| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 1 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 2 | <html> | 
|  | 3 | <head> | 
|  | 4 | <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /> | 
|  | 5 | <link rel="stylesheet" href="llvm.css" type="text/css" media="screen" /> | 
|  | 6 | <title>Alias Analysis Infrastructure in LLVM</title> | 
|  | 7 | </head> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 8 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 9 | <body> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 10 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 11 | <div class="doc_title"> | 
|  | 12 | Alias Analysis Infrastructure in LLVM | 
|  | 13 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 14 |  | 
|  | 15 | <ol> | 
|  | 16 | <li><a href="#introduction">Introduction</a> | 
|  | 17 |  | 
|  | 18 | <li><a href="#overview">AliasAnalysis Overview</a> | 
|  | 19 | <ul> | 
|  | 20 | <li><a href="#pointers">Representation of Pointers</a> | 
|  | 21 | <li><a href="#MustMayNo">Must, May, and No Alias Responses</a> | 
|  | 22 | <li><a href="#ModRefInfo">The <tt>getModRefInfo</tt> methods</a> | 
|  | 23 | </ul> | 
|  | 24 |  | 
|  | 25 | <li><a href="#writingnew">Writing a new AliasAnalysis Implementation</a> | 
|  | 26 | <ul> | 
|  | 27 | <li><a href="#passsubclasses">Different Pass styles</a> | 
|  | 28 | <li><a href="#requiredcalls">Required initialization calls</a> | 
|  | 29 | <li><a href="#interfaces">Interfaces which may be specified</a> | 
|  | 30 | <li><a href="#chaining">The AliasAnalysis chaining behavior</a> | 
|  | 31 | <li><a href="#implefficiency">Efficiency Issues</a> | 
|  | 32 | </ul> | 
|  | 33 |  | 
|  | 34 | <li><a href="#using">Using AliasAnalysis results</a> | 
|  | 35 | <ul> | 
|  | 36 | <li><a href="#loadvn">Using the <tt>-load-vn</tt> Pass</a> | 
|  | 37 | <li><a href="#ast">Using the <tt>AliasSetTracker</tt> class</a> | 
|  | 38 | <li><a href="#direct">Using the AliasAnalysis interface directly</a> | 
|  | 39 | </ul> | 
|  | 40 | <li><a href="#tools">Helpful alias analysis related tools</a> | 
|  | 41 | <ul> | 
|  | 42 | <li><a href="#no-aa">The <tt>-no-aa</tt> pass</a> | 
|  | 43 | <li><a href="#print-alias-sets">The <tt>-print-alias-sets</tt> pass</a> | 
|  | 44 | <li><a href="#count-aa">The <tt>-count-aa</tt> pass</a> | 
|  | 45 | <li><a href="#aa-eval">The <tt>-aa-eval</tt> pass</a> | 
|  | 46 | </ul> | 
|  | 47 | </ul> | 
|  | 48 |  | 
|  | 49 | <p><b>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></b><p> | 
|  | 50 | </ol><p> | 
|  | 51 |  | 
|  | 52 |  | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 53 | <!-- *********************************************************************** --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 54 | <div class="doc_section"> | 
|  | 55 | <a name="introduction">Introduction</a> | 
|  | 56 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 57 | <!-- *********************************************************************** --> | 
|  | 58 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 59 | <div class="doc_text"> | 
|  | 60 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 61 | Alias Analysis (or Pointer Analysis) is a technique which attempts to determine | 
|  | 62 | whether or not two pointers ever can point to the same object in memory. | 
|  | 63 | Traditionally, Alias Analyses respond to a query with either a <a | 
|  | 64 | href="#MustNoMay">Must, May, or No</a> alias response, indicating that two | 
|  | 65 | pointers do point to the same object, might point to the same object, or are | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 66 | known not to point to the same object. | 
|  | 67 | </p> | 
|  | 68 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 69 | The <a href="/doxygen/classAliasAnalysis.html">AliasAnalysis</a> class is the | 
|  | 70 | centerpiece of the LLVM Alias Analysis related infrastructure.  This class is | 
|  | 71 | the common interface between clients of alias analysis information and the | 
|  | 72 | implementations providing it.  In addition to simple alias analysis information, | 
|  | 73 | this class exposes Mod/Ref information from those implementations which can | 
|  | 74 | provide it, allowing for powerful analyses and transformations to work well | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 75 | together. | 
|  | 76 | </p> | 
|  | 77 | <p> | 
| Misha Brukman | 5560c9d | 2003-08-18 14:43:39 +0000 | [diff] [blame] | 78 | This document contains information necessary to successfully implement this | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 79 | interface, use it, and to test both sides.  It also explains some of the finer | 
|  | 80 | points about what exactly results mean.  If you feel that something is unclear | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 81 | or should be added, please <a href="mailto:sabre@nondot.org">let me | 
|  | 82 | know</a>. | 
|  | 83 | </p> | 
|  | 84 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 85 |  | 
|  | 86 | <!-- *********************************************************************** --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 87 | <div class="doc_section"> | 
