[SLP][REVEC] Fix the number of elements in the mask of a

ShuffleVectorInst is not a power of 2.

The following shufflevector should not be vectorized when
slp-vectorize-non-power-of-2 is enabled.

shufflevector <8 x float> %1, <8 x float> poison, <3 x i32> <i32 0, i32 1, i32 2>
shufflevector <8 x float> %1, <8 x float> poison, <3 x i32> <i32 4, i32 5, i32 6>

Author: HanKuanChen

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

@@ -344,6 +344,8 @@ static unsigned getShufflevectorNumGroups(ArrayRef<Value *> VL) {
   unsigned SVNumElements =
       cast<FixedVectorType>(SV->getOperand(0)->getType())->getNumElements();
   unsigned ShuffleMaskSize = SV->getShuffleMask().size();
+  if (SVNumElements % ShuffleMaskSize != 0)
+    return 0;
   unsigned GroupSize = SVNumElements / ShuffleMaskSize;
   if (GroupSize == 0 || (VL.size() % GroupSize) != 0)
     return 0;

llvm/test/Transforms/SLPVectorizer/RISCV/revec.ll

@@ -137,6 +137,64 @@ for.body:
 }
 
 define ptr @test4() {
+; POWEROF2-LABEL: @test4(
+; POWEROF2-NEXT:    [[TMP1:%.*]] = fadd <8 x float> zeroinitializer, zeroinitializer
+; POWEROF2-NEXT:    [[TMP2:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
+; POWEROF2-NEXT:    [[TMP3:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 5, i32 6>
+; POWEROF2-NEXT:    [[TMP4:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 4, i32 0>
+; POWEROF2-NEXT:    [[TMP5:%.*]] = call <4 x float> @llvm.vector.insert.v4f32.v2f32(<4 x float> poison, <2 x float> [[TMP2]], i64 0)
+; POWEROF2-NEXT:    [[TMP6:%.*]] = call <4 x float> @llvm.vector.insert.v4f32.v2f32(<4 x float> [[TMP5]], <2 x float> [[TMP3]], i64 2)
+; POWEROF2-NEXT:    br label [[TMP8:%.*]]
+; POWEROF2:       7:
+; POWEROF2-NEXT:    br label [[TMP8]]
+; POWEROF2:       8:
+; POWEROF2-NEXT:    [[TMP9:%.*]] = phi <2 x float> [ poison, [[TMP7:%.*]] ], [ [[TMP4]], [[TMP0:%.*]] ]
+; POWEROF2-NEXT:    [[TMP10:%.*]] = phi <4 x float> [ poison, [[TMP7]] ], [ [[TMP6]], [[TMP0]] ]
+; POWEROF2-NEXT:    br label [[TMP11:%.*]]
+; POWEROF2:       11:
+; POWEROF2-NEXT:    [[TMP12:%.*]] = call <2 x float> @llvm.vector.extract.v2f32.v4f32(<4 x float> [[TMP10]], i64 0)
+; POWEROF2-NEXT:    [[TMP13:%.*]] = fmul <2 x float> [[TMP12]], zeroinitializer
+; POWEROF2-NEXT:    [[TMP14:%.*]] = call <2 x float> @llvm.vector.extract.v2f32.v4f32(<4 x float> [[TMP10]], i64 2)
+; POWEROF2-NEXT:    [[TMP15:%.*]] = fmul <2 x float> zeroinitializer, [[TMP14]]
+; POWEROF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x float> [[TMP9]], i32 1
+; POWEROF2-NEXT:    [[TMP17:%.*]] = fmul float 0.000000e+00, [[TMP16]]
+; POWEROF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x float> [[TMP9]], i32 0
+; POWEROF2-NEXT:    [[TMP19:%.*]] = fmul float [[TMP18]], 0.000000e+00
+; POWEROF2-NEXT:    [[TMP20:%.*]] = extractelement <2 x float> [[TMP13]], i32 0
