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><!--
19 <li>The <tt>-time-passes</tt> option
20 <li>How to use the LLVM Makefile system
21 <li>How to write a regression test
25 <li><a href="#apis">Important and useful LLVM APIs</a>
27 <li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
28 and <tt>dyn_cast<></tt> templates</a> </li>
29 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro & <tt>-debug</tt>
32 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
33 and the <tt>-debug-only</tt> option</a> </li>
36 <li><a href="#Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
38 <li>The <tt>InstVisitor</tt> template
39 <li>The general graph API
43 <li><a href="#common">Helpful Hints for Common Operations</a>
45 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
47 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
48 in a <tt>Function</tt></a> </li>
49 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
50 in a <tt>BasicBlock</tt></a> </li>
51 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
52 in a <tt>Function</tt></a> </li>
53 <li><a href="#iterate_convert">Turning an iterator into a
54 class pointer</a> </li>
55 <li><a href="#iterate_complex">Finding call sites: a more
56 complex example</a> </li>
57 <li><a href="#calls_and_invokes">Treating calls and invokes
58 the same way</a> </li>
59 <li><a href="#iterate_chains">Iterating over def-use &
60 use-def chains</a> </li>
63 <li><a href="#simplechanges">Making simple changes</a>
65 <li><a href="#schanges_creating">Creating and inserting new
66 <tt>Instruction</tt>s</a> </li>
67 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
68 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
69 with another <tt>Value</tt></a> </li>
72 <li>Working with the Control Flow Graph
74 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
81 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
83 <li><a href="#Value">The <tt>Value</tt> class</a>
85 <li><a href="#User">The <tt>User</tt> class</a>
87 <li><a href="#Instruction">The <tt>Instruction</tt> class</a>
89 <li><a href="#GetElementPtrInst">The <tt>GetElementPtrInst</tt>
92 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
94 <li><a href="#BasicBlock">The <tt>BasicBlock</tt>class</a></li>
95 <li><a href="#Function">The <tt>Function</tt> class</a></li>
96 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class
99 <li><a href="#Module">The <tt>Module</tt> class</a></li>
100 <li><a href="#Constant">The <tt>Constant</tt> class</a></li>
101 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
102 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
105 <li><a href="#SymbolTable">The <tt>SymbolTable</tt> class </a></li>
106 <li>The <tt>ilist</tt> and <tt>iplist</tt> classes
108 <li>Creating, inserting, moving and deleting from LLVM lists </li>
111 <li>Important iterator invalidation semantics to be aware of.</li>
115 <div class="doc_author">
116 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
117 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
118 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
119 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
122 <!-- *********************************************************************** -->
123 <div class="doc_section">
124 <a name="introduction">Introduction </a>
126 <!-- *********************************************************************** -->
128 <div class="doc_text">
130 <p>This document is meant to highlight some of the important classes and
131 interfaces available in the LLVM source-base. This manual is not
132 intended to explain what LLVM is, how it works, and what LLVM code looks
133 like. It assumes that you know the basics of LLVM and are interested
134 in writing transformations or otherwise analyzing or manipulating the
137 <p>This document should get you oriented so that you can find your
138 way in the continuously growing source code that makes up the LLVM
139 infrastructure. Note that this manual is not intended to serve as a
140 replacement for reading the source code, so if you think there should be
141 a method in one of these classes to do something, but it's not listed,
142 check the source. Links to the <a href="/doxygen/">doxygen</a> sources
143 are provided to make this as easy as possible.</p>
145 <p>The first section of this document describes general information that is
146 useful to know when working in the LLVM infrastructure, and the second describes
147 the Core LLVM classes. In the future this manual will be extended with
148 information describing how to use extension libraries, such as dominator
149 information, CFG traversal routines, and useful utilities like the <tt><a
150 href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
154 <!-- *********************************************************************** -->
155 <div class="doc_section">
156 <a name="general">General Information</a>
158 <!-- *********************************************************************** -->
160 <div class="doc_text">
162 <p>This section contains general information that is useful if you are working
163 in the LLVM source-base, but that isn't specific to any particular API.</p>
167 <!-- ======================================================================= -->
168 <div class="doc_subsection">
169 <a name="stl">The C++ Standard Template Library</a>
172 <div class="doc_text">
174 <p>LLVM makes heavy use of the C++ Standard Template Library (STL),
175 perhaps much more than you are used to, or have seen before. Because of
176 this, you might want to do a little background reading in the
177 techniques used and capabilities of the library. There are many good
178 pages that discuss the STL, and several books on the subject that you
179 can get, so it will not be discussed in this document.</p>
181 <p>Here are some useful links:</p>
185 <li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
186 reference</a> - an excellent reference for the STL and other parts of the
187 standard C++ library.</li>
189 <li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
190 O'Reilly book in the making. It has a decent <a
191 href="http://www.tempest-sw.com/cpp/ch13-libref.html">Standard Library
192 Reference</a> that rivals Dinkumware's, and is actually free until the book is
195 <li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
198 <li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
200 href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
203 <li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
206 <li><a href="http://www.linux.com.cn/Bruce_Eckel/TICPPv2/Contents.htm">
207 Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
212 <p>You are also encouraged to take a look at the <a
213 href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
214 to write maintainable code more than where to put your curly braces.</p>
218 <!-- ======================================================================= -->
219 <div class="doc_subsection">
220 <a name="stl">Other useful references</a>
223 <div class="doc_text">
225 <p>LLVM is currently using CVS as its source versioning system. You may find
226 this reference handy:</p>
229 <li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
230 Branch and Tag Primer</a></li>
235 <!-- *********************************************************************** -->
236 <div class="doc_section">
237 <a name="apis">Important and useful LLVM APIs</a>
239 <!-- *********************************************************************** -->
241 <div class="doc_text">
243 <p>Here we highlight some LLVM APIs that are generally useful and good to
244 know about when writing transformations.</p>
248 <!-- ======================================================================= -->
249 <div class="doc_subsection">
250 <a name="isa">The isa<>, cast<> and dyn_cast<> templates</a>
253 <div class="doc_text">
255 <p>The LLVM source-base makes extensive use of a custom form of RTTI.
256 These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
257 operator, but they don't have some drawbacks (primarily stemming from
258 the fact that <tt>dynamic_cast<></tt> only works on classes that
259 have a v-table). Because they are used so often, you must know what they
260 do and how they work. All of these templates are defined in the <a
261 href="/doxygen/Casting_8h-source.html"><tt>Support/Casting.h</tt></a>
262 file (note that you very rarely have to include this file directly).</p>
265 <dt><tt>isa<></tt>: </dt>
267 <dd>The <tt>isa<></tt> operator works exactly like the Java
268 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
269 a reference or pointer points to an instance of the specified class. This can
270 be very useful for constraint checking of various sorts (example below).</dd>
272 <dt><tt>cast<></tt>: </dt>
274 <dd>The <tt>cast<></tt> operator is a "checked cast" operation. It
275 converts a pointer or reference from a base class to a derived cast, causing
276 an assertion failure if it is not really an instance of the right type. This
277 should be used in cases where you have some information that makes you believe
278 that something is of the right type. An example of the <tt>isa<></tt>
279 and <tt>cast<></tt> template is:
282 static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
283 if (isa<<a href="#Constant">Constant</a>>(V) || isa<<a href="#Argument">Argument</a>>(V) || isa<<a href="#GlobalValue">GlobalValue</a>>(V))
286 <i>// Otherwise, it must be an instruction...</i>
287 return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
290 <p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
291 by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
296 <dt><tt>dyn_cast<></tt>:</dt>
298 <dd>The <tt>dyn_cast<></tt> operator is a "checking cast" operation. It
299 checks to see if the operand is of the specified type, and if so, returns a
300 pointer to it (this operator does not work with references). If the operand is
301 not of the correct type, a null pointer is returned. Thus, this works very
302 much like the <tt>dynamic_cast</tt> operator in C++, and should be used in the
303 same circumstances. Typically, the <tt>dyn_cast<></tt> operator is used
304 in an <tt>if</tt> statement or some other flow control statement like this:
307 if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
312 <p> This form of the <tt>if</tt> statement effectively combines together a
313 call to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
314 statement, which is very convenient.</p>
316 <p> Another common example is:</p>
319 <i>// Loop over all of the phi nodes in a basic block</i>
320 BasicBlock::iterator BBI = BB->begin();
321 for (; <a href="#PhiNode">PHINode</a> *PN = dyn_cast<<a href="#PHINode">PHINode</a>>(BBI); ++BBI)
322 std::cerr << *PN;
325 <p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
326 <tt>dynamic_cast</tt> or Java's <tt>instanceof</tt> operator, can be abused.