|  | 88 | <a name="overview">AliasAnalysis Overview</a> | 
|  | 89 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 90 | <!-- *********************************************************************** --> | 
|  | 91 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 92 | <div class="doc_text"> | 
|  | 93 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 94 | The <a href="/doxygen/classAliasAnalysis.html">AliasAnalysis</a> class defines | 
|  | 95 | the interface that Alias Analysis implementations should support.  This class | 
|  | 96 | exports two important enums: <tt>AliasResult</tt> and <tt>ModRefResult</tt> | 
|  | 97 | which represent the result of an alias query or a mod/ref query, | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 98 | respectively. | 
|  | 99 | </p> | 
|  | 100 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 101 | The AliasAnalysis interface exposes information about memory, represented in | 
|  | 102 | several different ways.  In particular, memory objects are represented as a | 
|  | 103 | starting address and size, and function calls are represented as the actual | 
|  | 104 | <tt>call</tt> or <tt>invoke</tt> instructions that performs the call.  The | 
|  | 105 | AliasAnalysis interface also exposes some helper methods which allow you to get | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 106 | mod/ref information for arbitrary instructions. | 
|  | 107 | </p> | 
|  | 108 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 109 |  | 
|  | 110 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 111 | <div class="doc_subsection"> | 
|  | 112 | <a name="pointers">Representation of Pointers</a> | 
|  | 113 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 114 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 115 | <div class="doc_text"> | 
|  | 116 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 117 | Most importantly, the AliasAnalysis class provides several methods which are | 
|  | 118 | used to query whether or not pointers alias, whether function calls can modify | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 119 | or read memory, etc. | 
|  | 120 | </p> | 
|  | 121 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 122 | Representing memory objects as a starting address and a size is critically | 
|  | 123 | important for precise Alias Analyses.  For example, consider this (silly) C | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 124 | code: | 
|  | 125 | </p> | 
|  | 126 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 127 | <pre> | 
|  | 128 | int i; | 
|  | 129 | char C[2]; | 
|  | 130 | char A[10]; | 
|  | 131 | /* ... */ | 
|  | 132 | for (i = 0; i != 10; ++i) { | 
|  | 133 | C[0] = A[i];          /* One byte store */ | 
|  | 134 | C[1] = A[9-i];        /* One byte store */ | 
|  | 135 | } | 
|  | 136 | </pre> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 137 | </p> | 
|  | 138 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 139 | In this case, the <tt>basicaa</tt> pass will disambiguate the stores to | 
|  | 140 | <tt>C[0]</tt> and <tt>C[1]</tt> because they are accesses to two distinct | 
|  | 141 | locations one byte apart, and the accesses are each one byte.  In this case, the | 
|  | 142 | LICM pass can use store motion to remove the stores from the loop.  In | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 143 | constrast, the following code: | 
|  | 144 | </p> | 
|  | 145 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 146 | <pre> | 
|  | 147 | int i; | 
|  | 148 | char C[2]; | 
|  | 149 | char A[10]; | 
|  | 150 | /* ... */ | 
|  | 151 | for (i = 0; i != 10; ++i) { | 
|  | 152 | ((short*)C)[0] = A[i];  /* Two byte store! */ | 
|  | 153 | C[1] = A[9-i];          /* One byte store */ | 
|  | 154 | } | 
|  | 155 | </pre> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 156 | </p> | 
|  | 157 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 158 | In this case, the two stores to C do alias each other, because the access to the | 
|  | 159 | <tt>&C[0]</tt> element is a two byte access.  If size information wasn't | 
|  | 160 | available in the query, even the first case would have to conservatively assume | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 161 | that the accesses alias. | 
|  | 162 | </p> | 
|  | 163 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 164 |  | 
|  | 165 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 166 | <div class="doc_subsection"> | 
|  | 167 | <a name="MustMayNo">Must, May, and No Alias Responses</a> | 
|  | 168 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 169 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 170 | <div class="doc_text"> | 
|  | 171 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 172 | An Alias Analysis implementation can return one of three responses: MustAlias, | 
|  | 173 | MayAlias, and NoAlias.  The No and May alias results are obvious: if the two | 
|  | 174 | pointers may never equal each other, return NoAlias, if they might, return | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 175 | MayAlias. | 
|  | 176 | </p> | 
