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 & <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
46 <li><a href="#common">Helpful Hints for Common Operations</a>
48 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
50 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
51 in a <tt>Function</tt></a> </li>
52 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
53 in a <tt>BasicBlock</tt></a> </li>
54 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
55 in a <tt>Function</tt></a> </li>
56 <li><a href="#iterate_convert">Turning an iterator into a
57 class pointer</a> </li>
58 <li><a href="#iterate_complex">Finding call sites: a more
59 complex example</a> </li>
60 <li><a href="#calls_and_invokes">Treating calls and invokes
61 the same way</a> </li>
62 <li><a href="#iterate_chains">Iterating over def-use &
63 use-def chains</a> </li>
66 <li><a href="#simplechanges">Making simple changes</a>
68 <li><a href="#schanges_creating">Creating and inserting new
69 <tt>Instruction</tt>s</a> </li>
70 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
71 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
72 with another <tt>Value</tt></a> </li>
76 <li>Working with the Control Flow Graph
78 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
85 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
87 <li><a href="#Value">The <tt>Value</tt> class</a>
89 <li><a href="#User">The <tt>User</tt> class</a>
91 <li><a href="#Instruction">The <tt>Instruction</tt> class</a>
93 <li><a href="#GetElementPtrInst">The <tt>GetElementPtrInst</tt> class</a></li>
96 <li><a href="#Module">The <tt>Module</tt> class</a></li>
97 <li><a href="#Constant">The <tt>Constant</tt> class</a>
99 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
101 <li><a href="#BasicBlock">The <tt>BasicBlock</tt>class</a></li>
102 <li><a href="#Function">The <tt>Function</tt> class</a></li>
103 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
110 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
111 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
116 <li><a href="#SymbolTable">The <tt>SymbolTable</tt> class </a></li>
117 <li>The <tt>ilist</tt> and <tt>iplist</tt> classes
119 <li>Creating, inserting, moving and deleting from LLVM lists </li>
122 <li>Important iterator invalidation semantics to be aware of.</li>
125 <div class="doc_author">
126 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
127 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
128 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
129 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
132 <!-- *********************************************************************** -->
133 <div class="doc_section">
134 <a name="introduction">Introduction </a>
136 <!-- *********************************************************************** -->
138 <div class="doc_text">
140 <p>This document is meant to highlight some of the important classes and
141 interfaces available in the LLVM source-base. This manual is not
142 intended to explain what LLVM is, how it works, and what LLVM code looks
143 like. It assumes that you know the basics of LLVM and are interested
144 in writing transformations or otherwise analyzing or manipulating the
147 <p>This document should get you oriented so that you can find your
148 way in the continuously growing source code that makes up the LLVM
149 infrastructure. Note that this manual is not intended to serve as a
150 replacement for reading the source code, so if you think there should be
151 a method in one of these classes to do something, but it's not listed,
152 check the source. Links to the <a href="/doxygen/">doxygen</a> sources
153 are provided to make this as easy as possible.</p>
155 <p>The first section of this document describes general information that is
156 useful to know when working in the LLVM infrastructure, and the second describes
157 the Core LLVM classes. In the future this manual will be extended with
158 information describing how to use extension libraries, such as dominator
159 information, CFG traversal routines, and useful utilities like the <tt><a
160 href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
164 <!-- *********************************************************************** -->
165 <div class="doc_section">
166 <a name="general">General Information</a>
168 <!-- *********************************************************************** -->
170 <div class="doc_text">
172 <p>This section contains general information that is useful if you are working
173 in the LLVM source-base, but that isn't specific to any particular API.</p>
177 <!-- ======================================================================= -->
178 <div class="doc_subsection">
179 <a name="stl">The C++ Standard Template Library</a>
182 <div class="doc_text">
184 <p>LLVM makes heavy use of the C++ Standard Template Library (STL),
185 perhaps much more than you are used to, or have seen before. Because of
186 this, you might want to do a little background reading in the
187 techniques used and capabilities of the library. There are many good
188 pages that discuss the STL, and several books on the subject that you
189 can get, so it will not be discussed in this document.</p>
191 <p>Here are some useful links:</p>
195 <li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
196 reference</a> - an excellent reference for the STL and other parts of the
197 standard C++ library.</li>
199 <li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
200 O'Reilly book in the making. It has a decent
202 Reference that rivals Dinkumware's, and is unfortunately no longer free since the book has been
205 <li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
208 <li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
210 href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
213 <li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
216 <li><a href="http://64.78.49.204/">
217 Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
222 <p>You are also encouraged to take a look at the <a
223 href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
224 to write maintainable code more than where to put your curly braces.</p>
228 <!-- ======================================================================= -->
229 <div class="doc_subsection">
230 <a name="stl">Other useful references</a>
233 <div class="doc_text">
236 <li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
237 Branch and Tag Primer</a></li>
238 <li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
239 static and shared libraries across platforms</a></li>
244 <!-- *********************************************************************** -->
245 <div class="doc_section">
246 <a name="apis">Important and useful LLVM APIs</a>
248 <!-- *********************************************************************** -->
250 <div class="doc_text">
252 <p>Here we highlight some LLVM APIs that are generally useful and good to
253 know about when writing transformations.</p>
257 <!-- ======================================================================= -->
258 <div class="doc_subsection">
259 <a name="isa">The isa<>, cast<> and dyn_cast<> templates</a>
262 <div class="doc_text">
264 <p>The LLVM source-base makes extensive use of a custom form of RTTI.
265 These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
266 operator, but they don't have some drawbacks (primarily stemming from
267 the fact that <tt>dynamic_cast<></tt> only works on classes that
268 have a v-table). Because they are used so often, you must know what they
269 do and how they work. All of these templates are defined in the <a
270 href="/doxygen/Casting_8h-source.html"><tt>Support/Casting.h</tt></a>
271 file (note that you very rarely have to include this file directly).</p>
274 <dt><tt>isa<></tt>: </dt>
276 <dd>The <tt>isa<></tt> operator works exactly like the Java
277 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
278 a reference or pointer points to an instance of the specified class. This can
279 be very useful for constraint checking of various sorts (example below).</dd>
281 <dt><tt>cast<></tt>: </dt>
283 <dd>The <tt>cast<></tt> operator is a "checked cast" operation. It
284 converts a pointer or reference from a base class to a derived cast, causing
285 an assertion failure if it is not really an instance of the right type. This
286 should be used in cases where you have some information that makes you believe
287 that something is of the right type. An example of the <tt>isa<></tt>
288 and <tt>cast<></tt> template is:
291 static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
292 if (isa<<a href="#Constant">Constant</a>>(V) || isa<<a href="#Argument">Argument</a>>(V) || isa<<a href="#GlobalValue">GlobalValue</a>>(V))
295 <i>// Otherwise, it must be an instruction...</i>
296 return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
300 <p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
301 by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
306 <dt><tt>dyn_cast<></tt>:</dt>
308 <dd>The <tt>dyn_cast<></tt> operator is a "checking cast" operation. It
309 checks to see if the operand is of the specified type, and if so, returns a
310 pointer to it (this operator does not work with references). If the operand is
311 not of the correct type, a null pointer is returned. Thus, this works very
312 much like the <tt>dynamic_cast</tt> operator in C++, and should be used in the
313 same circumstances. Typically, the <tt>dyn_cast<></tt> operator is used
314 in an <tt>if</tt> statement or some other flow control statement like this:
317 if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
322 <p> This form of the <tt>if</tt> statement effectively combines together a
323 call to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
324 statement, which is very convenient.</p>
326 <p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
327 <tt>dynamic_cast</tt> or Java's <tt>instanceof</tt> operator, can be abused.
