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

blob: 7845d997b626f93e55e3040dc2c7549196a0feb2 [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>
Owen Anderson	60f8757	2009-06-16 17:40:28 +0000	[diff] [blame]	5	<meta http-equiv="Content-type" content="text/html;charset=UTF-8">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	6	<title>LLVM Programmer's Manual</title>
				7	<link rel="stylesheet" href="llvm.css" type="text/css">
				8	</head>
				9	<body>
				10
				11	<div class="doc_title">
				12	LLVM Programmer's Manual
				13	</div>
				14
				15	<ol>
				16	<li><a href="#introduction">Introduction</a></li>
				17	<li><a href="#general">General Information</a>
				18	<ul>
				19	<li><a href="#stl">The C++ Standard Template Library</a></li>
				20	<!--
				21	<li>The <tt>-time-passes</tt> option</li>
				22	<li>How to use the LLVM Makefile system</li>
				23	<li>How to write a regression test</li>
				24
				25	-->
				26	</ul>
				27	</li>
				28	<li><a href="#apis">Important and useful LLVM APIs</a>
				29	<ul>
				30	<li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
				31	and <tt>dyn_cast<></tt> templates</a> </li>
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	32	<li><a href="#string_apis">Passing strings (the <tt>StringRef</tt>
Benjamin Kramer	7b2136d	2009-08-05 15:42:44 +0000	[diff] [blame]	33	and <tt>Twine</tt> classes)</a>
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	34	<ul>
				35	<li><a href="#StringRef">The <tt>StringRef</tt> class</a> </li>
				36	<li><a href="#Twine">The <tt>Twine</tt> class</a> </li>
				37	</ul>
Benjamin Kramer	7b2136d	2009-08-05 15:42:44 +0000	[diff] [blame]	38	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	39	<li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
				40	option</a>
				41	<ul>
				42	<li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
				43	and the <tt>-debug-only</tt> option</a> </li>
				44	</ul>
				45	</li>
				46	<li><a href="#Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
				47	option</a></li>
				48	<!--
				49	<li>The <tt>InstVisitor</tt> template
				50	<li>The general graph API
				51	-->
				52	<li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
				53	</ul>
				54	</li>
				55	<li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
				56	<ul>
				57	<li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
				58	<ul>
				59	<li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
				60	<li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
				61	<li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
				62	<li><a href="#dss_vector"><vector></a></li>
				63	<li><a href="#dss_deque"><deque></a></li>
				64	<li><a href="#dss_list"><list></a></li>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	65	<li><a href="#dss_ilist">llvm/ADT/ilist.h</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	66	<li><a href="#dss_other">Other Sequential Container Options</a></li>
				67	</ul></li>
				68	<li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
				69	<ul>
				70	<li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
				71	<li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
				72	<li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattner	77ab46d	2007-09-30 00:58:59 +0000	[diff] [blame]	73	<li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	74	<li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
				75	<li><a href="#dss_set"><set></a></li>
				76	<li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
				77	<li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
				78	<li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
				79	</ul></li>
				80	<li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
				81	<ul>
				82	<li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
				83	<li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
				84	<li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
				85	<li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
Jeffrey Yasskin	50e846d	2009-10-22 22:11:22 +0000	[diff] [blame]	86	<li><a href="#dss_valuemap">"llvm/ADT/ValueMap.h"</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	87	<li><a href="#dss_map"><map></a></li>
				88	<li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
				89	</ul></li>
Chris Lattner	d8b95b7	2009-07-25 07:22:20 +0000	[diff] [blame]	90	<li><a href="#ds_string">String-like containers</a>
Benjamin Kramer	7b2136d	2009-08-05 15:42:44 +0000	[diff] [blame]	91	<!--<ul>
				92	todo
				93	</ul>--></li>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	94	<li><a href="#ds_bit">BitVector-like containers</a>
				95	<ul>
				96	<li><a href="#dss_bitvector">A dense bitvector</a></li>
Dan Gohman	c83a2b4	2010-01-05 18:24:00 +0000	[diff] [blame^]	97	<li><a href="#dss_smallbitvector">A "small" dense bitvector</a></li>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	98	<li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
				99	</ul></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	100	</ul>
				101	</li>
				102	<li><a href="#common">Helpful Hints for Common Operations</a>
				103	<ul>
				104	<li><a href="#inspection">Basic Inspection and Traversal Routines</a>
				105	<ul>
				106	<li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
				107	in a <tt>Function</tt></a> </li>
				108	<li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
				109	in a <tt>BasicBlock</tt></a> </li>
				110	<li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
				111	in a <tt>Function</tt></a> </li>
				112	<li><a href="#iterate_convert">Turning an iterator into a
				113	class pointer</a> </li>
				114	<li><a href="#iterate_complex">Finding call sites: a more
				115	complex example</a> </li>
				116	<li><a href="#calls_and_invokes">Treating calls and invokes
				117	the same way</a> </li>
				118	<li><a href="#iterate_chains">Iterating over def-use &
				119	use-def chains</a> </li>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	120	<li><a href="#iterate_preds">Iterating over predecessors &
				121	successors of blocks</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	122	</ul>
				123	</li>
				124	<li><a href="#simplechanges">Making simple changes</a>
				125	<ul>
				126	<li><a href="#schanges_creating">Creating and inserting new
				127	<tt>Instruction</tt>s</a> </li>
				128	<li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
				129	<li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
				130	with another <tt>Value</tt></a> </li>
				131	<li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
				132	</ul>
				133	</li>
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	134	<li><a href="#create_types">How to Create Types</a></li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	135	<!--
				136	<li>Working with the Control Flow Graph
				137	<ul>
				138	<li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
				139	<li>
				140	<li>
				141	</ul>
				142	-->
				143	</ul>
				144	</li>
				145
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	146	<li><a href="#threading">Threads and LLVM</a>
				147	<ul>
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	148	<li><a href="#startmultithreaded">Entering and Exiting Multithreaded Mode
				149	</a></li>
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	150	<li><a href="#shutdown">Ending execution with <tt>llvm_shutdown()</tt></a></li>
				151	<li><a href="#managedstatic">Lazy initialization with <tt>ManagedStatic</tt></a></li>
Owen Anderson	3f2da0e	2009-08-19 17:58:52 +0000	[diff] [blame]	152	<li><a href="#llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a></li>
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	153	</ul>
				154	</li>
				155
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	156	<li><a href="#advanced">Advanced Topics</a>
				157	<ul>
				158	<li><a href="#TypeResolve">LLVM Type Resolution</a>
				159	<ul>
				160	<li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
				161	<li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
				162	<li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
				163	<li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
				164	</ul></li>
				165
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	166	<li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes</a></li>
				167	<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]	168	</ul></li>
				169
				170	<li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
				171	<ul>
				172	<li><a href="#Type">The <tt>Type</tt> class</a> </li>
				173	<li><a href="#Module">The <tt>Module</tt> class</a></li>
				174	<li><a href="#Value">The <tt>Value</tt> class</a>
				175	<ul>
				176	<li><a href="#User">The <tt>User</tt> class</a>
				177	<ul>
				178	<li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
				179	<li><a href="#Constant">The <tt>Constant</tt> class</a>
				180	<ul>
				181	<li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
				182	<ul>
				183	<li><a href="#Function">The <tt>Function</tt> class</a></li>
				184	<li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
				185	</ul>
				186	</li>
				187	</ul>
				188	</li>
				189	</ul>
				190	</li>
				191	<li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
				192	<li><a href="#Argument">The <tt>Argument</tt> class</a></li>
				193	</ul>
				194	</li>
				195	</ul>
				196	</li>
				197	</ol>
				198
				199	<div class="doc_author">
				200	<p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
				201	<a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	202	<a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	203	<a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>,
				204	<a href="mailto:rspencer@x10sys.com">Reid Spencer</a> and
				205	<a href="mailto:owen@apple.com">Owen Anderson</a></p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	206	</div>
				207
				208	<!-- *********************************************************************** -->
				209	<div class="doc_section">
				210	<a name="introduction">Introduction </a>
				211	</div>
				212	<!-- *********************************************************************** -->
				213
				214	<div class="doc_text">
				215
				216	<p>This document is meant to highlight some of the important classes and
				217	interfaces available in the LLVM source-base. This manual is not
				218	intended to explain what LLVM is, how it works, and what LLVM code looks
				219	like. It assumes that you know the basics of LLVM and are interested
				220	in writing transformations or otherwise analyzing or manipulating the
				221	code.</p>
				222
				223	<p>This document should get you oriented so that you can find your
				224	way in the continuously growing source code that makes up the LLVM
				225	infrastructure. Note that this manual is not intended to serve as a
				226	replacement for reading the source code, so if you think there should be
				227	a method in one of these classes to do something, but it's not listed,
				228	check the source. Links to the <a href="/doxygen/">doxygen</a> sources
				229	are provided to make this as easy as possible.</p>
				230
				231	<p>The first section of this document describes general information that is
				232	useful to know when working in the LLVM infrastructure, and the second describes
				233	the Core LLVM classes. In the future this manual will be extended with
				234	information describing how to use extension libraries, such as dominator
				235	information, CFG traversal routines, and useful utilities like the <tt><a
				236	href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
				237
				238	</div>
				239
				240	<!-- *********************************************************************** -->
				241	<div class="doc_section">
				242	<a name="general">General Information</a>
				243	</div>
				244	<!-- *********************************************************************** -->
				245
				246	<div class="doc_text">
				247
				248	<p>This section contains general information that is useful if you are working
				249	in the LLVM source-base, but that isn't specific to any particular API.</p>
				250
				251	</div>
				252
				253	<!-- ======================================================================= -->
				254	<div class="doc_subsection">
				255	<a name="stl">The C++ Standard Template Library</a>
				256	</div>
				257
				258	<div class="doc_text">
				259
				260	<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
				261	perhaps much more than you are used to, or have seen before. Because of
				262	this, you might want to do a little background reading in the
				263	techniques used and capabilities of the library. There are many good
				264	pages that discuss the STL, and several books on the subject that you
				265	can get, so it will not be discussed in this document.</p>
				266
				267	<p>Here are some useful links:</p>
				268
				269	<ol>
				270
				271	<li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
				272	reference</a> - an excellent reference for the STL and other parts of the
				273	standard C++ library.</li>
				274
				275	<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]	276	O'Reilly book in the making. It has a decent Standard Library
				277	Reference that rivals Dinkumware's, and is unfortunately no longer free since the
				278	book has been published.</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	279
				280	<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
				281	Questions</a></li>
				282
				283	<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
				284	Contains a useful <a
				285	href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
				286	STL</a>.</li>
				287
				288	<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
				289	Page</a></li>
				290
				291	<li><a href="http://64.78.49.204/">
				292	Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
				293	the book).</a></li>
				294
				295	</ol>
				296
				297	<p>You are also encouraged to take a look at the <a
				298	href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
				299	to write maintainable code more than where to put your curly braces.</p>
				300
				301	</div>
				302
				303	<!-- ======================================================================= -->
				304	<div class="doc_subsection">
				305	<a name="stl">Other useful references</a>
				306	</div>
				307
				308	<div class="doc_text">
				309
				310	<ol>
				311	<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
				312	Branch and Tag Primer</a></li>
				313	<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
				314	static and shared libraries across platforms</a></li>
				315	</ol>
				316
				317	</div>
				318
				319	<!-- *********************************************************************** -->
				320	<div class="doc_section">
				321	<a name="apis">Important and useful LLVM APIs</a>
				322	</div>
				323	<!-- *********************************************************************** -->
				324
				325	<div class="doc_text">
				326
				327	<p>Here we highlight some LLVM APIs that are generally useful and good to
				328	know about when writing transformations.</p>
				329
				330	</div>
				331
				332	<!-- ======================================================================= -->
				333	<div class="doc_subsection">
				334	<a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
				335	<tt>dyn_cast<></tt> templates</a>
				336	</div>
				337
				338	<div class="doc_text">
				339
				340	<p>The LLVM source-base makes extensive use of a custom form of RTTI.
				341	These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
				342	operator, but they don't have some drawbacks (primarily stemming from
				343	the fact that <tt>dynamic_cast<></tt> only works on classes that
				344	have a v-table). Because they are used so often, you must know what they
				345	do and how they work. All of these templates are defined in the <a
				346	href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
				347	file (note that you very rarely have to include this file directly).</p>
				348
				349	<dl>
				350	<dt><tt>isa<></tt>: </dt>
				351
				352	<dd><p>The <tt>isa<></tt> operator works exactly like the Java
				353	"<tt>instanceof</tt>" operator. It returns true or false depending on whether
				354	a reference or pointer points to an instance of the specified class. This can
				355	be very useful for constraint checking of various sorts (example below).</p>
				356	</dd>
				357
				358	<dt><tt>cast<></tt>: </dt>
				359
				360	<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]	361	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]	362	an assertion failure if it is not really an instance of the right type. This
				363	should be used in cases where you have some information that makes you believe
				364	that something is of the right type. An example of the <tt>isa<></tt>
				365	and <tt>cast<></tt> template is:</p>
				366
				367	<div class="doc_code">
				368	<pre>
				369	static bool isLoopInvariant(const <a href="#Value">Value</a> V, const Loop L) {
				370	if (isa<<a href="#Constant">Constant</a>>(V) \|\| isa<<a href="#Argument">Argument</a>>(V) \|\| isa<<a href="#GlobalValue">GlobalValue</a>>(V))
				371	return true;
				372
				373	// <i>Otherwise, it must be an instruction...</i>
				374	return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
				375	}
				376	</pre>
				377	</div>
				378
				379	<p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
				380	by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
				381	operator.</p>
				382
				383	</dd>
				384
				385	<dt><tt>dyn_cast<></tt>:</dt>
				386
				387	<dd><p>The <tt>dyn_cast<></tt> operator is a "checking cast" operation.
				388	It checks to see if the operand is of the specified type, and if so, returns a
				389	pointer to it (this operator does not work with references). If the operand is
				390	not of the correct type, a null pointer is returned. Thus, this works very
				391	much like the <tt>dynamic_cast<></tt> operator in C++, and should be
				392	used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
				393	operator is used in an <tt>if</tt> statement or some other flow control
				394	statement like this:</p>
				395
				396	<div class="doc_code">
				397	<pre>
				398	if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
				399	// <i>...</i>
				400	}
				401	</pre>
				402	</div>
				403
				404	<p>This form of the <tt>if</tt> statement effectively combines together a call
				405	to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
				406	statement, which is very convenient.</p>
				407
				408	<p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
				409	<tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
				410	abused. In particular, you should not use big chained <tt>if/then/else</tt>
				411	blocks to check for lots of different variants of classes. If you find
				412	yourself wanting to do this, it is much cleaner and more efficient to use the
				413	<tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
				414
				415	</dd>
				416
				417	<dt><tt>cast_or_null<></tt>: </dt>
				418
				419	<dd><p>The <tt>cast_or_null<></tt> operator works just like the
				420	<tt>cast<></tt> operator, except that it allows for a null pointer as an
				421	argument (which it then propagates). This can sometimes be useful, allowing
				422	you to combine several null checks into one.</p></dd>
				423
				424	<dt><tt>dyn_cast_or_null<></tt>: </dt>
				425
				426	<dd><p>The <tt>dyn_cast_or_null<></tt> operator works just like the
				427	<tt>dyn_cast<></tt> operator, except that it allows for a null pointer
				428	as an argument (which it then propagates). This can sometimes be useful,
				429	allowing you to combine several null checks into one.</p></dd>
				430
				431	</dl>
				432
				433	<p>These five templates can be used with any classes, whether they have a
				434	v-table or not. To add support for these templates, you simply need to add
				435	<tt>classof</tt> static methods to the class you are interested casting
				436	to. Describing this is currently outside the scope of this document, but there
				437	are lots of examples in the LLVM source base.</p>
				438
				439	</div>
				440
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	441
				442	<!-- ======================================================================= -->
				443	<div class="doc_subsection">
				444	<a name="string_apis">Passing strings (the <tt>StringRef</tt>
				445	and <tt>Twine</tt> classes)</a>
				446	</div>
				447
				448	<div class="doc_text">
				449
				450	<p>Although LLVM generally does not do much string manipulation, we do have
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	451	several important APIs which take strings. Two important examples are the
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	452	Value class -- which has names for instructions, functions, etc. -- and the
				453	StringMap class which is used extensively in LLVM and Clang.</p>
				454
				455	<p>These are generic classes, and they need to be able to accept strings which
				456	may have embedded null characters. Therefore, they cannot simply take
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	457	a <tt>const char *</tt>, and taking a <tt>const std::string&</tt> requires
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	458	clients to perform a heap allocation which is usually unnecessary. Instead,
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	459	many LLVM APIs use a <tt>const StringRef&</tt> or a <tt>const
				460	Twine&</tt> for passing strings efficiently.</p>
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	461
				462	</div>
				463
				464	<!-- _______________________________________________________________________ -->
				465	<div class="doc_subsubsection">
				466	<a name="StringRef">The <tt>StringRef</tt> class</a>
				467	</div>
				468
				469	<div class="doc_text">
				470
				471	<p>The <tt>StringRef</tt> data type represents a reference to a constant string
				472	(a character array and a length) and supports the common operations available
				473	on <tt>std:string</tt>, but does not require heap allocation.</p>
				474
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	475	<p>It can be implicitly constructed using a C style null-terminated string,
				476	an <tt>std::string</tt>, or explicitly with a character pointer and length.
