<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p><br>
</p>
<blockquote type="cite"
cite="mid:CAF7ks-PPBzYGOmxb191=GZAaE36oCZks-X+V=BZWceRAVC-4Hg@mail.gmail.com">
<meta http-equiv="content-type" content="text/html; charset=utf-8">
<div dir="ltr">
<div dir="ltr">Patch to remove the code here <a
href="https://reviews.llvm.org/D53743"
moz-do-not-send="true">https://reviews.llvm.org/D53743</a><br
clear="all">
<div>
<div dir="ltr" class="gmail_signature">~Craig</div>
</div>
<br>
</div>
</div>
</blockquote>
Not sure exactly why this is not needed anymore, but I have
confirmed that benchmarks (SPEC) on SystemZ are completely
unaffected by this patch. /Jonas<br>
<br>
<br>
<blockquote type="cite"
cite="mid:CAF7ks-PPBzYGOmxb191=GZAaE36oCZks-X+V=BZWceRAVC-4Hg@mail.gmail.com"><br>
<div class="gmail_quote">
<div dir="ltr">On Thu, Oct 25, 2018 at 5:56 PM Craig Topper <<a
href="mailto:craig.topper@gmail.com" moz-do-not-send="true">craig.topper@gmail.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">
<div dir="ltr">
<div dir="ltr">
<div dir="ltr">It seems to occur
because getSETCCWidenedResultTy checks if the mask
type should be widened, but then convertMask uses that
VT as a result VT without doing anything to widen the
inputs. Removing that code
from getSETCCWidenedResultTy doesn't regress any lit
tests. <br clear="all">
<div>
<div dir="ltr"
class="m_-6271848459583471592gmail_signature"><br>
</div>
<div dir="ltr"
class="m_-6271848459583471592gmail_signature">~Craig</div>
</div>
<br>
</div>
</div>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr">On Thu, Oct 25, 2018 at 5:49 PM Friedman, Eli
<<a href="mailto:efriedma@codeaurora.org"
target="_blank" moz-do-not-send="true">efriedma@codeaurora.org</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">On
10/25/2018 5:20 PM, Craig Topper wrote:<br>
> Should SETCC result types and input types always have
the same number <br>
> of vector elements or both be scalar?<br>
<br>
No. At least, the documentation for ISD::SETCC doesn't
indicate it's <br>
allowed, and it's not obvious what it would mean.<br>
<br>
> I ask because it
appears DAGTypeLegalizer::WidenVSELECTAndMask can <br>
> create a setcc where this isn't true.<br>
<br>
This comes from <a href="https://reviews.llvm.org/D29489"
rel="noreferrer" target="_blank" moz-do-not-send="true">https://reviews.llvm.org/D29489</a>
. Looking over my <br>
comments from that review, the patch isn't intentionally
creating <br>
mismatched SETCCs, though.<br>
<br>
-Eli<br>
<br>
-- <br>
Employee of Qualcomm Innovation Center, Inc.<br>
Qualcomm Innovation Center, Inc. is a member of Code
Aurora Forum, a Linux Foundation Collaborative Project<br>
<br>
</blockquote>
</div>
</blockquote>
</div>
</blockquote>
<br>
</body>
</html>