328 In particular you should not use big chained <tt>if/then/else</tt> blocks to
329 check for lots of different variants of classes. If you find yourself
330 wanting to do this, it is much cleaner and more efficient to use the
331 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
335 <dt><tt>cast_or_null<></tt>: </dt>
337 <dd>The <tt>cast_or_null<></tt> operator works just like the
338 <tt>cast<></tt> operator, except that it allows for a null pointer as
339 an argument (which it then propagates). This can sometimes be useful,
340 allowing you to combine several null checks into one.</dd>
342 <dt><tt>dyn_cast_or_null<></tt>: </dt>
344 <dd>The <tt>dyn_cast_or_null<></tt> operator works just like the
345 <tt>dyn_cast<></tt> operator, except that it allows for a null pointer
346 as an argument (which it then propagates). This can sometimes be useful,
347 allowing you to combine several null checks into one.</dd>
351 <p>These five templates can be used with any classes, whether they have a
352 v-table or not. To add support for these templates, you simply need to add
353 <tt>classof</tt> static methods to the class you are interested casting
354 to. Describing this is currently outside the scope of this document, but there
355 are lots of examples in the LLVM source base.</p>
359 <!-- ======================================================================= -->
360 <div class="doc_subsection">
361 <a name="DEBUG">The <tt>DEBUG()</tt> macro & <tt>-debug</tt> option</a>
364 <div class="doc_text">
366 <p>Often when working on your pass you will put a bunch of debugging printouts
367 and other code into your pass. After you get it working, you want to remove
368 it... but you may need it again in the future (to work out new bugs that you run
371 <p> Naturally, because of this, you don't want to delete the debug printouts,
372 but you don't want them to always be noisy. A standard compromise is to comment
373 them out, allowing you to enable them if you need them in the future.</p>
375 <p>The "<tt><a href="/doxygen/Debug_8h-source.html">Support/Debug.h</a></tt>"
376 file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
377 this problem. Basically, you can put arbitrary code into the argument of the
378 <tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
379 tool) is run with the '<tt>-debug</tt>' command line argument:</p>
381 <pre> ... <br> DEBUG(std::cerr << "I am here!\n");<br> ...<br></pre>
383 <p>Then you can run your pass like this:</p>
385 <pre> $ opt < a.bc > /dev/null -mypass<br> <no output><br> $ opt < a.bc > /dev/null -mypass -debug<br> I am here!<br> $<br></pre>
387 <p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
388 to not have to create "yet another" command line option for the debug output for
389 your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
390 so they do not cause a performance impact at all (for the same reason, they
391 should also not contain side-effects!).</p>
393 <p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
394 enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
395 "<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
396 program hasn't been started yet, you can always just run it with
401 <!-- _______________________________________________________________________ -->
402 <div class="doc_subsubsection">
403 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE()</tt> and
404 the <tt>-debug-only</tt> option</a>
407 <div class="doc_text">
409 <p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
410 just turns on <b>too much</b> information (such as when working on the code
411 generator). If you want to enable debug information with more fine-grained
412 control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
413 option as follows:</p>
415 <pre> ...<br> DEBUG(std::cerr << "No debug type\n");<br> #undef DEBUG_TYPE<br> #define DEBUG_TYPE "foo"<br> DEBUG(std::cerr << "'foo' debug type\n");<br> #undef DEBUG_TYPE<br> #define DEBUG_TYPE "bar"<br> DEBUG(std::cerr << "'bar' debug type\n");<br> #undef DEBUG_TYPE<br> #define DEBUG_TYPE ""<br> DEBUG(std::cerr << "No debug type (2)\n");<br> ...<br></pre>
417 <p>Then you can run your pass like this:</p>
419 <pre> $ opt < a.bc > /dev/null -mypass<br> <no output><br> $ opt < a.bc > /dev/null -mypass -debug<br> No debug type<br> 'foo' debug type<br> 'bar' debug type<br> No debug type (2)<br> $ opt < a.bc > /dev/null -mypass -debug-only=foo<br> 'foo' debug type<br> $ opt < a.bc > /dev/null -mypass -debug-only=bar<br> 'bar' debug type<br> $<br></pre>
421 <p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
422 a file, to specify the debug type for the entire module (if you do this before
423 you <tt>#include "Support/Debug.h"</tt>, you don't have to insert the ugly
424 <tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
425 "bar", because there is no system in place to ensure that names do not
426 conflict. If two different modules use the same string, they will all be turned
427 on when the name is specified. This allows, for example, all debug information
428 for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
429 even if the source lives in multiple files.</p>
433 <!-- ======================================================================= -->
434 <div class="doc_subsection">
435 <a name="Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
439 <div class="doc_text">
442 href="/doxygen/Statistic_8h-source.html">Support/Statistic.h</a></tt>" file
443 provides a template named <tt>Statistic</tt> that is used as a unified way to
444 keep track of what the LLVM compiler is doing and how effective various
445 optimizations are. It is useful to see what optimizations are contributing to
446 making a particular program run faster.</p>
448 <p>Often you may run your pass on some big program, and you're interested to see
449 how many times it makes a certain transformation. Although you can do this with
450 hand inspection, or some ad-hoc method, this is a real pain and not very useful
451 for big programs. Using the <tt>Statistic</tt> template makes it very easy to
452 keep track of this information, and the calculated information is presented in a
453 uniform manner with the rest of the passes being executed.</p>
455 <p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
456 it are as follows:</p>
459 <li>Define your statistic like this:
460 <pre>static Statistic<> NumXForms("mypassname", "The # of times I did stuff");<br></pre>
462 <p>The <tt>Statistic</tt> template can emulate just about any data-type,
463 but if you do not specify a template argument, it defaults to acting like
464 an unsigned int counter (this is usually what you want).</p></li>
466 <li>Whenever you make a transformation, bump the counter:
467 <pre> ++NumXForms; // I did stuff<br></pre>
471 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
472 statistics gathered, use the '<tt>-stats</tt>' option:</p>
474 <pre> $ opt -stats -mypassname < program.bc > /dev/null<br> ... statistic output ...<br></pre>
476 <p> When running <tt>gccas</tt> on a C file from the SPEC benchmark
477 suite, it gives a report that looks like this:</p>
479 <pre> 7646 bytecodewriter - Number of normal instructions<br> 725 bytecodewriter - Number of oversized instructions<br> 129996 bytecodewriter - Number of bytecode bytes written<br> 2817 raise - Number of insts DCEd or constprop'd<br> 3213 raise - Number of cast-of-self removed<br> 5046 raise - Number of expression trees converted<br> 75 raise - Number of other getelementptr's formed<br> 138 raise - Number of load/store peepholes<br> 42 deadtypeelim - Number of unused typenames removed from symtab<br> 392 funcresolve - Number of varargs functions resolved<br> 27 globaldce - Number of global variables removed<br> 2 adce - Number of basic blocks removed<br> 134 cee - Number of branches revectored<br> 49 cee - Number of setcc instruction eliminated<br> 532 gcse - Number of loads removed<br> 2919 gcse - Number of instructions removed<br> 86 indvars - Number of canonical indvars added<br> 87 indvars - Number of aux indvars removed<br> 25 instcombine - Number of dead inst eliminate<br> 434 instcombine - Number of insts combined<br> 248 licm - Number of load insts hoisted<br> 1298 licm - Number of insts hoisted to a loop pre-header<br> 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)<br> 75 mem2reg - Number of alloca's promoted<br> 1444 cfgsimplify - Number of blocks simplified<br></pre>
481 <p>Obviously, with so many optimizations, having a unified framework for this
482 stuff is very nice. Making your pass fit well into the framework makes it more
483 maintainable and useful.</p>
487 <!-- *********************************************************************** -->
488 <div class="doc_section">
489 <a name="common">Helpful Hints for Common Operations</a>
491 <!-- *********************************************************************** -->
493 <div class="doc_text">
495 <p>This section describes how to perform some very simple transformations of
496 LLVM code. This is meant to give examples of common idioms used, showing the
497 practical side of LLVM transformations. <p> Because this is a "how-to" section,
498 you should also read about the main classes that you will be working with. The
499 <a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
500 and descriptions of the main classes that you should know about.</p>
504 <!-- NOTE: this section should be heavy on example code -->
505 <!-- ======================================================================= -->
506 <div class="doc_subsection">
507 <a name="inspection">Basic Inspection and Traversal Routines</a>
510 <div class="doc_text">
512 <p>The LLVM compiler infrastructure have many different data structures that may
513 be traversed. Following the example of the C++ standard template library, the
514 techniques used to traverse these various data structures are all basically the
515 same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
516 method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
517 function returns an iterator pointing to one past the last valid element of the
518 sequence, and there is some <tt>XXXiterator</tt> data type that is common
519 between the two operations.</p>
521 <p>Because the pattern for iteration is common across many different aspects of
522 the program representation, the standard template library algorithms may be used
523 on them, and it is easier to remember how to iterate. First we show a few common
524 examples of the data structures that need to be traversed. Other data
525 structures are traversed in very similar ways.</p>
529 <!-- _______________________________________________________________________ -->
530 <div class="doc_subsubsection">
531 <a name="iterate_function">Iterating over the </a><a
532 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
533 href="#Function"><tt>Function</tt></a>
536 <div class="doc_text">
538 <p>It's quite common to have a <tt>Function</tt> instance that you'd like to
539 transform in some way; in particular, you'd like to manipulate its
540 <tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
541 the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
542 an example that prints the name of a <tt>BasicBlock</tt> and the number of
543 <tt>Instruction</tt>s it contains:</p>
545 <pre> // func is a pointer to a Function instance<br> for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i) {<br><br> // print out the name of the basic block if it has one, and then the<br> // number of instructions that it contains<br><br> cerr << "Basic block (name=" << i->getName() << ") has " <br> << i->size() << " instructions.\n";<br> }<br></pre>
547 <p>Note that i can be used as if it were a pointer for the purposes of
548 invoking member functions of the <tt>Instruction</tt> class. This is
549 because the indirection operator is overloaded for the iterator
550 classes. In the above code, the expression <tt>i->size()</tt> is
551 exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
555 <!-- _______________________________________________________________________ -->
556 <div class="doc_subsubsection">
557 <a name="iterate_basicblock">Iterating over the </a><a
558 href="#Instruction"><tt>Instruction</tt></a>s in a <a
559 href="#BasicBlock"><tt>BasicBlock</tt></a>
562 <div class="doc_text">
564 <p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
565 easy to iterate over the individual instructions that make up
566 <tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
567 a <tt>BasicBlock</tt>:</p>
570 // blk is a pointer to a BasicBlock instance
571 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
572 // the next statement works since operator<<(ostream&,...)
