Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 1 | <?xml version="1.0" encoding="ISO-8859-1" ?>
|
| 2 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
|
| 3 | <html xmlns="http://www.w3.org/1999/xhtml" lang="en">
|
| 4 | <head>
|
| 5 | <meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1" />
|
| 6 | <link rel="stylesheet" href=".resources/doc.css" charset="ISO-8859-1" type="text/css" />
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 7 | <link rel="stylesheet" href="../coverage/.resources/prettify.css" charset="ISO-8859-1" type="text/css" />
|
Marc R. Hoffmann | d7d2f75 | 2010-05-06 21:12:31 +0000 | [diff] [blame] | 8 | <link rel="shortcut icon" href=".resources/report.gif" type="image/gif" />
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 9 | <script type="text/javascript" src="../coverage/.resources/prettify.js"></script>
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 10 | <title>JaCoCo - Control Flow Analysis</title>
|
| 11 | </head>
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 12 | <body onload="prettyPrint()">
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 13 |
|
| 14 | <div class="breadcrumb">
|
Marc R. Hoffmann | d7d2f75 | 2010-05-06 21:12:31 +0000 | [diff] [blame] | 15 | <a href="../index.html" class="el_report">JaCoCo</a> >
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 16 | <a href="index.html" class="el_group">Documentation</a> >
|
| 17 | <span class="el_source">Control Flow Analysis</span>
|
| 18 | </div>
|
| 19 | <div id="content">
|
| 20 |
|
| 21 | <h1>Control Flow Analysis for Java Methods</h1>
|
| 22 |
|
| 23 | <p style="font-weight:bold;">
|
| 24 | DRAFT - This document does not reflect the current JaCoCo implementation.
|
| 25 | </p>
|
| 26 |
|
| 27 | <p class="hint">
|
| 28 | Implementing a coverage tool for branch coverage requires detailed analysis
|
| 29 | of the internal control flow of Java methods. Due to the architecture of
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 30 | JaCoCo this analysis needs to happen on compiled class files (bytecode).
|
| 31 | This document defines graph structures for control flow analysis of Java
|
| 32 | bytecode and discusses strategies for probe insertion.
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 33 | Marc R. Hoffmann, July 2010
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 34 | </p>
|
| 35 |
|
| 36 | <h2>Motivation and Requirements</h2>
|
| 37 |
|
| 38 | <ul>
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 39 | <li>Branch Coverage</li>
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 40 | <li>Exception Detection</li>
|
| 41 | </ul>
|
| 42 |
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 43 | <h2>From Statement Coverage to Branch Coverage</h2>
|
| 44 |
|
| 45 | <p>
|
| 46 | A
|
| 47 | JaCoCo till version 0.4.x provides statement coverage
|
| 48 | As as starting point
|
| 49 | differnce between statement coverage and branch coverage.
|
| 50 | probe insertion strategy.
|
| 51 |
|
| 52 | </p>
|
| 53 |
|
| 54 | <pre class="source lang-java">
|
| 55 | <span class="nr"> 1</span>public void example() {
|
| 56 | <span class="nr"> 2</span> a();
|
| 57 | <span class="nr"> 3</span> if (condition()) {
|
| 58 | <span class="nr"> 4</span> b();
|
| 59 | <span class="nr"> 5</span> }
|
| 60 | <span class="nr"> 6</span> c();
|
| 61 | <span class="nr"> 7</span>}
|
| 62 | </pre>
|
| 63 |
|
| 64 |
|
| 65 | <pre class="source">
|
| 66 | <span class="nr"> 1</span>public example() : void
|
| 67 | <span class="nr"> 2</span> L0
|
| 68 | <span class="nr"> 3</span> INVOKESTATIC Example.a() : void
|
| 69 | <span class="nr"> 4</span> L1
|
| 70 | <span class="nr"> 5</span> INVOKESTATIC Example.condition() : boolean
|
| 71 | <span class="nr"> 6</span> IFEQ L3
|
| 72 | <span class="nr"> 7</span> L2
|
| 73 | <span class="nr"> 8</span> INVOKESTATIC Example.b() : void
|
| 74 | <span class="nr"> 9</span> L3
|
| 75 | <span class="nr"> 10</span> INVOKESTATIC Example.c() : void
|
