[SLP]Add better minbitwidth analysis for udiv/urem instructions.

Adds improved bitwidth analysis for udiv/urem instructions. The
analysis is based on similar version in InstCombiner.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: #85928

Author: alexey-bataev

llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

@@ -14190,6 +14190,28 @@ bool BoUpSLP::collectValuesToDemote(
       return false;
     break;
   }
+  case Instruction::UDiv:
+  case Instruction::URem: {
+    if (ITE->UserTreeIndices.size() > 1 && !IsPotentiallyTruncated(I, BitWidth))
+      return false;
+    // UDiv and URem can be truncated if all the truncated bits are zero.
+    if (!AttemptCheckBitwidth(
+            [&](unsigned BitWidth, unsigned OrigBitWidth) {
+              assert(BitWidth <= OrigBitWidth && "Unexpected bitwidths!");
+              APInt Mask = APInt::getBitsSetFrom(OrigBitWidth, BitWidth);
+              return MaskedValueIsZero(I->getOperand(0), Mask,
+                                       SimplifyQuery(*DL)) &&
+                     MaskedValueIsZero(I->getOperand(1), Mask,
+                                       SimplifyQuery(*DL));
+            },
+            NeedToExit))
+      return false;
+    if (NeedToExit)
+      return true;
+    if (!ProcessOperands({I->getOperand(0), I->getOperand(1)}, NeedToExit))
+      return false;
+    break;
+  }
 
   // We can demote selects if we can demote their true and false values.
   case Instruction::Select: {

llvm/test/Transforms/SLPVectorizer/X86/reorder-possible-strided-node.ll

@@ -116,9 +116,7 @@ define void @test_div() {
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i32>, ptr [[ARRAYIDX22]], align 4
 ; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <4 x i32> [[TMP1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:    [[TMP3:%.*]] = mul <4 x i32> [[TMP2]], [[TMP0]]
-; CHECK-NEXT:    [[TMP4:%.*]] = zext <4 x i32> [[TMP3]] to <4 x i64>
-; CHECK-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[TMP4]], <i64 1, i64 2, i64 1, i64 2>
-; CHECK-NEXT:    [[TMP6:%.*]] = trunc <4 x i64> [[TMP5]] to <4 x i32>
+; CHECK-NEXT:    [[TMP6:%.*]] = udiv <4 x i32> [[TMP3]], <i32 1, i32 2, i32 1, i32 2>
 ; CHECK-NEXT:    store <4 x i32> [[TMP6]], ptr getelementptr inbounds ([4 x i32], ptr null, i64 8, i64 0), align 16
 ; CHECK-NEXT:    ret void
 ;
@@ -170,9 +168,7 @@ define void @test_rem() {
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i32>, ptr [[ARRAYIDX22]], align 4
 ; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <4 x i32> [[TMP1]], <4 x i32> poison, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:    [[TMP3:%.*]] = mul <4 x i32> [[TMP2]], [[TMP0]]
-; CHECK-NEXT:    [[TMP4:%.*]] = zext <4 x i32> [[TMP3]] to <4 x i64>
-; CHECK-NEXT:    [[TMP5:%.*]] = urem <4 x i64> [[TMP4]], <i64 1, i64 2, i64 1, i64 1>
-; CHECK-NEXT:    [[TMP6:%.*]] = trunc <4 x i64> [[TMP5]] to <4 x i32>
+; CHECK-NEXT:    [[TMP6:%.*]] = urem <4 x i32> [[TMP3]], <i32 1, i32 2, i32 1, i32 1>
 ; CHECK-NEXT:    store <4 x i32> [[TMP6]], ptr getelementptr inbounds ([4 x i32], ptr null, i64 8, i64 0), align 16
 ; CHECK-NEXT:    ret void
 ;