Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / c9ad6fbab265d97f9568dd45117793ed7c62249c / . / docs / WritingAnLLVMPass.html
blob: 0d4dc09d2efe3b3f6f9d65090bd7828daa05e316 [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
6 <title>Writing an LLVM Pass</title>
7 <link rel="stylesheet" href="llvm.css" type="text/css">
8</head>
9<body>
10
11<div class="doc_title">
12 Writing an LLVM Pass
13</div>
14
15<ol>
16 <li><a href="#introduction">Introduction - What is a pass?</a></li>
17 <li><a href="#quickstart">Quick Start - Writing hello world</a>
18 <ul>
19 <li><a href="#makefile">Setting up the build environment</a></li>
20 <li><a href="#basiccode">Basic code required</a></li>
21 <li><a href="#running">Running a pass with <tt>opt</tt></a></li>
22 </ul></li>
23 <li><a href="#passtype">Pass classes and requirements</a>
24 <ul>
25 <li><a href="#ImmutablePass">The <tt>ImmutablePass</tt> class</a></li>
26 <li><a href="#ModulePass">The <tt>ModulePass</tt> class</a>
27 <ul>
28 <li><a href="#runOnModule">The <tt>runOnModule</tt> method</a></li>
29 </ul></li>
30 <li><a href="#CallGraphSCCPass">The <tt>CallGraphSCCPass</tt> class</a>
31 <ul>
32 <li><a href="#doInitialization_scc">The <tt>doInitialization(CallGraph
33 &amp;)</tt> method</a></li>
34 <li><a href="#runOnSCC">The <tt>runOnSCC</tt> method</a></li>
35 <li><a href="#doFinalization_scc">The <tt>doFinalization(CallGraph
36 &amp;)</tt> method</a></li>
37 </ul></li>
38 <li><a href="#FunctionPass">The <tt>FunctionPass</tt> class</a>
39 <ul>
40 <li><a href="#doInitialization_mod">The <tt>doInitialization(Module
41 &amp;)</tt> method</a></li>
42 <li><a href="#runOnFunction">The <tt>runOnFunction</tt> method</a></li>
43 <li><a href="#doFinalization_mod">The <tt>doFinalization(Module
44 &amp;)</tt> method</a></li>
45 </ul></li>
46 <li><a href="#LoopPass">The <tt>LoopPass</tt> class</a>
47 <ul>
48 <li><a href="#doInitialization_loop">The <tt>doInitialization(Loop *,
49 LPPassManager &amp;)</tt> method</a></li>
50 <li><a href="#runOnLoop">The <tt>runOnLoop</tt> method</a></li>
51 <li><a href="#doFinalization_loop">The <tt>doFinalization()
52 </tt> method</a></li>
53 </ul></li>
54 <li><a href="#BasicBlockPass">The <tt>BasicBlockPass</tt> class</a>
55 <ul>
56 <li><a href="#doInitialization_fn">The <tt>doInitialization(Function
57 &amp;)</tt> method</a></li>
58 <li><a href="#runOnBasicBlock">The <tt>runOnBasicBlock</tt>
59 method</a></li>
60 <li><a href="#doFinalization_fn">The <tt>doFinalization(Function
61 &amp;)</tt> method</a></li>
62 </ul></li>
63 <li><a href="#MachineFunctionPass">The <tt>MachineFunctionPass</tt>
64 class</a>
65 <ul>
66 <li><a href="#runOnMachineFunction">The
67 <tt>runOnMachineFunction(MachineFunction &amp;)</tt> method</a></li>
68 </ul></li>
69 </ul>
70 <li><a href="#registration">Pass Registration</a>
71 <ul>
72 <li><a href="#print">The <tt>print</tt> method</a></li>
73 </ul></li>
74 <li><a href="#interaction">Specifying interactions between passes</a>
75 <ul>
76 <li><a href="#getAnalysisUsage">The <tt>getAnalysisUsage</tt>
77 method</a></li>
78 <li><a href="#AU::addRequired">The <tt>AnalysisUsage::addRequired&lt;&gt;</tt> and <tt>AnalysisUsage::addRequiredTransitive&lt;&gt;</tt> methods</a></li>
79 <li><a href="#AU::addPreserved">The <tt>AnalysisUsage::addPreserved&lt;&gt;</tt> method</a></li>
80 <li><a href="#AU::examples">Example implementations of <tt>getAnalysisUsage</tt></a></li>
Duncan Sands4e0d6a72009-01-28 13:14:17 +0000[diff] [blame]81 <li><a href="#getAnalysis">The <tt>getAnalysis&lt;&gt;</tt> and
82<tt>getAnalysisIfAvailable&lt;&gt;</tt> methods</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]83 </ul></li>
84 <li><a href="#analysisgroup">Implementing Analysis Groups</a>
85 <ul>
86 <li><a href="#agconcepts">Analysis Group Concepts</a></li>
87 <li><a href="#registerag">Using <tt>RegisterAnalysisGroup</tt></a></li>
88 </ul></li>
89 <li><a href="#passStatistics">Pass Statistics</a>
90 <li><a href="#passmanager">What PassManager does</a>
91 <ul>
92 <li><a href="#releaseMemory">The <tt>releaseMemory</tt> method</a></li>
93 </ul></li>
94 <li><a href="#registering">Registering dynamically loaded passes</a>
95 <ul>
96 <li><a href="#registering_existing">Using existing registries</a></li>
97 <li><a href="#registering_new">Creating new registries</a></li>
98 </ul></li>
99 <li><a href="#debughints">Using GDB with dynamically loaded passes</a>
100 <ul>
101 <li><a href="#breakpoint">Setting a breakpoint in your pass</a></li>
102 <li><a href="#debugmisc">Miscellaneous Problems</a></li>
103 </ul></li>
104 <li><a href="#future">Future extensions planned</a>
105 <ul>
106 <li><a href="#SMP">Multithreaded LLVM</a></li>
107 </ul></li>
108</ol>
109
110<div class="doc_author">
111 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a> and
112 <a href="mailto:jlaskey@mac.com">Jim Laskey</a></p>
113</div>
114
115<!-- *********************************************************************** -->
116<div class="doc_section">
117 <a name="introduction">Introduction - What is a pass?</a>
118</div>
119<!-- *********************************************************************** -->
120
121<div class="doc_text">
122
123<p>The LLVM Pass Framework is an important part of the LLVM system, because LLVM
124passes are where most of the interesting parts of the compiler exist. Passes
125perform the transformations and optimizations that make up the compiler, they
126build the analysis results that are used by these transformations, and they are,
127above all, a structuring technique for compiler code.</p>
128
129<p>All LLVM passes are subclasses of the <tt><a
130href="http://llvm.org/doxygen/classllvm_1_1Pass.html">Pass</a></tt>
131class, which implement functionality by overriding virtual methods inherited
132from <tt>Pass</tt>. Depending on how your pass works, you should inherit from
133the <tt><a href="#ModulePass">ModulePass</a></tt>, <tt><a
134href="#CallGraphSCCPass">CallGraphSCCPass</a></tt>, <tt><a
135href="#FunctionPass">FunctionPass</a></tt>, or <tt><a
136href="#LoopPass">LoopPass</a></tt>, or <tt><a
137href="#BasicBlockPass">BasicBlockPass</a></tt> classes, which gives the system
138more information about what your pass does, and how it can be combined with
139other passes. One of the main features of the LLVM Pass Framework is that it
140schedules passes to run in an efficient way based on the constraints that your
141pass meets (which are indicated by which class they derive from).</p>
142
143<p>We start by showing you how to construct a pass, everything from setting up
144the code, to compiling, loading, and executing it. After the basics are down,
145more advanced features are discussed.</p>
146
147</div>
148
149<!-- *********************************************************************** -->
150<div class="doc_section">
151 <a name="quickstart">Quick Start - Writing hello world</a>
152</div>
153<!-- *********************************************************************** -->
154
155<div class="doc_text">
156
157<p>Here we describe how to write the "hello world" of passes. The "Hello" pass
158is designed to simply print out the name of non-external functions that exist in
159the program being compiled. It does not modify the program at all, it just
160inspects it. The source code and files for this pass are available in the LLVM
161source tree in the <tt>lib/Transforms/Hello</tt> directory.</p>
162
163</div>
164
165<!-- ======================================================================= -->
166<div class="doc_subsection">
167 <a name="makefile">Setting up the build environment</a>
168</div>
169
170<div class="doc_text">
171
172 <p>First, you need to create a new directory somewhere in the LLVM source
173 base. For this example, we'll assume that you made
174 <tt>lib/Transforms/Hello</tt>. Next, you must set up a build script
175 (Makefile) that will compile the source code for the new pass. To do this,
176 copy the following into <tt>Makefile</tt>:</p>
177 <hr/>
178
179<div class="doc_code"><pre>
180# Makefile for hello pass
181
Edwin Törökb26e2792009-10-12 13:37:29 +0000[diff] [blame]182# Path to top level of LLVM hierarchy
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]183LEVEL = ../../..
184
185# Name of the library to build
186LIBRARYNAME = Hello
187
188# Make the shared library become a loadable module so the tools can
189# dlopen/dlsym on the resulting library.
190LOADABLE_MODULE = 1
191
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]192# Include the makefile implementation stuff
193include $(LEVEL)/Makefile.common
194</pre></div>
195
196<p>This makefile specifies that all of the <tt>.cpp</tt> files in the current
197directory are to be compiled and linked together into a
198<tt>Debug/lib/Hello.so</tt> shared object that can be dynamically loaded by
199the <tt>opt</tt> or <tt>bugpoint</tt> tools via their <tt>-load</tt> options.
200If your operating system uses a suffix other than .so (such as windows or
201Mac OS/X), the appropriate extension will be used.</p>
202
203<p>Now that we have the build scripts set up, we just need to write the code for
204the pass itself.</p>
205
206</div>
207
208<!-- ======================================================================= -->
209<div class="doc_subsection">
210 <a name="basiccode">Basic code required</a>
211</div>
212
213<div class="doc_text">
214
215<p>Now that we have a way to compile our new pass, we just have to write it.