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	477	For example, the <tt>StringRef</tt> find function is declared as:</p>
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	478
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	479	<div class="doc_code">
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	480	iterator find(const StringRef &Key);
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	481	</div>
				482
				483	<p>and clients can call it using any one of:</p>
				484
				485	<div class="doc_code">
				486	<pre>
				487	Map.find("foo"); <i>// Lookup "foo"</i>
				488	Map.find(std::string("bar")); <i>// Lookup "bar"</i>
				489	Map.find(StringRef("\0baz", 4)); <i>// Lookup "\0baz"</i>
				490	</pre>
				491	</div>
				492
				493	<p>Similarly, APIs which need to return a string may return a <tt>StringRef</tt>
				494	instance, which can be used directly or converted to an <tt>std::string</tt>
				495	using the <tt>str</tt> member function. See
				496	"<tt><a href="/doxygen/classllvm_1_1StringRef_8h-source.html">llvm/ADT/StringRef.h</a></tt>"
				497	for more information.</p>
				498
				499	<p>You should rarely use the <tt>StringRef</tt> class directly, because it contains
				500	pointers to external memory it is not generally safe to store an instance of the
Chris Lattner	c00114f	2009-07-25 07:16:59 +0000	[diff] [blame]	501	class (unless you know that the external storage will not be freed).</p>
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	502
				503	</div>
				504
				505	<!-- _______________________________________________________________________ -->
				506	<div class="doc_subsubsection">
				507	<a name="Twine">The <tt>Twine</tt> class</a>
				508	</div>
				509
				510	<div class="doc_text">
				511
				512	<p>The <tt>Twine</tt> class is an efficient way for APIs to accept concatenated
				513	strings. For example, a common LLVM paradigm is to name one instruction based on
				514	the name of another instruction with a suffix, for example:</p>
				515
				516	<div class="doc_code">
				517	<pre>
				518	New = CmpInst::Create(<i>...</i>, SO->getName() + ".cmp");
				519	</pre>
				520	</div>
				521
				522	<p>The <tt>Twine</tt> class is effectively a
				523	lightweight <a href="http://en.wikipedia.org/wiki/Rope_(computer_science)">rope</a>
				524	which points to temporary (stack allocated) objects. Twines can be implicitly
				525	constructed as the result of the plus operator applied to strings (i.e., a C
				526	strings, an <tt>std::string</tt>, or a <tt>StringRef</tt>). The twine delays the
				527	actual concatentation of strings until it is actually required, at which point
				528	it can be efficiently rendered directly into a character array. This avoids
				529	unnecessary heap allocation involved in constructing the temporary results of
				530	string concatenation. See
				531	"<tt><a href="/doxygen/classllvm_1_1Twine_8h-source.html">llvm/ADT/Twine.h</a></tt>"
Benjamin Kramer	7b2136d	2009-08-05 15:42:44 +0000	[diff] [blame]	532	for more information.</p>
Daniel Dunbar	e3572ba	2009-07-25 04:41:11 +0000	[diff] [blame]	533
				534	<p>As with a <tt>StringRef</tt>, <tt>Twine</tt> objects point to external memory
				535	and should almost never be stored or mentioned directly. They are intended
				536	solely for use when defining a function which should be able to efficiently
				537	accept concatenated strings.</p>
				538
				539	</div>
				540
				541
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	542	<!-- ======================================================================= -->
				543	<div class="doc_subsection">
				544	<a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
				545	</div>
				546
				547	<div class="doc_text">
				548
				549	<p>Often when working on your pass you will put a bunch of debugging printouts
				550	and other code into your pass. After you get it working, you want to remove
				551	it, but you may need it again in the future (to work out new bugs that you run
				552	across).</p>
				553
				554	<p> Naturally, because of this, you don't want to delete the debug printouts,
				555	but you don't want them to always be noisy. A standard compromise is to comment
				556	them out, allowing you to enable them if you need them in the future.</p>
				557
				558	<p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
				559	file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
				560	this problem. Basically, you can put arbitrary code into the argument of the
				561	<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
				562	tool) is run with the '<tt>-debug</tt>' command line argument:</p>
				563
				564	<div class="doc_code">
				565	<pre>
Daniel Dunbar	e32f9b2	2009-07-25 01:55:32 +0000	[diff] [blame]	566	DEBUG(errs() << "I am here!\n");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	567	</pre>
				568	</div>
				569
				570	<p>Then you can run your pass like this:</p>
				571
				572	<div class="doc_code">
				573	<pre>
				574	$ opt < a.bc > /dev/null -mypass
				575	<i><no output></i>
				576	$ opt < a.bc > /dev/null -mypass -debug
				577	I am here!
				578	</pre>
				579	</div>
				580
				581	<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
				582	to not have to create "yet another" command line option for the debug output for
				583	your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
				584	so they do not cause a performance impact at all (for the same reason, they
				585	should also not contain side-effects!).</p>
				586
				587	<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
				588	enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
				589	"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
				590	program hasn't been started yet, you can always just run it with
				591	<tt>-debug</tt>.</p>
				592
				593	</div>
				594
				595	<!-- _______________________________________________________________________ -->
				596	<div class="doc_subsubsection">
				597	<a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
				598	the <tt>-debug-only</tt> option</a>
				599	</div>
				600
				601	<div class="doc_text">
				602
				603	<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
				604	just turns on <b>too much</b> information (such as when working on the code
				605	generator). If you want to enable debug information with more fine-grained
				606	control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
				607	option as follows:</p>
				608
				609	<div class="doc_code">
				610	<pre>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	611	#undef DEBUG_TYPE
Daniel Dunbar	e32f9b2	2009-07-25 01:55:32 +0000	[diff] [blame]	612	DEBUG(errs() << "No debug type\n");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	613	#define DEBUG_TYPE "foo"
Daniel Dunbar	e32f9b2	2009-07-25 01:55:32 +0000	[diff] [blame]	614	DEBUG(errs() << "'foo' debug type\n");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	615	#undef DEBUG_TYPE
				616	#define DEBUG_TYPE "bar"
Daniel Dunbar	e32f9b2	2009-07-25 01:55:32 +0000	[diff] [blame]	617	DEBUG(errs() << "'bar' debug type\n"));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	618	#undef DEBUG_TYPE
				619	#define DEBUG_TYPE ""
Daniel Dunbar	e32f9b2	2009-07-25 01:55:32 +0000	[diff] [blame]	620	DEBUG(errs() << "No debug type (2)\n");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	621	</pre>
				622	</div>
				623
				624	<p>Then you can run your pass like this:</p>
				625
				626	<div class="doc_code">
				627	<pre>
				628	$ opt < a.bc > /dev/null -mypass
				629	<i><no output></i>
				630	$ opt < a.bc > /dev/null -mypass -debug
				631	No debug type
				632	'foo' debug type
				633	'bar' debug type
				634	No debug type (2)
				635	$ opt < a.bc > /dev/null -mypass -debug-only=foo
				636	'foo' debug type
				637	$ opt < a.bc > /dev/null -mypass -debug-only=bar
				638	'bar' debug type
				639	</pre>
				640	</div>
				641
				642	<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
				643	a file, to specify the debug type for the entire module (if you do this before
				644	you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
				645	<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
				646	"bar", because there is no system in place to ensure that names do not
				647	conflict. If two different modules use the same string, they will all be turned
				648	on when the name is specified. This allows, for example, all debug information
				649	for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
				650	even if the source lives in multiple files.</p>
				651
Daniel Dunbar	8350419	2009-08-07 23:48:59 +0000	[diff] [blame]	652	<p>The <tt>DEBUG_WITH_TYPE</tt> macro is also available for situations where you
				653	would like to set <tt>DEBUG_TYPE</tt>, but only for one specific <tt>DEBUG</tt>
				654	statement. It takes an additional first parameter, which is the type to use. For
Benjamin Kramer	5fb9d7e	2009-10-12 14:46:08 +0000	[diff] [blame]	655	example, the preceding example could be written as:</p>
Daniel Dunbar	8350419	2009-08-07 23:48:59 +0000	[diff] [blame]	656
				657
				658	<div class="doc_code">
				659	<pre>
				660	DEBUG_WITH_TYPE("", errs() << "No debug type\n");
				661	DEBUG_WITH_TYPE("foo", errs() << "'foo' debug type\n");
				662	DEBUG_WITH_TYPE("bar", errs() << "'bar' debug type\n"));
				663	DEBUG_WITH_TYPE("", errs() << "No debug type (2)\n");
				664	</pre>
				665	</div>
				666
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	667	</div>
				668
				669	<!-- ======================================================================= -->
				670	<div class="doc_subsection">
				671	<a name="Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
				672	option</a>
				673	</div>
				674
				675	<div class="doc_text">
				676
				677	<p>The "<tt><a
				678	href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
				679	provides a class named <tt>Statistic</tt> that is used as a unified way to
				680	keep track of what the LLVM compiler is doing and how effective various
				681	optimizations are. It is useful to see what optimizations are contributing to
				682	making a particular program run faster.</p>
				683
				684	<p>Often you may run your pass on some big program, and you're interested to see
				685	how many times it makes a certain transformation. Although you can do this with
				686	hand inspection, or some ad-hoc method, this is a real pain and not very useful
				687	for big programs. Using the <tt>Statistic</tt> class makes it very easy to
				688	keep track of this information, and the calculated information is presented in a
				689	uniform manner with the rest of the passes being executed.</p>
				690
				691	<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
				692	it are as follows:</p>
				693
				694	<ol>
				695	<li><p>Define your statistic like this:</p>
				696
				697	<div class="doc_code">
				698	<pre>
				699	#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
				700	STATISTIC(NumXForms, "The # of times I did stuff");
				701	</pre>
				702	</div>
				703
				704	<p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
				705	specified by the first argument. The pass name is taken from the DEBUG_TYPE
				706	macro, and the description is taken from the second argument. The variable
				707	defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
				708
				709	<li><p>Whenever you make a transformation, bump the counter:</p>
				710
				711	<div class="doc_code">
				712	<pre>
				713	++NumXForms; // <i>I did stuff!</i>
				714	</pre>
				715	</div>
				716
				717	</li>
				718	</ol>
				719
				720	<p>That's all you have to do. To get '<tt>opt</tt>' to print out the
				721	statistics gathered, use the '<tt>-stats</tt>' option:</p>
				722
				723	<div class="doc_code">
				724	<pre>
				725	$ opt -stats -mypassname < program.bc > /dev/null
				726	<i>... statistics output ...</i>
				727	</pre>
				728	</div>
				729
				730	<p> When running <tt>opt</tt> on a C file from the SPEC benchmark
				731	suite, it gives a report that looks like this:</p>
				732
				733	<div class="doc_code">
				734	<pre>
				735	7646 bitcodewriter - Number of normal instructions
				736	725 bitcodewriter - Number of oversized instructions
				737	129996 bitcodewriter - Number of bitcode bytes written
				738	2817 raise - Number of insts DCEd or constprop'd
				739	3213 raise - Number of cast-of-self removed
				740	5046 raise - Number of expression trees converted
				741	75 raise - Number of other getelementptr's formed
				742	138 raise - Number of load/store peepholes
				743	42 deadtypeelim - Number of unused typenames removed from symtab
				744	392 funcresolve - Number of varargs functions resolved
				745	27 globaldce - Number of global variables removed
				746	2 adce - Number of basic blocks removed
				747	134 cee - Number of branches revectored
				748	49 cee - Number of setcc instruction eliminated
				749	532 gcse - Number of loads removed
				750	2919 gcse - Number of instructions removed
				751	86 indvars - Number of canonical indvars added
				752	87 indvars - Number of aux indvars removed
				753	25 instcombine - Number of dead inst eliminate
				754	434 instcombine - Number of insts combined
				755	248 licm - Number of load insts hoisted
				756	1298 licm - Number of insts hoisted to a loop pre-header
				757	3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
				758	75 mem2reg - Number of alloca's promoted
				759	1444 cfgsimplify - Number of blocks simplified
				760	</pre>
				761	</div>
				762
				763	<p>Obviously, with so many optimizations, having a unified framework for this
				764	stuff is very nice. Making your pass fit well into the framework makes it more
				765	maintainable and useful.</p>
				766
				767	</div>
				768
				769	<!-- ======================================================================= -->
				770	<div class="doc_subsection">
				771	<a name="ViewGraph">Viewing graphs while debugging code</a>
				772	</div>
				773
				774	<div class="doc_text">
				775
				776	<p>Several of the important data structures in LLVM are graphs: for example
				777	CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
				778	LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
				779	<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
				780	DAGs</a>. In many cases, while debugging various parts of the compiler, it is
				781	nice to instantly visualize these graphs.</p>
				782
				783	<p>LLVM provides several callbacks that are available in a debug build to do
				784	exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
				785	the current LLVM tool will pop up a window containing the CFG for the function
				786	where each basic block is a node in the graph, and each node contains the
				787	instructions in the block. Similarly, there also exists
				788	<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
				789	<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
				790	and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
				791	you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
				792	up a window. Alternatively, you can sprinkle calls to these functions in your
				793	code in places you want to debug.</p>
				794
				795	<p>Getting this to work requires a small amount of configuration. On Unix
				796	systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
				797	toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
				798	Mac OS/X, download and install the Mac OS/X <a
				799	href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
				800	<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
				801	it) to your path. Once in your system and path are set up, rerun the LLVM
				802	configure script and rebuild LLVM to enable this functionality.</p>
				803
				804	<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
				805	<i>interesting</i> nodes in large complex graphs. From gdb, if you
				806	<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
				807	next <tt>call DAG.viewGraph()</tt> would highlight the node in the
				808	specified color (choices of colors can be found at <a
				809	href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
				810	complex node attributes can be provided with <tt>call
				811	DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
				812	found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
				813	Attributes</a>.) If you want to restart and clear all the current graph
				814	attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
				815
				816	</div>
				817
				818	<!-- *********************************************************************** -->
				819	<div class="doc_section">
				820	<a name="datastructure">Picking the Right Data Structure for a Task</a>
				821	</div>
				822	<!-- *********************************************************************** -->
				823
				824	<div class="doc_text">
				825
				826	<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
				827	and we commonly use STL data structures. This section describes the trade-offs
				828	you should consider when you pick one.</p>
				829
				830	<p>
				831	The first step is a choose your own adventure: do you want a sequential
				832	container, a set-like container, or a map-like container? The most important
				833	thing when choosing a container is the algorithmic properties of how you plan to
				834	access the container. Based on that, you should use:</p>
				835
				836	<ul>
				837	<li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
				838	of an value based on another value. Map-like containers also support
				839	efficient queries for containment (whether a key is in the map). Map-like
				840	containers generally do not support efficient reverse mapping (values to
				841	keys). If you need that, use two maps. Some map-like containers also
				842	support efficient iteration through the keys in sorted order. Map-like
				843	containers are the most expensive sort, only use them if you need one of
				844	these capabilities.</li>
				845
				846	<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
				847	stuff into a container that automatically eliminates duplicates. Some
				848	set-like containers support efficient iteration through the elements in
				849	sorted order. Set-like containers are more expensive than sequential
				850	containers.
				851	</li>
				852
				853	<li>a <a href="#ds_sequential">sequential</a> container provides
				854	the most efficient way to add elements and keeps track of the order they are
				855	added to the collection. They permit duplicates and support efficient
				856	iteration, but do not support efficient look-up based on a key.
				857	</li>
				858
Chris Lattner	d8b95b7	2009-07-25 07:22:20 +0000	[diff] [blame]	859	<li>a <a href="#ds_string">string</a> container is a specialized sequential
				860	container or reference structure that is used for character or byte
				861	arrays.</li>
				862
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	863	<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
				864	perform set operations on sets of numeric id's, while automatically
				865	eliminating duplicates. Bit containers require a maximum of 1 bit for each
				866	identifier you want to store.
				867	</li>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	868	</ul>
				869
				870	<p>
				871	Once the proper category of container is determined, you can fine tune the
				872	memory use, constant factors, and cache behaviors of access by intelligently
				873	picking a member of the category. Note that constant factors and cache behavior
				874	can be a big deal. If you have a vector that usually only contains a few
				875	elements (but could contain many), for example, it's much better to use
				876	<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
				877	. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
				878	cost of adding the elements to the container. </p>
				879
				880	</div>
				881
				882	<!-- ======================================================================= -->
				883	<div class="doc_subsection">
				884	<a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
				885	</div>
				886
				887	<div class="doc_text">
				888	There are a variety of sequential containers available for you, based on your
				889	needs. Pick the first in this section that will do what you want.
				890	</div>
				891
				892	<!-- _______________________________________________________________________ -->
				893	<div class="doc_subsubsection">
				894	<a name="dss_fixedarrays">Fixed Size Arrays</a>
				895	</div>
				896
				897	<div class="doc_text">
				898	<p>Fixed size arrays are very simple and very fast. They are good if you know
				899	exactly how many elements you have, or you have a (low) upper bound on how many
				900	you have.</p>
				901	</div>
				902
				903	<!-- _______________________________________________________________________ -->
				904	<div class="doc_subsubsection">
				905	<a name="dss_heaparrays">Heap Allocated Arrays</a>
				906	</div>
				907
				908	<div class="doc_text">
				909	<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
				910	the number of elements is variable, if you know how many elements you will need
				911	before the array is allocated, and if the array is usually large (if not,
				912	consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
				913	allocated array is the cost of the new/delete (aka malloc/free). Also note that
				914	if you are allocating an array of a type with a constructor, the constructor and
				915	destructors will be run for every element in the array (re-sizable vectors only
				916	construct those elements actually used).</p>
				917	</div>
				918
				919	<!-- _______________________________________________________________________ -->
				920	<div class="doc_subsubsection">
				921	<a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
				922	</div>
				923
				924	<div class="doc_text">
				925	<p><tt>SmallVector<Type, N></tt> is a simple class that looks and smells
				926	just like <tt>vector<Type></tt>:
				927	it supports efficient iteration, lays out elements in memory order (so you can
				928	do pointer arithmetic between elements), supports efficient push_back/pop_back
				929	operations, supports efficient random access to its elements, etc.</p>
				930
				931	<p>The advantage of SmallVector is that it allocates space for
				932	some number of elements (N) <b>in the object itself</b>. Because of this, if
				933	the SmallVector is dynamically smaller than N, no malloc is performed. This can
				934	be a big win in cases where the malloc/free call is far more expensive than the
				935	code that fiddles around with the elements.</p>
				936
				937	<p>This is good for vectors that are "usually small" (e.g. the number of
				938	predecessors/successors of a block is usually less than 8). On the other hand,
				939	this makes the size of the SmallVector itself large, so you don't want to
				940	allocate lots of them (doing so will waste a lot of space). As such,
				941	SmallVectors are most useful when on the stack.</p>
				942
				943	<p>SmallVector also provides a nice portable and efficient replacement for
				944	<tt>alloca</tt>.</p>
				945
				946	</div>
				947
				948	<!-- _______________________________________________________________________ -->
				949	<div class="doc_subsubsection">
				950	<a name="dss_vector"><vector></a>
				951	</div>
				952
				953	<div class="doc_text">
				954	<p>
				955	std::vector is well loved and respected. It is useful when SmallVector isn't:
				956	when the size of the vector is often large (thus the small optimization will
				957	rarely be a benefit) or if you will be allocating many instances of the vector
				958	itself (which would waste space for elements that aren't in the container).
				959	vector is also useful when interfacing with code that expects vectors :).