573 // is overloaded for Instruction&
574 std::cerr << *i << "\n";
577 <p>However, this isn't really the best way to print out the contents of a
578 <tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
579 anything you'll care about, you could have just invoked the print routine on the
580 basic block itself: <tt>std::cerr << *blk << "\n";</tt>.</p>
584 <!-- _______________________________________________________________________ -->
585 <div class="doc_subsubsection">
586 <a name="iterate_institer">Iterating over the </a><a
587 href="#Instruction"><tt>Instruction</tt></a>s in a <a
588 href="#Function"><tt>Function</tt></a>
591 <div class="doc_text">
593 <p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
594 <tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
595 <tt>InstIterator</tt> should be used instead. You'll need to include <a
596 href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
597 and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
598 small example that shows how to dump all instructions in a function to the standard error stream:<p>
600 <pre>#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"<br>...<br>// Suppose F is a ptr to a function<br>for (inst_iterator i = inst_begin(F), e = inst_end(F); i != e; ++i)<br> cerr << *i << "\n";<br></pre>
601 Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
602 worklist with its initial contents. For example, if you wanted to
603 initialize a worklist to contain all instructions in a <tt>Function</tt>
604 F, all you would need to do is something like:
605 <pre>std::set<Instruction*> worklist;<br>worklist.insert(inst_begin(F), inst_end(F));<br></pre>
607 <p>The STL set <tt>worklist</tt> would now contain all instructions in the
608 <tt>Function</tt> pointed to by F.</p>
612 <!-- _______________________________________________________________________ -->
613 <div class="doc_subsubsection">
614 <a name="iterate_convert">Turning an iterator into a class pointer (and
618 <div class="doc_text">
620 <p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
621 instance when all you've got at hand is an iterator. Well, extracting
622 a reference or a pointer from an iterator is very straight-forward.
623 Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
624 is a <tt>BasicBlock::const_iterator</tt>:</p>
626 <pre> Instruction& inst = *i; // grab reference to instruction reference<br> Instruction* pinst = &*i; // grab pointer to instruction reference<br> const Instruction& inst = *j;<br></pre>
628 <p>However, the iterators you'll be working with in the LLVM framework are
629 special: they will automatically convert to a ptr-to-instance type whenever they
630 need to. Instead of dereferencing the iterator and then taking the address of
631 the result, you can simply assign the iterator to the proper pointer type and
632 you get the dereference and address-of operation as a result of the assignment
633 (behind the scenes, this is a result of overloading casting mechanisms). Thus
634 the last line of the last example,</p>
636 <pre>Instruction* pinst = &*i;</pre>
638 <p>is semantically equivalent to</p>
640 <pre>Instruction* pinst = i;</pre>
642 <p>It's also possible to turn a class pointer into the corresponding iterator,
643 and this is a constant time operation (very efficient). The following code
644 snippet illustrates use of the conversion constructors provided by LLVM
645 iterators. By using these, you can explicitly grab the iterator of something
646 without actually obtaining it via iteration over some structure:</p>
648 <pre>void printNextInstruction(Instruction* inst) {<br> BasicBlock::iterator it(inst);<br> ++it; // after this line, it refers to the instruction after *inst.<br> if (it != inst->getParent()->end()) cerr << *it << "\n";<br>}<br></pre>
652 <!--_______________________________________________________________________-->
653 <div class="doc_subsubsection">
654 <a name="iterate_complex">Finding call sites: a slightly more complex
658 <div class="doc_text">
660 <p>Say that you're writing a FunctionPass and would like to count all the
661 locations in the entire module (that is, across every <tt>Function</tt>) where a
662 certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
663 learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
664 much more straight-forward manner, but this example will allow us to explore how
665 you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudocode, this
666 is what we want to do:</p>
668 <pre>initialize callCounter to zero<br>for each Function f in the Module<br> for each BasicBlock b in f<br> for each Instruction i in b<br> if (i is a CallInst and calls the given function)<br> increment callCounter<br></pre>
670 <p>And the actual code is (remember, since we're writing a
671 <tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
672 override the <tt>runOnFunction</tt> method...):</p>
674 <pre>Function* targetFunc = ...;<br><br>class OurFunctionPass : public FunctionPass {<br> public:<br> OurFunctionPass(): callCounter(0) { }<br><br> virtual runOnFunction(Function& F) {<br> for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {<br> for (BasicBlock::iterator i = b->begin(); ie = b->end(); i != ie; ++i) {<br> if (<a
675 href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
676 href="#CallInst">CallInst</a>>(&*i)) {<br> // we know we've encountered a call instruction, so we<br> // need to determine if it's a call to the<br> // function pointed to by m_func or not.<br> <br> if (callInst->getCalledFunction() == targetFunc)<br> ++callCounter;<br> }<br> }<br> }<br> <br> private:<br> unsigned callCounter;<br>};<br></pre>
680 <!--_______________________________________________________________________-->
681 <div class="doc_subsubsection">
682 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
685 <div class="doc_text">
687 <p>You may have noticed that the previous example was a bit oversimplified in
688 that it did not deal with call sites generated by 'invoke' instructions. In
689 this, and in other situations, you may find that you want to treat
690 <tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
691 most-specific common base class is <tt>Instruction</tt>, which includes lots of
692 less closely-related things. For these cases, LLVM provides a handy wrapper
694 href="http://llvm.cs.uiuc.edu/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
695 It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
696 methods that provide functionality common to <tt>CallInst</tt>s and
697 <tt>InvokeInst</tt>s.</p>
699 <p>This class has "value semantics": it should be passed by value, not by
700 reference and it should not be dynamically allocated or deallocated using
701 <tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
702 assignable and constructable, with costs equivalents to that of a bare pointer.
