[ValueTracking] Support scalable vector splats in computeKnownBits (#170345)

Similar to #170325, this patch
adds support for scalable vector splats in computeKnownBits.

Author: dtcxzyw

llvm/lib/Analysis/ValueTracking.cpp

@@ -2249,6 +2249,11 @@ static void computeKnownBitsFromOperator(const Operator *I,
     break;
   }
   case Instruction::ShuffleVector: {
+    if (auto *Splat = getSplatValue(I)) {
+      computeKnownBits(Splat, Known, Q, Depth + 1);
+      break;
+    }
+
     auto *Shuf = dyn_cast<ShuffleVectorInst>(I);
     // FIXME: Do we need to handle ConstantExpr involving shufflevectors?
     if (!Shuf) {

llvm/test/Transforms/InstCombine/known-bits.ll

@@ -2425,6 +2425,23 @@ exit:
   ret i8 %or2
 }
 
+define <vscale x 4 x i32> @scalable_add_to_disjoint_or(i8 %x, <vscale x 4 x i32> range(i32 0, 256) %rhs) {
+; CHECK-LABEL: @scalable_add_to_disjoint_or(
+; CHECK-NEXT:    [[EXTX:%.*]] = zext i8 [[X:%.*]] to i32
+; CHECK-NEXT:    [[SHIFT:%.*]] = shl nuw nsw i32 [[EXTX]], 8
+; CHECK-NEXT:    [[INSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[SHIFT]], i64 0
+; CHECK-NEXT:    [[SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[INSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+; CHECK-NEXT:    [[ADD:%.*]] = or disjoint <vscale x 4 x i32> [[SPLAT]], [[RHS:%.*]]
+; CHECK-NEXT:    ret <vscale x 4 x i32> [[ADD]]
+;
+  %extx = zext i8 %x to i32
+  %shift = shl nuw nsw i32 %extx, 8
+  %insert = insertelement <vscale x 4 x i32> poison, i32 %shift, i32 0
+  %splat = shufflevector <vscale x 4 x i32> %insert, <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
+  %add = add <vscale x 4 x i32> %splat, %rhs
+  ret <vscale x 4 x i32> %add
+}
+
 declare void @dummy()
 declare void @use(i1)
 declare void @sink(i8)

llvm/test/Transforms/LoopVectorize/AArch64/sve-interleaved-accesses.ll

@@ -1197,8 +1197,8 @@ define void @PR27626_5(ptr %a, i32 %x, i32 %y, i32 %z, i64 %n) #1 {
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[INDUCTION]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP13:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], splat (i64 -1)
-; CHECK-NEXT:    [[TMP14:%.*]] = add <vscale x 4 x i64> [[VEC_IND]], splat (i64 -3)
+; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 -1)
+; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 -3)
 ; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], <vscale x 4 x i64> [[VEC_IND]]
 ; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP13]]
 ; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], <vscale x 4 x i64> [[TMP14]]
@@ -1286,7 +1286,7 @@ define void @PR34743(ptr %a, ptr %b, i64 %n) #1 {
 ; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <vscale x 4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_MASKED_GATHER4:%.*]], [[VECTOR_BODY]] ]
 ; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <vscale x 4 x i64> [ [[TMP15]], [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP18:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT:    [[TMP18:%.*]] = or disjoint <vscale x 4 x i64> [[VEC_IND]], splat (i64 1)
 ; CHECK-NEXT:    [[TMP19:%.*]] = add nuw nsw <vscale x 4 x i64> [[VEC_IND]], splat (i64 2)
 ; CHECK-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i16, ptr [[A]], <vscale x 4 x i64> [[TMP18]]
 ; CHECK-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <vscale x 4 x i16> @llvm.masked.gather.nxv4i16.nxv4p0(<vscale x 4 x ptr> align 4 [[TMP20]], <vscale x 4 x i1> splat (i1 true), <vscale x 4 x i16> poison), !alias.scope [[META34:![0-9]+]]