1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Programmer's Manual</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
10 <div class="doc_title">
11 LLVM Programmer's Manual
15 <li><a href="#introduction">Introduction</a></li>
16 <li><a href="#general">General Information</a>
18 <li><a href="#stl">The C++ Standard Template Library</a></li>
20 <li>The <tt>-time-passes</tt> option</li>
21 <li>How to use the LLVM Makefile system</li>
22 <li>How to write a regression test</li>
27 <li><a href="#apis">Important and useful LLVM APIs</a>
29 <li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
30 and <tt>dyn_cast<></tt> templates</a> </li>
31 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
34 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
35 and the <tt>-debug-only</tt> option</a> </li>
38 <li><a href="#Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
41 <li>The <tt>InstVisitor</tt> template
42 <li>The general graph API
44 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
47 <li><a href="#common">Helpful Hints for Common Operations</a>
49 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
51 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
52 in a <tt>Function</tt></a> </li>
53 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
54 in a <tt>BasicBlock</tt></a> </li>
55 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
56 in a <tt>Function</tt></a> </li>
57 <li><a href="#iterate_convert">Turning an iterator into a
58 class pointer</a> </li>
59 <li><a href="#iterate_complex">Finding call sites: a more
60 complex example</a> </li>
61 <li><a href="#calls_and_invokes">Treating calls and invokes
62 the same way</a> </li>
63 <li><a href="#iterate_chains">Iterating over def-use &
64 use-def chains</a> </li>
67 <li><a href="#simplechanges">Making simple changes</a>
69 <li><a href="#schanges_creating">Creating and inserting new
70 <tt>Instruction</tt>s</a> </li>
71 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
72 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
73 with another <tt>Value</tt></a> </li>
77 <li>Working with the Control Flow Graph
79 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
87 <li><a href="#advanced">Advanced Topics</a>
89 <li><a href="#TypeResolve">LLVM Type Resolution</a>
91 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
92 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
93 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
94 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
97 <li><a href="#SymbolTable">The <tt>SymbolTable</tt> class </a></li>
100 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
102 <li><a href="#Value">The <tt>Value</tt> class</a>
104 <li><a href="#User">The <tt>User</tt> class</a>
106 <li><a href="#Instruction">The <tt>Instruction</tt> class</a>
108 <li><a href="#GetElementPtrInst">The <tt>GetElementPtrInst</tt> class</a></li>
111 <li><a href="#Module">The <tt>Module</tt> class</a></li>
112 <li><a href="#Constant">The <tt>Constant</tt> class</a>
114 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
116 <li><a href="#BasicBlock">The <tt>BasicBlock</tt>class</a></li>
117 <li><a href="#Function">The <tt>Function</tt> class</a></li>
118 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
125 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
126 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
133 <div class="doc_author">
134 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
135 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
136 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
137 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
140 <!-- *********************************************************************** -->
141 <div class="doc_section">
142 <a name="introduction">Introduction </a>
144 <!-- *********************************************************************** -->
146 <div class="doc_text">
148 <p>This document is meant to highlight some of the important classes and
149 interfaces available in the LLVM source-base. This manual is not
150 intended to explain what LLVM is, how it works, and what LLVM code looks
151 like. It assumes that you know the basics of LLVM and are interested
152 in writing transformations or otherwise analyzing or manipulating the
155 <p>This document should get you oriented so that you can find your
156 way in the continuously growing source code that makes up the LLVM
157 infrastructure. Note that this manual is not intended to serve as a
158 replacement for reading the source code, so if you think there should be
159 a method in one of these classes to do something, but it's not listed,
160 check the source. Links to the <a href="/doxygen/">doxygen</a> sources
161 are provided to make this as easy as possible.</p>
163 <p>The first section of this document describes general information that is
164 useful to know when working in the LLVM infrastructure, and the second describes
165 the Core LLVM classes. In the future this manual will be extended with
166 information describing how to use extension libraries, such as dominator
167 information, CFG traversal routines, and useful utilities like the <tt><a
168 href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
172 <!-- *********************************************************************** -->
173 <div class="doc_section">
174 <a name="general">General Information</a>
176 <!-- *********************************************************************** -->
178 <div class="doc_text">
180 <p>This section contains general information that is useful if you are working
181 in the LLVM source-base, but that isn't specific to any particular API.</p>
185 <!-- ======================================================================= -->
186 <div class="doc_subsection">
187 <a name="stl">The C++ Standard Template Library</a>
190 <div class="doc_text">
192 <p>LLVM makes heavy use of the C++ Standard Template Library (STL),
193 perhaps much more than you are used to, or have seen before. Because of
194 this, you might want to do a little background reading in the
195 techniques used and capabilities of the library. There are many good
196 pages that discuss the STL, and several books on the subject that you
197 can get, so it will not be discussed in this document.</p>
199 <p>Here are some useful links:</p>
203 <li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
204 reference</a> - an excellent reference for the STL and other parts of the
205 standard C++ library.</li>
207 <li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
208 O'Reilly book in the making. It has a decent
210 Reference that rivals Dinkumware's, and is unfortunately no longer free since the book has been
213 <li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
216 <li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
218 href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
221 <li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
224 <li><a href="http://64.78.49.204/">
225 Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
230 <p>You are also encouraged to take a look at the <a
231 href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
232 to write maintainable code more than where to put your curly braces.</p>
236 <!-- ======================================================================= -->
237 <div class="doc_subsection">
238 <a name="stl">Other useful references</a>
241 <div class="doc_text">
244 <li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
245 Branch and Tag Primer</a></li>
246 <li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
247 static and shared libraries across platforms</a></li>
252 <!-- *********************************************************************** -->
253 <div class="doc_section">
254 <a name="apis">Important and useful LLVM APIs</a>
256 <!-- *********************************************************************** -->
258 <div class="doc_text">
260 <p>Here we highlight some LLVM APIs that are generally useful and good to
261 know about when writing transformations.</p>
265 <!-- ======================================================================= -->
266 <div class="doc_subsection">
267 <a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
268 <tt>dyn_cast<></tt> templates</a>
271 <div class="doc_text">
273 <p>The LLVM source-base makes extensive use of a custom form of RTTI.
274 These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
275 operator, but they don't have some drawbacks (primarily stemming from
276 the fact that <tt>dynamic_cast<></tt> only works on classes that
277 have a v-table). Because they are used so often, you must know what they
278 do and how they work. All of these templates are defined in the <a
279 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
280 file (note that you very rarely have to include this file directly).</p>
283 <dt><tt>isa<></tt>: </dt>
285 <dd><p>The <tt>isa<></tt> operator works exactly like the Java
286 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
287 a reference or pointer points to an instance of the specified class. This can
288 be very useful for constraint checking of various sorts (example below).</p>
291 <dt><tt>cast<></tt>: </dt>
293 <dd><p>The <tt>cast<></tt> operator is a "checked cast" operation. It
294 converts a pointer or reference from a base class to a derived cast, causing
295 an assertion failure if it is not really an instance of the right type. This
296 should be used in cases where you have some information that makes you believe
297 that something is of the right type. An example of the <tt>isa<></tt>
298 and <tt>cast<></tt> template is:</p>
300 <div class="doc_code">
302 static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
303 if (isa<<a href="#Constant">Constant</a>>(V) || isa<<a href="#Argument">Argument</a>>(V) || isa<<a href="#GlobalValue">GlobalValue</a>>(V))
306 <i>// Otherwise, it must be an instruction...</i>
307 return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
312 <p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
313 by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
318 <dt><tt>dyn_cast<></tt>:</dt>
320 <dd><p>The <tt>dyn_cast<></tt> operator is a "checking cast" operation.
