1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM Programmer's Manual</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
10 <div class="doc_title">
11 LLVM Programmer's Manual
15 <li><a href="#introduction">Introduction</a></li>
16 <li><a href="#general">General Information</a>
18 <li><a href="#stl">The C++ Standard Template Library</a></li>
20 <li>The <tt>-time-passes</tt> option</li>
21 <li>How to use the LLVM Makefile system</li>
22 <li>How to write a regression test</li>
27 <li><a href="#apis">Important and useful LLVM APIs</a>
29 <li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
30 and <tt>dyn_cast<></tt> templates</a> </li>
31 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
34 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
35 and the <tt>-debug-only</tt> option</a> </li>
38 <li><a href="#Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
41 <li>The <tt>InstVisitor</tt> template
42 <li>The general graph API
44 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
47 <li><a href="#common">Helpful Hints for Common Operations</a>
49 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
51 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
52 in a <tt>Function</tt></a> </li>
53 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
54 in a <tt>BasicBlock</tt></a> </li>
55 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
56 in a <tt>Function</tt></a> </li>
57 <li><a href="#iterate_convert">Turning an iterator into a
58 class pointer</a> </li>
59 <li><a href="#iterate_complex">Finding call sites: a more
60 complex example</a> </li>
61 <li><a href="#calls_and_invokes">Treating calls and invokes
62 the same way</a> </li>
63 <li><a href="#iterate_chains">Iterating over def-use &
64 use-def chains</a> </li>
67 <li><a href="#simplechanges">Making simple changes</a>
69 <li><a href="#schanges_creating">Creating and inserting new
70 <tt>Instruction</tt>s</a> </li>
71 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
72 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
73 with another <tt>Value</tt></a> </li>
77 <li>Working with the Control Flow Graph
79 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
87 <li><a href="#advanced">Advanced Topics</a>
89 <li><a href="#TypeResolve">LLVM Type Resolution</a>
91 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
92 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
93 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
94 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
97 <li><a href="#SymbolTable">The <tt>SymbolTable</tt> class </a></li>
100 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
102 <li><a href="#Value">The <tt>Value</tt> class</a>
104 <li><a href="#User">The <tt>User</tt> class</a>
106 <li><a href="#Instruction">The <tt>Instruction</tt> class</a>
108 <li><a href="#GetElementPtrInst">The <tt>GetElementPtrInst</tt> class</a></li>
111 <li><a href="#Module">The <tt>Module</tt> class</a></li>
112 <li><a href="#Constant">The <tt>Constant</tt> class</a>
114 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
116 <li><a href="#BasicBlock">The <tt>BasicBlock</tt>class</a></li>
117 <li><a href="#Function">The <tt>Function</tt> class</a></li>
118 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
125 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
126 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
133 <div class="doc_author">
134 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
135 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
136 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
137 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
140 <!-- *********************************************************************** -->
141 <div class="doc_section">
142 <a name="introduction">Introduction </a>
144 <!-- *********************************************************************** -->
146 <div class="doc_text">
148 <p>This document is meant to highlight some of the important classes and
149 interfaces available in the LLVM source-base. This manual is not
150 intended to explain what LLVM is, how it works, and what LLVM code looks
151 like. It assumes that you know the basics of LLVM and are interested
152 in writing transformations or otherwise analyzing or manipulating the
155 <p>This document should get you oriented so that you can find your
156 way in the continuously growing source code that makes up the LLVM
157 infrastructure. Note that this manual is not intended to serve as a
158 replacement for reading the source code, so if you think there should be
159 a method in one of these classes to do something, but it's not listed,
160 check the source. Links to the <a href="/doxygen/">doxygen</a> sources
161 are provided to make this as easy as possible.</p>
163 <p>The first section of this document describes general information that is
164 useful to know when working in the LLVM infrastructure, and the second describes
165 the Core LLVM classes. In the future this manual will be extended with
166 information describing how to use extension libraries, such as dominator
167 information, CFG traversal routines, and useful utilities like the <tt><a
168 href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
172 <!-- *********************************************************************** -->
173 <div class="doc_section">
174 <a name="general">General Information</a>
176 <!-- *********************************************************************** -->
178 <div class="doc_text">
180 <p>This section contains general information that is useful if you are working
181 in the LLVM source-base, but that isn't specific to any particular API.</p>
185 <!-- ======================================================================= -->
186 <div class="doc_subsection">
187 <a name="stl">The C++ Standard Template Library</a>
190 <div class="doc_text">
192 <p>LLVM makes heavy use of the C++ Standard Template Library (STL),
193 perhaps much more than you are used to, or have seen before. Because of
194 this, you might want to do a little background reading in the
195 techniques used and capabilities of the library. There are many good
196 pages that discuss the STL, and several books on the subject that you
197 can get, so it will not be discussed in this document.</p>
199 <p>Here are some useful links:</p>
203 <li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
204 reference</a> - an excellent reference for the STL and other parts of the
205 standard C++ library.</li>
207 <li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
208 O'Reilly book in the making. It has a decent
210 Reference that rivals Dinkumware's, and is unfortunately no longer free since the book has been
213 <li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
216 <li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
218 href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
221 <li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
224 <li><a href="http://64.78.49.204/">
225 Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
230 <p>You are also encouraged to take a look at the <a
231 href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
232 to write maintainable code more than where to put your curly braces.</p>
236 <!-- ======================================================================= -->
237 <div class="doc_subsection">
238 <a name="stl">Other useful references</a>
241 <div class="doc_text">
244 <li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
245 Branch and Tag Primer</a></li>
246 <li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
247 static and shared libraries across platforms</a></li>
252 <!-- *********************************************************************** -->
253 <div class="doc_section">
254 <a name="apis">Important and useful LLVM APIs</a>
256 <!-- *********************************************************************** -->
258 <div class="doc_text">
260 <p>Here we highlight some LLVM APIs that are generally useful and good to
261 know about when writing transformations.</p>
265 <!-- ======================================================================= -->
266 <div class="doc_subsection">
267 <a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
268 <tt>dyn_cast<></tt> templates</a>
271 <div class="doc_text">
273 <p>The LLVM source-base makes extensive use of a custom form of RTTI.