216Start out with:</p>
217
218<div class="doc_code"><pre>
219<b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>"
220<b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>"
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]221<b>#include</b> "<a href="http://llvm.org/doxygen/raw__ostream_8h.html">llvm/Support/raw_ostream.h</a>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]222</pre></div>
223
224<p>Which are needed because we are writing a <tt><a
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]225href="http://llvm.org/doxygen/classllvm_1_1Pass.html">Pass</a></tt>,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]226we are operating on <tt><a
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]227href="http://llvm.org/doxygen/classllvm_1_1Function.html">Function</a></tt>'s,
228and we will be doing some printing.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]229
230<p>Next we have:</p>
231<div class="doc_code"><pre>
232<b>using namespace llvm;</b>
233</pre></div>
234<p>... which is required because the functions from the include files
235live in the llvm namespace.
236</p>
237
238<p>Next we have:</p>
239
240<div class="doc_code"><pre>
241<b>namespace</b> {
242</pre></div>
243
244<p>... which starts out an anonymous namespace. Anonymous namespaces are to C++
245what the "<tt>static</tt>" keyword is to C (at global scope). It makes the
246things declared inside of the anonymous namespace only visible to the current
247file. If you're not familiar with them, consult a decent C++ book for more
248information.</p>
249
250<p>Next, we declare our pass itself:</p>
251
252<div class="doc_code"><pre>
253 <b>struct</b> Hello : <b>public</b> <a href="#FunctionPass">FunctionPass</a> {
254</pre></div><p>
255
256<p>This declares a "<tt>Hello</tt>" class that is a subclass of <tt><a
257href="http://llvm.org/doxygen/classllvm_1_1FunctionPass.html">FunctionPass</a></tt>.
258The different builtin pass subclasses are described in detail <a
259href="#passtype">later</a>, but for now, know that <a
260href="#FunctionPass"><tt>FunctionPass</tt></a>'s operate a function at a
261time.</p>
262
263<div class="doc_code"><pre>
264 static char ID;
Dan Gohmanc74a1972009-02-18 05:09:16 +0000[diff] [blame]265 Hello() : FunctionPass(&amp;ID) {}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]266</pre></div><p>
267
268<p> This declares pass identifier used by LLVM to identify pass. This allows LLVM to
269avoid using expensive C++ runtime information.</p>
270
271<div class="doc_code"><pre>
272 <b>virtual bool</b> <a href="#runOnFunction">runOnFunction</a>(Function &amp;F) {
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]273 errs() &lt;&lt; "<i>Hello: </i>" &lt;&lt; F.getName() &lt;&lt; "\n";
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]274 <b>return false</b>;
275 }
276 }; <i>// end of struct Hello</i>
277</pre></div>
278
279<p>We declare a "<a href="#runOnFunction"><tt>runOnFunction</tt></a>" method,
280which overloads an abstract virtual method inherited from <a
281href="#FunctionPass"><tt>FunctionPass</tt></a>. This is where we are supposed
282to do our thing, so we just print out our message with the name of each
283function.</p>
284
285<div class="doc_code"><pre>
286 char Hello::ID = 0;
287</pre></div>
288
289<p> We initialize pass ID here. LLVM uses ID's address to identify pass so
290initialization value is not important.</p>
291
292<div class="doc_code"><pre>
Devang Patel3aab76e2008-03-19 21:56:59 +0000[diff] [blame]293 RegisterPass&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>",
294 false /* Only looks at CFG */,
295 false /* Analysis Pass */);
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]296} <i>// end of anonymous namespace</i>
297</pre></div>
298
299<p>Lastly, we <a href="#registration">register our class</a> <tt>Hello</tt>,
300giving it a command line
Devang Patel3aab76e2008-03-19 21:56:59 +0000[diff] [blame]301argument "<tt>hello</tt>", and a name "<tt>Hello World Pass</tt>".
302Last two RegisterPass arguments are optional. Their default value is false.
303If a pass walks CFG without modifying it then third argument is set to true.
304If a pass is an analysis pass, for example dominator tree pass, then true
305is supplied as fourth argument. </p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]306
307<p>As a whole, the <tt>.cpp</tt> file looks like:</p>
308
309<div class="doc_code"><pre>
310<b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>"
311<b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>"
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]312<b>#include</b> "<a href="http://llvm.org/doxygen/raw__ostream_8h.html">llvm/Support/raw_ostream.h</a>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]313
314<b>using namespace llvm;</b>
315
316<b>namespace</b> {
317 <b>struct Hello</b> : <b>public</b> <a href="#FunctionPass">FunctionPass</a> {
318
319 static char ID;
Dan Gohmanc74a1972009-02-18 05:09:16 +0000[diff] [blame]320 Hello() : FunctionPass(&amp;ID) {}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]321
322 <b>virtual bool</b> <a href="#runOnFunction">runOnFunction</a>(Function &amp;F) {
Chris Lattner1efd4fd62009-09-08 05:14:44 +0000[diff] [blame]323 errs() &lt;&lt; "<i>Hello: </i>" &lt;&lt; F.getName() &lt;&lt; "\n";
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]324 <b>return false</b>;
325 }
326 };
327
Devang Patel8e46f052007-07-25 21:05:39 +0000[diff] [blame]328 char Hello::ID = 0;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]329 RegisterPass&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>");
330}
331</pre></div>
332
333<p>Now that it's all together, compile the file with a simple "<tt>gmake</tt>"
334command in the local directory and you should get a new
335"<tt>Debug/lib/Hello.so</tt> file. Note that everything in this file is
336contained in an anonymous namespace: this reflects the fact that passes are self
337contained units that do not need external interfaces (although they can have
338them) to be useful.</p>
339
340</div>
341
342<!-- ======================================================================= -->
343<div class="doc_subsection">
344 <a name="running">Running a pass with <tt>opt</tt></a>
345</div>
346
347<div class="doc_text">
348
349<p>Now that you have a brand new shiny shared object file, we can use the
350<tt>opt</tt> command to run an LLVM program through your pass. Because you
351registered your pass with the <tt>RegisterPass</tt> template, you will be able to
352use the <tt>opt</tt> tool to access it, once loaded.</p>
353
354<p>To test it, follow the example at the end of the <a
355href="GettingStarted.html">Getting Started Guide</a> to compile "Hello World" to
356LLVM. We can now run the bitcode file (<tt>hello.bc</tt>) for the program
357through our transformation like this (or course, any bitcode file will
358work):</p>
359
360<div class="doc_code"><pre>
361$ opt -load ../../../Debug/lib/Hello.so -hello &lt; hello.bc &gt; /dev/null
362Hello: __main
363Hello: puts
364Hello: main
365</pre></div>
366
367<p>The '<tt>-load</tt>' option specifies that '<tt>opt</tt>' should load your
368pass as a shared object, which makes '<tt>-hello</tt>' a valid command line
369argument (which is one reason you need to <a href="#registration">register your
370pass</a>). Because the hello pass does not modify the program in any
371interesting way, we just throw away the result of <tt>opt</tt> (sending it to
372<tt>/dev/null</tt>).</p>
373
374<p>To see what happened to the other string you registered, try running
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]375<tt>opt</tt> with the <tt>-help</tt> option:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]376
377<div class="doc_code"><pre>
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]378$ opt -load ../../../Debug/lib/Hello.so -help
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]379OVERVIEW: llvm .bc -&gt; .bc modular optimizer
380
381USAGE: opt [options] &lt;input bitcode&gt;
382
383OPTIONS:
384 Optimizations available:
385...
386 -funcresolve - Resolve Functions
387 -gcse - Global Common Subexpression Elimination
388 -globaldce - Dead Global Elimination
389 <b>-hello - Hello World Pass</b>
390 -indvars - Canonicalize Induction Variables
391 -inline - Function Integration/Inlining
392 -instcombine - Combine redundant instructions
393...
394</pre></div>
395
396<p>The pass name get added as the information string for your pass, giving some
397documentation to users of <tt>opt</tt>. Now that you have a working pass, you
398would go ahead and make it do the cool transformations you want. Once you get
399it all working and tested, it may become useful to find out how fast your pass
400is. The <a href="#passManager"><tt>PassManager</tt></a> provides a nice command
401line option (<tt>--time-passes</tt>) that allows you to get information about
402the execution time of your pass along with the other passes you queue up. For
403example:</p>
404
405<div class="doc_code"><pre>
406$ opt -load ../../../Debug/lib/Hello.so -hello -time-passes &lt; hello.bc &gt; /dev/null
407Hello: __main
408Hello: puts
409Hello: main
410===============================================================================
411 ... Pass execution timing report ...