327 In particular you should not use big chained <tt>if/then/else</tt> blocks to
328 check for lots of different variants of classes. If you find yourself
329 wanting to do this, it is much cleaner and more efficient to use the
330 InstVisitor class to dispatch over the instruction type directly.</p>
334 <dt><tt>cast_or_null<></tt>: </dt>
336 <dd>The <tt>cast_or_null<></tt> operator works just like the
337 <tt>cast<></tt> operator, except that it allows for a null pointer as
338 an argument (which it then propagates). This can sometimes be useful,
339 allowing you to combine several null checks into one.</dd>
341 <dt><tt>dyn_cast_or_null<></tt>: </dt>
343 <dd>The <tt>dyn_cast_or_null<></tt> operator works just like the
344 <tt>dyn_cast<></tt> operator, except that it allows for a null pointer
345 as an argument (which it then propagates). This can sometimes be useful,
346 allowing you to combine several null checks into one.</dd>
350 <p>These five templates can be used with any classes, whether they have a
351 v-table or not. To add support for these templates, you simply need to add
352 <tt>classof</tt> static methods to the class you are interested casting
353 to. Describing this is currently outside the scope of this document, but there
354 are lots of examples in the LLVM source base.</p>
358 <!-- ======================================================================= -->
359 <div class="doc_subsection">
360 <a name="DEBUG">The <tt>DEBUG()</tt> macro & <tt>-debug</tt> option</a>
363 <div class="doc_text">
365 <p>Often when working on your pass you will put a bunch of debugging printouts
366 and other code into your pass. After you get it working, you want to remove
367 it... but you may need it again in the future (to work out new bugs that you run
370 <p> Naturally, because of this, you don't want to delete the debug printouts,
371 but you don't want them to always be noisy. A standard compromise is to comment
372 them out, allowing you to enable them if you need them in the future.</p>
374 <p>The "<tt><a href="/doxygen/Debug_8h-source.html">Support/Debug.h</a></tt>"
375 file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
376 this problem. Basically, you can put arbitrary code into the argument of the
377 <tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
378 tool) is run with the '<tt>-debug</tt>' command line argument:</p>
380 <pre> ... <br> DEBUG(std::cerr << "I am here!\n");<br> ...<br></pre>
382 <p>Then you can run your pass like this:</p>
384 <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>
386 <p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
387 to not have to create "yet another" command line option for the debug output for
388 your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
389 so they do not cause a performance impact at all (for the same reason, they
390 should also not contain side-effects!).</p>
392 <p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
393 enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
394 "<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
395 program hasn't been started yet, you can always just run it with
400 <!-- _______________________________________________________________________ -->
401 <div class="doc_subsubsection">
402 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE()</tt> and
403 the <tt>-debug-only</tt> option</a>
406 <div class="doc_text">
408 <p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
409 just turns on <b>too much</b> information (such as when working on the code
410 generator). If you want to enable debug information with more fine-grained
411 control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
412 option as follows:</p>
414 <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>
416 <p>Then you can run your pass like this:</p>
418 <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>
420 <p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
421 a file, to specify the debug type for the entire module (if you do this before
422 you <tt>#include "Support/Debug.h"</tt>, you don't have to insert the ugly
423 <tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
424 "bar", because there is no system in place to ensure that names do not
425 conflict. If two different modules use the same string, they will all be turned
426 on when the name is specified. This allows, for example, all debug information
427 for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
428 even if the source lives in multiple files.</p>
432 <!-- ======================================================================= -->
433 <div class="doc_subsection">
434 <a name="Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
438 <div class="doc_text">
441 href="/doxygen/Statistic_8h-source.html">Support/Statistic.h</a></tt>" file
442 provides a template named <tt>Statistic</tt> that is used as a unified way to
443 keep track of what the LLVM compiler is doing and how effective various
444 optimizations are. It is useful to see what optimizations are contributing to
445 making a particular program run faster.</p>
447 <p>Often you may run your pass on some big program, and you're interested to see
448 how many times it makes a certain transformation. Although you can do this with
449 hand inspection, or some ad-hoc method, this is a real pain and not very useful
450 for big programs. Using the <tt>Statistic</tt> template makes it very easy to
451 keep track of this information, and the calculated information is presented in a
452 uniform manner with the rest of the passes being executed.</p>
454 <p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
455 it are as follows:</p>
458 <li>Define your statistic like this:
459 <pre>static Statistic<> NumXForms("mypassname", "The # of times I did stuff");<br></pre>
461 <p>The <tt>Statistic</tt> template can emulate just about any data-type,
462 but if you do not specify a template argument, it defaults to acting like
463 an unsigned int counter (this is usually what you want).</p></li>
465 <li>Whenever you make a transformation, bump the counter:
466 <pre> ++NumXForms; // I did stuff<br></pre>
470 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
471 statistics gathered, use the '<tt>-stats</tt>' option:</p>
473 <pre> $ opt -stats -mypassname < program.bc > /dev/null<br> ... statistic output ...<br></pre>
475 <p> When running <tt>gccas</tt> on a C file from the SPEC benchmark
476 suite, it gives a report that looks like this:</p>
478 <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>
480 <p>Obviously, with so many optimizations, having a unified framework for this
481 stuff is very nice. Making your pass fit well into the framework makes it more
482 maintainable and useful.</p>
486 <!-- *********************************************************************** -->
487 <div class="doc_section">
488 <a name="common">Helpful Hints for Common Operations</a>
490 <!-- *********************************************************************** -->
492 <div class="doc_text">
494 <p>This section describes how to perform some very simple transformations of
495 LLVM code. This is meant to give examples of common idioms used, showing the
496 practical side of LLVM transformations. <p> Because this is a "how-to" section,
497 you should also read about the main classes that you will be working with. The
498 <a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
499 and descriptions of the main classes that you should know about.</p>
503 <!-- NOTE: this section should be heavy on example code -->
504 <!-- ======================================================================= -->
505 <div class="doc_subsection">
506 <a name="inspection">Basic Inspection and Traversal Routines</a>
509 <div class="doc_text">
511 <p>The LLVM compiler infrastructure have many different data structures that may
512 be traversed. Following the example of the C++ standard template library, the
513 techniques used to traverse these various data structures are all basically the
514 same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
515 method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
516 function returns an iterator pointing to one past the last valid element of the
517 sequence, and there is some <tt>XXXiterator</tt> data type that is common
518 between the two operations.</p>
520 <p>Because the pattern for iteration is common across many different aspects of
521 the program representation, the standard template library algorithms may be used
522 on them, and it is easier to remember how to iterate. First we show a few common
523 examples of the data structures that need to be traversed. Other data
524 structures are traversed in very similar ways.</p>
528 <!-- _______________________________________________________________________ -->
529 <div class="doc_subsubsection">
530 <a name="iterate_function">Iterating over the </a><a
531 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
532 href="#Function"><tt>Function</tt></a>
535 <div class="doc_text">
537 <p>It's quite common to have a <tt>Function</tt> instance that you'd like to
538 transform in some way; in particular, you'd like to manipulate its
539 <tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
540 the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
541 an example that prints the name of a <tt>BasicBlock</tt> and the number of
542 <tt>Instruction</tt>s it contains:</p>
544 <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>
546 <p>Note that i can be used as if it were a pointer for the purposes of
547 invoking member functions of the <tt>Instruction</tt> class. This is
548 because the indirection operator is overloaded for the iterator
549 classes. In the above code, the expression <tt>i->size()</tt> is
550 exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
554 <!-- _______________________________________________________________________ -->
555 <div class="doc_subsubsection">
556 <a name="iterate_basicblock">Iterating over the </a><a
557 href="#Instruction"><tt>Instruction</tt></a>s in a <a
558 href="#BasicBlock"><tt>BasicBlock</tt></a>
561 <div class="doc_text">
563 <p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
564 easy to iterate over the individual instructions that make up
565 <tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
566 a <tt>BasicBlock</tt>:</p>
568 <pre> // blk is a pointer to a BasicBlock instance<br> for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)<br> // the next statement works since operator<<(ostream&,...) <br> // is overloaded for Instruction&<br> cerr << *i << "\n";<br></pre>
570 <p>However, this isn't really the best way to print out the contents of a
571 <tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
572 anything you'll care about, you could have just invoked the print routine on the
573 basic block itself: <tt>cerr << *blk << "\n";</tt>.</p>
575 <p>Note that currently operator<< is implemented for <tt>Value*</tt>, so
576 it will print out the contents of the pointer, instead of the pointer value you
577 might expect. This is a deprecated interface that will be removed in the
578 future, so it's best not to depend on it. To print out the pointer value for
579 now, you must cast to <tt>void*</tt>.</p>
583 <!-- _______________________________________________________________________ -->
584 <div class="doc_subsubsection">
585 <a name="iterate_institer">Iterating over the </a><a
586 href="#Instruction"><tt>Instruction</tt></a>s in a <a
587 href="#Function"><tt>Function</tt></a>
590 <div class="doc_text">
592 <p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
593 <tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
594 <tt>InstIterator</tt> should be used instead. You'll need to include <a
595 href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
596 and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
597 small example that shows how to dump all instructions in a function to the standard error stream:<p>
599 <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>
600 Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
601 worklist with its initial contents. For example, if you wanted to
602 initialize a worklist to contain all instructions in a <tt>Function</tt>
603 F, all you would need to do is something like:
604 <pre>std::set<Instruction*> worklist;<br>worklist.insert(inst_begin(F), inst_end(F));<br></pre>
606 <p>The STL set <tt>worklist</tt> would now contain all instructions in the
607 <tt>Function</tt> pointed to by F.</p>
611 <!-- _______________________________________________________________________ -->
612 <div class="doc_subsubsection">
613 <a name="iterate_convert">Turning an iterator into a class pointer (and
617 <div class="doc_text">
619 <p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
620 instance when all you've got at hand is an iterator. Well, extracting
621 a reference or a pointer from an iterator is very straight-forward.