274 These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
275 operator, but they don't have some drawbacks (primarily stemming from
276 the fact that <tt>dynamic_cast<></tt> only works on classes that
277 have a v-table). Because they are used so often, you must know what they
278 do and how they work. All of these templates are defined in the <a
279 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
280 file (note that you very rarely have to include this file directly).</p>
283 <dt><tt>isa<></tt>: </dt>
285 <dd>The <tt>isa<></tt> operator works exactly like the Java
286 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
287 a reference or pointer points to an instance of the specified class. This can
288 be very useful for constraint checking of various sorts (example below).</dd>
290 <dt><tt>cast<></tt>: </dt>
292 <dd>The <tt>cast<></tt> operator is a "checked cast" operation. It
293 converts a pointer or reference from a base class to a derived cast, causing
294 an assertion failure if it is not really an instance of the right type. This
295 should be used in cases where you have some information that makes you believe
296 that something is of the right type. An example of the <tt>isa<></tt>
297 and <tt>cast<></tt> template is:
300 static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
301 if (isa<<a href="#Constant">Constant</a>>(V) || isa<<a href="#Argument">Argument</a>>(V) || isa<<a href="#GlobalValue">GlobalValue</a>>(V))
304 <i>// Otherwise, it must be an instruction...</i>
305 return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
309 <p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
310 by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
315 <dt><tt>dyn_cast<></tt>:</dt>
317 <dd>The <tt>dyn_cast<></tt> operator is a "checking cast" operation. It
318 checks to see if the operand is of the specified type, and if so, returns a
319 pointer to it (this operator does not work with references). If the operand is
320 not of the correct type, a null pointer is returned. Thus, this works very
321 much like the <tt>dynamic_cast<></tt> operator in C++, and should be
322 used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
323 operator is used in an <tt>if</tt> statement or some other flow control
327 if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
332 <p>This form of the <tt>if</tt> statement effectively combines together a call
333 to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
334 statement, which is very convenient.</p>
336 <p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
337 <tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
338 abused. In particular, you should not use big chained <tt>if/then/else</tt>
339 blocks to check for lots of different variants of classes. If you find
340 yourself wanting to do this, it is much cleaner and more efficient to use the
341 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
345 <dt><tt>cast_or_null<></tt>: </dt>
347 <dd>The <tt>cast_or_null<></tt> operator works just like the
348 <tt>cast<></tt> operator, except that it allows for a null pointer as an
349 argument (which it then propagates). This can sometimes be useful, allowing
350 you to combine several null checks into one.</dd>
352 <dt><tt>dyn_cast_or_null<></tt>: </dt>
354 <dd>The <tt>dyn_cast_or_null<></tt> operator works just like the
355 <tt>dyn_cast<></tt> operator, except that it allows for a null pointer
356 as an argument (which it then propagates). This can sometimes be useful,
357 allowing you to combine several null checks into one.</dd>
361 <p>These five templates can be used with any classes, whether they have a
362 v-table or not. To add support for these templates, you simply need to add
363 <tt>classof</tt> static methods to the class you are interested casting
364 to. Describing this is currently outside the scope of this document, but there
365 are lots of examples in the LLVM source base.</p>
369 <!-- ======================================================================= -->
370 <div class="doc_subsection">
371 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
374 <div class="doc_text">
376 <p>Often when working on your pass you will put a bunch of debugging printouts
377 and other code into your pass. After you get it working, you want to remove
378 it... but you may need it again in the future (to work out new bugs that you run
381 <p> Naturally, because of this, you don't want to delete the debug printouts,
382 but you don't want them to always be noisy. A standard compromise is to comment
383 them out, allowing you to enable them if you need them in the future.</p>
385 <p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
386 file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
387 this problem. Basically, you can put arbitrary code into the argument of the
388 <tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
389 tool) is run with the '<tt>-debug</tt>' command line argument:</p>
391 <pre> ... <br> DEBUG(std::cerr << "I am here!\n");<br> ...<br></pre>
393 <p>Then you can run your pass like this:</p>
395 <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>
397 <p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
398 to not have to create "yet another" command line option for the debug output for
399 your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
400 so they do not cause a performance impact at all (for the same reason, they
401 should also not contain side-effects!).</p>
403 <p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
404 enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
405 "<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
406 program hasn't been started yet, you can always just run it with
411 <!-- _______________________________________________________________________ -->
412 <div class="doc_subsubsection">
413 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
414 the <tt>-debug-only</tt> option</a>
417 <div class="doc_text">
419 <p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
420 just turns on <b>too much</b> information (such as when working on the code
421 generator). If you want to enable debug information with more fine-grained
422 control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
423 option as follows:</p>
425 <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>
427 <p>Then you can run your pass like this:</p>
429 <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>
431 <p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
432 a file, to specify the debug type for the entire module (if you do this before
433 you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
434 <tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
435 "bar", because there is no system in place to ensure that names do not
436 conflict. If two different modules use the same string, they will all be turned
437 on when the name is specified. This allows, for example, all debug information
438 for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
439 even if the source lives in multiple files.</p>
443 <!-- ======================================================================= -->
444 <div class="doc_subsection">
445 <a name="Statistic">The <tt>Statistic</tt> template & <tt>-stats</tt>
449 <div class="doc_text">
452 href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
453 provides a template named <tt>Statistic</tt> that is used as a unified way to
454 keep track of what the LLVM compiler is doing and how effective various
455 optimizations are. It is useful to see what optimizations are contributing to
456 making a particular program run faster.</p>
458 <p>Often you may run your pass on some big program, and you're interested to see
459 how many times it makes a certain transformation. Although you can do this with
460 hand inspection, or some ad-hoc method, this is a real pain and not very useful
461 for big programs. Using the <tt>Statistic</tt> template makes it very easy to
462 keep track of this information, and the calculated information is presented in a
463 uniform manner with the rest of the passes being executed.</p>
465 <p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
466 it are as follows:</p>
469 <li>Define your statistic like this:
470 <pre>static Statistic<> NumXForms("mypassname", "The # of times I did stuff");<br></pre>
472 <p>The <tt>Statistic</tt> template can emulate just about any data-type,
473 but if you do not specify a template argument, it defaults to acting like
474 an unsigned int counter (this is usually what you want).</p></li>
476 <li>Whenever you make a transformation, bump the counter:
477 <pre> ++NumXForms; // I did stuff<br></pre>
481 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
482 statistics gathered, use the '<tt>-stats</tt>' option:</p>
484 <pre> $ opt -stats -mypassname < program.bc > /dev/null<br> ... statistic output ...<br></pre>
486 <p> When running <tt>gccas</tt> on a C file from the SPEC benchmark
487 suite, it gives a report that looks like this:</p>
489 <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>
491 <p>Obviously, with so many optimizations, having a unified framework for this
492 stuff is very nice. Making your pass fit well into the framework makes it more
493 maintainable and useful.</p>
497 <!-- ======================================================================= -->
498 <div class="doc_subsection">
499 <a name="ViewGraph">Viewing graphs while debugging code</a>
502 <div class="doc_text">
504 <p>Several of the important data structures in LLVM are graphs: for example
505 CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
506 LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
507 <a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
508 DAGs</a>. In many cases, while debugging various parts of the compiler, it is
509 nice to instantly visualize these graphs.</p>
511 <p>LLVM provides several callbacks that are available in a debug build to do
512 exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
513 the current LLVM tool will pop up a window containing the CFG for the function
514 where each basic block is a node in the graph, and each node contains the
515 instructions in the block. Similarly, there also exists
516 <tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
517 <tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
518 and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
519 you can usually use something like "<tt>call DAG.viewGraph()</tt>" to pop
520 up a window. Alternatively, you can sprinkle calls to these functions in your
521 code in places you want to debug.</p>
523 <p>Getting this to work requires a small amount of configuration. On Unix
524 systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
525 toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
526 Mac OS/X, download and install the Mac OS/X <a
527 href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
528 <tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or whereever you install
529 it) to your path. Once in your system and path are set up, rerun the LLVM
530 configure script and rebuild LLVM to enable this functionality.</p>
535 <!-- *********************************************************************** -->
536 <div class="doc_section">
537 <a name="common">Helpful Hints for Common Operations</a>
539 <!-- *********************************************************************** -->
541 <div class="doc_text">
543 <p>This section describes how to perform some very simple transformations of
544 LLVM code. This is meant to give examples of common idioms used, showing the
545 practical side of LLVM transformations. <p> Because this is a "how-to" section,
546 you should also read about the main classes that you will be working with. The
547 <a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
548 and descriptions of the main classes that you should know about.</p>
552 <!-- NOTE: this section should be heavy on example code -->
553 <!-- ======================================================================= -->
554 <div class="doc_subsection">
555 <a name="inspection">Basic Inspection and Traversal Routines</a>
558 <div class="doc_text">
560 <p>The LLVM compiler infrastructure have many different data structures that may
561 be traversed. Following the example of the C++ standard template library, the
562 techniques used to traverse these various data structures are all basically the
563 same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
564 method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
565 function returns an iterator pointing to one past the last valid element of the
566 sequence, and there is some <tt>XXXiterator</tt> data type that is common
567 between the two operations.</p>
569 <p>Because the pattern for iteration is common across many different aspects of
570 the program representation, the standard template library algorithms may be used
571 on them, and it is easier to remember how to iterate. First we show a few common
572 examples of the data structures that need to be traversed. Other data
573 structures are traversed in very similar ways.</p>
577 <!-- _______________________________________________________________________ -->
578 <div class="doc_subsubsection">
579 <a name="iterate_function">Iterating over the </a><a
580 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
581 href="#Function"><tt>Function</tt></a>
584 <div class="doc_text">
586 <p>It's quite common to have a <tt>Function</tt> instance that you'd like to
587 transform in some way; in particular, you'd like to manipulate its
588 <tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
589 the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
590 an example that prints the name of a <tt>BasicBlock</tt> and the number of
591 <tt>Instruction</tt>s it contains:</p>
593 <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> std::cerr << "Basic block (name=" << i->getName() << ") has " <br> << i->size() << " instructions.\n";<br> }<br></pre>
595 <p>Note that i can be used as if it were a pointer for the purposes of
596 invoking member functions of the <tt>Instruction</tt> class. This is
597 because the indirection operator is overloaded for the iterator
598 classes. In the above code, the expression <tt>i->size()</tt> is
599 exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
603 <!-- _______________________________________________________________________ -->
604 <div class="doc_subsubsection">
605 <a name="iterate_basicblock">Iterating over the </a><a
606 href="#Instruction"><tt>Instruction</tt></a>s in a <a
607 href="#BasicBlock"><tt>BasicBlock</tt></a>
610 <div class="doc_text">
612 <p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
613 easy to iterate over the individual instructions that make up
614 <tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
615 a <tt>BasicBlock</tt>:</p>
618 // blk is a pointer to a BasicBlock instance
619 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
620 // the next statement works since operator<<(ostream&,...)