703 If you look at its definition, it has only a single pointer member.</p>
707 <!--_______________________________________________________________________-->
708 <div class="doc_subsubsection">
709 <a name="iterate_chains">Iterating over def-use & use-def chains</a>
712 <div class="doc_text">
714 <p>Frequently, we might have an instance of the <a
715 href="/doxygen/structllvm_1_1Value.html">Value Class</a> and we want to
716 determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
717 <tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
718 For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
719 particular function <tt>foo</tt>. Finding all of the instructions that
720 <i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
723 <pre>Function* F = ...;<br><br>for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i) {<br> if (Instruction *Inst = dyn_cast<Instruction>(*i)) {<br> cerr << "F is used in instruction:\n";<br> cerr << *Inst << "\n";<br> }<br>}<br></pre>
725 <p>Alternately, it's common to have an instance of the <a
726 href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
727 <tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
728 <tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
729 <tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
730 all of the values that a particular instruction uses (that is, the operands of
731 the particular <tt>Instruction</tt>):</p>
733 <pre>Instruction* pi = ...;<br><br>for (User::op_iterator i = pi->op_begin(), e = pi->op_end(); i != e; ++i) {<br> Value* v = *i;<br> ...<br>}<br></pre>
736 def-use chains ("finding all users of"): Value::use_begin/use_end
737 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
742 <!-- ======================================================================= -->
743 <div class="doc_subsection">
744 <a name="simplechanges">Making simple changes</a>
747 <div class="doc_text">
749 <p>There are some primitive transformation operations present in the LLVM
750 infrastructure that are worth knowing about. When performing
751 transformations, it's fairly common to manipulate the contents of basic
752 blocks. This section describes some of the common methods for doing so
753 and gives example code.</p>
757 <!--_______________________________________________________________________-->
758 <div class="doc_subsubsection">
759 <a name="schanges_creating">Creating and inserting new
760 <tt>Instruction</tt>s</a>
763 <div class="doc_text">
765 <p><i>Instantiating Instructions</i></p>
767 <p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
768 constructor for the kind of instruction to instantiate and provide the necessary
769 parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
770 (const-ptr-to) <tt>Type</tt>. Thus:</p>
772 <pre>AllocaInst* ai = new AllocaInst(Type::IntTy);</pre>
774 <p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
775 one integer in the current stack frame, at runtime. Each <tt>Instruction</tt>
776 subclass is likely to have varying default parameters which change the semantics
777 of the instruction, so refer to the <a
778 href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
779 Instruction</a> that you're interested in instantiating.</p>
781 <p><i>Naming values</i></p>
783 <p>It is very useful to name the values of instructions when you're able to, as
784 this facilitates the debugging of your transformations. If you end up looking
785 at generated LLVM machine code, you definitely want to have logical names
786 associated with the results of instructions! By supplying a value for the
787 <tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
788 associate a logical name with the result of the instruction's execution at
789 runtime. For example, say that I'm writing a transformation that dynamically
790 allocates space for an integer on the stack, and that integer is going to be
791 used as some kind of index by some other code. To accomplish this, I place an
792 <tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
793 <tt>Function</tt>, and I'm intending to use it within the same
794 <tt>Function</tt>. I might do:</p>
796 <pre>AllocaInst* pa = new AllocaInst(Type::IntTy, 0, "indexLoc");</pre>
798 <p>where <tt>indexLoc</tt> is now the logical name of the instruction's
799 execution value, which is a pointer to an integer on the runtime stack.</p>
801 <p><i>Inserting instructions</i></p>
803 <p>There are essentially two ways to insert an <tt>Instruction</tt>
804 into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
807 <li>Insertion into an explicit instruction list
809 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
810 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
811 before <tt>*pi</tt>, we do the following: </p>
813 <pre> BasicBlock *pb = ...;<br> Instruction *pi = ...;<br> Instruction *newInst = new Instruction(...);<br> pb->getInstList().insert(pi, newInst); // inserts newInst before pi in pb<br></pre>
815 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
816 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
817 classes provide constructors which take a pointer to a
818 <tt>BasicBlock</tt> to be appended to. For example code that
821 <pre> BasicBlock *pb = ...;<br> Instruction *newInst = new Instruction(...);<br> pb->getInstList().push_back(newInst); // appends newInst to pb<br></pre>
825 <pre> BasicBlock *pb = ...;<br> Instruction *newInst = new Instruction(..., pb);<br></pre>
827 <p>which is much cleaner, especially if you are creating
828 long instruction streams.</p></li>
830 <li>Insertion into an implicit instruction list
832 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
833 are implicitly associated with an existing instruction list: the instruction
834 list of the enclosing basic block. Thus, we could have accomplished the same
835 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
838 <pre> Instruction *pi = ...;<br> Instruction *newInst = new Instruction(...);<br> pi->getParent()->getInstList().insert(pi, newInst);<br></pre>
840 <p>In fact, this sequence of steps occurs so frequently that the
841 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
842 constructors which take (as a default parameter) a pointer to an
843 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
844 precede. That is, <tt>Instruction</tt> constructors are capable of
845 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
846 provided instruction, immediately before that instruction. Using an
847 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
848 parameter, the above code becomes:</p>
850 <pre>Instruction* pi = ...;<br>Instruction* newInst = new Instruction(..., pi);<br></pre>
852 <p>which is much cleaner, especially if you're creating a lot of
853 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
858 <!--_______________________________________________________________________-->
859 <div class="doc_subsubsection">
860 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
863 <div class="doc_text">
865 <p>Deleting an instruction from an existing sequence of instructions that form a
866 <a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
867 you must have a pointer to the instruction that you wish to delete. Second, you
868 need to obtain the pointer to that instruction's basic block. You use the
869 pointer to the basic block to get its list of instructions and then use the
870 erase function to remove your instruction. For example:</p>
872 <pre> <a href="#Instruction">Instruction</a> *I = .. ;<br> <a
873 href="#BasicBlock">BasicBlock</a> *BB = I->getParent();<br> BB->getInstList().erase(I);<br></pre>
877 <!--_______________________________________________________________________-->
878 <div class="doc_subsubsection">
879 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
883 <div class="doc_text">
885 <p><i>Replacing individual instructions</i></p>
887 <p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
888 permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
889 and <tt>ReplaceInstWithInst</tt>.</p>
891 <h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
894 <li><tt>ReplaceInstWithValue</tt>
896 <p>This function replaces all uses (within a basic block) of a given
897 instruction with a value, and then removes the original instruction. The
898 following example illustrates the replacement of the result of a particular
899 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
900 pointer to an integer.</p>
902 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,<br> Constant::getNullValue(PointerType::get(Type::IntTy)));<br></pre></li>
904 <li><tt>ReplaceInstWithInst</tt>
906 <p>This function replaces a particular instruction with another
907 instruction. The following example illustrates the replacement of one
908 <tt>AllocaInst</tt> with another.</p>
910 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,<br> new AllocaInst(Type::IntTy, 0, "ptrToReplacedInt"));<br></pre></li>
913 <p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
915 <p>You can use <tt>Value::replaceAllUsesWith</tt> and
916 <tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
917 doxygen documentation for the <a href="/doxygen/structllvm_1_1Value.html">Value Class</a>
918 and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
921 <!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
922 include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
923 ReplaceInstWithValue, ReplaceInstWithInst -->
927 <!-- *********************************************************************** -->
928 <div class="doc_section">
929 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
931 <!-- *********************************************************************** -->
933 <div class="doc_text">
935 <p>The Core LLVM classes are the primary means of representing the program
936 being inspected or transformed. The core LLVM classes are defined in
937 header files in the <tt>include/llvm/</tt> directory, and implemented in
938 the <tt>lib/VMCore</tt> directory.</p>
942 <!-- ======================================================================= -->
943 <div class="doc_subsection">
944 <a name="Value">The <tt>Value</tt> class</a>
949 <p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
951 doxygen info: <a href="/doxygen/structllvm_1_1Value.html">Value Class</a></p>
953 <p>The <tt>Value</tt> class is the most important class in the LLVM Source
954 base. It represents a typed value that may be used (among other things) as an
955 operand to an instruction. There are many different types of <tt>Value</tt>s,
956 such as <a href="#Constant"><tt>Constant</tt></a>s,<a
957 href="#Argument"><tt>Argument</tt></a>s. Even <a
958 href="#Instruction"><tt>Instruction</tt></a>s and <a
959 href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
961 <p>A particular <tt>Value</tt> may be used many times in the LLVM representation
962 for a program. For example, an incoming argument to a function (represented
963 with an instance of the <a href="#Argument">Argument</a> class) is "used" by
964 every instruction in the function that references the argument. To keep track
965 of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
966 href="#User"><tt>User</tt></a>s that is using it (the <a
967 href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
968 graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
969 def-use information in the program, and is accessible through the <tt>use_</tt>*
970 methods, shown below.</p>
972 <p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
973 and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
974 method. In addition, all LLVM values can be named. The "name" of the
975 <tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
977 <pre> %<b>foo</b> = add int 1, 2<br></pre>
979 <p><a name="#nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
980 that the name of any value may be missing (an empty string), so names should
981 <b>ONLY</b> be used for debugging (making the source code easier to read,
982 debugging printouts), they should not be used to keep track of values or map
983 between them. For this purpose, use a <tt>std::map</tt> of pointers to the
984 <tt>Value</tt> itself instead.</p>
986 <p>One important aspect of LLVM is that there is no distinction between an SSA
987 variable and the operation that produces it. Because of this, any reference to
988 the value produced by an instruction (or the value available as an incoming
989 argument, for example) is represented as a direct pointer to the instance of
991 represents this value. Although this may take some getting used to, it
992 simplifies the representation and makes it easier to manipulate.</p>
996 <!-- _______________________________________________________________________ -->
997 <div class="doc_subsubsection">
998 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
1001 <div class="doc_text">
1004 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
1006 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
1008 <tt>unsigned use_size()</tt> - Returns the number of users of the
1010 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
1011 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
1013 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
1015 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
1016 element in the list.