				960	</p>
				961
				962	<p>One worthwhile note about std::vector: avoid code like this:</p>
				963
				964	<div class="doc_code">
				965	<pre>
				966	for ( ... ) {
				967	std::vector<foo> V;
				968	use V;
				969	}
				970	</pre>
				971	</div>
				972
				973	<p>Instead, write this as:</p>
				974
				975	<div class="doc_code">
				976	<pre>
				977	std::vector<foo> V;
				978	for ( ... ) {
				979	use V;
				980	V.clear();
				981	}
				982	</pre>
				983	</div>
				984
				985	<p>Doing so will save (at least) one heap allocation and free per iteration of
				986	the loop.</p>
				987
				988	</div>
				989
				990	<!-- _______________________________________________________________________ -->
				991	<div class="doc_subsubsection">
				992	<a name="dss_deque"><deque></a>
				993	</div>
				994
				995	<div class="doc_text">
				996	<p>std::deque is, in some senses, a generalized version of std::vector. Like
				997	std::vector, it provides constant time random access and other similar
				998	properties, but it also provides efficient access to the front of the list. It
				999	does not guarantee continuity of elements within memory.</p>
				1000
				1001	<p>In exchange for this extra flexibility, std::deque has significantly higher
				1002	constant factor costs than std::vector. If possible, use std::vector or
				1003	something cheaper.</p>
				1004	</div>
				1005
				1006	<!-- _______________________________________________________________________ -->
				1007	<div class="doc_subsubsection">
				1008	<a name="dss_list"><list></a>
				1009	</div>
				1010
				1011	<div class="doc_text">
				1012	<p>std::list is an extremely inefficient class that is rarely useful.
				1013	It performs a heap allocation for every element inserted into it, thus having an
				1014	extremely high constant factor, particularly for small data types. std::list
				1015	also only supports bidirectional iteration, not random access iteration.</p>
				1016
				1017	<p>In exchange for this high cost, std::list supports efficient access to both
				1018	ends of the list (like std::deque, but unlike std::vector or SmallVector). In
				1019	addition, the iterator invalidation characteristics of std::list are stronger
				1020	than that of a vector class: inserting or removing an element into the list does
				1021	not invalidate iterator or pointers to other elements in the list.</p>
				1022	</div>
				1023
				1024	<!-- _______________________________________________________________________ -->
				1025	<div class="doc_subsubsection">
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	1026	<a name="dss_ilist">llvm/ADT/ilist.h</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1027	</div>
				1028
				1029	<div class="doc_text">
				1030	<p><tt>ilist<T></tt> implements an 'intrusive' doubly-linked list. It is
				1031	intrusive, because it requires the element to store and provide access to the
				1032	prev/next pointers for the list.</p>
				1033
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	1034	<p><tt>ilist</tt> has the same drawbacks as <tt>std::list</tt>, and additionally
				1035	requires an <tt>ilist_traits</tt> implementation for the element type, but it
				1036	provides some novel characteristics. In particular, it can efficiently store
				1037	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]	1038	removed from the list, and <tt>ilist</tt>s are guaranteed to support a
				1039	constant-time splice operation.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1040
Gabor Greif	e39f0e7	2009-03-12 09:47:03 +0000	[diff] [blame]	1041	<p>These properties are exactly what we want for things like
				1042	<tt>Instruction</tt>s and basic blocks, which is why these are implemented with
				1043	<tt>ilist</tt>s.</p>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	1044
				1045	Related classes of interest are explained in the following subsections:
				1046	<ul>
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	1047	<li><a href="#dss_ilist_traits">ilist_traits</a></li>
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	1048	<li><a href="#dss_iplist">iplist</a></li>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	1049	<li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
Gabor Greif	d970d69	2009-03-12 10:30:31 +0000	[diff] [blame]	1050	<li><a href="#dss_ilist_sentinel">Sentinels</a></li>
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	1051	</ul>
				1052	</div>
				1053
				1054	<!-- _______________________________________________________________________ -->
				1055	<div class="doc_subsubsection">
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	1056	<a name="dss_ilist_traits">ilist_traits</a>
				1057	</div>
				1058
				1059	<div class="doc_text">
				1060	<p><tt>ilist_traits<T></tt> is <tt>ilist<T></tt>'s customization
				1061	mechanism. <tt>iplist<T></tt> (and consequently <tt>ilist<T></tt>)
				1062	publicly derive from this traits class.</p>
				1063	</div>
				1064
				1065	<!-- _______________________________________________________________________ -->
				1066	<div class="doc_subsubsection">
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	1067	<a name="dss_iplist">iplist</a>
				1068	</div>
				1069
				1070	<div class="doc_text">
				1071	<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]	1072	supports a slightly narrower interface. Notably, inserters from
				1073	<tt>T&</tt> are absent.</p>
Gabor Greif	a17abe1	2009-02-27 13:28:07 +0000	[diff] [blame]	1074
				1075	<p><tt>ilist_traits<T></tt> is a public base of this class and can be
				1076	used for a wide variety of customizations.</p>
Gabor Greif	b6b21ec	2009-02-27 12:02:19 +0000	[diff] [blame]	1077	</div>
				1078
				1079	<!-- _______________________________________________________________________ -->
				1080	<div class="doc_subsubsection">
Gabor Greif	bb17f65	2009-02-27 11:37:41 +0000	[diff] [blame]	1081	<a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
				1082	</div>
				1083
				1084	<div class="doc_text">
				1085	<p><tt>ilist_node<T></tt> implements a the forward and backward links
				1086	that are expected by the <tt>ilist<T></tt> (and analogous containers)
				1087	in the default manner.</p>
				1088
				1089	<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]	1090	<tt>T</tt>, usually <tt>T</tt> publicly derives from
				1091	<tt>ilist_node<T></tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1092	</div>
				1093
				1094	<!-- _______________________________________________________________________ -->
				1095	<div class="doc_subsubsection">
Gabor Greif	d970d69	2009-03-12 10:30:31 +0000	[diff] [blame]	1096	<a name="dss_ilist_sentinel">Sentinels</a>
				1097	</div>
				1098
				1099	<div class="doc_text">
				1100	<p><tt>ilist</tt>s have another speciality that must be considered. To be a good
				1101	citizen in the C++ ecosystem, it needs to support the standard container
				1102	operations, such as <tt>begin</tt> and <tt>end</tt> iterators, etc. Also, the
				1103	<tt>operator--</tt> must work correctly on the <tt>end</tt> iterator in the
				1104	case of non-empty <tt>ilist</tt>s.</p>
				1105
				1106	<p>The only sensible solution to this problem is to allocate a so-called
				1107	<i>sentinel</i> along with the intrusive list, which serves as the <tt>end</tt>
				1108	iterator, providing the back-link to the last element. However conforming to the
				1109	C++ convention it is illegal to <tt>operator++</tt> beyond the sentinel and it
				1110	also must not be dereferenced.</p>
				1111
				1112	<p>These constraints allow for some implementation freedom to the <tt>ilist</tt>
				1113	how to allocate and store the sentinel. The corresponding policy is dictated
				1114	by <tt>ilist_traits<T></tt>. By default a <tt>T</tt> gets heap-allocated
				1115	whenever the need for a sentinel arises.</p>
				1116
				1117	<p>While the default policy is sufficient in most cases, it may break down when
				1118	<tt>T</tt> does not provide a default constructor. Also, in the case of many
				1119	instances of <tt>ilist</tt>s, the memory overhead of the associated sentinels
				1120	is wasted. To alleviate the situation with numerous and voluminous
				1121	<tt>T</tt>-sentinels, sometimes a trick is employed, leading to <i>ghostly
				1122	sentinels</i>.</p>
				1123
				1124	<p>Ghostly sentinels are obtained by specially-crafted <tt>ilist_traits<T></tt>
				1125	which superpose the sentinel with the <tt>ilist</tt> instance in memory. Pointer
				1126	arithmetic is used to obtain the sentinel, which is relative to the
				1127	<tt>ilist</tt>'s <tt>this</tt> pointer. The <tt>ilist</tt> is augmented by an
				1128	extra pointer, which serves as the back-link of the sentinel. This is the only
				1129	field in the ghostly sentinel which can be legally accessed.</p>
				1130	</div>
				1131
				1132	<!-- _______________________________________________________________________ -->
				1133	<div class="doc_subsubsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1134	<a name="dss_other">Other Sequential Container options</a>
				1135	</div>
				1136
				1137	<div class="doc_text">
				1138	<p>Other STL containers are available, such as std::string.</p>
				1139
				1140	<p>There are also various STL adapter classes such as std::queue,
				1141	std::priority_queue, std::stack, etc. These provide simplified access to an
				1142	underlying container but don't affect the cost of the container itself.</p>
				1143
				1144	</div>
				1145
				1146
				1147	<!-- ======================================================================= -->
				1148	<div class="doc_subsection">
				1149	<a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
				1150	</div>
				1151
				1152	<div class="doc_text">
				1153
				1154	<p>Set-like containers are useful when you need to canonicalize multiple values
				1155	into a single representation. There are several different choices for how to do
				1156	this, providing various trade-offs.</p>
				1157
				1158	</div>
				1159
				1160
				1161	<!-- _______________________________________________________________________ -->
				1162	<div class="doc_subsubsection">
				1163	<a name="dss_sortedvectorset">A sorted 'vector'</a>
				1164	</div>
				1165
				1166	<div class="doc_text">
				1167
				1168	<p>If you intend to insert a lot of elements, then do a lot of queries, a
				1169	great approach is to use a vector (or other sequential container) with
				1170	std::sort+std::unique to remove duplicates. This approach works really well if
				1171	your usage pattern has these two distinct phases (insert then query), and can be
				1172	coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
				1173	</p>
				1174
				1175	<p>
				1176	This combination provides the several nice properties: the result data is
				1177	contiguous in memory (good for cache locality), has few allocations, is easy to
				1178	address (iterators in the final vector are just indices or pointers), and can be
				1179	efficiently queried with a standard binary or radix search.</p>
				1180
				1181	</div>
				1182
				1183	<!-- _______________________________________________________________________ -->
				1184	<div class="doc_subsubsection">
				1185	<a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
				1186	</div>
				1187
				1188	<div class="doc_text">
				1189
				1190	<p>If you have a set-like data structure that is usually small and whose elements
				1191	are reasonably small, a <tt>SmallSet<Type, N></tt> is a good choice. This set
				1192	has space for N elements in place (thus, if the set is dynamically smaller than
				1193	N, no malloc traffic is required) and accesses them with a simple linear search.
				1194	When the set grows beyond 'N' elements, it allocates a more expensive representation that
				1195	guarantees efficient access (for most types, it falls back to std::set, but for
				1196	pointers it uses something far better, <a
				1197	href="#dss_smallptrset">SmallPtrSet</a>).</p>
				1198
				1199	<p>The magic of this class is that it handles small sets extremely efficiently,
				1200	but gracefully handles extremely large sets without loss of efficiency. The
				1201	drawback is that the interface is quite small: it supports insertion, queries
				1202	and erasing, but does not support iteration.</p>
				1203
				1204	</div>
				1205
				1206	<!-- _______________________________________________________________________ -->
				1207	<div class="doc_subsubsection">
				1208	<a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
				1209	</div>
				1210
				1211	<div class="doc_text">
				1212
				1213	<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
				1214	transparently implemented with a SmallPtrSet), but also supports iterators. If
				1215	more than 'N' insertions are performed, a single quadratically
				1216	probed hash table is allocated and grows as needed, providing extremely
				1217	efficient access (constant time insertion/deleting/queries with low constant
				1218	factors) and is very stingy with malloc traffic.</p>
				1219
				1220	<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
				1221	whenever an insertion occurs. Also, the values visited by the iterators are not
				1222	visited in sorted order.</p>
				1223
				1224	</div>
				1225
				1226	<!-- _______________________________________________________________________ -->
				1227	<div class="doc_subsubsection">
Chris Lattner	77ab46d	2007-09-30 00:58:59 +0000	[diff] [blame]	1228	<a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
				1229	</div>
				1230
				1231	<div class="doc_text">
				1232
				1233	<p>
				1234	DenseSet is a simple quadratically probed hash table. It excels at supporting
				1235	small values: it uses a single allocation to hold all of the pairs that
				1236	are currently inserted in the set. DenseSet is a great way to unique small
				1237	values that are not simple pointers (use <a
				1238	href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
				1239	the same requirements for the value type that <a
				1240	href="#dss_densemap">DenseMap</a> has.
				1241	</p>
				1242
				1243	</div>
				1244
				1245	<!-- _______________________________________________________________________ -->
				1246	<div class="doc_subsubsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1247	<a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
				1248	</div>
				1249
				1250	<div class="doc_text">
				1251
				1252	<p>
				1253	FoldingSet is an aggregate class that is really good at uniquing
				1254	expensive-to-create or polymorphic objects. It is a combination of a chained
				1255	hash table with intrusive links (uniqued objects are required to inherit from
				1256	FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
				1257	its ID process.</p>
				1258
				1259	<p>Consider a case where you want to implement a "getOrCreateFoo" method for
				1260	a complex object (for example, a node in the code generator). The client has a
				1261	description of what it wants to generate (it knows the opcode and all the
				1262	operands), but we don't want to 'new' a node, then try inserting it into a set
				1263	only to find out it already exists, at which point we would have to delete it
				1264	and return the node that already exists.
				1265	</p>
				1266
				1267	<p>To support this style of client, FoldingSet perform a query with a
				1268	FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
				1269	element that we want to query for. The query either returns the element
				1270	matching the ID or it returns an opaque ID that indicates where insertion should
				1271	take place. Construction of the ID usually does not require heap traffic.</p>
				1272
				1273	<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
				1274	in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
				1275	Because the elements are individually allocated, pointers to the elements are
				1276	stable: inserting or removing elements does not invalidate any pointers to other
				1277	elements.
				1278	</p>
				1279
				1280	</div>
				1281
				1282	<!-- _______________________________________________________________________ -->
				1283	<div class="doc_subsubsection">
				1284	<a name="dss_set"><set></a>
				1285	</div>
				1286
				1287	<div class="doc_text">
				1288
				1289	<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
				1290	many things but great at nothing. std::set allocates memory for each element
				1291	inserted (thus it is very malloc intensive) and typically stores three pointers
				1292	per element in the set (thus adding a large amount of per-element space
				1293	overhead). It offers guaranteed log(n) performance, which is not particularly
				1294	fast from a complexity standpoint (particularly if the elements of the set are
				1295	expensive to compare, like strings), and has extremely high constant factors for
				1296	lookup, insertion and removal.</p>
				1297
				1298	<p>The advantages of std::set are that its iterators are stable (deleting or
				1299	inserting an element from the set does not affect iterators or pointers to other
				1300	elements) and that iteration over the set is guaranteed to be in sorted order.
				1301	If the elements in the set are large, then the relative overhead of the pointers
				1302	and malloc traffic is not a big deal, but if the elements of the set are small,
				1303	std::set is almost never a good choice.</p>
				1304
				1305	</div>
				1306
				1307	<!-- _______________________________________________________________________ -->
				1308	<div class="doc_subsubsection">
				1309	<a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
				1310	</div>
				1311
				1312	<div class="doc_text">
				1313	<p>LLVM's SetVector<Type> is an adapter class that combines your choice of
				1314	a set-like container along with a <a href="#ds_sequential">Sequential
				1315	Container</a>. The important property
				1316	that this provides is efficient insertion with uniquing (duplicate elements are
				1317	ignored) with iteration support. It implements this by inserting elements into
				1318	both a set-like container and the sequential container, using the set-like
				1319	container for uniquing and the sequential container for iteration.
				1320	</p>
				1321
				1322	<p>The difference between SetVector and other sets is that the order of
				1323	iteration is guaranteed to match the order of insertion into the SetVector.
				1324	This property is really important for things like sets of pointers. Because
				1325	pointer values are non-deterministic (e.g. vary across runs of the program on
				1326	different machines), iterating over the pointers in the set will
				1327	not be in a well-defined order.</p>
				1328
				1329	<p>
				1330	The drawback of SetVector is that it requires twice as much space as a normal
				1331	set and has the sum of constant factors from the set-like container and the
				1332	sequential container that it uses. Use it only if you need to iterate over
				1333	the elements in a deterministic order. SetVector is also expensive to delete
				1334	elements out of (linear time), unless you use it's "pop_back" method, which is
				1335	faster.
				1336	</p>
				1337
				1338	<p>SetVector is an adapter class that defaults to using std::vector and std::set
				1339	for the underlying containers, so it is quite expensive. However,
				1340	<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
				1341	defaults to using a SmallVector and SmallSet of a specified size. If you use
				1342	this, and if your sets are dynamically smaller than N, you will save a lot of
				1343	heap traffic.</p>
				1344
				1345	</div>
				1346
				1347	<!-- _______________________________________________________________________ -->
				1348	<div class="doc_subsubsection">
				1349	<a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
				1350	</div>
				1351
				1352	<div class="doc_text">
				1353
				1354	<p>
				1355	UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
				1356	retains a unique ID for each element inserted into the set. It internally
				1357	contains a map and a vector, and it assigns a unique ID for each value inserted
				1358	into the set.</p>
				1359
				1360	<p>UniqueVector is very expensive: its cost is the sum of the cost of
				1361	maintaining both the map and vector, it has high complexity, high constant
				1362	factors, and produces a lot of malloc traffic. It should be avoided.</p>
				1363
				1364	</div>
				1365
				1366
				1367	<!-- _______________________________________________________________________ -->
				1368	<div class="doc_subsubsection">
				1369	<a name="dss_otherset">Other Set-Like Container Options</a>
				1370	</div>
				1371
				1372	<div class="doc_text">
				1373
				1374	<p>
				1375	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]	1376	"hash_set" like containers (whether from C++ TR1 or from the SGI library). We
				1377	never use hash_set and unordered_set because they are generally very expensive
				1378	(each insertion requires a malloc) and very non-portable.
				1379	</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1380
				1381	<p>std::multiset is useful if you're not interested in elimination of
				1382	duplicates, but has all the drawbacks of std::set. A sorted vector (where you
				1383	don't delete duplicate entries) or some other approach is almost always
				1384	better.</p>
				1385
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1386	</div>
				1387
				1388	<!-- ======================================================================= -->
				1389	<div class="doc_subsection">
				1390	<a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
				1391	</div>
				1392
				1393	<div class="doc_text">
				1394	Map-like containers are useful when you want to associate data to a key. As
				1395	usual, there are a lot of different ways to do this. :)
				1396	</div>
				1397
				1398	<!-- _______________________________________________________________________ -->
				1399	<div class="doc_subsubsection">
				1400	<a name="dss_sortedvectormap">A sorted 'vector'</a>
				1401	</div>
				1402
				1403	<div class="doc_text">
				1404
				1405	<p>
				1406	If your usage pattern follows a strict insert-then-query approach, you can
				1407	trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
				1408	for set-like containers</a>. The only difference is that your query function
				1409	(which uses std::lower_bound to get efficient log(n) lookup) should only compare
				1410	the key, not both the key and value. This yields the same advantages as sorted
				1411	vectors for sets.
