1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <meta http-equiv="Content-type" content="text/html;charset=UTF-8">
6 <title>LLVM Programmer's Manual</title>
7 <link rel="stylesheet" href="llvm.css" type="text/css">
11 <div class="doc_title">
12 LLVM Programmer's Manual
16 <li><a href="#introduction">Introduction</a></li>
17 <li><a href="#general">General Information</a>
19 <li><a href="#stl">The C++ Standard Template Library</a></li>
21 <li>The <tt>-time-passes</tt> option</li>
22 <li>How to use the LLVM Makefile system</li>
23 <li>How to write a regression test</li>
28 <li><a href="#apis">Important and useful LLVM APIs</a>
30 <li><a href="#isa">The <tt>isa<></tt>, <tt>cast<></tt>
31 and <tt>dyn_cast<></tt> templates</a> </li>
32 <li><a href="#string_apis">Passing strings (the <tt>StringRef</tt>
33 and <tt>Twine</tt> classes)</a>
35 <li><a href="#StringRef">The <tt>StringRef</tt> class</a> </li>
36 <li><a href="#Twine">The <tt>Twine</tt> class</a> </li>
39 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
42 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
43 and the <tt>-debug-only</tt> option</a> </li>
46 <li><a href="#Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
49 <li>The <tt>InstVisitor</tt> template
50 <li>The general graph API
52 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
55 <li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
57 <li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
59 <li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
60 <li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
61 <li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
62 <li><a href="#dss_vector"><vector></a></li>
63 <li><a href="#dss_deque"><deque></a></li>
64 <li><a href="#dss_list"><list></a></li>
65 <li><a href="#dss_ilist">llvm/ADT/ilist.h</a></li>
66 <li><a href="#dss_other">Other Sequential Container Options</a></li>
68 <li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
70 <li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
71 <li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
72 <li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
73 <li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
74 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
75 <li><a href="#dss_set"><set></a></li>
76 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
77 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
78 <li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
80 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
82 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
83 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
84 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
85 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
86 <li><a href="#dss_valuemap">"llvm/ADT/ValueMap.h"</a></li>
87 <li><a href="#dss_intervalmap">"llvm/ADT/IntervalMap.h"</a></li>
88 <li><a href="#dss_map"><map></a></li>
89 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
91 <li><a href="#ds_string">String-like containers</a>
95 <li><a href="#ds_bit">BitVector-like containers</a>
97 <li><a href="#dss_bitvector">A dense bitvector</a></li>
98 <li><a href="#dss_smallbitvector">A "small" dense bitvector</a></li>
99 <li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
103 <li><a href="#common">Helpful Hints for Common Operations</a>
105 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
107 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
108 in a <tt>Function</tt></a> </li>
109 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
110 in a <tt>BasicBlock</tt></a> </li>
111 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
112 in a <tt>Function</tt></a> </li>
113 <li><a href="#iterate_convert">Turning an iterator into a
114 class pointer</a> </li>
115 <li><a href="#iterate_complex">Finding call sites: a more
116 complex example</a> </li>
117 <li><a href="#calls_and_invokes">Treating calls and invokes
118 the same way</a> </li>
119 <li><a href="#iterate_chains">Iterating over def-use &
120 use-def chains</a> </li>
121 <li><a href="#iterate_preds">Iterating over predecessors &
122 successors of blocks</a></li>
125 <li><a href="#simplechanges">Making simple changes</a>
127 <li><a href="#schanges_creating">Creating and inserting new
128 <tt>Instruction</tt>s</a> </li>
129 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
130 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
131 with another <tt>Value</tt></a> </li>
132 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
135 <li><a href="#create_types">How to Create Types</a></li>
137 <li>Working with the Control Flow Graph
139 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
147 <li><a href="#threading">Threads and LLVM</a>
149 <li><a href="#startmultithreaded">Entering and Exiting Multithreaded Mode
151 <li><a href="#shutdown">Ending execution with <tt>llvm_shutdown()</tt></a></li>
152 <li><a href="#managedstatic">Lazy initialization with <tt>ManagedStatic</tt></a></li>
153 <li><a href="#llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a></li>
154 <li><a href="#jitthreading">Threads and the JIT</a></li>
158 <li><a href="#advanced">Advanced Topics</a>
160 <li><a href="#TypeResolve">LLVM Type Resolution</a>
162 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
163 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
164 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
165 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
168 <li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes</a></li>
169 <li><a href="#UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a></li>
172 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
174 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
175 <li><a href="#Module">The <tt>Module</tt> class</a></li>
176 <li><a href="#Value">The <tt>Value</tt> class</a>
178 <li><a href="#User">The <tt>User</tt> class</a>
180 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
181 <li><a href="#Constant">The <tt>Constant</tt> class</a>
183 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
185 <li><a href="#Function">The <tt>Function</tt> class</a></li>
186 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
193 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
194 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
201 <div class="doc_author">
202 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
203 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
204 <a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
205 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>,
206 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a> and
207 <a href="mailto:owen@apple.com">Owen Anderson</a></p>
210 <!-- *********************************************************************** -->
211 <div class="doc_section">
212 <a name="introduction">Introduction </a>
214 <!-- *********************************************************************** -->
216 <div class="doc_text">
218 <p>This document is meant to highlight some of the important classes and
219 interfaces available in the LLVM source-base. This manual is not
220 intended to explain what LLVM is, how it works, and what LLVM code looks
221 like. It assumes that you know the basics of LLVM and are interested
222 in writing transformations or otherwise analyzing or manipulating the
225 <p>This document should get you oriented so that you can find your
226 way in the continuously growing source code that makes up the LLVM
227 infrastructure. Note that this manual is not intended to serve as a
228 replacement for reading the source code, so if you think there should be
229 a method in one of these classes to do something, but it's not listed,
230 check the source. Links to the <a href="/doxygen/">doxygen</a> sources
231 are provided to make this as easy as possible.</p>
233 <p>The first section of this document describes general information that is
234 useful to know when working in the LLVM infrastructure, and the second describes
235 the Core LLVM classes. In the future this manual will be extended with
236 information describing how to use extension libraries, such as dominator
237 information, CFG traversal routines, and useful utilities like the <tt><a
238 href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
242 <!-- *********************************************************************** -->
243 <div class="doc_section">
244 <a name="general">General Information</a>
246 <!-- *********************************************************************** -->
248 <div class="doc_text">
250 <p>This section contains general information that is useful if you are working
251 in the LLVM source-base, but that isn't specific to any particular API.</p>
255 <!-- ======================================================================= -->
256 <div class="doc_subsection">
257 <a name="stl">The C++ Standard Template Library</a>
260 <div class="doc_text">
262 <p>LLVM makes heavy use of the C++ Standard Template Library (STL),
263 perhaps much more than you are used to, or have seen before. Because of
264 this, you might want to do a little background reading in the
265 techniques used and capabilities of the library. There are many good
266 pages that discuss the STL, and several books on the subject that you
267 can get, so it will not be discussed in this document.</p>
269 <p>Here are some useful links:</p>
273 <li><a href="http://www.dinkumware.com/manuals/#Standard C++ Library">Dinkumware
274 C++ Library reference</a> - an excellent reference for the STL and other parts
275 of the standard C++ library.</li>
277 <li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
278 O'Reilly book in the making. It has a decent Standard Library
279 Reference that rivals Dinkumware's, and is unfortunately no longer free since the
280 book has been published.</li>
282 <li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
285 <li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
287 href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
290 <li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
293 <li><a href="http://64.78.49.204/">
294 Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
299 <p>You are also encouraged to take a look at the <a
300 href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
301 to write maintainable code more than where to put your curly braces.</p>
305 <!-- ======================================================================= -->
306 <div class="doc_subsection">
307 <a name="stl">Other useful references</a>
310 <div class="doc_text">
313 <li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
314 static and shared libraries across platforms</a></li>
319 <!-- *********************************************************************** -->
320 <div class="doc_section">
321 <a name="apis">Important and useful LLVM APIs</a>
323 <!-- *********************************************************************** -->
325 <div class="doc_text">
327 <p>Here we highlight some LLVM APIs that are generally useful and good to
328 know about when writing transformations.</p>
332 <!-- ======================================================================= -->
333 <div class="doc_subsection">
334 <a name="isa">The <tt>isa<></tt>, <tt>cast<></tt> and
335 <tt>dyn_cast<></tt> templates</a>
338 <div class="doc_text">
340 <p>The LLVM source-base makes extensive use of a custom form of RTTI.
341 These templates have many similarities to the C++ <tt>dynamic_cast<></tt>
342 operator, but they don't have some drawbacks (primarily stemming from
343 the fact that <tt>dynamic_cast<></tt> only works on classes that
344 have a v-table). Because they are used so often, you must know what they
345 do and how they work. All of these templates are defined in the <a
346 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
347 file (note that you very rarely have to include this file directly).</p>
350 <dt><tt>isa<></tt>: </dt>
352 <dd><p>The <tt>isa<></tt> operator works exactly like the Java
353 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
354 a reference or pointer points to an instance of the specified class. This can
355 be very useful for constraint checking of various sorts (example below).</p>
358 <dt><tt>cast<></tt>: </dt>
360 <dd><p>The <tt>cast<></tt> operator is a "checked cast" operation. It
361 converts a pointer or reference from a base class to a derived class, causing
362 an assertion failure if it is not really an instance of the right type. This
363 should be used in cases where you have some information that makes you believe
364 that something is of the right type. An example of the <tt>isa<></tt>
365 and <tt>cast<></tt> template is:</p>
367 <div class="doc_code">
369 static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
370 if (isa<<a href="#Constant">Constant</a>>(V) || isa<<a href="#Argument">Argument</a>>(V) || isa<<a href="#GlobalValue">GlobalValue</a>>(V))
373 // <i>Otherwise, it must be an instruction...</i>
374 return !L->contains(cast<<a href="#Instruction">Instruction</a>>(V)->getParent());
379 <p>Note that you should <b>not</b> use an <tt>isa<></tt> test followed
380 by a <tt>cast<></tt>, for that use the <tt>dyn_cast<></tt>
385 <dt><tt>dyn_cast<></tt>:</dt>
387 <dd><p>The <tt>dyn_cast<></tt> operator is a "checking cast" operation.
388 It checks to see if the operand is of the specified type, and if so, returns a
389 pointer to it (this operator does not work with references). If the operand is
390 not of the correct type, a null pointer is returned. Thus, this works very
391 much like the <tt>dynamic_cast<></tt> operator in C++, and should be
392 used in the same circumstances. Typically, the <tt>dyn_cast<></tt>
393 operator is used in an <tt>if</tt> statement or some other flow control
394 statement like this:</p>
396 <div class="doc_code">
398 if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast<<a href="#AllocationInst">AllocationInst</a>>(Val)) {
404 <p>This form of the <tt>if</tt> statement effectively combines together a call
405 to <tt>isa<></tt> and a call to <tt>cast<></tt> into one
406 statement, which is very convenient.</p>
408 <p>Note that the <tt>dyn_cast<></tt> operator, like C++'s
409 <tt>dynamic_cast<></tt> or Java's <tt>instanceof</tt> operator, can be
410 abused. In particular, you should not use big chained <tt>if/then/else</tt>
411 blocks to check for lots of different variants of classes. If you find
412 yourself wanting to do this, it is much cleaner and more efficient to use the
413 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
417 <dt><tt>cast_or_null<></tt>: </dt>
419 <dd><p>The <tt>cast_or_null<></tt> operator works just like the
420 <tt>cast<></tt> operator, except that it allows for a null pointer as an
421 argument (which it then propagates). This can sometimes be useful, allowing
422 you to combine several null checks into one.</p></dd>
424 <dt><tt>dyn_cast_or_null<></tt>: </dt>
426 <dd><p>The <tt>dyn_cast_or_null<></tt> operator works just like the
427 <tt>dyn_cast<></tt> operator, except that it allows for a null pointer
428 as an argument (which it then propagates). This can sometimes be useful,
429 allowing you to combine several null checks into one.</p></dd>
433 <p>These five templates can be used with any classes, whether they have a
434 v-table or not. To add support for these templates, you simply need to add
435 <tt>classof</tt> static methods to the class you are interested casting
436 to. Describing this is currently outside the scope of this document, but there
437 are lots of examples in the LLVM source base.</p>
442 <!-- ======================================================================= -->
443 <div class="doc_subsection">
444 <a name="string_apis">Passing strings (the <tt>StringRef</tt>
445 and <tt>Twine</tt> classes)</a>
448 <div class="doc_text">
450 <p>Although LLVM generally does not do much string manipulation, we do have
451 several important APIs which take strings. Two important examples are the
452 Value class -- which has names for instructions, functions, etc. -- and the
453 StringMap class which is used extensively in LLVM and Clang.</p>
455 <p>These are generic classes, and they need to be able to accept strings which
456 may have embedded null characters. Therefore, they cannot simply take
457 a <tt>const char *</tt>, and taking a <tt>const std::string&</tt> requires
458 clients to perform a heap allocation which is usually unnecessary. Instead,
459 many LLVM APIs use a <tt>StringRef</tt> or a <tt>const Twine&</tt> for
460 passing strings efficiently.</p>
464 <!-- _______________________________________________________________________ -->
465 <div class="doc_subsubsection">
466 <a name="StringRef">The <tt>StringRef</tt> class</a>
469 <div class="doc_text">
471 <p>The <tt>StringRef</tt> data type represents a reference to a constant string
472 (a character array and a length) and supports the common operations available
473 on <tt>std:string</tt>, but does not require heap allocation.</p>
475 <p>It can be implicitly constructed using a C style null-terminated string,
476 an <tt>std::string</tt>, or explicitly with a character pointer and length.