321 It checks to see if the operand is of the specified type, and if so, returns a
322 pointer to it (this operator does not work with references). If the operand is
323 not of the correct type, a null pointer is returned. Thus, this works very
324 much like the <tt>dynamic_cast<></tt> operator in C++, and should be
325 used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
326 operator is used in an <tt>if</tt> statement or some other flow control
327 statement like this:</p>
329 <div class="doc_code">
331 if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
337 <p>This form of the <tt>if</tt> statement effectively combines together a call
338 to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
339 statement, which is very convenient.</p>
341 <p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
342 <tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
343 abused. In particular, you should not use big chained <tt>if/then/else</tt>
344 blocks to check for lots of different variants of classes. If you find
345 yourself wanting to do this, it is much cleaner and more efficient to use the
346 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
350 <dt><tt>cast_or_null<></tt>: </dt>
352 <dd><p>The <tt>cast_or_null<></tt> operator works just like the
353 <tt>cast<></tt> operator, except that it allows for a null pointer as an
354 argument (which it then propagates). This can sometimes be useful, allowing
355 you to combine several null checks into one.</p></dd>
357 <dt><tt>dyn_cast_or_null<></tt>: </dt>
359 <dd><p>The <tt>dyn_cast_or_null<></tt> operator works just like the
360 <tt>dyn_cast<></tt> operator, except that it allows for a null pointer
361 as an argument (which it then propagates). This can sometimes be useful,
362 allowing you to combine several null checks into one.</p></dd>
366 <p>These five templates can be used with any classes, whether they have a
367 v-table or not. To add support for these templates, you simply need to add
368 <tt>classof</tt> static methods to the class you are interested casting
369 to. Describing this is currently outside the scope of this document, but there
370 are lots of examples in the LLVM source base.</p>
374 <!-- ======================================================================= -->
375 <div class="doc_subsection">
376 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
379 <div class="doc_text">
381 <p>Often when working on your pass you will put a bunch of debugging printouts
382 and other code into your pass. After you get it working, you want to remove
383 it, but you may need it again in the future (to work out new bugs that you run
386 <p> Naturally, because of this, you don't want to delete the debug printouts,
387 but you don't want them to always be noisy. A standard compromise is to comment
388 them out, allowing you to enable them if you need them in the future.</p>
390 <p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
391 file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
392 this problem. Basically, you can put arbitrary code into the argument of the
393 <tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
394 tool) is run with the '<tt>-debug</tt>' command line argument:</p>
396 <div class="doc_code">
398 DEBUG(std::cerr << "I am here!\n");
402 <p>Then you can run your pass like this:</p>
404 <div class="doc_code">
406 $ opt < a.bc > /dev/null -mypass
408 $ opt < a.bc > /dev/null -mypass -debug
413 <p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
414 to not have to create "yet another" command line option for the debug output for
415 your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
416 so they do not cause a performance impact at all (for the same reason, they
417 should also not contain side-effects!).</p>
419 <p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
420 enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
421 "<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
422 program hasn't been started yet, you can always just run it with
427 <!-- _______________________________________________________________________ -->
428 <div class="doc_subsubsection">
429 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
430 the <tt>-debug-only</tt> option</a>
433 <div class="doc_text">
435 <p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
436 just turns on <b>too much</b> information (such as when working on the code
437 generator). If you want to enable debug information with more fine-grained
438 control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
439 option as follows:</p>
441 <div class="doc_code">
443 DEBUG(std::cerr << "No debug type\n");
445 #define DEBUG_TYPE "foo"
446 DEBUG(std::cerr << "'foo' debug type\n");
448 #define DEBUG_TYPE "bar"
449 DEBUG(std::cerr << "'bar' debug type\n");
451 #define DEBUG_TYPE ""
452 DEBUG(std::cerr << "No debug type (2)\n");
456 <p>Then you can run your pass like this:</p>
458 <div class="doc_code">
460 $ opt < a.bc > /dev/null -mypass
462 $ opt < a.bc > /dev/null -mypass -debug
467 $ opt < a.bc > /dev/null -mypass -debug-only=foo
469 $ opt < a.bc > /dev/null -mypass -debug-only=bar
474 <p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
475 a file, to specify the debug type for the entire module (if you do this before
476 you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
477 <tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
478 "bar", because there is no system in place to ensure that names do not
479 conflict. If two different modules use the same string, they will all be turned
480 on when the name is specified. This allows, for example, all debug information
481 for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
482 even if the source lives in multiple files.</p>
486 <!-- ======================================================================= -->
487 <div class="doc_subsection">
488 <a name="Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
492 <div class="doc_text">
495 href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
496 provides a template named <tt>Statistic</tt> that is used as a unified way to
497 keep track of what the LLVM compiler is doing and how effective various
498 optimizations are. It is useful to see what optimizations are contributing to
499 making a particular program run faster.</p>
501 <p>Often you may run your pass on some big program, and you're interested to see
502 how many times it makes a certain transformation. Although you can do this with
503 hand inspection, or some ad-hoc method, this is a real pain and not very useful
504 for big programs. Using the <tt>Statistic</tt> template makes it very easy to
505 keep track of this information, and the calculated information is presented in a
506 uniform manner with the rest of the passes being executed.</p>
508 <p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
509 it are as follows:</p>
512 <li><p>Define your statistic like this:</p>
514 <div class="doc_code">
516 static Statistic<> NumXForms("mypassname", "The # of times I did stuff");
520 <p>The <tt>Statistic</tt> template can emulate just about any data-type,
521 but if you do not specify a template argument, it defaults to acting like
522 an unsigned int counter (this is usually what you want).</p></li>
524 <li><p>Whenever you make a transformation, bump the counter:</p>
526 <div class="doc_code">
528 ++NumXForms; // I did stuff!
535 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
536 statistics gathered, use the '<tt>-stats</tt>' option:</p>
538 <div class="doc_code">
540 $ opt -stats -mypassname < program.bc > /dev/null
541 ... statistic output ...
545 <p> When running <tt>gccas</tt> on a C file from the SPEC benchmark
546 suite, it gives a report that looks like this:</p>
548 <div class="doc_code">
550 7646 bytecodewriter - Number of normal instructions
551 725 bytecodewriter - Number of oversized instructions
552 129996 bytecodewriter - Number of bytecode bytes written
553 2817 raise - Number of insts DCEd or constprop'd
554 3213 raise - Number of cast-of-self removed
555 5046 raise - Number of expression trees converted
556 75 raise - Number of other getelementptr's formed
557 138 raise - Number of load/store peepholes
558 42 deadtypeelim - Number of unused typenames removed from symtab
559 392 funcresolve - Number of varargs functions resolved
560 27 globaldce - Number of global variables removed
561 2 adce - Number of basic blocks removed
562 134 cee - Number of branches revectored
563 49 cee - Number of setcc instruction eliminated
564 532 gcse - Number of loads removed
565 2919 gcse - Number of instructions removed
566 86 indvars - Number of canonical indvars added
567 87 indvars - Number of aux indvars removed
568 25 instcombine - Number of dead inst eliminate
569 434 instcombine - Number of insts combined
570 248 licm - Number of load insts hoisted
571 1298 licm - Number of insts hoisted to a loop pre-header
572 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
573 75 mem2reg - Number of alloca's promoted
574 1444 cfgsimplify - Number of blocks simplified
578 <p>Obviously, with so many optimizations, having a unified framework for this
579 stuff is very nice. Making your pass fit well into the framework makes it more
580 maintainable and useful.</p>
584 <!-- ======================================================================= -->
585 <div class="doc_subsection">
586 <a name="ViewGraph">Viewing graphs while debugging code</a>
589 <div class="doc_text">
591 <p>Several of the important data structures in LLVM are graphs: for example
592 CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
593 LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
594 <a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
595 DAGs</a>. In many cases, while debugging various parts of the compiler, it is
596 nice to instantly visualize these graphs.</p>
598 <p>LLVM provides several callbacks that are available in a debug build to do
599 exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
600 the current LLVM tool will pop up a window containing the CFG for the function
601 where each basic block is a node in the graph, and each node contains the
602 instructions in the block. Similarly, there also exists
603 <tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
604 <tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
605 and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
606 you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
607 up a window. Alternatively, you can sprinkle calls to these functions in your
608 code in places you want to debug.</p>
610 <p>Getting this to work requires a small amount of configuration. On Unix
611 systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
612 toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
613 Mac OS/X, download and install the Mac OS/X <a
614 href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
615 <tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or whereever you install
616 it) to your path. Once in your system and path are set up, rerun the LLVM
617 configure script and rebuild LLVM to enable this functionality.</p>
619 <p><tt>SelectionDAG</tt> has been extended to make it easier to locate
620 <i>interesting</i> nodes in large complex graphs. From gdb, if you
621 <tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
622 next <tt>call DAG.viewGraph()</tt> would hilight the node in the
623 specified color (choices of colors can be found at <a
624 href="http://www.graphviz.org/doc/info/colors.html">Colors<a>.) More
625 complex node attributes can be provided with <tt>call
626 DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
627 found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
628 Attributes</a>.) If you want to restart and clear all the current graph
629 attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
634 <!-- *********************************************************************** -->
635 <div class="doc_section">
636 <a name="common">Helpful Hints for Common Operations</a>
638 <!-- *********************************************************************** -->
640 <div class="doc_text">
642 <p>This section describes how to perform some very simple transformations of
643 LLVM code. This is meant to give examples of common idioms used, showing the
644 practical side of LLVM transformations. <p> Because this is a "how-to" section,
645 you should also read about the main classes that you will be working with. The
646 <a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
647 and descriptions of the main classes that you should know about.</p>
651 <!-- NOTE: this section should be heavy on example code -->
652 <!-- ======================================================================= -->
653 <div class="doc_subsection">
654 <a name="inspection">Basic Inspection and Traversal Routines</a>
657 <div class="doc_text">
659 <p>The LLVM compiler infrastructure have many different data structures that may
660 be traversed. Following the example of the C++ standard template library, the
661 techniques used to traverse these various data structures are all basically the
662 same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
663 method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
664 function returns an iterator pointing to one past the last valid element of the
665 sequence, and there is some <tt>XXXiterator</tt> data type that is common
666 between the two operations.</p>
668 <p>Because the pattern for iteration is common across many different aspects of
669 the program representation, the standard template library algorithms may be used
670 on them, and it is easier to remember how to iterate. First we show a few common
671 examples of the data structures that need to be traversed. Other data
672 structures are traversed in very similar ways.</p>
676 <!-- _______________________________________________________________________ -->
677 <div class="doc_subsubsection">
678 <a name="iterate_function">Iterating over the </a><a
679 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
680 href="#Function"><tt>Function</tt></a>
683 <div class="doc_text">
685 <p>It's quite common to have a <tt>Function</tt> instance that you'd like to
686 transform in some way; in particular, you'd like to manipulate its
687 <tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
688 the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
689 an example that prints the name of a <tt>BasicBlock</tt> and the number of
690 <tt>Instruction</tt>s it contains:</p>
692 <div class="doc_code">
694 // func is a pointer to a Function instance
695 for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i) {
696 // print out the name of the basic block if it has one, and then the
697 // number of instructions that it contains
698 std::cerr << "Basic block (name=" << i->getName() << ") has "
699 << i->size() << " instructions.\n";
704 <p>Note that i can be used as if it were a pointer for the purposes of
705 invoking member functions of the <tt>Instruction</tt> class. This is
706 because the indirection operator is overloaded for the iterator
707 classes. In the above code, the expression <tt>i->size()</tt> is
708 exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
712 <!-- _______________________________________________________________________ -->
713 <div class="doc_subsubsection">
714 <a name="iterate_basicblock">Iterating over the </a><a
715 href="#Instruction"><tt>Instruction</tt></a>s in a <a
716 href="#BasicBlock"><tt>BasicBlock</tt></a>
719 <div class="doc_text">
721 <p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
722 easy to iterate over the individual instructions that make up
723 <tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
724 a <tt>BasicBlock</tt>:</p>
726 <div class="doc_code">
728 // blk is a pointer to a BasicBlock instance
729 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
730 // the next statement works since operator<<(ostream&,...)