412===============================================================================
413 Total Execution Time: 0.02 seconds (0.0479059 wall clock)
414
415 ---User Time--- --System Time-- --User+System-- ---Wall Time--- --- Pass Name ---
416 0.0100 (100.0%) 0.0000 ( 0.0%) 0.0100 ( 50.0%) 0.0402 ( 84.0%) Bitcode Writer
417 0.0000 ( 0.0%) 0.0100 (100.0%) 0.0100 ( 50.0%) 0.0031 ( 6.4%) Dominator Set Construction
418 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0013 ( 2.7%) Module Verifier
419 <b> 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0033 ( 6.9%) Hello World Pass</b>
420 0.0100 (100.0%) 0.0100 (100.0%) 0.0200 (100.0%) 0.0479 (100.0%) TOTAL
421</pre></div>
422
423<p>As you can see, our implementation above is pretty fast :). The additional
424passes listed are automatically inserted by the '<tt>opt</tt>' tool to verify
425that the LLVM emitted by your pass is still valid and well formed LLVM, which
426hasn't been broken somehow.</p>
427
428<p>Now that you have seen the basics of the mechanics behind passes, we can talk
429about some more details of how they work and how to use them.</p>
430
431</div>
432
433<!-- *********************************************************************** -->
434<div class="doc_section">
435 <a name="passtype">Pass classes and requirements</a>
436</div>
437<!-- *********************************************************************** -->
438
439<div class="doc_text">
440
441<p>One of the first things that you should do when designing a new pass is to
442decide what class you should subclass for your pass. The <a
443href="#basiccode">Hello World</a> example uses the <tt><a
444href="#FunctionPass">FunctionPass</a></tt> class for its implementation, but we
445did not discuss why or when this should occur. Here we talk about the classes
446available, from the most general to the most specific.</p>
447
448<p>When choosing a superclass for your Pass, you should choose the <b>most
449specific</b> class possible, while still being able to meet the requirements
450listed. This gives the LLVM Pass Infrastructure information necessary to
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]451optimize how passes are run, so that the resultant compiler isn't unnecessarily
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]452slow.</p>
453
454</div>
455
456<!-- ======================================================================= -->
457<div class="doc_subsection">
458 <a name="ImmutablePass">The <tt>ImmutablePass</tt> class</a>
459</div>
460
461<div class="doc_text">
462
463<p>The most plain and boring type of pass is the "<tt><a
464href="http://llvm.org/doxygen/classllvm_1_1ImmutablePass.html">ImmutablePass</a></tt>"
465class. This pass type is used for passes that do not have to be run, do not
466change state, and never need to be updated. This is not a normal type of
467transformation or analysis, but can provide information about the current
468compiler configuration.</p>
469
470<p>Although this pass class is very infrequently used, it is important for
471providing information about the current target machine being compiled for, and
472other static information that can affect the various transformations.</p>
473
474<p><tt>ImmutablePass</tt>es never invalidate other transformations, are never
475invalidated, and are never "run".</p>
476
477</div>
478
479<!-- ======================================================================= -->
480<div class="doc_subsection">
481 <a name="ModulePass">The <tt>ModulePass</tt> class</a>
482</div>
483
484<div class="doc_text">
485
486<p>The "<tt><a
487href="http://llvm.org/doxygen/classllvm_1_1ModulePass.html">ModulePass</a></tt>"
488class is the most general of all superclasses that you can use. Deriving from
489<tt>ModulePass</tt> indicates that your pass uses the entire program as a unit,
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]490referring to function bodies in no predictable order, or adding and removing
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]491functions. Because nothing is known about the behavior of <tt>ModulePass</tt>
Daniel Dunbarb7750c52009-07-01 23:38:44 +0000[diff] [blame]492subclasses, no optimization can be done for their execution.</p>
493
494<p>A module pass can use function level passes (e.g. dominators) using
495the getAnalysis interface
496<tt>getAnalysis&lt;DominatorTree&gt;(llvm::Function *)</tt> to provide the
497function to retrieve analysis result for, if the function pass does not require
Devang Patel80eb3972009-08-10 16:37:29 +0000[diff] [blame]498any module or immutable passes. Note that this can only be done for functions for which the
Daniel Dunbarb7750c52009-07-01 23:38:44 +0000[diff] [blame]499analysis ran, e.g. in the case of dominators you should only ask for the
500DominatorTree for function definitions, not declarations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]501
502<p>To write a correct <tt>ModulePass</tt> subclass, derive from
503<tt>ModulePass</tt> and overload the <tt>runOnModule</tt> method with the
504following signature:</p>
505
506</div>
507
508<!-- _______________________________________________________________________ -->
509<div class="doc_subsubsection">
510 <a name="runOnModule">The <tt>runOnModule</tt> method</a>
511</div>
512
513<div class="doc_text">
514
515<div class="doc_code"><pre>
516 <b>virtual bool</b> runOnModule(Module &amp;M) = 0;
517</pre></div>
518
519<p>The <tt>runOnModule</tt> method performs the interesting work of the pass.
520It should return true if the module was modified by the transformation and
521false otherwise.</p>
522
523</div>
524
525<!-- ======================================================================= -->
526<div class="doc_subsection">
527 <a name="CallGraphSCCPass">The <tt>CallGraphSCCPass</tt> class</a>
528</div>
529
530<div class="doc_text">
531
532<p>The "<tt><a
533href="http://llvm.org/doxygen/classllvm_1_1CallGraphSCCPass.html">CallGraphSCCPass</a></tt>"
534is used by passes that need to traverse the program bottom-up on the call graph
535(callees before callers). Deriving from CallGraphSCCPass provides some
536mechanics for building and traversing the CallGraph, but also allows the system
537to optimize execution of CallGraphSCCPass's. If your pass meets the
538requirements outlined below, and doesn't meet the requirements of a <tt><a
539href="#FunctionPass">FunctionPass</a></tt> or <tt><a
540href="#BasicBlockPass">BasicBlockPass</a></tt>, you should derive from
541<tt>CallGraphSCCPass</tt>.</p>
542
543<p><b>TODO</b>: explain briefly what SCC, Tarjan's algo, and B-U mean.</p>
544
545<p>To be explicit, <tt>CallGraphSCCPass</tt> subclasses are:</p>
546
547<ol>
548
549<li>... <em>not allowed</em> to modify any <tt>Function</tt>s that are not in
550the current SCC.</li>
551
552<li>... <em>not allowed</em> to inspect any Function's other than those in the
553current SCC and the direct callees of the SCC.</li>
554
555<li>... <em>required</em> to preserve the current CallGraph object, updating it
556to reflect any changes made to the program.</li>
557
558<li>... <em>not allowed</em> to add or remove SCC's from the current Module,
559though they may change the contents of an SCC.</li>
560
561<li>... <em>allowed</em> to add or remove global variables from the current
562Module.</li>
563
564<li>... <em>allowed</em> to maintain state across invocations of
565 <a href="#runOnSCC"><tt>runOnSCC</tt></a> (including global data).</li>
566</ol>
567
568<p>Implementing a <tt>CallGraphSCCPass</tt> is slightly tricky in some cases
569because it has to handle SCCs with more than one node in it. All of the virtual
570methods described below should return true if they modified the program, or
571false if they didn't.</p>
572
573</div>
574
575<!-- _______________________________________________________________________ -->
576<div class="doc_subsubsection">
577 <a name="doInitialization_scc">The <tt>doInitialization(CallGraph &amp;)</tt>
578 method</a>
579</div>
580
581<div class="doc_text">
582
583<div class="doc_code"><pre>
584 <b>virtual bool</b> doInitialization(CallGraph &amp;CG);
585</pre></div>
586
587<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
588<tt>CallGraphSCCPass</tt>'s are not allowed to do. They can add and remove
589functions, get pointers to functions, etc. The <tt>doInitialization</tt> method
590is designed to do simple initialization type of stuff that does not depend on
591the SCCs being processed. The <tt>doInitialization</tt> method call is not
592scheduled to overlap with any other pass executions (thus it should be very
593fast).</p>
594
595</div>
596
597<!-- _______________________________________________________________________ -->
598<div class="doc_subsubsection">
599 <a name="runOnSCC">The <tt>runOnSCC</tt> method</a>
600</div>
601
602<div class="doc_text">
603
604<div class="doc_code"><pre>
Chris Lattner25cb59a2010-04-16 23:07:44 +0000[diff] [blame]605 <b>virtual bool</b> runOnSCC(CallGraphSCC &amp;SCC) = 0;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]606</pre></div>
607
608<p>The <tt>runOnSCC</tt> method performs the interesting work of the pass, and
609should return true if the module was modified by the transformation, false
610otherwise.</p>
611
612</div>
613
614<!-- _______________________________________________________________________ -->
615<div class="doc_subsubsection">
616 <a name="doFinalization_scc">The <tt>doFinalization(CallGraph
617 &amp;)</tt> method</a>
618</div>
619
620<div class="doc_text">
621
622<div class="doc_code"><pre>
623 <b>virtual bool</b> doFinalization(CallGraph &amp;CG);
624</pre></div>
625
626<p>The <tt>doFinalization</tt> method is an infrequently used method that is
627called when the pass framework has finished calling <a
628href="#runOnFunction"><tt>runOnFunction</tt></a> for every function in the
629program being compiled.</p>
630
631</div>
632
633<!-- ======================================================================= -->
634<div class="doc_subsection">
635 <a name="FunctionPass">The <tt>FunctionPass</tt> class</a>
636</div>
637
638<div class="doc_text">
639
640<p>In contrast to <tt>ModulePass</tt> subclasses, <tt><a
641href="http://llvm.org/doxygen/classllvm_1_1Pass.html">FunctionPass</a></tt>
642subclasses do have a predictable, local behavior that can be expected by the
643system. All <tt>FunctionPass</tt> execute on each function in the program
644independent of all of the other functions in the program.