|  | 177 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 178 | The Must Alias response is trickier though.  In LLVM, the Must Alias response | 
|  | 179 | may only be returned if the two memory objects are guaranteed to always start at | 
|  | 180 | exactly the same location.  If two memory objects overlap, but do not start at | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 181 | the same location, MayAlias must be returned. | 
|  | 182 | </p> | 
|  | 183 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 184 |  | 
|  | 185 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 186 | <div class="doc_subsection"> | 
|  | 187 | <a name="ModRefInfo">The <tt>getModRefInfo</tt> methods</a> | 
|  | 188 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 189 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 190 | <div class="doc_text"> | 
|  | 191 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 192 | The <tt>getModRefInfo</tt> methods return information about whether the | 
|  | 193 | execution of an instruction can read or modify a memory location.  Mod/Ref | 
|  | 194 | information is always conservative: if an action <b>may</b> read a location, Ref | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 195 | is returned. | 
|  | 196 | </p> | 
|  | 197 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 198 |  | 
|  | 199 | <!-- *********************************************************************** --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 200 | <div class="doc_section"> | 
|  | 201 | <a name="writingnew">Writing a new AliasAnalysis Implementation</a> | 
|  | 202 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 203 | <!-- *********************************************************************** --> | 
|  | 204 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 205 | <div class="doc_text"> | 
|  | 206 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 207 | Writing a new alias analysis implementation for LLVM is quite straight-forward. | 
|  | 208 | There are already several implementations that you can use for examples, and the | 
|  | 209 | following information should help fill in any details.  For a minimal example, | 
|  | 210 | take a look at the <a href="/doxygen/structNoAA.html"><tt>no-aa</tt></a> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 211 | implementation. | 
|  | 212 | </p> | 
|  | 213 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 214 |  | 
|  | 215 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 216 | <div class="doc_subsection"> | 
|  | 217 | <a name="passsubclasses">Different Pass styles</a> | 
|  | 218 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 219 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 220 | <div class="doc_text"> | 
|  | 221 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 222 | The first step to determining what type of <a href="WritingAnLLVMPass.html">LLVM | 
|  | 223 | pass</a> you need to use for your Alias Analysis.  As is the case with most | 
|  | 224 | other analyses and transformations, the answer should be fairly obvious from | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 225 | what type of problem you are trying to solve: | 
|  | 226 | </p> | 
|  | 227 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 228 | <ol> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 229 | <li>If you require interprocedural analysis, it should be a | 
|  | 230 | <tt>Pass</tt>.</li> | 
|  | 231 | <li>If you are a global analysis, subclass <tt>FunctionPass</tt>.</li> | 
|  | 232 | <li>If you are a local pass, subclass <tt>BasicBlockPass</tt>.</li> | 
|  | 233 | <li>If you don't need to look at the program at all, subclass | 
|  | 234 | <tt>ImmutablePass</tt>.</li> | 
|  | 235 | </ol> | 
|  | 236 | </p> | 
|  | 237 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 238 | In addition to the pass that you subclass, you should also inherit from the | 
| Misha Brukman | 700fd49 | 2003-05-07 21:47:16 +0000 | [diff] [blame] | 239 | <tt>AliasAnalysis</tt> interface, of course, and use the | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 240 | <tt>RegisterAnalysisGroup</tt> template to register as an implementation of | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 241 | <tt>AliasAnalysis</tt>. | 
|  | 242 | </p> | 
|  | 243 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 244 |  | 
|  | 245 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 246 | <div class="doc_subsection"> | 
|  | 247 | <a name="requiredcalls">Required initialization calls</a> | 
|  | 248 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 249 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 250 | <div class="doc_text"> | 
|  | 251 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 252 | Your subclass of AliasAnalysis is required to invoke two methods on the | 
|  | 253 | AliasAnalysis base class: <tt>getAnalysisUsage</tt> and | 
|  | 254 | <tt>InitializeAliasAnalysis</tt>.  In particular, your implementation of | 
|  | 255 | <tt>getAnalysisUsage</tt> should explicitly call into the | 
|  | 256 | <tt>AliasAnalysis::getAnalysisUsage</tt> method in addition to doing any | 
