				2996	<div class="doc_subsubsection">Important Subclasses of Constant </div>
				2997	<div class="doc_text">
				2998	<ul>
				2999	<li>ConstantInt : This subclass of Constant represents an integer constant of
				3000	any width.
				3001	<ul>
				3002	<li><tt>const APInt& getValue() const</tt>: Returns the underlying
				3003	value of this constant, an APInt value.</li>
				3004	<li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
				3005	value to an int64_t via sign extension. If the value (not the bit width)
				3006	of the APInt is too large to fit in an int64_t, an assertion will result.
				3007	For this reason, use of this method is discouraged.</li>
				3008	<li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
				3009	value to a uint64_t via zero extension. IF the value (not the bit width)
				3010	of the APInt is too large to fit in a uint64_t, an assertion will result.
				3011	For this reason, use of this method is discouraged.</li>
				3012	<li><tt>static ConstantInt* get(const APInt& Val)</tt>: Returns the
				3013	ConstantInt object that represents the value provided by <tt>Val</tt>.
				3014	The type is implied as the IntegerType that corresponds to the bit width
				3015	of <tt>Val</tt>.</li>
				3016	<li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
				3017	Returns the ConstantInt object that represents the value provided by
				3018	<tt>Val</tt> for integer type <tt>Ty</tt>.</li>
				3019	</ul>
				3020	</li>
				3021	<li>ConstantFP : This class represents a floating point constant.
				3022	<ul>
				3023	<li><tt>double getValue() const</tt>: Returns the underlying value of
				3024	this constant. </li>
				3025	</ul>
				3026	</li>
				3027	<li>ConstantArray : This represents a constant array.
				3028	<ul>
				3029	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3030	a vector of component constants that makeup this array. </li>
				3031	</ul>
				3032	</li>
				3033	<li>ConstantStruct : This represents a constant struct.
				3034	<ul>
				3035	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3036	a vector of component constants that makeup this array. </li>
				3037	</ul>
				3038	</li>
				3039	<li>GlobalValue : This represents either a global variable or a function. In
				3040	either case, the value is a constant fixed address (after linking).
				3041	</li>
				3042	</ul>
				3043	</div>
				3044
				3045
				3046	<!-- ======================================================================= -->
				3047	<div class="doc_subsection">
				3048	<a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
				3049	</div>
				3050
				3051	<div class="doc_text">
				3052
				3053	<p><tt>#include "<a
				3054	href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
				3055	doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
				3056	Class</a><br>
				3057	Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
				3058	<a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
				3059
				3060	<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
				3061	href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
				3062	visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
				3063	Because they are visible at global scope, they are also subject to linking with
				3064	other globals defined in different translation units. To control the linking
				3065	process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
				3066	<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
				3067	defined by the <tt>LinkageTypes</tt> enumeration.</p>
				3068
				3069	<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
				3070	<tt>static</tt> in C), it is not visible to code outside the current translation
				3071	unit, and does not participate in linking. If it has external linkage, it is
				3072	visible to external code, and does participate in linking. In addition to
				3073	linkage information, <tt>GlobalValue</tt>s keep track of which <a
				3074	href="#Module"><tt>Module</tt></a> they are currently part of.</p>
				3075
				3076	<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
				3077	by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
				3078	global is always a pointer to its contents. It is important to remember this
				3079	when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
				3080	be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
				3081	subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
				3082	i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
				3083	the address of the first element of this array and the value of the
				3084	<tt>GlobalVariable</tt> are the same, they have different types. The
				3085	<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
				3086	is <tt>i32.</tt> Because of this, accessing a global value requires you to
				3087	dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
				3088	can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
				3089	Language Reference Manual</a>.</p>
				3090
				3091	</div>
				3092
				3093	<!-- _______________________________________________________________________ -->
				3094	<div class="doc_subsubsection">
				3095	<a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
				3096	class</a>
				3097	</div>
				3098
				3099	<div class="doc_text">
				3100
				3101	<ul>
				3102	<li><tt>bool hasInternalLinkage() const</tt><br>
				3103	<tt>bool hasExternalLinkage() const</tt><br>
				3104	<tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