621 // is overloaded for Instruction&
622 std::cerr << *i << "\n";
625 <p>However, this isn't really the best way to print out the contents of a
626 <tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
627 anything you'll care about, you could have just invoked the print routine on the
628 basic block itself: <tt>std::cerr << *blk << "\n";</tt>.</p>
632 <!-- _______________________________________________________________________ -->
633 <div class="doc_subsubsection">
634 <a name="iterate_institer">Iterating over the </a><a
635 href="#Instruction"><tt>Instruction</tt></a>s in a <a
636 href="#Function"><tt>Function</tt></a>
639 <div class="doc_text">
641 <p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
642 <tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
643 <tt>InstIterator</tt> should be used instead. You'll need to include <a
644 href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
645 and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
646 small example that shows how to dump all instructions in a function to the standard error stream:<p>
648 <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> std::cerr << *i << "\n";<br></pre>
649 Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
650 worklist with its initial contents. For example, if you wanted to
651 initialize a worklist to contain all instructions in a <tt>Function</tt>
652 F, all you would need to do is something like:
653 <pre>std::set<Instruction*> worklist;<br>worklist.insert(inst_begin(F), inst_end(F));<br></pre>
655 <p>The STL set <tt>worklist</tt> would now contain all instructions in the
656 <tt>Function</tt> pointed to by F.</p>
660 <!-- _______________________________________________________________________ -->
661 <div class="doc_subsubsection">
662 <a name="iterate_convert">Turning an iterator into a class pointer (and
666 <div class="doc_text">
668 <p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
669 instance when all you've got at hand is an iterator. Well, extracting
670 a reference or a pointer from an iterator is very straight-forward.
671 Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
672 is a <tt>BasicBlock::const_iterator</tt>:</p>
674 <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>
676 <p>However, the iterators you'll be working with in the LLVM framework are
677 special: they will automatically convert to a ptr-to-instance type whenever they
678 need to. Instead of dereferencing the iterator and then taking the address of
679 the result, you can simply assign the iterator to the proper pointer type and
680 you get the dereference and address-of operation as a result of the assignment
681 (behind the scenes, this is a result of overloading casting mechanisms). Thus
682 the last line of the last example,</p>
684 <pre>Instruction* pinst = &*i;</pre>
686 <p>is semantically equivalent to</p>
688 <pre>Instruction* pinst = i;</pre>
690 <p>It's also possible to turn a class pointer into the corresponding iterator,
691 and this is a constant time operation (very efficient). The following code
692 snippet illustrates use of the conversion constructors provided by LLVM
693 iterators. By using these, you can explicitly grab the iterator of something
694 without actually obtaining it via iteration over some structure:</p>
696 <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()) std::cerr << *it << "\n";<br>}<br></pre>
700 <!--_______________________________________________________________________-->
701 <div class="doc_subsubsection">
702 <a name="iterate_complex">Finding call sites: a slightly more complex
706 <div class="doc_text">
708 <p>Say that you're writing a FunctionPass and would like to count all the
709 locations in the entire module (that is, across every <tt>Function</tt>) where a
710 certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
711 learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
712 much more straight-forward manner, but this example will allow us to explore how
713 you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudocode, this
714 is what we want to do:</p>
716 <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>
718 <p>And the actual code is (remember, since we're writing a
719 <tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
720 override the <tt>runOnFunction</tt> method...):</p>
722 <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
723 href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
724 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>
728 <!--_______________________________________________________________________-->
729 <div class="doc_subsubsection">
730 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
733 <div class="doc_text">
735 <p>You may have noticed that the previous example was a bit oversimplified in
736 that it did not deal with call sites generated by 'invoke' instructions. In
737 this, and in other situations, you may find that you want to treat
738 <tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
739 most-specific common base class is <tt>Instruction</tt>, which includes lots of
740 less closely-related things. For these cases, LLVM provides a handy wrapper
742 href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
743 It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
744 methods that provide functionality common to <tt>CallInst</tt>s and
745 <tt>InvokeInst</tt>s.</p>
747 <p>This class has "value semantics": it should be passed by value, not by
748 reference and it should not be dynamically allocated or deallocated using
749 <tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
750 assignable and constructable, with costs equivalents to that of a bare pointer.
751 If you look at its definition, it has only a single pointer member.</p>
755 <!--_______________________________________________________________________-->
756 <div class="doc_subsubsection">
757 <a name="iterate_chains">Iterating over def-use & use-def chains</a>
760 <div class="doc_text">
762 <p>Frequently, we might have an instance of the <a
763 href="/doxygen/structllvm_1_1Value.html">Value Class</a> and we want to
764 determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
765 <tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
766 For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
767 particular function <tt>foo</tt>. Finding all of the instructions that
768 <i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
771 <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> std::cerr << "F is used in instruction:\n";<br> std::cerr << *Inst << "\n";<br> }<br>}<br></pre>
773 <p>Alternately, it's common to have an instance of the <a
774 href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
775 <tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
776 <tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
777 <tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
778 all of the values that a particular instruction uses (that is, the operands of
779 the particular <tt>Instruction</tt>):</p>
781 <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>
784 def-use chains ("finding all users of"): Value::use_begin/use_end
785 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
790 <!-- ======================================================================= -->
791 <div class="doc_subsection">
792 <a name="simplechanges">Making simple changes</a>
795 <div class="doc_text">
797 <p>There are some primitive transformation operations present in the LLVM
798 infrastructure that are worth knowing about. When performing
799 transformations, it's fairly common to manipulate the contents of basic
800 blocks. This section describes some of the common methods for doing so
801 and gives example code.</p>
805 <!--_______________________________________________________________________-->
806 <div class="doc_subsubsection">
807 <a name="schanges_creating">Creating and inserting new
808 <tt>Instruction</tt>s</a>
811 <div class="doc_text">
813 <p><i>Instantiating Instructions</i></p>
815 <p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
816 constructor for the kind of instruction to instantiate and provide the necessary
817 parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
818 (const-ptr-to) <tt>Type</tt>. Thus:</p>
820 <pre>AllocaInst* ai = new AllocaInst(Type::IntTy);</pre>
822 <p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
823 one integer in the current stack frame, at runtime. Each <tt>Instruction</tt>
824 subclass is likely to have varying default parameters which change the semantics
825 of the instruction, so refer to the <a
826 href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
827 Instruction</a> that you're interested in instantiating.</p>
829 <p><i>Naming values</i></p>
831 <p>It is very useful to name the values of instructions when you're able to, as
832 this facilitates the debugging of your transformations. If you end up looking
833 at generated LLVM machine code, you definitely want to have logical names
834 associated with the results of instructions! By supplying a value for the
835 <tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
836 associate a logical name with the result of the instruction's execution at
837 runtime. For example, say that I'm writing a transformation that dynamically
838 allocates space for an integer on the stack, and that integer is going to be
839 used as some kind of index by some other code. To accomplish this, I place an
840 <tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
841 <tt>Function</tt>, and I'm intending to use it within the same
842 <tt>Function</tt>. I might do:</p>
844 <pre>AllocaInst* pa = new AllocaInst(Type::IntTy, 0, "indexLoc");</pre>
846 <p>where <tt>indexLoc</tt> is now the logical name of the instruction's
847 execution value, which is a pointer to an integer on the runtime stack.</p>
849 <p><i>Inserting instructions</i></p>
851 <p>There are essentially two ways to insert an <tt>Instruction</tt>
852 into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
855 <li>Insertion into an explicit instruction list
857 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
858 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
859 before <tt>*pi</tt>, we do the following: </p>
861 <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>
863 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
864 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
865 classes provide constructors which take a pointer to a
866 <tt>BasicBlock</tt> to be appended to. For example code that
869 <pre> BasicBlock *pb = ...;<br> Instruction *newInst = new Instruction(...);<br> pb->getInstList().push_back(newInst); // appends newInst to pb<br></pre>
873 <pre> BasicBlock *pb = ...;<br> Instruction *newInst = new Instruction(..., pb);<br></pre>
875 <p>which is much cleaner, especially if you are creating
876 long instruction streams.</p></li>
878 <li>Insertion into an implicit instruction list
880 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
881 are implicitly associated with an existing instruction list: the instruction
882 list of the enclosing basic block. Thus, we could have accomplished the same
883 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