+; POWEROF2-NEXT:    [[TMP21:%.*]] = fadd reassoc nsz float [[TMP20]], [[TMP17]]
+; POWEROF2-NEXT:    [[TMP22:%.*]] = extractelement <2 x float> [[TMP15]], i32 0
+; POWEROF2-NEXT:    [[TMP23:%.*]] = fadd reassoc nsz float [[TMP22]], [[TMP19]]
+; POWEROF2-NEXT:    [[TMP24:%.*]] = extractelement <2 x float> [[TMP13]], i32 1
+; POWEROF2-NEXT:    [[TMP25:%.*]] = fadd reassoc nsz float [[TMP21]], [[TMP24]]
+; POWEROF2-NEXT:    [[TMP26:%.*]] = extractelement <2 x float> [[TMP15]], i32 1
+; POWEROF2-NEXT:    [[TMP27:%.*]] = fadd reassoc nsz float [[TMP23]], [[TMP26]]
+; POWEROF2-NEXT:    [[TMP28:%.*]] = tail call float @llvm.sqrt.f32(float [[TMP25]])
+; POWEROF2-NEXT:    [[TMP29:%.*]] = tail call float @llvm.sqrt.f32(float [[TMP27]])
+; POWEROF2-NEXT:    ret ptr null
+;
+; NONPOWEROF2-LABEL: @test4(
+; NONPOWEROF2-NEXT:    [[TMP1:%.*]] = fadd <8 x float> zeroinitializer, zeroinitializer
+; NONPOWEROF2-NEXT:    [[TMP2:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <3 x i32> <i32 0, i32 1, i32 2>
+; NONPOWEROF2-NEXT:    [[TMP3:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <3 x i32> <i32 4, i32 5, i32 6>
+; NONPOWEROF2-NEXT:    [[TMP4:%.*]] = call <6 x float> @llvm.vector.insert.v6f32.v3f32(<6 x float> poison, <3 x float> [[TMP2]], i64 0)
+; NONPOWEROF2-NEXT:    [[TMP5:%.*]] = call <6 x float> @llvm.vector.insert.v6f32.v3f32(<6 x float> [[TMP4]], <3 x float> [[TMP3]], i64 3)
+; NONPOWEROF2-NEXT:    br label [[TMP7:%.*]]
+; NONPOWEROF2:       6:
+; NONPOWEROF2-NEXT:    br label [[TMP7]]
+; NONPOWEROF2:       7:
+; NONPOWEROF2-NEXT:    [[TMP8:%.*]] = phi <6 x float> [ poison, [[TMP6:%.*]] ], [ [[TMP5]], [[TMP0:%.*]] ]
+; NONPOWEROF2-NEXT:    br label [[TMP9:%.*]]
+; NONPOWEROF2:       9:
+; NONPOWEROF2-NEXT:    [[TMP10:%.*]] = call <3 x float> @llvm.vector.extract.v3f32.v6f32(<6 x float> [[TMP8]], i64 0)
+; NONPOWEROF2-NEXT:    [[TMP11:%.*]] = fmul <3 x float> zeroinitializer, [[TMP10]]
+; NONPOWEROF2-NEXT:    [[TMP12:%.*]] = call <3 x float> @llvm.vector.extract.v3f32.v6f32(<6 x float> [[TMP8]], i64 3)
+; NONPOWEROF2-NEXT:    [[TMP13:%.*]] = fmul <3 x float> zeroinitializer, [[TMP12]]
+; NONPOWEROF2-NEXT:    [[TMP14:%.*]] = call reassoc nsz float @llvm.vector.reduce.fadd.v3f32(float 0.000000e+00, <3 x float> [[TMP11]])
+; NONPOWEROF2-NEXT:    [[TMP15:%.*]] = call reassoc nsz float @llvm.vector.reduce.fadd.v3f32(float 0.000000e+00, <3 x float> [[TMP13]])
+; NONPOWEROF2-NEXT:    [[TMP16:%.*]] = tail call float @llvm.sqrt.f32(float [[TMP14]])
+; NONPOWEROF2-NEXT:    [[TMP17:%.*]] = tail call float @llvm.sqrt.f32(float [[TMP15]])
+; NONPOWEROF2-NEXT:    ret ptr null
+;
   %1 = fadd <8 x float> zeroinitializer, zeroinitializer
   %2 = extractelement <8 x float> %1, i64 0
   %3 = extractelement <8 x float> %1, i64 1