731 // is overloaded for Instruction&
732 std::cerr << *i << "\n";
736 <p>However, this isn't really the best way to print out the contents of a
737 <tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
738 anything you'll care about, you could have just invoked the print routine on the
739 basic block itself: <tt>std::cerr << *blk << "\n";</tt>.</p>
743 <!-- _______________________________________________________________________ -->
744 <div class="doc_subsubsection">
745 <a name="iterate_institer">Iterating over the </a><a
746 href="#Instruction"><tt>Instruction</tt></a>s in a <a
747 href="#Function"><tt>Function</tt></a>
750 <div class="doc_text">
752 <p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
753 <tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
754 <tt>InstIterator</tt> should be used instead. You'll need to include <a
755 href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
756 and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
757 small example that shows how to dump all instructions in a function to the standard error stream:<p>
759 <div class="doc_code">
761 #include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
763 // Suppose F is a ptr to a function
764 for (inst_iterator i = inst_begin(F), e = inst_end(F); i != e; ++i)
765 std::cerr << *i << "\n";
769 <p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
770 worklist with its initial contents. For example, if you wanted to
771 initialize a worklist to contain all instructions in a <tt>Function</tt>
772 F, all you would need to do is something like:</p>
774 <div class="doc_code">
776 std::set<Instruction*> worklist;
777 worklist.insert(inst_begin(F), inst_end(F));
781 <p>The STL set <tt>worklist</tt> would now contain all instructions in the
782 <tt>Function</tt> pointed to by F.</p>
786 <!-- _______________________________________________________________________ -->
787 <div class="doc_subsubsection">
788 <a name="iterate_convert">Turning an iterator into a class pointer (and
792 <div class="doc_text">
794 <p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
795 instance when all you've got at hand is an iterator. Well, extracting
796 a reference or a pointer from an iterator is very straight-forward.
797 Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
798 is a <tt>BasicBlock::const_iterator</tt>:</p>
800 <div class="doc_code">
802 Instruction& inst = *i; // grab reference to instruction reference
803 Instruction* pinst = &*i; // grab pointer to instruction reference
804 const Instruction& inst = *j;
808 <p>However, the iterators you'll be working with in the LLVM framework are
809 special: they will automatically convert to a ptr-to-instance type whenever they
810 need to. Instead of dereferencing the iterator and then taking the address of
811 the result, you can simply assign the iterator to the proper pointer type and
812 you get the dereference and address-of operation as a result of the assignment
813 (behind the scenes, this is a result of overloading casting mechanisms). Thus
814 the last line of the last example,</p>
816 <div class="doc_code">
818 Instruction* pinst = &*i;
822 <p>is semantically equivalent to</p>
824 <div class="doc_code">
826 Instruction* pinst = i;
830 <p>It's also possible to turn a class pointer into the corresponding iterator,
831 and this is a constant time operation (very efficient). The following code
832 snippet illustrates use of the conversion constructors provided by LLVM
833 iterators. By using these, you can explicitly grab the iterator of something
834 without actually obtaining it via iteration over some structure:</p>
836 <div class="doc_code">
838 void printNextInstruction(Instruction* inst) {
839 BasicBlock::iterator it(inst);
840 ++it; // after this line, it refers to the instruction after *inst.
841 if (it != inst->getParent()->end()) std::cerr << *it << "\n";
848 <!--_______________________________________________________________________-->
849 <div class="doc_subsubsection">
850 <a name="iterate_complex">Finding call sites: a slightly more complex
854 <div class="doc_text">
856 <p>Say that you're writing a FunctionPass and would like to count all the
857 locations in the entire module (that is, across every <tt>Function</tt>) where a
858 certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
859 learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
860 much more straight-forward manner, but this example will allow us to explore how
861 you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudocode, this
862 is what we want to do:</p>
864 <div class="doc_code">
866 initialize callCounter to zero
867 for each Function f in the Module
868 for each BasicBlock b in f
869 for each Instruction i in b
870 if (i is a CallInst and calls the given function)
871 increment callCounter
875 <p>And the actual code is (remember, because we're writing a
876 <tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
877 override the <tt>runOnFunction</tt> method):</p>
879 <div class="doc_code">
881 Function* targetFunc = ...;
883 class OurFunctionPass : public FunctionPass {
885 OurFunctionPass(): callCounter(0) { }
887 virtual runOnFunction(Function& F) {
888 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
889 for (BasicBlock::iterator i = b->begin(); ie = b->end(); i != ie; ++i) {
890 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
891 href="#CallInst">CallInst</a>>(&*i)) {
892 // we know we've encountered a call instruction, so we
893 // need to determine if it's a call to the
894 // function pointed to by m_func or not.
896 if (callInst->getCalledFunction() == targetFunc)
904 unsigned callCounter;
911 <!--_______________________________________________________________________-->
912 <div class="doc_subsubsection">
913 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
916 <div class="doc_text">
918 <p>You may have noticed that the previous example was a bit oversimplified in
919 that it did not deal with call sites generated by 'invoke' instructions. In
920 this, and in other situations, you may find that you want to treat
921 <tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
922 most-specific common base class is <tt>Instruction</tt>, which includes lots of
923 less closely-related things. For these cases, LLVM provides a handy wrapper
925 href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
926 It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
927 methods that provide functionality common to <tt>CallInst</tt>s and
928 <tt>InvokeInst</tt>s.</p>
930 <p>This class has "value semantics": it should be passed by value, not by
931 reference and it should not be dynamically allocated or deallocated using
932 <tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
933 assignable and constructable, with costs equivalents to that of a bare pointer.