				3105	<p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
				3106	<p> </p>
				3107	</li>
				3108	<li><tt><a href="#Module">Module</a> *getParent()</tt>
				3109	<p> This returns the <a href="#Module"><tt>Module</tt></a> that the
				3110	GlobalValue is currently embedded into.</p></li>
				3111	</ul>
				3112
				3113	</div>
				3114
				3115	<!-- ======================================================================= -->
				3116	<div class="doc_subsection">
				3117	<a name="Function">The <tt>Function</tt> class</a>
				3118	</div>
				3119
				3120	<div class="doc_text">
				3121
				3122	<p><tt>#include "<a
				3123	href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
				3124	info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
				3125	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3126	<a href="#Constant"><tt>Constant</tt></a>,
				3127	<a href="#User"><tt>User</tt></a>,
				3128	<a href="#Value"><tt>Value</tt></a></p>
				3129
				3130	<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
				3131	actually one of the more complex classes in the LLVM heirarchy because it must
				3132	keep track of a large amount of data. The <tt>Function</tt> class keeps track
				3133	of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
				3134	<a href="#Argument"><tt>Argument</tt></a>s, and a
				3135	<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
				3136
				3137	<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
				3138	commonly used part of <tt>Function</tt> objects. The list imposes an implicit
				3139	ordering of the blocks in the function, which indicate how the code will be
				3140	layed out by the backend. Additionally, the first <a
				3141	href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
				3142	<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
				3143	block. There are no implicit exit nodes, and in fact there may be multiple exit
				3144	nodes from a single <tt>Function</tt>. If the <a
				3145	href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
				3146	the <tt>Function</tt> is actually a function declaration: the actual body of the
				3147	function hasn't been linked in yet.</p>
				3148
				3149	<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
				3150	<tt>Function</tt> class also keeps track of the list of formal <a
				3151	href="#Argument"><tt>Argument</tt></a>s that the function receives. This
				3152	container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
				3153	nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
				3154	the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
				3155
				3156	<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
				3157	LLVM feature that is only used when you have to look up a value by name. Aside
				3158	from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
				3159	internally to make sure that there are not conflicts between the names of <a
				3160	href="#Instruction"><tt>Instruction</tt></a>s, <a
				3161	href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
				3162	href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
				3163
				3164	<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
				3165	and therefore also a <a href="#Constant">Constant</a>. The value of the function
				3166	is its address (after linking) which is guaranteed to be constant.</p>
				3167	</div>
				3168
				3169	<!-- _______________________________________________________________________ -->
				3170	<div class="doc_subsubsection">
				3171	<a name="m_Function">Important Public Members of the <tt>Function</tt>
				3172	class</a>
				3173	</div>
				3174
				3175	<div class="doc_text">
				3176
				3177	<ul>
				3178	<li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
				3179	Ty, LinkageTypes Linkage, const std::string &N = "", Module Parent = 0)</tt>
				3180
				3181	<p>Constructor used when you need to create new <tt>Function</tt>s to add
				3182	the the program. The constructor must specify the type of the function to
				3183	create and what type of linkage the function should have. The <a
				3184	href="#FunctionType"><tt>FunctionType</tt></a> argument
				3185	specifies the formal arguments and return value for the function. The same
				3186	<a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
				3187	create multiple functions. The <tt>Parent</tt> argument specifies the Module
				3188	in which the function is defined. If this argument is provided, the function
				3189	will automatically be inserted into that module's list of
				3190	functions.</p></li>
				3191
Chris Lattner	5e70957	2008-11-25 18:34:50 +0000	[diff] [blame]	3192	<li><tt>bool isDeclaration()</tt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3193
				3194	<p>Return whether or not the <tt>Function</tt> has a body defined. If the
				3195	function is "external", it does not have a body, and thus must be resolved
				3196	by linking with a function defined in a different translation unit.</p></li>
				3197
				3198	<li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
				3199	<tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
				3200
				3201	<tt>begin()</tt>, <tt>end()</tt>
				3202	<tt>size()</tt>, <tt>empty()</tt>
				3203
				3204	<p>These are forwarding methods that make it easy to access the contents of
				3205	a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
				3206	list.</p></li>
				3207
				3208	<li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
