<div dir="ltr"><div><div>Hello everyone,<br><br></div>I am working on loop modeling improvements in the Static Analyzer and sent recently some patches on loop widening. (<a href="https://reviews.llvm.org/D36690" target="_blank">D36690</a>)</div><div>There is already a widening solution (widen-loops config) which invalidates every MemRegion in the process. My version only widen loops which meets specific requirements (e.g. does not contain any pointers). It is hidden behind a new flag 'widen-loops-conservative'. (It is conservative in the sense that it only widen specific loops.)</div><div><br></div><div>In general case the difference of the 2 version is:<div class="gmail-m_7869373378371586527gmail-m_7873275108326179076gmail-remarkup-table-wrap"><table class="gmail-m_7869373378371586527gmail-m_7873275108326179076gmail-remarkup-table">
<tbody><tr><td>widen-loops</td><td>Invalidate everything</td></tr>
<tr><td>widen-loops-conservative</td><td>Only invalidate modified variables</td></tr>
</tbody></table></div><br></div><div>But a huge difference when there are pointers (most precisely if it encounters a statement which can result a modified variable but it is not handled yet):</div><div><div class="gmail-m_7869373378371586527gmail-m_7873275108326179076gmail-remarkup-table-wrap"><table class="gmail-m_7869373378371586527gmail-m_7873275108326179076gmail-remarkup-table">
<tbody><tr><td>widen-loops</td><td>Invalidate everything</td></tr>
<tr><td>widen-loops-conservative</td><td>Invalidate nothing (don't widen)</td></tr>
</tbody></table></div><br>I thought the new flag is necessary since this two options are very different. However, Sean (who created the current version of widening in <a href="https://reviews.llvm.org/D12358" target="_blank">D12358</a>) suggested to replace the 'loop-widening' option with the new functionality. For considering this I collected the relevant statistics (note: the 'widen-loops' config encountered some crashes on C++ projects so the statistics are only collected on the files which has passed):<br></div><div><b>Widen </b>states for the current implementation<b>, Widen v2 </b>states for the new (<a href="https://reviews.llvm.org/D36690" target="_blank">D36690</a>) implementation.<br></div><div><br></div><div>Coverage measurements (% of reachable basic blocks statistics)<br>
<table width="606" cellspacing="0" border="0" height="82">
<colgroup width="107"></colgroup>
<colgroup width="92"></colgroup>
<colgroup width="111"></colgroup>
<colgroup width="80"></colgroup>
<colgroup width="97"></colgroup>
<colgroup width="80" span="2"></colgroup>
<tbody><tr>
<td align="center" valign="bottom" height="18"><font color="#000000"><br></font></td>
<td align="center" valign="bottom"><font color="#000000">curl </font></td>
<td align="center" valign="bottom"><font color="#000000">libpng</font></td>
<td align="center" valign="bottom"><font color="#000000">vim</font></td>
<td align="center" valign="bottom"><font color="#000000">bitcoin</font></td>
<td align="center" valign="bottom"><font color="#000000">ffmpeg</font></td>
<td align="center" valign="bottom"><font color="#000000">xerces</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="20"><font color="#000000">Normal</font></td>
<td align="center" valign="bottom"><font color="#000000">58.05</font></td>
<td align="center" valign="bottom"><font color="#000000">55.07</font></td>
<td align="center" valign="bottom"><font color="#000000">51.12</font></td>
<td align="center" valign="bottom"><font color="#000000">69.37</font></td>
<td align="center" valign="bottom"><font color="#000000">49.78</font></td>
<td align="center" valign="bottom"><font color="#000000">74.86</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="18"><font color="#000000">Widen</font></td>
<td align="center" valign="bottom"><font color="#000000">75.8</font></td>
<td align="center" valign="bottom"><font color="#000000">56.69</font></td>
<td align="center" valign="bottom"><font color="#000000">51.54</font></td>
<td align="center" valign="bottom"><font color="#000000">72.06</font></td>
<td align="center" valign="bottom"><font color="#000000">65.63</font></td>
<td align="center" valign="bottom"><font color="#000000">76.06</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="18"><font color="#000000">Widen v2</font></td>
<td align="center" valign="bottom"><font color="#000000">70.7</font></td>
<td align="center" valign="bottom"><font color="#000000">55.92</font></td>
<td align="center" valign="bottom"><font color="#000000">51.77</font></td>
<td align="center" valign="bottom"><font color="#000000">69.67</font></td>
<td align="center" valign="bottom"><font color="#000000">59.53</font></td>
<td align="center" valign="bottom"><font color="#000000">75.04</font></td>
</tr>
</tbody></table>
<br></div><div>The number of founded bugs:<br></div><div>
<table width="594" cellspacing="0" border="0" height="80">
<colgroup width="107"></colgroup>
<colgroup width="92"></colgroup>
<colgroup width="111"></colgroup>
<colgroup width="80"></colgroup>
<colgroup width="97"></colgroup>
<colgroup width="80" span="2"></colgroup>
<tbody><tr>
<td align="center" valign="bottom" height="18"><font color="#000000"><br></font></td>
<td align="center" valign="bottom"><font color="#000000">curl </font></td>
<td align="center" valign="bottom"><font color="#000000">libpng</font></td>
<td align="center" valign="bottom"><font color="#000000">vim</font></td>
<td align="center" valign="bottom"><font color="#000000">bitcoin</font></td>
<td align="center" valign="bottom"><font color="#000000">ffmpeg</font></td>
<td align="center" valign="bottom"><font color="#000000">xerces</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="18"><font color="#000000">Normal</font></td>
<td align="center" valign="bottom"><font color="#000000">35</font></td>
<td align="center" valign="bottom"><font color="#000000">32</font></td>
<td align="center" valign="bottom"><font color="#000000">81</font></td>
<td align="center" valign="bottom"><font color="#000000">9</font></td>
<td align="center" valign="bottom"><font color="#000000">375</font></td>
<td align="center" valign="bottom"><font color="#000000">52</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="18"><font color="#000000">Widen</font></td>
<td align="center" valign="bottom"><font color="#000000">35</font></td>
<td align="center" valign="bottom"><font color="#000000">36</font></td>
<td align="center" valign="bottom"><font color="#000000">94</font></td>
<td align="center" valign="bottom"><font color="#000000">12</font></td>
<td align="center" valign="bottom"><font color="#000000">456</font></td>
<td align="center" valign="bottom"><font color="#000000">57</font></td>
</tr>
<tr>
<td align="right" valign="bottom" height="18"><font color="#000000">Widen v2</font></td>
<td align="center" valign="bottom"><font color="#000000">27</font></td>
<td align="center" valign="bottom"><font color="#000000">34</font></td>
<td align="center" valign="bottom"><font color="#000000">84</font></td>
<td align="center" valign="bottom"><font color="#000000">10</font></td>
<td align="center" valign="bottom"><font color="#000000">369</font></td>
<td align="center" valign="bottom"><font color="#000000">51</font></td>
</tr>
</tbody></table>
</div><div><br>The new findings are mostly resulted by the invalidation process and not the coverage increasement. So in general they are false positives which was founded by invalidating the informations (on MemRegions) we have.<br><br></div><div>Although there is a small observable coverage loss it's offset by the number of the false positives not discovered.<br></div><div>In conclusion it would be beneficial to replace the current implementation of the 'widen-loops' option.<br><br></div><div>What do you think?<br><br></div><div>Cheers,<br></div><div>Peter<br></div></div>