				1412	</p>
				1413	</div>
				1414
				1415	<!-- _______________________________________________________________________ -->
				1416	<div class="doc_subsubsection">
				1417	<a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
				1418	</div>
				1419
				1420	<div class="doc_text">
				1421
				1422	<p>
				1423	Strings are commonly used as keys in maps, and they are difficult to support
				1424	efficiently: they are variable length, inefficient to hash and compare when
				1425	long, expensive to copy, etc. StringMap is a specialized container designed to
				1426	cope with these issues. It supports mapping an arbitrary range of bytes to an
				1427	arbitrary other object.</p>
				1428
				1429	<p>The StringMap implementation uses a quadratically-probed hash table, where
				1430	the buckets store a pointer to the heap allocated entries (and some other
				1431	stuff). The entries in the map must be heap allocated because the strings are
				1432	variable length. The string data (key) and the element object (value) are
				1433	stored in the same allocation with the string data immediately after the element
				1434	object. This container guarantees the "<tt>(char*)(&Value+1)</tt>" points
				1435	to the key string for a value.</p>
				1436
				1437	<p>The StringMap is very fast for several reasons: quadratic probing is very
				1438	cache efficient for lookups, the hash value of strings in buckets is not
				1439	recomputed when lookup up an element, StringMap rarely has to touch the
				1440	memory for unrelated objects when looking up a value (even when hash collisions
				1441	happen), hash table growth does not recompute the hash values for strings
				1442	already in the table, and each pair in the map is store in a single allocation
				1443	(the string data is stored in the same allocation as the Value of a pair).</p>
				1444
				1445	<p>StringMap also provides query methods that take byte ranges, so it only ever
				1446	copies a string if a value is inserted into the table.</p>
				1447	</div>
				1448
				1449	<!-- _______________________________________________________________________ -->
				1450	<div class="doc_subsubsection">
				1451	<a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
				1452	</div>
				1453
				1454	<div class="doc_text">
				1455	<p>
				1456	IndexedMap is a specialized container for mapping small dense integers (or
				1457	values that can be mapped to small dense integers) to some other type. It is
				1458	internally implemented as a vector with a mapping function that maps the keys to
				1459	the dense integer range.
				1460	</p>
				1461
				1462	<p>
				1463	This is useful for cases like virtual registers in the LLVM code generator: they
				1464	have a dense mapping that is offset by a compile-time constant (the first
				1465	virtual register ID).</p>
				1466
				1467	</div>
				1468
				1469	<!-- _______________________________________________________________________ -->
				1470	<div class="doc_subsubsection">
				1471	<a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
				1472	</div>
				1473
				1474	<div class="doc_text">
				1475
				1476	<p>
				1477	DenseMap is a simple quadratically probed hash table. It excels at supporting
				1478	small keys and values: it uses a single allocation to hold all of the pairs that
				1479	are currently inserted in the map. DenseMap is a great way to map pointers to
				1480	pointers, or map other small types to each other.
				1481	</p>
				1482
				1483	<p>
				1484	There are several aspects of DenseMap that you should be aware of, however. The
				1485	iterators in a densemap are invalidated whenever an insertion occurs, unlike
				1486	map. Also, because DenseMap allocates space for a large number of key/value
				1487	pairs (it starts with 64 by default), it will waste a lot of space if your keys
				1488	or values are large. Finally, you must implement a partial specialization of
Chris Lattner	92eea07	2007-09-17 18:34:04 +0000	[diff] [blame]	1489	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]	1490	is required to tell DenseMap about two special marker values (which can never be
				1491	inserted into the map) that it needs internally.</p>
				1492
				1493	</div>
				1494
				1495	<!-- _______________________________________________________________________ -->
				1496	<div class="doc_subsubsection">
Jeffrey Yasskin	50e846d	2009-10-22 22:11:22 +0000	[diff] [blame]	1497	<a name="dss_valuemap">"llvm/ADT/ValueMap.h"</a>
				1498	</div>
				1499
				1500	<div class="doc_text">
				1501
				1502	<p>
				1503	ValueMap is a wrapper around a <a href="#dss_densemap">DenseMap</a> mapping
				1504	Value*s (or subclasses) to another type. When a Value is deleted or RAUW'ed,
				1505	ValueMap will update itself so the new version of the key is mapped to the same
				1506	value, just as if the key were a WeakVH. You can configure exactly how this
				1507	happens, and what else happens on these two events, by passing
				1508	a <code>Config</code> parameter to the ValueMap template.</p>
				1509
				1510	</div>
				1511
				1512	<!-- _______________________________________________________________________ -->
				1513	<div class="doc_subsubsection">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1514	<a name="dss_map"><map></a>
				1515	</div>
				1516
				1517	<div class="doc_text">
				1518
				1519	<p>
				1520	std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
				1521	a single allocation per pair inserted into the map, it offers log(n) lookup with
				1522	an extremely large constant factor, imposes a space penalty of 3 pointers per
				1523	pair in the map, etc.</p>
				1524
				1525	<p>std::map is most useful when your keys or values are very large, if you need
				1526	to iterate over the collection in sorted order, or if you need stable iterators
				1527	into the map (i.e. they don't get invalidated if an insertion or deletion of
				1528	another element takes place).</p>
				1529
				1530	</div>
				1531
				1532	<!-- _______________________________________________________________________ -->
				1533	<div class="doc_subsubsection">
				1534	<a name="dss_othermap">Other Map-Like Container Options</a>
				1535	</div>
				1536
				1537	<div class="doc_text">
				1538
				1539	<p>
				1540	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]	1541	"hash_map" like containers (whether from C++ TR1 or from the SGI library). We
				1542	never use hash_set and unordered_set because they are generally very expensive
				1543	(each insertion requires a malloc) and very non-portable.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1544
				1545	<p>std::multimap is useful if you want to map a key to multiple values, but has
				1546	all the drawbacks of std::map. A sorted vector or some other approach is almost
				1547	always better.</p>
				1548
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1549	</div>
				1550
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1551	<!-- ======================================================================= -->
				1552	<div class="doc_subsection">
Chris Lattner	d8b95b7	2009-07-25 07:22:20 +0000	[diff] [blame]	1553	<a name="ds_string">String-like containers</a>
				1554	</div>
				1555
				1556	<div class="doc_text">
				1557
				1558	<p>
				1559	TODO: const char* vs stringref vs smallstring vs std::string. Describe twine,
				1560	xref to #string_apis.
				1561	</p>
				1562
				1563	</div>
				1564
				1565	<!-- ======================================================================= -->
				1566	<div class="doc_subsection">
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1567	<a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
				1568	</div>
				1569
				1570	<div class="doc_text">
Chris Lattner	62a4eae	2007-09-25 22:37:50 +0000	[diff] [blame]	1571	<p>Unlike the other containers, there are only two bit storage containers, and
				1572	choosing when to use each is relatively straightforward.</p>
				1573
				1574	<p>One additional option is
				1575	<tt>std::vector<bool></tt>: we discourage its use for two reasons 1) the
				1576	implementation in many common compilers (e.g. commonly available versions of
				1577	GCC) is extremely inefficient and 2) the C++ standards committee is likely to
				1578	deprecate this container and/or change it significantly somehow. In any case,
				1579	please don't use it.</p>
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1580	</div>
				1581
				1582	<!-- _______________________________________________________________________ -->
				1583	<div class="doc_subsubsection">
				1584	<a name="dss_bitvector">BitVector</a>
				1585	</div>
				1586
				1587	<div class="doc_text">
Dan Gohman	c83a2b4	2010-01-05 18:24:00 +0000	[diff] [blame^]	1588	<p> The BitVector container provides a dynamic size set of bits for manipulation.
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1589	It supports individual bit setting/testing, as well as set operations. The set
				1590	operations take time O(size of bitvector), but operations are performed one word
				1591	at a time, instead of one bit at a time. This makes the BitVector very fast for
				1592	set operations compared to other containers. Use the BitVector when you expect
				1593	the number of set bits to be high (IE a dense set).
				1594	</p>
				1595	</div>
				1596
				1597	<!-- _______________________________________________________________________ -->
				1598	<div class="doc_subsubsection">
Dan Gohman	c83a2b4	2010-01-05 18:24:00 +0000	[diff] [blame^]	1599	<a name="dss_smallbitvector">SmallBitVector</a>
				1600	</div>
				1601
				1602	<div class="doc_text">
				1603	<p> The SmallBitVector container provides the same interface as BitVector, but
				1604	it is optimized for the case where only a small number of bits, less than
				1605	25 or so, are needed. It also transparently supports larger bit counts, but
				1606	slightly less efficiently than a plain BitVector, so SmallBitVector should
				1607	only be used when larger counts are rare.
				1608	</p>
				1609
				1610	<p>
				1611	At this time, SmallBitVector does not support set operations (and, or, xor),
				1612	and its operator[] does not provide an assignable lvalue.
				1613	</p>
				1614	</div>
				1615
				1616	<!-- _______________________________________________________________________ -->
				1617	<div class="doc_subsubsection">
Daniel Berlin	7ea44dc	2007-09-24 17:52:25 +0000	[diff] [blame]	1618	<a name="dss_sparsebitvector">SparseBitVector</a>
				1619	</div>
				1620
				1621	<div class="doc_text">
				1622	<p> The SparseBitVector container is much like BitVector, with one major
				1623	difference: Only the bits that are set, are stored. This makes the
				1624	SparseBitVector much more space efficient than BitVector when the set is sparse,
				1625	as well as making set operations O(number of set bits) instead of O(size of
				1626	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
				1627	(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).
				1628	</p>
				1629	</div>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1630
				1631	<!-- *********************************************************************** -->
				1632	<div class="doc_section">
				1633	<a name="common">Helpful Hints for Common Operations</a>
				1634	</div>
				1635	<!-- *********************************************************************** -->
				1636
				1637	<div class="doc_text">
				1638
				1639	<p>This section describes how to perform some very simple transformations of
				1640	LLVM code. This is meant to give examples of common idioms used, showing the
				1641	practical side of LLVM transformations. <p> Because this is a "how-to" section,
				1642	you should also read about the main classes that you will be working with. The
				1643	<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
				1644	and descriptions of the main classes that you should know about.</p>
				1645
				1646	</div>
				1647
				1648	<!-- NOTE: this section should be heavy on example code -->
				1649	<!-- ======================================================================= -->
				1650	<div class="doc_subsection">
				1651	<a name="inspection">Basic Inspection and Traversal Routines</a>
				1652	</div>
				1653
				1654	<div class="doc_text">
				1655
				1656	<p>The LLVM compiler infrastructure have many different data structures that may
				1657	be traversed. Following the example of the C++ standard template library, the
				1658	techniques used to traverse these various data structures are all basically the
				1659	same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
				1660	method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
				1661	function returns an iterator pointing to one past the last valid element of the
				1662	sequence, and there is some <tt>XXXiterator</tt> data type that is common
				1663	between the two operations.</p>
				1664
				1665	<p>Because the pattern for iteration is common across many different aspects of
				1666	the program representation, the standard template library algorithms may be used
				1667	on them, and it is easier to remember how to iterate. First we show a few common
				1668	examples of the data structures that need to be traversed. Other data
				1669	structures are traversed in very similar ways.</p>
				1670
				1671	</div>
				1672
				1673	<!-- _______________________________________________________________________ -->
				1674	<div class="doc_subsubsection">
				1675	<a name="iterate_function">Iterating over the </a><a
				1676	href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
				1677	href="#Function"><tt>Function</tt></a>
				1678	</div>
				1679
				1680	<div class="doc_text">
				1681
				1682	<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
				1683	transform in some way; in particular, you'd like to manipulate its
				1684	<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
				1685	the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
				1686	an example that prints the name of a <tt>BasicBlock</tt> and the number of
				1687	<tt>Instruction</tt>s it contains:</p>
				1688
				1689	<div class="doc_code">
				1690	<pre>
				1691	// <i>func is a pointer to a Function instance</i>
				1692	for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i)
				1693	// <i>Print out the name of the basic block if it has one, and then the</i>
				1694	// <i>number of instructions that it contains</i>
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1695	errs() << "Basic block (name=" << i->getName() << ") has "
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1696	<< i->size() << " instructions.\n";
				1697	</pre>
				1698	</div>
				1699
				1700	<p>Note that i can be used as if it were a pointer for the purposes of
				1701	invoking member functions of the <tt>Instruction</tt> class. This is
				1702	because the indirection operator is overloaded for the iterator
				1703	classes. In the above code, the expression <tt>i->size()</tt> is
				1704	exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
				1705
				1706	</div>
				1707
				1708	<!-- _______________________________________________________________________ -->
				1709	<div class="doc_subsubsection">
				1710	<a name="iterate_basicblock">Iterating over the </a><a
				1711	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1712	href="#BasicBlock"><tt>BasicBlock</tt></a>
				1713	</div>
				1714
				1715	<div class="doc_text">
				1716
				1717	<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
				1718	easy to iterate over the individual instructions that make up
				1719	<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
				1720	a <tt>BasicBlock</tt>:</p>
				1721
				1722	<div class="doc_code">
				1723	<pre>
				1724	// <i>blk is a pointer to a BasicBlock instance</i>
				1725	for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
				1726	// <i>The next statement works since operator<<(ostream&,...)</i>
				1727	// <i>is overloaded for Instruction&</i>
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1728	errs() << *i << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1729	</pre>
				1730	</div>
				1731
				1732	<p>However, this isn't really the best way to print out the contents of a
				1733	<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
				1734	anything you'll care about, you could have just invoked the print routine on the
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1735	basic block itself: <tt>errs() << *blk << "\n";</tt>.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1736
				1737	</div>
				1738
				1739	<!-- _______________________________________________________________________ -->
				1740	<div class="doc_subsubsection">
				1741	<a name="iterate_institer">Iterating over the </a><a
				1742	href="#Instruction"><tt>Instruction</tt></a>s in a <a
				1743	href="#Function"><tt>Function</tt></a>
				1744	</div>
				1745
				1746	<div class="doc_text">
				1747
				1748	<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
				1749	<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
				1750	<tt>InstIterator</tt> should be used instead. You'll need to include <a
				1751	href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
				1752	and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
				1753	small example that shows how to dump all instructions in a function to the standard error stream:<p>
				1754
				1755	<div class="doc_code">
				1756	<pre>
				1757	#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
				1758
				1759	// <i>F is a pointer to a Function instance</i>
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1760	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1761	errs() << *I << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1762	</pre>
				1763	</div>
				1764
				1765	<p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
				1766	work list with its initial contents. For example, if you wanted to
				1767	initialize a work list to contain all instructions in a <tt>Function</tt>
				1768	F, all you would need to do is something like:</p>
				1769
				1770	<div class="doc_code">
				1771	<pre>
				1772	std::set<Instruction*> worklist;
Chris Lattner	fd62ee7	2008-06-04 18:20:42 +0000	[diff] [blame]	1773	// or better yet, SmallPtrSet<Instruction*, 64> worklist;
				1774
				1775	for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
				1776	worklist.insert(&*I);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1777	</pre>
				1778	</div>
				1779
				1780	<p>The STL set <tt>worklist</tt> would now contain all instructions in the
				1781	<tt>Function</tt> pointed to by F.</p>
				1782
				1783	</div>
				1784
				1785	<!-- _______________________________________________________________________ -->
				1786	<div class="doc_subsubsection">
				1787	<a name="iterate_convert">Turning an iterator into a class pointer (and
				1788	vice-versa)</a>
				1789	</div>
				1790
				1791	<div class="doc_text">
				1792
				1793	<p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
				1794	instance when all you've got at hand is an iterator. Well, extracting
				1795	a reference or a pointer from an iterator is very straight-forward.
				1796	Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
				1797	is a <tt>BasicBlock::const_iterator</tt>:</p>
				1798
				1799	<div class="doc_code">
				1800	<pre>
				1801	Instruction& inst = *i; // <i>Grab reference to instruction reference</i>
				1802	Instruction* pinst = &*i; // <i>Grab pointer to instruction reference</i>
				1803	const Instruction& inst = *j;
				1804	</pre>
				1805	</div>
				1806
				1807	<p>However, the iterators you'll be working with in the LLVM framework are
				1808	special: they will automatically convert to a ptr-to-instance type whenever they
				1809	need to. Instead of dereferencing the iterator and then taking the address of
				1810	the result, you can simply assign the iterator to the proper pointer type and
				1811	you get the dereference and address-of operation as a result of the assignment
				1812	(behind the scenes, this is a result of overloading casting mechanisms). Thus
				1813	the last line of the last example,</p>
				1814
				1815	<div class="doc_code">
				1816	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1817	Instruction pinst = &i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1818	</pre>
				1819	</div>
				1820
				1821	<p>is semantically equivalent to</p>
				1822
				1823	<div class="doc_code">
				1824	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1825	Instruction *pinst = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1826	</pre>
				1827	</div>
				1828
				1829	<p>It's also possible to turn a class pointer into the corresponding iterator,
				1830	and this is a constant time operation (very efficient). The following code
				1831	snippet illustrates use of the conversion constructors provided by LLVM
				1832	iterators. By using these, you can explicitly grab the iterator of something
				1833	without actually obtaining it via iteration over some structure:</p>
				1834
				1835	<div class="doc_code">
				1836	<pre>
				1837	void printNextInstruction(Instruction* inst) {
				1838	BasicBlock::iterator it(inst);
				1839	++it; // <i>After this line, it refers to the instruction after *inst</i>
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1840	if (it != inst->getParent()->end()) errs() << *it << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1841	}
				1842	</pre>
				1843	</div>
				1844
				1845	</div>
				1846
				1847	<!--_______________________________________________________________________-->
				1848	<div class="doc_subsubsection">
				1849	<a name="iterate_complex">Finding call sites: a slightly more complex
				1850	example</a>
				1851	</div>
				1852
				1853	<div class="doc_text">
				1854
				1855	<p>Say that you're writing a FunctionPass and would like to count all the
				1856	locations in the entire module (that is, across every <tt>Function</tt>) where a
				1857	certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
				1858	learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
				1859	much more straight-forward manner, but this example will allow us to explore how
				1860	you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
				1861	is what we want to do:</p>
				1862
				1863	<div class="doc_code">
				1864	<pre>
				1865	initialize callCounter to zero
				1866	for each Function f in the Module
				1867	for each BasicBlock b in f
				1868	for each Instruction i in b
				1869	if (i is a CallInst and calls the given function)
				1870	increment callCounter
				1871	</pre>
				1872	</div>
				1873
				1874	<p>And the actual code is (remember, because we're writing a
				1875	<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
				1876	override the <tt>runOnFunction</tt> method):</p>
				1877
				1878	<div class="doc_code">
				1879	<pre>
				1880	Function* targetFunc = ...;
				1881
				1882	class OurFunctionPass : public FunctionPass {
				1883	public:
				1884	OurFunctionPass(): callCounter(0) { }
				1885
				1886	virtual runOnFunction(Function& F) {
				1887	for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
Eric Christopher	5fbecf7	2008-11-08 08:20:49 +0000	[diff] [blame]	1888	for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) {
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1889	if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
				1890	href="#CallInst">CallInst</a>>(&*i)) {
				1891	// <i>We know we've encountered a call instruction, so we</i>
				1892	// <i>need to determine if it's a call to the</i>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1893	// <i>function pointed to by m_func or not.</i>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1894	if (callInst->getCalledFunction() == targetFunc)
				1895	++callCounter;
				1896	}
				1897	}
				1898	}
				1899	}
				1900
				1901	private:
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1902	unsigned callCounter;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1903	};
				1904	</pre>
				1905	</div>
				1906
				1907	</div>
				1908
				1909	<!--_______________________________________________________________________-->
				1910	<div class="doc_subsubsection">
				1911	<a name="calls_and_invokes">Treating calls and invokes the same way</a>
				1912	</div>
				1913
				1914	<div class="doc_text">
				1915
				1916	<p>You may have noticed that the previous example was a bit oversimplified in
				1917	that it did not deal with call sites generated by 'invoke' instructions. In
				1918	this, and in other situations, you may find that you want to treat
				1919	<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
				1920	most-specific common base class is <tt>Instruction</tt>, which includes lots of
				1921	less closely-related things. For these cases, LLVM provides a handy wrapper
				1922	class called <a
				1923	href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
				1924	It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
				1925	methods that provide functionality common to <tt>CallInst</tt>s and
				1926	<tt>InvokeInst</tt>s.</p>
				1927
				1928	<p>This class has "value semantics": it should be passed by value, not by
				1929	reference and it should not be dynamically allocated or deallocated using
				1930	<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
				1931	assignable and constructable, with costs equivalents to that of a bare pointer.