|  | 257 | declaring any pass dependencies your pass has.  Thus you should have something | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 258 | like this: | 
|  | 259 | </p> | 
|  | 260 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 261 | <pre> | 
|  | 262 | void getAnalysisUsage(AnalysisUsage &AU) const { | 
|  | 263 | AliasAnalysis::getAnalysisUsage(AU); | 
|  | 264 | <i>// declare your dependencies here.</i> | 
|  | 265 | } | 
|  | 266 | </pre> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 267 | </p> | 
|  | 268 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 269 | Additionally, your must invoke the <tt>InitializeAliasAnalysis</tt> method from | 
|  | 270 | your analysis run method (<tt>run</tt> for a <tt>Pass</tt>, | 
|  | 271 | <tt>runOnFunction</tt> for a <tt>FunctionPass</tt>, <tt>runOnBasicBlock</tt> for | 
|  | 272 | a <tt>BasicBlockPass</tt>, or <tt>InitializeAliasAnalysis</tt> for an | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 273 | <tt>ImmutablePass</tt>).  For example (as part of a <tt>Pass</tt>): | 
|  | 274 | </p> | 
|  | 275 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 276 | <pre> | 
|  | 277 | bool run(Module &M) { | 
|  | 278 | InitializeAliasAnalysis(this); | 
|  | 279 | <i>// Perform analysis here...</i> | 
|  | 280 | return false; | 
|  | 281 | } | 
|  | 282 | </pre> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 283 | </p> | 
|  | 284 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 285 |  | 
|  | 286 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 287 | <div class="doc_subsection"> | 
|  | 288 | <a name="interfaces">Interfaces which may be specified</a> | 
|  | 289 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 290 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 291 | <div class="doc_text"> | 
|  | 292 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 293 | All of the <a href="/doxygen/classAliasAnalysis.html">AliasAnalysis</a> virtual | 
|  | 294 | methods default to providing conservatively correct information (returning "May" | 
|  | 295 | Alias and "Mod/Ref" for alias and mod/ref queries respectively).  Depending on | 
|  | 296 | the capabilities of the analysis you are implementing, you just override the | 
|  | 297 | interfaces you can improve. | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 298 | </p> | 
|  | 299 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 300 |  | 
|  | 301 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 302 | <div class="doc_subsection"> | 
|  | 303 | <a name="chaining">The AliasAnalysis chaining behavior</a> | 
|  | 304 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 305 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 306 | <div class="doc_text"> | 
|  | 307 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 308 | With only two special exceptions (the <tt>basicaa</tt> and <a | 
|  | 309 | href="#no-aa"><tt>no-aa</tt></a> passes) every alias analysis pass should chain | 
|  | 310 | to another alias analysis implementation (for example, you could specify | 
|  | 311 | "<tt>-basic-aa -ds-aa -andersens-aa -licm</tt>" to get the maximum benefit from | 
|  | 312 | the three alias analyses).  To do this, simply "Require" AliasAnalysis in your | 
|  | 313 | <tt>getAnalysisUsage</tt> method, and if you need to return a conservative | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 314 | MayAlias or Mod/Ref result, simply chain to a lower analysis. | 
|  | 315 | </p> | 
|  | 316 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 317 |  | 
|  | 318 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 319 | <div class="doc_subsection"> | 
|  | 320 | <a name="implefficiency">Efficiency Issues</a> | 
|  | 321 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 322 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 323 | <div class="doc_text"> | 
|  | 324 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 325 | From the LLVM perspective, the only thing you need to do to provide an efficient | 
|  | 326 | alias analysis is to make sure that alias analysis <b>queries</b> are serviced | 
|  | 327 | quickly.  The actual calculation of the alias analysis results (the "run" | 
|  | 328 | method) is only performed once, but many (perhaps duplicate) queries may be | 
|  | 329 | performed.  Because of this, try to move as much computation to the run method | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 330 | as possible (within reason). | 
|  | 331 | </p> | 
|  | 332 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 333 |  | 
|  | 334 | <!-- *********************************************************************** --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 335 | <div class="doc_section"> | 
|  | 336 | <a name="using">Using AliasAnalysis results</a> | 
|  | 337 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 338 | <!-- *********************************************************************** --> | 
|  | 339 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 340 | <div class="doc_text"> | 
|  | 341 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 342 | There are several different ways to use alias analysis results.  In order of | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 343 | preference, these are... | 