622 Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
623 is a <tt>BasicBlock::const_iterator</tt>:</p>
625 <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>
627 <p>However, the iterators you'll be working with in the LLVM framework are
628 special: they will automatically convert to a ptr-to-instance type whenever they
629 need to. Instead of dereferencing the iterator and then taking the address of
630 the result, you can simply assign the iterator to the proper pointer type and
631 you get the dereference and address-of operation as a result of the assignment
632 (behind the scenes, this is a result of overloading casting mechanisms). Thus
633 the last line of the last example,</p>
635 <pre>Instruction* pinst = &*i;</pre>
637 <p>is semantically equivalent to</p>
639 <pre>Instruction* pinst = i;</pre>
641 <p>It's also possible to turn a class pointer into the corresponding iterator,
642 and this is a constant time operation (very efficient). The following code
643 snippet illustrates use of the conversion constructors provided by LLVM
644 iterators. By using these, you can explicitly grab the iterator of something
645 without actually obtaining it via iteration over some structure:</p>
647 <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>
651 <!--_______________________________________________________________________-->
652 <div class="doc_subsubsection">
653 <a name="iterate_complex">Finding call sites: a slightly more complex
657 <div class="doc_text">
659 <p>Say that you're writing a FunctionPass and would like to count all the
660 locations in the entire module (that is, across every <tt>Function</tt>) where a
661 certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
662 learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
663 much more straight-forward manner, but this example will allow us to explore how
664 you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudocode, this
665 is what we want to do:</p>
667 <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>
669 <p>And the actual code is (remember, since we're writing a
670 <tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
671 override the <tt>runOnFunction</tt> method...):</p>
673 <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
674 href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
675 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>
679 <!--_______________________________________________________________________-->
680 <div class="doc_subsubsection">
681 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
684 <div class="doc_text">
686 <p>You may have noticed that the previous example was a bit oversimplified in
687 that it did not deal with call sites generated by 'invoke' instructions. In
688 this, and in other situations, you may find that you want to treat
689 <tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
690 most-specific common base class is <tt>Instruction</tt>, which includes lots of
691 less closely-related things. For these cases, LLVM provides a handy wrapper
693 href="http://llvm.cs.uiuc.edu/doxygen/classCallSite.html"><tt>CallSite</tt></a>.
694 It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
695 methods that provide functionality common to <tt>CallInst</tt>s and
696 <tt>InvokeInst</tt>s.</p>
698 <p>This class has "value semantics": it should be passed by value, not by
699 reference and it should not be dynamically allocated or deallocated using
700 <tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
701 assignable and constructable, with costs equivalents to that of a bare pointer.
702 If you look at its definition, it has only a single pointer member.</p>
706 <!--_______________________________________________________________________-->
707 <div class="doc_subsubsection">
708 <a name="iterate_chains">Iterating over def-use & use-def chains</a>
711 <div class="doc_text">
713 <p>Frequently, we might have an instance of the <a
714 href="/doxygen/classValue.html">Value Class</a> and we want to determine which
715 <tt>User</tt>s use the <tt>Value</tt>. The list of all <tt>User</tt>s of a
716 particular <tt>Value</tt> is called a <i>def-use</i> chain. For example, let's
717 say we have a <tt>Function*</tt> named <tt>F</tt> to a particular function
718 <tt>foo</tt>. Finding all of the instructions that <i>use</i> <tt>foo</tt> is as
719 simple as iterating over the <i>def-use</i> chain of <tt>F</tt>:</p>
721 <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>
723 <p>Alternately, it's common to have an instance of the <a
724 href="/doxygen/classUser.html">User Class</a> and need to know what
725 <tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
726 <tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
727 <tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
728 all of the values that a particular instruction uses (that is, the operands of
729 the particular <tt>Instruction</tt>):</p>
731 <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>
734 def-use chains ("finding all users of"): Value::use_begin/use_end
735 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
740 <!-- ======================================================================= -->
741 <div class="doc_subsection">
742 <a name="simplechanges">Making simple changes</a>
745 <div class="doc_text">
747 <p>There are some primitive transformation operations present in the LLVM
748 infrastructure that are worth knowing about. When performing
749 transformations, it's fairly common to manipulate the contents of basic
750 blocks. This section describes some of the common methods for doing so
751 and gives example code.</p>
755 <!--_______________________________________________________________________-->
756 <div class="doc_subsubsection">
757 <a name="schanges_creating">Creating and inserting new
758 <tt>Instruction</tt>s</a>
761 <div class="doc_text">
763 <p><i>Instantiating Instructions</i></p>
765 <p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
766 constructor for the kind of instruction to instantiate and provide the necessary
767 parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
768 (const-ptr-to) <tt>Type</tt>. Thus:</p>
770 <pre>AllocaInst* ai = new AllocaInst(Type::IntTy);</pre>
772 <p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
773 one integer in the current stack frame, at runtime. Each <tt>Instruction</tt>
774 subclass is likely to have varying default parameters which change the semantics
775 of the instruction, so refer to the <a
776 href="/doxygen/classInstruction.html">doxygen documentation for the subclass of
777 Instruction</a> that you're interested in instantiating.</p>
779 <p><i>Naming values</i></p>
781 <p>It is very useful to name the values of instructions when you're able to, as
782 this facilitates the debugging of your transformations. If you end up looking
783 at generated LLVM machine code, you definitely want to have logical names
784 associated with the results of instructions! By supplying a value for the
785 <tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
786 associate a logical name with the result of the instruction's execution at
787 runtime. For example, say that I'm writing a transformation that dynamically
788 allocates space for an integer on the stack, and that integer is going to be
789 used as some kind of index by some other code. To accomplish this, I place an
790 <tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
791 <tt>Function</tt>, and I'm intending to use it within the same
792 <tt>Function</tt>. I might do:</p>
794 <pre>AllocaInst* pa = new AllocaInst(Type::IntTy, 0, "indexLoc");</pre>
796 <p>where <tt>indexLoc</tt> is now the logical name of the instruction's
797 execution value, which is a pointer to an integer on the runtime stack.</p>
799 <p><i>Inserting instructions</i></p>
801 <p>There are essentially two ways to insert an <tt>Instruction</tt>
802 into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
805 <li>Insertion into an explicit instruction list
807 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
808 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
809 before <tt>*pi</tt>, we do the following: </p>
811 <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></li>
813 <li>Insertion into an implicit instruction list
815 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
816 are implicitly associated with an existing instruction list: the instruction
817 list of the enclosing basic block. Thus, we could have accomplished the same
818 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
821 <pre> Instruction *pi = ...;<br> Instruction *newInst = new Instruction(...);<br> pi->getParent()->getInstList().insert(pi, newInst);<br></pre>
823 <p>In fact, this sequence of steps occurs so frequently that the
824 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
825 constructors which take (as a default parameter) a pointer to an
826 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
827 precede. That is, <tt>Instruction</tt> constructors are capable of
828 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
829 provided instruction, immediately before that instruction. Using an
830 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
831 parameter, the above code becomes:</p>
833 <pre>Instruction* pi = ...;<br>Instruction* newInst = new Instruction(..., pi);<br></pre>
835 <p>which is much cleaner, especially if you're creating a lot of
836 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
841 <!--_______________________________________________________________________-->
842 <div class="doc_subsubsection">
843 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
846 <div class="doc_text">
848 <p>Deleting an instruction from an existing sequence of instructions that form a
849 <a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
850 you must have a pointer to the instruction that you wish to delete. Second, you
851 need to obtain the pointer to that instruction's basic block. You use the
852 pointer to the basic block to get its list of instructions and then use the
853 erase function to remove your instruction. For example:</p>
855 <pre> <a href="#Instruction">Instruction</a> *I = .. ;<br> <a
856 href="#BasicBlock">BasicBlock</a> *BB = I->getParent();<br> BB->getInstList().erase(I);<br></pre>
860 <!--_______________________________________________________________________-->
861 <div class="doc_subsubsection">
862 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
866 <div class="doc_text">
868 <p><i>Replacing individual instructions</i></p>
870 <p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
871 permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