				1932	If you look at its definition, it has only a single pointer member.</p>
				1933
				1934	</div>
				1935
				1936	<!--_______________________________________________________________________-->
				1937	<div class="doc_subsubsection">
				1938	<a name="iterate_chains">Iterating over def-use & use-def chains</a>
				1939	</div>
				1940
				1941	<div class="doc_text">
				1942
				1943	<p>Frequently, we might have an instance of the <a
				1944	href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
				1945	determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
				1946	<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
				1947	For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
				1948	particular function <tt>foo</tt>. Finding all of the instructions that
				1949	<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
				1950	of <tt>F</tt>:</p>
				1951
				1952	<div class="doc_code">
				1953	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1954	Function *F = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1955
				1956	for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i)
				1957	if (Instruction Inst = dyn_cast<Instruction>(i)) {
Chris Lattner	599c392	2009-09-08 05:15:50 +0000	[diff] [blame]	1958	errs() << "F is used in instruction:\n";
				1959	errs() << *Inst << "\n";
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1960	}
				1961	</pre>
				1962	</div>
				1963
				1964	<p>Alternately, it's common to have an instance of the <a
				1965	href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
				1966	<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
				1967	<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
				1968	<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
				1969	all of the values that a particular instruction uses (that is, the operands of
				1970	the particular <tt>Instruction</tt>):</p>
				1971
				1972	<div class="doc_code">
				1973	<pre>
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1974	Instruction *pi = ...;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1975
				1976	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]	1977	Value v = i;
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	1978	// <i>...</i>
				1979	}
				1980	</pre>
				1981	</div>
				1982
				1983	<!--
				1984	def-use chains ("finding all users of"): Value::use_begin/use_end
				1985	use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
				1986	-->
				1987
				1988	</div>
				1989
Chris Lattner	0665e1f	2008-01-03 16:56:04 +0000	[diff] [blame]	1990	<!--_______________________________________________________________________-->
				1991	<div class="doc_subsubsection">
				1992	<a name="iterate_preds">Iterating over predecessors &
				1993	successors of blocks</a>
				1994	</div>
				1995
				1996	<div class="doc_text">
				1997
				1998	<p>Iterating over the predecessors and successors of a block is quite easy
				1999	with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
				2000	this to iterate over all predecessors of BB:</p>
				2001
				2002	<div class="doc_code">
				2003	<pre>
				2004	#include "llvm/Support/CFG.h"
				2005	BasicBlock *BB = ...;
				2006
				2007	for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
				2008	BasicBlock Pred = PI;
				2009	// <i>...</i>
				2010	}
				2011	</pre>
				2012	</div>
				2013
				2014	<p>Similarly, to iterate over successors use
				2015	succ_iterator/succ_begin/succ_end.</p>
				2016
				2017	</div>
				2018
				2019
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2020	<!-- ======================================================================= -->
				2021	<div class="doc_subsection">
				2022	<a name="simplechanges">Making simple changes</a>
				2023	</div>
				2024
				2025	<div class="doc_text">
				2026
				2027	<p>There are some primitive transformation operations present in the LLVM
				2028	infrastructure that are worth knowing about. When performing
				2029	transformations, it's fairly common to manipulate the contents of basic
				2030	blocks. This section describes some of the common methods for doing so
				2031	and gives example code.</p>
				2032
				2033	</div>
				2034
				2035	<!--_______________________________________________________________________-->
				2036	<div class="doc_subsubsection">
				2037	<a name="schanges_creating">Creating and inserting new
				2038	<tt>Instruction</tt>s</a>
				2039	</div>
				2040
				2041	<div class="doc_text">
				2042
				2043	<p><i>Instantiating Instructions</i></p>
				2044
				2045	<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
				2046	constructor for the kind of instruction to instantiate and provide the necessary
				2047	parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
				2048	(const-ptr-to) <tt>Type</tt>. Thus:</p>
				2049
				2050	<div class="doc_code">
				2051	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2052	AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2053	</pre>
				2054	</div>
				2055
				2056	<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
				2057	one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
				2058	subclass is likely to have varying default parameters which change the semantics
				2059	of the instruction, so refer to the <a
				2060	href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
				2061	Instruction</a> that you're interested in instantiating.</p>
				2062
				2063	<p><i>Naming values</i></p>
				2064
				2065	<p>It is very useful to name the values of instructions when you're able to, as
				2066	this facilitates the debugging of your transformations. If you end up looking
				2067	at generated LLVM machine code, you definitely want to have logical names
				2068	associated with the results of instructions! By supplying a value for the
				2069	<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
				2070	associate a logical name with the result of the instruction's execution at
				2071	run time. For example, say that I'm writing a transformation that dynamically
				2072	allocates space for an integer on the stack, and that integer is going to be
				2073	used as some kind of index by some other code. To accomplish this, I place an
				2074	<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
				2075	<tt>Function</tt>, and I'm intending to use it within the same
				2076	<tt>Function</tt>. I might do:</p>
				2077
				2078	<div class="doc_code">
				2079	<pre>
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2080	AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2081	</pre>
				2082	</div>
				2083
				2084	<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
				2085	execution value, which is a pointer to an integer on the run time stack.</p>
				2086
				2087	<p><i>Inserting instructions</i></p>
				2088
				2089	<p>There are essentially two ways to insert an <tt>Instruction</tt>
				2090	into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
				2091
				2092	<ul>
				2093	<li>Insertion into an explicit instruction list
				2094
				2095	<p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
				2096	<tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
				2097	before <tt>*pi</tt>, we do the following: </p>
				2098
				2099	<div class="doc_code">
				2100	<pre>
				2101	BasicBlock *pb = ...;
				2102	Instruction *pi = ...;
				2103	Instruction *newInst = new Instruction(...);
				2104
				2105	pb->getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
				2106	</pre>
				2107	</div>
				2108
				2109	<p>Appending to the end of a <tt>BasicBlock</tt> is so common that
				2110	the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
				2111	classes provide constructors which take a pointer to a
				2112	<tt>BasicBlock</tt> to be appended to. For example code that
				2113	looked like: </p>
				2114
				2115	<div class="doc_code">
				2116	<pre>
				2117	BasicBlock *pb = ...;
				2118	Instruction *newInst = new Instruction(...);
				2119
				2120	pb->getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
				2121	</pre>
				2122	</div>
				2123
				2124	<p>becomes: </p>
				2125
				2126	<div class="doc_code">
				2127	<pre>
				2128	BasicBlock *pb = ...;
				2129	Instruction *newInst = new Instruction(..., pb);
				2130	</pre>
				2131	</div>
				2132
				2133	<p>which is much cleaner, especially if you are creating
				2134	long instruction streams.</p></li>
				2135
				2136	<li>Insertion into an implicit instruction list
				2137
				2138	<p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
				2139	are implicitly associated with an existing instruction list: the instruction
				2140	list of the enclosing basic block. Thus, we could have accomplished the same
				2141	thing as the above code without being given a <tt>BasicBlock</tt> by doing:
				2142	</p>
				2143
				2144	<div class="doc_code">
				2145	<pre>
				2146	Instruction *pi = ...;
				2147	Instruction *newInst = new Instruction(...);
				2148
				2149	pi->getParent()->getInstList().insert(pi, newInst);
				2150	</pre>
				2151	</div>
				2152
				2153	<p>In fact, this sequence of steps occurs so frequently that the
				2154	<tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
				2155	constructors which take (as a default parameter) a pointer to an
				2156	<tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
				2157	precede. That is, <tt>Instruction</tt> constructors are capable of
				2158	inserting the newly-created instance into the <tt>BasicBlock</tt> of a
				2159	provided instruction, immediately before that instruction. Using an
				2160	<tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
				2161	parameter, the above code becomes:</p>
				2162
				2163	<div class="doc_code">
				2164	<pre>
				2165	Instruction* pi = ...;
				2166	Instruction* newInst = new Instruction(..., pi);
				2167	</pre>
				2168	</div>
				2169
				2170	<p>which is much cleaner, especially if you're creating a lot of
				2171	instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
				2172	</ul>
				2173
				2174	</div>
				2175
				2176	<!--_______________________________________________________________________-->
				2177	<div class="doc_subsubsection">
				2178	<a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
				2179	</div>
				2180
				2181	<div class="doc_text">
				2182
				2183	<p>Deleting an instruction from an existing sequence of instructions that form a
				2184	<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
				2185	you must have a pointer to the instruction that you wish to delete. Second, you
				2186	need to obtain the pointer to that instruction's basic block. You use the
				2187	pointer to the basic block to get its list of instructions and then use the
				2188	erase function to remove your instruction. For example:</p>
				2189
				2190	<div class="doc_code">
				2191	<pre>
				2192	<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner	3db8f77	2008-02-15 22:57:17 +0000	[diff] [blame]	2193	I->eraseFromParent();
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2194	</pre>
				2195	</div>
				2196
				2197	</div>
				2198
				2199	<!--_______________________________________________________________________-->
				2200	<div class="doc_subsubsection">
				2201	<a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
				2202	<tt>Value</tt></a>
				2203	</div>
				2204
				2205	<div class="doc_text">
				2206
				2207	<p><i>Replacing individual instructions</i></p>
				2208
				2209	<p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
				2210	permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
				2211	and <tt>ReplaceInstWithInst</tt>.</p>
				2212
				2213	<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
				2214
				2215	<ul>
				2216	<li><tt>ReplaceInstWithValue</tt>
				2217
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	2218	<p>This function replaces all uses of a given instruction with a value,
				2219	and then removes the original instruction. The following example
				2220	illustrates the replacement of the result of a particular
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2221	<tt>AllocaInst</tt> that allocates memory for a single integer with a null
				2222	pointer to an integer.</p>
				2223
				2224	<div class="doc_code">
				2225	<pre>
				2226	AllocaInst* instToReplace = ...;
				2227	BasicBlock::iterator ii(instToReplace);
				2228
				2229	ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	2230	Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2231	</pre></div></li>
				2232
				2233	<li><tt>ReplaceInstWithInst</tt>
				2234
				2235	<p>This function replaces a particular instruction with another
Nick Lewycky	48d4b03	2008-09-15 06:31:52 +0000	[diff] [blame]	2236	instruction, inserting the new instruction into the basic block at the
				2237	location where the old instruction was, and replacing any uses of the old
				2238	instruction with the new instruction. The following example illustrates
				2239	the replacement of one <tt>AllocaInst</tt> with another.</p>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2240
				2241	<div class="doc_code">
				2242	<pre>
				2243	AllocaInst* instToReplace = ...;
				2244	BasicBlock::iterator ii(instToReplace);
				2245
				2246	ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2247	new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2248	</pre></div></li>
				2249	</ul>
				2250
				2251	<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
				2252
				2253	<p>You can use <tt>Value::replaceAllUsesWith</tt> and
				2254	<tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
				2255	doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
				2256	and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
				2257	information.</p>
				2258
				2259	<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
				2260	include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
				2261	ReplaceInstWithValue, ReplaceInstWithInst -->
				2262
				2263	</div>
				2264
				2265	<!--_______________________________________________________________________-->
				2266	<div class="doc_subsubsection">
				2267	<a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
				2268	</div>
				2269
				2270	<div class="doc_text">
				2271
				2272	<p>Deleting a global variable from a module is just as easy as deleting an
				2273	Instruction. First, you must have a pointer to the global variable that you wish
				2274	to delete. You use this pointer to erase it from its parent, the module.
				2275	For example:</p>
				2276
				2277	<div class="doc_code">
				2278	<pre>
				2279	<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
				2280
				2281	GV->eraseFromParent();
				2282	</pre>
				2283	</div>
				2284
				2285	</div>
				2286
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2287	<!-- ======================================================================= -->
				2288	<div class="doc_subsection">
				2289	<a name="create_types">How to Create Types</a>
				2290	</div>
				2291
				2292	<div class="doc_text">
				2293
				2294	<p>In generating IR, you may need some complex types. If you know these types
Misha Brukman	0672513	2009-05-01 20:40:51 +0000	[diff] [blame]	2295	statically, you can use <tt>TypeBuilder<...>::get()</tt>, defined
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2296	in <tt>llvm/Support/TypeBuilder.h</tt>, to retrieve them. <tt>TypeBuilder</tt>
				2297	has two forms depending on whether you're building types for cross-compilation
Misha Brukman	0672513	2009-05-01 20:40:51 +0000	[diff] [blame]	2298	or native library use. <tt>TypeBuilder<T, true></tt> requires
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2299	that <tt>T</tt> be independent of the host environment, meaning that it's built
				2300	out of types from
				2301	the <a href="/doxygen/namespacellvm_1_1types.html"><tt>llvm::types</tt></a>
				2302	namespace and pointers, functions, arrays, etc. built of
Misha Brukman	0672513	2009-05-01 20:40:51 +0000	[diff] [blame]	2303	those. <tt>TypeBuilder<T, false></tt> additionally allows native C types
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2304	whose size may depend on the host compiler. For example,</p>
				2305
				2306	<div class="doc_code">
				2307	<pre>
Misha Brukman	0672513	2009-05-01 20:40:51 +0000	[diff] [blame]	2308	FunctionType ft = TypeBuilder<types::i<8>(types::i<32>), true>::get();
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2309	</pre>
				2310	</div>
				2311
				2312	<p>is easier to read and write than the equivalent</p>
				2313
				2314	<div class="doc_code">
				2315	<pre>
Owen Anderson	60f8757	2009-06-16 17:40:28 +0000	[diff] [blame]	2316	std::vector<const Type*> params;
Jeffrey Yasskin	7ebb6ac	2009-04-30 22:33:41 +0000	[diff] [blame]	2317	params.push_back(PointerType::getUnqual(Type::Int32Ty));
				2318	FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false);
				2319	</pre>
				2320	</div>
				2321
				2322	<p>See the <a href="/doxygen/TypeBuilder_8h-source.html#l00001">class
				2323	comment</a> for more details.</p>
				2324
				2325	</div>
				2326
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2327	<!-- *********************************************************************** -->
				2328	<div class="doc_section">
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2329	<a name="threading">Threads and LLVM</a>
				2330	</div>
				2331	<!-- *********************************************************************** -->
				2332
				2333	<div class="doc_text">
				2334	<p>
				2335	This section describes the interaction of the LLVM APIs with multithreading,
				2336	both on the part of client applications, and in the JIT, in the hosted
				2337	application.
				2338	</p>
				2339
				2340	<p>
				2341	Note that LLVM's support for multithreading is still relatively young. Up
				2342	through version 2.5, the execution of threaded hosted applications was
				2343	supported, but not threaded client access to the APIs. While this use case is
				2344	now supported, clients <em>must</em> adhere to the guidelines specified below to
				2345	ensure proper operation in multithreaded mode.
				2346	</p>
				2347
				2348	<p>
				2349	Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic
				2350	intrinsics in order to support threaded operation. If you need a
				2351	multhreading-capable LLVM on a platform without a suitably modern system
				2352	compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and
				2353	using the resultant compiler to build a copy of LLVM with multithreading
				2354	support.
				2355	</p>
				2356	</div>
				2357
				2358	<!-- ======================================================================= -->
				2359	<div class="doc_subsection">
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	2360	<a name="startmultithreaded">Entering and Exiting Multithreaded Mode</a>
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2361	</div>
				2362
				2363	<div class="doc_text">
				2364
				2365	<p>
				2366	In order to properly protect its internal data structures while avoiding
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	2367	excessive locking overhead in the single-threaded case, the LLVM must intialize
				2368	certain data structures necessary to provide guards around its internals. To do
				2369	so, the client program must invoke <tt>llvm_start_multithreaded()</tt> before
				2370	making any concurrent LLVM API calls. To subsequently tear down these
				2371	structures, use the <tt>llvm_stop_multithreaded()</tt> call. You can also use
				2372	the <tt>llvm_is_multithreaded()</tt> call to check the status of multithreaded
				2373	mode.
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2374	</p>
				2375
				2376	<p>
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	2377	Note that both of these calls must be made <em>in isolation</em>. That is to
				2378	say that no other LLVM API calls may be executing at any time during the
				2379	execution of <tt>llvm_start_multithreaded()</tt> or <tt>llvm_stop_multithreaded
				2380	</tt>. It's is the client's responsibility to enforce this isolation.
				2381	</p>
				2382
				2383	<p>
				2384	The return value of <tt>llvm_start_multithreaded()</tt> indicates the success or
				2385	failure of the initialization. Failure typically indicates that your copy of
				2386	LLVM was built without multithreading support, typically because GCC atomic
				2387	intrinsics were not found in your system compiler. In this case, the LLVM API
				2388	will not be safe for concurrent calls. However, it <em>will</em> be safe for
				2389	hosting threaded applications in the JIT, though care must be taken to ensure
				2390	that side exits and the like do not accidentally result in concurrent LLVM API
				2391	calls.
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2392	</p>
				2393	</div>
				2394
				2395	<!-- ======================================================================= -->
				2396	<div class="doc_subsection">
				2397	<a name="shutdown">Ending Execution with <tt>llvm_shutdown()</tt></a>
				2398	</div>
				2399
				2400	<div class="doc_text">
				2401	<p>
				2402	When you are done using the LLVM APIs, you should call <tt>llvm_shutdown()</tt>
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	2403	to deallocate memory used for internal structures. This will also invoke
				2404	<tt>llvm_stop_multithreaded()</tt> if LLVM is operating in multithreaded mode.