| 76 | <span class="nr"> 11</span> L4
|
| 77 | <span class="nr"> 11</span> RETURN
|
| 78 | </pre>
|
| 79 |
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 80 | <h2>The Control Flow Graph</h2>
|
| 81 |
|
| 82 | <ul>
|
| 83 | <li>Virtual Entry and Exit Nodes</li>
|
| 84 | <li>Subsequent Instructions</li>
|
| 85 | <li>(Conditional) Jump</li>
|
| 86 | <li>Table/Lookup Switch</li>
|
| 87 | <li>Exception Handlers</li>
|
| 88 | <li>Unhandled Exceptions</li>
|
| 89 | </ul>
|
| 90 |
|
| 91 | <h2>Probe Insertion</h2>
|
| 92 |
|
| 93 | <p>
|
| 94 | Code coverage analysis is a runtime metric that provides execution details
|
| 95 | of the software under test. This requires detailed recording about the
|
Marc R. Hoffmann | ae8c2a8 | 2010-10-05 16:20:10 +0000 | [diff] [blame^] | 96 | instructions (instruction coverage) that have been executed. For branch
|
| 97 | coverage also the outcome of decisions has to be recorded. In any case
|
| 98 | execution data is collected by so called probes:
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 99 | </p>
|
| 100 |
|
| 101 | <p class="hint">
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 102 | A <b>probe</b> is a sequence of bytecode instructions that can be inserted
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 103 | into a Java method. When the probe is executed, this fact is recorded and can
|
| 104 | be reported by the coverage runtime.
|
| 105 | </p>
|
| 106 |
|
| 107 | <p>
|
| 108 | The only purpose of the probe is to record that it has been executed at least
|
| 109 | once. The probe does not record the number of times it has been called or
|
| 110 | collect any timing information. The latter is out of scope for code coverage
|
| 111 | analysis and more in the objective of a performance analysis tool. Typically
|
| 112 | multiple probes needs to be inserted into each method, therefore probes needs
|
| 113 | to be identified. Also the probe implementation and the storage mechanism it
|
| 114 | depends on needs to be thread safe as multi-threaded execution is a common
|
| 115 | scenario for java applications (albeit not for plain unit tests). Probes must
|
| 116 | not have any side effects on the original code of the method. Also they should
|
| 117 | add minimal overhead.
|
| 118 | </p>
|
| 119 |
|
| 120 | <p>
|
| 121 | So to summarize the requirements for execution probes:
|
| 122 | </p>
|
| 123 |
|
| 124 | <ul>
|
| 125 | <li>Record execution</li>
|
| 126 | <li>Identification for different probes</li>
|
| 127 | <li>Thread safe</li>
|
| 128 | <li>No side effects on application code</li>
|
| 129 | <li>Minimal runtime overhead</li>
|
| 130 | </ul>
|
| 131 |
|
| 132 | <p>
|
| 133 | JaCoCo implements probes with a <code>boolean[]</code> array instance per
|
| 134 | class. Each probe corresponds to a entry in this array. Whenever the probe is
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 135 | executed the entry is set to <code>true</code> with the following four
|
| 136 | bytecode instructions:
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 137 | </p>
|
| 138 |
|
| 139 | <pre class="source">
|
| 140 | ALOAD probearray
|
| 141 | xPUSH probeid
|
| 142 | ICONST_1
|
| 143 | BASTORE
|
| 144 | </pre>
|
| 145 |
|
| 146 | <p>
|
| 147 | Note that this probe code is thread safe, does not modify the operand stack
|
| 148 | or modify local variables and is also thread safe. It does also not leave the
|
| 149 | method though an external call. The only prerequisite is that the probe array
|
| 150 | is available as a local variable. For this at the beginning of each method
|
| 151 | additional instrumentation code needs to be added to obtain the array instance
|
Marc R. Hoffmann | 9772829 | 2010-04-27 20:29:23 +0000 | [diff] [blame] | 152 | associated with the belonging class. To avoid code duplication the
|
| 153 | initialization is delegated to a static private method
|
| 154 | <code>$jacocoinit()</code> which is added to every non-interface class.
