[LLVM][ComplexDeinterleaving] Update splat identification to include vector ConstantInt/FP.

Author: paulwalker-arm

llvm/lib/CodeGen/ComplexDeinterleavingPass.cpp

@@ -2005,6 +2005,9 @@ ComplexDeinterleavingGraph::identifySplat(Value *R, Value *I) {
     if (isa<ConstantDataVector>(V))
       return true;
 
+    if (isa<ConstantInt>(V) || isa<ConstantFP>(V))
+      return isa<VectorType>(V->getType());
+
     VectorType *VTy;
     ArrayRef<int> Mask;
     // Splats are represented differently depending on whether the repeated

llvm/test/CodeGen/AArch64/complex-deinterleaving-splat-scalable.ll

@@ -1,54 +1,33 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s --mattr=+sve -o - | FileCheck %s -check-prefixes=CHECK,SHUFFLE_SPLAT
-; RUN: llc -use-constant-int-for-scalable-splat -use-constant-fp-for-scalable-splat < %s --mattr=+sve -o - | FileCheck %s -check-prefixes=CHECK,CONST_SPLAT
+; RUN: llc < %s --mattr=+sve -o - | FileCheck %s
+; RUN: llc -use-constant-int-for-scalable-splat -use-constant-fp-for-scalable-splat < %s --mattr=+sve -o - | FileCheck %s
 
 target triple = "aarch64"
 
 ; a[i] * b[i] * (11.0 + 3.0.i);
 ;
 define <vscale x 4 x double> @complex_mul_const(<vscale x 4 x double> %a, <vscale x 4 x double> %b) {
-; SHUFFLE_SPLAT-LABEL: complex_mul_const:
-; SHUFFLE_SPLAT:       // %bb.0: // %entry
-; SHUFFLE_SPLAT-NEXT:    movi v4.2d, #0000000000000000
-; SHUFFLE_SPLAT-NEXT:    movi v5.2d, #0000000000000000
-; SHUFFLE_SPLAT-NEXT:    ptrue p0.d
-; SHUFFLE_SPLAT-NEXT:    fmov z6.d, #3.00000000
-; SHUFFLE_SPLAT-NEXT:    fmov z7.d, #11.00000000
-; SHUFFLE_SPLAT-NEXT:    fcmla z4.d, p0/m, z0.d, z2.d, #0
-; SHUFFLE_SPLAT-NEXT:    fcmla z5.d, p0/m, z1.d, z3.d, #0
-; SHUFFLE_SPLAT-NEXT:    fcmla z4.d, p0/m, z0.d, z2.d, #90
-; SHUFFLE_SPLAT-NEXT:    movi v2.2d, #0000000000000000
-; SHUFFLE_SPLAT-NEXT:    fcmla z5.d, p0/m, z1.d, z3.d, #90
-; SHUFFLE_SPLAT-NEXT:    zip2 z1.d, z7.d, z6.d
-; SHUFFLE_SPLAT-NEXT:    movi v0.2d, #0000000000000000
-; SHUFFLE_SPLAT-NEXT:    zip1 z3.d, z7.d, z6.d
-; SHUFFLE_SPLAT-NEXT:    fcmla z2.d, p0/m, z5.d, z1.d, #0
-; SHUFFLE_SPLAT-NEXT:    fcmla z0.d, p0/m, z4.d, z3.d, #0
-; SHUFFLE_SPLAT-NEXT:    fcmla z2.d, p0/m, z5.d, z1.d, #90
-; SHUFFLE_SPLAT-NEXT:    fcmla z0.d, p0/m, z4.d, z3.d, #90
-; SHUFFLE_SPLAT-NEXT:    mov z1.d, z2.d
-; SHUFFLE_SPLAT-NEXT:    ret
-;
-; CONST_SPLAT-LABEL: complex_mul_const:
-; CONST_SPLAT:       // %bb.0: // %entry
-; CONST_SPLAT-NEXT:    uzp1 z4.d, z2.d, z3.d
-; CONST_SPLAT-NEXT:    uzp2 z5.d, z0.d, z1.d
-; CONST_SPLAT-NEXT:    uzp2 z2.d, z2.d, z3.d
-; CONST_SPLAT-NEXT:    uzp1 z0.d, z0.d, z1.d
-; CONST_SPLAT-NEXT:    ptrue p0.d
-; CONST_SPLAT-NEXT:    fmul z3.d, z4.d, z5.d
-; CONST_SPLAT-NEXT:    fmul z1.d, z2.d, z5.d
-; CONST_SPLAT-NEXT:    fmov z5.d, #11.00000000
-; CONST_SPLAT-NEXT:    fmad z2.d, p0/m, z0.d, z3.d
-; CONST_SPLAT-NEXT:    fmov z3.d, #3.00000000
-; CONST_SPLAT-NEXT:    fnmsb z0.d, p0/m, z4.d, z1.d
-; CONST_SPLAT-NEXT:    fmul z1.d, z2.d, z5.d
-; CONST_SPLAT-NEXT:    fmul z2.d, z2.d, z3.d
-; CONST_SPLAT-NEXT:    fmla z1.d, p0/m, z0.d, z3.d
-; CONST_SPLAT-NEXT:    fnmls z2.d, p0/m, z0.d, z5.d
-; CONST_SPLAT-NEXT:    zip1 z0.d, z2.d, z1.d
-; CONST_SPLAT-NEXT:    zip2 z1.d, z2.d, z1.d
-; CONST_SPLAT-NEXT:    ret
+; CHECK-LABEL: complex_mul_const:
+; CHECK:       // %bb.0: // %entry
+; CHECK-NEXT:    movi v4.2d, #0000000000000000
+; CHECK-NEXT:    movi v5.2d, #0000000000000000
+; CHECK-NEXT:    ptrue p0.d
+; CHECK-NEXT:    fmov z6.d, #3.00000000
+; CHECK-NEXT:    fmov z7.d, #11.00000000
+; CHECK-NEXT:    fcmla z4.d, p0/m, z0.d, z2.d, #0
+; CHECK-NEXT:    fcmla z5.d, p0/m, z1.d, z3.d, #0
+; CHECK-NEXT:    fcmla z4.d, p0/m, z0.d, z2.d, #90
+; CHECK-NEXT:    movi v2.2d, #0000000000000000
+; CHECK-NEXT:    fcmla z5.d, p0/m, z1.d, z3.d, #90
+; CHECK-NEXT:    zip2 z1.d, z7.d, z6.d
+; CHECK-NEXT:    movi v0.2d, #0000000000000000
+; CHECK-NEXT:    zip1 z3.d, z7.d, z6.d
+; CHECK-NEXT:    fcmla z2.d, p0/m, z5.d, z1.d, #0
+; CHECK-NEXT:    fcmla z0.d, p0/m, z4.d, z3.d, #0
+; CHECK-NEXT:    fcmla z2.d, p0/m, z5.d, z1.d, #90
+; CHECK-NEXT:    fcmla z0.d, p0/m, z4.d, z3.d, #90
+; CHECK-NEXT:    mov z1.d, z2.d
+; CHECK-NEXT:    ret
 entry:
   %strided.vec = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %a)
   %0 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 0