				2405	As such, <tt>llvm_shutdown()</tt> requires the same isolation guarantees as
				2406	<tt>llvm_stop_multithreaded()</tt>.
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2407	</p>
				2408
				2409	<p>
				2410	Note that, if you use scope-based shutdown, you can use the
				2411	<tt>llvm_shutdown_obj</tt> class, which calls <tt>llvm_shutdown()</tt> in its
				2412	destructor.
				2413	</div>
				2414
				2415	<!-- ======================================================================= -->
				2416	<div class="doc_subsection">
				2417	<a name="managedstatic">Lazy Initialization with <tt>ManagedStatic</tt></a>
				2418	</div>
				2419
				2420	<div class="doc_text">
				2421	<p>
				2422	<tt>ManagedStatic</tt> is a utility class in LLVM used to implement static
				2423	initialization of static resources, such as the global type tables. Before the
				2424	invocation of <tt>llvm_shutdown()</tt>, it implements a simple lazy
				2425	initialization scheme. Once <tt>llvm_start_multithreaded()</tt> returns,
				2426	however, it uses double-checked locking to implement thread-safe lazy
				2427	initialization.
				2428	</p>
				2429
				2430	<p>
				2431	Note that, because no other threads are allowed to issue LLVM API calls before
				2432	<tt>llvm_start_multithreaded()</tt> returns, it is possible to have
				2433	<tt>ManagedStatic</tt>s of <tt>llvm::sys::Mutex</tt>s.
				2434	</p>
Owen Anderson	b418690	2009-06-16 18:04:19 +0000	[diff] [blame]	2435
				2436	<p>
				2437	The <tt>llvm_acquire_global_lock()</tt> and <tt>llvm_release_global_lock</tt>
				2438	APIs provide access to the global lock used to implement the double-checked
				2439	locking for lazy initialization. These should only be used internally to LLVM,
				2440	and only if you know what you're doing!
				2441	</p>
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2442	</div>
				2443
Owen Anderson	3f2da0e	2009-08-19 17:58:52 +0000	[diff] [blame]	2444	<!-- ======================================================================= -->
				2445	<div class="doc_subsection">
				2446	<a name="llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a>
				2447	</div>
				2448
				2449	<div class="doc_text">
				2450	<p>
				2451	<tt>LLVMContext</tt> is an opaque class in the LLVM API which clients can use
				2452	to operate multiple, isolated instances of LLVM concurrently within the same
				2453	address space. For instance, in a hypothetical compile-server, the compilation
				2454	of an individual translation unit is conceptually independent from all the
				2455	others, and it would be desirable to be able to compile incoming translation
				2456	units concurrently on independent server threads. Fortunately,
				2457	<tt>LLVMContext</tt> exists to enable just this kind of scenario!
				2458	</p>
				2459
				2460	<p>
				2461	Conceptually, <tt>LLVMContext</tt> provides isolation. Every LLVM entity
				2462	(<tt>Module</tt>s, <tt>Value</tt>s, <tt>Type</tt>s, <tt>Constant</tt>s, etc.)
Chris Lattner	9ca98ce	2009-08-20 03:10:14 +0000	[diff] [blame]	2463	in LLVM's in-memory IR belongs to an <tt>LLVMContext</tt>. Entities in
Owen Anderson	3f2da0e	2009-08-19 17:58:52 +0000	[diff] [blame]	2464	different contexts <em>cannot</em> interact with each other: <tt>Module</tt>s in
				2465	different contexts cannot be linked together, <tt>Function</tt>s cannot be added
				2466	to <tt>Module</tt>s in different contexts, etc. What this means is that is is
				2467	safe to compile on multiple threads simultaneously, as long as no two threads
				2468	operate on entities within the same context.
				2469	</p>
				2470
				2471	<p>
				2472	In practice, very few places in the API require the explicit specification of a
				2473	<tt>LLVMContext</tt>, other than the <tt>Type</tt> creation/lookup APIs.
				2474	Because every <tt>Type</tt> carries a reference to its owning context, most
				2475	other entities can determine what context they belong to by looking at their
				2476	own <tt>Type</tt>. If you are adding new entities to LLVM IR, please try to
				2477	maintain this interface design.
				2478	</p>
				2479
				2480	<p>
				2481	For clients that do <em>not</em> require the benefits of isolation, LLVM
				2482	provides a convenience API <tt>getGlobalContext()</tt>. This returns a global,
				2483	lazily initialized <tt>LLVMContext</tt> that may be used in situations where
				2484	isolation is not a concern.
				2485	</p>
				2486	</div>
				2487
Owen Anderson	e8370c0	2009-06-16 01:17:16 +0000	[diff] [blame]	2488	<!-- *********************************************************************** -->
				2489	<div class="doc_section">
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2490	<a name="advanced">Advanced Topics</a>
				2491	</div>
				2492	<!-- *********************************************************************** -->
				2493
				2494	<div class="doc_text">
				2495	<p>
				2496	This section describes some of the advanced or obscure API's that most clients
				2497	do not need to be aware of. These API's tend manage the inner workings of the
				2498	LLVM system, and only need to be accessed in unusual circumstances.
				2499	</p>
				2500	</div>
				2501
				2502	<!-- ======================================================================= -->
				2503	<div class="doc_subsection">
				2504	<a name="TypeResolve">LLVM Type Resolution</a>
				2505	</div>
				2506
				2507	<div class="doc_text">
				2508
				2509	<p>
				2510	The LLVM type system has a very simple goal: allow clients to compare types for
				2511	structural equality with a simple pointer comparison (aka a shallow compare).
				2512	This goal makes clients much simpler and faster, and is used throughout the LLVM
				2513	system.
				2514	</p>
				2515
				2516	<p>
				2517	Unfortunately achieving this goal is not a simple matter. In particular,
				2518	recursive types and late resolution of opaque types makes the situation very
				2519	difficult to handle. Fortunately, for the most part, our implementation makes
				2520	most clients able to be completely unaware of the nasty internal details. The
				2521	primary case where clients are exposed to the inner workings of it are when
				2522	building a recursive type. In addition to this case, the LLVM bitcode reader,
				2523	assembly parser, and linker also have to be aware of the inner workings of this
				2524	system.
				2525	</p>
				2526
				2527	<p>
				2528	For our purposes below, we need three concepts. First, an "Opaque Type" is
				2529	exactly as defined in the <a href="LangRef.html#t_opaque">language
				2530	reference</a>. Second an "Abstract Type" is any type which includes an
				2531	opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
				2532	Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
				2533	float }</tt>").
				2534	</p>
				2535
				2536	</div>
				2537
				2538	<!-- ______________________________________________________________________ -->
				2539	<div class="doc_subsubsection">
				2540	<a name="BuildRecType">Basic Recursive Type Construction</a>
				2541	</div>
				2542
				2543	<div class="doc_text">
				2544
				2545	<p>
				2546	Because the most common question is "how do I build a recursive type with LLVM",
				2547	we answer it now and explain it as we go. Here we include enough to cause this
				2548	to be emitted to an output .ll file:
				2549	</p>
				2550
				2551	<div class="doc_code">
				2552	<pre>
				2553	%mylist = type { %mylist*, i32 }
				2554	</pre>
				2555	</div>
				2556
				2557	<p>
				2558	To build this, use the following LLVM APIs:
				2559	</p>
				2560
				2561	<div class="doc_code">
				2562	<pre>
				2563	// <i>Create the initial outer struct</i>
				2564	<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
				2565	std::vector<const Type*> Elts;
Daniel Dunbar	8ce7962	2008-10-03 22:17:25 +0000	[diff] [blame]	2566	Elts.push_back(PointerType::getUnqual(StructTy));
Nick Lewycky	c004e5b	2007-12-03 01:52:52 +0000	[diff] [blame]	2567	Elts.push_back(Type::Int32Ty);
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2568	StructType *NewSTy = StructType::get(Elts);
				2569
				2570	// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
				2571	// <i>the struct and the opaque type are actually the same.</i>
				2572	cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
				2573
				2574	// <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
				2575	// <i>kept up-to-date</i>
				2576	NewSTy = cast<StructType>(StructTy.get());
				2577
				2578	// <i>Add a name for the type to the module symbol table (optional)</i>
				2579	MyModule->addTypeName("mylist", NewSTy);
				2580	</pre>
				2581	</div>
				2582
				2583	<p>
				2584	This code shows the basic approach used to build recursive types: build a
				2585	non-recursive type using 'opaque', then use type unification to close the cycle.
				2586	The type unification step is performed by the <tt><a
				2587	href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
				2588	described next. After that, we describe the <a
				2589	href="#PATypeHolder">PATypeHolder class</a>.
				2590	</p>
				2591
				2592	</div>
				2593
				2594	<!-- ______________________________________________________________________ -->
				2595	<div class="doc_subsubsection">
				2596	<a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
				2597	</div>
				2598
				2599	<div class="doc_text">
				2600	<p>
				2601	The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
				2602	While this method is actually a member of the DerivedType class, it is most
				2603	often used on OpaqueType instances. Type unification is actually a recursive
				2604	process. After unification, types can become structurally isomorphic to
				2605	existing types, and all duplicates are deleted (to preserve pointer equality).
				2606	</p>
				2607
				2608	<p>
				2609	In the example above, the OpaqueType object is definitely deleted.
				2610	Additionally, if there is an "{ \2*, i32}" type already created in the system,
				2611	the pointer and struct type created are <b>also</b> deleted. Obviously whenever
				2612	a type is deleted, any "Type*" pointers in the program are invalidated. As
				2613	such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
				2614	live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
				2615	types can never move or be deleted). To deal with this, the <a
				2616	href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
				2617	reference to a possibly refined type, and the <a
				2618	href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
				2619	complex datastructures.
				2620	</p>
				2621
				2622	</div>
				2623
				2624	<!-- ______________________________________________________________________ -->
				2625	<div class="doc_subsubsection">
				2626	<a name="PATypeHolder">The PATypeHolder Class</a>
				2627	</div>
				2628
				2629	<div class="doc_text">
				2630	<p>
				2631	PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
				2632	happily goes about nuking types that become isomorphic to existing types, it
				2633	automatically updates all PATypeHolder objects to point to the new type. In the
				2634	example above, this allows the code to maintain a pointer to the resultant
				2635	resolved recursive type, even though the Type*'s are potentially invalidated.
				2636	</p>
				2637
				2638	<p>
				2639	PATypeHolder is an extremely light-weight object that uses a lazy union-find
				2640	implementation to update pointers. For example the pointer from a Value to its
				2641	Type is maintained by PATypeHolder objects.
				2642	</p>
				2643
				2644	</div>
				2645
				2646	<!-- ______________________________________________________________________ -->
				2647	<div class="doc_subsubsection">
				2648	<a name="AbstractTypeUser">The AbstractTypeUser Class</a>
				2649	</div>
				2650
				2651	<div class="doc_text">
				2652
				2653	<p>
				2654	Some data structures need more to perform more complex updates when types get
				2655	resolved. To support this, a class can derive from the AbstractTypeUser class.
				2656	This class
				2657	allows it to get callbacks when certain types are resolved. To register to get
				2658	callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
				2659	methods can be called on a type. Note that these methods only work for <i>
				2660	abstract</i> types. Concrete types (those that do not include any opaque
				2661	objects) can never be refined.
				2662	</p>
				2663	</div>
				2664
				2665
				2666	<!-- ======================================================================= -->
				2667	<div class="doc_subsection">
				2668	<a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
				2669	<tt>TypeSymbolTable</tt> classes</a>
				2670	</div>
				2671
				2672	<div class="doc_text">
				2673	<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
				2674	ValueSymbolTable</a></tt> class provides a symbol table that the <a
				2675	href="#Function"><tt>Function</tt></a> and <a href="#Module">
				2676	<tt>Module</tt></a> classes use for naming value definitions. The symbol table
				2677	can provide a name for any <a href="#Value"><tt>Value</tt></a>.
				2678	The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
				2679	TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
				2680	names for types.</p>
				2681
				2682	<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
				2683	by most clients. It should only be used when iteration over the symbol table
				2684	names themselves are required, which is very special purpose. Note that not
				2685	all LLVM
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2686	<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]	2687	an empty name) do not exist in the symbol table.
				2688	</p>
				2689
				2690	<p>These symbol tables support iteration over the values/types in the symbol
				2691	table with <tt>begin/end/iterator</tt> and supports querying to see if a
				2692	specific name is in the symbol table (with <tt>lookup</tt>). The
				2693	<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
				2694	simply call <tt>setName</tt> on a value, which will autoinsert it into the
				2695	appropriate symbol table. For types, use the Module::addTypeName method to
				2696	insert entries into the symbol table.</p>
				2697
				2698	</div>
				2699
				2700
				2701
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2702	<!-- ======================================================================= -->
				2703	<div class="doc_subsection">
				2704	<a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
				2705	</div>
				2706
				2707	<div class="doc_text">
				2708	<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]	2709	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]	2710	towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
				2711	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]	2712	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]	2713	addition and removal.</p>
				2714
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2715	<!-- ______________________________________________________________________ -->
				2716	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2717	<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]	2718	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2719
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2720	<div class="doc_text">
				2721	<p>
				2722	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]	2723	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]	2724	(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
				2725	that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
				2726	</p>
				2727	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2728
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2729	<p>
				2730	We have 2 different layouts in the <tt>User</tt> (sub)classes:
				2731	<ul>
				2732	<li><p>Layout a)
				2733	The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
				2734	object and there are a fixed number of them.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2735
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2736	<li><p>Layout b)
				2737	The <tt>Use</tt> object(s) are referenced by a pointer to an
				2738	array from the <tt>User</tt> object and there may be a variable
				2739	number of them.</p>
				2740	</ul>
				2741	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2742	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]	2743	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]	2744	we stick to this redundancy for the sake of simplicity.
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2745	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]	2746	has. (Theoretically this information can also be calculated
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2747	given the scheme presented below.)</p>
				2748	<p>
				2749	Special forms of allocation operators (<tt>operator new</tt>)
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2750	enforce the following memory layouts:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2751
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2752	<ul>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2753	<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]	2754
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2755	<pre>
				2756	...---.---.---.---.-------...
				2757	\| P \| P \| P \| P \| User
				2758	'''---'---'---'---'-------'''
				2759	</pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2760
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2761	<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]	2762	<pre>
				2763	.-------...
				2764	\| User
				2765	'-------'''
				2766	\|
				2767	v
				2768	.---.---.---.---...
				2769	\| P \| P \| P \| P \|
				2770	'---'---'---'---'''
				2771	</pre>
				2772	</ul>
				2773	<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
				2774	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]	2775
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2776	<!-- ______________________________________________________________________ -->
				2777	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2778	<a name="Waymarking">The waymarking algorithm</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2779	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2780
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2781	<div class="doc_text">
				2782	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2783	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]	2784	their <tt>User</tt> objects, there must be a fast and exact method to
				2785	recover it. This is accomplished by the following scheme:</p>
				2786	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2787
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2788	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]	2789	start of the <tt>User</tt> object:
				2790	<ul>
				2791	<li><tt>00</tt> —> binary digit 0</li>
				2792	<li><tt>01</tt> —> binary digit 1</li>
				2793	<li><tt>10</tt> —> stop and calculate (<tt>s</tt>)</li>
				2794	<li><tt>11</tt> —> full stop (<tt>S</tt>)</li>
				2795	</ul>
				2796	<p>
				2797	Given a <tt>Use*</tt>, all we have to do is to walk till we get
				2798	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]	2799	we have to walk to the next stop picking up digits
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2800	and calculating the offset:</p>
				2801	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2802	.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
				2803	\| 1 \| s \| 1 \| 0 \| 1 \| 0 \| s \| 1 \| 1 \| 0 \| s \| 1 \| 1 \| s \| 1 \| S \| User (or User*)
				2804	'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
				2805	\|+15 \|+10 \|+6 \|+3 \|+1
				2806	\| \| \| \| \|__>
				2807	\| \| \| \|__________>
				2808	\| \| \|______________________>
				2809	\| \|______________________________________>
				2810	\|__________________________________________________________>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2811	</pre>
				2812	<p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2813	Only the significant number of bits need to be stored between the
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2814	stops, so that the <i>worst case is 20 memory accesses</i> when there are
				2815	1000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2816
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2817	<!-- ______________________________________________________________________ -->
				2818	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2819	<a name="ReferenceImpl">Reference implementation</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2820	</div>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2821
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2822	<div class="doc_text">
				2823	<p>
				2824	The following literate Haskell fragment demonstrates the concept:</p>
				2825	</div>
				2826
				2827	<div class="doc_code">
				2828	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2829	> import Test.QuickCheck
				2830	>
				2831	> digits :: Int -> [Char] -> [Char]
				2832	> digits 0 acc = '0' : acc
				2833	> digits 1 acc = '1' : acc
				2834	> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
				2835	>
				2836	> dist :: Int -> [Char] -> [Char]
				2837	> dist 0 [] = ['S']
				2838	> dist 0 acc = acc
				2839	> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
				2840	> dist n acc = dist (n - 1) $ dist 1 acc
				2841	>
				2842	> takeLast n ss = reverse $ take n $ reverse ss
				2843	>
				2844	> test = takeLast 40 $ dist 20 []
				2845	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2846	</pre>
				2847	</div>
				2848	<p>
				2849	Printing <test> gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
				2850	<p>
				2851	The reverse algorithm computes the length of the string just by examining
				2852	a certain prefix:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2853
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2854	<div class="doc_code">
				2855	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2856	> pref :: [Char] -> Int
				2857	> pref "S" = 1
				2858	> pref ('s':'1':rest) = decode 2 1 rest
				2859	> pref (_:rest) = 1 + pref rest
				2860	>
				2861	> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
				2862	> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
				2863	> decode walk acc _ = walk + acc
				2864	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2865	</pre>
				2866	</div>
				2867	<p>
				2868	Now, as expected, printing <pref test> gives <tt>40</tt>.</p>
				2869	<p>
				2870	We can <i>quickCheck</i> this with following property:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2871
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2872	<div class="doc_code">
				2873	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2874	> testcase = dist 2000 []
				2875	> testcaseLength = length testcase
				2876	>
				2877	> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
				2878	> where arr = takeLast n testcase
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2879	>
				2880	</pre>
				2881	</div>
				2882	<p>
				2883	As expected <quickCheck identityProp> gives:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2884
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2885	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2886	*Main> quickCheck identityProp
				2887	OK, passed 100 tests.
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2888	</pre>
				2889	<p>
				2890	Let's be a bit more exhaustive:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2891
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2892	<div class="doc_code">
				2893	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2894	>
				2895	> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
				2896	>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2897	</pre>
				2898	</div>
				2899	<p>
				2900	And here is the result of <deepCheck identityProp>:</p>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2901
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2902	<pre>
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2903	*Main> deepCheck identityProp
				2904	OK, passed 500 tests.
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2905	</pre>
				2906
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2907	<!-- ______________________________________________________________________ -->
				2908	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2909	<a name="Tagging">Tagging considerations</a>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2910	</div>
				2911
				2912	<p>
				2913	To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
				2914	never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
				2915	new <tt>Use**</tt> on every modification. Accordingly getters must strip the
				2916	tag bits.</p>
				2917	<p>
Gabor Greif	e247e36	2008-06-25 00:10:22 +0000	[diff] [blame]	2918	For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
				2919	Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
				2920	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]	2921	the LSBit set. (Portability is relying on the fact that all known compilers place the
				2922	<tt>vptr</tt> in the first word of the instances.)</p>
Gabor Greif	93b462b	2008-06-18 13:44:57 +0000	[diff] [blame]	2923
Gabor Greif	92e8776	2008-06-16 21:06:12 +0000	[diff] [blame]	2924	</div>
				2925
				2926	<!-- *********************************************************************** -->
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2927	<div class="doc_section">
				2928	<a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
				2929	</div>
				2930	<!-- *********************************************************************** -->
				2931
				2932	<div class="doc_text">
				2933	<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
				2934	<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
				2935
				2936	<p>The Core LLVM classes are the primary means of representing the program
				2937	being inspected or transformed. The core LLVM classes are defined in
				2938	header files in the <tt>include/llvm/</tt> directory, and implemented in
				2939	the <tt>lib/VMCore</tt> directory.</p>
				2940
				2941	</div>
				2942
				2943	<!-- ======================================================================= -->
				2944	<div class="doc_subsection">
				2945	<a name="Type">The <tt>Type</tt> class and Derived Types</a>
				2946	</div>
				2947
				2948	<div class="doc_text">
				2949
				2950	<p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
				2951	a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
				2952	through its subclasses. Certain primitive types (<tt>VoidType</tt>,
				2953	<tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
				2954	subclasses. They are hidden because they offer no useful functionality beyond
				2955	what the <tt>Type</tt> class offers except to distinguish themselves from
				2956	other subclasses of <tt>Type</tt>.</p>
				2957	<p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
				2958	named, but this is not a requirement. There exists exactly
				2959	one instance of a given shape at any one time. This allows type equality to
				2960	be performed with address equality of the Type Instance. That is, given two
				2961	<tt>Type*</tt> values, the types are identical if the pointers are identical.
