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

blob: a990462c253872967016f288f8b35f1ab517028e [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
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame^]	250	O'Reilly book in the making. It has a decent Standard Library
				251	Reference that rivals Dinkumware's, and is unfortunately no longer free since the
				252	book has been published.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	253
				254	<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
				255	Questions</a></li>
				256
				257	<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
				258	Contains a useful <a
				259	href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
				260	STL</a>.</li>
				261
				262	<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
				263	Page</a></li>
				264
				265	<li><a href="http://64.78.49.204/">
				266	Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
				267	the book).</a></li>
				268
				269	</ol>
				270
				271	<p>You are also encouraged to take a look at the <a
				272	href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
				273	to write maintainable code more than where to put your curly braces.</p>
				274
				275	</div>
				276
				277	<!-- ======================================================================= -->
				278	<div class="doc_subsection">
				279	<a name="stl">Other useful references</a>
				280	</div>
				281
				282	<div class="doc_text">
				283
				284	<ol>
				285	<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
				286	Branch and Tag Primer</a></li>
				287	<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
				288	static and shared libraries across platforms</a></li>
				289	</ol>
				290
				291	</div>
				292
				293	<!-- *********************************************************************** -->
				294	<div class="doc_section">
				295	<a name="apis">Important and useful LLVM APIs</a>
				296	</div>
				297	<!-- *********************************************************************** -->
				298
				299	<div class="doc_text">
				300
				301	<p>Here we highlight some LLVM APIs that are generally useful and good to
				302	know about when writing transformations.</p>
				303
				304	</div>
				305
				306	<!-- ======================================================================= -->
				307	<div class="doc_subsection">
				308	<a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
				309	<tt>dyn_cast<></tt> templates</a>
				310	</div>
				311
				312	<div class="doc_text">
				313
				314	<p>The LLVM source-base makes extensive use of a custom form of RTTI.
				315	These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
				316	operator, but they don't have some drawbacks (primarily stemming from
				317	the fact that <tt>dynamic_cast<></tt> only works on classes that
				318	have a v-table). Because they are used so often, you must know what they
				319	do and how they work. All of these templates are defined in the <a
				320	href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
				321	file (note that you very rarely have to include this file directly).</p>
				322
				323	<dl>
				324	<dt><tt>isa<></tt>: </dt>
				325
				326	<dd><p>The <tt>isa<></tt> operator works exactly like the Java
				327	"<tt>instanceof</tt>" operator. It returns true or false depending on whether
				328	a reference or pointer points to an instance of the specified class. This can
				329	be very useful for constraint checking of various sorts (example below).</p>
				330	</dd>
				331
				332	<dt><tt>cast<></tt>: </dt>
				333
				334	<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]	335	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]	336	an assertion failure if it is not really an instance of the right type. This
				337	should be used in cases where you have some information that makes you believe
				338	that something is of the right type. An example of the <tt>isa<></tt>
				339	and <tt>cast<></tt> template is:</p>
				340
				341	<div class="doc_code">
				342	<pre>
				343	static bool isLoopInvariant(const <a href="#Value">Value</a> V, const Loop L) {
				344	if (isa<<a href="#Constant">Constant</a>>(V) \|\| isa<<a href="#Argument">Argument</a>>(V) \|\| isa<<a href="#GlobalValue">GlobalValue</a>>(V))
				345	return true;
				346
				347	// <i>Otherwise, it must be an instruction...</i>
				348	return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
				349	}
				350	</pre>
				351	</div>
				352
				353	<p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
				354	by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
				355	operator.</p>
				356
				357	</dd>
				358
				359	<dt><tt>dyn_cast<></tt>:</dt>
				360
				361	<dd><p>The <tt>dyn_cast<></tt> operator is a "checking cast" operation.
				362	It checks to see if the operand is of the specified type, and if so, returns a
				363	pointer to it (this operator does not work with references). If the operand is
				364	not of the correct type, a null pointer is returned. Thus, this works very
				365	much like the <tt>dynamic_cast<></tt> operator in C++, and should be
				366	used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
				367	operator is used in an <tt>if</tt> statement or some other flow control
				368	statement like this:</p>
				369
				370	<div class="doc_code">
				371	<pre>
				372	if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
				373	// <i>...</i>
				374	}
				375	</pre>
				376	</div>
				377
				378	<p>This form of the <tt>if</tt> statement effectively combines together a call
				379	to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
				380	statement, which is very convenient.</p>
				381
				382	<p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
				383	<tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
				384	abused. In particular, you should not use big chained <tt>if/then/else</tt>
				385	blocks to check for lots of different variants of classes. If you find
				386	yourself wanting to do this, it is much cleaner and more efficient to use the
				387	<tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
				388
				389	</dd>
				390
				391	<dt><tt>cast_or_null<></tt>: </dt>
				392
				393	<dd><p>The <tt>cast_or_null<></tt> operator works just like the
				394	<tt>cast<></tt> operator, except that it allows for a null pointer as an
				395	argument (which it then propagates). This can sometimes be useful, allowing
				396	you to combine several null checks into one.</p></dd>
				397
				398	<dt><tt>dyn_cast_or_null<></tt>: </dt>
				399
				400	<dd><p>The <tt>dyn_cast_or_null<></tt> operator works just like the
				401	<tt>dyn_cast<></tt> operator, except that it allows for a null pointer
				402	as an argument (which it then propagates). This can sometimes be useful,
				403	allowing you to combine several null checks into one.</p></dd>
				404
				405	</dl>
				406
				407	<p>These five templates can be used with any classes, whether they have a
				408	v-table or not. To add support for these templates, you simply need to add
				409	<tt>classof</tt> static methods to the class you are interested casting
				410	to. Describing this is currently outside the scope of this document, but there
				411	are lots of examples in the LLVM source base.</p>
				412
				413	</div>
				414
				415	<!-- ======================================================================= -->
				416	<div class="doc_subsection">
				417	<a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
				418	</div>
				419
				420	<div class="doc_text">
				421
				422	<p>Often when working on your pass you will put a bunch of debugging printouts
				423	and other code into your pass. After you get it working, you want to remove
				424	it, but you may need it again in the future (to work out new bugs that you run
				425	across).</p>
				426
				427	<p> Naturally, because of this, you don't want to delete the debug printouts,
				428	but you don't want them to always be noisy. A standard compromise is to comment
				429	them out, allowing you to enable them if you need them in the future.</p>
				430
				431	<p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
				432	file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
				433	this problem. Basically, you can put arbitrary code into the argument of the
				434	<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
				435	tool) is run with the '<tt>-debug</tt>' command line argument:</p>
				436
				437	<div class="doc_code">
				438	<pre>
				439	DOUT << "I am here!\n";
				440	</pre>
				441	</div>
				442
				443	<p>Then you can run your pass like this:</p>
				444
				445	<div class="doc_code">
				446	<pre>
				447	$ opt < a.bc > /dev/null -mypass
				448	<i><no output></i>
				449	$ opt < a.bc > /dev/null -mypass -debug
				450	I am here!
				451	</pre>
				452	</div>
				453
				454	<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
				455	to not have to create "yet another" command line option for the debug output for
				456	your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
				457	so they do not cause a performance impact at all (for the same reason, they
				458	should also not contain side-effects!).</p>
				459
				460	<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
				461	enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
				462	"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
				463	program hasn't been started yet, you can always just run it with
				464	<tt>-debug</tt>.</p>
				465
				466	</div>
				467
				468	<!-- _______________________________________________________________________ -->
				469	<div class="doc_subsubsection">
				470	<a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
				471	the <tt>-debug-only</tt> option</a>
				472	</div>
				473
				474	<div class="doc_text">
				475
				476	<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
				477	just turns on <b>too much</b> information (such as when working on the code
				478	generator). If you want to enable debug information with more fine-grained
				479	control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
				480	option as follows:</p>
				481
				482	<div class="doc_code">
				483	<pre>
				484	DOUT << "No debug type\n";
				485	#undef DEBUG_TYPE
				486	#define DEBUG_TYPE "foo"
				487	DOUT << "'foo' debug type\n";
				488	#undef DEBUG_TYPE
				489	#define DEBUG_TYPE "bar"
				490	DOUT << "'bar' debug type\n";
				491	#undef DEBUG_TYPE
				492	#define DEBUG_TYPE ""
				493	DOUT << "No debug type (2)\n";
				494	</pre>
				495	</div>
				496
				497	<p>Then you can run your pass like this:</p>
				498
				499	<div class="doc_code">
				500	<pre>
				501	$ opt < a.bc > /dev/null -mypass
				502	<i><no output></i>
				503	$ opt < a.bc > /dev/null -mypass -debug
				504	No debug type
				505	'foo' debug type
				506	'bar' debug type
				507	No debug type (2)
				508	$ opt < a.bc > /dev/null -mypass -debug-only=foo
				509	'foo' debug type
				510	$ opt < a.bc > /dev/null -mypass -debug-only=bar
				511	'bar' debug type
				512	</pre>
				513	</div>
				514
				515	<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
				516	a file, to specify the debug type for the entire module (if you do this before
				517	you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
				518	<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
				519	"bar", because there is no system in place to ensure that names do not
				520	conflict. If two different modules use the same string, they will all be turned
				521	on when the name is specified. This allows, for example, all debug information
				522	for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
				523	even if the source lives in multiple files.</p>
				524
				525	</div>
				526
				527	<!-- ======================================================================= -->
				528	<div class="doc_subsection">
				529	<a name="Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
				530	option</a>
				531	</div>
				532
				533	<div class="doc_text">
				534
				535	<p>The "<tt><a
				536	href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
				537	provides a class named <tt>Statistic</tt> that is used as a unified way to
				538	keep track of what the LLVM compiler is doing and how effective various
				539	optimizations are. It is useful to see what optimizations are contributing to
				540	making a particular program run faster.</p>
				541
				542	<p>Often you may run your pass on some big program, and you're interested to see
				543	how many times it makes a certain transformation. Although you can do this with
				544	hand inspection, or some ad-hoc method, this is a real pain and not very useful
				545	for big programs. Using the <tt>Statistic</tt> class makes it very easy to
				546	keep track of this information, and the calculated information is presented in a
				547	uniform manner with the rest of the passes being executed.</p>
				548
				549	<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
				550	it are as follows:</p>
				551
				552	<ol>
				553	<li><p>Define your statistic like this:</p>
				554
				555	<div class="doc_code">
				556	<pre>
				557	#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
				558	STATISTIC(NumXForms, "The # of times I did stuff");
				559	</pre>
				560	</div>
				561
				562	<p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
				563	specified by the first argument. The pass name is taken from the DEBUG_TYPE
				564	macro, and the description is taken from the second argument. The variable
				565	defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
				566
				567	<li><p>Whenever you make a transformation, bump the counter:</p>
				568
				569	<div class="doc_code">
				570	<pre>
				571	++NumXForms; // <i>I did stuff!</i>
				572	</pre>
				573	</div>
				574
				575	</li>
				576	</ol>
				577
				578	<p>That's all you have to do. To get '<tt>opt</tt>' to print out the
				579	statistics gathered, use the '<tt>-stats</tt>' option:</p>
				580
				581	<div class="doc_code">
				582	<pre>
				583	$ opt -stats -mypassname < program.bc > /dev/null
				584	<i>... statistics output ...</i>
				585	</pre>
				586	</div>
				587
				588	<p> When running <tt>opt</tt> on a C file from the SPEC benchmark
				589	suite, it gives a report that looks like this:</p>
				590
				591	<div class="doc_code">
				592	<pre>
				593	7646 bitcodewriter - Number of normal instructions
				594	725 bitcodewriter - Number of oversized instructions
				595	129996 bitcodewriter - Number of bitcode bytes written
				596	2817 raise - Number of insts DCEd or constprop'd
				597	3213 raise - Number of cast-of-self removed
				598	5046 raise - Number of expression trees converted
				599	75 raise - Number of other getelementptr's formed
				600	138 raise - Number of load/store peepholes
				601	42 deadtypeelim - Number of unused typenames removed from symtab
				602	392 funcresolve - Number of varargs functions resolved
				603	27 globaldce - Number of global variables removed
				604	2 adce - Number of basic blocks removed
				605	134 cee - Number of branches revectored
				606	49 cee - Number of setcc instruction eliminated
				607	532 gcse - Number of loads removed
				608	2919 gcse - Number of instructions removed
				609	86 indvars - Number of canonical indvars added
				610	87 indvars - Number of aux indvars removed
				611	25 instcombine - Number of dead inst eliminate
				612	434 instcombine - Number of insts combined
				613	248 licm - Number of load insts hoisted
				614	1298 licm - Number of insts hoisted to a loop pre-header
				615	3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
				616	75 mem2reg - Number of alloca's promoted
				617	1444 cfgsimplify - Number of blocks simplified
				618	</pre>
				619	</div>
				620
				621	<p>Obviously, with so many optimizations, having a unified framework for this
				622	stuff is very nice. Making your pass fit well into the framework makes it more
				623	maintainable and useful.</p>
				624
				625	</div>
				626
				627	<!-- ======================================================================= -->
				628	<div class="doc_subsection">
				629	<a name="ViewGraph">Viewing graphs while debugging code</a>
				630	</div>
				631
				632	<div class="doc_text">
				633
				634	<p>Several of the important data structures in LLVM are graphs: for example
				635	CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
				636	LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
				637	<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
				638	DAGs</a>. In many cases, while debugging various parts of the compiler, it is
				639	nice to instantly visualize these graphs.</p>
				640
				641	<p>LLVM provides several callbacks that are available in a debug build to do
				642	exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
				643	the current LLVM tool will pop up a window containing the CFG for the function
				644	where each basic block is a node in the graph, and each node contains the
				645	instructions in the block. Similarly, there also exists
				646	<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
				647	<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
				648	and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
				649	you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
				650	up a window. Alternatively, you can sprinkle calls to these functions in your
				651	code in places you want to debug.</p>
				652
				653	<p>Getting this to work requires a small amount of configuration. On Unix
				654	systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
				655	toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
				656	Mac OS/X, download and install the Mac OS/X <a
				657	href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
				658	<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
				659	it) to your path. Once in your system and path are set up, rerun the LLVM
				660	configure script and rebuild LLVM to enable this functionality.</p>
				661
				662	<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
				663	<i>interesting</i> nodes in large complex graphs. From gdb, if you
				664	<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
				665	next <tt>call DAG.viewGraph()</tt> would highlight the node in the
				666	specified color (choices of colors can be found at <a
				667	href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
				668	complex node attributes can be provided with <tt>call
				669	DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
				670	found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
				671	Attributes</a>.) If you want to restart and clear all the current graph
				672	attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
				673
				674	</div>
				675
				676	<!-- *********************************************************************** -->
				677	<div class="doc_section">
				678	<a name="datastructure">Picking the Right Data Structure for a Task</a>
				679	</div>
				680	<!-- *********************************************************************** -->
				681
				682	<div class="doc_text">
				683
				684	<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
				685	and we commonly use STL data structures. This section describes the trade-offs
				686	you should consider when you pick one.</p>
				687
				688	<p>
				689	The first step is a choose your own adventure: do you want a sequential
				690	container, a set-like container, or a map-like container? The most important
				691	thing when choosing a container is the algorithmic properties of how you plan to
				692	access the container. Based on that, you should use:</p>
				693
				694	<ul>
				695	<li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
				696	of an value based on another value. Map-like containers also support
				697	efficient queries for containment (whether a key is in the map). Map-like
				698	containers generally do not support efficient reverse mapping (values to
				699	keys). If you need that, use two maps. Some map-like containers also
				700	support efficient iteration through the keys in sorted order. Map-like
				701	containers are the most expensive sort, only use them if you need one of
				702	these capabilities.</li>
				703
				704	<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
				705	stuff into a container that automatically eliminates duplicates. Some
				706	set-like containers support efficient iteration through the elements in
				707	sorted order. Set-like containers are more expensive than sequential
				708	containers.