934 If you look at its definition, it has only a single pointer member.</p>
938 <!--_______________________________________________________________________-->
939 <div class="doc_subsubsection">
940 <a name="iterate_chains">Iterating over def-use & use-def chains</a>
943 <div class="doc_text">
945 <p>Frequently, we might have an instance of the <a
946 href="/doxygen/structllvm_1_1Value.html">Value Class</a> and we want to
947 determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
948 <tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
949 For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
950 particular function <tt>foo</tt>. Finding all of the instructions that
951 <i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
954 <div class="doc_code">
958 for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i) {
959 if (Instruction *Inst = dyn_cast<Instruction>(*i)) {
960 std::cerr << "F is used in instruction:\n";
961 std::cerr << *Inst << "\n";
967 <p>Alternately, it's common to have an instance of the <a
968 href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
969 <tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
970 <tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
971 <tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
972 all of the values that a particular instruction uses (that is, the operands of
973 the particular <tt>Instruction</tt>):</p>
975 <div class="doc_code">
977 Instruction* pi = ...;
979 for (User::op_iterator i = pi->op_begin(), e = pi->op_end(); i != e; ++i) {
987 def-use chains ("finding all users of"): Value::use_begin/use_end
988 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
993 <!-- ======================================================================= -->
994 <div class="doc_subsection">
995 <a name="simplechanges">Making simple changes</a>
998 <div class="doc_text">
1000 <p>There are some primitive transformation operations present in the LLVM
1001 infrastructure that are worth knowing about. When performing
1002 transformations, it's fairly common to manipulate the contents of basic
1003 blocks. This section describes some of the common methods for doing so
1004 and gives example code.</p>
1008 <!--_______________________________________________________________________-->
1009 <div class="doc_subsubsection">
1010 <a name="schanges_creating">Creating and inserting new
1011 <tt>Instruction</tt>s</a>
1014 <div class="doc_text">
1016 <p><i>Instantiating Instructions</i></p>
1018 <p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
1019 constructor for the kind of instruction to instantiate and provide the necessary
1020 parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
1021 (const-ptr-to) <tt>Type</tt>. Thus:</p>
1023 <div class="doc_code">
1025 AllocaInst* ai = new AllocaInst(Type::IntTy);
1029 <p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
1030 one integer in the current stack frame, at runtime. Each <tt>Instruction</tt>
1031 subclass is likely to have varying default parameters which change the semantics
1032 of the instruction, so refer to the <a
1033 href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
1034 Instruction</a> that you're interested in instantiating.</p>
1036 <p><i>Naming values</i></p>
1038 <p>It is very useful to name the values of instructions when you're able to, as
1039 this facilitates the debugging of your transformations. If you end up looking
1040 at generated LLVM machine code, you definitely want to have logical names
1041 associated with the results of instructions! By supplying a value for the
1042 <tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
1043 associate a logical name with the result of the instruction's execution at
1044 runtime. For example, say that I'm writing a transformation that dynamically
1045 allocates space for an integer on the stack, and that integer is going to be
1046 used as some kind of index by some other code. To accomplish this, I place an
1047 <tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
1048 <tt>Function</tt>, and I'm intending to use it within the same
1049 <tt>Function</tt>. I might do:</p>
1051 <div class="doc_code">
1053 AllocaInst* pa = new AllocaInst(Type::IntTy, 0, "indexLoc");
1057 <p>where <tt>indexLoc</tt> is now the logical name of the instruction's
1058 execution value, which is a pointer to an integer on the runtime stack.</p>
1060 <p><i>Inserting instructions</i></p>
1062 <p>There are essentially two ways to insert an <tt>Instruction</tt>
1063 into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
1066 <li>Insertion into an explicit instruction list
1068 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
1069 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
1070 before <tt>*pi</tt>, we do the following: </p>
1072 <div class="doc_code">
1074 BasicBlock *pb = ...;
1075 Instruction *pi = ...;
1076 Instruction *newInst = new Instruction(...);
1078 pb->getInstList().insert(pi, newInst); // inserts newInst before pi in pb
1082 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
1083 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
1084 classes provide constructors which take a pointer to a
1085 <tt>BasicBlock</tt> to be appended to. For example code that
1088 <div class="doc_code">
1090 BasicBlock *pb = ...;
1091 Instruction *newInst = new Instruction(...);
1093 pb->getInstList().push_back(newInst); // appends newInst to pb
1099 <div class="doc_code">
1101 BasicBlock *pb = ...;
1102 Instruction *newInst = new Instruction(..., pb);
1106 <p>which is much cleaner, especially if you are creating
1107 long instruction streams.</p></li>
1109 <li>Insertion into an implicit instruction list
1111 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
1112 are implicitly associated with an existing instruction list: the instruction
1113 list of the enclosing basic block. Thus, we could have accomplished the same
1114 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
1117 <div class="doc_code">
1119 Instruction *pi = ...;
1120 Instruction *newInst = new Instruction(...);
1122 pi->getParent()->getInstList().insert(pi, newInst);
1126 <p>In fact, this sequence of steps occurs so frequently that the
1127 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
1128 constructors which take (as a default parameter) a pointer to an
1129 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
1130 precede. That is, <tt>Instruction</tt> constructors are capable of
1131 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
1132 provided instruction, immediately before that instruction. Using an
1133 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
1134 parameter, the above code becomes:</p>
1136 <div class="doc_code">
1138 Instruction* pi = ...;
1139 Instruction* newInst = new Instruction(..., pi);
1143 <p>which is much cleaner, especially if you're creating a lot of
1144 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
1149 <!--_______________________________________________________________________-->
1150 <div class="doc_subsubsection">
1151 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
1154 <div class="doc_text">
1156 <p>Deleting an instruction from an existing sequence of instructions that form a
1157 <a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
1158 you must have a pointer to the instruction that you wish to delete. Second, you
1159 need to obtain the pointer to that instruction's basic block. You use the
1160 pointer to the basic block to get its list of instructions and then use the
1161 erase function to remove your instruction. For example:</p>
1163 <div class="doc_code">
1165 <a href="#Instruction">Instruction</a> *I = .. ;
1166 <a href="#BasicBlock">BasicBlock</a> *BB = I->getParent();
1168 BB->getInstList().erase(I);
1174 <!--_______________________________________________________________________-->
1175 <div class="doc_subsubsection">
1176 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
1180 <div class="doc_text">
1182 <p><i>Replacing individual instructions</i></p>
1184 <p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
1185 permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
1186 and <tt>ReplaceInstWithInst</tt>.</p>
1188 <h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
1191 <li><tt>ReplaceInstWithValue</tt>
1193 <p>This function replaces all uses (within a basic block) of a given
1194 instruction with a value, and then removes the original instruction. The
1195 following example illustrates the replacement of the result of a particular
1196 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
1197 pointer to an integer.</p>
1199 <div class="doc_code">
1201 AllocaInst* instToReplace = ...;
1202 BasicBlock::iterator ii(instToReplace);
1204 ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
1205 Constant::getNullValue(PointerType::get(Type::IntTy)));
1208 <li><tt>ReplaceInstWithInst</tt>
1210 <p>This function replaces a particular instruction with another
1211 instruction. The following example illustrates the replacement of one
1212 <tt>AllocaInst</tt> with another.</p>
1214 <div class="doc_code">
1216 AllocaInst* instToReplace = ...;
1217 BasicBlock::iterator ii(instToReplace);
1219 ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
1220 new AllocaInst(Type::IntTy, 0, "ptrToReplacedInt"));
1224 <p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
1226 <p>You can use <tt>Value::replaceAllUsesWith</tt> and
1227 <tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
1228 doxygen documentation for the <a href="/doxygen/structllvm_1_1Value.html">Value Class</a>
1229 and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
1232 <!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
1233 include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
1234 ReplaceInstWithValue, ReplaceInstWithInst -->
1238 <!-- *********************************************************************** -->
1239 <div class="doc_section">
1240 <a name="advanced">Advanced Topics</a>
1242 <!-- *********************************************************************** -->
1244 <div class="doc_text">
1246 This section describes some of the advanced or obscure API's that most clients
1247 do not need to be aware of. These API's tend manage the inner workings of the
1248 LLVM system, and only need to be accessed in unusual circumstances.
1252 <!-- ======================================================================= -->
1253 <div class="doc_subsection">
1254 <a name="TypeResolve">LLVM Type Resolution</a>
1257 <div class="doc_text">
1260 The LLVM type system has a very simple goal: allow clients to compare types for
1261 structural equality with a simple pointer comparison (aka a shallow compare).
1262 This goal makes clients much simpler and faster, and is used throughout the LLVM
1267 Unfortunately achieving this goal is not a simple matter. In particular,
1268 recursive types and late resolution of opaque types makes the situation very
1269 difficult to handle. Fortunately, for the most part, our implementation makes
1270 most clients able to be completely unaware of the nasty internal details. The
1271 primary case where clients are exposed to the inner workings of it are when
1272 building a recursive type. In addition to this case, the LLVM bytecode reader,
1273 assembly parser, and linker also have to be aware of the inner workings of this
1278 For our purposes below, we need three concepts. First, an "Opaque Type" is
1279 exactly as defined in the <a href="LangRef.html#t_opaque">language
1280 reference</a>. Second an "Abstract Type" is any type which includes an
1281 opaque type as part of its type graph (for example "<tt>{ opaque, int }</tt>").
1282 Third, a concrete type is a type that is not an abstract type (e.g. "<tt>[ int,
1288 <!-- ______________________________________________________________________ -->
1289 <div class="doc_subsubsection">
1290 <a name="BuildRecType">Basic Recursive Type Construction</a>
1293 <div class="doc_text">
1296 Because the most common question is "how do I build a recursive type with LLVM",
1297 we answer it now and explain it as we go. Here we include enough to cause this
1298 to be emitted to an output .ll file:
1301 <div class="doc_code">
1303 %mylist = type { %mylist*, int }
1308 To build this, use the following LLVM APIs:
1311 <div class="doc_code">
1313 //<i> Create the initial outer struct.</i>
1314 <a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
1315 std::vector<const Type*> Elts;
1316 Elts.push_back(PointerType::get(StructTy));
1317 Elts.push_back(Type::IntTy);
1318 StructType *NewSTy = StructType::get(Elts);
1320 //<i> At this point, NewSTy = "{ opaque*, int }". Tell VMCore that</i>
1321 //<i> the struct and the opaque type are actually the same.</i>
1322 cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
1324 // <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
1325 // <i>kept up-to-date.</i>
1326 NewSTy = cast<StructType>(StructTy.get());
1328 // <i>Add a name for the type to the module symbol table (optional).</i>
1329 MyModule->addTypeName("mylist", NewSTy);
1334 This code shows the basic approach used to build recursive types: build a
1335 non-recursive type using 'opaque', then use type unification to close the cycle.
1336 The type unification step is performed by the <tt><a
1337 ref="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
1338 described next. After that, we describe the <a
1339 href="#PATypeHolder">PATypeHolder class</a>.
1344 <!-- ______________________________________________________________________ -->
1345 <div class="doc_subsubsection">
1346 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
1349 <div class="doc_text">
1351 The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
1352 While this method is actually a member of the DerivedType class, it is most
1353 often used on OpaqueType instances. Type unification is actually a recursive
1354 process. After unification, types can become structurally isomorphic to
1355 existing types, and all duplicates are deleted (to preserve pointer equality).