				2962	</p>
				2963	</div>
				2964
				2965	<!-- _______________________________________________________________________ -->
				2966	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2967	<a name="m_Type">Important Public Methods</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2968	</div>
				2969
				2970	<div class="doc_text">
				2971
				2972	<ul>
				2973	<li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
				2974
				2975	<li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
				2976	floating point types.</li>
				2977
				2978	<li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
				2979	an OpaqueType anywhere in its definition).</li>
				2980
				2981	<li><tt>bool isSized()</tt>: Return true if the type has known size. Things
				2982	that don't have a size are abstract types, labels and void.</li>
				2983
				2984	</ul>
				2985	</div>
				2986
				2987	<!-- _______________________________________________________________________ -->
				2988	<div class="doc_subsubsection">
Gabor Greif	50626fc	2009-01-05 16:05:32 +0000	[diff] [blame]	2989	<a name="derivedtypes">Important Derived Types</a>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	2990	</div>
				2991	<div class="doc_text">
				2992	<dl>
				2993	<dt><tt>IntegerType</tt></dt>
				2994	<dd>Subclass of DerivedType that represents integer types of any bit width.
				2995	Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
				2996	<tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
				2997	<ul>
				2998	<li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
				2999	type of a specific bit width.</li>
				3000	<li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
				3001	type.</li>
				3002	</ul>
				3003	</dd>
				3004	<dt><tt>SequentialType</tt></dt>
				3005	<dd>This is subclassed by ArrayType and PointerType
				3006	<ul>
				3007	<li><tt>const Type * getElementType() const</tt>: Returns the type of each
				3008	of the elements in the sequential type. </li>
				3009	</ul>
				3010	</dd>
				3011	<dt><tt>ArrayType</tt></dt>
				3012	<dd>This is a subclass of SequentialType and defines the interface for array
				3013	types.
				3014	<ul>
				3015	<li><tt>unsigned getNumElements() const</tt>: Returns the number of
				3016	elements in the array. </li>
				3017	</ul>
				3018	</dd>
				3019	<dt><tt>PointerType</tt></dt>
				3020	<dd>Subclass of SequentialType for pointer types.</dd>
				3021	<dt><tt>VectorType</tt></dt>
				3022	<dd>Subclass of SequentialType for vector types. A
				3023	vector type is similar to an ArrayType but is distinguished because it is
Benjamin Kramer	5fb9d7e	2009-10-12 14:46:08 +0000	[diff] [blame]	3024	a first class type whereas ArrayType is not. Vector types are used for
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3025	vector operations and are usually small vectors of of an integer or floating
				3026	point type.</dd>
				3027	<dt><tt>StructType</tt></dt>
				3028	<dd>Subclass of DerivedTypes for struct types.</dd>
				3029	<dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
				3030	<dd>Subclass of DerivedTypes for function types.
				3031	<ul>
				3032	<li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
				3033	function</li>
				3034	<li><tt> const Type * getReturnType() const</tt>: Returns the
				3035	return type of the function.</li>
				3036	<li><tt>const Type * getParamType (unsigned i)</tt>: Returns
				3037	the type of the ith parameter.</li>
				3038	<li><tt> const unsigned getNumParams() const</tt>: Returns the
				3039	number of formal parameters.</li>
				3040	</ul>
				3041	</dd>
				3042	<dt><tt>OpaqueType</tt></dt>
				3043	<dd>Sublcass of DerivedType for abstract types. This class
				3044	defines no content and is used as a placeholder for some other type. Note
				3045	that OpaqueType is used (temporarily) during type resolution for forward
				3046	references of types. Once the referenced type is resolved, the OpaqueType
				3047	is replaced with the actual type. OpaqueType can also be used for data
				3048	abstraction. At link time opaque types can be resolved to actual types
				3049	of the same name.</dd>
				3050	</dl>
				3051	</div>
				3052
				3053
				3054
				3055	<!-- ======================================================================= -->
				3056	<div class="doc_subsection">
				3057	<a name="Module">The <tt>Module</tt> class</a>
				3058	</div>
				3059
				3060	<div class="doc_text">
				3061
				3062	<p><tt>#include "<a
				3063	href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
				3064	<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
				3065
				3066	<p>The <tt>Module</tt> class represents the top level structure present in LLVM
				3067	programs. An LLVM module is effectively either a translation unit of the
				3068	original program or a combination of several translation units merged by the
				3069	linker. The <tt>Module</tt> class keeps track of a list of <a
				3070	href="#Function"><tt>Function</tt></a>s, a list of <a
				3071	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
				3072	href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
				3073	helpful member functions that try to make common operations easy.</p>
				3074
				3075	</div>
				3076
				3077	<!-- _______________________________________________________________________ -->
				3078	<div class="doc_subsubsection">
				3079	<a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
				3080	</div>
				3081
				3082	<div class="doc_text">
				3083
				3084	<ul>
				3085	<li><tt>Module::Module(std::string name = "")</tt></li>
				3086	</ul>
				3087
				3088	<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
				3089	provide a name for it (probably based on the name of the translation unit).</p>
				3090
				3091	<ul>
				3092	<li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
				3093	<tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
				3094
				3095	<tt>begin()</tt>, <tt>end()</tt>
				3096	<tt>size()</tt>, <tt>empty()</tt>
				3097
				3098	<p>These are forwarding methods that make it easy to access the contents of
				3099	a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
				3100	list.</p></li>
				3101
				3102	<li><tt>Module::FunctionListType &getFunctionList()</tt>
				3103
				3104	<p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
				3105	necessary to use when you need to update the list or perform a complex
				3106	action that doesn't have a forwarding method.</p>
				3107
				3108	<p><!-- Global Variable --></p></li>
				3109	</ul>
				3110
				3111	<hr>
				3112
				3113	<ul>
				3114	<li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
				3115
				3116	<tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
				3117
				3118	<tt>global_begin()</tt>, <tt>global_end()</tt>
				3119	<tt>global_size()</tt>, <tt>global_empty()</tt>
				3120
				3121	<p> These are forwarding methods that make it easy to access the contents of
				3122	a <tt>Module</tt> object's <a
				3123	href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
				3124
				3125	<li><tt>Module::GlobalListType &getGlobalList()</tt>
				3126
				3127	<p>Returns the list of <a
				3128	href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
				3129	use when you need to update the list or perform a complex action that
				3130	doesn't have a forwarding method.</p>
				3131
				3132	<p><!-- Symbol table stuff --> </p></li>
				3133	</ul>
				3134
				3135	<hr>
				3136
				3137	<ul>
				3138	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				3139
				3140	<p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				3141	for this <tt>Module</tt>.</p>
				3142
				3143	<p><!-- Convenience methods --></p></li>
				3144	</ul>
				3145
				3146	<hr>
				3147
				3148	<ul>
				3149	<li><tt><a href="#Function">Function</a> *getFunction(const std::string
				3150	&Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
				3151
				3152	<p>Look up the specified function in the <tt>Module</tt> <a
				3153	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
				3154	<tt>null</tt>.</p></li>
				3155
				3156	<li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
				3157	std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
				3158
				3159	<p>Look up the specified function in the <tt>Module</tt> <a
				3160	href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
				3161	external declaration for the function and return it.</p></li>
				3162
				3163	<li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
				3164
				3165	<p>If there is at least one entry in the <a
				3166	href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
				3167	href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
				3168	string.</p></li>
				3169
				3170	<li><tt>bool addTypeName(const std::string &Name, const <a
				3171	href="#Type">Type</a> *Ty)</tt>
				3172
				3173	<p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				3174	mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
				3175	name, true is returned and the <a
				3176	href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
				3177	</ul>
				3178
				3179	</div>
				3180
				3181
				3182	<!-- ======================================================================= -->
				3183	<div class="doc_subsection">
				3184	<a name="Value">The <tt>Value</tt> class</a>
				3185	</div>
				3186
				3187	<div class="doc_text">
				3188
				3189	<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
				3190	<br>
				3191	doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
				3192
				3193	<p>The <tt>Value</tt> class is the most important class in the LLVM Source
				3194	base. It represents a typed value that may be used (among other things) as an
				3195	operand to an instruction. There are many different types of <tt>Value</tt>s,
				3196	such as <a href="#Constant"><tt>Constant</tt></a>s,<a
				3197	href="#Argument"><tt>Argument</tt></a>s. Even <a
				3198	href="#Instruction"><tt>Instruction</tt></a>s and <a
				3199	href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
				3200
				3201	<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
				3202	for a program. For example, an incoming argument to a function (represented
				3203	with an instance of the <a href="#Argument">Argument</a> class) is "used" by
				3204	every instruction in the function that references the argument. To keep track
				3205	of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
				3206	href="#User"><tt>User</tt></a>s that is using it (the <a
				3207	href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
				3208	graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
				3209	def-use information in the program, and is accessible through the <tt>use_</tt>*
				3210	methods, shown below.</p>
				3211
				3212	<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
				3213	and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
				3214	method. In addition, all LLVM values can be named. The "name" of the
				3215	<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
				3216
				3217	<div class="doc_code">
				3218	<pre>
				3219	%<b>foo</b> = add i32 1, 2
				3220	</pre>
				3221	</div>
				3222
				3223	<p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
				3224	that the name of any value may be missing (an empty string), so names should
				3225	<b>ONLY</b> be used for debugging (making the source code easier to read,
				3226	debugging printouts), they should not be used to keep track of values or map
				3227	between them. For this purpose, use a <tt>std::map</tt> of pointers to the
				3228	<tt>Value</tt> itself instead.</p>
				3229
				3230	<p>One important aspect of LLVM is that there is no distinction between an SSA
				3231	variable and the operation that produces it. Because of this, any reference to
				3232	the value produced by an instruction (or the value available as an incoming
				3233	argument, for example) is represented as a direct pointer to the instance of
				3234	the class that
				3235	represents this value. Although this may take some getting used to, it
				3236	simplifies the representation and makes it easier to manipulate.</p>
				3237
				3238	</div>
				3239
				3240	<!-- _______________________________________________________________________ -->
				3241	<div class="doc_subsubsection">
				3242	<a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
				3243	</div>
				3244
				3245	<div class="doc_text">
				3246
				3247	<ul>
				3248	<li><tt>Value::use_iterator</tt> - Typedef for iterator over the
				3249	use-list<br>
				3250	<tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
				3251	the use-list<br>
				3252	<tt>unsigned use_size()</tt> - Returns the number of users of the
				3253	value.<br>
				3254	<tt>bool use_empty()</tt> - Returns true if there are no users.<br>
				3255	<tt>use_iterator use_begin()</tt> - Get an iterator to the start of
				3256	the use-list.<br>
				3257	<tt>use_iterator use_end()</tt> - Get an iterator to the end of the
				3258	use-list.<br>
				3259	<tt><a href="#User">User</a> *use_back()</tt> - Returns the last
				3260	element in the list.
				3261	<p> These methods are the interface to access the def-use
				3262	information in LLVM. As with all other iterators in LLVM, the naming
				3263	conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
				3264	</li>
				3265	<li><tt><a href="#Type">Type</a> *getType() const</tt>
				3266	<p>This method returns the Type of the Value.</p>
				3267	</li>
				3268	<li><tt>bool hasName() const</tt><br>
				3269	<tt>std::string getName() const</tt><br>
				3270	<tt>void setName(const std::string &Name)</tt>
				3271	<p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
				3272	be aware of the <a href="#nameWarning">precaution above</a>.</p>
				3273	</li>
				3274	<li><tt>void replaceAllUsesWith(Value *V)</tt>
				3275
				3276	<p>This method traverses the use list of a <tt>Value</tt> changing all <a
				3277	href="#User"><tt>User</tt>s</a> of the current value to refer to
				3278	"<tt>V</tt>" instead. For example, if you detect that an instruction always
				3279	produces a constant value (for example through constant folding), you can
				3280	replace all uses of the instruction with the constant like this:</p>
				3281
				3282	<div class="doc_code">
				3283	<pre>
				3284	Inst->replaceAllUsesWith(ConstVal);
				3285	</pre>
				3286	</div>
				3287
				3288	</ul>
				3289
				3290	</div>
				3291
				3292	<!-- ======================================================================= -->
				3293	<div class="doc_subsection">
				3294	<a name="User">The <tt>User</tt> class</a>
				3295	</div>
				3296
				3297	<div class="doc_text">
				3298
				3299	<p>
				3300	<tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
				3301	doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
				3302	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				3303
				3304	<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
				3305	refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
				3306	that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
				3307	referring to. The <tt>User</tt> class itself is a subclass of
				3308	<tt>Value</tt>.</p>
				3309
				3310	<p>The operands of a <tt>User</tt> point directly to the LLVM <a
				3311	href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
				3312	Single Assignment (SSA) form, there can only be one definition referred to,
				3313	allowing this direct connection. This connection provides the use-def
				3314	information in LLVM.</p>
				3315
				3316	</div>
				3317
				3318	<!-- _______________________________________________________________________ -->
				3319	<div class="doc_subsubsection">
				3320	<a name="m_User">Important Public Members of the <tt>User</tt> class</a>
				3321	</div>
				3322
				3323	<div class="doc_text">
				3324
				3325	<p>The <tt>User</tt> class exposes the operand list in two ways: through
				3326	an index access interface and through an iterator based interface.</p>
				3327
				3328	<ul>
				3329	<li><tt>Value *getOperand(unsigned i)</tt><br>
				3330	<tt>unsigned getNumOperands()</tt>
				3331	<p> These two methods expose the operands of the <tt>User</tt> in a
				3332	convenient form for direct access.</p></li>
				3333
				3334	<li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
				3335	list<br>
				3336	<tt>op_iterator op_begin()</tt> - Get an iterator to the start of
				3337	the operand list.<br>
				3338	<tt>op_iterator op_end()</tt> - Get an iterator to the end of the
				3339	operand list.
				3340	<p> Together, these methods make up the iterator based interface to
				3341	the operands of a <tt>User</tt>.</p></li>
				3342	</ul>
				3343
				3344	</div>
				3345
				3346	<!-- ======================================================================= -->
				3347	<div class="doc_subsection">
				3348	<a name="Instruction">The <tt>Instruction</tt> class</a>
				3349	</div>
				3350
				3351	<div class="doc_text">
				3352
				3353	<p><tt>#include "</tt><tt><a
				3354	href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
				3355	doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
				3356	Superclasses: <a href="#User"><tt>User</tt></a>, <a
				3357	href="#Value"><tt>Value</tt></a></p>
				3358
				3359	<p>The <tt>Instruction</tt> class is the common base class for all LLVM
				3360	instructions. It provides only a few methods, but is a very commonly used
				3361	class. The primary data tracked by the <tt>Instruction</tt> class itself is the
				3362	opcode (instruction type) and the parent <a
				3363	href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
				3364	into. To represent a specific type of instruction, one of many subclasses of
				3365	<tt>Instruction</tt> are used.</p>
				3366
				3367	<p> Because the <tt>Instruction</tt> class subclasses the <a
				3368	href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
				3369	way as for other <a href="#User"><tt>User</tt></a>s (with the
				3370	<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
				3371	<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
				3372	the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
				3373	file contains some meta-data about the various different types of instructions
				3374	in LLVM. It describes the enum values that are used as opcodes (for example
				3375	<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
				3376	concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
				3377	example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
				3378	href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
				3379	this file confuses doxygen, so these enum values don't show up correctly in the
				3380	<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
				3381
				3382	</div>
				3383
				3384	<!-- _______________________________________________________________________ -->
				3385	<div class="doc_subsubsection">
				3386	<a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
				3387	class</a>
				3388	</div>
				3389	<div class="doc_text">
				3390	<ul>
				3391	<li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
				3392	<p>This subclasses represents all two operand instructions whose operands
				3393	must be the same type, except for the comparison instructions.</p></li>
				3394	<li><tt><a name="CastInst">CastInst</a></tt>
				3395	<p>This subclass is the parent of the 12 casting instructions. It provides
				3396	common operations on cast instructions.</p>
				3397	<li><tt><a name="CmpInst">CmpInst</a></tt>
				3398	<p>This subclass respresents the two comparison instructions,
				3399	<a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
				3400	<a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
				3401	<li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
				3402	<p>This subclass is the parent of all terminator instructions (those which
				3403	can terminate a block).</p>
				3404	</ul>
				3405	</div>
				3406
				3407	<!-- _______________________________________________________________________ -->
				3408	<div class="doc_subsubsection">
				3409	<a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
				3410	class</a>
				3411	</div>
				3412
				3413	<div class="doc_text">
				3414
				3415	<ul>
				3416	<li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
				3417	<p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
				3418	this <tt>Instruction</tt> is embedded into.</p></li>
				3419	<li><tt>bool mayWriteToMemory()</tt>
				3420	<p>Returns true if the instruction writes to memory, i.e. it is a
				3421	<tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
				3422	<li><tt>unsigned getOpcode()</tt>
				3423	<p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
				3424	<li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
				3425	<p>Returns another instance of the specified instruction, identical
				3426	in all ways to the original except that the instruction has no parent
				3427	(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
				3428	and it has no name</p></li>
				3429	</ul>
				3430
				3431	</div>
				3432
				3433	<!-- ======================================================================= -->
				3434	<div class="doc_subsection">
				3435	<a name="Constant">The <tt>Constant</tt> class and subclasses</a>
				3436	</div>
				3437
				3438	<div class="doc_text">
				3439
				3440	<p>Constant represents a base class for different types of constants. It
				3441	is subclassed by ConstantInt, ConstantArray, etc. for representing
				3442	the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
				3443	a subclass, which represents the address of a global variable or function.
