<html>
<head>
<base href="https://bugs.freedesktop.org/" />
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW - PHINode containing itself causes segfault in LLVM when compiling Blender OpenCL kernel with R600 backend"
href="https://bugs.freedesktop.org/show_bug.cgi?id=84232#c3">Comment # 3</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW - PHINode containing itself causes segfault in LLVM when compiling Blender OpenCL kernel with R600 backend"
href="https://bugs.freedesktop.org/show_bug.cgi?id=84232">bug 84232</a>
from <span class="vcard"><a class="email" href="mailto:arsenm2@gmail.com" title="Matt Arsenault <arsenm2@gmail.com>"> <span class="fn">Matt Arsenault</span></a>
</span></b>
<pre>(In reply to Vitaliy Filippov from <a href="show_bug.cgi?id=84232#c2">comment #2</a>)
<span class="quote">>
> Unhandled loop condition!
> UNREACHABLE executed at
> /var/home/vitali/setup/llvm-toolchain-3.5-3.5/lib/Target/R600/
> SIAnnotateControlFlow.cpp:259!
>
> This is probably related to bugpoint replacing the branch conditions with
> constants, which makes conditional branches effectively unconditional :) and
> SIAnnotateControlFlow::handleLoopCondition also doesn't handle this case.
>
> Looks like an easy to fix problem?</span >
The StructurizeCFG pass can actually create branches on constants, although it
shouldn't, which I've seen several times before. I've had a partial fix for
SIAnnotateControlFlow handling branches on constants for a while, but I've been
putting off fixing it until the less noticeable StructurizeCFG problem is
fixed.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are the assignee for the bug.</li>
</ul>
</body>
</html>