477 For example, the <tt>StringRef</tt> find function is declared as:</p>
479 <pre class="doc_code">
480 iterator find(StringRef Key);
483 <p>and clients can call it using any one of:</p>
485 <pre class="doc_code">
486 Map.find("foo"); <i>// Lookup "foo"</i>
487 Map.find(std::string("bar")); <i>// Lookup "bar"</i>
488 Map.find(StringRef("\0baz", 4)); <i>// Lookup "\0baz"</i>
491 <p>Similarly, APIs which need to return a string may return a <tt>StringRef</tt>
492 instance, which can be used directly or converted to an <tt>std::string</tt>
493 using the <tt>str</tt> member function. See
494 "<tt><a href="/doxygen/classllvm_1_1StringRef_8h-source.html">llvm/ADT/StringRef.h</a></tt>"
495 for more information.</p>
497 <p>You should rarely use the <tt>StringRef</tt> class directly, because it contains
498 pointers to external memory it is not generally safe to store an instance of the
499 class (unless you know that the external storage will not be freed). StringRef is
500 small and pervasive enough in LLVM that it should always be passed by value.</p>
504 <!-- _______________________________________________________________________ -->
505 <div class="doc_subsubsection">
506 <a name="Twine">The <tt>Twine</tt> class</a>
509 <div class="doc_text">
511 <p>The <tt>Twine</tt> class is an efficient way for APIs to accept concatenated
512 strings. For example, a common LLVM paradigm is to name one instruction based on
513 the name of another instruction with a suffix, for example:</p>
515 <div class="doc_code">
517 New = CmpInst::Create(<i>...</i>, SO->getName() + ".cmp");
521 <p>The <tt>Twine</tt> class is effectively a
522 lightweight <a href="http://en.wikipedia.org/wiki/Rope_(computer_science)">rope</a>
523 which points to temporary (stack allocated) objects. Twines can be implicitly
524 constructed as the result of the plus operator applied to strings (i.e., a C
525 strings, an <tt>std::string</tt>, or a <tt>StringRef</tt>). The twine delays the
526 actual concatenation of strings until it is actually required, at which point
527 it can be efficiently rendered directly into a character array. This avoids
528 unnecessary heap allocation involved in constructing the temporary results of
529 string concatenation. See
530 "<tt><a href="/doxygen/classllvm_1_1Twine_8h-source.html">llvm/ADT/Twine.h</a></tt>"
531 for more information.</p>
533 <p>As with a <tt>StringRef</tt>, <tt>Twine</tt> objects point to external memory
534 and should almost never be stored or mentioned directly. They are intended
535 solely for use when defining a function which should be able to efficiently
536 accept concatenated strings.</p>
541 <!-- ======================================================================= -->
542 <div class="doc_subsection">
543 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
546 <div class="doc_text">
548 <p>Often when working on your pass you will put a bunch of debugging printouts
549 and other code into your pass. After you get it working, you want to remove
550 it, but you may need it again in the future (to work out new bugs that you run
553 <p> Naturally, because of this, you don't want to delete the debug printouts,
554 but you don't want them to always be noisy. A standard compromise is to comment
555 them out, allowing you to enable them if you need them in the future.</p>
557 <p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
558 file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
559 this problem. Basically, you can put arbitrary code into the argument of the
560 <tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
561 tool) is run with the '<tt>-debug</tt>' command line argument:</p>
563 <div class="doc_code">
565 DEBUG(errs() << "I am here!\n");
569 <p>Then you can run your pass like this:</p>
571 <div class="doc_code">
573 $ opt < a.bc > /dev/null -mypass
574 <i><no output></i>
575 $ opt < a.bc > /dev/null -mypass -debug
580 <p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
581 to not have to create "yet another" command line option for the debug output for
582 your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
583 so they do not cause a performance impact at all (for the same reason, they
584 should also not contain side-effects!).</p>
586 <p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
587 enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
588 "<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
589 program hasn't been started yet, you can always just run it with
594 <!-- _______________________________________________________________________ -->
595 <div class="doc_subsubsection">
596 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
597 the <tt>-debug-only</tt> option</a>
600 <div class="doc_text">
602 <p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
603 just turns on <b>too much</b> information (such as when working on the code
604 generator). If you want to enable debug information with more fine-grained
605 control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
606 option as follows:</p>
608 <div class="doc_code">
611 DEBUG(errs() << "No debug type\n");
612 #define DEBUG_TYPE "foo"
613 DEBUG(errs() << "'foo' debug type\n");
615 #define DEBUG_TYPE "bar"
616 DEBUG(errs() << "'bar' debug type\n"));
618 #define DEBUG_TYPE ""
619 DEBUG(errs() << "No debug type (2)\n");
623 <p>Then you can run your pass like this:</p>
625 <div class="doc_code">
627 $ opt < a.bc > /dev/null -mypass
628 <i><no output></i>
629 $ opt < a.bc > /dev/null -mypass -debug
634 $ opt < a.bc > /dev/null -mypass -debug-only=foo
636 $ opt < a.bc > /dev/null -mypass -debug-only=bar
641 <p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
642 a file, to specify the debug type for the entire module (if you do this before
643 you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
644 <tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
645 "bar", because there is no system in place to ensure that names do not
646 conflict. If two different modules use the same string, they will all be turned
647 on when the name is specified. This allows, for example, all debug information
648 for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
649 even if the source lives in multiple files.</p>
651 <p>The <tt>DEBUG_WITH_TYPE</tt> macro is also available for situations where you
652 would like to set <tt>DEBUG_TYPE</tt>, but only for one specific <tt>DEBUG</tt>
653 statement. It takes an additional first parameter, which is the type to use. For
654 example, the preceding example could be written as:</p>
657 <div class="doc_code">
659 DEBUG_WITH_TYPE("", errs() << "No debug type\n");
660 DEBUG_WITH_TYPE("foo", errs() << "'foo' debug type\n");
661 DEBUG_WITH_TYPE("bar", errs() << "'bar' debug type\n"));
662 DEBUG_WITH_TYPE("", errs() << "No debug type (2)\n");
668 <!-- ======================================================================= -->
669 <div class="doc_subsection">
670 <a name="Statistic">The <tt>Statistic</tt> class & <tt>-stats</tt>
674 <div class="doc_text">
677 href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
678 provides a class named <tt>Statistic</tt> that is used as a unified way to
679 keep track of what the LLVM compiler is doing and how effective various
680 optimizations are. It is useful to see what optimizations are contributing to
681 making a particular program run faster.</p>
683 <p>Often you may run your pass on some big program, and you're interested to see
684 how many times it makes a certain transformation. Although you can do this with
685 hand inspection, or some ad-hoc method, this is a real pain and not very useful
686 for big programs. Using the <tt>Statistic</tt> class makes it very easy to
687 keep track of this information, and the calculated information is presented in a
688 uniform manner with the rest of the passes being executed.</p>
690 <p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
691 it are as follows:</p>
694 <li><p>Define your statistic like this:</p>
696 <div class="doc_code">
698 #define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
699 STATISTIC(NumXForms, "The # of times I did stuff");
703 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
704 specified by the first argument. The pass name is taken from the DEBUG_TYPE
705 macro, and the description is taken from the second argument. The variable
706 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
708 <li><p>Whenever you make a transformation, bump the counter:</p>
710 <div class="doc_code">
712 ++NumXForms; // <i>I did stuff!</i>
719 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
720 statistics gathered, use the '<tt>-stats</tt>' option:</p>
722 <div class="doc_code">
724 $ opt -stats -mypassname < program.bc > /dev/null
725 <i>... statistics output ...</i>
729 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
730 suite, it gives a report that looks like this:</p>
732 <div class="doc_code">
734 7646 bitcodewriter - Number of normal instructions
735 725 bitcodewriter - Number of oversized instructions
736 129996 bitcodewriter - Number of bitcode bytes written
737 2817 raise - Number of insts DCEd or constprop'd
738 3213 raise - Number of cast-of-self removed
739 5046 raise - Number of expression trees converted
740 75 raise - Number of other getelementptr's formed
741 138 raise - Number of load/store peepholes
742 42 deadtypeelim - Number of unused typenames removed from symtab
743 392 funcresolve - Number of varargs functions resolved
744 27 globaldce - Number of global variables removed
745 2 adce - Number of basic blocks removed
746 134 cee - Number of branches revectored
747 49 cee - Number of setcc instruction eliminated
748 532 gcse - Number of loads removed
749 2919 gcse - Number of instructions removed
750 86 indvars - Number of canonical indvars added
751 87 indvars - Number of aux indvars removed
752 25 instcombine - Number of dead inst eliminate
753 434 instcombine - Number of insts combined
754 248 licm - Number of load insts hoisted
755 1298 licm - Number of insts hoisted to a loop pre-header
756 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
757 75 mem2reg - Number of alloca's promoted
758 1444 cfgsimplify - Number of blocks simplified
762 <p>Obviously, with so many optimizations, having a unified framework for this
763 stuff is very nice. Making your pass fit well into the framework makes it more
764 maintainable and useful.</p>
768 <!-- ======================================================================= -->
769 <div class="doc_subsection">
770 <a name="ViewGraph">Viewing graphs while debugging code</a>
773 <div class="doc_text">
775 <p>Several of the important data structures in LLVM are graphs: for example
776 CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
777 LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
778 <a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
779 DAGs</a>. In many cases, while debugging various parts of the compiler, it is
780 nice to instantly visualize these graphs.</p>
782 <p>LLVM provides several callbacks that are available in a debug build to do
783 exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
784 the current LLVM tool will pop up a window containing the CFG for the function
785 where each basic block is a node in the graph, and each node contains the
786 instructions in the block. Similarly, there also exists
787 <tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
788 <tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
789 and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
790 you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
791 up a window. Alternatively, you can sprinkle calls to these functions in your
792 code in places you want to debug.</p>
794 <p>Getting this to work requires a small amount of configuration. On Unix
795 systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
796 toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
797 Mac OS/X, download and install the Mac OS/X <a
798 href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
799 <tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
800 it) to your path. Once in your system and path are set up, rerun the LLVM
801 configure script and rebuild LLVM to enable this functionality.</p>
803 <p><tt>SelectionDAG</tt> has been extended to make it easier to locate
804 <i>interesting</i> nodes in large complex graphs. From gdb, if you
805 <tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
806 next <tt>call DAG.viewGraph()</tt> would highlight the node in the
807 specified color (choices of colors can be found at <a
808 href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
809 complex node attributes can be provided with <tt>call
810 DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
811 found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
812 Attributes</a>.) If you want to restart and clear all the current graph
813 attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
817 <!-- *********************************************************************** -->
818 <div class="doc_section">
819 <a name="datastructure">Picking the Right Data Structure for a Task</a>
821 <!-- *********************************************************************** -->
823 <div class="doc_text">
825 <p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
826 and we commonly use STL data structures. This section describes the trade-offs
827 you should consider when you pick one.</p>
830 The first step is a choose your own adventure: do you want a sequential
831 container, a set-like container, or a map-like container? The most important
832 thing when choosing a container is the algorithmic properties of how you plan to
833 access the container. Based on that, you should use:</p>
836 <li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
837 of an value based on another value. Map-like containers also support
838 efficient queries for containment (whether a key is in the map). Map-like
839 containers generally do not support efficient reverse mapping (values to
840 keys). If you need that, use two maps. Some map-like containers also
841 support efficient iteration through the keys in sorted order. Map-like
842 containers are the most expensive sort, only use them if you need one of
843 these capabilities.</li>
845 <li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
846 stuff into a container that automatically eliminates duplicates. Some
847 set-like containers support efficient iteration through the elements in
848 sorted order. Set-like containers are more expensive than sequential
852 <li>a <a href="#ds_sequential">sequential</a> container provides
853 the most efficient way to add elements and keeps track of the order they are
854 added to the collection. They permit duplicates and support efficient
855 iteration, but do not support efficient look-up based on a key.
858 <li>a <a href="#ds_string">string</a> container is a specialized sequential
859 container or reference structure that is used for character or byte
862 <li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
863 perform set operations on sets of numeric id's, while automatically
864 eliminating duplicates. Bit containers require a maximum of 1 bit for each
865 identifier you want to store.
870 Once the proper category of container is determined, you can fine tune the
871 memory use, constant factors, and cache behaviors of access by intelligently
872 picking a member of the category. Note that constant factors and cache behavior
873 can be a big deal. If you have a vector that usually only contains a few
874 elements (but could contain many), for example, it's much better to use
875 <a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
876 . Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
877 cost of adding the elements to the container. </p>
881 <!-- ======================================================================= -->
882 <div class="doc_subsection">
883 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
886 <div class="doc_text">
887 There are a variety of sequential containers available for you, based on your
888 needs. Pick the first in this section that will do what you want.
891 <!-- _______________________________________________________________________ -->
892 <div class="doc_subsubsection">
893 <a name="dss_fixedarrays">Fixed Size Arrays</a>
896 <div class="doc_text">
897 <p>Fixed size arrays are very simple and very fast. They are good if you know
898 exactly how many elements you have, or you have a (low) upper bound on how many
902 <!-- _______________________________________________________________________ -->
903 <div class="doc_subsubsection">
904 <a name="dss_heaparrays">Heap Allocated Arrays</a>
907 <div class="doc_text">
908 <p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
909 the number of elements is variable, if you know how many elements you will need
910 before the array is allocated, and if the array is usually large (if not,
911 consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
912 allocated array is the cost of the new/delete (aka malloc/free). Also note that
913 if you are allocating an array of a type with a constructor, the constructor and
914 destructors will be run for every element in the array (re-sizable vectors only
915 construct those elements actually used).</p>
918 <!-- _______________________________________________________________________ -->
919 <div class="doc_subsubsection">
920 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
923 <div class="doc_text">
924 <p><tt>SmallVector<Type, N></tt> is a simple class that looks and smells
925 just like <tt>vector<Type></tt>:
926 it supports efficient iteration, lays out elements in memory order (so you can
927 do pointer arithmetic between elements), supports efficient push_back/pop_back
928 operations, supports efficient random access to its elements, etc.</p>
930 <p>The advantage of SmallVector is that it allocates space for
931 some number of elements (N) <b>in the object itself</b>. Because of this, if
932 the SmallVector is dynamically smaller than N, no malloc is performed. This can
933 be a big win in cases where the malloc/free call is far more expensive than the
934 code that fiddles around with the elements.</p>
936 <p>This is good for vectors that are "usually small" (e.g. the number of
937 predecessors/successors of a block is usually less than 8). On the other hand,
938 this makes the size of the SmallVector itself large, so you don't want to
939 allocate lots of them (doing so will waste a lot of space). As such,
940 SmallVectors are most useful when on the stack.</p>
942 <p>SmallVector also provides a nice portable and efficient replacement for
947 <!-- _______________________________________________________________________ -->
948 <div class="doc_subsubsection">
949 <a name="dss_vector"><vector></a>
952 <div class="doc_text">
954 std::vector is well loved and respected. It is useful when SmallVector isn't:
955 when the size of the vector is often large (thus the small optimization will
956 rarely be a benefit) or if you will be allocating many instances of the vector
957 itself (which would waste space for elements that aren't in the container).
958 vector is also useful when interfacing with code that expects vectors :).
961 <p>One worthwhile note about std::vector: avoid code like this:</p>
963 <div class="doc_code">
966 std::vector<foo> V;
972 <p>Instead, write this as:</p>
974 <div class="doc_code">
976 std::vector<foo> V;
984 <p>Doing so will save (at least) one heap allocation and free per iteration of
989 <!-- _______________________________________________________________________ -->
990 <div class="doc_subsubsection">
991 <a name="dss_deque"><deque></a>
994 <div class="doc_text">
995 <p>std::deque is, in some senses, a generalized version of std::vector. Like
996 std::vector, it provides constant time random access and other similar
997 properties, but it also provides efficient access to the front of the list. It
998 does not guarantee continuity of elements within memory.</p>
1000 <p>In exchange for this extra flexibility, std::deque has significantly higher
1001 constant factor costs than std::vector. If possible, use std::vector or
1002 something cheaper.</p>
1005 <!-- _______________________________________________________________________ -->
1006 <div class="doc_subsubsection">
1007 <a name="dss_list"><list></a>
1010 <div class="doc_text">
1011 <p>std::list is an extremely inefficient class that is rarely useful.