1017 <p> These methods are the interface to access the def-use
1018 information in LLVM. As with all other iterators in LLVM, the naming
1019 conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
1021 <li><tt><a href="#Type">Type</a> *getType() const</tt>
1022 <p>This method returns the Type of the Value.</p>
1024 <li><tt>bool hasName() const</tt><br>
1025 <tt>std::string getName() const</tt><br>
1026 <tt>void setName(const std::string &Name)</tt>
1027 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
1028 be aware of the <a href="#nameWarning">precaution above</a>.</p>
1030 <li><tt>void replaceAllUsesWith(Value *V)</tt>
1032 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
1033 href="#User"><tt>User</tt>s</a> of the current value to refer to
1034 "<tt>V</tt>" instead. For example, if you detect that an instruction always
1035 produces a constant value (for example through constant folding), you can
1036 replace all uses of the instruction with the constant like this:</p>
1038 <pre> Inst->replaceAllUsesWith(ConstVal);<br></pre>
1043 <!-- ======================================================================= -->
1044 <div class="doc_subsection">
1045 <a name="User">The <tt>User</tt> class</a>
1048 <div class="doc_text">
1051 <tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
1052 doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
1053 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1055 <p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
1056 refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
1057 that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
1058 referring to. The <tt>User</tt> class itself is a subclass of
1061 <p>The operands of a <tt>User</tt> point directly to the LLVM <a
1062 href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
1063 Single Assignment (SSA) form, there can only be one definition referred to,
1064 allowing this direct connection. This connection provides the use-def
1065 information in LLVM.</p>
1069 <!-- _______________________________________________________________________ -->
1070 <div class="doc_subsubsection">
1071 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
1074 <div class="doc_text">
1076 <p>The <tt>User</tt> class exposes the operand list in two ways: through
1077 an index access interface and through an iterator based interface.</p>
1080 <li><tt>Value *getOperand(unsigned i)</tt><br>
1081 <tt>unsigned getNumOperands()</tt>
1082 <p> These two methods expose the operands of the <tt>User</tt> in a
1083 convenient form for direct access.</p></li>
1085 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
1087 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
1088 the operand list.<br>
1089 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
1091 <p> Together, these methods make up the iterator based interface to
1092 the operands of a <tt>User</tt>.</p></li>
1097 <!-- ======================================================================= -->
1098 <div class="doc_subsection">
1099 <a name="Instruction">The <tt>Instruction</tt> class</a>
1102 <div class="doc_text">
1104 <p><tt>#include "</tt><tt><a
1105 href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
1106 doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
1107 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1108 href="#Value"><tt>Value</tt></a></p>
1110 <p>The <tt>Instruction</tt> class is the common base class for all LLVM
1111 instructions. It provides only a few methods, but is a very commonly used
1112 class. The primary data tracked by the <tt>Instruction</tt> class itself is the
1113 opcode (instruction type) and the parent <a
1114 href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
1115 into. To represent a specific type of instruction, one of many subclasses of
1116 <tt>Instruction</tt> are used.</p>
1118 <p> Because the <tt>Instruction</tt> class subclasses the <a
1119 href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
1120 way as for other <a href="#User"><tt>User</tt></a>s (with the
1121 <tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
1122 <tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
1123 the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
1124 file contains some meta-data about the various different types of instructions
1125 in LLVM. It describes the enum values that are used as opcodes (for example
1126 <tt>Instruction::Add</tt> and <tt>Instruction::SetLE</tt>), as well as the
1127 concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
1128 example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
1129 href="#SetCondInst">SetCondInst</a></tt>). Unfortunately, the use of macros in
1130 this file confuses doxygen, so these enum values don't show up correctly in the
1131 <a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
1135 <!-- _______________________________________________________________________ -->
1136 <div class="doc_subsubsection">
1137 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
1141 <div class="doc_text">
1144 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
1145 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
1146 this <tt>Instruction</tt> is embedded into.</p></li>
1147 <li><tt>bool mayWriteToMemory()</tt>
1148 <p>Returns true if the instruction writes to memory, i.e. it is a
1149 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
1150 <li><tt>unsigned getOpcode()</tt>
1151 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
1152 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
1153 <p>Returns another instance of the specified instruction, identical
1154 in all ways to the original except that the instruction has no parent
1155 (ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
1156 and it has no name</p></li>
1161 <!-- ======================================================================= -->
1162 <div class="doc_subsection">
1163 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
1166 <div class="doc_text">
1169 href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
1170 doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
1172 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1174 <p>This class represents a single entry multiple exit section of the code,
1175 commonly known as a basic block by the compiler community. The
1176 <tt>BasicBlock</tt> class maintains a list of <a
1177 href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
1178 Matching the language definition, the last element of this list of instructions
1179 is always a terminator instruction (a subclass of the <a
1180 href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
1182 <p>In addition to tracking the list of instructions that make up the block, the
1183 <tt>BasicBlock</tt> class also keeps track of the <a
1184 href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
1186 <p>Note that <tt>BasicBlock</tt>s themselves are <a
1187 href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
1188 like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
1193 <!-- _______________________________________________________________________ -->
1194 <div class="doc_subsubsection">
1195 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
1199 <div class="doc_text">
1203 <li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
1204 href="#Function">Function</a> *Parent = 0)</tt>
1206 <p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
1207 insertion into a function. The constructor optionally takes a name for the new