872 and <tt>ReplaceInstWithInst</tt>.</p>
874 <h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
877 <li><tt>ReplaceInstWithValue</tt>
879 <p>This function replaces all uses (within a basic block) of a given
880 instruction with a value, and then removes the original instruction. The
881 following example illustrates the replacement of the result of a particular
882 <tt>AllocaInst</tt> that allocates memory for a single integer with an null
883 pointer to an integer.</p>
885 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,<br> Constant::getNullValue(PointerType::get(Type::IntTy)));<br></pre></li>
887 <li><tt>ReplaceInstWithInst</tt>
889 <p>This function replaces a particular instruction with another
890 instruction. The following example illustrates the replacement of one
891 <tt>AllocaInst</tt> with another.</p>
893 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,<br> new AllocaInst(Type::IntTy, 0, "ptrToReplacedInt"));<br></pre></li>
896 <p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
898 <p>You can use <tt>Value::replaceAllUsesWith</tt> and
899 <tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
900 doxygen documentation for the <a href="/doxygen/classValue.html">Value Class</a>
901 and <a href="/doxygen/classUser.html">User Class</a>, respectively, for more
904 <!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
905 include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
906 ReplaceInstWithValue, ReplaceInstWithInst -->
910 <!-- *********************************************************************** -->
911 <div class="doc_section">
912 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
914 <!-- *********************************************************************** -->
916 <div class="doc_text">
918 <p>The Core LLVM classes are the primary means of representing the program
919 being inspected or transformed. The core LLVM classes are defined in
920 header files in the <tt>include/llvm/</tt> directory, and implemented in
921 the <tt>lib/VMCore</tt> directory.</p>
925 <!-- ======================================================================= -->
926 <div class="doc_subsection">
927 <a name="Value">The <tt>Value</tt> class</a>
932 <p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
934 doxygen info: <a href="/doxygen/classValue.html">Value Class</a></p>
936 <p>The <tt>Value</tt> class is the most important class in the LLVM Source
937 base. It represents a typed value that may be used (among other things) as an
938 operand to an instruction. There are many different types of <tt>Value</tt>s,
939 such as <a href="#Constant"><tt>Constant</tt></a>s,<a
940 href="#Argument"><tt>Argument</tt></a>s. Even <a
941 href="#Instruction"><tt>Instruction</tt></a>s and <a
942 href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
944 <p>A particular <tt>Value</tt> may be used many times in the LLVM representation
945 for a program. For example, an incoming argument to a function (represented
946 with an instance of the <a href="#Argument">Argument</a> class) is "used" by
947 every instruction in the function that references the argument. To keep track
948 of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
949 href="#User"><tt>User</tt></a>s that is using it (the <a
950 href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
951 graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
952 def-use information in the program, and is accessible through the <tt>use_</tt>*
953 methods, shown below.</p>
955 <p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
956 and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
957 method. In addition, all LLVM values can be named. The "name" of the
958 <tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
960 <pre> %<b>foo</b> = add int 1, 2<br></pre>
962 <p><a name="#nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
963 that the name of any value may be missing (an empty string), so names should
964 <b>ONLY</b> be used for debugging (making the source code easier to read,
965 debugging printouts), they should not be used to keep track of values or map
966 between them. For this purpose, use a <tt>std::map</tt> of pointers to the
967 <tt>Value</tt> itself instead.</p>
969 <p>One important aspect of LLVM is that there is no distinction between an SSA
970 variable and the operation that produces it. Because of this, any reference to
971 the value produced by an instruction (or the value available as an incoming
972 argument, for example) is represented as a direct pointer to the instance of
974 represents this value. Although this may take some getting used to, it
975 simplifies the representation and makes it easier to manipulate.</p>
979 <!-- _______________________________________________________________________ -->
980 <div class="doc_subsubsection">
981 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
984 <div class="doc_text">
987 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
989 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
991 <tt>unsigned use_size()</tt> - Returns the number of users of the
993 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
994 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
996 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
998 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
1000 <p> These methods are the interface to access the def-use
1001 information in LLVM. As with all other iterators in LLVM, the naming
1002 conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
1004 <li><tt><a href="#Type">Type</a> *getType() const</tt>
1005 <p>This method returns the Type of the Value.</p>
1007 <li><tt>bool hasName() const</tt><br>
1008 <tt>std::string getName() const</tt><br>
1009 <tt>void setName(const std::string &Name)</tt>
1010 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
1011 be aware of the <a href="#nameWarning">precaution above</a>.</p>
1013 <li><tt>void replaceAllUsesWith(Value *V)</tt>
1015 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
1016 href="#User"><tt>User</tt>s</a> of the current value to refer to
1017 "<tt>V</tt>" instead. For example, if you detect that an instruction always
1018 produces a constant value (for example through constant folding), you can
1019 replace all uses of the instruction with the constant like this:</p>
1021 <pre> Inst->replaceAllUsesWith(ConstVal);<br></pre>
1026 <!-- ======================================================================= -->
1027 <div class="doc_subsection">
1028 <a name="User">The <tt>User</tt> class</a>
1031 <div class="doc_text">
1034 <tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
1035 doxygen info: <a href="/doxygen/classUser.html">User Class</a><br>
1036 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1038 <p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
1039 refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
1040 that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
1041 referring to. The <tt>User</tt> class itself is a subclass of
1044 <p>The operands of a <tt>User</tt> point directly to the LLVM <a
1045 href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
1046 Single Assignment (SSA) form, there can only be one definition referred to,
1047 allowing this direct connection. This connection provides the use-def
1048 information in LLVM.</p>
1052 <!-- _______________________________________________________________________ -->
1053 <div class="doc_subsubsection">
1054 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
1057 <div class="doc_text">
1059 <p>The <tt>User</tt> class exposes the operand list in two ways: through
1060 an index access interface and through an iterator based interface.</p>
1063 <li><tt>Value *getOperand(unsigned i)</tt><br>
1064 <tt>unsigned getNumOperands()</tt>
1065 <p> These two methods expose the operands of the <tt>User</tt> in a
1066 convenient form for direct access.</p></li>
1068 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
1070 <tt>User::op_const_iterator</tt> <tt>use_iterator op_begin()</tt> -
1071 Get an iterator to the start of the operand list.<br>
1072 <tt>use_iterator op_end()</tt> - Get an iterator to the end of the
1074 <p> Together, these methods make up the iterator based interface to
1075 the operands of a <tt>User</tt>.</p></li>
1080 <!-- ======================================================================= -->
1081 <div class="doc_subsection">
1082 <a name="Instruction">The <tt>Instruction</tt> class</a>
1085 <div class="doc_text">
1087 <p><tt>#include "</tt><tt><a
1088 href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
1089 doxygen info: <a href="/doxygen/classInstruction.html">Instruction Class</a><br>
1090 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1091 href="#Value"><tt>Value</tt></a></p>
1093 <p>The <tt>Instruction</tt> class is the common base class for all LLVM
1094 instructions. It provides only a few methods, but is a very commonly used
1095 class. The primary data tracked by the <tt>Instruction</tt> class itself is the
1096 opcode (instruction type) and the parent <a
1097 href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
1098 into. To represent a specific type of instruction, one of many subclasses of
1099 <tt>Instruction</tt> are used.</p>
1101 <p> Because the <tt>Instruction</tt> class subclasses the <a
1102 href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
1103 way as for other <a href="#User"><tt>User</tt></a>s (with the
1104 <tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
1105 <tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
1106 the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
1107 file contains some meta-data about the various different types of instructions
1108 in LLVM. It describes the enum values that are used as opcodes (for example
1109 <tt>Instruction::Add</tt> and <tt>Instruction::SetLE</tt>), as well as the
1110 concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
1111 example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
1112 href="#SetCondInst">SetCondInst</a></tt>). Unfortunately, the use of macros in
1113 this file confuses doxygen, so these enum values don't show up correctly in the
1114 <a href="/doxygen/classInstruction.html">doxygen output</a>.</p>
1118 <!-- _______________________________________________________________________ -->
1119 <div class="doc_subsubsection">
1120 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
1124 <div class="doc_text">
1127 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
1128 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
1129 this <tt>Instruction</tt> is embedded into.</p></li>