1012 It performs a heap allocation for every element inserted into it, thus having an
1013 extremely high constant factor, particularly for small data types. std::list
1014 also only supports bidirectional iteration, not random access iteration.</p>
1016 <p>In exchange for this high cost, std::list supports efficient access to both
1017 ends of the list (like std::deque, but unlike std::vector or SmallVector). In
1018 addition, the iterator invalidation characteristics of std::list are stronger
1019 than that of a vector class: inserting or removing an element into the list does
1020 not invalidate iterator or pointers to other elements in the list.</p>
1023 <!-- _______________________________________________________________________ -->
1024 <div class="doc_subsubsection">
1025 <a name="dss_ilist">llvm/ADT/ilist.h</a>
1028 <div class="doc_text">
1029 <p><tt>ilist<T></tt> implements an 'intrusive' doubly-linked list. It is
1030 intrusive, because it requires the element to store and provide access to the
1031 prev/next pointers for the list.</p>
1033 <p><tt>ilist</tt> has the same drawbacks as <tt>std::list</tt>, and additionally
1034 requires an <tt>ilist_traits</tt> implementation for the element type, but it
1035 provides some novel characteristics. In particular, it can efficiently store
1036 polymorphic objects, the traits class is informed when an element is inserted or
1037 removed from the list, and <tt>ilist</tt>s are guaranteed to support a
1038 constant-time splice operation.</p>
1040 <p>These properties are exactly what we want for things like
1041 <tt>Instruction</tt>s and basic blocks, which is why these are implemented with
1042 <tt>ilist</tt>s.</p>
1044 Related classes of interest are explained in the following subsections:
1046 <li><a href="#dss_ilist_traits">ilist_traits</a></li>
1047 <li><a href="#dss_iplist">iplist</a></li>
1048 <li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
1049 <li><a href="#dss_ilist_sentinel">Sentinels</a></li>
1053 <!-- _______________________________________________________________________ -->
1054 <div class="doc_subsubsection">
1055 <a name="dss_ilist_traits">ilist_traits</a>
1058 <div class="doc_text">
1059 <p><tt>ilist_traits<T></tt> is <tt>ilist<T></tt>'s customization
1060 mechanism. <tt>iplist<T></tt> (and consequently <tt>ilist<T></tt>)
1061 publicly derive from this traits class.</p>
1064 <!-- _______________________________________________________________________ -->
1065 <div class="doc_subsubsection">
1066 <a name="dss_iplist">iplist</a>
1069 <div class="doc_text">
1070 <p><tt>iplist<T></tt> is <tt>ilist<T></tt>'s base and as such
1071 supports a slightly narrower interface. Notably, inserters from
1072 <tt>T&</tt> are absent.</p>
1074 <p><tt>ilist_traits<T></tt> is a public base of this class and can be
1075 used for a wide variety of customizations.</p>
1078 <!-- _______________________________________________________________________ -->
1079 <div class="doc_subsubsection">
1080 <a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
1083 <div class="doc_text">
1084 <p><tt>ilist_node<T></tt> implements a the forward and backward links
1085 that are expected by the <tt>ilist<T></tt> (and analogous containers)
1086 in the default manner.</p>
1088 <p><tt>ilist_node<T></tt>s are meant to be embedded in the node type
1089 <tt>T</tt>, usually <tt>T</tt> publicly derives from
1090 <tt>ilist_node<T></tt>.</p>
1093 <!-- _______________________________________________________________________ -->
1094 <div class="doc_subsubsection">
1095 <a name="dss_ilist_sentinel">Sentinels</a>
1098 <div class="doc_text">
1099 <p><tt>ilist</tt>s have another specialty that must be considered. To be a good
1100 citizen in the C++ ecosystem, it needs to support the standard container
1101 operations, such as <tt>begin</tt> and <tt>end</tt> iterators, etc. Also, the
1102 <tt>operator--</tt> must work correctly on the <tt>end</tt> iterator in the
1103 case of non-empty <tt>ilist</tt>s.</p>
1105 <p>The only sensible solution to this problem is to allocate a so-called
1106 <i>sentinel</i> along with the intrusive list, which serves as the <tt>end</tt>
1107 iterator, providing the back-link to the last element. However conforming to the
1108 C++ convention it is illegal to <tt>operator++</tt> beyond the sentinel and it
1109 also must not be dereferenced.</p>
1111 <p>These constraints allow for some implementation freedom to the <tt>ilist</tt>
1112 how to allocate and store the sentinel. The corresponding policy is dictated
1113 by <tt>ilist_traits<T></tt>. By default a <tt>T</tt> gets heap-allocated
1114 whenever the need for a sentinel arises.</p>
1116 <p>While the default policy is sufficient in most cases, it may break down when
1117 <tt>T</tt> does not provide a default constructor. Also, in the case of many
1118 instances of <tt>ilist</tt>s, the memory overhead of the associated sentinels
1119 is wasted. To alleviate the situation with numerous and voluminous
1120 <tt>T</tt>-sentinels, sometimes a trick is employed, leading to <i>ghostly
1123 <p>Ghostly sentinels are obtained by specially-crafted <tt>ilist_traits<T></tt>
1124 which superpose the sentinel with the <tt>ilist</tt> instance in memory. Pointer
1125 arithmetic is used to obtain the sentinel, which is relative to the
1126 <tt>ilist</tt>'s <tt>this</tt> pointer. The <tt>ilist</tt> is augmented by an
1127 extra pointer, which serves as the back-link of the sentinel. This is the only
1128 field in the ghostly sentinel which can be legally accessed.</p>
1131 <!-- _______________________________________________________________________ -->
1132 <div class="doc_subsubsection">
1133 <a name="dss_other">Other Sequential Container options</a>
1136 <div class="doc_text">
1137 <p>Other STL containers are available, such as std::string.</p>
1139 <p>There are also various STL adapter classes such as std::queue,
1140 std::priority_queue, std::stack, etc. These provide simplified access to an
1141 underlying container but don't affect the cost of the container itself.</p>
1146 <!-- ======================================================================= -->
1147 <div class="doc_subsection">
1148 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
1151 <div class="doc_text">
1153 <p>Set-like containers are useful when you need to canonicalize multiple values
1154 into a single representation. There are several different choices for how to do
1155 this, providing various trade-offs.</p>
1160 <!-- _______________________________________________________________________ -->
1161 <div class="doc_subsubsection">
1162 <a name="dss_sortedvectorset">A sorted 'vector'</a>
1165 <div class="doc_text">
1167 <p>If you intend to insert a lot of elements, then do a lot of queries, a
1168 great approach is to use a vector (or other sequential container) with
1169 std::sort+std::unique to remove duplicates. This approach works really well if
1170 your usage pattern has these two distinct phases (insert then query), and can be
1171 coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
1175 This combination provides the several nice properties: the result data is
1176 contiguous in memory (good for cache locality), has few allocations, is easy to
1177 address (iterators in the final vector are just indices or pointers), and can be
1178 efficiently queried with a standard binary or radix search.</p>
1182 <!-- _______________________________________________________________________ -->
1183 <div class="doc_subsubsection">
1184 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
1187 <div class="doc_text">
1189 <p>If you have a set-like data structure that is usually small and whose elements
1190 are reasonably small, a <tt>SmallSet<Type, N></tt> is a good choice. This set
1191 has space for N elements in place (thus, if the set is dynamically smaller than
1192 N, no malloc traffic is required) and accesses them with a simple linear search.
1193 When the set grows beyond 'N' elements, it allocates a more expensive representation that
1194 guarantees efficient access (for most types, it falls back to std::set, but for
1195 pointers it uses something far better, <a
1196 href="#dss_smallptrset">SmallPtrSet</a>).</p>
1198 <p>The magic of this class is that it handles small sets extremely efficiently,
1199 but gracefully handles extremely large sets without loss of efficiency. The
1200 drawback is that the interface is quite small: it supports insertion, queries
1201 and erasing, but does not support iteration.</p>
1205 <!-- _______________________________________________________________________ -->
1206 <div class="doc_subsubsection">
1207 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
1210 <div class="doc_text">
1212 <p>SmallPtrSet has all the advantages of <tt>SmallSet</tt> (and a <tt>SmallSet</tt> of pointers is
1213 transparently implemented with a <tt>SmallPtrSet</tt>), but also supports iterators. If
1214 more than 'N' insertions are performed, a single quadratically
1215 probed hash table is allocated and grows as needed, providing extremely
1216 efficient access (constant time insertion/deleting/queries with low constant
1217 factors) and is very stingy with malloc traffic.</p>
1219 <p>Note that, unlike <tt>std::set</tt>, the iterators of <tt>SmallPtrSet</tt> are invalidated
1220 whenever an insertion occurs. Also, the values visited by the iterators are not
1221 visited in sorted order.</p>
1225 <!-- _______________________________________________________________________ -->
1226 <div class="doc_subsubsection">
1227 <a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
1230 <div class="doc_text">
1233 DenseSet is a simple quadratically probed hash table. It excels at supporting
1234 small values: it uses a single allocation to hold all of the pairs that
1235 are currently inserted in the set. DenseSet is a great way to unique small
1236 values that are not simple pointers (use <a
1237 href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
1238 the same requirements for the value type that <a
1239 href="#dss_densemap">DenseMap</a> has.
1244 <!-- _______________________________________________________________________ -->
1245 <div class="doc_subsubsection">
1246 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
1249 <div class="doc_text">
1252 FoldingSet is an aggregate class that is really good at uniquing
1253 expensive-to-create or polymorphic objects. It is a combination of a chained
1254 hash table with intrusive links (uniqued objects are required to inherit from
1255 FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
1258 <p>Consider a case where you want to implement a "getOrCreateFoo" method for
1259 a complex object (for example, a node in the code generator). The client has a
1260 description of *what* it wants to generate (it knows the opcode and all the
1261 operands), but we don't want to 'new' a node, then try inserting it into a set
1262 only to find out it already exists, at which point we would have to delete it
1263 and return the node that already exists.
1266 <p>To support this style of client, FoldingSet perform a query with a
1267 FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1268 element that we want to query for. The query either returns the element
1269 matching the ID or it returns an opaque ID that indicates where insertion should
1270 take place. Construction of the ID usually does not require heap traffic.</p>
1272 <p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1273 in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1274 Because the elements are individually allocated, pointers to the elements are
1275 stable: inserting or removing elements does not invalidate any pointers to other
1281 <!-- _______________________________________________________________________ -->
1282 <div class="doc_subsubsection">
1283 <a name="dss_set"><set></a>
1286 <div class="doc_text">
1288 <p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1289 many things but great at nothing. std::set allocates memory for each element
1290 inserted (thus it is very malloc intensive) and typically stores three pointers
1291 per element in the set (thus adding a large amount of per-element space
1292 overhead). It offers guaranteed log(n) performance, which is not particularly
1293 fast from a complexity standpoint (particularly if the elements of the set are
1294 expensive to compare, like strings), and has extremely high constant factors for
1295 lookup, insertion and removal.</p>
1297 <p>The advantages of std::set are that its iterators are stable (deleting or
1298 inserting an element from the set does not affect iterators or pointers to other
1299 elements) and that iteration over the set is guaranteed to be in sorted order.
1300 If the elements in the set are large, then the relative overhead of the pointers
1301 and malloc traffic is not a big deal, but if the elements of the set are small,
1302 std::set is almost never a good choice.</p>
1306 <!-- _______________________________________________________________________ -->
1307 <div class="doc_subsubsection">
1308 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
1311 <div class="doc_text">
1312 <p>LLVM's SetVector<Type> is an adapter class that combines your choice of
1313 a set-like container along with a <a href="#ds_sequential">Sequential
1314 Container</a>. The important property
1315 that this provides is efficient insertion with uniquing (duplicate elements are
1316 ignored) with iteration support. It implements this by inserting elements into
1317 both a set-like container and the sequential container, using the set-like
1318 container for uniquing and the sequential container for iteration.
1321 <p>The difference between SetVector and other sets is that the order of
1322 iteration is guaranteed to match the order of insertion into the SetVector.
1323 This property is really important for things like sets of pointers. Because
1324 pointer values are non-deterministic (e.g. vary across runs of the program on
1325 different machines), iterating over the pointers in the set will
1326 not be in a well-defined order.</p>
1329 The drawback of SetVector is that it requires twice as much space as a normal
1330 set and has the sum of constant factors from the set-like container and the
1331 sequential container that it uses. Use it *only* if you need to iterate over
1332 the elements in a deterministic order. SetVector is also expensive to delete
1333 elements out of (linear time), unless you use it's "pop_back" method, which is
1337 <p>SetVector is an adapter class that defaults to using std::vector and std::set
1338 for the underlying containers, so it is quite expensive. However,
1339 <tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
1340 defaults to using a SmallVector and SmallSet of a specified size. If you use
1341 this, and if your sets are dynamically smaller than N, you will save a lot of
1346 <!-- _______________________________________________________________________ -->
1347 <div class="doc_subsubsection">
1348 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
1351 <div class="doc_text">
1354 UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1355 retains a unique ID for each element inserted into the set. It internally
1356 contains a map and a vector, and it assigns a unique ID for each value inserted
1359 <p>UniqueVector is very expensive: its cost is the sum of the cost of
1360 maintaining both the map and vector, it has high complexity, high constant
1361 factors, and produces a lot of malloc traffic. It should be avoided.</p>
1366 <!-- _______________________________________________________________________ -->
1367 <div class="doc_subsubsection">
1368 <a name="dss_otherset">Other Set-Like Container Options</a>
1371 <div class="doc_text">
1374 The STL provides several other options, such as std::multiset and the various
1375 "hash_set" like containers (whether from C++ TR1 or from the SGI library). We
1376 never use hash_set and unordered_set because they are generally very expensive
1377 (each insertion requires a malloc) and very non-portable.