1359 In the example above, the OpaqueType object is definitely deleted.
1360 Additionally, if there is an "{ \2*, int}" type already created in the system,
1361 the pointer and struct type created are <b>also</b> deleted. Obviously whenever
1362 a type is deleted, any "Type*" pointers in the program are invalidated. As
1363 such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
1364 live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
1365 types can never move or be deleted). To deal with this, the <a
1366 href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
1367 reference to a possibly refined type, and the <a
1368 href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
1369 complex datastructures.
1374 <!-- ______________________________________________________________________ -->
1375 <div class="doc_subsubsection">
1376 <a name="PATypeHolder">The PATypeHolder Class</a>
1379 <div class="doc_text">
1381 PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
1382 happily goes about nuking types that become isomorphic to existing types, it
1383 automatically updates all PATypeHolder objects to point to the new type. In the
1384 example above, this allows the code to maintain a pointer to the resultant
1385 resolved recursive type, even though the Type*'s are potentially invalidated.
1389 PATypeHolder is an extremely light-weight object that uses a lazy union-find
1390 implementation to update pointers. For example the pointer from a Value to its
1391 Type is maintained by PATypeHolder objects.
1396 <!-- ______________________________________________________________________ -->
1397 <div class="doc_subsubsection">
1398 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
1401 <div class="doc_text">
1404 Some data structures need more to perform more complex updates when types get
1405 resolved. The <a href="#SymbolTable">SymbolTable</a> class, for example, needs
1406 move and potentially merge type planes in its representation when a pointer
1410 To support this, a class can derive from the AbstractTypeUser class. This class
1411 allows it to get callbacks when certain types are resolved. To register to get
1412 callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
1413 methods can be called on a type. Note that these methods only work for <i>
1414 abstract</i> types. Concrete types (those that do not include an opaque objects
1415 somewhere) can never be refined.
1420 <!-- ======================================================================= -->
1421 <div class="doc_subsection">
1422 <a name="SymbolTable">The <tt>SymbolTable</tt> class</a>
1425 <div class="doc_text">
1426 <p>This class provides a symbol table that the <a
1427 href="#Function"><tt>Function</tt></a> and <a href="#Module">
1428 <tt>Module</tt></a> classes use for naming definitions. The symbol table can
1429 provide a name for any <a href="#Value"><tt>Value</tt></a> or <a
1430 href="#Type"><tt>Type</tt></a>. <tt>SymbolTable</tt> is an abstract data
1431 type. It hides the data it contains and provides access to it through a
1432 controlled interface.</p>
1434 <p>Note that the symbol table class is should not be directly accessed by most
1435 clients. It should only be used when iteration over the symbol table names
1436 themselves are required, which is very special purpose. Note that not all LLVM
1437 <a href="#Value">Value</a>s have names, and those without names (i.e. they have
1438 an empty name) do not exist in the symbol table.
1441 <p>To use the <tt>SymbolTable</tt> well, you need to understand the
1442 structure of the information it holds. The class contains two
1443 <tt>std::map</tt> objects. The first, <tt>pmap</tt>, is a map of
1444 <tt>Type*</tt> to maps of name (<tt>std::string</tt>) to <tt>Value*</tt>.
1445 The second, <tt>tmap</tt>, is a map of names to <tt>Type*</tt>. Thus, Values
1446 are stored in two-dimensions and accessed by <tt>Type</tt> and name. Types,
1447 however, are stored in a single dimension and accessed only by name.</p>
1449 <p>The interface of this class provides three basic types of operations:
1451 <li><em>Accessors</em>. Accessors provide read-only access to information
1452 such as finding a value for a name with the
1453 <a href="#SymbolTable_lookup">lookup</a> method.</li>
1454 <li><em>Mutators</em>. Mutators allow the user to add information to the
1455 <tt>SymbolTable</tt> with methods like
1456 <a href="#SymbolTable_insert"><tt>insert</tt></a>.</li>
1457 <li><em>Iterators</em>. Iterators allow the user to traverse the content
1458 of the symbol table in well defined ways, such as the method
1459 <a href="#SymbolTable_type_begin"><tt>type_begin</tt></a>.</li>
1464 <dt><tt>Value* lookup(const Type* Ty, const std::string& name) const</tt>:
1466 <dd>The <tt>lookup</tt> method searches the type plane given by the
1467 <tt>Ty</tt> parameter for a <tt>Value</tt> with the provided <tt>name</tt>.
1468 If a suitable <tt>Value</tt> is not found, null is returned.</dd>
1470 <dt><tt>Type* lookupType( const std::string& name) const</tt>:</dt>
1471 <dd>The <tt>lookupType</tt> method searches through the types for a
1472 <tt>Type</tt> with the provided <tt>name</tt>. If a suitable <tt>Type</tt>
1473 is not found, null is returned.</dd>
1475 <dt><tt>bool hasTypes() const</tt>:</dt>
1476 <dd>This function returns true if an entry has been made into the type
1479 <dt><tt>bool isEmpty() const</tt>:</dt>
1480 <dd>This function returns true if both the value and types maps are
1486 <dt><tt>void insert(Value *Val)</tt>:</dt>
1487 <dd>This method adds the provided value to the symbol table. The Value must
1488 have both a name and a type which are extracted and used to place the value
1489 in the correct type plane under the value's name.</dd>
1491 <dt><tt>void insert(const std::string& Name, Value *Val)</tt>:</dt>
1492 <dd> Inserts a constant or type into the symbol table with the specified
1493 name. There can be a many to one mapping between names and constants
1496 <dt><tt>void insert(const std::string& Name, Type *Typ)</tt>:</dt>
1497 <dd> Inserts a type into the symbol table with the specified name. There
1498 can be a many-to-one mapping between names and types. This method
1499 allows a type with an existing entry in the symbol table to get
1502 <dt><tt>void remove(Value* Val)</tt>:</dt>
1503 <dd> This method removes a named value from the symbol table. The
1504 type and name of the Value are extracted from \p N and used to
1505 lookup the Value in the correct type plane. If the Value is
1506 not in the symbol table, this method silently ignores the
1509 <dt><tt>void remove(Type* Typ)</tt>:</dt>
1510 <dd> This method removes a named type from the symbol table. The
1511 name of the type is extracted from \P T and used to look up
1512 the Type in the type map. If the Type is not in the symbol
1513 table, this method silently ignores the request.</dd>
1515 <dt><tt>Value* remove(const std::string& Name, Value *Val)</tt>:</dt>
1516 <dd> Remove a constant or type with the specified name from the
1519 <dt><tt>Type* remove(const std::string& Name, Type* T)</tt>:</dt>
1520 <dd> Remove a type with the specified name from the symbol table.
1521 Returns the removed Type.</dd>
1523 <dt><tt>Value *value_remove(const value_iterator& It)</tt>:</dt>
1524 <dd> Removes a specific value from the symbol table.
1525 Returns the removed value.</dd>
1527 <dt><tt>bool strip()</tt>:</dt>
1528 <dd> This method will strip the symbol table of its names leaving
1529 the type and values. </dd>
1531 <dt><tt>void clear()</tt>:</dt>
1532 <dd>Empty the symbol table completely.</dd>
1536 <p>The following functions describe three types of iterators you can obtain
1537 the beginning or end of the sequence for both const and non-const. It is
1538 important to keep track of the different kinds of iterators. There are
1539 three idioms worth pointing out:</p>
1542 <tr><th>Units</th><th>Iterator</th><th>Idiom</th></tr>
1544 <td align="left">Planes Of name/Value maps</td><td>PI</td>
1545 <td align="left"><pre><tt>
1546 for (SymbolTable::plane_const_iterator PI = ST.plane_begin(),
1547 PE = ST.plane_end(); PI != PE; ++PI ) {
1548 PI->first // This is the Type* of the plane
1549 PI->second // This is the SymbolTable::ValueMap of name/Value pairs
1554 <td align="left">All name/Type Pairs</td><td>TI</td>
1555 <td align="left"><pre><tt>
1556 for (SymbolTable::type_const_iterator TI = ST.type_begin(),
1557 TE = ST.type_end(); TI != TE; ++TI ) {
1558 TI->first // This is the name of the type
1559 TI->second // This is the Type* value associated with the name
1564 <td align="left">name/Value pairs in a plane</td><td>VI</td>
1565 <td align="left"><pre><tt>
1566 for (SymbolTable::value_const_iterator VI = ST.value_begin(SomeType),
1567 VE = ST.value_end(SomeType); VI != VE; ++VI ) {
1568 VI->first // This is the name of the Value
1569 VI->second // This is the Value* value associated with the name
1575 <p>Using the recommended iterator names and idioms will help you avoid
1576 making mistakes. Of particular note, make sure that whenever you use
1577 value_begin(SomeType) that you always compare the resulting iterator
1578 with value_end(SomeType) not value_end(SomeOtherType) or else you
1579 will loop infinitely.</p>
1583 <dt><tt>plane_iterator plane_begin()</tt>:</dt>
1584 <dd>Get an iterator that starts at the beginning of the type planes.