645<tt>FunctionPass</tt>'s do not require that they are executed in a particular
646order, and <tt>FunctionPass</tt>'s do not modify external functions.</p>
647
648<p>To be explicit, <tt>FunctionPass</tt> subclasses are not allowed to:</p>
649
650<ol>
651<li>Modify a Function other than the one currently being processed.</li>
652<li>Add or remove Function's from the current Module.</li>
653<li>Add or remove global variables from the current Module.</li>
654<li>Maintain state across invocations of
655 <a href="#runOnFunction"><tt>runOnFunction</tt></a> (including global data)</li>
656</ol>
657
658<p>Implementing a <tt>FunctionPass</tt> is usually straightforward (See the <a
659href="#basiccode">Hello World</a> pass for example). <tt>FunctionPass</tt>'s
660may overload three virtual methods to do their work. All of these methods
661should return true if they modified the program, or false if they didn't.</p>
662
663</div>
664
665<!-- _______________________________________________________________________ -->
666<div class="doc_subsubsection">
667 <a name="doInitialization_mod">The <tt>doInitialization(Module &amp;)</tt>
668 method</a>
669</div>
670
671<div class="doc_text">
672
673<div class="doc_code"><pre>
674 <b>virtual bool</b> doInitialization(Module &amp;M);
675</pre></div>
676
677<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
678<tt>FunctionPass</tt>'s are not allowed to do. They can add and remove
679functions, get pointers to functions, etc. The <tt>doInitialization</tt> method
680is designed to do simple initialization type of stuff that does not depend on
681the functions being processed. The <tt>doInitialization</tt> method call is not
682scheduled to overlap with any other pass executions (thus it should be very
683fast).</p>
684
685<p>A good example of how this method should be used is the <a
686href="http://llvm.org/doxygen/LowerAllocations_8cpp-source.html">LowerAllocations</a>
687pass. This pass converts <tt>malloc</tt> and <tt>free</tt> instructions into
688platform dependent <tt>malloc()</tt> and <tt>free()</tt> function calls. It
689uses the <tt>doInitialization</tt> method to get a reference to the malloc and
690free functions that it needs, adding prototypes to the module if necessary.</p>
691
692</div>
693
694<!-- _______________________________________________________________________ -->
695<div class="doc_subsubsection">
696 <a name="runOnFunction">The <tt>runOnFunction</tt> method</a>
697</div>
698
699<div class="doc_text">
700
701<div class="doc_code"><pre>
702 <b>virtual bool</b> runOnFunction(Function &amp;F) = 0;
703</pre></div><p>
704
705<p>The <tt>runOnFunction</tt> method must be implemented by your subclass to do
706the transformation or analysis work of your pass. As usual, a true value should
707be returned if the function is modified.</p>
708
709</div>
710
711<!-- _______________________________________________________________________ -->
712<div class="doc_subsubsection">
713 <a name="doFinalization_mod">The <tt>doFinalization(Module
714 &amp;)</tt> method</a>
715</div>
716
717<div class="doc_text">
718
719<div class="doc_code"><pre>
720 <b>virtual bool</b> doFinalization(Module &amp;M);
721</pre></div>
722
723<p>The <tt>doFinalization</tt> method is an infrequently used method that is
724called when the pass framework has finished calling <a
725href="#runOnFunction"><tt>runOnFunction</tt></a> for every function in the
726program being compiled.</p>
727
728</div>
729
730<!-- ======================================================================= -->
731<div class="doc_subsection">
732 <a name="LoopPass">The <tt>LoopPass</tt> class </a>
733</div>
734
735<div class="doc_text">
736
737<p> All <tt>LoopPass</tt> execute on each loop in the function independent of
738all of the other loops in the function. <tt>LoopPass</tt> processes loops in
739loop nest order such that outer most loop is processed last. </p>
740
741<p> <tt>LoopPass</tt> subclasses are allowed to update loop nest using
742<tt>LPPassManager</tt> interface. Implementing a loop pass is usually
743straightforward. <tt>Looppass</tt>'s may overload three virtual methods to
744do their work. All these methods should return true if they modified the
745program, or false if they didn't. </p>
746</div>
747
748<!-- _______________________________________________________________________ -->
749<div class="doc_subsubsection">
750 <a name="doInitialization_loop">The <tt>doInitialization(Loop *,
751 LPPassManager &amp;)</tt>
752 method</a>
753</div>
754
755<div class="doc_text">
756
757<div class="doc_code"><pre>
758 <b>virtual bool</b> doInitialization(Loop *, LPPassManager &amp;LPM);
759</pre></div>
760
761<p>The <tt>doInitialization</tt> method is designed to do simple initialization
762type of stuff that does not depend on the functions being processed. The
763<tt>doInitialization</tt> method call is not scheduled to overlap with any
764other pass executions (thus it should be very fast). LPPassManager
765interface should be used to access Function or Module level analysis
766information.</p>
767
768</div>
769
770
771<!-- _______________________________________________________________________ -->
772<div class="doc_subsubsection">
773 <a name="runOnLoop">The <tt>runOnLoop</tt> method</a>
774</div>
775
776<div class="doc_text">
777
778<div class="doc_code"><pre>
779 <b>virtual bool</b> runOnLoop(Loop *, LPPassManager &amp;LPM) = 0;
780</pre></div><p>
781
782<p>The <tt>runOnLoop</tt> method must be implemented by your subclass to do
783the transformation or analysis work of your pass. As usual, a true value should
784be returned if the function is modified. <tt>LPPassManager</tt> interface
785should be used to update loop nest.</p>
786
787</div>
788
789<!-- _______________________________________________________________________ -->
790<div class="doc_subsubsection">
791 <a name="doFinalization_loop">The <tt>doFinalization()</tt> method</a>
792</div>
793
794<div class="doc_text">
795
796<div class="doc_code"><pre>
797 <b>virtual bool</b> doFinalization();
798</pre></div>
799
800<p>The <tt>doFinalization</tt> method is an infrequently used method that is
801called when the pass framework has finished calling <a
802href="#runOnLoop"><tt>runOnLoop</tt></a> for every loop in the
803program being compiled. </p>
804
805</div>
806
807
808
809<!-- ======================================================================= -->
810<div class="doc_subsection">
811 <a name="BasicBlockPass">The <tt>BasicBlockPass</tt> class</a>
812</div>
813
814<div class="doc_text">
815
816<p><tt>BasicBlockPass</tt>'s are just like <a
817href="#FunctionPass"><tt>FunctionPass</tt></a>'s, except that they must limit
818their scope of inspection and modification to a single basic block at a time.
819As such, they are <b>not</b> allowed to do any of the following:</p>
820
821<ol>
822<li>Modify or inspect any basic blocks outside of the current one</li>
823<li>Maintain state across invocations of
824 <a href="#runOnBasicBlock"><tt>runOnBasicBlock</tt></a></li>
825<li>Modify the control flow graph (by altering terminator instructions)</li>
826<li>Any of the things forbidden for
827 <a href="#FunctionPass"><tt>FunctionPass</tt></a>es.</li>
828</ol>
829
830<p><tt>BasicBlockPass</tt>es are useful for traditional local and "peephole"
831optimizations. They may override the same <a
832href="#doInitialization_mod"><tt>doInitialization(Module &amp;)</tt></a> and <a
833href="#doFinalization_mod"><tt>doFinalization(Module &amp;)</tt></a> methods that <a
834href="#FunctionPass"><tt>FunctionPass</tt></a>'s have, but also have the following virtual methods that may also be implemented:</p>
835
836</div>
837
838<!-- _______________________________________________________________________ -->
839<div class="doc_subsubsection">
840 <a name="doInitialization_fn">The <tt>doInitialization(Function
841 &amp;)</tt> method</a>
842</div>
843
844<div class="doc_text">
845
846<div class="doc_code"><pre>
847 <b>virtual bool</b> doInitialization(Function &amp;F);
848</pre></div>
849
850<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
851<tt>BasicBlockPass</tt>'s are not allowed to do, but that
852<tt>FunctionPass</tt>'s can. The <tt>doInitialization</tt> method is designed
853to do simple initialization that does not depend on the
854BasicBlocks being processed. The <tt>doInitialization</tt> method call is not
855scheduled to overlap with any other pass executions (thus it should be very
856fast).</p>
857
858</div>
859
860<!-- _______________________________________________________________________ -->
861<div class="doc_subsubsection">
862 <a name="runOnBasicBlock">The <tt>runOnBasicBlock</tt> method</a>
863</div>
864
865<div class="doc_text">
866
867<div class="doc_code"><pre>
868 <b>virtual bool</b> runOnBasicBlock(BasicBlock &amp;BB) = 0;
869</pre></div>
870
871<p>Override this function to do the work of the <tt>BasicBlockPass</tt>. This
872function is not allowed to inspect or modify basic blocks other than the
873parameter, and are not allowed to modify the CFG. A true value must be returned
874if the basic block is modified.</p>
875
876</div>
877
878<!-- _______________________________________________________________________ -->
879<div class="doc_subsubsection">
880 <a name="doFinalization_fn">The <tt>doFinalization(Function &amp;)</tt>
881 method</a>
882</div>
883
884<div class="doc_text">
885
886<div class="doc_code"><pre>
887 <b>virtual bool</b> doFinalization(Function &amp;F);
888</pre></div>
889
890<p>The <tt>doFinalization</tt> method is an infrequently used method that is
891called when the pass framework has finished calling <a
892href="#runOnBasicBlock"><tt>runOnBasicBlock</tt></a> for every BasicBlock in the
893program being compiled. This can be used to perform per-function
894finalization.</p>
895
896</div>
897
898<!-- ======================================================================= -->
899<div class="doc_subsection">
900 <a name="MachineFunctionPass">The <tt>MachineFunctionPass</tt> class</a>
901</div>
902
903<div class="doc_text">
904
905<p>A <tt>MachineFunctionPass</tt> is a part of the LLVM code generator that
906executes on the machine-dependent representation of each LLVM function in the
Dan Gohmanc23b1a32010-03-10 01:29:39 +0000[diff] [blame]907program.</p>
908
909<p>Code generator passes are registered and initialized specially by
910<tt>TargetMachine::addPassesToEmitFile</tt> and similar routines, so they
911cannot generally be run from the <tt>opt</tt> or <tt>bugpoint</tt>
912commands.</p>
913
914<p>A <tt>MachineFunctionPass</tt> is also a <tt>FunctionPass</tt>, so all
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]915the restrictions that apply to a <tt>FunctionPass</tt> also apply to it.