1380 <p>std::multiset is useful if you're not interested in elimination of
1381 duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1382 don't delete duplicate entries) or some other approach is almost always
1387 <!-- ======================================================================= -->
1388 <div class="doc_subsection">
1389 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
1392 <div class="doc_text">
1393 Map-like containers are useful when you want to associate data to a key. As
1394 usual, there are a lot of different ways to do this. :)
1397 <!-- _______________________________________________________________________ -->
1398 <div class="doc_subsubsection">
1399 <a name="dss_sortedvectormap">A sorted 'vector'</a>
1402 <div class="doc_text">
1405 If your usage pattern follows a strict insert-then-query approach, you can
1406 trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1407 for set-like containers</a>. The only difference is that your query function
1408 (which uses std::lower_bound to get efficient log(n) lookup) should only compare
1409 the key, not both the key and value. This yields the same advantages as sorted
1414 <!-- _______________________________________________________________________ -->
1415 <div class="doc_subsubsection">
1416 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
1419 <div class="doc_text">
1422 Strings are commonly used as keys in maps, and they are difficult to support
1423 efficiently: they are variable length, inefficient to hash and compare when
1424 long, expensive to copy, etc. StringMap is a specialized container designed to
1425 cope with these issues. It supports mapping an arbitrary range of bytes to an
1426 arbitrary other object.</p>
1428 <p>The StringMap implementation uses a quadratically-probed hash table, where
1429 the buckets store a pointer to the heap allocated entries (and some other
1430 stuff). The entries in the map must be heap allocated because the strings are
1431 variable length. The string data (key) and the element object (value) are
1432 stored in the same allocation with the string data immediately after the element
1433 object. This container guarantees the "<tt>(char*)(&Value+1)</tt>" points
1434 to the key string for a value.</p>
1436 <p>The StringMap is very fast for several reasons: quadratic probing is very
1437 cache efficient for lookups, the hash value of strings in buckets is not
1438 recomputed when looking up an element, StringMap rarely has to touch the
1439 memory for unrelated objects when looking up a value (even when hash collisions
1440 happen), hash table growth does not recompute the hash values for strings
1441 already in the table, and each pair in the map is store in a single allocation
1442 (the string data is stored in the same allocation as the Value of a pair).</p>
1444 <p>StringMap also provides query methods that take byte ranges, so it only ever
1445 copies a string if a value is inserted into the table.</p>
1448 <!-- _______________________________________________________________________ -->
1449 <div class="doc_subsubsection">
1450 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
1453 <div class="doc_text">
1455 IndexedMap is a specialized container for mapping small dense integers (or
1456 values that can be mapped to small dense integers) to some other type. It is
1457 internally implemented as a vector with a mapping function that maps the keys to
1458 the dense integer range.
1462 This is useful for cases like virtual registers in the LLVM code generator: they
1463 have a dense mapping that is offset by a compile-time constant (the first
1464 virtual register ID).</p>
1468 <!-- _______________________________________________________________________ -->
1469 <div class="doc_subsubsection">
1470 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
1473 <div class="doc_text">
1476 DenseMap is a simple quadratically probed hash table. It excels at supporting
1477 small keys and values: it uses a single allocation to hold all of the pairs that
1478 are currently inserted in the map. DenseMap is a great way to map pointers to
1479 pointers, or map other small types to each other.
1483 There are several aspects of DenseMap that you should be aware of, however. The
1484 iterators in a densemap are invalidated whenever an insertion occurs, unlike
1485 map. Also, because DenseMap allocates space for a large number of key/value
1486 pairs (it starts with 64 by default), it will waste a lot of space if your keys
1487 or values are large. Finally, you must implement a partial specialization of
1488 DenseMapInfo for the key that you want, if it isn't already supported. This
1489 is required to tell DenseMap about two special marker values (which can never be
1490 inserted into the map) that it needs internally.</p>
1494 <!-- _______________________________________________________________________ -->
1495 <div class="doc_subsubsection">
1496 <a name="dss_valuemap">"llvm/ADT/ValueMap.h"</a>
1499 <div class="doc_text">
1502 ValueMap is a wrapper around a <a href="#dss_densemap">DenseMap</a> mapping
1503 Value*s (or subclasses) to another type. When a Value is deleted or RAUW'ed,
1504 ValueMap will update itself so the new version of the key is mapped to the same
1505 value, just as if the key were a WeakVH. You can configure exactly how this
1506 happens, and what else happens on these two events, by passing
1507 a <code>Config</code> parameter to the ValueMap template.</p>
1511 <!-- _______________________________________________________________________ -->
1512 <div class="doc_subsubsection">
1513 <a name="dss_intervalmap">"llvm/ADT/IntervalMap.h"</a>
1516 <div class="doc_text">
1518 <p> IntervalMap is a compact map for small keys and values. It maps key
1519 intervals instead of single keys, and it will automatically coalesce adjacent
1520 intervals. When then map only contains a few intervals, they are stored in the
1521 map object itself to avoid allocations.</p>
1523 <p> The IntervalMap iterators are quite big, so they should not be passed around
1524 as STL iterators. The heavyweight iterators allow a smaller data structure.</p>
1528 <!-- _______________________________________________________________________ -->
1529 <div class="doc_subsubsection">
1530 <a name="dss_map"><map></a>
1533 <div class="doc_text">
1536 std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1537 a single allocation per pair inserted into the map, it offers log(n) lookup with
1538 an extremely large constant factor, imposes a space penalty of 3 pointers per
1539 pair in the map, etc.</p>
1541 <p>std::map is most useful when your keys or values are very large, if you need
1542 to iterate over the collection in sorted order, or if you need stable iterators
1543 into the map (i.e. they don't get invalidated if an insertion or deletion of
1544 another element takes place).</p>
1548 <!-- _______________________________________________________________________ -->
1549 <div class="doc_subsubsection">
1550 <a name="dss_othermap">Other Map-Like Container Options</a>
1553 <div class="doc_text">
1556 The STL provides several other options, such as std::multimap and the various
1557 "hash_map" like containers (whether from C++ TR1 or from the SGI library). We
1558 never use hash_set and unordered_set because they are generally very expensive
1559 (each insertion requires a malloc) and very non-portable.</p>
1561 <p>std::multimap is useful if you want to map a key to multiple values, but has
1562 all the drawbacks of std::map. A sorted vector or some other approach is almost
1567 <!-- ======================================================================= -->
1568 <div class="doc_subsection">
1569 <a name="ds_string">String-like containers</a>
1572 <div class="doc_text">
1575 TODO: const char* vs stringref vs smallstring vs std::string. Describe twine,
1576 xref to #string_apis.
1581 <!-- ======================================================================= -->
1582 <div class="doc_subsection">
1583 <a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
1586 <div class="doc_text">
1587 <p>Unlike the other containers, there are only two bit storage containers, and
1588 choosing when to use each is relatively straightforward.</p>
1590 <p>One additional option is
1591 <tt>std::vector<bool></tt>: we discourage its use for two reasons 1) the
1592 implementation in many common compilers (e.g. commonly available versions of
1593 GCC) is extremely inefficient and 2) the C++ standards committee is likely to
1594 deprecate this container and/or change it significantly somehow. In any case,
1595 please don't use it.</p>
1598 <!-- _______________________________________________________________________ -->
1599 <div class="doc_subsubsection">
1600 <a name="dss_bitvector">BitVector</a>
1603 <div class="doc_text">
1604 <p> The BitVector container provides a dynamic size set of bits for manipulation.
1605 It supports individual bit setting/testing, as well as set operations. The set
1606 operations take time O(size of bitvector), but operations are performed one word
1607 at a time, instead of one bit at a time. This makes the BitVector very fast for
1608 set operations compared to other containers. Use the BitVector when you expect
1609 the number of set bits to be high (IE a dense set).
1613 <!-- _______________________________________________________________________ -->
1614 <div class="doc_subsubsection">
1615 <a name="dss_smallbitvector">SmallBitVector</a>
1618 <div class="doc_text">
1619 <p> The SmallBitVector container provides the same interface as BitVector, but
1620 it is optimized for the case where only a small number of bits, less than
1621 25 or so, are needed. It also transparently supports larger bit counts, but
1622 slightly less efficiently than a plain BitVector, so SmallBitVector should
1623 only be used when larger counts are rare.
1627 At this time, SmallBitVector does not support set operations (and, or, xor),
1628 and its operator[] does not provide an assignable lvalue.
1632 <!-- _______________________________________________________________________ -->
1633 <div class="doc_subsubsection">
1634 <a name="dss_sparsebitvector">SparseBitVector</a>
1637 <div class="doc_text">
1638 <p> The SparseBitVector container is much like BitVector, with one major
1639 difference: Only the bits that are set, are stored. This makes the
1640 SparseBitVector much more space efficient than BitVector when the set is sparse,
1641 as well as making set operations O(number of set bits) instead of O(size of
1642 universe). The downside to the SparseBitVector is that setting and testing of random bits is O(N), and on large SparseBitVectors, this can be slower than BitVector. In our implementation, setting or testing bits in sorted order
1643 (either forwards or reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends on size) of the current bit is also O(1). As a general statement, testing/setting bits in a SparseBitVector is O(distance away from last set bit).
1647 <!-- *********************************************************************** -->
1648 <div class="doc_section">
1649 <a name="common">Helpful Hints for Common Operations</a>
1651 <!-- *********************************************************************** -->
1653 <div class="doc_text">
1655 <p>This section describes how to perform some very simple transformations of
1656 LLVM code. This is meant to give examples of common idioms used, showing the
1657 practical side of LLVM transformations. <p> Because this is a "how-to" section,
1658 you should also read about the main classes that you will be working with. The
1659 <a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
1660 and descriptions of the main classes that you should know about.</p>
1664 <!-- NOTE: this section should be heavy on example code -->
1665 <!-- ======================================================================= -->
1666 <div class="doc_subsection">
1667 <a name="inspection">Basic Inspection and Traversal Routines</a>
1670 <div class="doc_text">
1672 <p>The LLVM compiler infrastructure have many different data structures that may
1673 be traversed. Following the example of the C++ standard template library, the
1674 techniques used to traverse these various data structures are all basically the
1675 same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
1676 method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
1677 function returns an iterator pointing to one past the last valid element of the
1678 sequence, and there is some <tt>XXXiterator</tt> data type that is common
1679 between the two operations.</p>
1681 <p>Because the pattern for iteration is common across many different aspects of
1682 the program representation, the standard template library algorithms may be used
1683 on them, and it is easier to remember how to iterate. First we show a few common
1684 examples of the data structures that need to be traversed. Other data
1685 structures are traversed in very similar ways.</p>
1689 <!-- _______________________________________________________________________ -->
1690 <div class="doc_subsubsection">
1691 <a name="iterate_function">Iterating over the </a><a
1692 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
1693 href="#Function"><tt>Function</tt></a>
1696 <div class="doc_text">
1698 <p>It's quite common to have a <tt>Function</tt> instance that you'd like to
1699 transform in some way; in particular, you'd like to manipulate its
1700 <tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
1701 the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
1702 an example that prints the name of a <tt>BasicBlock</tt> and the number of
1703 <tt>Instruction</tt>s it contains:</p>
1705 <div class="doc_code">
1707 // <i>func is a pointer to a Function instance</i>
1708 for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i)
1709 // <i>Print out the name of the basic block if it has one, and then the</i>
1710 // <i>number of instructions that it contains</i>
1711 errs() << "Basic block (name=" << i->getName() << ") has "
1712 << i->size() << " instructions.\n";
1716 <p>Note that i can be used as if it were a pointer for the purposes of
1717 invoking member functions of the <tt>Instruction</tt> class. This is
1718 because the indirection operator is overloaded for the iterator
1719 classes. In the above code, the expression <tt>i->size()</tt> is
1720 exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
1724 <!-- _______________________________________________________________________ -->
1725 <div class="doc_subsubsection">
1726 <a name="iterate_basicblock">Iterating over the </a><a
1727 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1728 href="#BasicBlock"><tt>BasicBlock</tt></a>
1731 <div class="doc_text">
1733 <p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
1734 easy to iterate over the individual instructions that make up
1735 <tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
1736 a <tt>BasicBlock</tt>:</p>
1738 <div class="doc_code">
1740 // <i>blk is a pointer to a BasicBlock instance</i>
1741 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
1742 // <i>The next statement works since operator<<(ostream&,...)</i>
1743 // <i>is overloaded for Instruction&</i>
1744 errs() << *i << "\n";
1748 <p>However, this isn't really the best way to print out the contents of a
1749 <tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
1750 anything you'll care about, you could have just invoked the print routine on the
1751 basic block itself: <tt>errs() << *blk << "\n";</tt>.</p>
1755 <!-- _______________________________________________________________________ -->
1756 <div class="doc_subsubsection">
1757 <a name="iterate_institer">Iterating over the </a><a
1758 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1759 href="#Function"><tt>Function</tt></a>
1762 <div class="doc_text">
1764 <p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
1765 <tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
1766 <tt>InstIterator</tt> should be used instead. You'll need to include <a
1767 href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
1768 and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
1769 small example that shows how to dump all instructions in a function to the standard error stream:<p>
1771 <div class="doc_code">
1773 #include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
1775 // <i>F is a pointer to a Function instance</i>
1776 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1777 errs() << *I << "\n";
1781 <p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
1782 work list with its initial contents. For example, if you wanted to
1783 initialize a work list to contain all instructions in a <tt>Function</tt>
1784 F, all you would need to do is something like:</p>
1786 <div class="doc_code">
1788 std::set<Instruction*> worklist;
1789 // or better yet, SmallPtrSet<Instruction*, 64> worklist;
1791 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1792 worklist.insert(&*I);
1796 <p>The STL set <tt>worklist</tt> would now contain all instructions in the
1797 <tt>Function</tt> pointed to by F.</p>
1801 <!-- _______________________________________________________________________ -->
1802 <div class="doc_subsubsection">
1803 <a name="iterate_convert">Turning an iterator into a class pointer (and
1807 <div class="doc_text">
1809 <p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
1810 instance when all you've got at hand is an iterator. Well, extracting
1811 a reference or a pointer from an iterator is very straight-forward.