1585 The iterator will iterate over the Type/ValueMap pairs in the
1588 <dt><tt>plane_const_iterator plane_begin() const</tt>:</dt>
1589 <dd>Get a const_iterator that starts at the beginning of the type
1590 planes. The iterator will iterate over the Type/ValueMap pairs
1591 in the type planes. </dd>
1593 <dt><tt>plane_iterator plane_end()</tt>:</dt>
1594 <dd>Get an iterator at the end of the type planes. This serves as
1595 the marker for end of iteration over the type planes.</dd>
1597 <dt><tt>plane_const_iterator plane_end() const</tt>:</dt>
1598 <dd>Get a const_iterator at the end of the type planes. This serves as
1599 the marker for end of iteration over the type planes.</dd>
1601 <dt><tt>value_iterator value_begin(const Type *Typ)</tt>:</dt>
1602 <dd>Get an iterator that starts at the beginning of a type plane.
1603 The iterator will iterate over the name/value pairs in the type plane.
1604 Note: The type plane must already exist before using this.</dd>
1606 <dt><tt>value_const_iterator value_begin(const Type *Typ) const</tt>:</dt>
1607 <dd>Get a const_iterator that starts at the beginning of a type plane.
1608 The iterator will iterate over the name/value pairs in the type plane.
1609 Note: The type plane must already exist before using this.</dd>
1611 <dt><tt>value_iterator value_end(const Type *Typ)</tt>:</dt>
1612 <dd>Get an iterator to the end of a type plane. This serves as the marker
1613 for end of iteration of the type plane.
1614 Note: The type plane must already exist before using this.</dd>
1616 <dt><tt>value_const_iterator value_end(const Type *Typ) const</tt>:</dt>
1617 <dd>Get a const_iterator to the end of a type plane. This serves as the
1618 marker for end of iteration of the type plane.
1619 Note: the type plane must already exist before using this.</dd>
1621 <dt><tt>type_iterator type_begin()</tt>:</dt>
1622 <dd>Get an iterator to the start of the name/Type map.</dd>
1624 <dt><tt>type_const_iterator type_begin() cons</tt>:</dt>
1625 <dd> Get a const_iterator to the start of the name/Type map.</dd>
1627 <dt><tt>type_iterator type_end()</tt>:</dt>
1628 <dd>Get an iterator to the end of the name/Type map. This serves as the
1629 marker for end of iteration of the types.</dd>
1631 <dt><tt>type_const_iterator type_end() const</tt>:</dt>
1632 <dd>Get a const-iterator to the end of the name/Type map. This serves
1633 as the marker for end of iteration of the types.</dd>
1635 <dt><tt>plane_const_iterator find(const Type* Typ ) const</tt>:</dt>
1636 <dd>This method returns a plane_const_iterator for iteration over
1637 the type planes starting at a specific plane, given by \p Ty.</dd>
1639 <dt><tt>plane_iterator find( const Type* Typ </tt>:</dt>
1640 <dd>This method returns a plane_iterator for iteration over the
1641 type planes starting at a specific plane, given by \p Ty.</dd>
1648 <!-- *********************************************************************** -->
1649 <div class="doc_section">
1650 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
1652 <!-- *********************************************************************** -->
1654 <div class="doc_text">
1656 <p>The Core LLVM classes are the primary means of representing the program
1657 being inspected or transformed. The core LLVM classes are defined in
1658 header files in the <tt>include/llvm/</tt> directory, and implemented in
1659 the <tt>lib/VMCore</tt> directory.</p>
1663 <!-- ======================================================================= -->
1664 <div class="doc_subsection">
1665 <a name="Value">The <tt>Value</tt> class</a>
1670 <p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
1672 doxygen info: <a href="/doxygen/structllvm_1_1Value.html">Value Class</a></p>
1674 <p>The <tt>Value</tt> class is the most important class in the LLVM Source
1675 base. It represents a typed value that may be used (among other things) as an
1676 operand to an instruction. There are many different types of <tt>Value</tt>s,
1677 such as <a href="#Constant"><tt>Constant</tt></a>s,<a
1678 href="#Argument"><tt>Argument</tt></a>s. Even <a
1679 href="#Instruction"><tt>Instruction</tt></a>s and <a
1680 href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
1682 <p>A particular <tt>Value</tt> may be used many times in the LLVM representation
1683 for a program. For example, an incoming argument to a function (represented
1684 with an instance of the <a href="#Argument">Argument</a> class) is "used" by
1685 every instruction in the function that references the argument. To keep track
1686 of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
1687 href="#User"><tt>User</tt></a>s that is using it (the <a
1688 href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
1689 graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
1690 def-use information in the program, and is accessible through the <tt>use_</tt>*
1691 methods, shown below.</p>
1693 <p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
1694 and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
1695 method. In addition, all LLVM values can be named. The "name" of the
1696 <tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
1698 <div class="doc_code">
1700 %<b>foo</b> = add int 1, 2
1704 <p><a name="#nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
1705 that the name of any value may be missing (an empty string), so names should
1706 <b>ONLY</b> be used for debugging (making the source code easier to read,
1707 debugging printouts), they should not be used to keep track of values or map
1708 between them. For this purpose, use a <tt>std::map</tt> of pointers to the
1709 <tt>Value</tt> itself instead.</p>
1711 <p>One important aspect of LLVM is that there is no distinction between an SSA
1712 variable and the operation that produces it. Because of this, any reference to
1713 the value produced by an instruction (or the value available as an incoming
1714 argument, for example) is represented as a direct pointer to the instance of
1716 represents this value. Although this may take some getting used to, it
1717 simplifies the representation and makes it easier to manipulate.</p>
1721 <!-- _______________________________________________________________________ -->
1722 <div class="doc_subsubsection">
1723 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
1726 <div class="doc_text">
1729 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
1731 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
1733 <tt>unsigned use_size()</tt> - Returns the number of users of the
1735 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
1736 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
1738 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
1740 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
1741 element in the list.
1742 <p> These methods are the interface to access the def-use
1743 information in LLVM. As with all other iterators in LLVM, the naming
1744 conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
1746 <li><tt><a href="#Type">Type</a> *getType() const</tt>
1747 <p>This method returns the Type of the Value.</p>
1749 <li><tt>bool hasName() const</tt><br>
1750 <tt>std::string getName() const</tt><br>
1751 <tt>void setName(const std::string &Name)</tt>
1752 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
1753 be aware of the <a href="#nameWarning">precaution above</a>.</p>
1755 <li><tt>void replaceAllUsesWith(Value *V)</tt>
1757 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
1758 href="#User"><tt>User</tt>s</a> of the current value to refer to
1759 "<tt>V</tt>" instead. For example, if you detect that an instruction always
1760 produces a constant value (for example through constant folding), you can
1761 replace all uses of the instruction with the constant like this:</p>
1763 <div class="doc_code">
1765 Inst->replaceAllUsesWith(ConstVal);
1773 <!-- ======================================================================= -->
1774 <div class="doc_subsection">
1775 <a name="User">The <tt>User</tt> class</a>
1778 <div class="doc_text">
1781 <tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
1782 doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
1783 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1785 <p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
1786 refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
1787 that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
1788 referring to. The <tt>User</tt> class itself is a subclass of
1791 <p>The operands of a <tt>User</tt> point directly to the LLVM <a
1792 href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
1793 Single Assignment (SSA) form, there can only be one definition referred to,
1794 allowing this direct connection. This connection provides the use-def
1795 information in LLVM.</p>
1799 <!-- _______________________________________________________________________ -->
1800 <div class="doc_subsubsection">
1801 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
1804 <div class="doc_text">
1806 <p>The <tt>User</tt> class exposes the operand list in two ways: through
1807 an index access interface and through an iterator based interface.</p>
1810 <li><tt>Value *getOperand(unsigned i)</tt><br>
1811 <tt>unsigned getNumOperands()</tt>
1812 <p> These two methods expose the operands of the <tt>User</tt> in a
1813 convenient form for direct access.</p></li>
1815 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
1817 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
1818 the operand list.<br>
1819 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
1821 <p> Together, these methods make up the iterator based interface to
1822 the operands of a <tt>User</tt>.</p></li>
1827 <!-- ======================================================================= -->
1828 <div class="doc_subsection">
1829 <a name="Instruction">The <tt>Instruction</tt> class</a>
1832 <div class="doc_text">
1834 <p><tt>#include "</tt><tt><a
1835 href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
1836 doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
1837 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1838 href="#Value"><tt>Value</tt></a></p>
1840 <p>The <tt>Instruction</tt> class is the common base class for all LLVM
1841 instructions. It provides only a few methods, but is a very commonly used
1842 class. The primary data tracked by the <tt>Instruction</tt> class itself is the
1843 opcode (instruction type) and the parent <a
1844 href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
1845 into. To represent a specific type of instruction, one of many subclasses of
1846 <tt>Instruction</tt> are used.</p>
1848 <p> Because the <tt>Instruction</tt> class subclasses the <a
1849 href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
1850 way as for other <a href="#User"><tt>User</tt></a>s (with the
1851 <tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
1852 <tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
1853 the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
1854 file contains some meta-data about the various different types of instructions
1855 in LLVM. It describes the enum values that are used as opcodes (for example
1856 <tt>Instruction::Add</tt> and <tt>Instruction::SetLE</tt>), as well as the
1857 concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
1858 example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
1859 href="#SetCondInst">SetCondInst</a></tt>). Unfortunately, the use of macros in
1860 this file confuses doxygen, so these enum values don't show up correctly in the
1861 <a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
1865 <!-- _______________________________________________________________________ -->
1866 <div class="doc_subsubsection">
1867 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
1871 <div class="doc_text">
1874 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
1875 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
1876 this <tt>Instruction</tt> is embedded into.</p></li>
1877 <li><tt>bool mayWriteToMemory()</tt>
1878 <p>Returns true if the instruction writes to memory, i.e. it is a
1879 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
1880 <li><tt>unsigned getOpcode()</tt>
1881 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
1882 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
1883 <p>Returns another instance of the specified instruction, identical
1884 in all ways to the original except that the instruction has no parent
1885 (ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
1886 and it has no name</p></li>
1891 <!-- ======================================================================= -->
1892 <div class="doc_subsection">
1893 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
1896 <div class="doc_text">
1899 href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
1900 doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
1902 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1904 <p>This class represents a single entry multiple exit section of the code,
1905 commonly known as a basic block by the compiler community. The
1906 <tt>BasicBlock</tt> class maintains a list of <a
1907 href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
1908 Matching the language definition, the last element of this list of instructions
1909 is always a terminator instruction (a subclass of the <a
1910 href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
1912 <p>In addition to tracking the list of instructions that make up the block, the
1913 <tt>BasicBlock</tt> class also keeps track of the <a
1914 href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
1916 <p>Note that <tt>BasicBlock</tt>s themselves are <a
1917 href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
1918 like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
1923 <!-- _______________________________________________________________________ -->
1924 <div class="doc_subsubsection">
1925 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
1929 <div class="doc_text">
1933 <li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
1934 href="#Function">Function</a> *Parent = 0)</tt>
1936 <p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
1937 insertion into a function. The constructor optionally takes a name for the new