				3209
				3210	<p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
				3211	is necessary to use when you need to update the list or perform a complex
				3212	action that doesn't have a forwarding method.</p></li>
				3213
				3214	<li><tt>Function::arg_iterator</tt> - Typedef for the argument list
				3215	iterator<br>
				3216	<tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
				3217
				3218	<tt>arg_begin()</tt>, <tt>arg_end()</tt>
				3219	<tt>arg_size()</tt>, <tt>arg_empty()</tt>
				3220
				3221	<p>These are forwarding methods that make it easy to access the contents of
				3222	a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
				3223	list.</p></li>
				3224
				3225	<li><tt>Function::ArgumentListType &getArgumentList()</tt>
				3226
				3227	<p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
				3228	necessary to use when you need to update the list or perform a complex
				3229	action that doesn't have a forwarding method.</p></li>
				3230
				3231	<li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
				3232
				3233	<p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
				3234	function. Because the entry block for the function is always the first
				3235	block, this returns the first block of the <tt>Function</tt>.</p></li>
				3236
				3237	<li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
				3238	<tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
				3239
				3240	<p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
				3241	<tt>Function</tt> and returns the return type of the function, or the <a
				3242	href="#FunctionType"><tt>FunctionType</tt></a> of the actual
				3243	function.</p></li>
				3244
				3245	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				3246
				3247	<p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				3248	for this <tt>Function</tt>.</p></li>
				3249	</ul>
				3250
				3251	</div>
				3252
				3253	<!-- ======================================================================= -->
				3254	<div class="doc_subsection">
				3255	<a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
				3256	</div>
				3257
				3258	<div class="doc_text">
				3259
				3260	<p><tt>#include "<a
				3261	href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
				3262	<br>
				3263	doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
				3264	Class</a><br>
				3265	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3266	<a href="#Constant"><tt>Constant</tt></a>,
				3267	<a href="#User"><tt>User</tt></a>,
				3268	<a href="#Value"><tt>Value</tt></a></p>
				3269
				3270	<p>Global variables are represented with the (suprise suprise)
				3271	<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
				3272	subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
				3273	always referenced by their address (global values must live in memory, so their
				3274	"name" refers to their constant address). See
				3275	<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
				3276	variables may have an initial value (which must be a
				3277	<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
				3278	they may be marked as "constant" themselves (indicating that their contents
				3279	never change at runtime).</p>
				3280	</div>
				3281
				3282	<!-- _______________________________________________________________________ -->
				3283	<div class="doc_subsubsection">
				3284	<a name="m_GlobalVariable">Important Public Members of the
				3285	<tt>GlobalVariable</tt> class</a>
				3286	</div>
				3287
				3288	<div class="doc_text">
				3289
				3290	<ul>
				3291	<li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
				3292	isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
				3293	Initializer = 0, const std::string &Name = "", Module Parent = 0)</tt>
				3294
				3295	<p>Create a new global variable of the specified type. If
				3296	<tt>isConstant</tt> is true then the global variable will be marked as
				3297	unchanging for the program. The Linkage parameter specifies the type of
				3298	linkage (internal, external, weak, linkonce, appending) for the variable. If
				3299	the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage,  then
				3300	the resultant global variable will have internal linkage. AppendingLinkage
				3301	concatenates together all instances (in different translation units) of the
				3302	variable into a single variable but is only applicable to arrays.  See
				3303	the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
				3304	further details on linkage types. Optionally an initializer, a name, and the
				3305	module to put the variable into may be specified for the global variable as
				3306	well.</p></li>
				3307
				3308	<li><tt>bool isConstant() const</tt>
				3309
				3310	<p>Returns true if this is a global variable that is known not to
				3311	be modified at runtime.</p></li>
				3312
				3313	<li><tt>bool hasInitializer()</tt>
				3314
				3315	<p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
				3316
				3317	<li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
				3318
				3319	<p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
				3320	to call this method if there is no initializer.</p></li>
				3321	</ul>
				3322
				3323	</div>
				3324
				3325
				3326	<!-- ======================================================================= -->
				3327	<div class="doc_subsection">
				3328	<a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
				3329	</div>