1812 Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
1813 is a <tt>BasicBlock::const_iterator</tt>:</p>
1815 <div class="doc_code">
1817 Instruction& inst = *i; // <i>Grab reference to instruction reference</i>
1818 Instruction* pinst = &*i; // <i>Grab pointer to instruction reference</i>
1819 const Instruction& inst = *j;
1823 <p>However, the iterators you'll be working with in the LLVM framework are
1824 special: they will automatically convert to a ptr-to-instance type whenever they
1825 need to. Instead of dereferencing the iterator and then taking the address of
1826 the result, you can simply assign the iterator to the proper pointer type and
1827 you get the dereference and address-of operation as a result of the assignment
1828 (behind the scenes, this is a result of overloading casting mechanisms). Thus
1829 the last line of the last example,</p>
1831 <div class="doc_code">
1833 Instruction *pinst = &*i;
1837 <p>is semantically equivalent to</p>
1839 <div class="doc_code">
1841 Instruction *pinst = i;
1845 <p>It's also possible to turn a class pointer into the corresponding iterator,
1846 and this is a constant time operation (very efficient). The following code
1847 snippet illustrates use of the conversion constructors provided by LLVM
1848 iterators. By using these, you can explicitly grab the iterator of something
1849 without actually obtaining it via iteration over some structure:</p>
1851 <div class="doc_code">
1853 void printNextInstruction(Instruction* inst) {
1854 BasicBlock::iterator it(inst);
1855 ++it; // <i>After this line, it refers to the instruction after *inst</i>
1856 if (it != inst->getParent()->end()) errs() << *it << "\n";
1861 <p>Unfortunately, these implicit conversions come at a cost; they prevent
1862 these iterators from conforming to standard iterator conventions, and thus
1863 from being usable with standard algorithms and containers. For example, they
1864 prevent the following code, where <tt>B</tt> is a <tt>BasicBlock</tt>,
1867 <div class="doc_code">
1869 llvm::SmallVector<llvm::Instruction *, 16>(B->begin(), B->end());
1873 <p>Because of this, these implicit conversions may be removed some day,
1874 and <tt>operator*</tt> changed to return a pointer instead of a reference.</p>
1878 <!--_______________________________________________________________________-->
1879 <div class="doc_subsubsection">
1880 <a name="iterate_complex">Finding call sites: a slightly more complex
1884 <div class="doc_text">
1886 <p>Say that you're writing a FunctionPass and would like to count all the
1887 locations in the entire module (that is, across every <tt>Function</tt>) where a
1888 certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
1889 learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
1890 much more straight-forward manner, but this example will allow us to explore how
1891 you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
1892 is what we want to do:</p>
1894 <div class="doc_code">
1896 initialize callCounter to zero
1897 for each Function f in the Module
1898 for each BasicBlock b in f
1899 for each Instruction i in b
1900 if (i is a CallInst and calls the given function)
1901 increment callCounter
1905 <p>And the actual code is (remember, because we're writing a
1906 <tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
1907 override the <tt>runOnFunction</tt> method):</p>
1909 <div class="doc_code">
1911 Function* targetFunc = ...;
1913 class OurFunctionPass : public FunctionPass {
1915 OurFunctionPass(): callCounter(0) { }
1917 virtual runOnFunction(Function& F) {
1918 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
1919 for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) {
1920 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a><<a
1921 href="#CallInst">CallInst</a>>(&*i)) {
1922 // <i>We know we've encountered a call instruction, so we</i>
1923 // <i>need to determine if it's a call to the</i>
1924 // <i>function pointed to by m_func or not.</i>
1925 if (callInst->getCalledFunction() == targetFunc)
1933 unsigned callCounter;
1940 <!--_______________________________________________________________________-->
1941 <div class="doc_subsubsection">
1942 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
1945 <div class="doc_text">
1947 <p>You may have noticed that the previous example was a bit oversimplified in
1948 that it did not deal with call sites generated by 'invoke' instructions. In
1949 this, and in other situations, you may find that you want to treat
1950 <tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
1951 most-specific common base class is <tt>Instruction</tt>, which includes lots of
1952 less closely-related things. For these cases, LLVM provides a handy wrapper
1954 href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
1955 It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
1956 methods that provide functionality common to <tt>CallInst</tt>s and
1957 <tt>InvokeInst</tt>s.</p>
1959 <p>This class has "value semantics": it should be passed by value, not by
1960 reference and it should not be dynamically allocated or deallocated using
1961 <tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
1962 assignable and constructable, with costs equivalents to that of a bare pointer.
1963 If you look at its definition, it has only a single pointer member.</p>
1967 <!--_______________________________________________________________________-->
1968 <div class="doc_subsubsection">
1969 <a name="iterate_chains">Iterating over def-use & use-def chains</a>
1972 <div class="doc_text">
1974 <p>Frequently, we might have an instance of the <a
1975 href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
1976 determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
1977 <tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
1978 For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
1979 particular function <tt>foo</tt>. Finding all of the instructions that
1980 <i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
1983 <div class="doc_code">
1987 for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i)
1988 if (Instruction *Inst = dyn_cast<Instruction>(*i)) {
1989 errs() << "F is used in instruction:\n";
1990 errs() << *Inst << "\n";
1995 <p>Note that dereferencing a <tt>Value::use_iterator</tt> is not a very cheap
1996 operation. Instead of performing <tt>*i</tt> above several times, consider
1997 doing it only once in the loop body and reusing its result.</p>
1999 <p>Alternatively, it's common to have an instance of the <a
2000 href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
2001 <tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
2002 <tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
2003 <tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
2004 all of the values that a particular instruction uses (that is, the operands of
2005 the particular <tt>Instruction</tt>):</p>
2007 <div class="doc_code">
2009 Instruction *pi = ...;
2011 for (User::op_iterator i = pi->op_begin(), e = pi->op_end(); i != e; ++i) {
2018 <p>Declaring objects as <tt>const</tt> is an important tool of enforcing
2019 mutation free algorithms (such as analyses, etc.). For this purpose above
2020 iterators come in constant flavors as <tt>Value::const_use_iterator</tt>
2021 and <tt>Value::const_op_iterator</tt>. They automatically arise when
2022 calling <tt>use/op_begin()</tt> on <tt>const Value*</tt>s or
2023 <tt>const User*</tt>s respectively. Upon dereferencing, they return
2024 <tt>const Use*</tt>s. Otherwise the above patterns remain unchanged.</p>
2028 <!--_______________________________________________________________________-->
2029 <div class="doc_subsubsection">
2030 <a name="iterate_preds">Iterating over predecessors &
2031 successors of blocks</a>
2034 <div class="doc_text">
2036 <p>Iterating over the predecessors and successors of a block is quite easy
2037 with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
2038 this to iterate over all predecessors of BB:</p>
2040 <div class="doc_code">
2042 #include "llvm/Support/CFG.h"
2043 BasicBlock *BB = ...;
2045 for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
2046 BasicBlock *Pred = *PI;
2052 <p>Similarly, to iterate over successors use
2053 succ_iterator/succ_begin/succ_end.</p>
2058 <!-- ======================================================================= -->
2059 <div class="doc_subsection">
2060 <a name="simplechanges">Making simple changes</a>
2063 <div class="doc_text">
2065 <p>There are some primitive transformation operations present in the LLVM
2066 infrastructure that are worth knowing about. When performing
2067 transformations, it's fairly common to manipulate the contents of basic
2068 blocks. This section describes some of the common methods for doing so
2069 and gives example code.</p>
2073 <!--_______________________________________________________________________-->
2074 <div class="doc_subsubsection">
2075 <a name="schanges_creating">Creating and inserting new
2076 <tt>Instruction</tt>s</a>
2079 <div class="doc_text">
2081 <p><i>Instantiating Instructions</i></p>
2083 <p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
2084 constructor for the kind of instruction to instantiate and provide the necessary
2085 parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
2086 (const-ptr-to) <tt>Type</tt>. Thus:</p>
2088 <div class="doc_code">
2090 AllocaInst* ai = new AllocaInst(Type::Int32Ty);
2094 <p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
2095 one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
2096 subclass is likely to have varying default parameters which change the semantics
2097 of the instruction, so refer to the <a
2098 href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
2099 Instruction</a> that you're interested in instantiating.</p>
2101 <p><i>Naming values</i></p>
2103 <p>It is very useful to name the values of instructions when you're able to, as
2104 this facilitates the debugging of your transformations. If you end up looking
2105 at generated LLVM machine code, you definitely want to have logical names
2106 associated with the results of instructions! By supplying a value for the
2107 <tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
2108 associate a logical name with the result of the instruction's execution at
2109 run time. For example, say that I'm writing a transformation that dynamically
2110 allocates space for an integer on the stack, and that integer is going to be
2111 used as some kind of index by some other code. To accomplish this, I place an
2112 <tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
2113 <tt>Function</tt>, and I'm intending to use it within the same
2114 <tt>Function</tt>. I might do:</p>
2116 <div class="doc_code">
2118 AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
2122 <p>where <tt>indexLoc</tt> is now the logical name of the instruction's
2123 execution value, which is a pointer to an integer on the run time stack.</p>
2125 <p><i>Inserting instructions</i></p>
2127 <p>There are essentially two ways to insert an <tt>Instruction</tt>
2128 into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
2131 <li>Insertion into an explicit instruction list
2133 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
2134 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
2135 before <tt>*pi</tt>, we do the following: </p>
2137 <div class="doc_code">
2139 BasicBlock *pb = ...;
2140 Instruction *pi = ...;
2141 Instruction *newInst = new Instruction(...);
2143 pb->getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
2147 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
2148 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
2149 classes provide constructors which take a pointer to a
2150 <tt>BasicBlock</tt> to be appended to. For example code that
2153 <div class="doc_code">
2155 BasicBlock *pb = ...;
2156 Instruction *newInst = new Instruction(...);
2158 pb->getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
2164 <div class="doc_code">
2166 BasicBlock *pb = ...;
2167 Instruction *newInst = new Instruction(..., pb);
2171 <p>which is much cleaner, especially if you are creating
2172 long instruction streams.</p></li>
2174 <li>Insertion into an implicit instruction list
2176 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
2177 are implicitly associated with an existing instruction list: the instruction
2178 list of the enclosing basic block. Thus, we could have accomplished the same
2179 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
2182 <div class="doc_code">
2184 Instruction *pi = ...;
2185 Instruction *newInst = new Instruction(...);
2187 pi->getParent()->getInstList().insert(pi, newInst);
2191 <p>In fact, this sequence of steps occurs so frequently that the
2192 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
2193 constructors which take (as a default parameter) a pointer to an
2194 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
2195 precede. That is, <tt>Instruction</tt> constructors are capable of
2196 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
2197 provided instruction, immediately before that instruction. Using an
2198 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
2199 parameter, the above code becomes:</p>
2201 <div class="doc_code">
2203 Instruction* pi = ...;
2204 Instruction* newInst = new Instruction(..., pi);
2208 <p>which is much cleaner, especially if you're creating a lot of
2209 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
2214 <!--_______________________________________________________________________-->
2215 <div class="doc_subsubsection">
2216 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
2219 <div class="doc_text">
2221 <p>Deleting an instruction from an existing sequence of instructions that form a
2222 <a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward: just
2223 call the instruction's eraseFromParent() method. For example:</p>
2225 <div class="doc_code">
2227 <a href="#Instruction">Instruction</a> *I = .. ;
2228 I->eraseFromParent();
2232 <p>This unlinks the instruction from its containing basic block and deletes
2233 it. If you'd just like to unlink the instruction from its containing basic
2234 block but not delete it, you can use the <tt>removeFromParent()</tt> method.</p>
2238 <!--_______________________________________________________________________-->
2239 <div class="doc_subsubsection">
2240 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
2244 <div class="doc_text">
2246 <p><i>Replacing individual instructions</i></p>
2248 <p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
2249 permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
2250 and <tt>ReplaceInstWithInst</tt>.</p>
2252 <h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
2255 <li><tt>ReplaceInstWithValue</tt>
2257 <p>This function replaces all uses of a given instruction with a value,
2258 and then removes the original instruction. The following example
2259 illustrates the replacement of the result of a particular
2260 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
2261 pointer to an integer.</p>
2263 <div class="doc_code">
2265 AllocaInst* instToReplace = ...;
2266 BasicBlock::iterator ii(instToReplace);
2268 ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
2269 Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
2272 <li><tt>ReplaceInstWithInst</tt>
2274 <p>This function replaces a particular instruction with another
2275 instruction, inserting the new instruction into the basic block at the
2276 location where the old instruction was, and replacing any uses of the old
2277 instruction with the new instruction. The following example illustrates
2278 the replacement of one <tt>AllocaInst</tt> with another.</p>
2280 <div class="doc_code">
2282 AllocaInst* instToReplace = ...;
2283 BasicBlock::iterator ii(instToReplace);
2285 ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
2286 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
2290 <p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
2292 <p>You can use <tt>Value::replaceAllUsesWith</tt> and
2293 <tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
2294 doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
2295 and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
2298 <!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
2299 include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
2300 ReplaceInstWithValue, ReplaceInstWithInst -->
2304 <!--_______________________________________________________________________-->
2305 <div class="doc_subsubsection">
2306 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
2309 <div class="doc_text">
2311 <p>Deleting a global variable from a module is just as easy as deleting an
2312 Instruction. First, you must have a pointer to the global variable that you wish
2313 to delete. You use this pointer to erase it from its parent, the module.
2316 <div class="doc_code">
2318 <a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
2320 GV->eraseFromParent();
2326 <!-- ======================================================================= -->
2327 <div class="doc_subsection">
2328 <a name="create_types">How to Create Types</a>
2331 <div class="doc_text">
2333 <p>In generating IR, you may need some complex types. If you know these types
2334 statically, you can use <tt>TypeBuilder<...>::get()</tt>, defined
2335 in <tt>llvm/Support/TypeBuilder.h</tt>, to retrieve them. <tt>TypeBuilder</tt>
2336 has two forms depending on whether you're building types for cross-compilation
2337 or native library use. <tt>TypeBuilder<T, true></tt> requires
2338 that <tt>T</tt> be independent of the host environment, meaning that it's built
2340 the <a href="/doxygen/namespacellvm_1_1types.html"><tt>llvm::types</tt></a>
2341 namespace and pointers, functions, arrays, etc. built of
2342 those. <tt>TypeBuilder<T, false></tt> additionally allows native C types
2343 whose size may depend on the host compiler. For example,</p>
2345 <div class="doc_code">
2347 FunctionType *ft = TypeBuilder<types::i<8>(types::i<32>*), true>::get();
2351 <p>is easier to read and write than the equivalent</p>
2353 <div class="doc_code">
2355 std::vector<const Type*> params;
2356 params.push_back(PointerType::getUnqual(Type::Int32Ty));
2357 FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false);
2361 <p>See the <a href="/doxygen/TypeBuilder_8h-source.html#l00001">class
2362 comment</a> for more details.</p>
2366 <!-- *********************************************************************** -->
2367 <div class="doc_section">
2368 <a name="threading">Threads and LLVM</a>
2370 <!-- *********************************************************************** -->
2372 <div class="doc_text">
2374 This section describes the interaction of the LLVM APIs with multithreading,
2375 both on the part of client applications, and in the JIT, in the hosted
2380 Note that LLVM's support for multithreading is still relatively young. Up
2381 through version 2.5, the execution of threaded hosted applications was
2382 supported, but not threaded client access to the APIs. While this use case is
2383 now supported, clients <em>must</em> adhere to the guidelines specified below to
2384 ensure proper operation in multithreaded mode.
2388 Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic
2389 intrinsics in order to support threaded operation. If you need a
2390 multhreading-capable LLVM on a platform without a suitably modern system
2391 compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and
2392 using the resultant compiler to build a copy of LLVM with multithreading
2397 <!-- ======================================================================= -->
2398 <div class="doc_subsection">
2399 <a name="startmultithreaded">Entering and Exiting Multithreaded Mode</a>
2402 <div class="doc_text">
2405 In order to properly protect its internal data structures while avoiding
2406 excessive locking overhead in the single-threaded case, the LLVM must intialize
2407 certain data structures necessary to provide guards around its internals. To do
2408 so, the client program must invoke <tt>llvm_start_multithreaded()</tt> before
2409 making any concurrent LLVM API calls. To subsequently tear down these
2410 structures, use the <tt>llvm_stop_multithreaded()</tt> call. You can also use
2411 the <tt>llvm_is_multithreaded()</tt> call to check the status of multithreaded
2416 Note that both of these calls must be made <em>in isolation</em>. That is to
2417 say that no other LLVM API calls may be executing at any time during the
2418 execution of <tt>llvm_start_multithreaded()</tt> or <tt>llvm_stop_multithreaded
2419 </tt>. It's is the client's responsibility to enforce this isolation.