916<tt>MachineFunctionPass</tt>es also have additional restrictions. In particular,
917<tt>MachineFunctionPass</tt>es are not allowed to do any of the following:</p>
918
919<ol>
Dan Gohmanc23b1a32010-03-10 01:29:39 +0000[diff] [blame]920<li>Modify or create any LLVM IR Instructions, BasicBlocks, Arguments,
921 Functions, GlobalVariables, GlobalAliases, or Modules.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]922<li>Modify a MachineFunction other than the one currently being processed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]923<li>Maintain state across invocations of <a
924href="#runOnMachineFunction"><tt>runOnMachineFunction</tt></a> (including global
925data)</li>
926</ol>
927
928</div>
929
930<!-- _______________________________________________________________________ -->
931<div class="doc_subsubsection">
932 <a name="runOnMachineFunction">The <tt>runOnMachineFunction(MachineFunction
933 &amp;MF)</tt> method</a>
934</div>
935
936<div class="doc_text">
937
938<div class="doc_code"><pre>
939 <b>virtual bool</b> runOnMachineFunction(MachineFunction &amp;MF) = 0;
940</pre></div>
941
942<p><tt>runOnMachineFunction</tt> can be considered the main entry point of a
943<tt>MachineFunctionPass</tt>; that is, you should override this method to do the
944work of your <tt>MachineFunctionPass</tt>.</p>
945
946<p>The <tt>runOnMachineFunction</tt> method is called on every
947<tt>MachineFunction</tt> in a <tt>Module</tt>, so that the
948<tt>MachineFunctionPass</tt> may perform optimizations on the machine-dependent
949representation of the function. If you want to get at the LLVM <tt>Function</tt>
950for the <tt>MachineFunction</tt> you're working on, use
951<tt>MachineFunction</tt>'s <tt>getFunction()</tt> accessor method -- but
952remember, you may not modify the LLVM <tt>Function</tt> or its contents from a
953<tt>MachineFunctionPass</tt>.</p>
954
955</div>
956
957<!-- *********************************************************************** -->
958<div class="doc_section">
959 <a name="registration">Pass registration</a>
960</div>
961<!-- *********************************************************************** -->
962
963<div class="doc_text">
964
965<p>In the <a href="#basiccode">Hello World</a> example pass we illustrated how
966pass registration works, and discussed some of the reasons that it is used and
967what it does. Here we discuss how and why passes are registered.</p>
968
969<p>As we saw above, passes are registered with the <b><tt>RegisterPass</tt></b>
970template, which requires you to pass at least two
971parameters. The first parameter is the name of the pass that is to be used on
972the command line to specify that the pass should be added to a program (for
973example, with <tt>opt</tt> or <tt>bugpoint</tt>). The second argument is the
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]974name of the pass, which is to be used for the <tt>-help</tt> output of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]975programs, as
976well as for debug output generated by the <tt>--debug-pass</tt> option.</p>
977
978<p>If you want your pass to be easily dumpable, you should
979implement the virtual <tt>print</tt> method:</p>
980
981</div>
982
983<!-- _______________________________________________________________________ -->
984<div class="doc_subsubsection">
985 <a name="print">The <tt>print</tt> method</a>
986</div>
987
988<div class="doc_text">
989
990<div class="doc_code"><pre>
Edwin Török72a8fd22008-10-28 17:29:23 +0000[diff] [blame]991 <b>virtual void</b> print(std::ostream &amp;O, <b>const</b> Module *M) <b>const</b>;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]992</pre></div>
993
994<p>The <tt>print</tt> method must be implemented by "analyses" in order to print
995a human readable version of the analysis results. This is useful for debugging
996an analysis itself, as well as for other people to figure out how an analysis
997works. Use the <tt>opt -analyze</tt> argument to invoke this method.</p>
998
999<p>The <tt>llvm::OStream</tt> parameter specifies the stream to write the results on,
1000and the <tt>Module</tt> parameter gives a pointer to the top level module of the
1001program that has been analyzed. Note however that this pointer may be null in
1002certain circumstances (such as calling the <tt>Pass::dump()</tt> from a
1003debugger), so it should only be used to enhance debug output, it should not be
1004depended on.</p>
1005
1006</div>
1007
1008<!-- *********************************************************************** -->
1009<div class="doc_section">
1010 <a name="interaction">Specifying interactions between passes</a>
1011</div>
1012<!-- *********************************************************************** -->
1013
1014<div class="doc_text">
1015
John Criswella99e43f2007-12-03 19:34:25 +0000[diff] [blame]1016<p>One of the main responsibilities of the <tt>PassManager</tt> is to make sure
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1017that passes interact with each other correctly. Because <tt>PassManager</tt>
1018tries to <a href="#passmanager">optimize the execution of passes</a> it must
1019know how the passes interact with each other and what dependencies exist between
1020the various passes. To track this, each pass can declare the set of passes that
1021are required to be executed before the current pass, and the passes which are
1022invalidated by the current pass.</p>
1023
1024<p>Typically this functionality is used to require that analysis results are
1025computed before your pass is run. Running arbitrary transformation passes can
1026invalidate the computed analysis results, which is what the invalidation set
1027specifies. If a pass does not implement the <tt><a
1028href="#getAnalysisUsage">getAnalysisUsage</a></tt> method, it defaults to not
1029having any prerequisite passes, and invalidating <b>all</b> other passes.</p>
1030
1031</div>
1032
1033<!-- _______________________________________________________________________ -->
1034<div class="doc_subsubsection">
1035 <a name="getAnalysisUsage">The <tt>getAnalysisUsage</tt> method</a>
1036</div>
1037
1038<div class="doc_text">
1039
1040<div class="doc_code"><pre>
1041 <b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;Info) <b>const</b>;
1042</pre></div>
1043
1044<p>By implementing the <tt>getAnalysisUsage</tt> method, the required and
1045invalidated sets may be specified for your transformation. The implementation
1046should fill in the <tt><a
1047href="http://llvm.org/doxygen/classllvm_1_1AnalysisUsage.html">AnalysisUsage</a></tt>
1048object with information about which passes are required and not invalidated. To
1049do this, a pass may call any of the following methods on the AnalysisUsage
1050object:</p>
1051</div>
1052
1053<!-- _______________________________________________________________________ -->
1054<div class="doc_subsubsection">
1055 <a name="AU::addRequired">The <tt>AnalysisUsage::addRequired&lt;&gt;</tt> and <tt>AnalysisUsage::addRequiredTransitive&lt;&gt;</tt> methods</a>
1056</div>
1057
1058<div class="doc_text">
1059<p>
1060If your pass requires a previous pass to be executed (an analysis for example),
1061it can use one of these methods to arrange for it to be run before your pass.
1062LLVM has many different types of analyses and passes that can be required,
1063spanning the range from <tt>DominatorSet</tt> to <tt>BreakCriticalEdges</tt>.
1064Requiring <tt>BreakCriticalEdges</tt>, for example, guarantees that there will
1065be no critical edges in the CFG when your pass has been run.
1066</p>
1067
1068<p>
1069Some analyses chain to other analyses to do their job. For example, an <a
1070href="AliasAnalysis.html">AliasAnalysis</a> implementation is required to <a
1071href="AliasAnalysis.html#chaining">chain</a> to other alias analysis passes. In
1072cases where analyses chain, the <tt>addRequiredTransitive</tt> method should be
1073used instead of the <tt>addRequired</tt> method. This informs the PassManager
1074that the transitively required pass should be alive as long as the requiring
1075pass is.
1076</p>
1077</div>
1078
1079<!-- _______________________________________________________________________ -->
1080<div class="doc_subsubsection">
1081 <a name="AU::addPreserved">The <tt>AnalysisUsage::addPreserved&lt;&gt;</tt> method</a>
1082</div>
1083
1084<div class="doc_text">
1085<p>
1086One of the jobs of the PassManager is to optimize how and when analyses are run.
1087In particular, it attempts to avoid recomputing data unless it needs to. For
1088this reason, passes are allowed to declare that they preserve (i.e., they don't
1089invalidate) an existing analysis if it's available. For example, a simple
1090constant folding pass would not modify the CFG, so it can't possibly affect the
1091results of dominator analysis. By default, all passes are assumed to invalidate
1092all others.
1093</p>
1094
1095<p>
1096The <tt>AnalysisUsage</tt> class provides several methods which are useful in
1097certain circumstances that are related to <tt>addPreserved</tt>. In particular,
1098the <tt>setPreservesAll</tt> method can be called to indicate that the pass does
1099not modify the LLVM program at all (which is true for analyses), and the
1100<tt>setPreservesCFG</tt> method can be used by transformations that change
1101instructions in the program but do not modify the CFG or terminator instructions
1102(note that this property is implicitly set for <a
1103href="#BasicBlockPass">BasicBlockPass</a>'s).
1104</p>
1105
1106<p>
1107<tt>addPreserved</tt> is particularly useful for transformations like
1108<tt>BreakCriticalEdges</tt>. This pass knows how to update a small set of loop
1109and dominator related analyses if they exist, so it can preserve them, despite
1110the fact that it hacks on the CFG.
1111</p>
1112</div>
1113
1114<!-- _______________________________________________________________________ -->
1115<div class="doc_subsubsection">
1116 <a name="AU::examples">Example implementations of <tt>getAnalysisUsage</tt></a>
1117</div>
1118
1119<div class="doc_text">
1120
1121<div class="doc_code"><pre>
1122 <i>// This is an example implementation from an analysis, which does not modify
1123 // the program at all, yet has a prerequisite.</i>
1124 <b>void</b> <a href="http://llvm.org/doxygen/classllvm_1_1PostDominanceFrontier.html">PostDominanceFrontier</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
1125 AU.setPreservesAll();
1126 AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1PostDominatorTree.html">PostDominatorTree</a>&gt;();
1127 }
1128</pre></div>
1129
1130<p>and:</p>
1131
1132<div class="doc_code"><pre>
1133 <i>// This example modifies the program, but does not modify the CFG</i>
1134 <b>void</b> <a href="http://llvm.org/doxygen/structLICM.html">LICM</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
1135 AU.setPreservesCFG();
1136 AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1LoopInfo.html">LoopInfo</a>&gt;();
1137 }
1138</pre></div>
1139
1140</div>
1141
1142<!-- _______________________________________________________________________ -->
1143<div class="doc_subsubsection">
Duncan Sands4e0d6a72009-01-28 13:14:17 +0000[diff] [blame]1144 <a name="getAnalysis">The <tt>getAnalysis&lt;&gt;</tt> and
1145<tt>getAnalysisIfAvailable&lt;&gt;</tt> methods</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1146</div>
1147
1148<div class="doc_text">
1149
1150<p>The <tt>Pass::getAnalysis&lt;&gt;</tt> method is automatically inherited by
1151your class, providing you with access to the passes that you declared that you
1152required with the <a href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a>
1153method. It takes a single template argument that specifies which pass class you
1154want, and returns a reference to that pass. For example:</p>
1155
1156<div class="doc_code"><pre>
1157 bool LICM::runOnFunction(Function &amp;F) {
1158 LoopInfo &amp;LI = getAnalysis&lt;LoopInfo&gt;();
1159 ...