1130 <li><tt>bool mayWriteToMemory()</tt>
1131 <p>Returns true if the instruction writes to memory, i.e. it is a
1132 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
1133 <li><tt>unsigned getOpcode()</tt>
1134 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
1135 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
1136 <p>Returns another instance of the specified instruction, identical
1137 in all ways to the original except that the instruction has no parent
1138 (ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
1139 and it has no name</p></li>
1144 <!-- ======================================================================= -->
1145 <div class="doc_subsection">
1146 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
1149 <div class="doc_text">
1151 <p><tt>#include "<a href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
1152 doxygen info: <a href="/doxygen/classBasicBlock.html">BasicBlock Class</a><br>
1153 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1155 <p>This class represents a single entry multiple exit section of the code,
1156 commonly known as a basic block by the compiler community. The
1157 <tt>BasicBlock</tt> class maintains a list of <a
1158 href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
1159 Matching the language definition, the last element of this list of instructions
1160 is always a terminator instruction (a subclass of the <a
1161 href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
1163 <p>In addition to tracking the list of instructions that make up the block, the
1164 <tt>BasicBlock</tt> class also keeps track of the <a
1165 href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
1167 <p>Note that <tt>BasicBlock</tt>s themselves are <a
1168 href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
1169 like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
1174 <!-- _______________________________________________________________________ -->
1175 <div class="doc_subsubsection">
1176 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
1180 <div class="doc_text">
1183 <li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
1184 href="#Function">Function</a> *Parent = 0)</tt>
1185 <p>The <tt>BasicBlock</tt> constructor is used to create new basic
1186 blocks for insertion into a function. The constructor optionally takes
1187 a name for the new block, and a <a href="#Function"><tt>Function</tt></a>
1188 to insert it into. If the <tt>Parent</tt> parameter is specified, the
1189 new <tt>BasicBlock</tt> is automatically inserted at the end of the
1190 specified <a href="#Function"><tt>Function</tt></a>, if not specified,
1191 the BasicBlock must be manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p>
1193 <li><tt>BasicBlock::iterator</tt> - Typedef for instruction list
1195 <tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
1196 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,<tt>size()</tt>,<tt>empty()</tt>,<tt>rbegin()</tt>,<tt>rend()
1197 - </tt>STL style functions for accessing the instruction list.
1198 <p> These methods and typedefs are forwarding functions that have
1199 the same semantics as the standard library methods of the same names.
1200 These methods expose the underlying instruction list of a basic block in
1201 a way that is easy to manipulate. To get the full complement of
1202 container operations (including operations to update the list), you must
1203 use the <tt>getInstList()</tt> method.</p></li>
1204 <li><tt>BasicBlock::InstListType &getInstList()</tt>
1205 <p> This method is used to get access to the underlying container
1206 that actually holds the Instructions. This method must be used when
1207 there isn't a forwarding function in the <tt>BasicBlock</tt> class for
1208 the operation that you would like to perform. Because there are no
1209 forwarding functions for "updating" operations, you need to use this if
1210 you want to update the contents of a <tt>BasicBlock</tt>.</p></li>
1211 <li><tt><a href="#Function">Function</a> *getParent()</tt>
1212 <p> Returns a pointer to <a href="#Function"><tt>Function</tt></a>
1213 the block is embedded into, or a null pointer if it is homeless.</p></li>
1214 <li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
1215 <p> Returns a pointer to the terminator instruction that appears at
1216 the end of the <tt>BasicBlock</tt>. If there is no terminator
1217 instruction, or if the last instruction in the block is not a
1218 terminator, then a null pointer is returned.</p></li>
1223 <!-- ======================================================================= -->
1224 <div class="doc_subsection">
1225 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
1228 <div class="doc_text">
1231 href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
1232 doxygen info: <a href="/doxygen/classGlobalValue.html">GlobalValue Class</a><br>
1233 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1234 href="#Value"><tt>Value</tt></a></p>
1236 <p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
1237 href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
1238 visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
1239 Because they are visible at global scope, they are also subject to linking with
1240 other globals defined in different translation units. To control the linking
1241 process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
1242 <tt>GlobalValue</tt>s know whether they have internal or external linkage, as
1243 defined by the <tt>LinkageTypes</tt> enumerator.</p>
1245 <p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
1246 <tt>static</tt> in C), it is not visible to code outside the current translation
1247 unit, and does not participate in linking. If it has external linkage, it is
1248 visible to external code, and does participate in linking. In addition to
1249 linkage information, <tt>GlobalValue</tt>s keep track of which <a
1250 href="#Module"><tt>Module</tt></a> they are currently part of.</p>
1252 <p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
1253 by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
1254 global is always a pointer to its contents. It is important to remember this
1255 when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
1256 be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
1257 subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
1258 int]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
1259 the address of the first element of this array and the value of the
1260 <tt>GlobalVariable</tt> are the same, they have different types. The
1261 <tt>GlobalVariable</tt>'s type is <tt>[24 x int]</tt>. The first element's type
1262 is <tt>int.</tt> Because of this, accessing a global value requires you to
1263 dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
1264 can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
1265 Language Reference Manual</a>.</p>
1269 <!-- _______________________________________________________________________ -->
1270 <div class="doc_subsubsection">
1271 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
1275 <div class="doc_text">
1278 <li><tt>bool hasInternalLinkage() const</tt><br>
1279 <tt>bool hasExternalLinkage() const</tt><br>
1280 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
1281 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
1284 <li><tt><a href="#Module">Module</a> *getParent()</tt>
1285 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
1286 GlobalValue is currently embedded into.</p></li>
1291 <!-- ======================================================================= -->
1292 <div class="doc_subsection">
1293 <a name="Function">The <tt>Function</tt> class</a>
1296 <div class="doc_text">
1299 href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
1300 info: <a href="/doxygen/classFunction.html">Function Class</a><br> Superclasses:
1301 <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1302 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1304 <p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
1305 actually one of the more complex classes in the LLVM heirarchy because it must
1306 keep track of a large amount of data. The <tt>Function</tt> class keeps track
1307 of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal <a
1308 href="#Argument"><tt>Argument</tt></a>s, and a <a
1309 href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
1311 <p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
1312 commonly used part of <tt>Function</tt> objects. The list imposes an implicit
1313 ordering of the blocks in the function, which indicate how the code will be
1314 layed out by the backend. Additionally, the first <a
1315 href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
1316 <tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
1317 block. There are no implicit exit nodes, and in fact there may be multiple exit
1318 nodes from a single <tt>Function</tt>. If the <a
1319 href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
1320 the <tt>Function</tt> is actually a function declaration: the actual body of the
1321 function hasn't been linked in yet.</p>
1323 <p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
1324 <tt>Function</tt> class also keeps track of the list of formal <a
1325 href="#Argument"><tt>Argument</tt></a>s that the function receives. This
1326 container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
1327 nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
1328 the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
1330 <p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
1331 LLVM feature that is only used when you have to look up a value by name. Aside
1332 from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
1333 internally to make sure that there are not conflicts between the names of <a
1334 href="#Instruction"><tt>Instruction</tt></a>s, <a
1335 href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
1336 href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
1340 <!-- _______________________________________________________________________ -->
1341 <div class="doc_subsubsection">
1342 <a name="m_Function">Important Public Members of the <tt>Function</tt>
1346 <div class="doc_text">
1349 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
1350 *Ty, bool isInternal, const std::string &N = "", Module* Parent = 0)</tt>
1352 <p>Constructor used when you need to create new <tt>Function</tt>s to add
1353 the the program. The constructor must specify the type of the function to
1354 create and whether or not it should start out with internal or external
1355 linkage. The <a href="#FunctionType"><tt>FunctionType</tt></a> argument
