[llvm] r358165 - [X86] SimplifyDemandedVectorElts - add VPPERM support

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Thu Apr 11 06:30:38 2019
New Revision: 358165

URL: http://llvm.org/viewvc/llvm-project?rev=358165&view=rev
Log:
[X86] SimplifyDemandedVectorElts - add VPPERM support

We need to add support for all variable shuffle mask ops, but VPPERM is the only one that already has test coverage.

Modified:
    llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
    llvm/trunk/test/CodeGen/X86/oddshuffles.ll

Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=358165&r1=358164&r2=358165&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Thu Apr 11 06:30:38 2019
@@ -33234,6 +33234,15 @@ bool X86TargetLowering::SimplifyDemanded
       return true;
     break;
   }
+  case X86ISD::VPPERM: {
+    // TODO - simplify other variable shuffle masks.
+    SDValue Mask = Op.getOperand(2);
+    APInt MaskUndef, MaskZero;
+    if (SimplifyDemandedVectorElts(Mask, DemandedElts, MaskUndef, MaskZero, TLO,
+                                   Depth + 1))
+      return true;
+    break;
+  }
   case X86ISD::HADD:
   case X86ISD::HSUB:
   case X86ISD::FHADD:

Modified: llvm/trunk/test/CodeGen/X86/oddshuffles.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/oddshuffles.ll?rev=358165&r1=358164&r2=358165&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/oddshuffles.ll (original)
+++ llvm/trunk/test/CodeGen/X86/oddshuffles.ll Thu Apr 11 06:30:38 2019
@@ -1241,7 +1241,7 @@ define void @interleave_24i16_in(<24 x i
 ; XOP-NEXT:    vmovdqu (%rsi), %xmm0
 ; XOP-NEXT:    vmovdqu (%rdx), %xmm1
 ; XOP-NEXT:    vmovdqu (%rcx), %xmm2
-; XOP-NEXT:    vpperm {{.*#+}} xmm3 = xmm0[4,5,6,7],xmm1[6,7],xmm0[6,7,8,9],xmm1[8,9],xmm0[8,9,10,11]
+; XOP-NEXT:    vpperm {{.*#+}} xmm3 = xmm0[u,u,6,7],xmm1[6,7],xmm0[u,u,8,9],xmm1[8,9],xmm0[u,u,10,11]
 ; XOP-NEXT:    vpshufd {{.*#+}} xmm4 = xmm2[1,1,2,2]
 ; XOP-NEXT:    vpblendw {{.*#+}} xmm3 = xmm4[0],xmm3[1,2],xmm4[3],xmm3[4,5],xmm4[6],xmm3[7]
 ; XOP-NEXT:    vpunpcklwd {{.*#+}} xmm4 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]