1208 block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
1209 the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
1210 automatically inserted at the end of the specified <a
1211 href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
1212 manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
1214 <li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
1215 <tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
1216 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
1217 <tt>size()</tt>, <tt>empty()</tt>
1218 STL-style functions for accessing the instruction list.
1220 <p>These methods and typedefs are forwarding functions that have the same
1221 semantics as the standard library methods of the same names. These methods
1222 expose the underlying instruction list of a basic block in a way that is easy to
1223 manipulate. To get the full complement of container operations (including
1224 operations to update the list), you must use the <tt>getInstList()</tt>
1227 <li><tt>BasicBlock::InstListType &getInstList()</tt>
1229 <p>This method is used to get access to the underlying container that actually
1230 holds the Instructions. This method must be used when there isn't a forwarding
1231 function in the <tt>BasicBlock</tt> class for the operation that you would like
1232 to perform. Because there are no forwarding functions for "updating"
1233 operations, you need to use this if you want to update the contents of a
1234 <tt>BasicBlock</tt>.</p></li>
1236 <li><tt><a href="#Function">Function</a> *getParent()</tt>
1238 <p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
1239 embedded into, or a null pointer if it is homeless.</p></li>
1241 <li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
1243 <p> Returns a pointer to the terminator instruction that appears at the end of
1244 the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
1245 instruction in the block is not a terminator, then a null pointer is
1252 <!-- ======================================================================= -->
1253 <div class="doc_subsection">
1254 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
1257 <div class="doc_text">
1260 href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
1261 doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
1263 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1264 href="#Value"><tt>Value</tt></a></p>
1266 <p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
1267 href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
1268 visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
1269 Because they are visible at global scope, they are also subject to linking with
1270 other globals defined in different translation units. To control the linking
1271 process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
1272 <tt>GlobalValue</tt>s know whether they have internal or external linkage, as
1273 defined by the <tt>LinkageTypes</tt> enumeration.</p>
1275 <p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
1276 <tt>static</tt> in C), it is not visible to code outside the current translation
1277 unit, and does not participate in linking. If it has external linkage, it is
1278 visible to external code, and does participate in linking. In addition to
1279 linkage information, <tt>GlobalValue</tt>s keep track of which <a
1280 href="#Module"><tt>Module</tt></a> they are currently part of.</p>
1282 <p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
1283 by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
1284 global is always a pointer to its contents. It is important to remember this
1285 when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
1286 be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
1287 subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
1288 int]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
1289 the address of the first element of this array and the value of the
1290 <tt>GlobalVariable</tt> are the same, they have different types. The
1291 <tt>GlobalVariable</tt>'s type is <tt>[24 x int]</tt>. The first element's type
1292 is <tt>int.</tt> Because of this, accessing a global value requires you to
1293 dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
1294 can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
1295 Language Reference Manual</a>.</p>
1299 <!-- _______________________________________________________________________ -->
1300 <div class="doc_subsubsection">
1301 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
1305 <div class="doc_text">
1308 <li><tt>bool hasInternalLinkage() const</tt><br>
1309 <tt>bool hasExternalLinkage() const</tt><br>
1310 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
1311 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
1314 <li><tt><a href="#Module">Module</a> *getParent()</tt>
1315 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
1316 GlobalValue is currently embedded into.</p></li>
1321 <!-- ======================================================================= -->
1322 <div class="doc_subsection">
1323 <a name="Function">The <tt>Function</tt> class</a>
1326 <div class="doc_text">
1329 href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
1330 info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
1331 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1332 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1334 <p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
1335 actually one of the more complex classes in the LLVM heirarchy because it must
1336 keep track of a large amount of data. The <tt>Function</tt> class keeps track
1337 of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal <a
1338 href="#Argument"><tt>Argument</tt></a>s, and a <a
1339 href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
1341 <p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
1342 commonly used part of <tt>Function</tt> objects. The list imposes an implicit
1343 ordering of the blocks in the function, which indicate how the code will be
1344 layed out by the backend. Additionally, the first <a
1345 href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
1346 <tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
1347 block. There are no implicit exit nodes, and in fact there may be multiple exit
1348 nodes from a single <tt>Function</tt>. If the <a
1349 href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
1350 the <tt>Function</tt> is actually a function declaration: the actual body of the
1351 function hasn't been linked in yet.</p>
1353 <p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
1354 <tt>Function</tt> class also keeps track of the list of formal <a
1355 href="#Argument"><tt>Argument</tt></a>s that the function receives. This
1356 container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
1357 nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
1358 the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
1360 <p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
1361 LLVM feature that is only used when you have to look up a value by name. Aside
1362 from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
1363 internally to make sure that there are not conflicts between the names of <a
1364 href="#Instruction"><tt>Instruction</tt></a>s, <a
1365 href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
1366 href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
1368 <p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
1369 and therefore also a <a href="#Constant">Constant</a>. The value of the function
1370 is its address (after linking) which is guaranteed to be constant.</p>
1373 <!-- _______________________________________________________________________ -->
1374 <div class="doc_subsubsection">
1375 <a name="m_Function">Important Public Members of the <tt>Function</tt>
1379 <div class="doc_text">
1382 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
1383 *Ty, LinkageTypes Linkage, const std::string &N = "", Module* Parent = 0)</tt>
1385 <p>Constructor used when you need to create new <tt>Function</tt>s to add
1386 the the program. The constructor must specify the type of the function to
1387 create and what type of linkage the function should have. The <a
1388 href="#FunctionType"><tt>FunctionType</tt></a> argument
1389 specifies the formal arguments and return value for the function. The same
1390 <a href="#FunctionTypel"><tt>FunctionType</tt></a> value can be used to
1391 create multiple functions. The <tt>Parent</tt> argument specifies the Module
1392 in which the function is defined. If this argument is provided, the function
1393 will automatically be inserted into that module's list of
1396 <li><tt>bool isExternal()</tt>
1398 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
1399 function is "external", it does not have a body, and thus must be resolved
1400 by linking with a function defined in a different translation unit.</p></li>
1402 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
1403 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
1405 <tt>begin()</tt>, <tt>end()</tt>
1406 <tt>size()</tt>, <tt>empty()</tt>
1408 <p>These are forwarding methods that make it easy to access the contents of
1409 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
1412 <li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
1414 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
1415 is necessary to use when you need to update the list or perform a complex
1416 action that doesn't have a forwarding method.</p></li>
1418 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
1420 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
1422 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
1423 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
1425 <p>These are forwarding methods that make it easy to access the contents of
1426 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
1429 <li><tt>Function::ArgumentListType &getArgumentList()</tt>
1431 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
1432 necessary to use when you need to update the list or perform a complex
1433 action that doesn't have a forwarding method.</p></li>
1435 <li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
1437 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
1438 function. Because the entry block for the function is always the first
1439 block, this returns the first block of the <tt>Function</tt>.</p></li>
1441 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
1442 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
1444 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
1445 <tt>Function</tt> and returns the return type of the function, or the <a
1446 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
1449 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
1451 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1452 for this <tt>Function</tt>.</p></li>
1457 <!-- ======================================================================= -->
1458 <div class="doc_subsection">
1459 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
1462 <div class="doc_text">
1465 href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
1467 doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
1468 Class</a><br> Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1469 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1471 <p>Global variables are represented with the (suprise suprise)
1472 <tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
1473 subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
1474 always referenced by their address (global values must live in memory, so their
1475 "name" refers to their address). See <a
1476 href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global variables
1477 may have an initial value (which must be a <a
1478 href="#Constant"><tt>Constant</tt></a>), and if they have an initializer, they
1479 may be marked as "constant" themselves (indicating that their contents never
1480 change at runtime).</p>
1484 <!-- _______________________________________________________________________ -->
1485 <div class="doc_subsubsection">
1486 <a name="m_GlobalVariable">Important Public Members of the
1487 <tt>GlobalVariable</tt> class</a>
1490 <div class="doc_text">
1493 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
1494 isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
1495 *Initializer = 0, const std::string &Name = "", Module* Parent = 0)</tt>
1497 <p>Create a new global variable of the specified type. If
1498 <tt>isConstant</tt> is true then the global variable will be marked as
1499 unchanging for the program. The Linkage parameter specifies the type of
1500 linkage (internal, external, weak, linkonce, appending) for the variable. If
1501 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage, then
1502 the resultant global variable will have internal linkage. AppendingLinkage
1503 concatenates together all instances (in different translation units) of the
1504 variable into a single variable but is only applicable to arrays. See
1505 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
1506 further details on linkage types. Optionally an initializer, a name, and the
1507 module to put the variable into may be specified for the global variable as
1510 <li><tt>bool isConstant() const</tt>
1512 <p>Returns true if this is a global variable that is known not to
1513 be modified at runtime.</p></li>
1515 <li><tt>bool hasInitializer()</tt>
1517 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
1519 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
1521 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
1522 to call this method if there is no initializer.</p></li>
1527 <!-- ======================================================================= -->
1528 <div class="doc_subsection">
1529 <a name="Module">The <tt>Module</tt> class</a>
1532 <div class="doc_text">
1535 href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
1536 <a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
1538 <p>The <tt>Module</tt> class represents the top level structure present in LLVM
1539 programs. An LLVM module is effectively either a translation unit of the
1540 original program or a combination of several translation units merged by the
1541 linker. The <tt>Module</tt> class keeps track of a list of <a
1542 href="#Function"><tt>Function</tt></a>s, a list of <a
1543 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
1544 href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
1545 helpful member functions that try to make common operations easy.</p>
1549 <!-- _______________________________________________________________________ -->
1550 <div class="doc_subsubsection">
1551 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
1554 <div class="doc_text">
1557 <li><tt>Module::Module(std::string name = "")</tt></li>
1560 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
1561 provide a name for it (probably based on the name of the translation unit).</p>
1564 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
1565 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
1567 <tt>begin()</tt>, <tt>end()</tt>
1568 <tt>size()</tt>, <tt>empty()</tt>
1570 <p>These are forwarding methods that make it easy to access the contents of
1571 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
1574 <li><tt>Module::FunctionListType &getFunctionList()</tt>
1576 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
1577 necessary to use when you need to update the list or perform a complex
1578 action that doesn't have a forwarding method.</p>
1580 <p><!-- Global Variable --></p></li>
1586 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
1588 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
1590 <tt>global_begin()</tt>, <tt>global_end()</tt>
1591 <tt>global_size()</tt>, <tt>global_empty()</tt>
1593 <p> These are forwarding methods that make it easy to access the contents of
1594 a <tt>Module</tt> object's <a
1595 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
1597 <li><tt>Module::GlobalListType &getGlobalList()</tt>
1599 <p>Returns the list of <a
1600 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
1601 use when you need to update the list or perform a complex action that
1602 doesn't have a forwarding method.</p>
1604 <p><!-- Symbol table stuff --> </p></li>
1610 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
1612 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1613 for this <tt>Module</tt>.</p>
1615 <p><!-- Convenience methods --></p></li>
1621 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
1622 &Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
1624 <p>Look up the specified function in the <tt>Module</tt> <a
1625 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
1626 <tt>null</tt>.</p></li>
1628 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
1629 std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
1631 <p>Look up the specified function in the <tt>Module</tt> <a
1632 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
1633 external declaration for the function and return it.</p></li>
1635 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
1637 <p>If there is at least one entry in the <a
1638 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
1639 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
1642 <li><tt>bool addTypeName(const std::string &Name, const <a
1643 href="#Type">Type</a> *Ty)</tt>
1645 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1646 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
1647 name, true is returned and the <a
1648 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
1653 <!-- ======================================================================= -->
1654 <div class="doc_subsection">
1655 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
1658 <div class="doc_text">
1660 <p>Constant represents a base class for different types of constants. It
1661 is subclassed by ConstantBool, ConstantInt, ConstantSInt, ConstantUInt,
1662 ConstantArray etc for representing the various types of Constants.</p>
1666 <!-- _______________________________________________________________________ -->
1667 <div class="doc_subsubsection">
1668 <a name="m_Constant">Important Public Methods</a>
1670 <div class="doc_text">
1673 <!-- _______________________________________________________________________ -->
1674 <div class="doc_subsubsection">Important Subclasses of Constant </div>
1675 <div class="doc_text">
1677 <li>ConstantSInt : This subclass of Constant represents a signed integer
1680 <li><tt>int64_t getValue() const</tt>: Returns the underlying value of
1681 this constant. </li>
1684 <li>ConstantUInt : This class represents an unsigned integer.