1356 specifies the formal arguments and return value for the function. The same
1357 <a href="#FunctionTypel"><tt>FunctionType</tt></a> value can be used to
1358 create multiple functions. The <tt>Parent</tt> argument specifies the Module
1359 in which the function is defined. If this argument is provided, the function
1360 will automatically be inserted into that module's list of
1363 <li><tt>bool isExternal()</tt>
1365 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
1366 function is "external", it does not have a body, and thus must be resolved
1367 by linking with a function defined in a different translation unit.</p></li>
1369 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
1370 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
1372 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
1373 <tt>size()</tt>, <tt>empty()</tt>, <tt>rbegin()</tt>, <tt>rend()</tt>
1375 <p>These are forwarding methods that make it easy to access the contents of
1376 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
1379 <li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
1381 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
1382 is necessary to use when you need to update the list or perform a complex
1383 action that doesn't have a forwarding method.</p></li>
1385 <li><tt>Function::aiterator</tt> - Typedef for the argument list
1387 <tt>Function::const_aiterator</tt> - Typedef for const_iterator.<br>
1389 <tt>abegin()</tt>, <tt>aend()</tt>, <tt>afront()</tt>, <tt>aback()</tt>,
1390 <tt>asize()</tt>, <tt>aempty()</tt>, <tt>arbegin()</tt>, <tt>arend()</tt>
1392 <p>These are forwarding methods that make it easy to access the contents of
1393 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
1396 <li><tt>Function::ArgumentListType &getArgumentList()</tt>
1398 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
1399 necessary to use when you need to update the list or perform a complex
1400 action that doesn't have a forwarding method.</p></li>
1402 <li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
1404 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
1405 function. Because the entry block for the function is always the first
1406 block, this returns the first block of the <tt>Function</tt>.</p></li>
1408 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
1409 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
1411 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
1412 <tt>Function</tt> and returns the return type of the function, or the <a
1413 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
1416 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
1418 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1419 for this <tt>Function</tt>.</p></li>
1424 <!-- ======================================================================= -->
1425 <div class="doc_subsection">
1426 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
1429 <div class="doc_text">
1432 href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
1434 doxygen info: <a href="/doxygen/classGlobalVariable.html">GlobalVariable
1435 Class</a><br> Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1436 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1438 <p>Global variables are represented with the (suprise suprise)
1439 <tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
1440 subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
1441 always referenced by their address (global values must live in memory, so their
1442 "name" refers to their address). See <a
1443 href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global variables
1444 may have an initial value (which must be a <a
1445 href="#Constant"><tt>Constant</tt></a>), and if they have an initializer, they
1446 may be marked as "constant" themselves (indicating that their contents never
1447 change at runtime).</p>
1451 <!-- _______________________________________________________________________ -->
1452 <div class="doc_subsubsection">
1453 <a name="m_GlobalVariable">Important Public Members of the
1454 <tt>GlobalVariable</tt> class</a>
1457 <div class="doc_text">
1460 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
1461 isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
1462 *Initializer = 0, const std::string &Name = "", Module* Parent = 0)</tt>
1464 <p>Create a new global variable of the specified type. If
1465 <tt>isConstant</tt> is true then the global variable will be marked as
1466 unchanging for the program. The Linkage parameter specifies the type of
1467 linkage (internal, external, weak, linkonce, appending) for the variable. If
1468 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage, then
1469 the resultant global variable will have internal linkage. AppendingLinkage
1470 concatenates together all instances (in different translation units) of the
1471 variable into a single variable but is only applicable to arrays. See
1472 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
1473 further details on linkage types. Optionally an initializer, a name, and the
1474 module to put the variable into may be specified for the global variable as
1477 <li><tt>bool isConstant() const</tt>
1479 <p>Returns true if this is a global variable that is known not to
1480 be modified at runtime.</p></li>
1482 <li><tt>bool hasInitializer()</tt>
1484 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
1486 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
1488 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
1489 to call this method if there is no initializer.</p></li>
1494 <!-- ======================================================================= -->
1495 <div class="doc_subsection">
1496 <a name="Module">The <tt>Module</tt> class</a>
1499 <div class="doc_text">
1502 href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
1503 <a href="/doxygen/classModule.html">Module Class</a></p>
1505 <p>The <tt>Module</tt> class represents the top level structure present in LLVM
1506 programs. An LLVM module is effectively either a translation unit of the
1507 original program or a combination of several translation units merged by the
1508 linker. The <tt>Module</tt> class keeps track of a list of <a
1509 href="#Function"><tt>Function</tt></a>s, a list of <a
1510 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
1511 href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
1512 helpful member functions that try to make common operations easy.</p>
1516 <!-- _______________________________________________________________________ -->
1517 <div class="doc_subsubsection">
1518 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
1521 <div class="doc_text">
1524 <li><tt>Module::Module(std::string name = "")</tt></li>
1527 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
1528 provide a name for it (probably based on the name of the translation unit).</p>
1531 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
1532 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
1534 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
1535 <tt>size()</tt>, <tt>empty()</tt>, <tt>rbegin()</tt>, <tt>rend()</tt>
1537 <p>These are forwarding methods that make it easy to access the contents of
1538 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
1541 <li><tt>Module::FunctionListType &getFunctionList()</tt>
1543 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
1544 necessary to use when you need to update the list or perform a complex
1545 action that doesn't have a forwarding method.</p>
1547 <p><!-- Global Variable --></p></li>
1553 <li><tt>Module::giterator</tt> - Typedef for global variable list iterator<br>
1555 <tt>Module::const_giterator</tt> - Typedef for const_iterator.<br>
1557 <tt>gbegin()</tt>, <tt>gend()</tt>, <tt>gfront()</tt>, <tt>gback()</tt>,
1558 <tt>gsize()</tt>, <tt>gempty()</tt>, <tt>grbegin()</tt>, <tt>grend()</tt>
1560 <p> These are forwarding methods that make it easy to access the contents of
1561 a <tt>Module</tt> object's <a
1562 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
1564 <li><tt>Module::GlobalListType &getGlobalList()</tt>
1566 <p>Returns the list of <a
1567 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
1568 use when you need to update the list or perform a complex action that
1569 doesn't have a forwarding method.</p>
1571 <p><!-- Symbol table stuff --> </p></li>
1577 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
1579 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1580 for this <tt>Module</tt>.</p>
1582 <p><!-- Convenience methods --></p></li>
1588 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
1589 &Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
1591 <p>Look up the specified function in the <tt>Module</tt> <a
1592 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
1593 <tt>null</tt>.</p></li>
1595 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
1596 std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
1598 <p>Look up the specified function in the <tt>Module</tt> <a
1599 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
1600 external declaration for the function and return it.</p></li>
1602 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
1604 <p>If there is at least one entry in the <a
1605 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
1606 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
1609 <li><tt>bool addTypeName(const std::string &Name, const <a
1610 href="#Type">Type</a> *Ty)</tt>
1612 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1613 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
1614 name, true is returned and the <a
1615 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
1620 <!-- ======================================================================= -->
1621 <div class="doc_subsection">
1622 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
1625 <div class="doc_text">
1627 <p>Constant represents a base class for different types of constants. It
1628 is subclassed by ConstantBool, ConstantInt, ConstantSInt, ConstantUInt,
1629 ConstantArray etc for representing the various types of Constants.</p>
1633 <!-- _______________________________________________________________________ -->
1634 <div class="doc_subsubsection">
1635 <a name="m_Value">Important Public Methods</a>
1638 <div class="doc_text">
1641 <li><tt>bool isConstantExpr()</tt>: Returns true if it is a
1643 <hr> Important Subclasses of Constant
1646 <li>ConstantSInt : This subclass of Constant represents a signed
1649 <li><tt>int64_t getValue() const</tt>: Returns the underlying value of
1650 this constant. </li>
1653 <li>ConstantUInt : This class represents an unsigned integer.