2423 The return value of <tt>llvm_start_multithreaded()</tt> indicates the success or
2424 failure of the initialization. Failure typically indicates that your copy of
2425 LLVM was built without multithreading support, typically because GCC atomic
2426 intrinsics were not found in your system compiler. In this case, the LLVM API
2427 will not be safe for concurrent calls. However, it <em>will</em> be safe for
2428 hosting threaded applications in the JIT, though <a href="#jitthreading">care
2429 must be taken</a> to ensure that side exits and the like do not accidentally
2430 result in concurrent LLVM API calls.
2434 <!-- ======================================================================= -->
2435 <div class="doc_subsection">
2436 <a name="shutdown">Ending Execution with <tt>llvm_shutdown()</tt></a>
2439 <div class="doc_text">
2441 When you are done using the LLVM APIs, you should call <tt>llvm_shutdown()</tt>
2442 to deallocate memory used for internal structures. This will also invoke
2443 <tt>llvm_stop_multithreaded()</tt> if LLVM is operating in multithreaded mode.
2444 As such, <tt>llvm_shutdown()</tt> requires the same isolation guarantees as
2445 <tt>llvm_stop_multithreaded()</tt>.
2449 Note that, if you use scope-based shutdown, you can use the
2450 <tt>llvm_shutdown_obj</tt> class, which calls <tt>llvm_shutdown()</tt> in its
2454 <!-- ======================================================================= -->
2455 <div class="doc_subsection">
2456 <a name="managedstatic">Lazy Initialization with <tt>ManagedStatic</tt></a>
2459 <div class="doc_text">
2461 <tt>ManagedStatic</tt> is a utility class in LLVM used to implement static
2462 initialization of static resources, such as the global type tables. Before the
2463 invocation of <tt>llvm_shutdown()</tt>, it implements a simple lazy
2464 initialization scheme. Once <tt>llvm_start_multithreaded()</tt> returns,
2465 however, it uses double-checked locking to implement thread-safe lazy
2470 Note that, because no other threads are allowed to issue LLVM API calls before
2471 <tt>llvm_start_multithreaded()</tt> returns, it is possible to have
2472 <tt>ManagedStatic</tt>s of <tt>llvm::sys::Mutex</tt>s.
2476 The <tt>llvm_acquire_global_lock()</tt> and <tt>llvm_release_global_lock</tt>
2477 APIs provide access to the global lock used to implement the double-checked
2478 locking for lazy initialization. These should only be used internally to LLVM,
2479 and only if you know what you're doing!
2483 <!-- ======================================================================= -->
2484 <div class="doc_subsection">
2485 <a name="llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a>
2488 <div class="doc_text">
2490 <tt>LLVMContext</tt> is an opaque class in the LLVM API which clients can use
2491 to operate multiple, isolated instances of LLVM concurrently within the same
2492 address space. For instance, in a hypothetical compile-server, the compilation
2493 of an individual translation unit is conceptually independent from all the
2494 others, and it would be desirable to be able to compile incoming translation
2495 units concurrently on independent server threads. Fortunately,
2496 <tt>LLVMContext</tt> exists to enable just this kind of scenario!
2500 Conceptually, <tt>LLVMContext</tt> provides isolation. Every LLVM entity
2501 (<tt>Module</tt>s, <tt>Value</tt>s, <tt>Type</tt>s, <tt>Constant</tt>s, etc.)
2502 in LLVM's in-memory IR belongs to an <tt>LLVMContext</tt>. Entities in
2503 different contexts <em>cannot</em> interact with each other: <tt>Module</tt>s in
2504 different contexts cannot be linked together, <tt>Function</tt>s cannot be added
2505 to <tt>Module</tt>s in different contexts, etc. What this means is that is is
2506 safe to compile on multiple threads simultaneously, as long as no two threads
2507 operate on entities within the same context.
2511 In practice, very few places in the API require the explicit specification of a
2512 <tt>LLVMContext</tt>, other than the <tt>Type</tt> creation/lookup APIs.
2513 Because every <tt>Type</tt> carries a reference to its owning context, most
2514 other entities can determine what context they belong to by looking at their
2515 own <tt>Type</tt>. If you are adding new entities to LLVM IR, please try to
2516 maintain this interface design.
2520 For clients that do <em>not</em> require the benefits of isolation, LLVM
2521 provides a convenience API <tt>getGlobalContext()</tt>. This returns a global,
2522 lazily initialized <tt>LLVMContext</tt> that may be used in situations where
2523 isolation is not a concern.
2527 <!-- ======================================================================= -->
2528 <div class="doc_subsection">
2529 <a name="jitthreading">Threads and the JIT</a>
2532 <div class="doc_text">
2534 LLVM's "eager" JIT compiler is safe to use in threaded programs. Multiple
2535 threads can call <tt>ExecutionEngine::getPointerToFunction()</tt> or
2536 <tt>ExecutionEngine::runFunction()</tt> concurrently, and multiple threads can
2537 run code output by the JIT concurrently. The user must still ensure that only
2538 one thread accesses IR in a given <tt>LLVMContext</tt> while another thread
2539 might be modifying it. One way to do that is to always hold the JIT lock while
2540 accessing IR outside the JIT (the JIT <em>modifies</em> the IR by adding
2541 <tt>CallbackVH</tt>s). Another way is to only
2542 call <tt>getPointerToFunction()</tt> from the <tt>LLVMContext</tt>'s thread.
2545 <p>When the JIT is configured to compile lazily (using
2546 <tt>ExecutionEngine::DisableLazyCompilation(false)</tt>), there is currently a
2547 <a href="http://llvm.org/bugs/show_bug.cgi?id=5184">race condition</a> in
2548 updating call sites after a function is lazily-jitted. It's still possible to
2549 use the lazy JIT in a threaded program if you ensure that only one thread at a
2550 time can call any particular lazy stub and that the JIT lock guards any IR
2551 access, but we suggest using only the eager JIT in threaded programs.
2555 <!-- *********************************************************************** -->
2556 <div class="doc_section">
2557 <a name="advanced">Advanced Topics</a>
2559 <!-- *********************************************************************** -->
2561 <div class="doc_text">
2563 This section describes some of the advanced or obscure API's that most clients
2564 do not need to be aware of. These API's tend manage the inner workings of the
2565 LLVM system, and only need to be accessed in unusual circumstances.
2569 <!-- ======================================================================= -->
2570 <div class="doc_subsection">
2571 <a name="TypeResolve">LLVM Type Resolution</a>
2574 <div class="doc_text">
2577 The LLVM type system has a very simple goal: allow clients to compare types for
2578 structural equality with a simple pointer comparison (aka a shallow compare).
2579 This goal makes clients much simpler and faster, and is used throughout the LLVM
2584 Unfortunately achieving this goal is not a simple matter. In particular,
2585 recursive types and late resolution of opaque types makes the situation very
2586 difficult to handle. Fortunately, for the most part, our implementation makes
2587 most clients able to be completely unaware of the nasty internal details. The
2588 primary case where clients are exposed to the inner workings of it are when
2589 building a recursive type. In addition to this case, the LLVM bitcode reader,
2590 assembly parser, and linker also have to be aware of the inner workings of this
2595 For our purposes below, we need three concepts. First, an "Opaque Type" is
2596 exactly as defined in the <a href="LangRef.html#t_opaque">language
2597 reference</a>. Second an "Abstract Type" is any type which includes an
2598 opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
2599 Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
2605 <!-- ______________________________________________________________________ -->
2606 <div class="doc_subsubsection">
2607 <a name="BuildRecType">Basic Recursive Type Construction</a>
2610 <div class="doc_text">
2613 Because the most common question is "how do I build a recursive type with LLVM",
2614 we answer it now and explain it as we go. Here we include enough to cause this
2615 to be emitted to an output .ll file:
2618 <div class="doc_code">
2620 %mylist = type { %mylist*, i32 }
2625 To build this, use the following LLVM APIs:
2628 <div class="doc_code">
2630 // <i>Create the initial outer struct</i>
2631 <a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
2632 std::vector<const Type*> Elts;
2633 Elts.push_back(PointerType::getUnqual(StructTy));
2634 Elts.push_back(Type::Int32Ty);
2635 StructType *NewSTy = StructType::get(Elts);
2637 // <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
2638 // <i>the struct and the opaque type are actually the same.</i>
2639 cast<OpaqueType>(StructTy.get())-><a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
2641 // <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
2642 // <i>kept up-to-date</i>
2643 NewSTy = cast<StructType>(StructTy.get());
2645 // <i>Add a name for the type to the module symbol table (optional)</i>
2646 MyModule->addTypeName("mylist", NewSTy);
2651 This code shows the basic approach used to build recursive types: build a
2652 non-recursive type using 'opaque', then use type unification to close the cycle.
2653 The type unification step is performed by the <tt><a
2654 href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
2655 described next. After that, we describe the <a
2656 href="#PATypeHolder">PATypeHolder class</a>.
2661 <!-- ______________________________________________________________________ -->
2662 <div class="doc_subsubsection">
2663 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
2666 <div class="doc_text">
2668 The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
2669 While this method is actually a member of the DerivedType class, it is most
2670 often used on OpaqueType instances. Type unification is actually a recursive
2671 process. After unification, types can become structurally isomorphic to
2672 existing types, and all duplicates are deleted (to preserve pointer equality).
2676 In the example above, the OpaqueType object is definitely deleted.
2677 Additionally, if there is an "{ \2*, i32}" type already created in the system,
2678 the pointer and struct type created are <b>also</b> deleted. Obviously whenever
2679 a type is deleted, any "Type*" pointers in the program are invalidated. As
2680 such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
2681 live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
2682 types can never move or be deleted). To deal with this, the <a
2683 href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
2684 reference to a possibly refined type, and the <a
2685 href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
2686 complex datastructures.
2691 <!-- ______________________________________________________________________ -->
2692 <div class="doc_subsubsection">
2693 <a name="PATypeHolder">The PATypeHolder Class</a>
2696 <div class="doc_text">
2698 PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
2699 happily goes about nuking types that become isomorphic to existing types, it
2700 automatically updates all PATypeHolder objects to point to the new type. In the
2701 example above, this allows the code to maintain a pointer to the resultant
2702 resolved recursive type, even though the Type*'s are potentially invalidated.
2706 PATypeHolder is an extremely light-weight object that uses a lazy union-find
2707 implementation to update pointers. For example the pointer from a Value to its
2708 Type is maintained by PATypeHolder objects.
2713 <!-- ______________________________________________________________________ -->
2714 <div class="doc_subsubsection">
2715 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
2718 <div class="doc_text">
2721 Some data structures need more to perform more complex updates when types get
2722 resolved. To support this, a class can derive from the AbstractTypeUser class.
2724 allows it to get callbacks when certain types are resolved. To register to get
2725 callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
2726 methods can be called on a type. Note that these methods only work for <i>
2727 abstract</i> types. Concrete types (those that do not include any opaque
2728 objects) can never be refined.
2733 <!-- ======================================================================= -->
2734 <div class="doc_subsection">
2735 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
2736 <tt>TypeSymbolTable</tt> classes</a>
2739 <div class="doc_text">
2740 <p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2741 ValueSymbolTable</a></tt> class provides a symbol table that the <a
2742 href="#Function"><tt>Function</tt></a> and <a href="#Module">
2743 <tt>Module</tt></a> classes use for naming value definitions. The symbol table
2744 can provide a name for any <a href="#Value"><tt>Value</tt></a>.
2745 The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
2746 TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
2747 names for types.</p>
2749 <p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2750 by most clients. It should only be used when iteration over the symbol table
2751 names themselves are required, which is very special purpose. Note that not
2753 <tt><a href="#Value">Value</a></tt>s have names, and those without names (i.e. they have
2754 an empty name) do not exist in the symbol table.
2757 <p>These symbol tables support iteration over the values/types in the symbol
2758 table with <tt>begin/end/iterator</tt> and supports querying to see if a
2759 specific name is in the symbol table (with <tt>lookup</tt>). The
2760 <tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
2761 simply call <tt>setName</tt> on a value, which will autoinsert it into the
2762 appropriate symbol table. For types, use the Module::addTypeName method to
2763 insert entries into the symbol table.</p>
2769 <!-- ======================================================================= -->
2770 <div class="doc_subsection">
2771 <a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
2774 <div class="doc_text">
2775 <p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1User.html">
2776 User</a></tt> class provides a basis for expressing the ownership of <tt>User</tt>
2777 towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
2778 Value</a></tt>s. The <tt><a href="http://llvm.org/doxygen/classllvm_1_1Use.html">
2779 Use</a></tt> helper class is employed to do the bookkeeping and to facilitate <i>O(1)</i>
2780 addition and removal.</p>
2782 <!-- ______________________________________________________________________ -->
2783 <div class="doc_subsubsection">
2784 <a name="Use2User">Interaction and relationship between <tt>User</tt> and <tt>Use</tt> objects</a>
2787 <div class="doc_text">
2789 A subclass of <tt>User</tt> can choose between incorporating its <tt>Use</tt> objects
2790 or refer to them out-of-line by means of a pointer. A mixed variant
2791 (some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
2792 that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
2797 We have 2 different layouts in the <tt>User</tt> (sub)classes:
2800 The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
2801 object and there are a fixed number of them.</p>
2804 The <tt>Use</tt> object(s) are referenced by a pointer to an
2805 array from the <tt>User</tt> object and there may be a variable
2809 As of v2.4 each layout still possesses a direct pointer to the
2810 start of the array of <tt>Use</tt>s. Though not mandatory for layout a),
2811 we stick to this redundancy for the sake of simplicity.
2812 The <tt>User</tt> object also stores the number of <tt>Use</tt> objects it
2813 has. (Theoretically this information can also be calculated
2814 given the scheme presented below.)</p>
2816 Special forms of allocation operators (<tt>operator new</tt>)
2817 enforce the following memory layouts:</p>
2820 <li><p>Layout a) is modelled by prepending the <tt>User</tt> object by the <tt>Use[]</tt> array.</p>
2823 ...---.---.---.---.-------...