886 <pre> Instruction *pi = ...;<br> Instruction *newInst = new Instruction(...);<br> pi->getParent()->getInstList().insert(pi, newInst);<br></pre>
888 <p>In fact, this sequence of steps occurs so frequently that the
889 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
890 constructors which take (as a default parameter) a pointer to an
891 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
892 precede. That is, <tt>Instruction</tt> constructors are capable of
893 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
894 provided instruction, immediately before that instruction. Using an
895 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
896 parameter, the above code becomes:</p>
898 <pre>Instruction* pi = ...;<br>Instruction* newInst = new Instruction(..., pi);<br></pre>
900 <p>which is much cleaner, especially if you're creating a lot of
901 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
906 <!--_______________________________________________________________________-->
907 <div class="doc_subsubsection">
908 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
911 <div class="doc_text">
913 <p>Deleting an instruction from an existing sequence of instructions that form a
914 <a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
915 you must have a pointer to the instruction that you wish to delete. Second, you
916 need to obtain the pointer to that instruction's basic block. You use the
917 pointer to the basic block to get its list of instructions and then use the
918 erase function to remove your instruction. For example:</p>
920 <pre> <a href="#Instruction">Instruction</a> *I = .. ;<br> <a
921 href="#BasicBlock">BasicBlock</a> *BB = I->getParent();<br> BB->getInstList().erase(I);<br></pre>
925 <!--_______________________________________________________________________-->
926 <div class="doc_subsubsection">
927 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
931 <div class="doc_text">
933 <p><i>Replacing individual instructions</i></p>
935 <p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
936 permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
937 and <tt>ReplaceInstWithInst</tt>.</p>
939 <h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
942 <li><tt>ReplaceInstWithValue</tt>
944 <p>This function replaces all uses (within a basic block) of a given
945 instruction with a value, and then removes the original instruction. The
946 following example illustrates the replacement of the result of a particular
947 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
948 pointer to an integer.</p>
950 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,<br> Constant::getNullValue(PointerType::get(Type::IntTy)));<br></pre></li>
952 <li><tt>ReplaceInstWithInst</tt>
954 <p>This function replaces a particular instruction with another
955 instruction. The following example illustrates the replacement of one
956 <tt>AllocaInst</tt> with another.</p>
958 <pre>AllocaInst* instToReplace = ...;<br>BasicBlock::iterator ii(instToReplace);<br>ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,<br> new AllocaInst(Type::IntTy, 0, "ptrToReplacedInt"));<br></pre></li>
961 <p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
963 <p>You can use <tt>Value::replaceAllUsesWith</tt> and
964 <tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
965 doxygen documentation for the <a href="/doxygen/structllvm_1_1Value.html">Value Class</a>
966 and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
969 <!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
970 include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
971 ReplaceInstWithValue, ReplaceInstWithInst -->
975 <!-- *********************************************************************** -->
976 <div class="doc_section">
977 <a name="advanced">Advanced Topics</a>
979 <!-- *********************************************************************** -->
981 <div class="doc_text">
983 This section describes some of the advanced or obscure API's that most clients
984 do not need to be aware of. These API's tend manage the inner workings of the
985 LLVM system, and only need to be accessed in unusual circumstances.
989 <!-- ======================================================================= -->
990 <div class="doc_subsection">
991 <a name="TypeResolve">LLVM Type Resolution</a>
994 <div class="doc_text">
997 The LLVM type system has a very simple goal: allow clients to compare types for
998 structural equality with a simple pointer comparison (aka a shallow compare).
999 This goal makes clients much simpler and faster, and is used throughout the LLVM
1004 Unfortunately achieving this goal is not a simple matter. In particular,
1005 recursive types and late resolution of opaque types makes the situation very
1006 difficult to handle. Fortunately, for the most part, our implementation makes
1007 most clients able to be completely unaware of the nasty internal details. The
1008 primary case where clients are exposed to the inner workings of it are when
1009 building a recursive type. In addition to this case, the LLVM bytecode reader,
1010 assembly parser, and linker also have to be aware of the inner workings of this
1015 For our purposes below, we need three concepts. First, an "Opaque Type" is
1016 exactly as defined in the <a href="LangRef.html#t_opaque">language
1017 reference</a>. Second an "Abstract Type" is any type which includes an
1018 opaque type as part of its type graph (for example "<tt>{ opaque, int }</tt>").
1019 Third, a concrete type is a type that is not an abstract type (e.g. "<tt>[ int,
1025 <!-- ______________________________________________________________________ -->
1026 <div class="doc_subsubsection">
1027 <a name="BuildRecType">Basic Recursive Type Construction</a>
1030 <div class="doc_text">
1033 Because the most common question is "how do I build a recursive type with LLVM",
1034 we answer it now and explain it as we go. Here we include enough to cause this
1035 to be emitted to an output .ll file:
1039 %mylist = type { %mylist*, int }
1043 To build this, use the following LLVM APIs:
1047 //<i> Create the initial outer struct.</i>
1048 <a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
1049 std::vector<const Type*> Elts;
1050 Elts.push_back(PointerType::get(StructTy));
1051 Elts.push_back(Type::IntTy);
1052 StructType *NewSTy = StructType::get(Elts);
1054 //<i> At this point, NewSTy = "{ opaque*, int }". Tell VMCore that</i>
1055 //<i> the struct and the opaque type are actually the same.</i>
1056 cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
1058 // <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
1059 // <i>kept up-to-date.</i>
1060 NewSTy = cast<StructType>(StructTy.get());
1062 // <i>Add a name for the type to the module symbol table (optional).</i>
1063 MyModule->addTypeName("mylist", NewSTy);
1067 This code shows the basic approach used to build recursive types: build a
1068 non-recursive type using 'opaque', then use type unification to close the cycle.
1069 The type unification step is performed by the <tt><a
1070 ref="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
1071 described next. After that, we describe the <a
1072 href="#PATypeHolder">PATypeHolder class</a>.
1077 <!-- ______________________________________________________________________ -->
1078 <div class="doc_subsubsection">
1079 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
1082 <div class="doc_text">
1084 The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
1085 While this method is actually a member of the DerivedType class, it is most
1086 often used on OpaqueType instances. Type unification is actually a recursive
1087 process. After unification, types can become structurally isomorphic to
1088 existing types, and all duplicates are deleted (to preserve pointer equality).
1092 In the example above, the OpaqueType object is definitely deleted.
1093 Additionally, if there is an "{ \2*, int}" type already created in the system,
1094 the pointer and struct type created are <b>also</b> deleted. Obviously whenever
1095 a type is deleted, any "Type*" pointers in the program are invalidated. As
1096 such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
1097 live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
1098 types can never move or be deleted). To deal with this, the <a
1099 href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
1100 reference to a possibly refined type, and the <a
1101 href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
1102 complex datastructures.
1107 <!-- ______________________________________________________________________ -->
1108 <div class="doc_subsubsection">
1109 <a name="PATypeHolder">The PATypeHolder Class</a>
1112 <div class="doc_text">
1114 PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
1115 happily goes about nuking types that become isomorphic to existing types, it
1116 automatically updates all PATypeHolder objects to point to the new type. In the
1117 example above, this allows the code to maintain a pointer to the resultant
1118 resolved recursive type, even though the Type*'s are potentially invalidated.
1122 PATypeHolder is an extremely light-weight object that uses a lazy union-find
1123 implementation to update pointers. For example the pointer from a Value to its
1124 Type is maintained by PATypeHolder objects.
1129 <!-- ______________________________________________________________________ -->
1130 <div class="doc_subsubsection">
1131 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
1134 <div class="doc_text">
1137 Some data structures need more to perform more complex updates when types get
1138 resolved. The <a href="#SymbolTable">SymbolTable</a> class, for example, needs
1139 move and potentially merge type planes in its representation when a pointer
1143 To support this, a class can derive from the AbstractTypeUser class. This class
1144 allows it to get callbacks when certain types are resolved. To register to get
1145 callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
1146 methods can be called on a type. Note that these methods only work for <i>
1147 abstract</i> types. Concrete types (those that do not include an opaque objects
1148 somewhere) can never be refined.
1153 <!-- ======================================================================= -->
1154 <div class="doc_subsection">
1155 <a name="SymbolTable">The <tt>SymbolTable</tt> class</a>
1158 <div class="doc_text">
1159 <p>This class provides a symbol table that the <a
1160 href="#Function"><tt>Function</tt></a> and <a href="#Module">
1161 <tt>Module</tt></a> classes use for naming definitions. The symbol table can
1162 provide a name for any <a href="#Value"><tt>Value</tt></a> or <a
1163 href="#Type"><tt>Type</tt></a>. <tt>SymbolTable</tt> is an abstract data
1164 type. It hides the data it contains and provides access to it through a
1165 controlled interface.</p>
1167 <p>Note that the symbol table class is should not be directly accessed by most
1168 clients. It should only be used when iteration over the symbol table names
1169 themselves are required, which is very special purpose. Note that not all LLVM
1170 <a href="#Value">Value</a>s have names, and those without names (i.e. they have
1171 an empty name) do not exist in the symbol table.