				3444	</p>
				3445
				3446	</div>
				3447
				3448	<!-- _______________________________________________________________________ -->
				3449	<div class="doc_subsubsection">Important Subclasses of Constant </div>
				3450	<div class="doc_text">
				3451	<ul>
				3452	<li>ConstantInt : This subclass of Constant represents an integer constant of
				3453	any width.
				3454	<ul>
				3455	<li><tt>const APInt& getValue() const</tt>: Returns the underlying
				3456	value of this constant, an APInt value.</li>
				3457	<li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
				3458	value to an int64_t via sign extension. If the value (not the bit width)
				3459	of the APInt is too large to fit in an int64_t, an assertion will result.
				3460	For this reason, use of this method is discouraged.</li>
				3461	<li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
				3462	value to a uint64_t via zero extension. IF the value (not the bit width)
				3463	of the APInt is too large to fit in a uint64_t, an assertion will result.
				3464	For this reason, use of this method is discouraged.</li>
				3465	<li><tt>static ConstantInt* get(const APInt& Val)</tt>: Returns the
				3466	ConstantInt object that represents the value provided by <tt>Val</tt>.
				3467	The type is implied as the IntegerType that corresponds to the bit width
				3468	of <tt>Val</tt>.</li>
				3469	<li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
				3470	Returns the ConstantInt object that represents the value provided by
				3471	<tt>Val</tt> for integer type <tt>Ty</tt>.</li>
				3472	</ul>
				3473	</li>
				3474	<li>ConstantFP : This class represents a floating point constant.
				3475	<ul>
				3476	<li><tt>double getValue() const</tt>: Returns the underlying value of
				3477	this constant. </li>
				3478	</ul>
				3479	</li>
				3480	<li>ConstantArray : This represents a constant array.
				3481	<ul>
				3482	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3483	a vector of component constants that makeup this array. </li>
				3484	</ul>
				3485	</li>
				3486	<li>ConstantStruct : This represents a constant struct.
				3487	<ul>
				3488	<li><tt>const std::vector<Use> &getValues() const</tt>: Returns
				3489	a vector of component constants that makeup this array. </li>
				3490	</ul>
				3491	</li>
				3492	<li>GlobalValue : This represents either a global variable or a function. In
				3493	either case, the value is a constant fixed address (after linking).
				3494	</li>
				3495	</ul>
				3496	</div>
				3497
				3498
				3499	<!-- ======================================================================= -->
				3500	<div class="doc_subsection">
				3501	<a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
				3502	</div>
				3503
				3504	<div class="doc_text">
				3505
				3506	<p><tt>#include "<a
				3507	href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
				3508	doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
				3509	Class</a><br>
				3510	Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
				3511	<a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
				3512
				3513	<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
				3514	href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
				3515	visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
				3516	Because they are visible at global scope, they are also subject to linking with
				3517	other globals defined in different translation units. To control the linking
				3518	process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
				3519	<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
				3520	defined by the <tt>LinkageTypes</tt> enumeration.</p>
				3521
				3522	<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
				3523	<tt>static</tt> in C), it is not visible to code outside the current translation
				3524	unit, and does not participate in linking. If it has external linkage, it is
				3525	visible to external code, and does participate in linking. In addition to
				3526	linkage information, <tt>GlobalValue</tt>s keep track of which <a
				3527	href="#Module"><tt>Module</tt></a> they are currently part of.</p>
				3528
				3529	<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
				3530	by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
				3531	global is always a pointer to its contents. It is important to remember this
				3532	when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
				3533	be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
				3534	subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
				3535	i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
				3536	the address of the first element of this array and the value of the
				3537	<tt>GlobalVariable</tt> are the same, they have different types. The
				3538	<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
				3539	is <tt>i32.</tt> Because of this, accessing a global value requires you to
				3540	dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
				3541	can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
				3542	Language Reference Manual</a>.</p>
				3543
				3544	</div>
				3545
				3546	<!-- _______________________________________________________________________ -->
				3547	<div class="doc_subsubsection">
				3548	<a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
				3549	class</a>
				3550	</div>
				3551
				3552	<div class="doc_text">
				3553
				3554	<ul>
				3555	<li><tt>bool hasInternalLinkage() const</tt><br>
				3556	<tt>bool hasExternalLinkage() const</tt><br>
				3557	<tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
				3558	<p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
				3559	<p> </p>
				3560	</li>
				3561	<li><tt><a href="#Module">Module</a> *getParent()</tt>
				3562	<p> This returns the <a href="#Module"><tt>Module</tt></a> that the
				3563	GlobalValue is currently embedded into.</p></li>
				3564	</ul>
				3565
				3566	</div>
				3567
				3568	<!-- ======================================================================= -->
				3569	<div class="doc_subsection">
				3570	<a name="Function">The <tt>Function</tt> class</a>
				3571	</div>
				3572
				3573	<div class="doc_text">
				3574
				3575	<p><tt>#include "<a
				3576	href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
				3577	info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
				3578	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3579	<a href="#Constant"><tt>Constant</tt></a>,
				3580	<a href="#User"><tt>User</tt></a>,
				3581	<a href="#Value"><tt>Value</tt></a></p>
				3582
				3583	<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
Edwin Török	b26e279	2009-10-12 13:37:29 +0000	[diff] [blame]	3584	actually one of the more complex classes in the LLVM hierarchy because it must
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3585	keep track of a large amount of data. The <tt>Function</tt> class keeps track
				3586	of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
				3587	<a href="#Argument"><tt>Argument</tt></a>s, and a
				3588	<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
				3589
				3590	<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
				3591	commonly used part of <tt>Function</tt> objects. The list imposes an implicit
				3592	ordering of the blocks in the function, which indicate how the code will be
Benjamin Kramer	5fb9d7e	2009-10-12 14:46:08 +0000	[diff] [blame]	3593	laid out by the backend. Additionally, the first <a
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3594	href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
				3595	<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
				3596	block. There are no implicit exit nodes, and in fact there may be multiple exit
				3597	nodes from a single <tt>Function</tt>. If the <a
				3598	href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
				3599	the <tt>Function</tt> is actually a function declaration: the actual body of the
				3600	function hasn't been linked in yet.</p>
				3601
				3602	<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
				3603	<tt>Function</tt> class also keeps track of the list of formal <a
				3604	href="#Argument"><tt>Argument</tt></a>s that the function receives. This
				3605	container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
				3606	nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
				3607	the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
				3608
				3609	<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
				3610	LLVM feature that is only used when you have to look up a value by name. Aside
				3611	from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
				3612	internally to make sure that there are not conflicts between the names of <a
				3613	href="#Instruction"><tt>Instruction</tt></a>s, <a
				3614	href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
				3615	href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
				3616
				3617	<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
				3618	and therefore also a <a href="#Constant">Constant</a>. The value of the function
				3619	is its address (after linking) which is guaranteed to be constant.</p>
				3620	</div>
				3621
				3622	<!-- _______________________________________________________________________ -->
				3623	<div class="doc_subsubsection">
				3624	<a name="m_Function">Important Public Members of the <tt>Function</tt>
				3625	class</a>
				3626	</div>
				3627
				3628	<div class="doc_text">
				3629
				3630	<ul>
				3631	<li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
				3632	Ty, LinkageTypes Linkage, const std::string &N = "", Module Parent = 0)</tt>
				3633
				3634	<p>Constructor used when you need to create new <tt>Function</tt>s to add
				3635	the the program. The constructor must specify the type of the function to
				3636	create and what type of linkage the function should have. The <a
				3637	href="#FunctionType"><tt>FunctionType</tt></a> argument
				3638	specifies the formal arguments and return value for the function. The same
				3639	<a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
				3640	create multiple functions. The <tt>Parent</tt> argument specifies the Module
				3641	in which the function is defined. If this argument is provided, the function
				3642	will automatically be inserted into that module's list of
				3643	functions.</p></li>
				3644
Chris Lattner	5e70957	2008-11-25 18:34:50 +0000	[diff] [blame]	3645	<li><tt>bool isDeclaration()</tt>
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3646
				3647	<p>Return whether or not the <tt>Function</tt> has a body defined. If the
				3648	function is "external", it does not have a body, and thus must be resolved
				3649	by linking with a function defined in a different translation unit.</p></li>
				3650
				3651	<li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
				3652	<tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
				3653
				3654	<tt>begin()</tt>, <tt>end()</tt>
				3655	<tt>size()</tt>, <tt>empty()</tt>
				3656
				3657	<p>These are forwarding methods that make it easy to access the contents of
				3658	a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
				3659	list.</p></li>
				3660
				3661	<li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
				3662
				3663	<p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
				3664	is necessary to use when you need to update the list or perform a complex
				3665	action that doesn't have a forwarding method.</p></li>
				3666
				3667	<li><tt>Function::arg_iterator</tt> - Typedef for the argument list
				3668	iterator<br>
				3669	<tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
				3670
				3671	<tt>arg_begin()</tt>, <tt>arg_end()</tt>
				3672	<tt>arg_size()</tt>, <tt>arg_empty()</tt>
				3673
				3674	<p>These are forwarding methods that make it easy to access the contents of
				3675	a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
				3676	list.</p></li>
				3677
				3678	<li><tt>Function::ArgumentListType &getArgumentList()</tt>
				3679
				3680	<p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
				3681	necessary to use when you need to update the list or perform a complex
				3682	action that doesn't have a forwarding method.</p></li>
				3683
				3684	<li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
				3685
				3686	<p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
				3687	function. Because the entry block for the function is always the first
				3688	block, this returns the first block of the <tt>Function</tt>.</p></li>
				3689
				3690	<li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
				3691	<tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
				3692
				3693	<p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
				3694	<tt>Function</tt> and returns the return type of the function, or the <a
				3695	href="#FunctionType"><tt>FunctionType</tt></a> of the actual
				3696	function.</p></li>
				3697
				3698	<li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
				3699
				3700	<p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
				3701	for this <tt>Function</tt>.</p></li>
				3702	</ul>
				3703
				3704	</div>
				3705
				3706	<!-- ======================================================================= -->
				3707	<div class="doc_subsection">
				3708	<a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
				3709	</div>
				3710
				3711	<div class="doc_text">
				3712
				3713	<p><tt>#include "<a
				3714	href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
				3715	<br>
				3716	doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
				3717	Class</a><br>
				3718	Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
				3719	<a href="#Constant"><tt>Constant</tt></a>,
				3720	<a href="#User"><tt>User</tt></a>,
				3721	<a href="#Value"><tt>Value</tt></a></p>
				3722
Benjamin Kramer	5fb9d7e	2009-10-12 14:46:08 +0000	[diff] [blame]	3723	<p>Global variables are represented with the (surprise surprise)
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3724	<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
				3725	subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
				3726	always referenced by their address (global values must live in memory, so their
				3727	"name" refers to their constant address). See
				3728	<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
				3729	variables may have an initial value (which must be a
				3730	<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
				3731	they may be marked as "constant" themselves (indicating that their contents
				3732	never change at runtime).</p>
				3733	</div>
				3734
				3735	<!-- _______________________________________________________________________ -->
				3736	<div class="doc_subsubsection">
				3737	<a name="m_GlobalVariable">Important Public Members of the
				3738	<tt>GlobalVariable</tt> class</a>
				3739	</div>
				3740
				3741	<div class="doc_text">
				3742
				3743	<ul>
				3744	<li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
				3745	isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
				3746	Initializer = 0, const std::string &Name = "", Module Parent = 0)</tt>
				3747
				3748	<p>Create a new global variable of the specified type. If
				3749	<tt>isConstant</tt> is true then the global variable will be marked as
				3750	unchanging for the program. The Linkage parameter specifies the type of
Duncan Sands	19d161f	2009-03-07 15:45:40 +0000	[diff] [blame]	3751	linkage (internal, external, weak, linkonce, appending) for the variable.
				3752	If the linkage is InternalLinkage, WeakAnyLinkage, WeakODRLinkage,
				3753	LinkOnceAnyLinkage or LinkOnceODRLinkage,  then the resultant
				3754	global variable will have internal linkage. AppendingLinkage concatenates
				3755	together all instances (in different translation units) of the variable
				3756	into a single variable but is only applicable to arrays.  See
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3757	the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
				3758	further details on linkage types. Optionally an initializer, a name, and the
				3759	module to put the variable into may be specified for the global variable as
				3760	well.</p></li>
				3761
				3762	<li><tt>bool isConstant() const</tt>
				3763
				3764	<p>Returns true if this is a global variable that is known not to
				3765	be modified at runtime.</p></li>
				3766
				3767	<li><tt>bool hasInitializer()</tt>
				3768
				3769	<p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
				3770
				3771	<li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
				3772
Benjamin Kramer	5fb9d7e	2009-10-12 14:46:08 +0000	[diff] [blame]	3773	<p>Returns the initial value for a <tt>GlobalVariable</tt>. It is not legal
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3774	to call this method if there is no initializer.</p></li>
				3775	</ul>
				3776
				3777	</div>
				3778
				3779
				3780	<!-- ======================================================================= -->
				3781	<div class="doc_subsection">
				3782	<a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
				3783	</div>
				3784
				3785	<div class="doc_text">
				3786
				3787	<p><tt>#include "<a
				3788	href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
Stefanus Du Toit	f017c85	2009-06-17 21:12:26 +0000	[diff] [blame]	3789	doxygen info: <a href="/doxygen/classllvm_1_1BasicBlock.html">BasicBlock
Dan Gohman	f17a25c	2007-07-18 16:29:46 +0000	[diff] [blame]	3790	Class</a><br>
				3791	Superclass: <a href="#Value"><tt>Value</tt></a></p>
				3792
				3793	<p>This class represents a single entry multiple exit section of the code,
				3794	commonly known as a basic block by the compiler community. The
				3795	<tt>BasicBlock</tt> class maintains a list of <a
				3796	href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
				3797	Matching the language definition, the last element of this list of instructions
				3798	is always a terminator instruction (a subclass of the <a
				3799	href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
				3800
				3801	<p>In addition to tracking the list of instructions that make up the block, the
				3802	<tt>BasicBlock</tt> class also keeps track of the <a
				3803	href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
				3804
				3805	<p>Note that <tt>BasicBlock</tt>s themselves are <a
				3806	href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
				3807	like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
				3808	<tt>label</tt>.</p>
				3809
				3810	</div>
				3811
				3812	<!-- _______________________________________________________________________ -->
				3813	<div class="doc_subsubsection">
				3814	<a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
				3815	class</a>
				3816	</div>
				3817
				3818	<div class="doc_text">
				3819	<ul>
				3820
				3821	<li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
				3822	href="#Function">Function</a> *Parent = 0)</tt>
				3823
				3824	<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
				3825	insertion into a function. The constructor optionally takes a name for the new
				3826	block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
				3827	the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
				3828	automatically inserted at the end of the specified <a
				3829	href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
				3830	manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
				3831
				3832	<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
				3833	<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
				3834	<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
				3835	<tt>size()</tt>, <tt>empty()</tt>
				3836	STL-style functions for accessing the instruction list.
				3837
				3838	<p>These methods and typedefs are forwarding functions that have the same
				3839	semantics as the standard library methods of the same names. These methods
				3840	expose the underlying instruction list of a basic block in a way that is easy to
				3841	manipulate. To get the full complement of container operations (including
				3842	operations to update the list), you must use the <tt>getInstList()</tt>
				3843	method.</p></li>
				3844
				3845	<li><tt>BasicBlock::InstListType &getInstList()</tt>
				3846
				3847	<p>This method is used to get access to the underlying container that actually
				3848	holds the Instructions. This method must be used when there isn't a forwarding
				3849	function in the <tt>BasicBlock</tt> class for the operation that you would like
				3850	to perform. Because there are no forwarding functions for "updating"
				3851	operations, you need to use this if you want to update the contents of a
				3852	<tt>BasicBlock</tt>.</p></li>
				3853
				3854	<li><tt><a href="#Function">Function</a> *getParent()</tt>
				3855
				3856	<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
				3857	embedded into, or a null pointer if it is homeless.</p></li>
				3858
				3859	<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
				3860
				3861	<p> Returns a pointer to the terminator instruction that appears at the end of
				3862	the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
				3863	instruction in the block is not a terminator, then a null pointer is
				3864	returned.</p></li>
				3865
				3866	</ul>
				3867
				3868	</div>
				3869
				3870
				3871	<!-- ======================================================================= -->
				3872	<div class="doc_subsection">
				3873	<a name="Argument">The <tt>Argument</tt> class</a>
				3874	</div>
				3875
				3876	<div class="doc_text">
				3877
				3878	<p>This subclass of Value defines the interface for incoming formal
				3879	arguments to a function. A Function maintains a list of its formal
				3880	arguments. An argument has a pointer to the parent Function.</p>
				3881
				3882	</div>
				3883
				3884	<!-- *********************************************************************** -->
				3885	<hr>
				3886	<address>
				3887	<a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman	947321d	2008-12-11 17:34:48 +0000	[diff] [blame]	3888	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]	3889	<a href="http://validator.w3.org/check/referer"><img
Gabor Greif	7dabe38	2008-06-18 14:05:31 +0000	[diff] [blame]	3890	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]	3891
				3892	<a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
				3893	<a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
				3894	<a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
				3895	Last modified: $Date$
				3896	</address>
				3897
				3898	</body>
				3899	</html>