1160 }
1161</pre></div>
1162
1163<p>This method call returns a reference to the pass desired. You may get a
1164runtime assertion failure if you attempt to get an analysis that you did not
1165declare as required in your <a
1166href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a> implementation. This
1167method can be called by your <tt>run*</tt> method implementation, or by any
1168other local method invoked by your <tt>run*</tt> method.
1169
1170A module level pass can use function level analysis info using this interface.
1171For example:</p>
1172
1173<div class="doc_code"><pre>
1174 bool ModuleLevelPass::runOnModule(Module &amp;M) {
1175 ...
1176 DominatorTree &amp;DT = getAnalysis&lt;DominatorTree&gt;(Func);
1177 ...
1178 }
1179</pre></div>
1180
1181<p>In above example, runOnFunction for DominatorTree is called by pass manager
1182before returning a reference to the desired pass.</p>
1183
1184<p>
1185If your pass is capable of updating analyses if they exist (e.g.,
1186<tt>BreakCriticalEdges</tt>, as described above), you can use the
Duncan Sands4e0d6a72009-01-28 13:14:17 +0000[diff] [blame]1187<tt>getAnalysisIfAvailable</tt> method, which returns a pointer to the analysis
1188if it is active. For example:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1189
1190<div class="doc_code"><pre>
1191 ...
Duncan Sands4e0d6a72009-01-28 13:14:17 +0000[diff] [blame]1192 if (DominatorSet *DS = getAnalysisIfAvailable&lt;DominatorSet&gt;()) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1193 <i>// A DominatorSet is active. This code will update it.</i>
1194 }
1195 ...
1196</pre></div>
1197
1198</div>
1199
1200<!-- *********************************************************************** -->
1201<div class="doc_section">
1202 <a name="analysisgroup">Implementing Analysis Groups</a>
1203</div>
1204<!-- *********************************************************************** -->
1205
1206<div class="doc_text">
1207
Chris Lattner942c3952007-11-16 05:32:05 +0000[diff] [blame]1208<p>Now that we understand the basics of how passes are defined, how they are
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1209used, and how they are required from other passes, it's time to get a little bit
1210fancier. All of the pass relationships that we have seen so far are very
1211simple: one pass depends on one other specific pass to be run before it can run.
1212For many applications, this is great, for others, more flexibility is
1213required.</p>
1214
1215<p>In particular, some analyses are defined such that there is a single simple
1216interface to the analysis results, but multiple ways of calculating them.
1217Consider alias analysis for example. The most trivial alias analysis returns
1218"may alias" for any alias query. The most sophisticated analysis a
1219flow-sensitive, context-sensitive interprocedural analysis that can take a
1220significant amount of time to execute (and obviously, there is a lot of room
1221between these two extremes for other implementations). To cleanly support
1222situations like this, the LLVM Pass Infrastructure supports the notion of
1223Analysis Groups.</p>
1224
1225</div>
1226
1227<!-- _______________________________________________________________________ -->
1228<div class="doc_subsubsection">
1229 <a name="agconcepts">Analysis Group Concepts</a>
1230</div>
1231
1232<div class="doc_text">
1233
1234<p>An Analysis Group is a single simple interface that may be implemented by
1235multiple different passes. Analysis Groups can be given human readable names
1236just like passes, but unlike passes, they need not derive from the <tt>Pass</tt>
1237class. An analysis group may have one or more implementations, one of which is
1238the "default" implementation.</p>
1239
1240<p>Analysis groups are used by client passes just like other passes are: the
1241<tt>AnalysisUsage::addRequired()</tt> and <tt>Pass::getAnalysis()</tt> methods.
1242In order to resolve this requirement, the <a href="#passmanager">PassManager</a>
1243scans the available passes to see if any implementations of the analysis group
1244are available. If none is available, the default implementation is created for
1245the pass to use. All standard rules for <A href="#interaction">interaction
1246between passes</a> still apply.</p>
1247
1248<p>Although <a href="#registration">Pass Registration</a> is optional for normal
1249passes, all analysis group implementations must be registered, and must use the
1250<A href="#registerag"><tt>RegisterAnalysisGroup</tt></a> template to join the
1251implementation pool. Also, a default implementation of the interface
1252<b>must</b> be registered with <A
1253href="#registerag"><tt>RegisterAnalysisGroup</tt></a>.</p>
1254
1255<p>As a concrete example of an Analysis Group in action, consider the <a
1256href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>
1257analysis group. The default implementation of the alias analysis interface (the
1258<tt><a
1259href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">basicaa</a></tt>
1260pass) just does a few simple checks that don't require significant analysis to
1261compute (such as: two different globals can never alias each other, etc).
1262Passes that use the <tt><a
1263href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a></tt>
1264interface (for example the <tt><a
1265href="http://llvm.org/doxygen/structGCSE.html">gcse</a></tt> pass), do
1266not care which implementation of alias analysis is actually provided, they just
1267use the designated interface.</p>
1268
1269<p>From the user's perspective, commands work just like normal. Issuing the
1270command '<tt>opt -gcse ...</tt>' will cause the <tt>basicaa</tt> class to be
1271instantiated and added to the pass sequence. Issuing the command '<tt>opt
1272-somefancyaa -gcse ...</tt>' will cause the <tt>gcse</tt> pass to use the
1273<tt>somefancyaa</tt> alias analysis (which doesn't actually exist, it's just a
1274hypothetical example) instead.</p>
1275
1276</div>
1277
1278<!-- _______________________________________________________________________ -->
1279<div class="doc_subsubsection">
1280 <a name="registerag">Using <tt>RegisterAnalysisGroup</tt></a>
1281</div>
1282
1283<div class="doc_text">
1284
1285<p>The <tt>RegisterAnalysisGroup</tt> template is used to register the analysis
1286group itself as well as add pass implementations to the analysis group. First,
1287an analysis should be registered, with a human readable name provided for it.
1288Unlike registration of passes, there is no command line argument to be specified
1289for the Analysis Group Interface itself, because it is "abstract":</p>
1290
1291<div class="doc_code"><pre>
1292 <b>static</b> RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>&gt; A("<i>Alias Analysis</i>");
1293</pre></div>
1294
1295<p>Once the analysis is registered, passes can declare that they are valid
1296implementations of the interface by using the following code:</p>
1297
1298<div class="doc_code"><pre>
1299<b>namespace</b> {
1300 //<i> Analysis Group implementations <b>must</b> be registered normally...</i>
1301 RegisterPass&lt;FancyAA&gt;
1302 B("<i>somefancyaa</i>", "<i>A more complex alias analysis implementation</i>");
1303
1304 //<i> Declare that we implement the AliasAnalysis interface</i>
1305 RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>&gt; C(B);
1306}
1307</pre></div>
1308
1309<p>This just shows a class <tt>FancyAA</tt> that is registered normally, then
1310uses the <tt>RegisterAnalysisGroup</tt> template to "join" the <tt><a
1311href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a></tt>
1312analysis group. Every implementation of an analysis group should join using
1313this template. A single pass may join multiple different analysis groups with
1314no problem.</p>
1315
1316<div class="doc_code"><pre>
1317<b>namespace</b> {
1318 //<i> Analysis Group implementations <b>must</b> be registered normally...</i>
1319 RegisterPass&lt;<a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a>&gt;
1320 D("<i>basicaa</i>", "<i>Basic Alias Analysis (default AA impl)</i>");
1321
1322 //<i> Declare that we implement the AliasAnalysis interface</i>
1323 RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>, <b>true</b>&gt; E(D);
1324}
1325</pre></div>
1326
1327<p>Here we show how the default implementation is specified (using the extra
1328argument to the <tt>RegisterAnalysisGroup</tt> template). There must be exactly
1329one default implementation available at all times for an Analysis Group to be
1330used. Only default implementation can derive from <tt>ImmutablePass</tt>.
1331Here we declare that the
1332 <tt><a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a></tt>
1333pass is the default implementation for the interface.</p>
1334
1335</div>
1336
1337<!-- *********************************************************************** -->
1338<div class="doc_section">
1339 <a name="passStatistics">Pass Statistics</a>
1340</div>
1341<!-- *********************************************************************** -->
1342
1343<div class="doc_text">
1344<p>The <a
1345href="http://llvm.org/doxygen/Statistic_8h-source.html"><tt>Statistic</tt></a>
1346class is designed to be an easy way to expose various success
1347metrics from passes. These statistics are printed at the end of a
1348run, when the -stats command line option is enabled on the command
1349line. See the <a href="http://llvm.org/docs/ProgrammersManual.html#Statistic">Statistics section</a> in the Programmer's Manual for details.