1938 block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
1939 the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
1940 automatically inserted at the end of the specified <a
1941 href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
1942 manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
1944 <li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
1945 <tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
1946 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
1947 <tt>size()</tt>, <tt>empty()</tt>
1948 STL-style functions for accessing the instruction list.
1950 <p>These methods and typedefs are forwarding functions that have the same
1951 semantics as the standard library methods of the same names. These methods
1952 expose the underlying instruction list of a basic block in a way that is easy to
1953 manipulate. To get the full complement of container operations (including
1954 operations to update the list), you must use the <tt>getInstList()</tt>
1957 <li><tt>BasicBlock::InstListType &getInstList()</tt>
1959 <p>This method is used to get access to the underlying container that actually
1960 holds the Instructions. This method must be used when there isn't a forwarding
1961 function in the <tt>BasicBlock</tt> class for the operation that you would like
1962 to perform. Because there are no forwarding functions for "updating"
1963 operations, you need to use this if you want to update the contents of a
1964 <tt>BasicBlock</tt>.</p></li>
1966 <li><tt><a href="#Function">Function</a> *getParent()</tt>
1968 <p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
1969 embedded into, or a null pointer if it is homeless.</p></li>
1971 <li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
1973 <p> Returns a pointer to the terminator instruction that appears at the end of
1974 the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
1975 instruction in the block is not a terminator, then a null pointer is
1982 <!-- ======================================================================= -->
1983 <div class="doc_subsection">
1984 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
1987 <div class="doc_text">
1990 href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
1991 doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
1993 Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
1994 <a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1996 <p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
1997 href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
1998 visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
1999 Because they are visible at global scope, they are also subject to linking with
2000 other globals defined in different translation units. To control the linking
2001 process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
2002 <tt>GlobalValue</tt>s know whether they have internal or external linkage, as
2003 defined by the <tt>LinkageTypes</tt> enumeration.</p>
2005 <p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
2006 <tt>static</tt> in C), it is not visible to code outside the current translation
2007 unit, and does not participate in linking. If it has external linkage, it is
2008 visible to external code, and does participate in linking. In addition to
2009 linkage information, <tt>GlobalValue</tt>s keep track of which <a
2010 href="#Module"><tt>Module</tt></a> they are currently part of.</p>
2012 <p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
2013 by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
2014 global is always a pointer to its contents. It is important to remember this
2015 when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
2016 be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
2017 subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
2018 int]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
2019 the address of the first element of this array and the value of the
2020 <tt>GlobalVariable</tt> are the same, they have different types. The
2021 <tt>GlobalVariable</tt>'s type is <tt>[24 x int]</tt>. The first element's type
2022 is <tt>int.</tt> Because of this, accessing a global value requires you to
2023 dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
2024 can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
2025 Language Reference Manual</a>.</p>
2029 <!-- _______________________________________________________________________ -->
2030 <div class="doc_subsubsection">
2031 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
2035 <div class="doc_text">
2038 <li><tt>bool hasInternalLinkage() const</tt><br>
2039 <tt>bool hasExternalLinkage() const</tt><br>
2040 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
2041 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
2044 <li><tt><a href="#Module">Module</a> *getParent()</tt>
2045 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
2046 GlobalValue is currently embedded into.</p></li>
2051 <!-- ======================================================================= -->
2052 <div class="doc_subsection">
2053 <a name="Function">The <tt>Function</tt> class</a>
2056 <div class="doc_text">
2059 href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
2060 info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
2061 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2062 <a href="#Constant"><tt>Constant</tt></a>,
2063 <a href="#User"><tt>User</tt></a>,
2064 <a href="#Value"><tt>Value</tt></a></p>
2066 <p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
2067 actually one of the more complex classes in the LLVM heirarchy because it must
2068 keep track of a large amount of data. The <tt>Function</tt> class keeps track
2069 of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
2070 <a href="#Argument"><tt>Argument</tt></a>s, and a
2071 <a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
2073 <p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
2074 commonly used part of <tt>Function</tt> objects. The list imposes an implicit
2075 ordering of the blocks in the function, which indicate how the code will be
2076 layed out by the backend. Additionally, the first <a
2077 href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
2078 <tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
2079 block. There are no implicit exit nodes, and in fact there may be multiple exit
2080 nodes from a single <tt>Function</tt>. If the <a
2081 href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
2082 the <tt>Function</tt> is actually a function declaration: the actual body of the
2083 function hasn't been linked in yet.</p>
2085 <p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
2086 <tt>Function</tt> class also keeps track of the list of formal <a
2087 href="#Argument"><tt>Argument</tt></a>s that the function receives. This
2088 container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
2089 nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
2090 the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
2092 <p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
2093 LLVM feature that is only used when you have to look up a value by name. Aside
2094 from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
2095 internally to make sure that there are not conflicts between the names of <a
2096 href="#Instruction"><tt>Instruction</tt></a>s, <a
2097 href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
2098 href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
2100 <p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
2101 and therefore also a <a href="#Constant">Constant</a>. The value of the function
2102 is its address (after linking) which is guaranteed to be constant.</p>
2105 <!-- _______________________________________________________________________ -->
2106 <div class="doc_subsubsection">
2107 <a name="m_Function">Important Public Members of the <tt>Function</tt>
2111 <div class="doc_text">
2114 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
2115 *Ty, LinkageTypes Linkage, const std::string &N = "", Module* Parent = 0)</tt>
2117 <p>Constructor used when you need to create new <tt>Function</tt>s to add
2118 the the program. The constructor must specify the type of the function to
2119 create and what type of linkage the function should have. The <a
2120 href="#FunctionType"><tt>FunctionType</tt></a> argument
2121 specifies the formal arguments and return value for the function. The same
2122 <a href="#FunctionTypel"><tt>FunctionType</tt></a> value can be used to
2123 create multiple functions. The <tt>Parent</tt> argument specifies the Module
2124 in which the function is defined. If this argument is provided, the function
2125 will automatically be inserted into that module's list of
2128 <li><tt>bool isExternal()</tt>
2130 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
2131 function is "external", it does not have a body, and thus must be resolved
2132 by linking with a function defined in a different translation unit.</p></li>
2134 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
2135 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
2137 <tt>begin()</tt>, <tt>end()</tt>
2138 <tt>size()</tt>, <tt>empty()</tt>
2140 <p>These are forwarding methods that make it easy to access the contents of
2141 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
2144 <li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
2146 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
2147 is necessary to use when you need to update the list or perform a complex
2148 action that doesn't have a forwarding method.</p></li>
2150 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
2152 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
2154 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
2155 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
2157 <p>These are forwarding methods that make it easy to access the contents of
2158 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
2161 <li><tt>Function::ArgumentListType &getArgumentList()</tt>
2163 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
2164 necessary to use when you need to update the list or perform a complex
2165 action that doesn't have a forwarding method.</p></li>
2167 <li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
2169 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
2170 function. Because the entry block for the function is always the first
2171 block, this returns the first block of the <tt>Function</tt>.</p></li>
2173 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
2174 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
2176 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
2177 <tt>Function</tt> and returns the return type of the function, or the <a
2178 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
2181 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2183 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2184 for this <tt>Function</tt>.</p></li>
2189 <!-- ======================================================================= -->
2190 <div class="doc_subsection">
2191 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
2194 <div class="doc_text">
2197 href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
2199 doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
2201 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2202 <a href="#Constant"><tt>Constant</tt></a>,
2203 <a href="#User"><tt>User</tt></a>,
2204 <a href="#Value"><tt>Value</tt></a></p>
2206 <p>Global variables are represented with the (suprise suprise)
2207 <tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
2208 subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
2209 always referenced by their address (global values must live in memory, so their
2210 "name" refers to their constant address). See
2211 <a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
2212 variables may have an initial value (which must be a
