[LV][NFCI]Use integer for cost/trip count calculations instead of double, fix possible UB.

Using fp type in the compiler is not the best idea, here it used with
the comparison for equal to 0 and may cause undefined behavior in some
cases.

Reviewers: fhahn

Reviewed By: fhahn

Pull Request: #87241

Author: alexey-bataev

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

@@ -9722,7 +9722,7 @@ static bool areRuntimeChecksProfitable(GeneratedRTChecks &Checks,
   }
 
   // The scalar cost should only be 0 when vectorizing with a user specified VF/IC. In those cases, runtime checks should always be generated.
-  double ScalarC = *VF.ScalarCost.getValue();
+  uint64_t ScalarC = *VF.ScalarCost.getValue();
   if (ScalarC == 0)
     return true;
 
@@ -9749,7 +9749,7 @@ static bool areRuntimeChecksProfitable(GeneratedRTChecks &Checks,
   //   RtC + VecC * (TC / VF) + EpiC <  ScalarC * TC
   //
   // Now we can compute the minimum required trip count TC as
-  //   (RtC + EpiC) / (ScalarC - (VecC / VF)) < TC
+  //   VF * (RtC + EpiC) / (ScalarC * VF - VecC) < TC
   //
   // For now we assume the epilogue cost EpiC = 0 for simplicity. Note that
   // the computations are performed on doubles, not integers and the result
@@ -9761,9 +9761,9 @@ static bool areRuntimeChecksProfitable(GeneratedRTChecks &Checks,
       AssumedMinimumVscale = *VScale;
     IntVF *= AssumedMinimumVscale;
   }
-  double VecCOverVF = double(*VF.Cost.getValue()) / IntVF;
-  double RtC = *CheckCost.getValue();
-  double MinTC1 = RtC / (ScalarC - VecCOverVF);
+  uint64_t RtC = *CheckCost.getValue();
+  uint64_t Div = ScalarC * IntVF - *VF.Cost.getValue();
+  uint64_t MinTC1 = Div == 0 ? 0 : divideCeil(RtC * IntVF, Div);
 
   // Second, compute a minimum iteration count so that the cost of the
   // runtime checks is only a fraction of the total scalar loop cost. This
@@ -9772,12 +9772,12 @@ static bool areRuntimeChecksProfitable(GeneratedRTChecks &Checks,
   // * TC. To bound the runtime check to be a fraction 1/X of the scalar
   // cost, compute
   //   RtC < ScalarC * TC * (1 / X)  ==>  RtC * X / ScalarC < TC
-  double MinTC2 = RtC * 10 / ScalarC;
+  uint64_t MinTC2 = divideCeil(RtC * 10, ScalarC);
 
   // Now pick the larger minimum. If it is not a multiple of VF and a scalar
   // epilogue is allowed, choose the next closest multiple of VF. This should
   // partly compensate for ignoring the epilogue cost.
-  uint64_t MinTC = std::ceil(std::max(MinTC1, MinTC2));
+  uint64_t MinTC = std::max(MinTC1, MinTC2);
   if (SEL == CM_ScalarEpilogueAllowed)
     MinTC = alignTo(MinTC, IntVF);
   VF.MinProfitableTripCount = ElementCount::getFixed(MinTC);

llvm/test/Transforms/LoopVectorize/vectorize-force-tail-with-evl.ll

@@ -9,6 +9,11 @@
 ; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue -force-vector-width=4 \
 ; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on -S < %s | FileCheck --check-prefix=NO-VP %s
 
+; RUN: opt -passes=loop-vectorize \
+; RUN: -force-tail-folding-style=none \
+; RUN: -prefer-predicate-over-epilogue=predicate-else-scalar-epilogue \
+; RUN: -force-target-supports-scalable-vectors -scalable-vectorization=on -S < %s | FileCheck --check-prefix=NO-VP-DEF %s
+
 ; The target does not support predicated vectorization.
 define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
 ; IF-EVL-LABEL: @foo(
@@ -80,6 +85,54 @@ define void @foo(ptr noalias %a, ptr noalias %b, ptr noalias %c, i64 %N) {
 ; NO-VP:       for.cond.cleanup:
 ; NO-VP-NEXT:    ret void
 ;
+; NO-VP-DEF-LABEL: @foo(
+; NO-VP-DEF-NEXT:  entry:
+; NO-VP-DEF-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-DEF-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], [[TMP0]]
+; NO-VP-DEF-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; NO-VP-DEF:       vector.ph:
+; NO-VP-DEF-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-DEF-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N]], [[TMP1]]
+; NO-VP-DEF-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[N_MOD_VF]]
+; NO-VP-DEF-NEXT:    [[TMP2:%.*]] = call i64 @llvm.vscale.i64()
+; NO-VP-DEF-NEXT:    br label [[VECTOR_BODY:%.*]]
+; NO-VP-DEF:       vector.body:
+; NO-VP-DEF-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; NO-VP-DEF-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 0
+; NO-VP-DEF-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[B:%.*]], i64 [[TMP3]]
+; NO-VP-DEF-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[TMP4]], i32 0
+; NO-VP-DEF-NEXT:    [[WIDE_LOAD:%.*]] = load <vscale x 1 x i32>, ptr [[TMP5]], align 4
+; NO-VP-DEF-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[C:%.*]], i64 [[TMP3]]
+; NO-VP-DEF-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[TMP6]], i32 0
+; NO-VP-DEF-NEXT:    [[WIDE_LOAD1:%.*]] = load <vscale x 1 x i32>, ptr [[TMP7]], align 4
+; NO-VP-DEF-NEXT:    [[TMP8:%.*]] = add nsw <vscale x 1 x i32> [[WIDE_LOAD1]], [[WIDE_LOAD]]
+; NO-VP-DEF-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i32, ptr [[A:%.*]], i64 [[TMP3]]
+; NO-VP-DEF-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, ptr [[TMP9]], i32 0
+; NO-VP-DEF-NEXT:    store <vscale x 1 x i32> [[TMP8]], ptr [[TMP10]], align 4
+; NO-VP-DEF-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], [[TMP2]]
+; NO-VP-DEF-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; NO-VP-DEF-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; NO-VP-DEF:       middle.block:
+; NO-VP-DEF-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N]], [[N_VEC]]
+; NO-VP-DEF-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
+; NO-VP-DEF:       scalar.ph:
+; NO-VP-DEF-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; NO-VP-DEF-NEXT:    br label [[FOR_BODY:%.*]]
+; NO-VP-DEF:       for.body:
+; NO-VP-DEF-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY]] ]
+; NO-VP-DEF-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[IV]]
+; NO-VP-DEF-NEXT:    [[TMP12:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
+; NO-VP-DEF-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[IV]]
+; NO-VP-DEF-NEXT:    [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX2]], align 4
+; NO-VP-DEF-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP12]]
+; NO-VP-DEF-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[IV]]
+; NO-VP-DEF-NEXT:    store i32 [[ADD]], ptr [[ARRAYIDX4]], align 4
+; NO-VP-DEF-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; NO-VP-DEF-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[IV_NEXT]], [[N]]
+; NO-VP-DEF-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
+; NO-VP-DEF:       for.cond.cleanup:
+; NO-VP-DEF-NEXT:    ret void
+;
 entry:
   br label %for.body