2824 | P | P | P | P | User
2825 '''---'---'---'---'-------'''
2828 <li><p>Layout b) is modelled by pointing at the <tt>Use[]</tt> array.</p>
2840 <i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
2841 is stored in each <tt>Use</tt> object in the member <tt>Use::Prev</tt>)</i>
2843 <!-- ______________________________________________________________________ -->
2844 <div class="doc_subsubsection">
2845 <a name="Waymarking">The waymarking algorithm</a>
2848 <div class="doc_text">
2850 Since the <tt>Use</tt> objects are deprived of the direct (back)pointer to
2851 their <tt>User</tt> objects, there must be a fast and exact method to
2852 recover it. This is accomplished by the following scheme:</p>
2855 A bit-encoding in the 2 LSBits (least significant bits) of the <tt>Use::Prev</tt> allows to find the
2856 start of the <tt>User</tt> object:
2858 <li><tt>00</tt> —> binary digit 0</li>
2859 <li><tt>01</tt> —> binary digit 1</li>
2860 <li><tt>10</tt> —> stop and calculate (<tt>s</tt>)</li>
2861 <li><tt>11</tt> —> full stop (<tt>S</tt>)</li>
2864 Given a <tt>Use*</tt>, all we have to do is to walk till we get
2865 a stop and we either have a <tt>User</tt> immediately behind or
2866 we have to walk to the next stop picking up digits
2867 and calculating the offset:</p>
2869 .---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
2870 | 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*)
2871 '---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
2872 |+15 |+10 |+6 |+3 |+1
2875 | | |______________________>
2876 | |______________________________________>
2877 |__________________________________________________________>
2880 Only the significant number of bits need to be stored between the
2881 stops, so that the <i>worst case is 20 memory accesses</i> when there are
2882 1000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
2884 <!-- ______________________________________________________________________ -->
2885 <div class="doc_subsubsection">
2886 <a name="ReferenceImpl">Reference implementation</a>
2889 <div class="doc_text">
2891 The following literate Haskell fragment demonstrates the concept:</p>
2894 <div class="doc_code">
2896 > import Test.QuickCheck
2898 > digits :: Int -> [Char] -> [Char]
2899 > digits 0 acc = '0' : acc
2900 > digits 1 acc = '1' : acc
2901 > digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
2903 > dist :: Int -> [Char] -> [Char]
2906 > dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
2907 > dist n acc = dist (n - 1) $ dist 1 acc
2909 > takeLast n ss = reverse $ take n $ reverse ss
2911 > test = takeLast 40 $ dist 20 []
2916 Printing <test> gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
2918 The reverse algorithm computes the length of the string just by examining
2919 a certain prefix:</p>
2921 <div class="doc_code">
2923 > pref :: [Char] -> Int
2925 > pref ('s':'1':rest) = decode 2 1 rest
2926 > pref (_:rest) = 1 + pref rest
2928 > decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
2929 > decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
2930 > decode walk acc _ = walk + acc
2935 Now, as expected, printing <pref test> gives <tt>40</tt>.</p>
2937 We can <i>quickCheck</i> this with following property:</p>
2939 <div class="doc_code">
2941 > testcase = dist 2000 []
2942 > testcaseLength = length testcase
2944 > identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
2945 > where arr = takeLast n testcase
2950 As expected <quickCheck identityProp> gives:</p>
2953 *Main> quickCheck identityProp
2954 OK, passed 100 tests.
2957 Let's be a bit more exhaustive:</p>
2959 <div class="doc_code">
2962 > deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
2967 And here is the result of <deepCheck identityProp>:</p>
2970 *Main> deepCheck identityProp
2971 OK, passed 500 tests.
2974 <!-- ______________________________________________________________________ -->
2975 <div class="doc_subsubsection">
2976 <a name="Tagging">Tagging considerations</a>
2980 To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
2981 never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
2982 new <tt>Use**</tt> on every modification. Accordingly getters must strip the
2985 For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
2986 Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
2987 that the first bytes of <tt>User</tt> (if interpreted as a pointer) never has
2988 the LSBit set. (Portability is relying on the fact that all known compilers place the
2989 <tt>vptr</tt> in the first word of the instances.)</p>
2993 <!-- *********************************************************************** -->
2994 <div class="doc_section">
2995 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
2997 <!-- *********************************************************************** -->
2999 <div class="doc_text">
3000 <p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
3001 <br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
3003 <p>The Core LLVM classes are the primary means of representing the program
3004 being inspected or transformed. The core LLVM classes are defined in
3005 header files in the <tt>include/llvm/</tt> directory, and implemented in
3006 the <tt>lib/VMCore</tt> directory.</p>
3010 <!-- ======================================================================= -->
3011 <div class="doc_subsection">
3012 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
3015 <div class="doc_text">
3017 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
3018 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
3019 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
3020 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
3021 subclasses. They are hidden because they offer no useful functionality beyond
3022 what the <tt>Type</tt> class offers except to distinguish themselves from
3023 other subclasses of <tt>Type</tt>.</p>
3024 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
3025 named, but this is not a requirement. There exists exactly
3026 one instance of a given shape at any one time. This allows type equality to
3027 be performed with address equality of the Type Instance. That is, given two
3028 <tt>Type*</tt> values, the types are identical if the pointers are identical.
3032 <!-- _______________________________________________________________________ -->
3033 <div class="doc_subsubsection">
3034 <a name="m_Type">Important Public Methods</a>
3037 <div class="doc_text">
3040 <li><tt>bool isIntegerTy() const</tt>: Returns true for any integer type.</li>
3042 <li><tt>bool isFloatingPointTy()</tt>: Return true if this is one of the five
3043 floating point types.</li>
3045 <li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
3046 an OpaqueType anywhere in its definition).</li>
3048 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
3049 that don't have a size are abstract types, labels and void.</li>
3054 <!-- _______________________________________________________________________ -->
3055 <div class="doc_subsubsection">
3056 <a name="derivedtypes">Important Derived Types</a>
3058 <div class="doc_text">
3060 <dt><tt>IntegerType</tt></dt>
3061 <dd>Subclass of DerivedType that represents integer types of any bit width.
3062 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
3063 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
3065 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
3066 type of a specific bit width.</li>
3067 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
3071 <dt><tt>SequentialType</tt></dt>
3072 <dd>This is subclassed by ArrayType and PointerType
3074 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
3075 of the elements in the sequential type. </li>
3078 <dt><tt>ArrayType</tt></dt>
3079 <dd>This is a subclass of SequentialType and defines the interface for array
3082 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
3083 elements in the array. </li>
3086 <dt><tt>PointerType</tt></dt>
3087 <dd>Subclass of SequentialType for pointer types.</dd>
3088 <dt><tt>VectorType</tt></dt>
3089 <dd>Subclass of SequentialType for vector types. A
3090 vector type is similar to an ArrayType but is distinguished because it is
3091 a first class type whereas ArrayType is not. Vector types are used for
3092 vector operations and are usually small vectors of of an integer or floating
3094 <dt><tt>StructType</tt></dt>
3095 <dd>Subclass of DerivedTypes for struct types.</dd>
3096 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
3097 <dd>Subclass of DerivedTypes for function types.
3099 <li><tt>bool isVarArg() const</tt>: Returns true if it's a vararg
3101 <li><tt> const Type * getReturnType() const</tt>: Returns the
3102 return type of the function.</li>
3103 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
3104 the type of the ith parameter.</li>
3105 <li><tt> const unsigned getNumParams() const</tt>: Returns the
3106 number of formal parameters.</li>
3109 <dt><tt>OpaqueType</tt></dt>
3110 <dd>Sublcass of DerivedType for abstract types. This class
3111 defines no content and is used as a placeholder for some other type. Note
3112 that OpaqueType is used (temporarily) during type resolution for forward
3113 references of types. Once the referenced type is resolved, the OpaqueType
3114 is replaced with the actual type. OpaqueType can also be used for data
3115 abstraction. At link time opaque types can be resolved to actual types
3116 of the same name.</dd>
3122 <!-- ======================================================================= -->
3123 <div class="doc_subsection">
3124 <a name="Module">The <tt>Module</tt> class</a>
3127 <div class="doc_text">
3130 href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
3131 <a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
3133 <p>The <tt>Module</tt> class represents the top level structure present in LLVM
3134 programs. An LLVM module is effectively either a translation unit of the
3135 original program or a combination of several translation units merged by the
3136 linker. The <tt>Module</tt> class keeps track of a list of <a
3137 href="#Function"><tt>Function</tt></a>s, a list of <a
3138 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
3139 href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
3140 helpful member functions that try to make common operations easy.</p>
3144 <!-- _______________________________________________________________________ -->
3145 <div class="doc_subsubsection">
3146 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
3149 <div class="doc_text">
3152 <li><tt>Module::Module(std::string name = "")</tt></li>
3155 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
3156 provide a name for it (probably based on the name of the translation unit).</p>
3159 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
3160 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
3162 <tt>begin()</tt>, <tt>end()</tt>
3163 <tt>size()</tt>, <tt>empty()</tt>
3165 <p>These are forwarding methods that make it easy to access the contents of
3166 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
3169 <li><tt>Module::FunctionListType &getFunctionList()</tt>
3171 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
3172 necessary to use when you need to update the list or perform a complex
3173 action that doesn't have a forwarding method.</p>
3175 <p><!-- Global Variable --></p></li>
3181 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
3183 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
3185 <tt>global_begin()</tt>, <tt>global_end()</tt>
3186 <tt>global_size()</tt>, <tt>global_empty()</tt>
3188 <p> These are forwarding methods that make it easy to access the contents of
3189 a <tt>Module</tt> object's <a
3190 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
3192 <li><tt>Module::GlobalListType &getGlobalList()</tt>
3194 <p>Returns the list of <a
3195 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
3196 use when you need to update the list or perform a complex action that
3197 doesn't have a forwarding method.</p>
3199 <p><!-- Symbol table stuff --> </p></li>
3205 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
3207 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3208 for this <tt>Module</tt>.</p>
3210 <p><!-- Convenience methods --></p></li>
3216 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
3217 &Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
3219 <p>Look up the specified function in the <tt>Module</tt> <a
3220 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
3221 <tt>null</tt>.</p></li>
3223 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
3224 std::string &Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
3226 <p>Look up the specified function in the <tt>Module</tt> <a
3227 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
3228 external declaration for the function and return it.</p></li>
3230 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
3232 <p>If there is at least one entry in the <a
3233 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
3234 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
3237 <li><tt>bool addTypeName(const std::string &Name, const <a
3238 href="#Type">Type</a> *Ty)</tt>
3240 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3241 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
3242 name, true is returned and the <a
3243 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
3249 <!-- ======================================================================= -->
3250 <div class="doc_subsection">
3251 <a name="Value">The <tt>Value</tt> class</a>
3254 <div class="doc_text">
3256 <p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
3258 doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
3260 <p>The <tt>Value</tt> class is the most important class in the LLVM Source
3261 base. It represents a typed value that may be used (among other things) as an
3262 operand to an instruction. There are many different types of <tt>Value</tt>s,
3263 such as <a href="#Constant"><tt>Constant</tt></a>s,<a
3264 href="#Argument"><tt>Argument</tt></a>s. Even <a
3265 href="#Instruction"><tt>Instruction</tt></a>s and <a
3266 href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
3268 <p>A particular <tt>Value</tt> may be used many times in the LLVM representation
3269 for a program. For example, an incoming argument to a function (represented
3270 with an instance of the <a href="#Argument">Argument</a> class) is "used" by
3271 every instruction in the function that references the argument. To keep track
3272 of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
3273 href="#User"><tt>User</tt></a>s that is using it (the <a
3274 href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
3275 graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
3276 def-use information in the program, and is accessible through the <tt>use_</tt>*
3277 methods, shown below.</p>
3279 <p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
3280 and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
3281 method. In addition, all LLVM values can be named. The "name" of the
3282 <tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
3284 <div class="doc_code">
3286 %<b>foo</b> = add i32 1, 2
3290 <p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
3291 that the name of any value may be missing (an empty string), so names should
3292 <b>ONLY</b> be used for debugging (making the source code easier to read,
3293 debugging printouts), they should not be used to keep track of values or map
3294 between them. For this purpose, use a <tt>std::map</tt> of pointers to the
3295 <tt>Value</tt> itself instead.</p>
3297 <p>One important aspect of LLVM is that there is no distinction between an SSA
3298 variable and the operation that produces it. Because of this, any reference to
3299 the value produced by an instruction (or the value available as an incoming
3300 argument, for example) is represented as a direct pointer to the instance of
3302 represents this value. Although this may take some getting used to, it
3303 simplifies the representation and makes it easier to manipulate.</p>
3307 <!-- _______________________________________________________________________ -->
3308 <div class="doc_subsubsection">
3309 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
3312 <div class="doc_text">
3315 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
3317 <tt>Value::const_use_iterator</tt> - Typedef for const_iterator over
3319 <tt>unsigned use_size()</tt> - Returns the number of users of the
3321 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
3322 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
3324 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
3326 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
3327 element in the list.
3328 <p> These methods are the interface to access the def-use
3329 information in LLVM. As with all other iterators in LLVM, the naming
3330 conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
3332 <li><tt><a href="#Type">Type</a> *getType() const</tt>
3333 <p>This method returns the Type of the Value.</p>
3335 <li><tt>bool hasName() const</tt><br>
3336 <tt>std::string getName() const</tt><br>
3337 <tt>void setName(const std::string &Name)</tt>
3338 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
3339 be aware of the <a href="#nameWarning">precaution above</a>.</p>
3341 <li><tt>void replaceAllUsesWith(Value *V)</tt>
3343 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
3344 href="#User"><tt>User</tt>s</a> of the current value to refer to
3345 "<tt>V</tt>" instead. For example, if you detect that an instruction always
3346 produces a constant value (for example through constant folding), you can
3347 replace all uses of the instruction with the constant like this:</p>
3349 <div class="doc_code">
3351 Inst->replaceAllUsesWith(ConstVal);
3359 <!-- ======================================================================= -->
3360 <div class="doc_subsection">
3361 <a name="User">The <tt>User</tt> class</a>
3364 <div class="doc_text">
3367 <tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
3368 doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
3369 Superclass: <a href="#Value"><tt>Value</tt></a></p>
3371 <p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
3372 refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
3373 that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
3374 referring to. The <tt>User</tt> class itself is a subclass of
3377 <p>The operands of a <tt>User</tt> point directly to the LLVM <a
3378 href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
3379 Single Assignment (SSA) form, there can only be one definition referred to,
3380 allowing this direct connection. This connection provides the use-def
3381 information in LLVM.</p>
3385 <!-- _______________________________________________________________________ -->
3386 <div class="doc_subsubsection">
3387 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
3390 <div class="doc_text">
3392 <p>The <tt>User</tt> class exposes the operand list in two ways: through
3393 an index access interface and through an iterator based interface.</p>
3396 <li><tt>Value *getOperand(unsigned i)</tt><br>
3397 <tt>unsigned getNumOperands()</tt>
3398 <p> These two methods expose the operands of the <tt>User</tt> in a
3399 convenient form for direct access.</p></li>
3401 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
3403 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
3404 the operand list.<br>
3405 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
3407 <p> Together, these methods make up the iterator based interface to
3408 the operands of a <tt>User</tt>.</p></li>
3413 <!-- ======================================================================= -->
3414 <div class="doc_subsection">
3415 <a name="Instruction">The <tt>Instruction</tt> class</a>
3418 <div class="doc_text">
3420 <p><tt>#include "</tt><tt><a
3421 href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
3422 doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
3423 Superclasses: <a href="#User"><tt>User</tt></a>, <a
3424 href="#Value"><tt>Value</tt></a></p>
3426 <p>The <tt>Instruction</tt> class is the common base class for all LLVM
3427 instructions. It provides only a few methods, but is a very commonly used
3428 class. The primary data tracked by the <tt>Instruction</tt> class itself is the
3429 opcode (instruction type) and the parent <a
3430 href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
3431 into. To represent a specific type of instruction, one of many subclasses of
3432 <tt>Instruction</tt> are used.</p>
3434 <p> Because the <tt>Instruction</tt> class subclasses the <a
3435 href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
3436 way as for other <a href="#User"><tt>User</tt></a>s (with the
3437 <tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
3438 <tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
3439 the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
3440 file contains some meta-data about the various different types of instructions
3441 in LLVM. It describes the enum values that are used as opcodes (for example
3442 <tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
3443 concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
3444 example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
3445 href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
3446 this file confuses doxygen, so these enum values don't show up correctly in the
3447 <a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
3451 <!-- _______________________________________________________________________ -->
3452 <div class="doc_subsubsection">
3453 <a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
3456 <div class="doc_text">
3458 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
3459 <p>This subclasses represents all two operand instructions whose operands
3460 must be the same type, except for the comparison instructions.</p></li>
3461 <li><tt><a name="CastInst">CastInst</a></tt>
3462 <p>This subclass is the parent of the 12 casting instructions. It provides
3463 common operations on cast instructions.</p>
3464 <li><tt><a name="CmpInst">CmpInst</a></tt>
3465 <p>This subclass respresents the two comparison instructions,
3466 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
3467 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
3468 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
3469 <p>This subclass is the parent of all terminator instructions (those which
3470 can terminate a block).</p>
3474 <!-- _______________________________________________________________________ -->
3475 <div class="doc_subsubsection">
3476 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
3480 <div class="doc_text">
3483 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
3484 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
3485 this <tt>Instruction</tt> is embedded into.</p></li>
3486 <li><tt>bool mayWriteToMemory()</tt>
3487 <p>Returns true if the instruction writes to memory, i.e. it is a
3488 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
3489 <li><tt>unsigned getOpcode()</tt>
3490 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
3491 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
3492 <p>Returns another instance of the specified instruction, identical
3493 in all ways to the original except that the instruction has no parent
3494 (ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
3495 and it has no name</p></li>
3500 <!-- ======================================================================= -->
3501 <div class="doc_subsection">
3502 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
3505 <div class="doc_text">
3507 <p>Constant represents a base class for different types of constants. It
3508 is subclassed by ConstantInt, ConstantArray, etc. for representing
3509 the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
3510 a subclass, which represents the address of a global variable or function.
