<html>
<head>
<base href="https://bugs.llvm.org/">
</head>
<body><span class="vcard"><a class="email" href="mailto:davide@freebsd.org" title="Davide Italiano <davide@freebsd.org>"> <span class="fn">Davide Italiano</span></a>
</span> changed
<a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - -reassociate runs forever"
href="https://bugs.llvm.org/show_bug.cgi?id=37390">bug 37390</a>
<br>
<table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>What</th>
<th>Removed</th>
<th>Added</th>
</tr>
<tr>
<td style="text-align:right;">Status</td>
<td>NEW
</td>
<td>RESOLVED
</td>
</tr>
<tr>
<td style="text-align:right;">Resolution</td>
<td>---
</td>
<td>FIXED
</td>
</tr></table>
<p>
<div>
<b><a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - -reassociate runs forever"
href="https://bugs.llvm.org/show_bug.cgi?id=37390#c6">Comment # 6</a>
on <a class="bz_bug_link
bz_status_RESOLVED bz_closed"
title="RESOLVED FIXED - -reassociate runs forever"
href="https://bugs.llvm.org/show_bug.cgi?id=37390">bug 37390</a>
from <span class="vcard"><a class="email" href="mailto:davide@freebsd.org" title="Davide Italiano <davide@freebsd.org>"> <span class="fn">Davide Italiano</span></a>
</span></b>
<pre>dcci@Davides-MacBook-Pro ~/w/l/llvm> git svn dcommit --interactive
Committing to <a href="https://llvm.org/svn/llvm-project/llvm/trunk">https://llvm.org/svn/llvm-project/llvm/trunk</a> ...
commit 8c88b7846a8120179850d6ba485ae731aba20982
Author: Davide Italiano <<a href="mailto:ditaliano@apple.com">ditaliano@apple.com</a>>
Date: Fri May 11 08:42:01 2018 -0700
[Reassociate] Prevent infinite loops when processing PHIs.
Phi nodes can reside in live blocks but one of their incoming
arguments can come from a dead block. Dead blocks and reassociate
don't play nice together. In fact, reassociate performs an RPO
as a first step to avoid processing dead blocks.
The reason why Reassociate might not fixpoint when examining
dead blocks is that the following:
%xor0 = xor i16 %xor1, undef
%xor1 = xor i16 %xor0, undef
is perfectly valid LLVM IR (if it appears in a dead block),
so the worklist algorithm keeps pushing the two instructions for
reexamination. Note that this is not Reassociate fault, at least
not entirely. It's llvm that has a weird definition of dominance.
Fixes PR37390.
create mode 100644 test/Transforms/Reassociate/infloop-deadphi.ll
Commit this patch to SVN? ([y]es (default)|[n]o|[q]uit|[a]ll): y
A test/Transforms/Reassociate/infloop-deadphi.ll
M lib/Transforms/Scalar/Reassociate.cpp
Committed r332100
M lib/Target/AMDGPU/AMDGPUISelLowering.cpp
A test/CodeGen/AMDGPU/partial-shift-shrink.ll</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>