1686 <li><tt>uint64_t getValue() const</tt>: Returns the underlying value of
1687 this constant. </li>
1690 <li>ConstantFP : This class represents a floating point constant.
1692 <li><tt>double getValue() const</tt>: Returns the underlying value of
1693 this constant. </li>
1696 <li>ConstantBool : This represents a boolean constant.
1698 <li><tt>bool getValue() const</tt>: Returns the underlying value of this
1702 <li>ConstantArray : This represents a constant array.
1704 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
1705 a vector of component constants that makeup this array. </li>
1708 <li>ConstantStruct : This represents a constant struct.
1710 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
1711 a vector of component constants that makeup this array. </li>
1714 <li>GlobalValue : This represents either a global variable or a function. In
1715 either case, the value is a constant fixed address (after linking).
1720 <!-- ======================================================================= -->
1721 <div class="doc_subsection">
1722 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
1725 <div class="doc_text">
1727 <p>Type as noted earlier is also a subclass of a Value class. Any primitive
1728 type (like int, short etc) in LLVM is an instance of Type Class. All other
1729 types are instances of subclasses of type like FunctionType, ArrayType
1730 etc. DerivedType is the interface for all such dervied types including
1731 FunctionType, ArrayType, PointerType, StructType. Types can have names. They can
1732 be recursive (StructType). There exists exactly one instance of any type
1733 structure at a time. This allows using pointer equality of Type *s for comparing
1738 <!-- _______________________________________________________________________ -->
1739 <div class="doc_subsubsection">
1740 <a name="m_Value">Important Public Methods</a>
1743 <div class="doc_text">
1747 <li><tt>bool isSigned() const</tt>: Returns whether an integral numeric type
1748 is signed. This is true for SByteTy, ShortTy, IntTy, LongTy. Note that this is
1749 not true for Float and Double. </li>
1751 <li><tt>bool isUnsigned() const</tt>: Returns whether a numeric type is
1752 unsigned. This is not quite the complement of isSigned... nonnumeric types
1753 return false as they do with isSigned. This returns true for UByteTy,
1754 UShortTy, UIntTy, and ULongTy. </li>
1756 <li><tt>bool isInteger() const</tt>: Equivalent to isSigned() || isUnsigned().</li>
1758 <li><tt>bool isIntegral() const</tt>: Returns true if this is an integral
1759 type, which is either Bool type or one of the Integer types.</li>
1761 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
1762 floating point types.</li>
1764 <li><tt>isLosslesslyConvertableTo (const Type *Ty) const</tt>: Return true if
1765 this type can be converted to 'Ty' without any reinterpretation of bits. For
1766 example, uint to int or one pointer type to another.</li>
1770 <!-- _______________________________________________________________________ -->
1771 <div class="doc_subsubsection">
1772 <a name="m_Value">Important Derived Types</a>
1774 <div class="doc_text">
1776 <li>SequentialType : This is subclassed by ArrayType and PointerType
1778 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
1779 of the elements in the sequential type. </li>
1782 <li>ArrayType : This is a subclass of SequentialType and defines interface for
1785 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
1786 elements in the array. </li>
1789 <li>PointerType : Subclass of SequentialType for pointer types. </li>
1790 <li>StructType : subclass of DerivedTypes for struct types </li>
1791 <li>FunctionType : subclass of DerivedTypes for function types.
1793 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
1795 <li><tt> const Type * getReturnType() const</tt>: Returns the
1796 return type of the function.</li>
1797 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
1798 the type of the ith parameter.</li>
1799 <li><tt> const unsigned getNumParams() const</tt>: Returns the
1800 number of formal parameters.</li>
1806 <!-- ======================================================================= -->
1807 <div class="doc_subsection">
1808 <a name="Argument">The <tt>Argument</tt> class</a>
1811 <div class="doc_text">
1813 <p>This subclass of Value defines the interface for incoming formal
1814 arguments to a function. A Function maintains a list of its formal
1815 arguments. An argument has a pointer to the parent Function.</p>
1819 <!-- ======================================================================= -->
1820 <div class="doc_subsection">
1821 <a name="SymbolTable">The <tt>SymbolTable</tt> class</a>
1823 <div class="doc_text">
1824 <p>This class provides a symbol table that the
1825 <a href="#Function"><tt>Function</tt></a> and <a href="#Module">
1826 <tt>Module</tt></a> classes use for naming definitions. The symbol table can
1827 provide a name for any <a href="#Value"><tt>Value</tt></a> or
1828 <a href="#Type"><tt>Type</tt></a>. <tt>SymbolTable</tt> is an abstract data
1829 type. It hides the data it contains and provides access to it through a
1830 controlled interface.</p>
1832 <p>To use the <tt>SymbolTable</tt> well, you need to understand the
1833 structure of the information it holds. The class contains two
1834 <tt>std::map</tt> objects. The first, <tt>pmap</tt>, is a map of
1835 <tt>Type*</tt> to maps of name (<tt>std::string</tt>) to <tt>Value*</tt>.
1836 The second, <tt>tmap</tt>, is a map of names to <tt>Type*</tt>. Thus, Values
1837 are stored in two-dimensions and accessed by <tt>Type</tt> and name. Types,
1838 however, are stored in a single dimension and accessed only by name.</p>
1840 <p>The interface of this class provides three basic types of operations:
1842 <li><em>Accessors</em>. Accessors provide read-only access to information
1843 such as finding a value for a name with the
1844 <a href="#SymbolTable_lookup">lookup</a> method.</li>
1845 <li><em>Mutators</em>. Mutators allow the user to add information to the
1846 <tt>SymbolTable</tt> with methods like
1847 <a href="#SymbolTable_insert"><tt>insert</tt></a>.</li>
1848 <li><em>Iterators</em>. Iterators allow the user to traverse the content
1849 of the symbol table in well defined ways, such as the method
1850 <a href="#SymbolTable_type_begin"><tt>type_begin</tt></a>.</li>
1855 <dt><tt>Value* lookup(const Type* Ty, const std::string& name) const</tt>:
1857 <dd>The <tt>lookup</tt> method searches the type plane given by the
1858 <tt>Ty</tt> parameter for a <tt>Value</tt> with the provided <tt>name</tt>.