|  | 344 | </p> | 
|  | 345 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 346 |  | 
|  | 347 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 348 | <div class="doc_subsection"> | 
|  | 349 | <a name="loadvn">Using the <tt>-load-vn</tt> Pass</a> | 
|  | 350 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 351 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 352 | <div class="doc_text"> | 
|  | 353 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 354 | The <tt>load-vn</tt> pass uses alias analysis to provide value numbering | 
|  | 355 | information for <tt>load</tt> instructions.  If your analysis or transformation | 
|  | 356 | can be modelled in a form that uses value numbering information, you don't have | 
|  | 357 | to do anything special to handle load instructions: just use the | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 358 | <tt>load-vn</tt> pass, which uses alias analysis. | 
|  | 359 | </p> | 
|  | 360 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 361 |  | 
|  | 362 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 363 | <div class="doc_subsection"> | 
|  | 364 | <a name="ast">Using the <tt>AliasSetTracker</tt> class</a> | 
|  | 365 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 366 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 367 | <div class="doc_text"> | 
|  | 368 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 369 | Many transformations need information about alias <b>sets</b> that are active in | 
|  | 370 | some scope, rather than information about pairwise aliasing.  The <tt><a | 
|  | 371 | href="/doxygen/classAliasSetTracker.html">AliasSetTracker</a></tt> class is used | 
|  | 372 | to efficiently build these Alias Sets from the pairwise alias analysis | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 373 | information provided by the AliasAnalysis interface. | 
|  | 374 | </p> | 
|  | 375 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 376 | First you initialize the AliasSetTracker by use the "<tt>add</tt>" methods to | 
|  | 377 | add information about various potentially aliasing instructions in the scope you | 
|  | 378 | are interested in.  Once all of the alias sets are completed, your pass should | 
|  | 379 | simply iterate through the constructed alias sets, using the AliasSetTracker | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 380 | <tt>begin()</tt>/<tt>end()</tt> methods. | 
|  | 381 | </p> | 
|  | 382 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 383 | The <tt>AliasSet</tt>s formed by the <tt>AliasSetTracker</tt> are guaranteed to | 
|  | 384 | be disjoint, calculate mod/ref information for the set, and keep track of | 
|  | 385 | whether or not all of the pointers in the set are Must aliases.  The | 
|  | 386 | AliasSetTracker also makes sure that sets are properly folded due to call | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 387 | instructions, and can provide a list of pointers in each set. | 
|  | 388 | </p> | 
|  | 389 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 390 | As an example user of this, the <a href="/doxygen/structLICM.html">Loop | 
|  | 391 | Invariant Code Motion</a> pass uses AliasSetTrackers to build alias information | 
|  | 392 | about each loop nest.  If an AliasSet in a loop is not modified, then all load | 
|  | 393 | instructions from that set may be hoisted out of the loop.  If any alias sets | 
|  | 394 | are stored <b>and</b> are must alias sets, then the stores may be sunk to | 
|  | 395 | outside of the loop.  Both of these transformations obviously only apply if the | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 396 | pointer argument is loop-invariant. | 
|  | 397 | </p> | 
|  | 398 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 399 |  | 
|  | 400 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 401 | <div class="doc_subsection"> | 
|  | 402 | <a name="direct">Using the AliasAnalysis interface directly</a> | 
|  | 403 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 404 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 405 | <div class="doc_text"> | 
|  | 406 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 407 | As a last resort, your pass could use the AliasAnalysis interface directly to | 
|  | 408 | service your pass.  If you find the need to do this, please <a | 
|  | 409 | href="mailto:sabre@nondot.org">let me know</a> so I can see if something new | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 410 | needs to be added to LLVM. | 
|  | 411 | </p> | 
|  | 412 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 413 |  | 
|  | 414 | <!-- *********************************************************************** --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 415 | <div class="doc_section"> | 
|  | 416 | <a name="tools">Helpful alias-analysis-related tools</a> | 
|  | 417 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 418 | <!-- *********************************************************************** --> | 
|  | 419 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 420 | <div class="doc_text"> | 
