<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - Regression in SVN r368898, "[RLEV] Rewrite loop exit values for multiple exit loops w/o overall loop exit count""
href="https://bugs.llvm.org/show_bug.cgi?id=43028">43028</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>Regression in SVN r368898, "[RLEV] Rewrite loop exit values for multiple exit loops w/o overall loop exit count"
</td>
</tr>
<tr>
<th>Product</th>
<td>libraries
</td>
</tr>
<tr>
<th>Version</th>
<td>trunk
</td>
</tr>
<tr>
<th>Hardware</th>
<td>PC
</td>
</tr>
<tr>
<th>OS</th>
<td>All
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>enhancement
</td>
</tr>
<tr>
<th>Priority</th>
<td>P
</td>
</tr>
<tr>
<th>Component</th>
<td>Scalar Optimizations
</td>
</tr>
<tr>
<th>Assignee</th>
<td>unassignedbugs@nondot.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>martin@martin.st
</td>
</tr>
<tr>
<th>CC</th>
<td>listmail@philipreames.com, llvm-bugs@lists.llvm.org, nikita.ppv@gmail.com
</td>
</tr></table>
<p>
<div>
<pre>Created <span class=""><a href="attachment.cgi?id=22385" name="attach_22385" title="Reproduction C code">attachment 22385</a> <a href="attachment.cgi?id=22385&action=edit" title="Reproduction C code">[details]</a></span>
Reproduction C code
Since SVN r368898, "[RLEV] Rewrite loop exit values for multiple exit loops w/o
overall loop exit count", compilation of GLEW for x86_64 has regressed further.
Prior to this commit, compilation of the attached sample file ("clang -target
x86_64-w64-mingw32 -w -c -O3 glew-preproc-x86_64-edited.c") took 2 minutes, now
it takes close to 6 minutes.
Prior to SVN r363883, compiling this file used to take only one minute (this
previous regression is tracked in PR42357, and a potential workaround was
posted in <a href="https://reviews.llvm.org/D64533">https://reviews.llvm.org/D64533</a>), and GCC compiles it in 5 seconds.
Whether it affects runtime is unknown, but originally (prior to r363883), the
output object file size was 650K, after that it was 1.2M, and now it's 1.5M.
The previous regression manifested itself for both 32 and 64 bit x86, but this
change only (notably) affected x86_64.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are on the CC list for the bug.</li>
</ul>
</body>
</html>