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 155 | </p>
|
| 156 |
|
| 157 | <p>
|
| 158 | The size of the probe code above depends on the position of the probe array
|
| 159 | variable and the value of the probe identifier as different opcodes can be
|
| 160 | used. As calculated in the table below the overhead per probe ranges between 4
|
| 161 | and 7 bytes of additional bytecode:
|
| 162 | </p>
|
| 163 |
|
| 164 | <table class="coverage">
|
| 165 | <thead>
|
| 166 | <tr>
|
| 167 | <td>Possible Opcodes</td>
|
| 168 | <td>Min. Size [bytes]</td>
|
| 169 | <td>Max. Size [bytes]</td>
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 170 | </tr>
|
| 171 | </thead>
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 172 | <tfoot>
|
| 173 | <tr>
|
| 174 | <td>Total:</td>
|
| 175 | <td>4</td>
|
| 176 | <td>7</td>
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 177 | </tr>
|
| 178 | </tfoot>
|
Marc R. Hoffmann | 9772829 | 2010-04-27 20:29:23 +0000 | [diff] [blame] | 179 | <tbody>
|
| 180 | <tr>
|
| 181 | <td><code>ALOAD_x</code>, <code>ALOAD</code> <sup>1</sup></td>
|
| 182 | <td>1</td>
|
| 183 | <td>2</td>
|
| 184 | </tr>
|
| 185 | <tr>
|
| 186 | <td><code>ICONST_x</code>, <code>BIPUSH</code>, <code>SIPUSH</code>, <code>LDC</code>, <code>LDC_W</code> <sup>2</sup></td>
|
| 187 | <td>1</td>
|
| 188 | <td>3</td>
|
| 189 | </tr>
|
| 190 | <tr>
|
| 191 | <td><code>ICONST_1</code></td>
|
| 192 | <td>1</td>
|
| 193 | <td>1</td>
|
| 194 | </tr>
|
| 195 | <tr>
|
| 196 | <td><code>BASTORE</code></td>
|
| 197 | <td>1</td>
|
| 198 | <td>1</td>
|
| 199 | </tr>
|
| 200 | </tbody>
|
Marc R. Hoffmann | 102e837 | 2010-04-27 15:46:23 +0000 | [diff] [blame] | 201 | </table>
|
| 202 |
|
| 203 | <p>
|
Marc R. Hoffmann | 9772829 | 2010-04-27 20:29:23 +0000 | [diff] [blame] | 204 | <sup>1</sup> The probe array is the first variable after the arguments.
|
| 205 | If the method arguments do not consume more that 3 slots the 1-byte opcode can
|
| 206 | be used.<br/>
|
| 207 | <sup>2</sup> 1-byte opcodes for ids 0 to 5, 2-byte opcode for ids up to 127,
|
| 208 | 3-byte opcode for ids up to 32767. Ids values of 32768 or more require an
|
| 209 | additional constant pool entry. For normal class files it is very unlikely to
|
| 210 | require more than 32,000 probes.
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 211 | </p>
|
| 212 |
|
| 213 | <ul>
|
Marc R. Hoffmann | 096bd1d | 2010-04-27 04:21:39 +0000 | [diff] [blame] | 214 | <li>Limitation: Only proves that the probe itself has been executed,
|
| 215 | assumptions about the surrounding application code is interpolation</li>
|
| 216 | <li>Probe in every edge of the control flow graph</li>
|
| 217 | <li>Every exit path known (branch coverage)</li>
|
| 218 | <li>Block entry known (exceptions within blocks)</li>
|
| 219 | </ul>
|
| 220 |
|
| 221 | <h2>Different Types of Edges</h2>
|
| 222 |
|
| 223 | <ul>
|
| 224 | <li>Probe insertion strategies</li>
|
| 225 | </ul>
|
| 226 |
|
| 227 | <h2>Runtime Overhead</h2>
|
| 228 |
|
| 229 | <ul>
|
| 230 | <li>Comparison to current basic block probes</li>
|
| 231 | </ul>
|
| 232 |
|
| 233 | </div>
|
| 234 | <div class="footer">
|
| 235 | <div class="versioninfo"><a href="@jacoco.home.url@">JaCoCo</a> @qualified.bundle.version@</div>
|
| 236 | <a href="license.html">Copyright</a> © @copyright.years@ Mountainminds GmbH & Co. KG and Contributors
|
| 237 | </div>
|
| 238 |
|
| 239 | </body>
|
| 240 | </html> |