				709	</li>
				710
				711	<li>a <a href="#ds_sequential">sequential</a> container provides
				712	the most efficient way to add elements and keeps track of the order they are
				713	added to the collection. They permit duplicates and support efficient
				714	iteration, but do not support efficient look-up based on a key.
				715	</li>
				716
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	717	<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
				718	perform set operations on sets of numeric id's, while automatically
				719	eliminating duplicates. Bit containers require a maximum of 1 bit for each
				720	identifier you want to store.
				721	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	722	</ul>
				723
				724	<p>
				725	Once the proper category of container is determined, you can fine tune the
				726	memory use, constant factors, and cache behaviors of access by intelligently
				727	picking a member of the category. Note that constant factors and cache behavior
				728	can be a big deal. If you have a vector that usually only contains a few
				729	elements (but could contain many), for example, it's much better to use
				730	<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
				731	. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
				732	cost of adding the elements to the container. </p>
				733
				734	</div>
				735
				736	<!-- ======================================================================= -->
				737	<div class="doc_subsection">
				738	<a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
				739	</div>
				740
				741	<div class="doc_text">
				742	There are a variety of sequential containers available for you, based on your
				743	needs. Pick the first in this section that will do what you want.
				744	</div>
				745
				746	<!-- _______________________________________________________________________ -->
				747	<div class="doc_subsubsection">
				748	<a name="dss_fixedarrays">Fixed Size Arrays</a>
				749	</div>
				750
				751	<div class="doc_text">
				752	<p>Fixed size arrays are very simple and very fast. They are good if you know
				753	exactly how many elements you have, or you have a (low) upper bound on how many
				754	you have.</p>
				755	</div>
				756
				757	<!-- _______________________________________________________________________ -->
				758	<div class="doc_subsubsection">
				759	<a name="dss_heaparrays">Heap Allocated Arrays</a>
				760	</div>
				761
				762	<div class="doc_text">
				763	<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
				764	the number of elements is variable, if you know how many elements you will need
				765	before the array is allocated, and if the array is usually large (if not,
				766	consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
				767	allocated array is the cost of the new/delete (aka malloc/free). Also note that
				768	if you are allocating an array of a type with a constructor, the constructor and
				769	destructors will be run for every element in the array (re-sizable vectors only
				770	construct those elements actually used).</p>
				771	</div>
				772
				773	<!-- _______________________________________________________________________ -->
				774	<div class="doc_subsubsection">
				775	<a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
				776	</div>
				777
				778	<div class="doc_text">
				779	<p><tt>SmallVector<Type, N></tt> is a simple class that looks and smells
				780	just like <tt>vector<Type></tt>:
				781	it supports efficient iteration, lays out elements in memory order (so you can
				782	do pointer arithmetic between elements), supports efficient push_back/pop_back
				783	operations, supports efficient random access to its elements, etc.</p>
				784
				785	<p>The advantage of SmallVector is that it allocates space for
				786	some number of elements (N) <b>in the object itself</b>. Because of this, if
				787	the SmallVector is dynamically smaller than N, no malloc is performed. This can
				788	be a big win in cases where the malloc/free call is far more expensive than the
				789	code that fiddles around with the elements.</p>
				790
				791	<p>This is good for vectors that are "usually small" (e.g. the number of
				792	predecessors/successors of a block is usually less than 8). On the other hand,
				793	this makes the size of the SmallVector itself large, so you don't want to
				794	allocate lots of them (doing so will waste a lot of space). As such,
				795	SmallVectors are most useful when on the stack.</p>
				796
				797	<p>SmallVector also provides a nice portable and efficient replacement for
				798	<tt>alloca</tt>.</p>
				799
				800	</div>
				801
				802	<!-- _______________________________________________________________________ -->
				803	<div class="doc_subsubsection">
				804	<a name="dss_vector"><vector></a>
				805	</div>
				806
				807	<div class="doc_text">
				808	<p>
				809	std::vector is well loved and respected. It is useful when SmallVector isn't:
				810	when the size of the vector is often large (thus the small optimization will
				811	rarely be a benefit) or if you will be allocating many instances of the vector
				812	itself (which would waste space for elements that aren't in the container).
				813	vector is also useful when interfacing with code that expects vectors :).
				814	</p>
				815
				816	<p>One worthwhile note about std::vector: avoid code like this:</p>
				817
				818	<div class="doc_code">
				819	<pre>
				820	for ( ... ) {
				821	std::vector<foo> V;
				822	use V;
				823	}
				824	</pre>
				825	</div>
				826
				827	<p>Instead, write this as:</p>
				828
				829	<div class="doc_code">
				830	<pre>
				831	std::vector<foo> V;
				832	for ( ... ) {
				833	use V;
				834	V.clear();
				835	}
				836	</pre>
				837	</div>
				838
				839	<p>Doing so will save (at least) one heap allocation and free per iteration of
				840	the loop.</p>
				841
				842	</div>
				843
				844	<!-- _______________________________________________________________________ -->
				845	<div class="doc_subsubsection">
				846	<a name="dss_deque"><deque></a>
				847	</div>
				848
				849	<div class="doc_text">
				850	<p>std::deque is, in some senses, a generalized version of std::vector. Like
				851	std::vector, it provides constant time random access and other similar
				852	properties, but it also provides efficient access to the front of the list. It
				853	does not guarantee continuity of elements within memory.</p>
				854
				855	<p>In exchange for this extra flexibility, std::deque has significantly higher
				856	constant factor costs than std::vector. If possible, use std::vector or
				857	something cheaper.</p>
				858	</div>
				859
				860	<!-- _______________________________________________________________________ -->
				861	<div class="doc_subsubsection">
				862	<a name="dss_list"><list></a>
				863	</div>
				864
				865	<div class="doc_text">
				866	<p>std::list is an extremely inefficient class that is rarely useful.
				867	It performs a heap allocation for every element inserted into it, thus having an
				868	extremely high constant factor, particularly for small data types. std::list
				869	also only supports bidirectional iteration, not random access iteration.</p>
				870
				871	<p>In exchange for this high cost, std::list supports efficient access to both
				872	ends of the list (like std::deque, but unlike std::vector or SmallVector). In
				873	addition, the iterator invalidation characteristics of std::list are stronger
				874	than that of a vector class: inserting or removing an element into the list does
				875	not invalidate iterator or pointers to other elements in the list.</p>
				876	</div>
				877
				878	<!-- _______________________________________________________________________ -->
				879	<div class="doc_subsubsection">
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	880	<a name="dss_ilist">llvm/ADT/ilist.h</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	881	</div>
				882
				883	<div class="doc_text">
				884	<p><tt>ilist<T></tt> implements an 'intrusive' doubly-linked list. It is
				885	intrusive, because it requires the element to store and provide access to the
				886	prev/next pointers for the list.</p>
				887
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	888	<p><tt>ilist</tt> has the same drawbacks as <tt>std::list</tt>, and additionally
				889	requires an <tt>ilist_traits</tt> implementation for the element type, but it
				890	provides some novel characteristics. In particular, it can efficiently store
				891	polymorphic objects, the traits class is informed when an element is inserted or
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame^]	892	removed from the list, and <tt>ilist</tt>s are guaranteed to support a
				893	constant-time splice operation.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	894
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame^]	895	<p>These properties are exactly what we want for things like
				896	<tt>Instruction</tt>s and basic blocks, which is why these are implemented with
				897	<tt>ilist</tt>s.</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>
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	901	<li><a href="#dss_ilist_traits">ilist_traits</a></li>
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	902	<li><a href="#dss_iplist">iplist</a></li>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	903	<li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
				904	</ul>
				905	</div>
				906
				907	<!-- _______________________________________________________________________ -->
				908	<div class="doc_subsubsection">
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	909	<a name="dss_ilist_traits">ilist_traits</a>
				910	</div>
				911
				912	<div class="doc_text">
				913	<p><tt>ilist_traits<T></tt> is <tt>ilist<T></tt>'s customization
				914	mechanism. <tt>iplist<T></tt> (and consequently <tt>ilist<T></tt>)
				915	publicly derive from this traits class.</p>
				916	</div>
				917
				918	<!-- _______________________________________________________________________ -->
				919	<div class="doc_subsubsection">
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	920	<a name="dss_iplist">iplist</a>
				921	</div>
				922
				923	<div class="doc_text">
				924	<p><tt>iplist<T></tt> is <tt>ilist<T></tt>'s base and as such
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame^]	925	supports a slightly narrower interface. Notably, inserters from
				926	<tt>T&</tt> are absent.</p>
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	927
				928	<p><tt>ilist_traits<T></tt> is a public base of this class and can be
				929	used for a wide variety of customizations.</p>
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	930	</div>
				931
				932	<!-- _______________________________________________________________________ -->
				933	<div class="doc_subsubsection">
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	934	<a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
				935	</div>
				936
				937	<div class="doc_text">
				938	<p><tt>ilist_node<T></tt> implements a the forward and backward links
				939	that are expected by the <tt>ilist<T></tt> (and analogous containers)
				940	in the default manner.</p>
				941
				942	<p><tt>ilist_node<T></tt>s are meant to be embedded in the node type
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame^]	943	<tt>T</tt>, usually <tt>T</tt> publicly derives from
				944	<tt>ilist_node<T></tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	945	</div>
				946
				947	<!-- _______________________________________________________________________ -->
				948	<div class="doc_subsubsection">
				949	<a name="dss_other">Other Sequential Container options</a>
				950	</div>
				951
				952	<div class="doc_text">
				953	<p>Other STL containers are available, such as std::string.</p>
				954
				955	<p>There are also various STL adapter classes such as std::queue,
				956	std::priority_queue, std::stack, etc. These provide simplified access to an
				957	underlying container but don't affect the cost of the container itself.</p>
				958
				959	</div>
				960
				961
				962	<!-- ======================================================================= -->
				963	<div class="doc_subsection">
				964	<a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
				965	</div>
				966
				967	<div class="doc_text">
				968
				969	<p>Set-like containers are useful when you need to canonicalize multiple values
				970	into a single representation. There are several different choices for how to do
				971	this, providing various trade-offs.</p>
				972
				973	</div>
				974
				975
				976	<!-- _______________________________________________________________________ -->
				977	<div class="doc_subsubsection">
				978	<a name="dss_sortedvectorset">A sorted 'vector'</a>
				979	</div>
				980
				981	<div class="doc_text">
				982
				983	<p>If you intend to insert a lot of elements, then do a lot of queries, a
				984	great approach is to use a vector (or other sequential container) with
				985	std::sort+std::unique to remove duplicates. This approach works really well if
				986	your usage pattern has these two distinct phases (insert then query), and can be
				987	coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
				988	</p>
				989
				990	<p>
				991	This combination provides the several nice properties: the result data is
				992	contiguous in memory (good for cache locality), has few allocations, is easy to
				993	address (iterators in the final vector are just indices or pointers), and can be
				994	efficiently queried with a standard binary or radix search.</p>
				995
				996	</div>
				997
				998	<!-- _______________________________________________________________________ -->
				999	<div class="doc_subsubsection">
				1000	<a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
				1001	</div>
				1002
				1003	<div class="doc_text">
				1004
				1005	<p>If you have a set-like data structure that is usually small and whose elements
				1006	are reasonably small, a <tt>SmallSet<Type, N></tt> is a good choice. This set
				1007	has space for N elements in place (thus, if the set is dynamically smaller than
				1008	N, no malloc traffic is required) and accesses them with a simple linear search.
				1009	When the set grows beyond 'N' elements, it allocates a more expensive representation that
				1010	guarantees efficient access (for most types, it falls back to std::set, but for
				1011	pointers it uses something far better, <a
				1012	href="#dss_smallptrset">SmallPtrSet</a>).</p>
				1013
				1014	<p>The magic of this class is that it handles small sets extremely efficiently,
				1015	but gracefully handles extremely large sets without loss of efficiency. The
				1016	drawback is that the interface is quite small: it supports insertion, queries
				1017	and erasing, but does not support iteration.</p>
				1018
				1019	</div>
				1020
				1021	<!-- _______________________________________________________________________ -->
				1022	<div class="doc_subsubsection">
				1023	<a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
				1024	</div>
				1025
				1026	<div class="doc_text">
				1027
				1028	<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
				1029	transparently implemented with a SmallPtrSet), but also supports iterators. If
				1030	more than 'N' insertions are performed, a single quadratically
				1031	probed hash table is allocated and grows as needed, providing extremely
				1032	efficient access (constant time insertion/deleting/queries with low constant
				1033	factors) and is very stingy with malloc traffic.</p>
				1034
				1035	<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
				1036	whenever an insertion occurs. Also, the values visited by the iterators are not
				1037	visited in sorted order.</p>
				1038
				1039	</div>
				1040
				1041	<!-- _______________________________________________________________________ -->
				1042	<div class="doc_subsubsection">
Chris Lattner	77ab46d	2007-09-30 00:58:59 +0000	[diff] [blame]	1043	<a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
				1044	</div>
				1045
				1046	<div class="doc_text">
				1047
				1048	<p>
				1049	DenseSet is a simple quadratically probed hash table. It excels at supporting
				1050	small values: it uses a single allocation to hold all of the pairs that
				1051	are currently inserted in the set. DenseSet is a great way to unique small
				1052	values that are not simple pointers (use <a
				1053	href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
				1054	the same requirements for the value type that <a
				1055	href="#dss_densemap">DenseMap</a> has.