1174 <p>To use the <tt>SymbolTable</tt> well, you need to understand the
1175 structure of the information it holds. The class contains two
1176 <tt>std::map</tt> objects. The first, <tt>pmap</tt>, is a map of
1177 <tt>Type*</tt> to maps of name (<tt>std::string</tt>) to <tt>Value*</tt>.
1178 The second, <tt>tmap</tt>, is a map of names to <tt>Type*</tt>. Thus, Values
1179 are stored in two-dimensions and accessed by <tt>Type</tt> and name. Types,
1180 however, are stored in a single dimension and accessed only by name.</p>
1182 <p>The interface of this class provides three basic types of operations:
1184 <li><em>Accessors</em>. Accessors provide read-only access to information
1185 such as finding a value for a name with the
1186 <a href="#SymbolTable_lookup">lookup</a> method.</li>
1187 <li><em>Mutators</em>. Mutators allow the user to add information to the
1188 <tt>SymbolTable</tt> with methods like
1189 <a href="#SymbolTable_insert"><tt>insert</tt></a>.</li>
1190 <li><em>Iterators</em>. Iterators allow the user to traverse the content
1191 of the symbol table in well defined ways, such as the method
1192 <a href="#SymbolTable_type_begin"><tt>type_begin</tt></a>.</li>
1197 <dt><tt>Value* lookup(const Type* Ty, const std::string& name) const</tt>:
1199 <dd>The <tt>lookup</tt> method searches the type plane given by the
1200 <tt>Ty</tt> parameter for a <tt>Value</tt> with the provided <tt>name</tt>.
1201 If a suitable <tt>Value</tt> is not found, null is returned.</dd>
1203 <dt><tt>Type* lookupType( const std::string& name) const</tt>:</dt>
1204 <dd>The <tt>lookupType</tt> method searches through the types for a
1205 <tt>Type</tt> with the provided <tt>name</tt>. If a suitable <tt>Type</tt>
1206 is not found, null is returned.</dd>
1208 <dt><tt>bool hasTypes() const</tt>:</dt>
1209 <dd>This function returns true if an entry has been made into the type
1212 <dt><tt>bool isEmpty() const</tt>:</dt>
1213 <dd>This function returns true if both the value and types maps are
1219 <dt><tt>void insert(Value *Val)</tt>:</dt>
1220 <dd>This method adds the provided value to the symbol table. The Value must
1221 have both a name and a type which are extracted and used to place the value
1222 in the correct type plane under the value's name.</dd>
1224 <dt><tt>void insert(const std::string& Name, Value *Val)</tt>:</dt>
1225 <dd> Inserts a constant or type into the symbol table with the specified
1226 name. There can be a many to one mapping between names and constants
1229 <dt><tt>void insert(const std::string& Name, Type *Typ)</tt>:</dt>
1230 <dd> Inserts a type into the symbol table with the specified name. There
1231 can be a many-to-one mapping between names and types. This method
1232 allows a type with an existing entry in the symbol table to get
1235 <dt><tt>void remove(Value* Val)</tt>:</dt>
1236 <dd> This method removes a named value from the symbol table. The
1237 type and name of the Value are extracted from \p N and used to
1238 lookup the Value in the correct type plane. If the Value is
1239 not in the symbol table, this method silently ignores the
1242 <dt><tt>void remove(Type* Typ)</tt>:</dt>
1243 <dd> This method removes a named type from the symbol table. The
1244 name of the type is extracted from \P T and used to look up
1245 the Type in the type map. If the Type is not in the symbol
1246 table, this method silently ignores the request.</dd>
1248 <dt><tt>Value* remove(const std::string& Name, Value *Val)</tt>:</dt>
1249 <dd> Remove a constant or type with the specified name from the
1252 <dt><tt>Type* remove(const std::string& Name, Type* T)</tt>:</dt>
1253 <dd> Remove a type with the specified name from the symbol table.
1254 Returns the removed Type.</dd>
1256 <dt><tt>Value *value_remove(const value_iterator& It)</tt>:</dt>
1257 <dd> Removes a specific value from the symbol table.
1258 Returns the removed value.</dd>
1260 <dt><tt>bool strip()</tt>:</dt>
1261 <dd> This method will strip the symbol table of its names leaving
1262 the type and values. </dd>
1264 <dt><tt>void clear()</tt>:</dt>
1265 <dd>Empty the symbol table completely.</dd>
1269 <p>The following functions describe three types of iterators you can obtain
1270 the beginning or end of the sequence for both const and non-const. It is
1271 important to keep track of the different kinds of iterators. There are
1272 three idioms worth pointing out:</p>
1274 <tr><th>Units</th><th>Iterator</th><th>Idiom</th></tr>
1276 <td align="left">Planes Of name/Value maps</td><td>PI</td>
1277 <td align="left"><pre><tt>
1278 for (SymbolTable::plane_const_iterator PI = ST.plane_begin(),
1279 PE = ST.plane_end(); PI != PE; ++PI ) {
1280 PI->first // This is the Type* of the plane
1281 PI->second // This is the SymbolTable::ValueMap of name/Value pairs
1285 <td align="left">All name/Type Pairs</td><td>TI</td>
1286 <td align="left"><pre><tt>
1287 for (SymbolTable::type_const_iterator TI = ST.type_begin(),
1288 TE = ST.type_end(); TI != TE; ++TI )
1289 TI->first // This is the name of the type
1290 TI->second // This is the Type* value associated with the name
1294 <td align="left">name/Value pairs in a plane</td><td>VI</td>
1295 <td align="left"><pre><tt>
1296 for (SymbolTable::value_const_iterator VI = ST.value_begin(SomeType),
1297 VE = ST.value_end(SomeType); VI != VE; ++VI )
1298 VI->first // This is the name of the Value
1299 VI->second // This is the Value* value associated with the name
1304 <p>Using the recommended iterator names and idioms will help you avoid
1305 making mistakes. Of particular note, make sure that whenever you use
1306 value_begin(SomeType) that you always compare the resulting iterator
1307 with value_end(SomeType) not value_end(SomeOtherType) or else you
1308 will loop infinitely.</p>
1312 <dt><tt>plane_iterator plane_begin()</tt>:</dt>
1313 <dd>Get an iterator that starts at the beginning of the type planes.
1314 The iterator will iterate over the Type/ValueMap pairs in the
1317 <dt><tt>plane_const_iterator plane_begin() const</tt>:</dt>
1318 <dd>Get a const_iterator that starts at the beginning of the type
1319 planes. The iterator will iterate over the Type/ValueMap pairs
1320 in the type planes. </dd>
1322 <dt><tt>plane_iterator plane_end()</tt>:</dt>
1323 <dd>Get an iterator at the end of the type planes. This serves as
1324 the marker for end of iteration over the type planes.</dd>
1326 <dt><tt>plane_const_iterator plane_end() const</tt>:</dt>
1327 <dd>Get a const_iterator at the end of the type planes. This serves as
1328 the marker for end of iteration over the type planes.</dd>
1330 <dt><tt>value_iterator value_begin(const Type *Typ)</tt>:</dt>
1331 <dd>Get an iterator that starts at the beginning of a type plane.
1332 The iterator will iterate over the name/value pairs in the type plane.
1333 Note: The type plane must already exist before using this.</dd>
1335 <dt><tt>value_const_iterator value_begin(const Type *Typ) const</tt>:</dt>
1336 <dd>Get a const_iterator that starts at the beginning of a type plane.
1337 The iterator will iterate over the name/value pairs in the type plane.
1338 Note: The type plane must already exist before using this.</dd>
1340 <dt><tt>value_iterator value_end(const Type *Typ)</tt>:</dt>
1341 <dd>Get an iterator to the end of a type plane. This serves as the marker
1342 for end of iteration of the type plane.
1343 Note: The type plane must already exist before using this.</dd>
1345 <dt><tt>value_const_iterator value_end(const Type *Typ) const</tt>:</dt>
1346 <dd>Get a const_iterator to the end of a type plane. This serves as the
1347 marker for end of iteration of the type plane.