1350
1351</div>
1352
1353
1354<!-- *********************************************************************** -->
1355<div class="doc_section">
1356 <a name="passmanager">What PassManager does</a>
1357</div>
1358<!-- *********************************************************************** -->
1359
1360<div class="doc_text">
1361
1362<p>The <a
1363href="http://llvm.org/doxygen/PassManager_8h-source.html"><tt>PassManager</tt></a>
1364<a
1365href="http://llvm.org/doxygen/classllvm_1_1PassManager.html">class</a>
1366takes a list of passes, ensures their <a href="#interaction">prerequisites</a>
1367are set up correctly, and then schedules passes to run efficiently. All of the
1368LLVM tools that run passes use the <tt>PassManager</tt> for execution of these
1369passes.</p>
1370
1371<p>The <tt>PassManager</tt> does two main things to try to reduce the execution
1372time of a series of passes:</p>
1373
1374<ol>
1375<li><b>Share analysis results</b> - The PassManager attempts to avoid
1376recomputing analysis results as much as possible. This means keeping track of
1377which analyses are available already, which analyses get invalidated, and which
1378analyses are needed to be run for a pass. An important part of work is that the
1379<tt>PassManager</tt> tracks the exact lifetime of all analysis results, allowing
1380it to <a href="#releaseMemory">free memory</a> allocated to holding analysis
1381results as soon as they are no longer needed.</li>
1382
1383<li><b>Pipeline the execution of passes on the program</b> - The
1384<tt>PassManager</tt> attempts to get better cache and memory usage behavior out
1385of a series of passes by pipelining the passes together. This means that, given
1386a series of consequtive <a href="#FunctionPass"><tt>FunctionPass</tt></a>'s, it
1387will execute all of the <a href="#FunctionPass"><tt>FunctionPass</tt></a>'s on
1388the first function, then all of the <a
1389href="#FunctionPass"><tt>FunctionPass</tt></a>es on the second function,
1390etc... until the entire program has been run through the passes.
1391
1392<p>This improves the cache behavior of the compiler, because it is only touching
1393the LLVM program representation for a single function at a time, instead of
1394traversing the entire program. It reduces the memory consumption of compiler,
1395because, for example, only one <a
1396href="http://llvm.org/doxygen/classllvm_1_1DominatorSet.html"><tt>DominatorSet</tt></a>
John Criswell8a726152007-12-10 20:26:29 +0000[diff] [blame]1397needs to be calculated at a time. This also makes it possible to implement
1398some <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1399href="#SMP">interesting enhancements</a> in the future.</p></li>
1400
1401</ol>
1402
1403<p>The effectiveness of the <tt>PassManager</tt> is influenced directly by how
1404much information it has about the behaviors of the passes it is scheduling. For
1405example, the "preserved" set is intentionally conservative in the face of an
1406unimplemented <a href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a> method.
1407Not implementing when it should be implemented will have the effect of not
1408allowing any analysis results to live across the execution of your pass.</p>
1409
1410<p>The <tt>PassManager</tt> class exposes a <tt>--debug-pass</tt> command line
1411options that is useful for debugging pass execution, seeing how things work, and
1412diagnosing when you should be preserving more analyses than you currently are
1413(To get information about all of the variants of the <tt>--debug-pass</tt>
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]1414option, just type '<tt>opt -help-hidden</tt>').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1415
1416<p>By using the <tt>--debug-pass=Structure</tt> option, for example, we can see
1417how our <a href="#basiccode">Hello World</a> pass interacts with other passes.
1418Lets try it out with the <tt>gcse</tt> and <tt>licm</tt> passes:</p>
1419
1420<div class="doc_code"><pre>
1421$ opt -load ../../../Debug/lib/Hello.so -gcse -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
1422Module Pass Manager
1423 Function Pass Manager
1424 Dominator Set Construction
1425 Immediate Dominators Construction
1426 Global Common Subexpression Elimination
1427-- Immediate Dominators Construction
1428-- Global Common Subexpression Elimination
1429 Natural Loop Construction
1430 Loop Invariant Code Motion
1431-- Natural Loop Construction
1432-- Loop Invariant Code Motion
1433 Module Verifier
1434-- Dominator Set Construction
1435-- Module Verifier
1436 Bitcode Writer
1437--Bitcode Writer
1438</pre></div>
1439
1440<p>This output shows us when passes are constructed and when the analysis
1441results are known to be dead (prefixed with '<tt>--</tt>'). Here we see that
1442GCSE uses dominator and immediate dominator information to do its job. The LICM
1443pass uses natural loop information, which uses dominator sets, but not immediate
1444dominators. Because immediate dominators are no longer useful after the GCSE
1445pass, it is immediately destroyed. The dominator sets are then reused to
1446compute natural loop information, which is then used by the LICM pass.</p>
1447
1448<p>After the LICM pass, the module verifier runs (which is automatically added
1449by the '<tt>opt</tt>' tool), which uses the dominator set to check that the
1450resultant LLVM code is well formed. After it finishes, the dominator set
1451information is destroyed, after being computed once, and shared by three
1452passes.</p>
1453
1454<p>Lets see how this changes when we run the <a href="#basiccode">Hello
1455World</a> pass in between the two passes:</p>
1456
1457<div class="doc_code"><pre>
1458$ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
1459Module Pass Manager
1460 Function Pass Manager
1461 Dominator Set Construction
1462 Immediate Dominators Construction
1463 Global Common Subexpression Elimination
1464<b>-- Dominator Set Construction</b>
1465-- Immediate Dominators Construction
1466-- Global Common Subexpression Elimination
1467<b> Hello World Pass
1468-- Hello World Pass
1469 Dominator Set Construction</b>
1470 Natural Loop Construction
1471 Loop Invariant Code Motion
1472-- Natural Loop Construction
1473-- Loop Invariant Code Motion
1474 Module Verifier
1475-- Dominator Set Construction
1476-- Module Verifier
1477 Bitcode Writer
1478--Bitcode Writer
1479Hello: __main
1480Hello: puts
1481Hello: main
1482</pre></div>
1483
1484<p>Here we see that the <a href="#basiccode">Hello World</a> pass has killed the
1485Dominator Set pass, even though it doesn't modify the code at all! To fix this,
1486we need to add the following <a
1487href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a> method to our pass:</p>
1488
1489<div class="doc_code"><pre>
1490 <i>// We don't modify the program, so we preserve all analyses</i>
1491 <b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
1492 AU.setPreservesAll();
1493 }
1494</pre></div>
1495
1496<p>Now when we run our pass, we get this output:</p>
1497
1498<div class="doc_code"><pre>
1499$ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
1500Pass Arguments: -gcse -hello -licm
1501Module Pass Manager
1502 Function Pass Manager
1503 Dominator Set Construction
1504 Immediate Dominators Construction
1505 Global Common Subexpression Elimination
1506-- Immediate Dominators Construction
1507-- Global Common Subexpression Elimination
1508 Hello World Pass
1509-- Hello World Pass
1510 Natural Loop Construction
1511 Loop Invariant Code Motion
1512-- Loop Invariant Code Motion
1513-- Natural Loop Construction
1514 Module Verifier
1515-- Dominator Set Construction
1516-- Module Verifier
1517 Bitcode Writer
1518--Bitcode Writer
1519Hello: __main
1520Hello: puts
1521Hello: main
1522</pre></div>
1523
1524<p>Which shows that we don't accidentally invalidate dominator information
1525anymore, and therefore do not have to compute it twice.</p>
1526
1527</div>
1528
1529<!-- _______________________________________________________________________ -->
1530<div class="doc_subsubsection">
1531 <a name="releaseMemory">The <tt>releaseMemory</tt> method</a>
1532</div>
1533
1534<div class="doc_text">
1535
1536<div class="doc_code"><pre>
1537 <b>virtual void</b> releaseMemory();
1538</pre></div>
1539
1540<p>The <tt>PassManager</tt> automatically determines when to compute analysis
1541results, and how long to keep them around for. Because the lifetime of the pass
1542object itself is effectively the entire duration of the compilation process, we
1543need some way to free analysis results when they are no longer useful. The
1544<tt>releaseMemory</tt> virtual method is the way to do this.</p>
1545
1546<p>If you are writing an analysis or any other pass that retains a significant
1547amount of state (for use by another pass which "requires" your pass and uses the
1548<a href="#getAnalysis">getAnalysis</a> method) you should implement
Dan Gohmand72d8ea2009-06-15 18:22:49 +0000[diff] [blame]1549<tt>releaseMemory</tt> to, well, release the memory allocated to maintain this
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1550internal state. This method is called after the <tt>run*</tt> method for the
1551class, before the next call of <tt>run*</tt> in your pass.</p>
1552
1553</div>
1554
1555<!-- *********************************************************************** -->
1556<div class="doc_section">
1557 <a name="registering">Registering dynamically loaded passes</a>
1558</div>
1559<!-- *********************************************************************** -->
1560
1561<div class="doc_text">
1562
1563<p><i>Size matters</i> when constructing production quality tools using llvm,
1564both for the purposes of distribution, and for regulating the resident code size
1565when running on the target system. Therefore, it becomes desirable to
1566selectively use some passes, while omitting others and maintain the flexibility
1567to change configurations later on. You want to be able to do all this, and,
1568provide feedback to the user. This is where pass registration comes into
1569play.</p>
1570
1571<p>The fundamental mechanisms for pass registration are the
1572<tt>MachinePassRegistry</tt> class and subclasses of
1573<tt>MachinePassRegistryNode</tt>.</p>
1574
1575<p>An instance of <tt>MachinePassRegistry</tt> is used to maintain a list of
1576<tt>MachinePassRegistryNode</tt> objects. This instance maintains the list and
1577communicates additions and deletions to the command line interface.</p>
1578
1579<p>An instance of <tt>MachinePassRegistryNode</tt> subclass is used to maintain
1580information provided about a particular pass. This information includes the
1581command line name, the command help string and the address of the function used
1582to create an instance of the pass. A global static constructor of one of these
1583instances <i>registers</i> with a corresponding <tt>MachinePassRegistry</tt>,
1584the static destructor <i>unregisters</i>. Thus a pass that is statically linked
1585in the tool will be registered at start up. A dynamically loaded pass will
1586register on load and unregister at unload.</p>
1587
1588</div>
1589
1590<!-- _______________________________________________________________________ -->
1591<div class="doc_subsection">
1592 <a name="registering_existing">Using existing registries</a>
1593</div>
1594
1595<div class="doc_text">
1596
1597<p>There are predefined registries to track instruction scheduling
1598(<tt>RegisterScheduler</tt>) and register allocation (<tt>RegisterRegAlloc</tt>)
1599machine passes. Here we will describe how to <i>register</i> a register
1600allocator machine pass.</p>
1601
1602<p>Implement your register allocator machine pass. In your register allocator
1603.cpp file add the following include;</p>
1604
1605<div class="doc_code"><pre>
1606 #include "llvm/CodeGen/RegAllocRegistry.h"
1607</pre></div>
1608
1609<p>Also in your register allocator .cpp file, define a creator function in the
1610form; </p>
1611
1612<div class="doc_code"><pre>
1613 FunctionPass *createMyRegisterAllocator() {
1614 return new MyRegisterAllocator();
1615 }
1616</pre></div>
1617
1618<p>Note that the signature of this function should match the type of
1619<tt>RegisterRegAlloc::FunctionPassCtor</tt>. In the same file add the
1620"installing" declaration, in the form;</p>
1621
1622<div class="doc_code"><pre>
1623 static RegisterRegAlloc myRegAlloc("myregalloc",
1624 " my register allocator help string",
1625 createMyRegisterAllocator);
1626</pre></div>
1627
1628<p>Note the two spaces prior to the help string produces a tidy result on the
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]1629-help query.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1630
1631<div class="doc_code"><pre>
Duncan Sandsc5c38f02010-02-18 14:08:13 +0000[diff] [blame]1632$ llc -help
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1633 ...