				1056	</p>
				1057
				1058	</div>
				1059
				1060	<!-- _______________________________________________________________________ -->
				1061	<div class="doc_subsubsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1062	<a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
				1063	</div>
				1064
				1065	<div class="doc_text">
				1066
				1067	<p>
				1068	FoldingSet is an aggregate class that is really good at uniquing
				1069	expensive-to-create or polymorphic objects. It is a combination of a chained
				1070	hash table with intrusive links (uniqued objects are required to inherit from
				1071	FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
				1072	its ID process.</p>
				1073
				1074	<p>Consider a case where you want to implement a "getOrCreateFoo" method for
				1075	a complex object (for example, a node in the code generator). The client has a
				1076	description of what it wants to generate (it knows the opcode and all the
				1077	operands), but we don't want to 'new' a node, then try inserting it into a set
				1078	only to find out it already exists, at which point we would have to delete it
				1079	and return the node that already exists.
				1080	</p>
				1081
				1082	<p>To support this style of client, FoldingSet perform a query with a
				1083	FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
				1084	element that we want to query for. The query either returns the element
				1085	matching the ID or it returns an opaque ID that indicates where insertion should
				1086	take place. Construction of the ID usually does not require heap traffic.</p>
				1087
				1088	<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
				1089	in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
				1090	Because the elements are individually allocated, pointers to the elements are
				1091	stable: inserting or removing elements does not invalidate any pointers to other
				1092	elements.
				1093	</p>
				1094
				1095	</div>
				1096
				1097	<!-- _______________________________________________________________________ -->
				1098	<div class="doc_subsubsection">
				1099	<a name="dss_set"><set></a>
				1100	</div>
				1101
				1102	<div class="doc_text">
				1103
				1104	<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
				1105	many things but great at nothing. std::set allocates memory for each element
				1106	inserted (thus it is very malloc intensive) and typically stores three pointers
				1107	per element in the set (thus adding a large amount of per-element space
				1108	overhead). It offers guaranteed log(n) performance, which is not particularly
				1109	fast from a complexity standpoint (particularly if the elements of the set are
				1110	expensive to compare, like strings), and has extremely high constant factors for
				1111	lookup, insertion and removal.</p>
				1112
				1113	<p>The advantages of std::set are that its iterators are stable (deleting or
				1114	inserting an element from the set does not affect iterators or pointers to other
				1115	elements) and that iteration over the set is guaranteed to be in sorted order.
				1116	If the elements in the set are large, then the relative overhead of the pointers
				1117	and malloc traffic is not a big deal, but if the elements of the set are small,
				1118	std::set is almost never a good choice.</p>
				1119
				1120	</div>
				1121
				1122	<!-- _______________________________________________________________________ -->
				1123	<div class="doc_subsubsection">
				1124	<a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
				1125	</div>
				1126
				1127	<div class="doc_text">
				1128	<p>LLVM's SetVector<Type> is an adapter class that combines your choice of
				1129	a set-like container along with a <a href="#ds_sequential">Sequential
				1130	Container</a>. The important property
				1131	that this provides is efficient insertion with uniquing (duplicate elements are
				1132	ignored) with iteration support. It implements this by inserting elements into
				1133	both a set-like container and the sequential container, using the set-like
				1134	container for uniquing and the sequential container for iteration.
				1135	</p>
				1136
				1137	<p>The difference between SetVector and other sets is that the order of
				1138	iteration is guaranteed to match the order of insertion into the SetVector.
				1139	This property is really important for things like sets of pointers. Because
				1140	pointer values are non-deterministic (e.g. vary across runs of the program on
				1141	different machines), iterating over the pointers in the set will
				1142	not be in a well-defined order.</p>
				1143
				1144	<p>
				1145	The drawback of SetVector is that it requires twice as much space as a normal
				1146	set and has the sum of constant factors from the set-like container and the
				1147	sequential container that it uses. Use it only if you need to iterate over
				1148	the elements in a deterministic order. SetVector is also expensive to delete
				1149	elements out of (linear time), unless you use it's "pop_back" method, which is
				1150	faster.
				1151	</p>
				1152
				1153	<p>SetVector is an adapter class that defaults to using std::vector and std::set
				1154	for the underlying containers, so it is quite expensive. However,
				1155	<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
				1156	defaults to using a SmallVector and SmallSet of a specified size. If you use
				1157	this, and if your sets are dynamically smaller than N, you will save a lot of
				1158	heap traffic.</p>
				1159
				1160	</div>
				1161
				1162	<!-- _______________________________________________________________________ -->
				1163	<div class="doc_subsubsection">
				1164	<a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
				1165	</div>
				1166
				1167	<div class="doc_text">
				1168
				1169	<p>
				1170	UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
				1171	retains a unique ID for each element inserted into the set. It internally
				1172	contains a map and a vector, and it assigns a unique ID for each value inserted
				1173	into the set.</p>
				1174
				1175	<p>UniqueVector is very expensive: its cost is the sum of the cost of
				1176	maintaining both the map and vector, it has high complexity, high constant
				1177	factors, and produces a lot of malloc traffic. It should be avoided.</p>
				1178
				1179	</div>
				1180
				1181
				1182	<!-- _______________________________________________________________________ -->
				1183	<div class="doc_subsubsection">
				1184	<a name="dss_otherset">Other Set-Like Container Options</a>
				1185	</div>
				1186
				1187	<div class="doc_text">
				1188
				1189	<p>
				1190	The STL provides several other options, such as std::multiset and the various
Chris Lattner	86a63d0	2009-03-09 05:20:45 +0000	[diff] [blame]	1191	"hash_set" like containers (whether from C++ TR1 or from the SGI library). We
				1192	never use hash_set and unordered_set because they are generally very expensive
				1193	(each insertion requires a malloc) and very non-portable.
				1194	</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1195
				1196	<p>std::multiset is useful if you're not interested in elimination of
				1197	duplicates, but has all the drawbacks of std::set. A sorted vector (where you
				1198	don't delete duplicate entries) or some other approach is almost always
				1199	better.</p>
				1200
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1201	</div>
				1202
				1203	<!-- ======================================================================= -->
				1204	<div class="doc_subsection">
				1205	<a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
				1206	</div>
				1207
				1208	<div class="doc_text">
				1209	Map-like containers are useful when you want to associate data to a key. As
				1210	usual, there are a lot of different ways to do this. :)
				1211	</div>
				1212
				1213	<!-- _______________________________________________________________________ -->
				1214	<div class="doc_subsubsection">
				1215	<a name="dss_sortedvectormap">A sorted 'vector'</a>
				1216	</div>
				1217
				1218	<div class="doc_text">
				1219
				1220	<p>
				1221	If your usage pattern follows a strict insert-then-query approach, you can
				1222	trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
				1223	for set-like containers</a>. The only difference is that your query function
				1224	(which uses std::lower_bound to get efficient log(n) lookup) should only compare
				1225	the key, not both the key and value. This yields the same advantages as sorted
				1226	vectors for sets.
				1227	</p>
				1228	</div>
				1229
				1230	<!-- _______________________________________________________________________ -->
				1231	<div class="doc_subsubsection">
				1232	<a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
				1233	</div>
				1234
				1235	<div class="doc_text">
				1236
				1237	<p>
				1238	Strings are commonly used as keys in maps, and they are difficult to support
				1239	efficiently: they are variable length, inefficient to hash and compare when
				1240	long, expensive to copy, etc. StringMap is a specialized container designed to
				1241	cope with these issues. It supports mapping an arbitrary range of bytes to an
				1242	arbitrary other object.</p>
				1243
				1244	<p>The StringMap implementation uses a quadratically-probed hash table, where
				1245	the buckets store a pointer to the heap allocated entries (and some other
				1246	stuff). The entries in the map must be heap allocated because the strings are
				1247	variable length. The string data (key) and the element object (value) are
				1248	stored in the same allocation with the string data immediately after the element
				1249	object. This container guarantees the "<tt>(char*)(&Value+1)</tt>" points
				1250	to the key string for a value.</p>
				1251
				1252	<p>The StringMap is very fast for several reasons: quadratic probing is very
				1253	cache efficient for lookups, the hash value of strings in buckets is not
				1254	recomputed when lookup up an element, StringMap rarely has to touch the
				1255	memory for unrelated objects when looking up a value (even when hash collisions
				1256	happen), hash table growth does not recompute the hash values for strings
				1257	already in the table, and each pair in the map is store in a single allocation
				1258	(the string data is stored in the same allocation as the Value of a pair).</p>
				1259
				1260	<p>StringMap also provides query methods that take byte ranges, so it only ever
				1261	copies a string if a value is inserted into the table.</p>
				1262	</div>
				1263
				1264	<!-- _______________________________________________________________________ -->
				1265	<div class="doc_subsubsection">
				1266	<a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
				1267	</div>
				1268
				1269	<div class="doc_text">
				1270	<p>
				1271	IndexedMap is a specialized container for mapping small dense integers (or
				1272	values that can be mapped to small dense integers) to some other type. It is
				1273	internally implemented as a vector with a mapping function that maps the keys to
				1274	the dense integer range.
				1275	</p>
				1276
				1277	<p>
				1278	This is useful for cases like virtual registers in the LLVM code generator: they
				1279	have a dense mapping that is offset by a compile-time constant (the first
				1280	virtual register ID).</p>
				1281
				1282	</div>
				1283
				1284	<!-- _______________________________________________________________________ -->
				1285	<div class="doc_subsubsection">
				1286	<a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
				1287	</div>
				1288
				1289	<div class="doc_text">
				1290
				1291	<p>
				1292	DenseMap is a simple quadratically probed hash table. It excels at supporting
				1293	small keys and values: it uses a single allocation to hold all of the pairs that
				1294	are currently inserted in the map. DenseMap is a great way to map pointers to
				1295	pointers, or map other small types to each other.
				1296	</p>
				1297
				1298	<p>
				1299	There are several aspects of DenseMap that you should be aware of, however. The
				1300	iterators in a densemap are invalidated whenever an insertion occurs, unlike
				1301	map. Also, because DenseMap allocates space for a large number of key/value
				1302	pairs (it starts with 64 by default), it will waste a lot of space if your keys
				1303	or values are large. Finally, you must implement a partial specialization of
Chris Lattner	92eea07	2007-09-17 18:34:04 +0000	[diff] [blame]	1304	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]	1305	is required to tell DenseMap about two special marker values (which can never be
				1306	inserted into the map) that it needs internally.</p>
				1307
				1308	</div>
				1309
				1310	<!-- _______________________________________________________________________ -->
				1311	<div class="doc_subsubsection">
				1312	<a name="dss_map"><map></a>
				1313	</div>
				1314
				1315	<div class="doc_text">
				1316
				1317	<p>
				1318	std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
				1319	a single allocation per pair inserted into the map, it offers log(n) lookup with
				1320	an extremely large constant factor, imposes a space penalty of 3 pointers per
				1321	pair in the map, etc.</p>
				1322
				1323	<p>std::map is most useful when your keys or values are very large, if you need
				1324	to iterate over the collection in sorted order, or if you need stable iterators
				1325	into the map (i.e. they don't get invalidated if an insertion or deletion of
				1326	another element takes place).</p>
				1327
				1328	</div>
				1329
				1330	<!-- _______________________________________________________________________ -->
				1331	<div class="doc_subsubsection">
				1332	<a name="dss_othermap">Other Map-Like Container Options</a>
				1333	</div>
				1334
				1335	<div class="doc_text">
				1336
				1337	<p>
				1338	The STL provides several other options, such as std::multimap and the various
Chris Lattner	86a63d0	2009-03-09 05:20:45 +0000	[diff] [blame]	1339	"hash_map" like containers (whether from C++ TR1 or from the SGI library). We
				1340	never use hash_set and unordered_set because they are generally very expensive
				1341	(each insertion requires a malloc) and very non-portable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1342
				1343	<p>std::multimap is useful if you want to map a key to multiple values, but has
				1344	all the drawbacks of std::map. A sorted vector or some other approach is almost
				1345	always better.</p>
				1346
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1347	</div>
				1348
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1349	<!-- ======================================================================= -->
				1350	<div class="doc_subsection">
				1351	<a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
				1352	</div>
				1353
				1354	<div class="doc_text">
Chris Lattner	62a4eae	2007-09-25 22:37:50 +0000	[diff] [blame]	1355	<p>Unlike the other containers, there are only two bit storage containers, and
				1356	choosing when to use each is relatively straightforward.</p>
				1357
				1358	<p>One additional option is
				1359	<tt>std::vector<bool></tt>: we discourage its use for two reasons 1) the
				1360	implementation in many common compilers (e.g. commonly available versions of
				1361	GCC) is extremely inefficient and 2) the C++ standards committee is likely to
				1362	deprecate this container and/or change it significantly somehow. In any case,
				1363	please don't use it.</p>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1364	</div>
				1365
				1366	<!-- _______________________________________________________________________ -->
				1367	<div class="doc_subsubsection">
				1368	<a name="dss_bitvector">BitVector</a>
				1369	</div>
				1370
				1371	<div class="doc_text">
				1372	<p> The BitVector container provides a fixed size set of bits for manipulation.
				1373	It supports individual bit setting/testing, as well as set operations. The set
				1374	operations take time O(size of bitvector), but operations are performed one word
				1375	at a time, instead of one bit at a time. This makes the BitVector very fast for
				1376	set operations compared to other containers. Use the BitVector when you expect
				1377	the number of set bits to be high (IE a dense set).
				1378	</p>
				1379	</div>
				1380
				1381	<!-- _______________________________________________________________________ -->
				1382	<div class="doc_subsubsection">
				1383	<a name="dss_sparsebitvector">SparseBitVector</a>
				1384	</div>
				1385
				1386	<div class="doc_text">
				1387	<p> The SparseBitVector container is much like BitVector, with one major
				1388	difference: Only the bits that are set, are stored. This makes the
				1389	SparseBitVector much more space efficient than BitVector when the set is sparse,
				1390	as well as making set operations O(number of set bits) instead of O(size of
				1391	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
				1392	(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).