1348 Note: the type plane must already exist before using this.</dd>
1350 <dt><tt>type_iterator type_begin()</tt>:</dt>
1351 <dd>Get an iterator to the start of the name/Type map.</dd>
1353 <dt><tt>type_const_iterator type_begin() cons</tt>:</dt>
1354 <dd> Get a const_iterator to the start of the name/Type map.</dd>
1356 <dt><tt>type_iterator type_end()</tt>:</dt>
1357 <dd>Get an iterator to the end of the name/Type map. This serves as the
1358 marker for end of iteration of the types.</dd>
1360 <dt><tt>type_const_iterator type_end() const</tt>:</dt>
1361 <dd>Get a const-iterator to the end of the name/Type map. This serves
1362 as the marker for end of iteration of the types.</dd>
1364 <dt><tt>plane_const_iterator find(const Type* Typ ) const</tt>:</dt>
1365 <dd>This method returns a plane_const_iterator for iteration over
1366 the type planes starting at a specific plane, given by \p Ty.</dd>
1368 <dt><tt>plane_iterator find( const Type* Typ </tt>:</dt>
1369 <dd>This method returns a plane_iterator for iteration over the
1370 type planes starting at a specific plane, given by \p Ty.</dd>
1377 <!-- *********************************************************************** -->
1378 <div class="doc_section">
1379 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
1381 <!-- *********************************************************************** -->
1383 <div class="doc_text">
1385 <p>The Core LLVM classes are the primary means of representing the program
1386 being inspected or transformed. The core LLVM classes are defined in
1387 header files in the <tt>include/llvm/</tt> directory, and implemented in
1388 the <tt>lib/VMCore</tt> directory.</p>
1392 <!-- ======================================================================= -->
1393 <div class="doc_subsection">
1394 <a name="Value">The <tt>Value</tt> class</a>
1399 <p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
1401 doxygen info: <a href="/doxygen/structllvm_1_1Value.html">Value Class</a></p>
1403 <p>The <tt>Value</tt> class is the most important class in the LLVM Source
1404 base. It represents a typed value that may be used (among other things) as an
1405 operand to an instruction. There are many different types of <tt>Value</tt>s,
1406 such as <a href="#Constant"><tt>Constant</tt></a>s,<a
1407 href="#Argument"><tt>Argument</tt></a>s. Even <a
1408 href="#Instruction"><tt>Instruction</tt></a>s and <a
1409 href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
1411 <p>A particular <tt>Value</tt> may be used many times in the LLVM representation
1412 for a program. For example, an incoming argument to a function (represented
1413 with an instance of the <a href="#Argument">Argument</a> class) is "used" by
1414 every instruction in the function that references the argument. To keep track
1415 of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
1416 href="#User"><tt>User</tt></a>s that is using it (the <a
1417 href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
1418 graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
1419 def-use information in the program, and is accessible through the <tt>use_</tt>*
1420 methods, shown below.</p>
1422 <p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
1423 and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
1424 method. In addition, all LLVM values can be named. The "name" of the
1425 <tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
1427 <pre> %<b>foo</b> = add int 1, 2<br></pre>
1429 <p><a name="#nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
1430 that the name of any value may be missing (an empty string), so names should
1431 <b>ONLY</b> be used for debugging (making the source code easier to read,
1432 debugging printouts), they should not be used to keep track of values or map
1433 between them. For this purpose, use a <tt>std::map</tt> of pointers to the
1434 <tt>Value</tt> itself instead.</p>
1436 <p>One important aspect of LLVM is that there is no distinction between an SSA
1437 variable and the operation that produces it. Because of this, any reference to
1438 the value produced by an instruction (or the value available as an incoming
1439 argument, for example) is represented as a direct pointer to the instance of
1441 represents this value. Although this may take some getting used to, it
1442 simplifies the representation and makes it easier to manipulate.</p>
1446 <!-- _______________________________________________________________________ -->
1447 <div class="doc_subsubsection">
1448 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
1451 <div class="doc_text">
1454 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
1456 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
1458 <tt>unsigned use_size()</tt> - Returns the number of users of the
1460 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
1461 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
1463 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
1465 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
1466 element in the list.
1467 <p> These methods are the interface to access the def-use
1468 information in LLVM. As with all other iterators in LLVM, the naming
1469 conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
1471 <li><tt><a href="#Type">Type</a> *getType() const</tt>
1472 <p>This method returns the Type of the Value.</p>
1474 <li><tt>bool hasName() const</tt><br>
1475 <tt>std::string getName() const</tt><br>
1476 <tt>void setName(const std::string &Name)</tt>
1477 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
1478 be aware of the <a href="#nameWarning">precaution above</a>.</p>
1480 <li><tt>void replaceAllUsesWith(Value *V)</tt>
1482 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
1483 href="#User"><tt>User</tt>s</a> of the current value to refer to
1484 "<tt>V</tt>" instead. For example, if you detect that an instruction always
1485 produces a constant value (for example through constant folding), you can
1486 replace all uses of the instruction with the constant like this:</p>
1488 <pre> Inst->replaceAllUsesWith(ConstVal);<br></pre>
1493 <!-- ======================================================================= -->
1494 <div class="doc_subsection">
1495 <a name="User">The <tt>User</tt> class</a>
1498 <div class="doc_text">
1501 <tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
1502 doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
1503 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1505 <p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
1506 refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
1507 that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
1508 referring to. The <tt>User</tt> class itself is a subclass of
1511 <p>The operands of a <tt>User</tt> point directly to the LLVM <a
1512 href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
1513 Single Assignment (SSA) form, there can only be one definition referred to,
1514 allowing this direct connection. This connection provides the use-def
1515 information in LLVM.</p>
1519 <!-- _______________________________________________________________________ -->
1520 <div class="doc_subsubsection">
1521 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
1524 <div class="doc_text">
1526 <p>The <tt>User</tt> class exposes the operand list in two ways: through
1527 an index access interface and through an iterator based interface.</p>
1530 <li><tt>Value *getOperand(unsigned i)</tt><br>
1531 <tt>unsigned getNumOperands()</tt>
1532 <p> These two methods expose the operands of the <tt>User</tt> in a
1533 convenient form for direct access.</p></li>
1535 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
1537 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
1538 the operand list.<br>
1539 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
1541 <p> Together, these methods make up the iterator based interface to
1542 the operands of a <tt>User</tt>.</p></li>
1547 <!-- ======================================================================= -->
1548 <div class="doc_subsection">
1549 <a name="Instruction">The <tt>Instruction</tt> class</a>
1552 <div class="doc_text">
1554 <p><tt>#include "</tt><tt><a
1555 href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
1556 doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
1557 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1558 href="#Value"><tt>Value</tt></a></p>
1560 <p>The <tt>Instruction</tt> class is the common base class for all LLVM
1561 instructions. It provides only a few methods, but is a very commonly used
1562 class. The primary data tracked by the <tt>Instruction</tt> class itself is the
1563 opcode (instruction type) and the parent <a
1564 href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
1565 into. To represent a specific type of instruction, one of many subclasses of
1566 <tt>Instruction</tt> are used.</p>
1568 <p> Because the <tt>Instruction</tt> class subclasses the <a
1569 href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
1570 way as for other <a href="#User"><tt>User</tt></a>s (with the
1571 <tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
1572 <tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
1573 the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
1574 file contains some meta-data about the various different types of instructions
1575 in LLVM. It describes the enum values that are used as opcodes (for example
1576 <tt>Instruction::Add</tt> and <tt>Instruction::SetLE</tt>), as well as the
1577 concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
1578 example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
1579 href="#SetCondInst">SetCondInst</a></tt>). Unfortunately, the use of macros in
1580 this file confuses doxygen, so these enum values don't show up correctly in the
1581 <a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
1585 <!-- _______________________________________________________________________ -->
1586 <div class="doc_subsubsection">
1587 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
1591 <div class="doc_text">
1594 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
1595 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
1596 this <tt>Instruction</tt> is embedded into.</p></li>
1597 <li><tt>bool mayWriteToMemory()</tt>
1598 <p>Returns true if the instruction writes to memory, i.e. it is a
1599 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
1600 <li><tt>unsigned getOpcode()</tt>
1601 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
1602 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
1603 <p>Returns another instance of the specified instruction, identical
1604 in all ways to the original except that the instruction has no parent
1605 (ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
1606 and it has no name</p></li>
1611 <!-- ======================================================================= -->
1612 <div class="doc_subsection">
1613 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
1616 <div class="doc_text">
1619 href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
1620 doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
1622 Superclass: <a href="#Value"><tt>Value</tt></a></p>
1624 <p>This class represents a single entry multiple exit section of the code,
1625 commonly known as a basic block by the compiler community. The
1626 <tt>BasicBlock</tt> class maintains a list of <a
1627 href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
1628 Matching the language definition, the last element of this list of instructions
1629 is always a terminator instruction (a subclass of the <a
1630 href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
1632 <p>In addition to tracking the list of instructions that make up the block, the
1633 <tt>BasicBlock</tt> class also keeps track of the <a
1634 href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
1636 <p>Note that <tt>BasicBlock</tt>s themselves are <a
1637 href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
1638 like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
1643 <!-- _______________________________________________________________________ -->
1644 <div class="doc_subsubsection">
1645 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
1649 <div class="doc_text">
1653 <li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
1654 href="#Function">Function</a> *Parent = 0)</tt>
1656 <p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
1657 insertion into a function. The constructor optionally takes a name for the new