1655 <li><tt>uint64_t getValue() const</tt>: Returns the underlying value
1656 of this constant. </li>
1659 <li>ConstantFP : This class represents a floating point constant.
1661 <li><tt>double getValue() const</tt>: Returns the underlying value of
1662 this constant. </li>
1665 <li>ConstantBool : This represents a boolean constant.
1667 <li><tt>bool getValue() const</tt>: Returns the underlying value of
1668 this constant. </li>
1671 <li>ConstantArray : This represents a constant array.
1673 <li><tt>const std::vector<Use> &getValues() const</tt>:
1674 Returns a Vecotr of component constants that makeup this array. </li>
1677 <li>ConstantStruct : This represents a constant struct.
1679 <li><tt>const std::vector<Use> &getValues() const</tt>:
1680 Returns a Vecotr of component constants that makeup this array. </li>
1683 <li>ConstantPointerRef : This represents a constant pointer value
1684 that is initialized to point to a global value, which lies at a
1685 constant fixed address.
1687 <li><tt>GlobalValue *getValue()</tt>: Returns the global
1688 value to which this pointer is pointing to. </li>
1697 <!-- ======================================================================= -->
1698 <div class="doc_subsection">
1699 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
1702 <div class="doc_text">
1704 <p>Type as noted earlier is also a subclass of a Value class. Any primitive
1705 type (like int, short etc) in LLVM is an instance of Type Class. All other
1706 types are instances of subclasses of type like FunctionType, ArrayType
1707 etc. DerivedType is the interface for all such dervied types including
1708 FunctionType, ArrayType, PointerType, StructType. Types can have names. They can
1709 be recursive (StructType). There exists exactly one instance of any type
1710 structure at a time. This allows using pointer equality of Type *s for comparing
1715 <!-- _______________________________________________________________________ -->
1716 <div class="doc_subsubsection">
1717 <a name="m_Value">Important Public Methods</a>
1720 <div class="doc_text">
1724 <li><tt>bool isSigned() const</tt>: Returns whether an integral numeric type
1725 is signed. This is true for SByteTy, ShortTy, IntTy, LongTy. Note that this is
1726 not true for Float and Double. </li>
1728 <li><tt>bool isUnsigned() const</tt>: Returns whether a numeric type is
1729 unsigned. This is not quite the complement of isSigned... nonnumeric types
1730 return false as they do with isSigned. This returns true for UByteTy,
1731 UShortTy, UIntTy, and ULongTy. </li>
1733 <li><tt>bool isInteger() const</tt>: Equilivent to isSigned() || isUnsigned(),
1734 but with only a single virtual function invocation.</li>
1736 <li><tt>bool isIntegral() const</tt>: Returns true if this is an integral
1737 type, which is either Bool type or one of the Integer types.</li>
1739 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
1740 floating point types.</li>
1742 <li><tt>isLosslesslyConvertableTo (const Type *Ty) const</tt>: Return true if
1743 this type can be converted to 'Ty' without any reinterpretation of bits. For
1744 example, uint to int or one pointer type to another.</li>
1747 <p>Derived Types</p>
1750 <li>SequentialType : This is subclassed by ArrayType and PointerType
1752 <li><tt>const Type * getElementType() const</tt>: Returns the type of
1753 each of the elements in the sequential type. </li>
1756 <li>ArrayType : This is a subclass of SequentialType and defines
1757 interface for array types.
1759 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
1760 elements in the array. </li>
1763 <li>PointerType : Subclass of SequentialType for pointer types. </li>
1764 <li>StructType : subclass of DerivedTypes for struct types </li>
1765 <li>FunctionType : subclass of DerivedTypes for function types.
1767 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
1769 <li><tt> const Type * getReturnType() const</tt>: Returns the
1770 return type of the function.</li>
1771 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
1772 the type of the ith parameter.</li>
1773 <li><tt> const unsigned getNumParams() const</tt>: Returns the
1774 number of formal parameters.</li>
1783 <!-- ======================================================================= -->
1784 <div class="doc_subsection">
1785 <a name="Argument">The <tt>Argument</tt> class</a>
1788 <div class="doc_text">
1790 <p>This subclass of Value defines the interface for incoming formal
1791 arguments to a function. A Function maitanis a list of its formal
1792 arguments. An argument has a pointer to the parent Function.</p>
1796 <!-- ======================================================================= -->
1797 <div class="doc_subsection">
1798 <a name="SymbolTable">The <tt>SymbolTable</tt> class</a>
1800 <div class="doc_text">
1801 <p>This class provides a symbol table that the
1802 <a href="#Function"><tt>Function</tt></a> and <a href="#Module">
1803 <tt>Module</tt></a> classes use for naming definitions. The symbol table can
1804 provide a name for any <a href="#Value"><tt>Value</tt></a> or
1805 <a href="#Type"><tt>Type</tt></a>. <tt>SymbolTable</tt> is an abstract data
1806 type. It hides the data it contains and provides access to it through a
1807 controlled interface.</p>
1809 <p>To use the <tt>SymbolTable</tt> well, you need to understand the
1810 structure of the information it holds. The class contains two
1811 <tt>std::map</tt> objects. The first, <tt>pmap</tt>, is a map of
1812 <tt>Type*</tt> to maps of name (<tt>std::string</tt>) to <tt>Value*</tt>.
1813 The second, <tt>tmap</tt>, is a map of names to <tt>Type*</tt>. Thus, Values
1814 are stored in two-dimensions and accessed by <tt>Type</tt> and name. Types,
1815 however, are stored in a single dimension and accessed only by name.</p>
1817 <p>The interface of this class provides three basic types of operations:
1819 <li><em>Accessors</em>. Accessors provide read-only access to information
1820 such as finding a value for a name with the
1821 <a href="#SymbolTable_lookup">lookup</a> method.</li>
1822 <li><em>Mutators</em>. Mutators allow the user to add information to the
1823 <tt>SymbolTable</tt> with methods like
1824 <a href="#SymbolTable_insert"><tt>insert</tt></a>.</li>
1825 <li><em>Iterators</em>. Iterators allow the user to traverse the content
1826 of the symbol table in well defined ways, such as the method
1827 <a href="#SymbolTable_type_begin"><tt>type_begin</tt></a>.</li>
1832 <dt><tt>Value* lookup(const Type* Ty, const std::string& name) const</tt>:
1834 <dd>The <tt>lookup</tt> method searches the type plane given by the
1835 <tt>Ty</tt> parameter for a <tt>Value</tt> with the provided <tt>name</tt>.