				1393	</p>
				1394	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1395
				1396	<!-- *********************************************************************** -->
				1397	<div class="doc_section">
				1398	<a name="common">Helpful Hints for Common Operations</a>
				1399	</div>
				1400	<!-- *********************************************************************** -->
				1401
				1402	<div class="doc_text">
				1403
				1404	<p>This section describes how to perform some very simple transformations of
				1405	LLVM code. This is meant to give examples of common idioms used, showing the
				1406	practical side of LLVM transformations. <p> Because this is a "how-to" section,
				1407	you should also read about the main classes that you will be working with. The
				1408	<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
				1409	and descriptions of the main classes that you should know about.</p>
				1410
				1411	</div>
				1412
				1413	<!-- NOTE: this section should be heavy on example code -->
				1414	<!-- ======================================================================= -->
				1415	<div class="doc_subsection">
				1416	<a name="inspection">Basic Inspection and Traversal Routines</a>
				1417	</div>
				1418
				1419	<div class="doc_text">
				1420
				1421	<p>The LLVM compiler infrastructure have many different data structures that may
				1422	be traversed. Following the example of the C++ standard template library, the
				1423	techniques used to traverse these various data structures are all basically the
				1424	same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
				1425	method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
				1426	function returns an iterator pointing to one past the last valid element of the
				1427	sequence, and there is some <tt>XXXiterator</tt> data type that is common
				1428	between the two operations.</p>
				1429
				1430	<p>Because the pattern for iteration is common across many different aspects of
				1431	the program representation, the standard template library algorithms may be used
				1432	on them, and it is easier to remember how to iterate. First we show a few common
				1433	examples of the data structures that need to be traversed. Other data
				1434	structures are traversed in very similar ways.</p>
				1435
				1436	</div>
				1437
				1438	<!-- _______________________________________________________________________ -->
				1439	<div class="doc_subsubsection">
				1440	<a name="iterate_function">Iterating over the </a><a
				1441	href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
				1442	href="#Function"><tt>Function</tt></a>
				1443	</div>
				1444
				1445	<div class="doc_text">
				1446
				1447	<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
				1448	transform in some way; in particular, you'd like to manipulate its
				1449	<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
				1450	the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
				1451	an example that prints the name of a <tt>BasicBlock</tt> and the number of
				1452	<tt>Instruction</tt>s it contains:</p>
				1453
				1454	<div class="doc_code">
				1455	<pre>
				1456	// <i>func is a pointer to a Function instance</i>
				1457	for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i)
				1458	// <i>Print out the name of the basic block if it has one, and then the</i>
				1459	// <i>number of instructions that it contains</i>
				1460	llvm::cerr << "Basic block (name=" << i->getName() << ") has "
				1461	<< i->size() << " instructions.\n";
				1462	</pre>
				1463	</div>
				1464
				1465	<p>Note that i can be used as if it were a pointer for the purposes of
				1466	invoking member functions of the <tt>Instruction</tt> class. This is
				1467	because the indirection operator is overloaded for the iterator
				1468	classes. In the above code, the expression <tt>i->size()</tt> is
				1469	exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
				1470
				1471	</div>
				1472
				1473	<!-- _______________________________________________________________________ -->
				1474	<div class="doc_subsubsection">
				1475	<a name="iterate_basicblock">Iterating over the </a><a
				1476	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1477	href="#BasicBlock"><tt>BasicBlock</tt></a>
				1478	</div>
				1479
				1480	<div class="doc_text">
				1481
				1482	<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
				1483	easy to iterate over the individual instructions that make up
				1484	<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
				1485	a <tt>BasicBlock</tt>:</p>
				1486
				1487	<div class="doc_code">
				1488	<pre>
				1489	// <i>blk is a pointer to a BasicBlock instance</i>
				1490	for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
				1491	// <i>The next statement works since operator<<(ostream&,...)</i>
				1492	// <i>is overloaded for Instruction&</i>
				1493	llvm::cerr << *i << "\n";
				1494	</pre>
				1495	</div>
				1496
				1497	<p>However, this isn't really the best way to print out the contents of a
				1498	<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
				1499	anything you'll care about, you could have just invoked the print routine on the
				1500	basic block itself: <tt>llvm::cerr << *blk << "\n";</tt>.</p>
				1501
				1502	</div>
				1503
				1504	<!-- _______________________________________________________________________ -->
				1505	<div class="doc_subsubsection">
				1506	<a name="iterate_institer">Iterating over the </a><a
				1507	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1508	href="#Function"><tt>Function</tt></a>
				1509	</div>
				1510
				1511	<div class="doc_text">
				1512
				1513	<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
				1514	<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
				1515	<tt>InstIterator</tt> should be used instead. You'll need to include <a
				1516	href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
				1517	and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
				1518	small example that shows how to dump all instructions in a function to the standard error stream:<p>
				1519
				1520	<div class="doc_code">
				1521	<pre>
				1522	#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
				1523
				1524	// <i>F is a pointer to a Function instance</i>
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1525	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
				1526	llvm::cerr << *I << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1527	</pre>
				1528	</div>
				1529
				1530	<p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
				1531	work list with its initial contents. For example, if you wanted to
				1532	initialize a work list to contain all instructions in a <tt>Function</tt>
				1533	F, all you would need to do is something like:</p>
				1534
				1535	<div class="doc_code">
				1536	<pre>
				1537	std::set<Instruction*> worklist;
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1538	// or better yet, SmallPtrSet<Instruction*, 64> worklist;
				1539
				1540	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
				1541	worklist.insert(&*I);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1542	</pre>
				1543	</div>
				1544
				1545	<p>The STL set <tt>worklist</tt> would now contain all instructions in the
				1546	<tt>Function</tt> pointed to by F.</p>
				1547
				1548	</div>
				1549
				1550	<!-- _______________________________________________________________________ -->
				1551	<div class="doc_subsubsection">
				1552	<a name="iterate_convert">Turning an iterator into a class pointer (and
				1553	vice-versa)</a>
				1554	</div>
				1555
				1556	<div class="doc_text">
				1557
				1558	<p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
				1559	instance when all you've got at hand is an iterator. Well, extracting
				1560	a reference or a pointer from an iterator is very straight-forward.
				1561	Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
				1562	is a <tt>BasicBlock::const_iterator</tt>:</p>
				1563
				1564	<div class="doc_code">
				1565	<pre>
				1566	Instruction& inst = *i; // <i>Grab reference to instruction reference</i>
				1567	Instruction* pinst = &*i; // <i>Grab pointer to instruction reference</i>
				1568	const Instruction& inst = *j;
				1569	</pre>
				1570	</div>
				1571
				1572	<p>However, the iterators you'll be working with in the LLVM framework are
				1573	special: they will automatically convert to a ptr-to-instance type whenever they
				1574	need to. Instead of dereferencing the iterator and then taking the address of
				1575	the result, you can simply assign the iterator to the proper pointer type and
				1576	you get the dereference and address-of operation as a result of the assignment
				1577	(behind the scenes, this is a result of overloading casting mechanisms). Thus
				1578	the last line of the last example,</p>
				1579
				1580	<div class="doc_code">
				1581	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1582	Instruction pinst = &i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1583	</pre>
				1584	</div>
				1585
				1586	<p>is semantically equivalent to</p>
				1587
				1588	<div class="doc_code">
				1589	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1590	Instruction *pinst = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1591	</pre>
				1592	</div>
				1593
				1594	<p>It's also possible to turn a class pointer into the corresponding iterator,
				1595	and this is a constant time operation (very efficient). The following code
				1596	snippet illustrates use of the conversion constructors provided by LLVM
				1597	iterators. By using these, you can explicitly grab the iterator of something
				1598	without actually obtaining it via iteration over some structure:</p>
				1599
				1600	<div class="doc_code">
				1601	<pre>
				1602	void printNextInstruction(Instruction* inst) {
				1603	BasicBlock::iterator it(inst);
				1604	++it; // <i>After this line, it refers to the instruction after *inst</i>
				1605	if (it != inst->getParent()->end()) llvm::cerr << *it << "\n";
				1606	}
				1607	</pre>
				1608	</div>
				1609
				1610	</div>
				1611
				1612	<!--_______________________________________________________________________-->
				1613	<div class="doc_subsubsection">
				1614	<a name="iterate_complex">Finding call sites: a slightly more complex
				1615	example</a>
				1616	</div>
				1617
				1618	<div class="doc_text">
				1619
				1620	<p>Say that you're writing a FunctionPass and would like to count all the
				1621	locations in the entire module (that is, across every <tt>Function</tt>) where a
				1622	certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
				1623	learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
				1624	much more straight-forward manner, but this example will allow us to explore how
				1625	you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
				1626	is what we want to do:</p>
				1627
				1628	<div class="doc_code">
				1629	<pre>
				1630	initialize callCounter to zero
				1631	for each Function f in the Module
				1632	for each BasicBlock b in f
				1633	for each Instruction i in b
				1634	if (i is a CallInst and calls the given function)
				1635	increment callCounter
				1636	</pre>
				1637	</div>
				1638
				1639	<p>And the actual code is (remember, because we're writing a
				1640	<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
				1641	override the <tt>runOnFunction</tt> method):</p>
				1642
				1643	<div class="doc_code">
				1644	<pre>
				1645	Function* targetFunc = ...;
				1646
				1647	class OurFunctionPass : public FunctionPass {
				1648	public:
				1649	OurFunctionPass(): callCounter(0) { }
				1650
				1651	virtual runOnFunction(Function& F) {
				1652	for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
Eric Christopher	5fbecf7	2008-11-08 08:20:49 +0000	[diff] [blame]	1653	for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) {
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1654	if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
				1655	href="#CallInst">CallInst</a>>(&*i)) {
				1656	// <i>We know we've encountered a call instruction, so we</i>
				1657	// <i>need to determine if it's a call to the</i>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1658	// <i>function pointed to by m_func or not.</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1659	if (callInst->getCalledFunction() == targetFunc)
				1660	++callCounter;
				1661	}
				1662	}
				1663	}
				1664	}
				1665
				1666	private:
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1667	unsigned callCounter;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1668	};
				1669	</pre>
				1670	</div>
				1671
				1672	</div>
				1673
				1674	<!--_______________________________________________________________________-->
				1675	<div class="doc_subsubsection">
				1676	<a name="calls_and_invokes">Treating calls and invokes the same way</a>
				1677	</div>
				1678
				1679	<div class="doc_text">
				1680
				1681	<p>You may have noticed that the previous example was a bit oversimplified in
				1682	that it did not deal with call sites generated by 'invoke' instructions. In
				1683	this, and in other situations, you may find that you want to treat
				1684	<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
				1685	most-specific common base class is <tt>Instruction</tt>, which includes lots of
				1686	less closely-related things. For these cases, LLVM provides a handy wrapper
				1687	class called <a
				1688	href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
				1689	It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
				1690	methods that provide functionality common to <tt>CallInst</tt>s and
				1691	<tt>InvokeInst</tt>s.</p>
				1692
				1693	<p>This class has "value semantics": it should be passed by value, not by
				1694	reference and it should not be dynamically allocated or deallocated using
				1695	<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
				1696	assignable and constructable, with costs equivalents to that of a bare pointer.