Duncan Sands82400092010-02-18 14:37:52 +0000[diff] [blame]1634 -regalloc - Register allocator to use (default=linearscan)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1635 =linearscan - linear scan register allocator
1636 =local - local register allocator
1637 =simple - simple register allocator
1638 =myregalloc - my register allocator help string
1639 ...
1640</pre></div>
1641
1642<p>And that's it. The user is now free to use <tt>-regalloc=myregalloc</tt> as
1643an option. Registering instruction schedulers is similar except use the
1644<tt>RegisterScheduler</tt> class. Note that the
1645<tt>RegisterScheduler::FunctionPassCtor</tt> is significantly different from
1646<tt>RegisterRegAlloc::FunctionPassCtor</tt>.</p>
1647
1648<p>To force the load/linking of your register allocator into the llc/lli tools,
1649add your creator function's global declaration to "Passes.h" and add a "pseudo"
1650call line to <tt>llvm/Codegen/LinkAllCodegenComponents.h</tt>.</p>
1651
1652</div>
1653
1654
1655<!-- _______________________________________________________________________ -->
1656<div class="doc_subsection">
1657 <a name="registering_new">Creating new registries</a>
1658</div>
1659
1660<div class="doc_text">
1661
1662<p>The easiest way to get started is to clone one of the existing registries; we
1663recommend <tt>llvm/CodeGen/RegAllocRegistry.h</tt>. The key things to modify
1664are the class name and the <tt>FunctionPassCtor</tt> type.</p>
1665
1666<p>Then you need to declare the registry. Example: if your pass registry is
1667<tt>RegisterMyPasses</tt> then define;</p>
1668
1669<div class="doc_code"><pre>
1670MachinePassRegistry RegisterMyPasses::Registry;
1671</pre></div>
1672
1673<p>And finally, declare the command line option for your passes. Example:</p>
1674
1675<div class="doc_code"><pre>
1676 cl::opt&lt;RegisterMyPasses::FunctionPassCtor, false,
Dan Gohman8e58bc52008-10-14 17:00:38 +0000[diff] [blame]1677 RegisterPassParser&lt;RegisterMyPasses&gt; &gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1678 MyPassOpt("mypass",
1679 cl::init(&amp;createDefaultMyPass),
1680 cl::desc("my pass option help"));
1681</pre></div>
1682
1683<p>Here the command option is "mypass", with createDefaultMyPass as the default
1684creator.</p>
1685
1686</div>
1687
1688<!-- *********************************************************************** -->
1689<div class="doc_section">
1690 <a name="debughints">Using GDB with dynamically loaded passes</a>
1691</div>
1692<!-- *********************************************************************** -->
1693
1694<div class="doc_text">
1695
1696<p>Unfortunately, using GDB with dynamically loaded passes is not as easy as it
1697should be. First of all, you can't set a breakpoint in a shared object that has
1698not been loaded yet, and second of all there are problems with inlined functions
1699in shared objects. Here are some suggestions to debugging your pass with
1700GDB.</p>
1701
1702<p>For sake of discussion, I'm going to assume that you are debugging a
1703transformation invoked by <tt>opt</tt>, although nothing described here depends
1704on that.</p>
1705
1706</div>
1707
1708<!-- _______________________________________________________________________ -->
1709<div class="doc_subsubsection">
1710 <a name="breakpoint">Setting a breakpoint in your pass</a>
1711</div>
1712
1713<div class="doc_text">
1714
1715<p>First thing you do is start <tt>gdb</tt> on the <tt>opt</tt> process:</p>
1716
1717<div class="doc_code"><pre>
1718$ <b>gdb opt</b>
1719GNU gdb 5.0
1720Copyright 2000 Free Software Foundation, Inc.
1721GDB is free software, covered by the GNU General Public License, and you are
1722welcome to change it and/or distribute copies of it under certain conditions.
1723Type "show copying" to see the conditions.
1724There is absolutely no warranty for GDB. Type "show warranty" for details.
1725This GDB was configured as "sparc-sun-solaris2.6"...
1726(gdb)
1727</pre></div>
1728
1729<p>Note that <tt>opt</tt> has a lot of debugging information in it, so it takes
1730time to load. Be patient. Since we cannot set a breakpoint in our pass yet
1731(the shared object isn't loaded until runtime), we must execute the process, and
1732have it stop before it invokes our pass, but after it has loaded the shared
1733object. The most foolproof way of doing this is to set a breakpoint in
1734<tt>PassManager::run</tt> and then run the process with the arguments you
1735want:</p>
1736
1737<div class="doc_code"><pre>
1738(gdb) <b>break llvm::PassManager::run</b>
1739Breakpoint 1 at 0x2413bc: file Pass.cpp, line 70.
1740(gdb) <b>run test.bc -load $(LLVMTOP)/llvm/Debug/lib/[libname].so -[passoption]</b>
1741Starting program: opt test.bc -load $(LLVMTOP)/llvm/Debug/lib/[libname].so -[passoption]
1742Breakpoint 1, PassManager::run (this=0xffbef174, M=@0x70b298) at Pass.cpp:70
174370 bool PassManager::run(Module &amp;M) { return PM-&gt;run(M); }
1744(gdb)
1745</pre></div>
1746
1747<p>Once the <tt>opt</tt> stops in the <tt>PassManager::run</tt> method you are
1748now free to set breakpoints in your pass so that you can trace through execution
1749or do other standard debugging stuff.</p>
1750
1751</div>
1752
1753<!-- _______________________________________________________________________ -->
1754<div class="doc_subsubsection">
1755 <a name="debugmisc">Miscellaneous Problems</a>
1756</div>
1757
1758<div class="doc_text">
1759
1760<p>Once you have the basics down, there are a couple of problems that GDB has,
1761some with solutions, some without.</p>
1762
1763<ul>
1764<li>Inline functions have bogus stack information. In general, GDB does a
1765pretty good job getting stack traces and stepping through inline functions.
1766When a pass is dynamically loaded however, it somehow completely loses this
1767capability. The only solution I know of is to de-inline a function (move it
1768from the body of a class to a .cpp file).</li>
1769
1770<li>Restarting the program breaks breakpoints. After following the information
1771above, you have succeeded in getting some breakpoints planted in your pass. Nex
1772thing you know, you restart the program (i.e., you type '<tt>run</tt>' again),
1773and you start getting errors about breakpoints being unsettable. The only way I
1774have found to "fix" this problem is to <tt>delete</tt> the breakpoints that are
1775already set in your pass, run the program, and re-set the breakpoints once
1776execution stops in <tt>PassManager::run</tt>.</li>
1777
1778</ul>
1779
1780<p>Hopefully these tips will help with common case debugging situations. If
1781you'd like to contribute some tips of your own, just contact <a
1782href="mailto:sabre@nondot.org">Chris</a>.</p>
1783
1784</div>
1785
1786<!-- *********************************************************************** -->
1787<div class="doc_section">
1788 <a name="future">Future extensions planned</a>
1789</div>
1790<!-- *********************************************************************** -->
1791
1792<div class="doc_text">
1793
1794<p>Although the LLVM Pass Infrastructure is very capable as it stands, and does
1795some nifty stuff, there are things we'd like to add in the future. Here is
1796where we are going:</p>
1797
1798</div>
1799
1800<!-- _______________________________________________________________________ -->
1801<div class="doc_subsubsection">
1802 <a name="SMP">Multithreaded LLVM</a>
1803</div>
1804
1805<div class="doc_text">
1806
1807<p>Multiple CPU machines are becoming more common and compilation can never be
1808fast enough: obviously we should allow for a multithreaded compiler. Because of
1809the semantics defined for passes above (specifically they cannot maintain state
1810across invocations of their <tt>run*</tt> methods), a nice clean way to
1811implement a multithreaded compiler would be for the <tt>PassManager</tt> class
1812to create multiple instances of each pass object, and allow the separate
1813instances to be hacking on different parts of the program at the same time.</p>
1814
1815<p>This implementation would prevent each of the passes from having to implement
1816multithreaded constructs, requiring only the LLVM core to have locking in a few
1817places (for global resources). Although this is a simple extension, we simply
1818haven't had time (or multiprocessor machines, thus a reason) to implement this.
1819Despite that, we have kept the LLVM passes SMP ready, and you should too.</p>
1820
1821</div>
1822
1823<!-- *********************************************************************** -->
1824<hr>
1825<address>
1826 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]1827 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1828 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman5c1cc642008-12-11 18:23:24 +0000[diff] [blame]1829 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1830
1831 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1832 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
1833 Last modified: $Date$
1834</address>
1835
1836</body>
1837</html>