2213 <a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
2214 they may be marked as "constant" themselves (indicating that their contents
2215 never change at runtime).</p>
2218 <!-- _______________________________________________________________________ -->
2219 <div class="doc_subsubsection">
2220 <a name="m_GlobalVariable">Important Public Members of the
2221 <tt>GlobalVariable</tt> class</a>
2224 <div class="doc_text">
2227 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
2228 isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
2229 *Initializer = 0, const std::string &Name = "", Module* Parent = 0)</tt>
2231 <p>Create a new global variable of the specified type. If
2232 <tt>isConstant</tt> is true then the global variable will be marked as
2233 unchanging for the program. The Linkage parameter specifies the type of
2234 linkage (internal, external, weak, linkonce, appending) for the variable. If
2235 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage, then
2236 the resultant global variable will have internal linkage. AppendingLinkage
2237 concatenates together all instances (in different translation units) of the
2238 variable into a single variable but is only applicable to arrays. See
2239 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
2240 further details on linkage types. Optionally an initializer, a name, and the
2241 module to put the variable into may be specified for the global variable as
2244 <li><tt>bool isConstant() const</tt>
2246 <p>Returns true if this is a global variable that is known not to
2247 be modified at runtime.</p></li>
2249 <li><tt>bool hasInitializer()</tt>
2251 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
2253 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
2255 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
2256 to call this method if there is no initializer.</p></li>
2261 <!-- ======================================================================= -->
2262 <div class="doc_subsection">
2263 <a name="Module">The <tt>Module</tt> class</a>
2266 <div class="doc_text">
2269 href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
2270 <a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
2272 <p>The <tt>Module</tt> class represents the top level structure present in LLVM
2273 programs. An LLVM module is effectively either a translation unit of the
2274 original program or a combination of several translation units merged by the
2275 linker. The <tt>Module</tt> class keeps track of a list of <a
2276 href="#Function"><tt>Function</tt></a>s, a list of <a
2277 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
2278 href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
2279 helpful member functions that try to make common operations easy.</p>
2283 <!-- _______________________________________________________________________ -->
2284 <div class="doc_subsubsection">
2285 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
2288 <div class="doc_text">
2291 <li><tt>Module::Module(std::string name = "")</tt></li>
2294 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
2295 provide a name for it (probably based on the name of the translation unit).</p>
2298 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
2299 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
2301 <tt>begin()</tt>, <tt>end()</tt>
2302 <tt>size()</tt>, <tt>empty()</tt>
2304 <p>These are forwarding methods that make it easy to access the contents of
2305 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
2308 <li><tt>Module::FunctionListType &getFunctionList()</tt>
2310 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
2311 necessary to use when you need to update the list or perform a complex
2312 action that doesn't have a forwarding method.</p>
2314 <p><!-- Global Variable --></p></li>
2320 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
2322 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
2324 <tt>global_begin()</tt>, <tt>global_end()</tt>
2325 <tt>global_size()</tt>, <tt>global_empty()</tt>
2327 <p> These are forwarding methods that make it easy to access the contents of
2328 a <tt>Module</tt> object's <a
2329 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
2331 <li><tt>Module::GlobalListType &getGlobalList()</tt>
2333 <p>Returns the list of <a
2334 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
2335 use when you need to update the list or perform a complex action that
2336 doesn't have a forwarding method.</p>
2338 <p><!-- Symbol table stuff --> </p></li>
2344 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2346 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2347 for this <tt>Module</tt>.</p>
2349 <p><!-- Convenience methods --></p></li>
2355 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
2356 &Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
2358 <p>Look up the specified function in the <tt>Module</tt> <a
2359 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
2360 <tt>null</tt>.</p></li>
2362 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
2363 std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
2365 <p>Look up the specified function in the <tt>Module</tt> <a
2366 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
2367 external declaration for the function and return it.</p></li>
2369 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
2371 <p>If there is at least one entry in the <a
2372 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
2373 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
2376 <li><tt>bool addTypeName(const std::string &Name, const <a
2377 href="#Type">Type</a> *Ty)</tt>
2379 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2380 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
2381 name, true is returned and the <a
2382 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
2387 <!-- ======================================================================= -->
2388 <div class="doc_subsection">
2389 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
2392 <div class="doc_text">
2394 <p>Constant represents a base class for different types of constants. It
2395 is subclassed by ConstantBool, ConstantInt, ConstantSInt, ConstantUInt,
2396 ConstantArray etc for representing the various types of Constants.</p>
2400 <!-- _______________________________________________________________________ -->
2401 <div class="doc_subsubsection">
2402 <a name="m_Constant">Important Public Methods</a>
2404 <div class="doc_text">
2407 <!-- _______________________________________________________________________ -->
2408 <div class="doc_subsubsection">Important Subclasses of Constant </div>
2409 <div class="doc_text">
2411 <li>ConstantSInt : This subclass of Constant represents a signed integer
2414 <li><tt>int64_t getValue() const</tt>: Returns the underlying value of
2415 this constant. </li>
2418 <li>ConstantUInt : This class represents an unsigned integer.
2420 <li><tt>uint64_t getValue() const</tt>: Returns the underlying value of
2421 this constant. </li>
2424 <li>ConstantFP : This class represents a floating point constant.
2426 <li><tt>double getValue() const</tt>: Returns the underlying value of
2427 this constant. </li>
2430 <li>ConstantBool : This represents a boolean constant.
2432 <li><tt>bool getValue() const</tt>: Returns the underlying value of this
2436 <li>ConstantArray : This represents a constant array.
2438 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
2439 a vector of component constants that makeup this array. </li>
2442 <li>ConstantStruct : This represents a constant struct.
2444 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
2445 a vector of component constants that makeup this array. </li>
2448 <li>GlobalValue : This represents either a global variable or a function. In
2449 either case, the value is a constant fixed address (after linking).
2454 <!-- ======================================================================= -->
2455 <div class="doc_subsection">
2456 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
2459 <div class="doc_text">
2461 <p>Type as noted earlier is also a subclass of a Value class. Any primitive
2462 type (like int, short etc) in LLVM is an instance of Type Class. All other
2463 types are instances of subclasses of type like FunctionType, ArrayType
2464 etc. DerivedType is the interface for all such dervied types including
2465 FunctionType, ArrayType, PointerType, StructType. Types can have names. They can
2466 be recursive (StructType). There exists exactly one instance of any type
2467 structure at a time. This allows using pointer equality of Type *s for comparing
2472 <!-- _______________________________________________________________________ -->
2473 <div class="doc_subsubsection">
2474 <a name="m_Value">Important Public Methods</a>
2477 <div class="doc_text">
2481 <li><tt>bool isSigned() const</tt>: Returns whether an integral numeric type
2482 is signed. This is true for SByteTy, ShortTy, IntTy, LongTy. Note that this is
2483 not true for Float and Double. </li>
2485 <li><tt>bool isUnsigned() const</tt>: Returns whether a numeric type is
2486 unsigned. This is not quite the complement of isSigned... nonnumeric types
2487 return false as they do with isSigned. This returns true for UByteTy,
2488 UShortTy, UIntTy, and ULongTy. </li>
2490 <li><tt>bool isInteger() const</tt>: Equivalent to isSigned() || isUnsigned().</li>
2492 <li><tt>bool isIntegral() const</tt>: Returns true if this is an integral
2493 type, which is either Bool type or one of the Integer types.</li>
2495 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
2496 floating point types.</li>
2498 <li><tt>isLosslesslyConvertableTo (const Type *Ty) const</tt>: Return true if
2499 this type can be converted to 'Ty' without any reinterpretation of bits. For
2500 example, uint to int or one pointer type to another.</li>
2504 <!-- _______________________________________________________________________ -->
2505 <div class="doc_subsubsection">
2506 <a name="m_Value">Important Derived Types</a>
2508 <div class="doc_text">
2510 <li>SequentialType : This is subclassed by ArrayType and PointerType
2512 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
2513 of the elements in the sequential type. </li>
2516 <li>ArrayType : This is a subclass of SequentialType and defines interface for
2519 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
2520 elements in the array. </li>
2523 <li>PointerType : Subclass of SequentialType for pointer types. </li>
2524 <li>StructType : subclass of DerivedTypes for struct types </li>
2525 <li>FunctionType : subclass of DerivedTypes for function types.
2527 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
2529 <li><tt> const Type * getReturnType() const</tt>: Returns the
2530 return type of the function.</li>
2531 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
2532 the type of the ith parameter.</li>
2533 <li><tt> const unsigned getNumParams() const</tt>: Returns the
2534 number of formal parameters.</li>
2540 <!-- ======================================================================= -->
2541 <div class="doc_subsection">
2542 <a name="Argument">The <tt>Argument</tt> class</a>
2545 <div class="doc_text">
2547 <p>This subclass of Value defines the interface for incoming formal
2548 arguments to a function. A Function maintains a list of its formal
2549 arguments. An argument has a pointer to the parent Function.</p>
2553 <!-- *********************************************************************** -->
2556 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2557 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
2558 <a href="http://validator.w3.org/check/referer"><img
2559 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
2561 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
2562 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
2563 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
2564 Last modified: $Date$