				1697	If you look at its definition, it has only a single pointer member.</p>
				1698
				1699	</div>
				1700
				1701	<!--_______________________________________________________________________-->
				1702	<div class="doc_subsubsection">
				1703	<a name="iterate_chains">Iterating over def-use & use-def chains</a>
				1704	</div>
				1705
				1706	<div class="doc_text">
				1707
				1708	<p>Frequently, we might have an instance of the <a
				1709	href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
				1710	determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
				1711	<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
				1712	For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
				1713	particular function <tt>foo</tt>. Finding all of the instructions that
				1714	<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
				1715	of <tt>F</tt>:</p>
				1716
				1717	<div class="doc_code">
				1718	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1719	Function *F = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1720
				1721	for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i)
				1722	if (Instruction Inst = dyn_cast<Instruction>(i)) {
				1723	llvm::cerr << "F is used in instruction:\n";
				1724	llvm::cerr << *Inst << "\n";
				1725	}
				1726	</pre>
				1727	</div>
				1728
				1729	<p>Alternately, it's common to have an instance of the <a
				1730	href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
				1731	<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
				1732	<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
				1733	<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
				1734	all of the values that a particular instruction uses (that is, the operands of
				1735	the particular <tt>Instruction</tt>):</p>
				1736
				1737	<div class="doc_code">
				1738	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1739	Instruction *pi = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1740
				1741	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]	1742	Value v = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1743	// <i>...</i>
				1744	}
				1745	</pre>
				1746	</div>
				1747
				1748	<!--
				1749	def-use chains ("finding all users of"): Value::use_begin/use_end
				1750	use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
				1751	-->
				1752
				1753	</div>
				1754
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1755	<!--_______________________________________________________________________-->
				1756	<div class="doc_subsubsection">
				1757	<a name="iterate_preds">Iterating over predecessors &
				1758	successors of blocks</a>
				1759	</div>
				1760
				1761	<div class="doc_text">
				1762
				1763	<p>Iterating over the predecessors and successors of a block is quite easy
				1764	with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
				1765	this to iterate over all predecessors of BB:</p>
				1766
				1767	<div class="doc_code">
				1768	<pre>
				1769	#include "llvm/Support/CFG.h"
				1770	BasicBlock *BB = ...;
				1771
				1772	for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
				1773	BasicBlock Pred = PI;
				1774	// <i>...</i>
				1775	}
				1776	</pre>
				1777	</div>
				1778
				1779	<p>Similarly, to iterate over successors use
				1780	succ_iterator/succ_begin/succ_end.</p>
				1781
				1782	</div>
				1783
				1784
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1785	<!-- ======================================================================= -->
				1786	<div class="doc_subsection">
				1787	<a name="simplechanges">Making simple changes</a>
				1788	</div>
				1789
				1790	<div class="doc_text">
				1791
				1792	<p>There are some primitive transformation operations present in the LLVM
				1793	infrastructure that are worth knowing about. When performing
				1794	transformations, it's fairly common to manipulate the contents of basic
				1795	blocks. This section describes some of the common methods for doing so
				1796	and gives example code.</p>
				1797
				1798	</div>
				1799
				1800	<!--_______________________________________________________________________-->
				1801	<div class="doc_subsubsection">
				1802	<a name="schanges_creating">Creating and inserting new
				1803	<tt>Instruction</tt>s</a>
				1804	</div>
				1805
				1806	<div class="doc_text">
				1807
				1808	<p><i>Instantiating Instructions</i></p>
				1809
				1810	<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
				1811	constructor for the kind of instruction to instantiate and provide the necessary
				1812	parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
				1813	(const-ptr-to) <tt>Type</tt>. Thus:</p>
				1814
				1815	<div class="doc_code">
				1816	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	1817	AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1818	</pre>
				1819	</div>
				1820
				1821	<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
				1822	one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
				1823	subclass is likely to have varying default parameters which change the semantics
				1824	of the instruction, so refer to the <a
				1825	href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
				1826	Instruction</a> that you're interested in instantiating.</p>
				1827
				1828	<p><i>Naming values</i></p>
				1829
				1830	<p>It is very useful to name the values of instructions when you're able to, as
				1831	this facilitates the debugging of your transformations. If you end up looking
				1832	at generated LLVM machine code, you definitely want to have logical names
				1833	associated with the results of instructions! By supplying a value for the
				1834	<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
				1835	associate a logical name with the result of the instruction's execution at
				1836	run time. For example, say that I'm writing a transformation that dynamically
				1837	allocates space for an integer on the stack, and that integer is going to be
				1838	used as some kind of index by some other code. To accomplish this, I place an
				1839	<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
				1840	<tt>Function</tt>, and I'm intending to use it within the same
				1841	<tt>Function</tt>. I might do:</p>
				1842
				1843	<div class="doc_code">
				1844	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	1845	AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1846	</pre>
				1847	</div>
				1848
				1849	<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
				1850	execution value, which is a pointer to an integer on the run time stack.</p>
				1851
				1852	<p><i>Inserting instructions</i></p>
				1853
				1854	<p>There are essentially two ways to insert an <tt>Instruction</tt>
				1855	into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
				1856
				1857	<ul>
				1858	<li>Insertion into an explicit instruction list
				1859
				1860	<p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
				1861	<tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
				1862	before <tt>*pi</tt>, we do the following: </p>
				1863
				1864	<div class="doc_code">
				1865	<pre>
				1866	BasicBlock *pb = ...;
				1867	Instruction *pi = ...;
				1868	Instruction *newInst = new Instruction(...);
				1869
				1870	pb->getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
				1871	</pre>
				1872	</div>
				1873
				1874	<p>Appending to the end of a <tt>BasicBlock</tt> is so common that
				1875	the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
				1876	classes provide constructors which take a pointer to a
				1877	<tt>BasicBlock</tt> to be appended to. For example code that
				1878	looked like: </p>
				1879
				1880	<div class="doc_code">
				1881	<pre>
				1882	BasicBlock *pb = ...;
				1883	Instruction *newInst = new Instruction(...);
				1884
				1885	pb->getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
				1886	</pre>
				1887	</div>
				1888
				1889	<p>becomes: </p>
				1890
				1891	<div class="doc_code">
				1892	<pre>
				1893	BasicBlock *pb = ...;
				1894	Instruction *newInst = new Instruction(..., pb);
				1895	</pre>
				1896	</div>
				1897
				1898	<p>which is much cleaner, especially if you are creating
				1899	long instruction streams.</p></li>
				1900
				1901	<li>Insertion into an implicit instruction list
				1902
				1903	<p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
				1904	are implicitly associated with an existing instruction list: the instruction
				1905	list of the enclosing basic block. Thus, we could have accomplished the same
				1906	thing as the above code without being given a <tt>BasicBlock</tt> by doing:
				1907	</p>
				1908
				1909	<div class="doc_code">
				1910	<pre>
				1911	Instruction *pi = ...;
				1912	Instruction *newInst = new Instruction(...);
				1913
				1914	pi->getParent()->getInstList().insert(pi, newInst);
				1915	</pre>
				1916	</div>
				1917
				1918	<p>In fact, this sequence of steps occurs so frequently that the
				1919	<tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
				1920	constructors which take (as a default parameter) a pointer to an
				1921	<tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
				1922	precede. That is, <tt>Instruction</tt> constructors are capable of
				1923	inserting the newly-created instance into the <tt>BasicBlock</tt> of a
				1924	provided instruction, immediately before that instruction. Using an
				1925	<tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
				1926	parameter, the above code becomes:</p>
				1927
				1928	<div class="doc_code">
				1929	<pre>
				1930	Instruction* pi = ...;
				1931	Instruction* newInst = new Instruction(..., pi);
				1932	</pre>
				1933	</div>
				1934
				1935	<p>which is much cleaner, especially if you're creating a lot of
				1936	instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
				1937	</ul>
				1938
				1939	</div>
				1940
				1941	<!--_______________________________________________________________________-->
				1942	<div class="doc_subsubsection">
				1943	<a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
				1944	</div>
				1945
				1946	<div class="doc_text">
				1947
				1948	<p>Deleting an instruction from an existing sequence of instructions that form a
				1949	<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
				1950	you must have a pointer to the instruction that you wish to delete. Second, you
				1951	need to obtain the pointer to that instruction's basic block. You use the
				1952	pointer to the basic block to get its list of instructions and then use the
				1953	erase function to remove your instruction. For example:</p>
				1954
				1955	<div class="doc_code">
				1956	<pre>
				1957	<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner	3db8f77	2008-02-15 22:57:17 +0000	[diff] [blame]	1958	I->eraseFromParent();
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1959	</pre>
				1960	</div>
				1961
				1962	</div>
				1963
				1964	<!--_______________________________________________________________________-->
				1965	<div class="doc_subsubsection">
				1966	<a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
				1967	<tt>Value</tt></a>
				1968	</div>
				1969
				1970	<div class="doc_text">
				1971
				1972	<p><i>Replacing individual instructions</i></p>
				1973
				1974	<p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
				1975	permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
				1976	and <tt>ReplaceInstWithInst</tt>.</p>
				1977
				1978	<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
				1979
				1980	<ul>
				1981	<li><tt>ReplaceInstWithValue</tt>
				1982
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	1983	<p>This function replaces all uses of a given instruction with a value,
				1984	and then removes the original instruction. The following example
				1985	illustrates the replacement of the result of a particular
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1986	<tt>AllocaInst</tt> that allocates memory for a single integer with a null
				1987	pointer to an integer.</p>
				1988
				1989	<div class="doc_code">
				1990	<pre>
				1991	AllocaInst* instToReplace = ...;
				1992	BasicBlock::iterator ii(instToReplace);
				1993
				1994	ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	1995	Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1996	</pre></div></li>
				1997
				1998	<li><tt>ReplaceInstWithInst</tt>
				1999
				2000	<p>This function replaces a particular instruction with another
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	2001	instruction, inserting the new instruction into the basic block at the
				2002	location where the old instruction was, and replacing any uses of the old
				2003	instruction with the new instruction. The following example illustrates
				2004	the replacement of one <tt>AllocaInst</tt> with another.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2005
				2006	<div class="doc_code">
				2007	<pre>
				2008	AllocaInst* instToReplace = ...;
				2009	BasicBlock::iterator ii(instToReplace);
				2010
				2011	ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2012	new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2013	</pre></div></li>
				2014	</ul>
				2015
				2016	<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
				2017
				2018	<p>You can use <tt>Value::replaceAllUsesWith</tt> and
				2019	<tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
				2020	doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
				2021	and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
				2022	information.</p>
				2023
				2024	<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
				2025	include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
				2026	ReplaceInstWithValue, ReplaceInstWithInst -->
				2027
				2028	</div>
				2029
				2030	<!--_______________________________________________________________________-->
				2031	<div class="doc_subsubsection">
				2032	<a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
				2033	</div>
				2034
				2035	<div class="doc_text">
				2036
				2037	<p>Deleting a global variable from a module is just as easy as deleting an
				2038	Instruction. First, you must have a pointer to the global variable that you wish
				2039	to delete. You use this pointer to erase it from its parent, the module.
				2040	For example:</p>
				2041
				2042	<div class="doc_code">
				2043	<pre>
				2044	<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
				2045
				2046	GV->eraseFromParent();
				2047	</pre>
				2048	</div>
				2049
				2050	</div>
				2051
				2052	<!-- *********************************************************************** -->
				2053	<div class="doc_section">
				2054	<a name="advanced">Advanced Topics</a>
				2055	</div>
				2056	<!-- *********************************************************************** -->
				2057
				2058	<div class="doc_text">
				2059	<p>
				2060	This section describes some of the advanced or obscure API's that most clients
				2061	do not need to be aware of. These API's tend manage the inner workings of the
				2062	LLVM system, and only need to be accessed in unusual circumstances.
				2063	</p>
				2064	</div>
				2065
				2066	<!-- ======================================================================= -->
				2067	<div class="doc_subsection">
				2068	<a name="TypeResolve">LLVM Type Resolution</a>
				2069	</div>
				2070
				2071	<div class="doc_text">
				2072
				2073	<p>
				2074	The LLVM type system has a very simple goal: allow clients to compare types for
				2075	structural equality with a simple pointer comparison (aka a shallow compare).
				2076	This goal makes clients much simpler and faster, and is used throughout the LLVM
				2077	system.
				2078	</p>
				2079
				2080	<p>
				2081	Unfortunately achieving this goal is not a simple matter. In particular,
				2082	recursive types and late resolution of opaque types makes the situation very
				2083	difficult to handle. Fortunately, for the most part, our implementation makes
				2084	most clients able to be completely unaware of the nasty internal details. The
				2085	primary case where clients are exposed to the inner workings of it are when
				2086	building a recursive type. In addition to this case, the LLVM bitcode reader,
				2087	assembly parser, and linker also have to be aware of the inner workings of this
				2088	system.
				2089	</p>
				2090
				2091	<p>
				2092	For our purposes below, we need three concepts. First, an "Opaque Type" is
				2093	exactly as defined in the <a href="LangRef.html#t_opaque">language
				2094	reference</a>. Second an "Abstract Type" is any type which includes an
				2095	opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
				2096	Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
				2097	float }</tt>").
				2098	</p>
				2099
				2100	</div>
				2101
				2102	<!-- ______________________________________________________________________ -->
				2103	<div class="doc_subsubsection">
				2104	<a name="BuildRecType">Basic Recursive Type Construction</a>
				2105	</div>
				2106
				2107	<div class="doc_text">
				2108
				2109	<p>
				2110	Because the most common question is "how do I build a recursive type with LLVM",
				2111	we answer it now and explain it as we go. Here we include enough to cause this
				2112	to be emitted to an output .ll file:
				2113	</p>
				2114
				2115	<div class="doc_code">
				2116	<pre>
				2117	%mylist = type { %mylist*, i32 }
				2118	</pre>
				2119	</div>
				2120
				2121	<p>
				2122	To build this, use the following LLVM APIs:
				2123	</p>
				2124
				2125	<div class="doc_code">
				2126	<pre>
				2127	// <i>Create the initial outer struct</i>
				2128	<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
				2129	std::vector<const Type*> Elts;
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	2130	Elts.push_back(PointerType::getUnqual(StructTy));
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2131	Elts.push_back(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2132	StructType *NewSTy = StructType::get(Elts);
				2133
				2134	// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
				2135	// <i>the struct and the opaque type are actually the same.</i>
				2136	cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
				2137
				2138	// <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
				2139	// <i>kept up-to-date</i>
				2140	NewSTy = cast<StructType>(StructTy.get());
				2141
				2142	// <i>Add a name for the type to the module symbol table (optional)</i>
				2143	MyModule->addTypeName("mylist", NewSTy);
				2144	</pre>
				2145	</div>
				2146
				2147	<p>
				2148	This code shows the basic approach used to build recursive types: build a
				2149	non-recursive type using 'opaque', then use type unification to close the cycle.
				2150	The type unification step is performed by the <tt><a
				2151	href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
				2152	described next. After that, we describe the <a
				2153	href="#PATypeHolder">PATypeHolder class</a>.
				2154	</p>
				2155
				2156	</div>
				2157
				2158	<!-- ______________________________________________________________________ -->
				2159	<div class="doc_subsubsection">
				2160	<a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
				2161	</div>
				2162
				2163	<div class="doc_text">
				2164	<p>
				2165	The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
				2166	While this method is actually a member of the DerivedType class, it is most
				2167	often used on OpaqueType instances. Type unification is actually a recursive
				2168	process. After unification, types can become structurally isomorphic to
				2169	existing types, and all duplicates are deleted (to preserve pointer equality).
				2170	</p>
				2171
				2172	<p>
				2173	In the example above, the OpaqueType object is definitely deleted.
				2174	Additionally, if there is an "{ \2*, i32}" type already created in the system,
				2175	the pointer and struct type created are <b>also</b> deleted. Obviously whenever
				2176	a type is deleted, any "Type*" pointers in the program are invalidated. As
				2177	such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
				2178	live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
				2179	types can never move or be deleted). To deal with this, the <a
				2180	href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
				2181	reference to a possibly refined type, and the <a
				2182	href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
				2183	complex datastructures.
				2184	</p>
				2185
				2186	</div>
				2187
				2188	<!-- ______________________________________________________________________ -->
				2189	<div class="doc_subsubsection">
				2190	<a name="PATypeHolder">The PATypeHolder Class</a>
				2191	</div>
				2192
				2193	<div class="doc_text">
				2194	<p>
				2195	PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
				2196	happily goes about nuking types that become isomorphic to existing types, it
				2197	automatically updates all PATypeHolder objects to point to the new type. In the
				2198	example above, this allows the code to maintain a pointer to the resultant
				2199	resolved recursive type, even though the Type*'s are potentially invalidated.
				2200	</p>
				2201
				2202	<p>
				2203	PATypeHolder is an extremely light-weight object that uses a lazy union-find
				2204	implementation to update pointers. For example the pointer from a Value to its
				2205	Type is maintained by PATypeHolder objects.
				2206	</p>
				2207
				2208	</div>
				2209
				2210	<!-- ______________________________________________________________________ -->
				2211	<div class="doc_subsubsection">
				2212	<a name="AbstractTypeUser">The AbstractTypeUser Class</a>
				2213	</div>
				2214
				2215	<div class="doc_text">
				2216
				2217	<p>
				2218	Some data structures need more to perform more complex updates when types get
				2219	resolved. To support this, a class can derive from the AbstractTypeUser class.