1836 If a suitable <tt>Value</tt> is not found, null is returned.</dd>
1838 <dt><tt>Type* lookupType( const std::string& name) const</tt>:</dt>
1839 <dd>The <tt>lookupType</tt> method searches through the types for a
1840 <tt>Type</tt> with the provided <tt>name</tt>. If a suitable <tt>Type</tt>
1841 is not found, null is returned.</dd>
1843 <dt><tt>bool hasTypes() const</tt>:</dt>
1844 <dd>This function returns true if an entry has been made into the type
1847 <dt><tt>bool isEmpty() const</tt>:</dt>
1848 <dd>This function returns true if both the value and types maps are
1851 <dt><tt>std::string get_name(const Value*) const</tt>:</dt>
1852 <dd>This function returns the name of the Value provided or the empty
1853 string if the Value is not in the symbol table.</dd>
1855 <dt><tt>std::string get_name(const Type*) const</tt>:</dt>
1856 <dd>This function returns the name of the Type provided or the empty
1857 string if the Type is not in the symbol table.</dd>
1862 <dt><tt>void insert(Value *Val)</tt>:</dt>
1863 <dd>This method adds the provided value to the symbol table. The Value must
1864 have both a name and a type which are extracted and used to place the value
1865 in the correct type plane under the value's name.</dd>
1867 <dt><tt>void insert(const std::string& Name, Value *Val)</tt>:</dt>
1868 <dd> Inserts a constant or type into the symbol table with the specified
1869 name. There can be a many to one mapping between names and constants
1872 <dt><tt>void insert(const std::string& Name, Type *Typ)</tt>:</dt>
1873 <dd> Inserts a type into the symbol table with the specified name. There
1874 can be a many-to-one mapping between names and types. This method
1875 allows a type with an existing entry in the symbol table to get
1878 <dt><tt>void remove(Value* Val)</tt>:</dt>
1879 <dd> This method removes a named value from the symbol table. The
1880 type and name of the Value are extracted from \p N and used to
1881 lookup the Value in the correct type plane. If the Value is
1882 not in the symbol table, this method silently ignores the
1885 <dt><tt>void remove(Type* Typ)</tt>:</dt>
1886 <dd> This method removes a named type from the symbol table. The
1887 name of the type is extracted from \P T and used to look up
1888 the Type in the type map. If the Type is not in the symbol
1889 table, this method silently ignores the request.</dd>
1891 <dt><tt>Value* remove(const std::string& Name, Value *Val)</tt>:</dt>
1892 <dd> Remove a constant or type with the specified name from the
1895 <dt><tt>Type* remove(const std::string& Name, Type* T)</tt>:</dt>
1896 <dd> Remove a type with the specified name from the symbol table.
1897 Returns the removed Type.</dd>
1899 <dt><tt>Value *value_remove(const value_iterator& It)</tt>:</dt>
1900 <dd> Removes a specific value from the symbol table.
1901 Returns the removed value.</dd>
1903 <dt><tt>bool strip()</tt>:</dt>
1904 <dd> This method will strip the symbol table of its names leaving
1905 the type and values. </dd>
1907 <dt><tt>void clear()</tt>:</dt>
1908 <dd>Empty the symbol table completely.</dd>
1912 <p>The following functions describe three types of iterators you can obtain
1913 the beginning or end of the sequence for both const and non-const. It is
1914 important to keep track of the different kinds of iterators. There are
1915 three idioms worth pointing out:</p>
1916 <table class="doc_table">
1917 <tr><th>Units</th><th>Iterator</th><th>Idiom</th></tr>
1919 <td>Planes Of name/Value maps</td><td>PI</td>
1921 for (SymbolTable::plane_const_iterator PI = ST.plane_begin(),
1922 PE = ST.plane_end(); PI != PE; ++PI ) {
1923 PI->first // This is the Type* of the plane
1924 PI->second // This is the SymbolTable::ValueMap of name/Value pairs
1928 <td>All name/Type Pairs</td><td>TI</td>
1930 for (SymbolTable::type_const_iterator TI = ST.type_begin(),
1931 TE = ST.type_end(); TI != TE; ++TI )
1932 TI->first // This is the name of the type
1933 TI->second // This is the Type* value associated with the name
1937 <td>name/Value pairs in a plane</td><td>VI</td>
1939 for (SymbolTable::value_const_iterator VI = ST.value_begin(SomeType),
1940 VE = ST.value_end(SomeType); VI != VE; ++VI )
1941 VI->first // This is the name of the Value
1942 VI->second // This is the Value* value associated with the name
1946 <p>Using the recommended iterator names and idioms will help you avoid
1947 making mistakes. Of particular note, make sure that whenever you use
1948 value_begin(SomeType) that you always compare the resulting iterator
1949 with value_end(SomeType) not value_end(SomeOtherType) or else you
1950 will loop infinitely.</p>
1954 <dt><tt>plane_iterator plane_begin()</tt>:</dt>
1955 <dd>Get an iterator that starts at the beginning of the type planes.
1956 The iterator will iterate over the Type/ValueMap pairs in the
1959 <dt><tt>plane_const_iterator plane_begin() const</tt>:</dt>
1960 <dd>Get a const_iterator that starts at the beginning of the type
1961 planes. The iterator will iterate over the Type/ValueMap pairs
1962 in the type planes. </dd>
1964 <dt><tt>plane_iterator plane_end()</tt>:</dt>
1965 <dd>Get an iterator at the end of the type planes. This serves as
1966 the marker for end of iteration over the type planes.</dd>
1968 <dt><tt>plane_const_iterator plane_end() const</tt>:</dt>
1969 <dd>Get a const_iterator at the end of the type planes. This serves as
1970 the marker for end of iteration over the type planes.</dd>
1972 <dt><tt>value_iterator value_begin(const Type *Typ)</tt>:</dt>
1973 <dd>Get an iterator that starts at the beginning of a type plane.
1974 The iterator will iterate over the name/value pairs in the type plane.
1975 Note: The type plane must already exist before using this.</dd>
1977 <dt><tt>value_const_iterator value_begin(const Type *Typ) const</tt>:</dt>
1978 <dd>Get a const_iterator that starts at the beginning of a type plane.
1979 The iterator will iterate over the name/value pairs in the type plane.
1980 Note: The type plane must already exist before using this.</dd>
1982 <dt><tt>value_iterator value_end(const Type *Typ)</tt>:</dt>
1983 <dd>Get an iterator to the end of a type plane. This serves as the marker
1984 for end of iteration of the type plane.
1985 Note: The type plane must already exist before using this.</dd>
1987 <dt><tt>value_const_iterator value_end(const Type *Typ) const</tt>:</dt>
1988 <dd>Get a const_iterator to the end of a type plane. This serves as the
1989 marker for end of iteration of the type plane.
1990 Note: the type plane must already exist before using this.</dd>
1992 <dt><tt>type_iterator type_begin()</tt>:</dt>
1993 <dd>Get an iterator to the start of the name/Type map.</dd>
1995 <dt><tt>type_const_iterator type_begin() cons</tt>:</dt>
1996 <dd> Get a const_iterator to the start of the name/Type map.</dd>
1998 <dt><tt>type_iterator type_end()</tt>:</dt>
1999 <dd>Get an iterator to the end of the name/Type map. This serves as the
2000 marker for end of iteration of the types.</dd>
2002 <dt><tt>type_const_iterator type_end() const</tt>:</dt>
2003 <dd>Get a const-iterator to the end of the name/Type map. This serves
2004 as the marker for end of iteration of the types.</dd>
2006 <dt><tt>plane_const_iterator find(const Type* Typ ) const</tt>:</dt>
2007 <dd>This method returns a plane_const_iterator for iteration over
2008 the type planes starting at a specific plane, given by \p Ty.</dd>
2010 <dt><tt>plane_iterator find( const Type* Typ </tt>:</dt>
2011 <dd>This method returns a plane_iterator for iteration over the
2012 type planes starting at a specific plane, given by \p Ty.</dd>
2014 <dt><tt>const ValueMap* findPlane( const Type* Typ ) cons</tt>:</dt>
2015 <dd>This method returns a ValueMap* for a specific type plane. This
2016 interface is deprecated and may go away in the future.</dd>
2020 <!-- *********************************************************************** -->
2023 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2024 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
2025 <a href="http://validator.w3.org/check/referer"><img
2026 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
2028 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
2029 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
2030 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br>
2031 Last modified: $Date$