1859 If a suitable <tt>Value</tt> is not found, null is returned.</dd>
1861 <dt><tt>Type* lookupType( const std::string& name) const</tt>:</dt>
1862 <dd>The <tt>lookupType</tt> method searches through the types for a
1863 <tt>Type</tt> with the provided <tt>name</tt>. If a suitable <tt>Type</tt>
1864 is not found, null is returned.</dd>
1866 <dt><tt>bool hasTypes() const</tt>:</dt>
1867 <dd>This function returns true if an entry has been made into the type
1870 <dt><tt>bool isEmpty() const</tt>:</dt>
1871 <dd>This function returns true if both the value and types maps are
1877 <dt><tt>void insert(Value *Val)</tt>:</dt>
1878 <dd>This method adds the provided value to the symbol table. The Value must
1879 have both a name and a type which are extracted and used to place the value
1880 in the correct type plane under the value's name.</dd>
1882 <dt><tt>void insert(const std::string& Name, Value *Val)</tt>:</dt>
1883 <dd> Inserts a constant or type into the symbol table with the specified
1884 name. There can be a many to one mapping between names and constants
1887 <dt><tt>void insert(const std::string& Name, Type *Typ)</tt>:</dt>
1888 <dd> Inserts a type into the symbol table with the specified name. There
1889 can be a many-to-one mapping between names and types. This method
1890 allows a type with an existing entry in the symbol table to get
1893 <dt><tt>void remove(Value* Val)</tt>:</dt>
1894 <dd> This method removes a named value from the symbol table. The
1895 type and name of the Value are extracted from \p N and used to
1896 lookup the Value in the correct type plane. If the Value is
1897 not in the symbol table, this method silently ignores the
1900 <dt><tt>void remove(Type* Typ)</tt>:</dt>
1901 <dd> This method removes a named type from the symbol table. The
1902 name of the type is extracted from \P T and used to look up
1903 the Type in the type map. If the Type is not in the symbol
1904 table, this method silently ignores the request.</dd>
1906 <dt><tt>Value* remove(const std::string& Name, Value *Val)</tt>:</dt>
1907 <dd> Remove a constant or type with the specified name from the
1910 <dt><tt>Type* remove(const std::string& Name, Type* T)</tt>:</dt>
1911 <dd> Remove a type with the specified name from the symbol table.
1912 Returns the removed Type.</dd>
1914 <dt><tt>Value *value_remove(const value_iterator& It)</tt>:</dt>
1915 <dd> Removes a specific value from the symbol table.
1916 Returns the removed value.</dd>
1918 <dt><tt>bool strip()</tt>:</dt>
1919 <dd> This method will strip the symbol table of its names leaving
1920 the type and values. </dd>
1922 <dt><tt>void clear()</tt>:</dt>
1923 <dd>Empty the symbol table completely.</dd>
1927 <p>The following functions describe three types of iterators you can obtain
1928 the beginning or end of the sequence for both const and non-const. It is
1929 important to keep track of the different kinds of iterators. There are
1930 three idioms worth pointing out:</p>
1932 <tr><th>Units</th><th>Iterator</th><th>Idiom</th></tr>
1934 <td align="left">Planes Of name/Value maps</td><td>PI</td>
1935 <td align="left"><pre><tt>
1936 for (SymbolTable::plane_const_iterator PI = ST.plane_begin(),
1937 PE = ST.plane_end(); PI != PE; ++PI ) {
1938 PI->first // This is the Type* of the plane
1939 PI->second // This is the SymbolTable::ValueMap of name/Value pairs
1943 <td align="left">All name/Type Pairs</td><td>TI</td>
1944 <td align="left"><pre><tt>
1945 for (SymbolTable::type_const_iterator TI = ST.type_begin(),
1946 TE = ST.type_end(); TI != TE; ++TI )
1947 TI->first // This is the name of the type
1948 TI->second // This is the Type* value associated with the name
1952 <td align="left">name/Value pairs in a plane</td><td>VI</td>
1953 <td align="left"><pre><tt>
1954 for (SymbolTable::value_const_iterator VI = ST.value_begin(SomeType),
1955 VE = ST.value_end(SomeType); VI != VE; ++VI )
1956 VI->first // This is the name of the Value
1957 VI->second // This is the Value* value associated with the name
1962 <p>Using the recommended iterator names and idioms will help you avoid
1963 making mistakes. Of particular note, make sure that whenever you use
1964 value_begin(SomeType) that you always compare the resulting iterator
1965 with value_end(SomeType) not value_end(SomeOtherType) or else you
1966 will loop infinitely.</p>
1970 <dt><tt>plane_iterator plane_begin()</tt>:</dt>
1971 <dd>Get an iterator that starts at the beginning of the type planes.
1972 The iterator will iterate over the Type/ValueMap pairs in the
1975 <dt><tt>plane_const_iterator plane_begin() const</tt>:</dt>
1976 <dd>Get a const_iterator that starts at the beginning of the type
1977 planes. The iterator will iterate over the Type/ValueMap pairs
1978 in the type planes. </dd>
1980 <dt><tt>plane_iterator plane_end()</tt>:</dt>
1981 <dd>Get an iterator at the end of the type planes. This serves as
1982 the marker for end of iteration over the type planes.</dd>
1984 <dt><tt>plane_const_iterator plane_end() const</tt>:</dt>
1985 <dd>Get a const_iterator at the end of the type planes. This serves as
1986 the marker for end of iteration over the type planes.</dd>
1988 <dt><tt>value_iterator value_begin(const Type *Typ)</tt>:</dt>
1989 <dd>Get an iterator that starts at the beginning of a type plane.
1990 The iterator will iterate over the name/value pairs in the type plane.
1991 Note: The type plane must already exist before using this.</dd>
1993 <dt><tt>value_const_iterator value_begin(const Type *Typ) const</tt>:</dt>
1994 <dd>Get a const_iterator that starts at the beginning of a type plane.
1995 The iterator will iterate over the name/value pairs in the type plane.
1996 Note: The type plane must already exist before using this.</dd>
1998 <dt><tt>value_iterator value_end(const Type *Typ)</tt>:</dt>
1999 <dd>Get an iterator to the end of a type plane. This serves as the marker
2000 for end of iteration of the type plane.
2001 Note: The type plane must already exist before using this.</dd>
2003 <dt><tt>value_const_iterator value_end(const Type *Typ) const</tt>:</dt>
2004 <dd>Get a const_iterator to the end of a type plane. This serves as the
2005 marker for end of iteration of the type plane.
2006 Note: the type plane must already exist before using this.</dd>
2008 <dt><tt>type_iterator type_begin()</tt>:</dt>
2009 <dd>Get an iterator to the start of the name/Type map.</dd>
2011 <dt><tt>type_const_iterator type_begin() cons</tt>:</dt>
2012 <dd> Get a const_iterator to the start of the name/Type map.</dd>
2014 <dt><tt>type_iterator type_end()</tt>:</dt>
2015 <dd>Get an iterator to the end of the name/Type map. This serves as the
2016 marker for end of iteration of the types.</dd>
2018 <dt><tt>type_const_iterator type_end() const</tt>:</dt>
2019 <dd>Get a const-iterator to the end of the name/Type map. This serves
2020 as the marker for end of iteration of the types.</dd>
2022 <dt><tt>plane_const_iterator find(const Type* Typ ) const</tt>:</dt>
2023 <dd>This method returns a plane_const_iterator for iteration over
2024 the type planes starting at a specific plane, given by \p Ty.</dd>
2026 <dt><tt>plane_iterator find( const Type* Typ </tt>:</dt>
2027 <dd>This method returns a plane_iterator for iteration over the
2028 type planes starting at a specific plane, given by \p Ty.</dd>
2033 <!-- *********************************************************************** -->
2036 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2037 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
2038 <a href="http://validator.w3.org/check/referer"><img
2039 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
2041 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
2042 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
2043 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
2044 Last modified: $Date$