				2220	This class
				2221	allows it to get callbacks when certain types are resolved. To register to get
				2222	callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
				2223	methods can be called on a type. Note that these methods only work for <i>
				2224	abstract</i> types. Concrete types (those that do not include any opaque
				2225	objects) can never be refined.
				2226	</p>
				2227	</div>
				2228
				2229
				2230	<!-- ======================================================================= -->
				2231	<div class="doc_subsection">
				2232	<a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
				2233	<tt>TypeSymbolTable</tt> classes</a>
				2234	</div>
				2235
				2236	<div class="doc_text">
				2237	<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
				2238	ValueSymbolTable</a></tt> class provides a symbol table that the <a
				2239	href="#Function"><tt>Function</tt></a> and <a href="#Module">
				2240	<tt>Module</tt></a> classes use for naming value definitions. The symbol table
				2241	can provide a name for any <a href="#Value"><tt>Value</tt></a>.
				2242	The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
				2243	TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
				2244	names for types.</p>
				2245
				2246	<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
				2247	by most clients. It should only be used when iteration over the symbol table
				2248	names themselves are required, which is very special purpose. Note that not
				2249	all LLVM
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2250	<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]	2251	an empty name) do not exist in the symbol table.
				2252	</p>
				2253
				2254	<p>These symbol tables support iteration over the values/types in the symbol
				2255	table with <tt>begin/end/iterator</tt> and supports querying to see if a
				2256	specific name is in the symbol table (with <tt>lookup</tt>). The
				2257	<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
				2258	simply call <tt>setName</tt> on a value, which will autoinsert it into the
				2259	appropriate symbol table. For types, use the Module::addTypeName method to
				2260	insert entries into the symbol table.</p>
				2261
				2262	</div>
				2263
				2264
				2265
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2266	<!-- ======================================================================= -->
				2267	<div class="doc_subsection">
				2268	<a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
				2269	</div>
				2270
				2271	<div class="doc_text">
				2272	<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]	2273	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]	2274	towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
				2275	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]	2276	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]	2277	addition and removal.</p>
				2278
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2279	<!-- ______________________________________________________________________ -->
				2280	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2281	<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]	2282	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2283
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2284	<div class="doc_text">
				2285	<p>
				2286	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]	2287	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]	2288	(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
				2289	that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
				2290	</p>
				2291	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2292
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2293	<p>
				2294	We have 2 different layouts in the <tt>User</tt> (sub)classes:
				2295	<ul>
				2296	<li><p>Layout a)
				2297	The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
				2298	object and there are a fixed number of them.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2299
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2300	<li><p>Layout b)
				2301	The <tt>Use</tt> object(s) are referenced by a pointer to an
				2302	array from the <tt>User</tt> object and there may be a variable
				2303	number of them.</p>
				2304	</ul>
				2305	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2306	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]	2307	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]	2308	we stick to this redundancy for the sake of simplicity.
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2309	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]	2310	has. (Theoretically this information can also be calculated
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2311	given the scheme presented below.)</p>
				2312	<p>
				2313	Special forms of allocation operators (<tt>operator new</tt>)
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2314	enforce the following memory layouts:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2315
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2316	<ul>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2317	<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]	2318
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2319	<pre>
				2320	...---.---.---.---.-------...
				2321	\| P \| P \| P \| P \| User
				2322	'''---'---'---'---'-------'''
				2323	</pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2324
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2325	<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]	2326	<pre>
				2327	.-------...
				2328	\| User
				2329	'-------'''
				2330	\|
				2331	v
				2332	.---.---.---.---...
				2333	\| P \| P \| P \| P \|
				2334	'---'---'---'---'''
				2335	</pre>
				2336	</ul>
				2337	<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
				2338	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]	2339
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2340	<!-- ______________________________________________________________________ -->
				2341	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2342	<a name="Waymarking">The waymarking algorithm</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2343	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2344
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2345	<div class="doc_text">
				2346	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2347	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]	2348	their <tt>User</tt> objects, there must be a fast and exact method to
				2349	recover it. This is accomplished by the following scheme:</p>
				2350	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2351
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2352	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]	2353	start of the <tt>User</tt> object:
				2354	<ul>
				2355	<li><tt>00</tt> —> binary digit 0</li>
				2356	<li><tt>01</tt> —> binary digit 1</li>
				2357	<li><tt>10</tt> —> stop and calculate (<tt>s</tt>)</li>
				2358	<li><tt>11</tt> —> full stop (<tt>S</tt>)</li>
				2359	</ul>
				2360	<p>
				2361	Given a <tt>Use*</tt>, all we have to do is to walk till we get
				2362	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]	2363	we have to walk to the next stop picking up digits
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2364	and calculating the offset:</p>
				2365	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2366	.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
				2367	\| 1 \| s \| 1 \| 0 \| 1 \| 0 \| s \| 1 \| 1 \| 0 \| s \| 1 \| 1 \| s \| 1 \| S \| User (or User*)
				2368	'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
				2369	\|+15 \|+10 \|+6 \|+3 \|+1
				2370	\| \| \| \| \|__>
				2371	\| \| \| \|__________>
				2372	\| \| \|______________________>
				2373	\| \|______________________________________>
				2374	\|__________________________________________________________>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2375	</pre>
				2376	<p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2377	Only the significant number of bits need to be stored between the
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2378	stops, so that the <i>worst case is 20 memory accesses</i> when there are
				2379	1000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2380
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2381	<!-- ______________________________________________________________________ -->
				2382	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2383	<a name="ReferenceImpl">Reference implementation</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2384	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2385
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2386	<div class="doc_text">
				2387	<p>
				2388	The following literate Haskell fragment demonstrates the concept:</p>
				2389	</div>
				2390
				2391	<div class="doc_code">
				2392	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2393	> import Test.QuickCheck
				2394	>
				2395	> digits :: Int -> [Char] -> [Char]
				2396	> digits 0 acc = '0' : acc
				2397	> digits 1 acc = '1' : acc
				2398	> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
				2399	>
				2400	> dist :: Int -> [Char] -> [Char]
				2401	> dist 0 [] = ['S']
				2402	> dist 0 acc = acc
				2403	> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
				2404	> dist n acc = dist (n - 1) $ dist 1 acc
				2405	>
				2406	> takeLast n ss = reverse $ take n $ reverse ss
				2407	>
				2408	> test = takeLast 40 $ dist 20 []
				2409	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2410	</pre>
				2411	</div>
				2412	<p>
				2413	Printing <test> gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
				2414	<p>
				2415	The reverse algorithm computes the length of the string just by examining
				2416	a certain prefix:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2417
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2418	<div class="doc_code">
				2419	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2420	> pref :: [Char] -> Int
				2421	> pref "S" = 1
				2422	> pref ('s':'1':rest) = decode 2 1 rest
				2423	> pref (_:rest) = 1 + pref rest
				2424	>
				2425	> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
				2426	> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
				2427	> decode walk acc _ = walk + acc
				2428	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2429	</pre>
				2430	</div>
				2431	<p>
				2432	Now, as expected, printing <pref test> gives <tt>40</tt>.</p>
				2433	<p>
				2434	We can <i>quickCheck</i> this with following property:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2435
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2436	<div class="doc_code">
				2437	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2438	> testcase = dist 2000 []
				2439	> testcaseLength = length testcase
				2440	>
				2441	> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
				2442	> where arr = takeLast n testcase
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2443	>
				2444	</pre>
				2445	</div>
				2446	<p>
				2447	As expected <quickCheck identityProp> gives:</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> quickCheck identityProp
				2451	OK, passed 100 tests.
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2452	</pre>
				2453	<p>
				2454	Let's be a bit more exhaustive:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2455
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2456	<div class="doc_code">
				2457	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2458	>
				2459	> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
				2460	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2461	</pre>
				2462	</div>
				2463	<p>
				2464	And here is the result of <deepCheck identityProp>:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2465
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2466	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2467	*Main> deepCheck identityProp
				2468	OK, passed 500 tests.
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2469	</pre>
				2470
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2471	<!-- ______________________________________________________________________ -->
				2472	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2473	<a name="Tagging">Tagging considerations</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2474	</div>
				2475
				2476	<p>
				2477	To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
				2478	never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
				2479	new <tt>Use**</tt> on every modification. Accordingly getters must strip the
				2480	tag bits.</p>
				2481	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2482	For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
				2483	Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
				2484	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]	2485	the LSBit set. (Portability is relying on the fact that all known compilers place the
				2486	<tt>vptr</tt> in the first word of the instances.)</p>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2487
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2488	</div>
				2489
				2490	<!-- *********************************************************************** -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2491	<div class="doc_section">
				2492	<a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
				2493	</div>
				2494	<!-- *********************************************************************** -->
				2495
				2496	<div class="doc_text">
				2497	<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
				2498	<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
				2499
				2500	<p>The Core LLVM classes are the primary means of representing the program
				2501	being inspected or transformed. The core LLVM classes are defined in
				2502	header files in the <tt>include/llvm/</tt> directory, and implemented in
				2503	the <tt>lib/VMCore</tt> directory.</p>
				2504
				2505	</div>
				2506
				2507	<!-- ======================================================================= -->
				2508	<div class="doc_subsection">
				2509	<a name="Type">The <tt>Type</tt> class and Derived Types</a>
				2510	</div>
				2511
				2512	<div class="doc_text">
				2513
				2514	<p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
				2515	a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
				2516	through its subclasses. Certain primitive types (<tt>VoidType</tt>,
				2517	<tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
				2518	subclasses. They are hidden because they offer no useful functionality beyond
				2519	what the <tt>Type</tt> class offers except to distinguish themselves from
				2520	other subclasses of <tt>Type</tt>.</p>
				2521	<p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
				2522	named, but this is not a requirement. There exists exactly
				2523	one instance of a given shape at any one time. This allows type equality to
				2524	be performed with address equality of the Type Instance. That is, given two
				2525	<tt>Type*</tt> values, the types are identical if the pointers are identical.
				2526	</p>
				2527	</div>
				2528
				2529	<!-- _______________________________________________________________________ -->
				2530	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2531	<a name="m_Type">Important Public Methods</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2532	</div>
				2533
				2534	<div class="doc_text">
				2535
				2536	<ul>
				2537	<li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
				2538
				2539	<li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
				2540	floating point types.</li>
				2541
				2542	<li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
				2543	an OpaqueType anywhere in its definition).</li>
				2544
				2545	<li><tt>bool isSized()</tt>: Return true if the type has known size. Things
				2546	that don't have a size are abstract types, labels and void.</li>
				2547
				2548	</ul>
				2549	</div>
				2550
				2551	<!-- _______________________________________________________________________ -->
				2552	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2553	<a name="derivedtypes">Important Derived Types</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2554	</div>
				2555	<div class="doc_text">
				2556	<dl>
				2557	<dt><tt>IntegerType</tt></dt>
				2558	<dd>Subclass of DerivedType that represents integer types of any bit width.
				2559	Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
				2560	<tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
				2561	<ul>
				2562	<li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
				2563	type of a specific bit width.</li>
				2564	<li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
				2565	type.</li>
				2566	</ul>
				2567	</dd>
				2568	<dt><tt>SequentialType</tt></dt>
				2569	<dd>This is subclassed by ArrayType and PointerType
				2570	<ul>
				2571	<li><tt>const Type * getElementType() const</tt>: Returns the type of each
				2572	of the elements in the sequential type. </li>
				2573	</ul>
				2574	</dd>
				2575	<dt><tt>ArrayType</tt></dt>
				2576	<dd>This is a subclass of SequentialType and defines the interface for array
				2577	types.
				2578	<ul>
				2579	<li><tt>unsigned getNumElements() const</tt>: Returns the number of
				2580	elements in the array. </li>
				2581	</ul>
				2582	</dd>
				2583	<dt><tt>PointerType</tt></dt>
				2584	<dd>Subclass of SequentialType for pointer types.</dd>
				2585	<dt><tt>VectorType</tt></dt>
				2586	<dd>Subclass of SequentialType for vector types. A
				2587	vector type is similar to an ArrayType but is distinguished because it is
				2588	a first class type wherease ArrayType is not. Vector types are used for
				2589	vector operations and are usually small vectors of of an integer or floating
				2590	point type.</dd>
				2591	<dt><tt>StructType</tt></dt>
				2592	<dd>Subclass of DerivedTypes for struct types.</dd>
				2593	<dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
				2594	<dd>Subclass of DerivedTypes for function types.