3515 <!-- _______________________________________________________________________ -->
3516 <div class="doc_subsubsection">Important Subclasses of Constant </div>
3517 <div class="doc_text">
3519 <li>ConstantInt : This subclass of Constant represents an integer constant of
3522 <li><tt>const APInt& getValue() const</tt>: Returns the underlying
3523 value of this constant, an APInt value.</li>
3524 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
3525 value to an int64_t via sign extension. If the value (not the bit width)
3526 of the APInt is too large to fit in an int64_t, an assertion will result.
3527 For this reason, use of this method is discouraged.</li>
3528 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
3529 value to a uint64_t via zero extension. IF the value (not the bit width)
3530 of the APInt is too large to fit in a uint64_t, an assertion will result.
3531 For this reason, use of this method is discouraged.</li>
3532 <li><tt>static ConstantInt* get(const APInt& Val)</tt>: Returns the
3533 ConstantInt object that represents the value provided by <tt>Val</tt>.
3534 The type is implied as the IntegerType that corresponds to the bit width
3535 of <tt>Val</tt>.</li>
3536 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
3537 Returns the ConstantInt object that represents the value provided by
3538 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
3541 <li>ConstantFP : This class represents a floating point constant.
3543 <li><tt>double getValue() const</tt>: Returns the underlying value of
3544 this constant. </li>
3547 <li>ConstantArray : This represents a constant array.
3549 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
3550 a vector of component constants that makeup this array. </li>
3553 <li>ConstantStruct : This represents a constant struct.
3555 <li><tt>const std::vector<Use> &getValues() const</tt>: Returns
3556 a vector of component constants that makeup this array. </li>
3559 <li>GlobalValue : This represents either a global variable or a function. In
3560 either case, the value is a constant fixed address (after linking).
3566 <!-- ======================================================================= -->
3567 <div class="doc_subsection">
3568 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
3571 <div class="doc_text">
3574 href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
3575 doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
3577 Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
3578 <a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
3580 <p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
3581 href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
3582 visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
3583 Because they are visible at global scope, they are also subject to linking with
3584 other globals defined in different translation units. To control the linking
3585 process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
3586 <tt>GlobalValue</tt>s know whether they have internal or external linkage, as
3587 defined by the <tt>LinkageTypes</tt> enumeration.</p>
3589 <p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
3590 <tt>static</tt> in C), it is not visible to code outside the current translation
3591 unit, and does not participate in linking. If it has external linkage, it is
3592 visible to external code, and does participate in linking. In addition to
3593 linkage information, <tt>GlobalValue</tt>s keep track of which <a
3594 href="#Module"><tt>Module</tt></a> they are currently part of.</p>
3596 <p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
3597 by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
3598 global is always a pointer to its contents. It is important to remember this
3599 when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
3600 be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
3601 subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
3602 i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
3603 the address of the first element of this array and the value of the
3604 <tt>GlobalVariable</tt> are the same, they have different types. The
3605 <tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
3606 is <tt>i32.</tt> Because of this, accessing a global value requires you to
3607 dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
3608 can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
3609 Language Reference Manual</a>.</p>
3613 <!-- _______________________________________________________________________ -->
3614 <div class="doc_subsubsection">
3615 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
3619 <div class="doc_text">
3622 <li><tt>bool hasInternalLinkage() const</tt><br>
3623 <tt>bool hasExternalLinkage() const</tt><br>
3624 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
3625 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
3628 <li><tt><a href="#Module">Module</a> *getParent()</tt>
3629 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
3630 GlobalValue is currently embedded into.</p></li>
3635 <!-- ======================================================================= -->
3636 <div class="doc_subsection">
3637 <a name="Function">The <tt>Function</tt> class</a>
3640 <div class="doc_text">
3643 href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
3644 info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
3645 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3646 <a href="#Constant"><tt>Constant</tt></a>,
3647 <a href="#User"><tt>User</tt></a>,
3648 <a href="#Value"><tt>Value</tt></a></p>
3650 <p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
3651 actually one of the more complex classes in the LLVM hierarchy because it must
3652 keep track of a large amount of data. The <tt>Function</tt> class keeps track
3653 of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
3654 <a href="#Argument"><tt>Argument</tt></a>s, and a
3655 <a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
3657 <p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
3658 commonly used part of <tt>Function</tt> objects. The list imposes an implicit
3659 ordering of the blocks in the function, which indicate how the code will be
3660 laid out by the backend. Additionally, the first <a
3661 href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
3662 <tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
3663 block. There are no implicit exit nodes, and in fact there may be multiple exit
3664 nodes from a single <tt>Function</tt>. If the <a
3665 href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
3666 the <tt>Function</tt> is actually a function declaration: the actual body of the
3667 function hasn't been linked in yet.</p>
3669 <p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
3670 <tt>Function</tt> class also keeps track of the list of formal <a
3671 href="#Argument"><tt>Argument</tt></a>s that the function receives. This
3672 container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
3673 nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
3674 the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
3676 <p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
3677 LLVM feature that is only used when you have to look up a value by name. Aside
3678 from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
3679 internally to make sure that there are not conflicts between the names of <a
3680 href="#Instruction"><tt>Instruction</tt></a>s, <a
3681 href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
3682 href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
3684 <p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
3685 and therefore also a <a href="#Constant">Constant</a>. The value of the function
3686 is its address (after linking) which is guaranteed to be constant.</p>
3689 <!-- _______________________________________________________________________ -->
3690 <div class="doc_subsubsection">
3691 <a name="m_Function">Important Public Members of the <tt>Function</tt>
3695 <div class="doc_text">
3698 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
3699 *Ty, LinkageTypes Linkage, const std::string &N = "", Module* Parent = 0)</tt>
3701 <p>Constructor used when you need to create new <tt>Function</tt>s to add
3702 the the program. The constructor must specify the type of the function to
3703 create and what type of linkage the function should have. The <a
3704 href="#FunctionType"><tt>FunctionType</tt></a> argument
3705 specifies the formal arguments and return value for the function. The same
3706 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
3707 create multiple functions. The <tt>Parent</tt> argument specifies the Module
3708 in which the function is defined. If this argument is provided, the function
3709 will automatically be inserted into that module's list of
3712 <li><tt>bool isDeclaration()</tt>
3714 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
3715 function is "external", it does not have a body, and thus must be resolved
3716 by linking with a function defined in a different translation unit.</p></li>
3718 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
3719 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
3721 <tt>begin()</tt>, <tt>end()</tt>
3722 <tt>size()</tt>, <tt>empty()</tt>
3724 <p>These are forwarding methods that make it easy to access the contents of
3725 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
3728 <li><tt>Function::BasicBlockListType &getBasicBlockList()</tt>
3730 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
3731 is necessary to use when you need to update the list or perform a complex
3732 action that doesn't have a forwarding method.</p></li>
3734 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
3736 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
3738 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
3739 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
3741 <p>These are forwarding methods that make it easy to access the contents of
3742 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
3745 <li><tt>Function::ArgumentListType &getArgumentList()</tt>
3747 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
3748 necessary to use when you need to update the list or perform a complex
3749 action that doesn't have a forwarding method.</p></li>
3751 <li><tt><a href="#BasicBlock">BasicBlock</a> &getEntryBlock()</tt>
3753 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
3754 function. Because the entry block for the function is always the first
3755 block, this returns the first block of the <tt>Function</tt>.</p></li>
3757 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
3758 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
3760 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
3761 <tt>Function</tt> and returns the return type of the function, or the <a
3762 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
3765 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
3767 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3768 for this <tt>Function</tt>.</p></li>
3773 <!-- ======================================================================= -->
3774 <div class="doc_subsection">
3775 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
3778 <div class="doc_text">
3781 href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
3783 doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
3785 Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3786 <a href="#Constant"><tt>Constant</tt></a>,
3787 <a href="#User"><tt>User</tt></a>,
3788 <a href="#Value"><tt>Value</tt></a></p>
3790 <p>Global variables are represented with the (surprise surprise)
3791 <tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
3792 subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
3793 always referenced by their address (global values must live in memory, so their
3794 "name" refers to their constant address). See
3795 <a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
3796 variables may have an initial value (which must be a
3797 <a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
3798 they may be marked as "constant" themselves (indicating that their contents
3799 never change at runtime).</p>
3802 <!-- _______________________________________________________________________ -->
3803 <div class="doc_subsubsection">
3804 <a name="m_GlobalVariable">Important Public Members of the
3805 <tt>GlobalVariable</tt> class</a>
3808 <div class="doc_text">
3811 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
3812 isConstant, LinkageTypes& Linkage, <a href="#Constant">Constant</a>
3813 *Initializer = 0, const std::string &Name = "", Module* Parent = 0)</tt>
3815 <p>Create a new global variable of the specified type. If
3816 <tt>isConstant</tt> is true then the global variable will be marked as
3817 unchanging for the program. The Linkage parameter specifies the type of
3818 linkage (internal, external, weak, linkonce, appending) for the variable.
3819 If the linkage is InternalLinkage, WeakAnyLinkage, WeakODRLinkage,
3820 LinkOnceAnyLinkage or LinkOnceODRLinkage, then the resultant
3821 global variable will have internal linkage. AppendingLinkage concatenates
3822 together all instances (in different translation units) of the variable
3823 into a single variable but is only applicable to arrays. See
3824 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
3825 further details on linkage types. Optionally an initializer, a name, and the
3826 module to put the variable into may be specified for the global variable as
3829 <li><tt>bool isConstant() const</tt>
3831 <p>Returns true if this is a global variable that is known not to
3832 be modified at runtime.</p></li>
3834 <li><tt>bool hasInitializer()</tt>
3836 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
3838 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
3840 <p>Returns the initial value for a <tt>GlobalVariable</tt>. It is not legal
3841 to call this method if there is no initializer.</p></li>
3847 <!-- ======================================================================= -->
3848 <div class="doc_subsection">
3849 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
3852 <div class="doc_text">
3855 href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
3856 doxygen info: <a href="/doxygen/classllvm_1_1BasicBlock.html">BasicBlock
3858 Superclass: <a href="#Value"><tt>Value</tt></a></p>
3860 <p>This class represents a single entry single exit section of the code,
3861 commonly known as a basic block by the compiler community. The
3862 <tt>BasicBlock</tt> class maintains a list of <a
3863 href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
3864 Matching the language definition, the last element of this list of instructions
3865 is always a terminator instruction (a subclass of the <a
3866 href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
3868 <p>In addition to tracking the list of instructions that make up the block, the
3869 <tt>BasicBlock</tt> class also keeps track of the <a
3870 href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
3872 <p>Note that <tt>BasicBlock</tt>s themselves are <a
3873 href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
3874 like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
3879 <!-- _______________________________________________________________________ -->
3880 <div class="doc_subsubsection">
3881 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
3885 <div class="doc_text">
3888 <li><tt>BasicBlock(const std::string &Name = "", </tt><tt><a
3889 href="#Function">Function</a> *Parent = 0)</tt>
3891 <p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
3892 insertion into a function. The constructor optionally takes a name for the new
3893 block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
3894 the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
3895 automatically inserted at the end of the specified <a
3896 href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
3897 manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
3899 <li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
3900 <tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
3901 <tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
3902 <tt>size()</tt>, <tt>empty()</tt>
3903 STL-style functions for accessing the instruction list.
3905 <p>These methods and typedefs are forwarding functions that have the same
3906 semantics as the standard library methods of the same names. These methods
3907 expose the underlying instruction list of a basic block in a way that is easy to
3908 manipulate. To get the full complement of container operations (including
3909 operations to update the list), you must use the <tt>getInstList()</tt>
3912 <li><tt>BasicBlock::InstListType &getInstList()</tt>
3914 <p>This method is used to get access to the underlying container that actually
3915 holds the Instructions. This method must be used when there isn't a forwarding
3916 function in the <tt>BasicBlock</tt> class for the operation that you would like
3917 to perform. Because there are no forwarding functions for "updating"
3918 operations, you need to use this if you want to update the contents of a
3919 <tt>BasicBlock</tt>.</p></li>
3921 <li><tt><a href="#Function">Function</a> *getParent()</tt>
3923 <p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
3924 embedded into, or a null pointer if it is homeless.</p></li>
3926 <li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
3928 <p> Returns a pointer to the terminator instruction that appears at the end of
3929 the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
3930 instruction in the block is not a terminator, then a null pointer is
3938 <!-- ======================================================================= -->
3939 <div class="doc_subsection">
3940 <a name="Argument">The <tt>Argument</tt> class</a>
3943 <div class="doc_text">
3945 <p>This subclass of Value defines the interface for incoming formal
3946 arguments to a function. A Function maintains a list of its formal
3947 arguments. An argument has a pointer to the parent Function.</p>
3951 <!-- *********************************************************************** -->
3954 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3955 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
3956 <a href="http://validator.w3.org/check/referer"><img
3957 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01 Strict"></a>
3959 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
3960 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
3961 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
3962 Last modified: $Date$