1658 block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
1659 the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
1660 automatically inserted at the end of the specified <a
1661 href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
1662 manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
1664 <li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
1665 <tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
1666 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
1667 <tt>size()</tt>, <tt>empty()</tt>
1668 STL-style functions for accessing the instruction list.
1670 <p>These methods and typedefs are forwarding functions that have the same
1671 semantics as the standard library methods of the same names. These methods
1672 expose the underlying instruction list of a basic block in a way that is easy to
1673 manipulate. To get the full complement of container operations (including
1674 operations to update the list), you must use the <tt>getInstList()</tt>
1677 <li><tt>BasicBlock::InstListType &getInstList()</tt>
1679 <p>This method is used to get access to the underlying container that actually
1680 holds the Instructions. This method must be used when there isn't a forwarding
1681 function in the <tt>BasicBlock</tt> class for the operation that you would like
1682 to perform. Because there are no forwarding functions for "updating"
1683 operations, you need to use this if you want to update the contents of a
1684 <tt>BasicBlock</tt>.</p></li>
1686 <li><tt><a href="#Function">Function</a> *getParent()</tt>
1688 <p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
1689 embedded into, or a null pointer if it is homeless.</p></li>
1691 <li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
1693 <p> Returns a pointer to the terminator instruction that appears at the end of
1694 the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
1695 instruction in the block is not a terminator, then a null pointer is
1702 <!-- ======================================================================= -->
1703 <div class="doc_subsection">
1704 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
1707 <div class="doc_text">
1710 href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
1711 doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
1713 Superclasses: <a href="#User"><tt>User</tt></a>, <a
1714 href="#Value"><tt>Value</tt></a></p>
1716 <p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
1717 href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
1718 visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
1719 Because they are visible at global scope, they are also subject to linking with
1720 other globals defined in different translation units. To control the linking
1721 process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
1722 <tt>GlobalValue</tt>s know whether they have internal or external linkage, as
1723 defined by the <tt>LinkageTypes</tt> enumeration.</p>
1725 <p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
1726 <tt>static</tt> in C), it is not visible to code outside the current translation
1727 unit, and does not participate in linking. If it has external linkage, it is
1728 visible to external code, and does participate in linking. In addition to
1729 linkage information, <tt>GlobalValue</tt>s keep track of which <a
1730 href="#Module"><tt>Module</tt></a> they are currently part of.</p>
1732 <p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
1733 by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
1734 global is always a pointer to its contents. It is important to remember this
1735 when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
1736 be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
1737 subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
1738 int]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
1739 the address of the first element of this array and the value of the
1740 <tt>GlobalVariable</tt> are the same, they have different types. The
1741 <tt>GlobalVariable</tt>'s type is <tt>[24 x int]</tt>. The first element's type
1742 is <tt>int.</tt> Because of this, accessing a global value requires you to
1743 dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
1744 can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
1745 Language Reference Manual</a>.</p>
1749 <!-- _______________________________________________________________________ -->
1750 <div class="doc_subsubsection">
1751 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
1755 <div class="doc_text">
1758 <li><tt>bool hasInternalLinkage() const</tt><br>
1759 <tt>bool hasExternalLinkage() const</tt><br>
1760 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
1761 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
1764 <li><tt><a href="#Module">Module</a> *getParent()</tt>
1765 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
1766 GlobalValue is currently embedded into.</p></li>
1771 <!-- ======================================================================= -->
1772 <div class="doc_subsection">
1773 <a name="Function">The <tt>Function</tt> class</a>
1776 <div class="doc_text">
1779 href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
1780 info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
1781 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1782 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1784 <p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
1785 actually one of the more complex classes in the LLVM heirarchy because it must
1786 keep track of a large amount of data. The <tt>Function</tt> class keeps track
1787 of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal <a
1788 href="#Argument"><tt>Argument</tt></a>s, and a <a
1789 href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
1791 <p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
1792 commonly used part of <tt>Function</tt> objects. The list imposes an implicit
1793 ordering of the blocks in the function, which indicate how the code will be
1794 layed out by the backend. Additionally, the first <a
1795 href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
1796 <tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
1797 block. There are no implicit exit nodes, and in fact there may be multiple exit
1798 nodes from a single <tt>Function</tt>. If the <a
1799 href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
1800 the <tt>Function</tt> is actually a function declaration: the actual body of the
1801 function hasn't been linked in yet.</p>
1803 <p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
1804 <tt>Function</tt> class also keeps track of the list of formal <a
1805 href="#Argument"><tt>Argument</tt></a>s that the function receives. This
1806 container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
1807 nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
1808 the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
1810 <p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
1811 LLVM feature that is only used when you have to look up a value by name. Aside
1812 from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
1813 internally to make sure that there are not conflicts between the names of <a
1814 href="#Instruction"><tt>Instruction</tt></a>s, <a
1815 href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
1816 href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
1818 <p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
1819 and therefore also a <a href="#Constant">Constant</a>. The value of the function
1820 is its address (after linking) which is guaranteed to be constant.</p>
1823 <!-- _______________________________________________________________________ -->
1824 <div class="doc_subsubsection">
1825 <a name="m_Function">Important Public Members of the <tt>Function</tt>
1829 <div class="doc_text">
1832 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
1833 *Ty, LinkageTypes Linkage, const std::string &N = "", Module* Parent = 0)</tt>
1835 <p>Constructor used when you need to create new <tt>Function</tt>s to add
1836 the the program. The constructor must specify the type of the function to
1837 create and what type of linkage the function should have. The <a
1838 href="#FunctionType"><tt>FunctionType</tt></a> argument
1839 specifies the formal arguments and return value for the function. The same
1840 <a href="#FunctionTypel"><tt>FunctionType</tt></a> value can be used to
1841 create multiple functions. The <tt>Parent</tt> argument specifies the Module
1842 in which the function is defined. If this argument is provided, the function
1843 will automatically be inserted into that module's list of
1846 <li><tt>bool isExternal()</tt>
1848 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
1849 function is "external", it does not have a body, and thus must be resolved
1850 by linking with a function defined in a different translation unit.</p></li>
1852 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
1853 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
1855 <tt>begin()</tt>, <tt>end()</tt>
1856 <tt>size()</tt>, <tt>empty()</tt>
1858 <p>These are forwarding methods that make it easy to access the contents of
1859 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
1862 <li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
1864 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
1865 is necessary to use when you need to update the list or perform a complex
1866 action that doesn't have a forwarding method.</p></li>
1868 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
1870 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
1872 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
1873 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
1875 <p>These are forwarding methods that make it easy to access the contents of
1876 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
1879 <li><tt>Function::ArgumentListType &getArgumentList()</tt>
1881 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
1882 necessary to use when you need to update the list or perform a complex
1883 action that doesn't have a forwarding method.</p></li>
1885 <li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
1887 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
1888 function. Because the entry block for the function is always the first
1889 block, this returns the first block of the <tt>Function</tt>.</p></li>
1891 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
1892 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
1894 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
1895 <tt>Function</tt> and returns the return type of the function, or the <a
1896 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
1899 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
1901 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
1902 for this <tt>Function</tt>.</p></li>
1907 <!-- ======================================================================= -->
1908 <div class="doc_subsection">
1909 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
1912 <div class="doc_text">
1915 href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
1917 doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
1918 Class</a><br> Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>, <a
1919 href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
1921 <p>Global variables are represented with the (suprise suprise)
1922 <tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
1923 subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
1924 always referenced by their address (global values must live in memory, so their
1925 "name" refers to their address). See <a
1926 href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global variables
1927 may have an initial value (which must be a <a
1928 href="#Constant"><tt>Constant</tt></a>), and if they have an initializer, they