				2595	<ul>
				2596	<li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
				2597	function</li>
				2598	<li><tt> const Type * getReturnType() const</tt>: Returns the
				2599	return type of the function.</li>
				2600	<li><tt>const Type * getParamType (unsigned i)</tt>: Returns
				2601	the type of the ith parameter.</li>
				2602	<li><tt> const unsigned getNumParams() const</tt>: Returns the
				2603	number of formal parameters.</li>
				2604	</ul>
				2605	</dd>
				2606	<dt><tt>OpaqueType</tt></dt>
				2607	<dd>Sublcass of DerivedType for abstract types. This class
				2608	defines no content and is used as a placeholder for some other type. Note
				2609	that OpaqueType is used (temporarily) during type resolution for forward
				2610	references of types. Once the referenced type is resolved, the OpaqueType
				2611	is replaced with the actual type. OpaqueType can also be used for data
				2612	abstraction. At link time opaque types can be resolved to actual types
				2613	of the same name.</dd>
				2614	</dl>
				2615	</div>
				2616
				2617
				2618
				2619	<!-- ======================================================================= -->
				2620	<div class="doc_subsection">
				2621	<a name="Module">The <tt>Module</tt> class</a>
				2622	</div>
				2623
				2624	<div class="doc_text">
				2625
				2626	<p><tt>#include "<a
				2627	href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
				2628	<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
				2629
				2630	<p>The <tt>Module</tt> class represents the top level structure present in LLVM
				2631	programs. An LLVM module is effectively either a translation unit of the
				2632	original program or a combination of several translation units merged by the
				2633	linker. The <tt>Module</tt> class keeps track of a list of <a
				2634	href="#Function"><tt>Function</tt></a>s, a list of <a
				2635	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
				2636	href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
				2637	helpful member functions that try to make common operations easy.</p>
				2638
				2639	</div>
				2640
				2641	<!-- _______________________________________________________________________ -->
				2642	<div class="doc_subsubsection">
				2643	<a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
				2644	</div>
				2645
				2646	<div class="doc_text">
				2647
				2648	<ul>
				2649	<li><tt>Module::Module(std::string name = "")</tt></li>
				2650	</ul>
				2651
				2652	<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
				2653	provide a name for it (probably based on the name of the translation unit).</p>
				2654
				2655	<ul>
				2656	<li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
				2657	<tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
				2658
				2659	<tt>begin()</tt>, <tt>end()</tt>
				2660	<tt>size()</tt>, <tt>empty()</tt>
				2661
				2662	<p>These are forwarding methods that make it easy to access the contents of
				2663	a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
				2664	list.</p></li>
				2665
				2666	<li><tt>Module::FunctionListType &getFunctionList()</tt>
				2667
				2668	<p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
				2669	necessary to use when you need to update the list or perform a complex
				2670	action that doesn't have a forwarding method.</p>
				2671
				2672	<p><!-- Global Variable --></p></li>
				2673	</ul>
				2674
				2675	<hr>
				2676
				2677	<ul>
				2678	<li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
				2679
				2680	<tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
				2681
				2682	<tt>global_begin()</tt>, <tt>global_end()</tt>
				2683	<tt>global_size()</tt>, <tt>global_empty()</tt>
				2684
				2685	<p> These are forwarding methods that make it easy to access the contents of
				2686	a <tt>Module</tt> object's <a
				2687	href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
				2688
				2689	<li><tt>Module::GlobalListType &getGlobalList()</tt>
				2690
				2691	<p>Returns the list of <a
				2692	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
				2693	use when you need to update the list or perform a complex action that
				2694	doesn't have a forwarding method.</p>
				2695
				2696	<p><!-- Symbol table stuff --> </p></li>
				2697	</ul>
				2698
				2699	<hr>
				2700
				2701	<ul>
				2702	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				2703
				2704	<p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				2705	for this <tt>Module</tt>.</p>
				2706
				2707	<p><!-- Convenience methods --></p></li>
				2708	</ul>
				2709
				2710	<hr>
				2711
				2712	<ul>
				2713	<li><tt><a href="#Function">Function</a> *getFunction(const std::string
				2714	&Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
				2715
				2716	<p>Look up the specified function in the <tt>Module</tt> <a
				2717	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
				2718	<tt>null</tt>.</p></li>
				2719
				2720	<li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
				2721	std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
				2722
				2723	<p>Look up the specified function in the <tt>Module</tt> <a
				2724	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
				2725	external declaration for the function and return it.</p></li>
				2726
				2727	<li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
				2728
				2729	<p>If there is at least one entry in the <a
				2730	href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
				2731	href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
				2732	string.</p></li>
				2733
				2734	<li><tt>bool addTypeName(const std::string &Name, const <a
				2735	href="#Type">Type</a> *Ty)</tt>
				2736
				2737	<p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				2738	mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
				2739	name, true is returned and the <a
				2740	href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
				2741	</ul>
				2742
				2743	</div>
				2744
				2745
				2746	<!-- ======================================================================= -->
				2747	<div class="doc_subsection">
				2748	<a name="Value">The <tt>Value</tt> class</a>
				2749	</div>
				2750
				2751	<div class="doc_text">
				2752
				2753	<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
				2754	<br>
				2755	doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
				2756
				2757	<p>The <tt>Value</tt> class is the most important class in the LLVM Source
				2758	base. It represents a typed value that may be used (among other things) as an
				2759	operand to an instruction. There are many different types of <tt>Value</tt>s,
				2760	such as <a href="#Constant"><tt>Constant</tt></a>s,<a
				2761	href="#Argument"><tt>Argument</tt></a>s. Even <a
				2762	href="#Instruction"><tt>Instruction</tt></a>s and <a
				2763	href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
				2764
				2765	<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
				2766	for a program. For example, an incoming argument to a function (represented
				2767	with an instance of the <a href="#Argument">Argument</a> class) is "used" by
				2768	every instruction in the function that references the argument. To keep track
				2769	of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
				2770	href="#User"><tt>User</tt></a>s that is using it (the <a
				2771	href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
				2772	graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
				2773	def-use information in the program, and is accessible through the <tt>use_</tt>*
				2774	methods, shown below.</p>
				2775
				2776	<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
				2777	and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
				2778	method. In addition, all LLVM values can be named. The "name" of the
				2779	<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
				2780
				2781	<div class="doc_code">
				2782	<pre>
				2783	%<b>foo</b> = add i32 1, 2
				2784	</pre>
				2785	</div>
				2786
				2787	<p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
				2788	that the name of any value may be missing (an empty string), so names should
				2789	<b>ONLY</b> be used for debugging (making the source code easier to read,
				2790	debugging printouts), they should not be used to keep track of values or map
				2791	between them. For this purpose, use a <tt>std::map</tt> of pointers to the
				2792	<tt>Value</tt> itself instead.</p>
				2793
				2794	<p>One important aspect of LLVM is that there is no distinction between an SSA
				2795	variable and the operation that produces it. Because of this, any reference to
				2796	the value produced by an instruction (or the value available as an incoming
				2797	argument, for example) is represented as a direct pointer to the instance of
				2798	the class that
				2799	represents this value. Although this may take some getting used to, it
				2800	simplifies the representation and makes it easier to manipulate.</p>
				2801
				2802	</div>
				2803
				2804	<!-- _______________________________________________________________________ -->
				2805	<div class="doc_subsubsection">
				2806	<a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
				2807	</div>
				2808
				2809	<div class="doc_text">
				2810
				2811	<ul>
				2812	<li><tt>Value::use_iterator</tt> - Typedef for iterator over the
				2813	use-list<br>
				2814	<tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
				2815	the use-list<br>
				2816	<tt>unsigned use_size()</tt> - Returns the number of users of the
				2817	value.<br>
				2818	<tt>bool use_empty()</tt> - Returns true if there are no users.<br>
				2819	<tt>use_iterator use_begin()</tt> - Get an iterator to the start of
				2820	the use-list.<br>
				2821	<tt>use_iterator use_end()</tt> - Get an iterator to the end of the
				2822	use-list.<br>
				2823	<tt><a href="#User">User</a> *use_back()</tt> - Returns the last
				2824	element in the list.
				2825	<p> These methods are the interface to access the def-use
				2826	information in LLVM. As with all other iterators in LLVM, the naming
				2827	conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
				2828	</li>
				2829	<li><tt><a href="#Type">Type</a> *getType() const</tt>
				2830	<p>This method returns the Type of the Value.</p>
				2831	</li>
				2832	<li><tt>bool hasName() const</tt><br>
				2833	<tt>std::string getName() const</tt><br>
				2834	<tt>void setName(const std::string &Name)</tt>
				2835	<p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
				2836	be aware of the <a href="#nameWarning">precaution above</a>.</p>
				2837	</li>
				2838	<li><tt>void replaceAllUsesWith(Value *V)</tt>
				2839
				2840	<p>This method traverses the use list of a <tt>Value</tt> changing all <a
				2841	href="#User"><tt>User</tt>s</a> of the current value to refer to
				2842	"<tt>V</tt>" instead. For example, if you detect that an instruction always
				2843	produces a constant value (for example through constant folding), you can
				2844	replace all uses of the instruction with the constant like this:</p>
				2845
				2846	<div class="doc_code">
				2847	<pre>
				2848	Inst->replaceAllUsesWith(ConstVal);
				2849	</pre>
				2850	</div>
				2851
				2852	</ul>
				2853
				2854	</div>
				2855
				2856	<!-- ======================================================================= -->
				2857	<div class="doc_subsection">
				2858	<a name="User">The <tt>User</tt> class</a>
				2859	</div>
				2860
				2861	<div class="doc_text">
				2862
				2863	<p>
				2864	<tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
				2865	doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
				2866	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				2867
				2868	<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
				2869	refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
				2870	that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
				2871	referring to. The <tt>User</tt> class itself is a subclass of
				2872	<tt>Value</tt>.</p>
				2873
				2874	<p>The operands of a <tt>User</tt> point directly to the LLVM <a
				2875	href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
				2876	Single Assignment (SSA) form, there can only be one definition referred to,
				2877	allowing this direct connection. This connection provides the use-def
				2878	information in LLVM.</p>
				2879
				2880	</div>
				2881
				2882	<!-- _______________________________________________________________________ -->
				2883	<div class="doc_subsubsection">
				2884	<a name="m_User">Important Public Members of the <tt>User</tt> class</a>
				2885	</div>
				2886
				2887	<div class="doc_text">
				2888
				2889	<p>The <tt>User</tt> class exposes the operand list in two ways: through
				2890	an index access interface and through an iterator based interface.</p>
				2891
				2892	<ul>
				2893	<li><tt>Value *getOperand(unsigned i)</tt><br>
				2894	<tt>unsigned getNumOperands()</tt>
				2895	<p> These two methods expose the operands of the <tt>User</tt> in a
				2896	convenient form for direct access.</p></li>
				2897
				2898	<li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
				2899	list<br>
				2900	<tt>op_iterator op_begin()</tt> - Get an iterator to the start of
				2901	the operand list.<br>
				2902	<tt>op_iterator op_end()</tt> - Get an iterator to the end of the
				2903	operand list.
				2904	<p> Together, these methods make up the iterator based interface to
				2905	the operands of a <tt>User</tt>.</p></li>
				2906	</ul>
				2907
				2908	</div>
				2909
				2910	<!-- ======================================================================= -->
				2911	<div class="doc_subsection">
				2912	<a name="Instruction">The <tt>Instruction</tt> class</a>
				2913	</div>
				2914
				2915	<div class="doc_text">
				2916
				2917	<p><tt>#include "</tt><tt><a
				2918	href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
				2919	doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
				2920	Superclasses: <a href="#User"><tt>User</tt></a>, <a
				2921	href="#Value"><tt>Value</tt></a></p>
				2922
				2923	<p>The <tt>Instruction</tt> class is the common base class for all LLVM
				2924	instructions. It provides only a few methods, but is a very commonly used
				2925	class. The primary data tracked by the <tt>Instruction</tt> class itself is the
				2926	opcode (instruction type) and the parent <a
				2927	href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
				2928	into. To represent a specific type of instruction, one of many subclasses of
				2929	<tt>Instruction</tt> are used.</p>
				2930
				2931	<p> Because the <tt>Instruction</tt> class subclasses the <a
				2932	href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
				2933	way as for other <a href="#User"><tt>User</tt></a>s (with the
				2934	<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
				2935	<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
				2936	the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
				2937	file contains some meta-data about the various different types of instructions
				2938	in LLVM. It describes the enum values that are used as opcodes (for example
				2939	<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
				2940	concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
				2941	example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
				2942	href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
				2943	this file confuses doxygen, so these enum values don't show up correctly in the
				2944	<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
				2945
				2946	</div>
				2947
				2948	<!-- _______________________________________________________________________ -->
				2949	<div class="doc_subsubsection">
				2950	<a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
				2951	class</a>
				2952	</div>
				2953	<div class="doc_text">
				2954	<ul>
				2955	<li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
				2956	<p>This subclasses represents all two operand instructions whose operands
				2957	must be the same type, except for the comparison instructions.</p></li>
				2958	<li><tt><a name="CastInst">CastInst</a></tt>
				2959	<p>This subclass is the parent of the 12 casting instructions. It provides
				2960	common operations on cast instructions.</p>
				2961	<li><tt><a name="CmpInst">CmpInst</a></tt>
				2962	<p>This subclass respresents the two comparison instructions,
				2963	<a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
				2964	<a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
				2965	<li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
				2966	<p>This subclass is the parent of all terminator instructions (those which
				2967	can terminate a block).</p>
				2968	</ul>
				2969	</div>
				2970
				2971	<!-- _______________________________________________________________________ -->
				2972	<div class="doc_subsubsection">
				2973	<a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
				2974	class</a>
				2975	</div>
				2976
				2977	<div class="doc_text">
				2978
				2979	<ul>
				2980	<li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
				2981	<p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
				2982	this <tt>Instruction</tt> is embedded into.</p></li>
				2983	<li><tt>bool mayWriteToMemory()</tt>
				2984	<p>Returns true if the instruction writes to memory, i.e. it is a
				2985	<tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
				2986	<li><tt>unsigned getOpcode()</tt>
				2987	<p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
				2988	<li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
				2989	<p>Returns another instance of the specified instruction, identical
				2990	in all ways to the original except that the instruction has no parent
				2991	(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
				2992	and it has no name</p></li>
				2993	</ul>
				2994
				2995	</div>
				2996
				2997	<!-- ======================================================================= -->
				2998	<div class="doc_subsection">
				2999	<a name="Constant">The <tt>Constant</tt> class and subclasses</a>
				3000	</div>
				3001
				3002	<div class="doc_text">
				3003
				3004	<p>Constant represents a base class for different types of constants. It
				3005	is subclassed by ConstantInt, ConstantArray, etc. for representing
				3006	the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
				3007	a subclass, which represents the address of a global variable or function.
				3008	</p>
				3009
				3010	</div>
				3011
				3012	<!-- _______________________________________________________________________ -->
				3013	<div class="doc_subsubsection">Important Subclasses of Constant </div>
				3014	<div class="doc_text">
				3015	<ul>
				3016	<li>ConstantInt : This subclass of Constant represents an integer constant of
				3017	any width.
				3018	<ul>
				3019	<li><tt>const APInt& getValue() const</tt>: Returns the underlying
				3020	value of this constant, an APInt value.</li>
				3021	<li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
				3022	value to an int64_t via sign extension. If the value (not the bit width)
				3023	of the APInt is too large to fit in an int64_t, an assertion will result.
				3024	For this reason, use of this method is discouraged.</li>
				3025	<li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
				3026	value to a uint64_t via zero extension. IF the value (not the bit width)
				3027	of the APInt is too large to fit in a uint64_t, an assertion will result.
				3028	For this reason, use of this method is discouraged.</li>
				3029	<li><tt>static ConstantInt* get(const APInt& Val)</tt>: Returns the
				3030	ConstantInt object that represents the value provided by <tt>Val</tt>.
				3031	The type is implied as the IntegerType that corresponds to the bit width
				3032	of <tt>Val</tt>.</li>
				3033	<li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
				3034	Returns the ConstantInt object that represents the value provided by
				3035	<tt>Val</tt> for integer type <tt>Ty</tt>.</li>
				3036	</ul>
				3037	</li>
				3038	<li>ConstantFP : This class represents a floating point constant.
				3039	<ul>
				3040	<li><tt>double getValue() const</tt>: Returns the underlying value of
				3041	this constant. </li>
				3042	</ul>
				3043	</li>
				3044	<li>ConstantArray : This represents a constant array.