|  | 421 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 422 | If you're going to be working with the AliasAnalysis infrastructure, there are | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 423 | several nice tools that may be useful for you and are worth knowing about... | 
|  | 424 | </p> | 
|  | 425 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 426 |  | 
|  | 427 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 428 | <div class="doc_subsection"> | 
|  | 429 | <a name="no-aa">The <tt>-no-aa</tt> pass</a> | 
|  | 430 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 431 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 432 | <div class="doc_text"> | 
|  | 433 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 434 | The <tt>-no-aa</tt> analysis is just like what it sounds: an alias analysis that | 
|  | 435 | never returns any useful information.  This pass can be useful if you think that | 
|  | 436 | alias analysis is doing something wrong and are trying to narrow down a problem. | 
|  | 437 | If you don't specify an alias analysis, the default will be to use the | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 438 | <tt>basicaa</tt> pass which does quite a bit of disambiguation on its own. | 
|  | 439 | </p> | 
|  | 440 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 441 |  | 
|  | 442 |  | 
|  | 443 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 444 | <div class="doc_subsection"> | 
|  | 445 | <a name="print-alias-sets">The <tt>-print-alias-sets</tt> pass</a> | 
|  | 446 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 447 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 448 | <div class="doc_text"> | 
|  | 449 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 450 | The <tt>-print-alias-sets</tt> pass is exposed as part of the <tt>analyze</tt> | 
|  | 451 | tool to print out the Alias Sets formed by the <a | 
|  | 452 | href="#ast"><tt>AliasSetTracker</tt></a> class.  This is useful if you're using | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 453 | the <tt>AliasSetTracker</tt>. | 
|  | 454 | </p> | 
|  | 455 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 456 |  | 
|  | 457 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 458 | <div class="doc_subsection"> | 
|  | 459 | <a name="count-aa">The <tt>-count-aa</tt> pass</a> | 
|  | 460 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 461 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 462 | <div class="doc_text"> | 
|  | 463 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 464 | The <tt>-count-aa</tt> pass is useful to see how many queries a particular pass | 
|  | 465 | is making and what kinds of responses are returned by the alias analysis.  An | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 466 | example usage is: | 
|  | 467 | </p> | 
|  | 468 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 469 | <pre> | 
|  | 470 | $ opt -basicaa -count-aa -ds-aa -count-aa -licm | 
|  | 471 | </pre> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 472 | </p> | 
|  | 473 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 474 | Which will print out how many queries (and what responses are returned) by the | 
|  | 475 | <tt>-licm</tt> pass (of the <tt>-ds-aa</tt> pass) and how many queries are made | 
|  | 476 | of the <tt>-basicaa</tt> pass by the <tt>-ds-aa</tt> pass.  This can be useful | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 477 | when evaluating an alias analysis for precision. | 
|  | 478 | </p> | 
|  | 479 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 480 |  | 
|  | 481 | <!-- ======================================================================= --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 482 | <div class="doc_subsection"> | 
|  | 483 | <a name="aa-eval">The <tt>-aa-eval</tt> pass</a> | 
|  | 484 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 485 |  | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 486 | <div class="doc_text"> | 
|  | 487 | <p> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 488 | The <tt>-aa-eval</tt> pass simply iterates through all pairs of pointers in a | 
|  | 489 | function and asks an alias analysis whether or not the pointers alias.  This | 
|  | 490 | gives an indication of the precision of the alias analysis.  Statistics are | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 491 | printed. | 
|  | 492 | </p> | 
|  | 493 | </div> | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 494 |  | 
| Chris Lattner | 9f64875 | 2003-03-04 19:37:49 +0000 | [diff] [blame] | 495 | <!-- *********************************************************************** --> | 
|  | 496 |  | 
|  | 497 | <hr><font size=-1> | 
|  | 498 | <address><a href="mailto:sabre@nondot.org">Chris Lattner</a></address> | 
|  | 499 | <!-- Created: Wed Feb 26 10:40:50 CST 2003 --> | 
|  | 500 | <!-- hhmts start --> | 
|  | 501 | Last modified: Tue Mar  4 13:36:53 CST 2003 | 
|  | 502 | <!-- hhmts end --> | 
| Misha Brukman | cd11e45 | 2003-10-22 23:27:16 +0000 | [diff] [blame^] | 503 | </font> | 
|  | 504 |  | 
|  | 505 | </body> | 
|  | 506 | </html> |