1929 may be marked as "constant" themselves (indicating that their contents never
1930 change at runtime).</p>
1934 <!-- _______________________________________________________________________ -->
1935 <div class="doc_subsubsection">
1936 <a name="m_GlobalVariable">Important Public Members of the
1937 <tt>GlobalVariable</tt> class</a>
1940 <div class="doc_text">
1943 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
1944 isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
1945 *Initializer = 0, const std::string &Name = "", Module* Parent = 0)</tt>
1947 <p>Create a new global variable of the specified type. If
1948 <tt>isConstant</tt> is true then the global variable will be marked as
1949 unchanging for the program. The Linkage parameter specifies the type of
1950 linkage (internal, external, weak, linkonce, appending) for the variable. If
1951 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage, then
1952 the resultant global variable will have internal linkage. AppendingLinkage
1953 concatenates together all instances (in different translation units) of the
1954 variable into a single variable but is only applicable to arrays. See
1955 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
1956 further details on linkage types. Optionally an initializer, a name, and the
1957 module to put the variable into may be specified for the global variable as
1960 <li><tt>bool isConstant() const</tt>
1962 <p>Returns true if this is a global variable that is known not to
1963 be modified at runtime.</p></li>
1965 <li><tt>bool hasInitializer()</tt>
1967 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
1969 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
1971 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
1972 to call this method if there is no initializer.</p></li>
1977 <!-- ======================================================================= -->
1978 <div class="doc_subsection">
1979 <a name="Module">The <tt>Module</tt> class</a>
1982 <div class="doc_text">
1985 href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
1986 <a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
1988 <p>The <tt>Module</tt> class represents the top level structure present in LLVM
1989 programs. An LLVM module is effectively either a translation unit of the
1990 original program or a combination of several translation units merged by the
1991 linker. The <tt>Module</tt> class keeps track of a list of <a
1992 href="#Function"><tt>Function</tt></a>s, a list of <a
1993 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
1994 href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
1995 helpful member functions that try to make common operations easy.</p>
1999 <!-- _______________________________________________________________________ -->
2000 <div class="doc_subsubsection">
2001 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
2004 <div class="doc_text">
2007 <li><tt>Module::Module(std::string name = "")</tt></li>
2010 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
2011 provide a name for it (probably based on the name of the translation unit).</p>
2014 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
2015 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
2017 <tt>begin()</tt>, <tt>end()</tt>
2018 <tt>size()</tt>, <tt>empty()</tt>
2020 <p>These are forwarding methods that make it easy to access the contents of
2021 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
2024 <li><tt>Module::FunctionListType &getFunctionList()</tt>
2026 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
2027 necessary to use when you need to update the list or perform a complex
2028 action that doesn't have a forwarding method.</p>
2030 <p><!-- Global Variable --></p></li>
2036 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
2038 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
2040 <tt>global_begin()</tt>, <tt>global_end()</tt>
2041 <tt>global_size()</tt>, <tt>global_empty()</tt>
2043 <p> These are forwarding methods that make it easy to access the contents of
2044 a <tt>Module</tt> object's <a
2045 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
2047 <li><tt>Module::GlobalListType &getGlobalList()</tt>
2049 <p>Returns the list of <a
2050 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
2051 use when you need to update the list or perform a complex action that
2052 doesn't have a forwarding method.</p>
2054 <p><!-- Symbol table stuff --> </p></li>
2060 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2062 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2063 for this <tt>Module</tt>.</p>
2065 <p><!-- Convenience methods --></p></li>
2071 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
2072 &Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
2074 <p>Look up the specified function in the <tt>Module</tt> <a
2075 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
2076 <tt>null</tt>.</p></li>
2078 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
2079 std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
2081 <p>Look up the specified function in the <tt>Module</tt> <a
2082 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
2083 external declaration for the function and return it.</p></li>
2085 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
2087 <p>If there is at least one entry in the <a
2088 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
2089 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
2092 <li><tt>bool addTypeName(const std::string &Name, const <a
2093 href="#Type">Type</a> *Ty)</tt>
2095 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2096 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
2097 name, true is returned and the <a
2098 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
2103 <!-- ======================================================================= -->
2104 <div class="doc_subsection">
2105 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
2108 <div class="doc_text">
2110 <p>Constant represents a base class for different types of constants. It
2111 is subclassed by ConstantBool, ConstantInt, ConstantSInt, ConstantUInt,
2112 ConstantArray etc for representing the various types of Constants.</p>
2116 <!-- _______________________________________________________________________ -->
2117 <div class="doc_subsubsection">
2118 <a name="m_Constant">Important Public Methods</a>
2120 <div class="doc_text">
2123 <!-- _______________________________________________________________________ -->
2124 <div class="doc_subsubsection">Important Subclasses of Constant </div>
2125 <div class="doc_text">
2127 <li>ConstantSInt : This subclass of Constant represents a signed integer
2130 <li><tt>int64_t getValue() const</tt>: Returns the underlying value of
2131 this constant. </li>
2134 <li>ConstantUInt : This class represents an unsigned integer.
2136 <li><tt>uint64_t getValue() const</tt>: Returns the underlying value of
2137 this constant. </li>
2140 <li>ConstantFP : This class represents a floating point constant.
2142 <li><tt>double getValue() const</tt>: Returns the underlying value of
2143 this constant. </li>
2146 <li>ConstantBool : This represents a boolean constant.
2148 <li><tt>bool getValue() const</tt>: Returns the underlying value of this
2152 <li>ConstantArray : This represents a constant array.
2154 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
2155 a vector of component constants that makeup this array. </li>
2158 <li>ConstantStruct : This represents a constant struct.
2160 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
2161 a vector of component constants that makeup this array. </li>
2164 <li>GlobalValue : This represents either a global variable or a function. In
2165 either case, the value is a constant fixed address (after linking).
2170 <!-- ======================================================================= -->
2171 <div class="doc_subsection">
2172 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
2175 <div class="doc_text">
2177 <p>Type as noted earlier is also a subclass of a Value class. Any primitive
2178 type (like int, short etc) in LLVM is an instance of Type Class. All other
2179 types are instances of subclasses of type like FunctionType, ArrayType
2180 etc. DerivedType is the interface for all such dervied types including
2181 FunctionType, ArrayType, PointerType, StructType. Types can have names. They can
2182 be recursive (StructType). There exists exactly one instance of any type
2183 structure at a time. This allows using pointer equality of Type *s for comparing
2188 <!-- _______________________________________________________________________ -->
2189 <div class="doc_subsubsection">
2190 <a name="m_Value">Important Public Methods</a>
2193 <div class="doc_text">
2197 <li><tt>bool isSigned() const</tt>: Returns whether an integral numeric type
2198 is signed. This is true for SByteTy, ShortTy, IntTy, LongTy. Note that this is
2199 not true for Float and Double. </li>
2201 <li><tt>bool isUnsigned() const</tt>: Returns whether a numeric type is
2202 unsigned. This is not quite the complement of isSigned... nonnumeric types
2203 return false as they do with isSigned. This returns true for UByteTy,
2204 UShortTy, UIntTy, and ULongTy. </li>
2206 <li><tt>bool isInteger() const</tt>: Equivalent to isSigned() || isUnsigned().</li>
2208 <li><tt>bool isIntegral() const</tt>: Returns true if this is an integral
2209 type, which is either Bool type or one of the Integer types.</li>
2211 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
2212 floating point types.</li>
2214 <li><tt>isLosslesslyConvertableTo (const Type *Ty) const</tt>: Return true if
2215 this type can be converted to 'Ty' without any reinterpretation of bits. For
2216 example, uint to int or one pointer type to another.</li>
2220 <!-- _______________________________________________________________________ -->
2221 <div class="doc_subsubsection">
2222 <a name="m_Value">Important Derived Types</a>
2224 <div class="doc_text">
2226 <li>SequentialType : This is subclassed by ArrayType and PointerType
2228 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
2229 of the elements in the sequential type. </li>
2232 <li>ArrayType : This is a subclass of SequentialType and defines interface for
2235 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
2236 elements in the array. </li>
2239 <li>PointerType : Subclass of SequentialType for pointer types. </li>
2240 <li>StructType : subclass of DerivedTypes for struct types </li>
2241 <li>FunctionType : subclass of DerivedTypes for function types.
2243 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
2245 <li><tt> const Type * getReturnType() const</tt>: Returns the
2246 return type of the function.</li>
2247 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
2248 the type of the ith parameter.</li>
2249 <li><tt> const unsigned getNumParams() const</tt>: Returns the
2250 number of formal parameters.</li>
2256 <!-- ======================================================================= -->
2257 <div class="doc_subsection">
2258 <a name="Argument">The <tt>Argument</tt> class</a>
2261 <div class="doc_text">
2263 <p>This subclass of Value defines the interface for incoming formal
2264 arguments to a function. A Function maintains a list of its formal
2265 arguments. An argument has a pointer to the parent Function.</p>
2269 <!-- *********************************************************************** -->
2272 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2273 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
2274 <a href="http://validator.w3.org/check/referer"><img
2275 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
2277 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
2278 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
2279 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
2280 Last modified: $Date$