				3045	<ul>
				3046	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3047	a vector of component constants that makeup this array. </li>
				3048	</ul>
				3049	</li>
				3050	<li>ConstantStruct : This represents a constant struct.
				3051	<ul>
				3052	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3053	a vector of component constants that makeup this array. </li>
				3054	</ul>
				3055	</li>
				3056	<li>GlobalValue : This represents either a global variable or a function. In
				3057	either case, the value is a constant fixed address (after linking).
				3058	</li>
				3059	</ul>
				3060	</div>
				3061
				3062
				3063	<!-- ======================================================================= -->
				3064	<div class="doc_subsection">
				3065	<a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
				3066	</div>
				3067
				3068	<div class="doc_text">
				3069
				3070	<p><tt>#include "<a
				3071	href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
				3072	doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
				3073	Class</a><br>
				3074	Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
				3075	<a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
				3076
				3077	<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
				3078	href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
				3079	visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
				3080	Because they are visible at global scope, they are also subject to linking with
				3081	other globals defined in different translation units. To control the linking
				3082	process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
				3083	<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
				3084	defined by the <tt>LinkageTypes</tt> enumeration.</p>
				3085
				3086	<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
				3087	<tt>static</tt> in C), it is not visible to code outside the current translation
				3088	unit, and does not participate in linking. If it has external linkage, it is
				3089	visible to external code, and does participate in linking. In addition to
				3090	linkage information, <tt>GlobalValue</tt>s keep track of which <a
				3091	href="#Module"><tt>Module</tt></a> they are currently part of.</p>
				3092
				3093	<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
				3094	by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
				3095	global is always a pointer to its contents. It is important to remember this
				3096	when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
				3097	be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
				3098	subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
				3099	i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
				3100	the address of the first element of this array and the value of the
				3101	<tt>GlobalVariable</tt> are the same, they have different types. The
				3102	<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
				3103	is <tt>i32.</tt> Because of this, accessing a global value requires you to
				3104	dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
				3105	can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
				3106	Language Reference Manual</a>.</p>
				3107
				3108	</div>
				3109
				3110	<!-- _______________________________________________________________________ -->
				3111	<div class="doc_subsubsection">
				3112	<a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
				3113	class</a>
				3114	</div>
				3115
				3116	<div class="doc_text">
				3117
				3118	<ul>
				3119	<li><tt>bool hasInternalLinkage() const</tt><br>
				3120	<tt>bool hasExternalLinkage() const</tt><br>
				3121	<tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
				3122	<p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
				3123	<p> </p>
				3124	</li>
				3125	<li><tt><a href="#Module">Module</a> *getParent()</tt>
				3126	<p> This returns the <a href="#Module"><tt>Module</tt></a> that the
				3127	GlobalValue is currently embedded into.</p></li>
				3128	</ul>
				3129
				3130	</div>
				3131
				3132	<!-- ======================================================================= -->
				3133	<div class="doc_subsection">
				3134	<a name="Function">The <tt>Function</tt> class</a>
				3135	</div>
				3136
				3137	<div class="doc_text">
				3138
				3139	<p><tt>#include "<a
				3140	href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
				3141	info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
				3142	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3143	<a href="#Constant"><tt>Constant</tt></a>,
				3144	<a href="#User"><tt>User</tt></a>,
				3145	<a href="#Value"><tt>Value</tt></a></p>
				3146
				3147	<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
				3148	actually one of the more complex classes in the LLVM heirarchy because it must
				3149	keep track of a large amount of data. The <tt>Function</tt> class keeps track
				3150	of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
				3151	<a href="#Argument"><tt>Argument</tt></a>s, and a
				3152	<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
				3153
				3154	<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
				3155	commonly used part of <tt>Function</tt> objects. The list imposes an implicit
				3156	ordering of the blocks in the function, which indicate how the code will be
				3157	layed out by the backend. Additionally, the first <a
				3158	href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
				3159	<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
				3160	block. There are no implicit exit nodes, and in fact there may be multiple exit
				3161	nodes from a single <tt>Function</tt>. If the <a
				3162	href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
				3163	the <tt>Function</tt> is actually a function declaration: the actual body of the
				3164	function hasn't been linked in yet.</p>
				3165
				3166	<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
				3167	<tt>Function</tt> class also keeps track of the list of formal <a
				3168	href="#Argument"><tt>Argument</tt></a>s that the function receives. This
				3169	container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
				3170	nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
				3171	the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
				3172
				3173	<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
				3174	LLVM feature that is only used when you have to look up a value by name. Aside
				3175	from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
				3176	internally to make sure that there are not conflicts between the names of <a
				3177	href="#Instruction"><tt>Instruction</tt></a>s, <a
				3178	href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
				3179	href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
				3180
				3181	<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
				3182	and therefore also a <a href="#Constant">Constant</a>. The value of the function
				3183	is its address (after linking) which is guaranteed to be constant.</p>
				3184	</div>
				3185
				3186	<!-- _______________________________________________________________________ -->
				3187	<div class="doc_subsubsection">
				3188	<a name="m_Function">Important Public Members of the <tt>Function</tt>
				3189	class</a>
				3190	</div>
				3191
				3192	<div class="doc_text">
				3193
				3194	<ul>
				3195	<li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
				3196	Ty, LinkageTypes Linkage, const std::string &N = "", Module Parent = 0)</tt>
				3197
				3198	<p>Constructor used when you need to create new <tt>Function</tt>s to add
				3199	the the program. The constructor must specify the type of the function to
				3200	create and what type of linkage the function should have. The <a
				3201	href="#FunctionType"><tt>FunctionType</tt></a> argument
				3202	specifies the formal arguments and return value for the function. The same
				3203	<a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
				3204	create multiple functions. The <tt>Parent</tt> argument specifies the Module
				3205	in which the function is defined. If this argument is provided, the function
				3206	will automatically be inserted into that module's list of
				3207	functions.</p></li>
				3208
Chris Lattner	5e70957	2008-11-25 18:34:50 +0000	[diff] [blame]	3209	<li><tt>bool isDeclaration()</tt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3210
				3211	<p>Return whether or not the <tt>Function</tt> has a body defined. If the
				3212	function is "external", it does not have a body, and thus must be resolved
				3213	by linking with a function defined in a different translation unit.</p></li>
				3214
				3215	<li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
				3216	<tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
				3217
				3218	<tt>begin()</tt>, <tt>end()</tt>
				3219	<tt>size()</tt>, <tt>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="#BasicBlock"><tt>BasicBlock</tt></a>
				3223	list.</p></li>
				3224
				3225	<li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
				3226
				3227	<p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
				3228	is 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>Function::arg_iterator</tt> - Typedef for the argument list
				3232	iterator<br>
				3233	<tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
				3234
				3235	<tt>arg_begin()</tt>, <tt>arg_end()</tt>
				3236	<tt>arg_size()</tt>, <tt>arg_empty()</tt>
				3237
				3238	<p>These are forwarding methods that make it easy to access the contents of
				3239	a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
				3240	list.</p></li>
				3241
				3242	<li><tt>Function::ArgumentListType &getArgumentList()</tt>
				3243
				3244	<p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
				3245	necessary to use when you need to update the list or perform a complex
				3246	action that doesn't have a forwarding method.</p></li>
				3247
				3248	<li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
				3249
				3250	<p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
				3251	function. Because the entry block for the function is always the first
				3252	block, this returns the first block of the <tt>Function</tt>.</p></li>
				3253
				3254	<li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
				3255	<tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
				3256
				3257	<p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
				3258	<tt>Function</tt> and returns the return type of the function, or the <a
				3259	href="#FunctionType"><tt>FunctionType</tt></a> of the actual
				3260	function.</p></li>
				3261
				3262	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				3263
				3264	<p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				3265	for this <tt>Function</tt>.</p></li>
				3266	</ul>
				3267
				3268	</div>
				3269
				3270	<!-- ======================================================================= -->
				3271	<div class="doc_subsection">
				3272	<a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
				3273	</div>
				3274
				3275	<div class="doc_text">
				3276
				3277	<p><tt>#include "<a
				3278	href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
				3279	<br>
				3280	doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
				3281	Class</a><br>
				3282	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3283	<a href="#Constant"><tt>Constant</tt></a>,
				3284	<a href="#User"><tt>User</tt></a>,
				3285	<a href="#Value"><tt>Value</tt></a></p>
				3286
				3287	<p>Global variables are represented with the (suprise suprise)
				3288	<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
				3289	subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
				3290	always referenced by their address (global values must live in memory, so their
				3291	"name" refers to their constant address). See
				3292	<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
				3293	variables may have an initial value (which must be a
				3294	<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
				3295	they may be marked as "constant" themselves (indicating that their contents
				3296	never change at runtime).</p>
				3297	</div>
				3298
				3299	<!-- _______________________________________________________________________ -->
				3300	<div class="doc_subsubsection">
				3301	<a name="m_GlobalVariable">Important Public Members of the
				3302	<tt>GlobalVariable</tt> class</a>
				3303	</div>
				3304
				3305	<div class="doc_text">
				3306
				3307	<ul>
				3308	<li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
				3309	isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
				3310	Initializer = 0, const std::string &Name = "", Module Parent = 0)</tt>
				3311
				3312	<p>Create a new global variable of the specified type. If
				3313	<tt>isConstant</tt> is true then the global variable will be marked as
				3314	unchanging for the program. The Linkage parameter specifies the type of
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	3315	linkage (internal, external, weak, linkonce, appending) for the variable.
				3316	If the linkage is InternalLinkage, WeakAnyLinkage, WeakODRLinkage,
				3317	LinkOnceAnyLinkage or LinkOnceODRLinkage,  then the resultant
				3318	global variable will have internal linkage. AppendingLinkage concatenates
				3319	together all instances (in different translation units) of the variable
				3320	into a single variable but is only applicable to arrays.  See
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3321	the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
				3322	further details on linkage types. Optionally an initializer, a name, and the
				3323	module to put the variable into may be specified for the global variable as
				3324	well.</p></li>
				3325
				3326	<li><tt>bool isConstant() const</tt>
				3327
				3328	<p>Returns true if this is a global variable that is known not to
				3329	be modified at runtime.</p></li>
				3330
				3331	<li><tt>bool hasInitializer()</tt>
				3332
				3333	<p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
				3334
				3335	<li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
				3336
				3337	<p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
				3338	to call this method if there is no initializer.</p></li>
				3339	</ul>
				3340
				3341	</div>
				3342
				3343
				3344	<!-- ======================================================================= -->
				3345	<div class="doc_subsection">
				3346	<a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
				3347	</div>
				3348
				3349	<div class="doc_text">
				3350
				3351	<p><tt>#include "<a
				3352	href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
				3353	doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
				3354	Class</a><br>
				3355	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				3356
				3357	<p>This class represents a single entry multiple exit section of the code,
				3358	commonly known as a basic block by the compiler community. The
				3359	<tt>BasicBlock</tt> class maintains a list of <a
				3360	href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
				3361	Matching the language definition, the last element of this list of instructions
				3362	is always a terminator instruction (a subclass of the <a
				3363	href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
				3364
				3365	<p>In addition to tracking the list of instructions that make up the block, the
				3366	<tt>BasicBlock</tt> class also keeps track of the <a
				3367	href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
				3368
				3369	<p>Note that <tt>BasicBlock</tt>s themselves are <a
				3370	href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
				3371	like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
				3372	<tt>label</tt>.</p>
				3373
				3374	</div>
				3375
				3376	<!-- _______________________________________________________________________ -->
				3377	<div class="doc_subsubsection">
				3378	<a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
				3379	class</a>
				3380	</div>
				3381
				3382	<div class="doc_text">
				3383	<ul>
				3384
				3385	<li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
				3386	href="#Function">Function</a> *Parent = 0)</tt>
				3387
				3388	<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
				3389	insertion into a function. The constructor optionally takes a name for the new
				3390	block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
				3391	the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
				3392	automatically inserted at the end of the specified <a
				3393	href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
				3394	manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
				3395
				3396	<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
				3397	<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
				3398	<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
				3399	<tt>size()</tt>, <tt>empty()</tt>
				3400	STL-style functions for accessing the instruction list.
				3401
				3402	<p>These methods and typedefs are forwarding functions that have the same
				3403	semantics as the standard library methods of the same names. These methods
				3404	expose the underlying instruction list of a basic block in a way that is easy to
				3405	manipulate. To get the full complement of container operations (including
				3406	operations to update the list), you must use the <tt>getInstList()</tt>
				3407	method.</p></li>
				3408
				3409	<li><tt>BasicBlock::InstListType &getInstList()</tt>
				3410
				3411	<p>This method is used to get access to the underlying container that actually
				3412	holds the Instructions. This method must be used when there isn't a forwarding
				3413	function in the <tt>BasicBlock</tt> class for the operation that you would like
				3414	to perform. Because there are no forwarding functions for "updating"
				3415	operations, you need to use this if you want to update the contents of a
				3416	<tt>BasicBlock</tt>.</p></li>
				3417
				3418	<li><tt><a href="#Function">Function</a> *getParent()</tt>
				3419
				3420	<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
				3421	embedded into, or a null pointer if it is homeless.</p></li>
				3422
				3423	<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
				3424
				3425	<p> Returns a pointer to the terminator instruction that appears at the end of
				3426	the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
				3427	instruction in the block is not a terminator, then a null pointer is
				3428	returned.</p></li>
				3429
				3430	</ul>
				3431
				3432	</div>
				3433
				3434
				3435	<!-- ======================================================================= -->
				3436	<div class="doc_subsection">
				3437	<a name="Argument">The <tt>Argument</tt> class</a>
				3438	</div>
				3439
				3440	<div class="doc_text">
				3441
				3442	<p>This subclass of Value defines the interface for incoming formal
				3443	arguments to a function. A Function maintains a list of its formal
				3444	arguments. An argument has a pointer to the parent Function.</p>
				3445
				3446	</div>
				3447
				3448	<!-- *********************************************************************** -->
				3449	<hr>
				3450	<address>
				3451	<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman	947321d	2008-12-11 17:34:48 +0000	[diff] [blame]	3452	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]	3453	<a href="http://validator.w3.org/check/referer"><img
Gabor Greif	7dabe38	2008-06-18 14:05:31 +0000	[diff] [blame]	3454	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]	3455
				3456	<a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
				3457	<a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
				3458	<a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
				3459	Last modified: $Date$
				3460	</address>
				3461
				3462	</body>
				3463	</html>