				3330
				3331	<div class="doc_text">
				3332
				3333	<p><tt>#include "<a
				3334	href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
				3335	doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
				3336	Class</a><br>
				3337	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				3338
				3339	<p>This class represents a single entry multiple exit section of the code,
				3340	commonly known as a basic block by the compiler community. The
				3341	<tt>BasicBlock</tt> class maintains a list of <a
				3342	href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
				3343	Matching the language definition, the last element of this list of instructions
				3344	is always a terminator instruction (a subclass of the <a
				3345	href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
				3346
				3347	<p>In addition to tracking the list of instructions that make up the block, the
				3348	<tt>BasicBlock</tt> class also keeps track of the <a
				3349	href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
				3350
				3351	<p>Note that <tt>BasicBlock</tt>s themselves are <a
				3352	href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
				3353	like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
				3354	<tt>label</tt>.</p>
				3355
				3356	</div>
				3357
				3358	<!-- _______________________________________________________________________ -->
				3359	<div class="doc_subsubsection">
				3360	<a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
				3361	class</a>
				3362	</div>
				3363
				3364	<div class="doc_text">
				3365	<ul>
				3366
				3367	<li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
				3368	href="#Function">Function</a> *Parent = 0)</tt>
				3369
				3370	<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
				3371	insertion into a function. The constructor optionally takes a name for the new
				3372	block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
				3373	the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
				3374	automatically inserted at the end of the specified <a
				3375	href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
				3376	manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
				3377
				3378	<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
				3379	<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
				3380	<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
				3381	<tt>size()</tt>, <tt>empty()</tt>
				3382	STL-style functions for accessing the instruction list.
				3383
				3384	<p>These methods and typedefs are forwarding functions that have the same
				3385	semantics as the standard library methods of the same names. These methods
				3386	expose the underlying instruction list of a basic block in a way that is easy to
				3387	manipulate. To get the full complement of container operations (including
				3388	operations to update the list), you must use the <tt>getInstList()</tt>
				3389	method.</p></li>
				3390
				3391	<li><tt>BasicBlock::InstListType &getInstList()</tt>
				3392
				3393	<p>This method is used to get access to the underlying container that actually
				3394	holds the Instructions. This method must be used when there isn't a forwarding
				3395	function in the <tt>BasicBlock</tt> class for the operation that you would like
				3396	to perform. Because there are no forwarding functions for "updating"
				3397	operations, you need to use this if you want to update the contents of a
				3398	<tt>BasicBlock</tt>.</p></li>
				3399
				3400	<li><tt><a href="#Function">Function</a> *getParent()</tt>
				3401
				3402	<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
				3403	embedded into, or a null pointer if it is homeless.</p></li>
				3404
				3405	<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
				3406
				3407	<p> Returns a pointer to the terminator instruction that appears at the end of
				3408	the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
				3409	instruction in the block is not a terminator, then a null pointer is
				3410	returned.</p></li>
				3411
				3412	</ul>
				3413
				3414	</div>
				3415
				3416
				3417	<!-- ======================================================================= -->
				3418	<div class="doc_subsection">
				3419	<a name="Argument">The <tt>Argument</tt> class</a>
				3420	</div>
				3421
				3422	<div class="doc_text">
				3423
				3424	<p>This subclass of Value defines the interface for incoming formal
				3425	arguments to a function. A Function maintains a list of its formal
				3426	arguments. An argument has a pointer to the parent Function.</p>
				3427
				3428	</div>
				3429
				3430	<!-- *********************************************************************** -->
				3431	<hr>
				3432	<address>
				3433	<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman	947321d	2008-12-11 17:34:48 +0000	[diff] [blame]	3434	src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3435	<a href="http://validator.w3.org/check/referer"><img
Gabor Greif	7dabe38	2008-06-18 14:05:31 +0000	[diff] [blame]	3436	src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01 Strict"></a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3437
				3438	<a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
				3439	<a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
				3440	<a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
				3441	Last modified: $Date$
				3442	</address>
				3443
				3444	</body>
				3445	</html>