From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Wed, 3 Jul 2019 10:04:16 +0000 (+0000)
Subject: [X86] LowerFunnelShift - use modulo constant shift amount.
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=9739a115f4b123f44e39649bee9c1bddc5984dd6;p=llvm

[X86] LowerFunnelShift - use modulo constant shift amount.

This avoids the use of getZExtValue and uses the modulo shift amount which is whats expected for funnel shifts anyhow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@365016 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 3807ad68ee9..65a72fea379 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -17831,7 +17831,7 @@ static SDValue LowerFunnelShift(SDValue Op, const X86Subtarget &Subtarget,
 
     APInt APIntShiftAmt;
     if (isConstantSplat(Amt, APIntShiftAmt)) {
-      uint64_t ShiftAmt = APIntShiftAmt.getZExtValue();
+      uint64_t ShiftAmt = APIntShiftAmt.urem(VT.getScalarSizeInBits());
       return DAG.getNode(IsFSHR ? X86ISD::VSHRD : X86ISD::VSHLD, DL, VT,
                          Op0, Op1, DAG.getConstant(ShiftAmt, DL, MVT::i8));
     }