[LoopVectorize] Ignore @llvm.assume for cost estimates and legality

A few minor changes to prevent @llvm.assume from interfering with loop
vectorization. First, treat @llvm.assume like the lifetime intrinsics, which
are scalarized (but don't otherwise interfere with the legality checking).
Second, ignore the cost of ephemeral instructions in the loop (these will go
away anyway during CodeGen).

Alignment assumptions and other uses of @llvm.assume can often end up inside of
loops that should be vectorized (this is not uncommon for assumptions generated
by __attribute__((align_value(n))), for example).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219741 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/Transforms/Utils/VectorUtils.h

@@ -99,7 +99,7 @@ getIntrinsicIDForCall(CallInst *CI, const TargetLibraryInfo *TLI) {
   if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(CI)) {
     Intrinsic::ID ID = II->getIntrinsicID();
     if (isTriviallyVectorizable(ID) || ID == Intrinsic::lifetime_start ||
-        ID == Intrinsic::lifetime_end)
+        ID == Intrinsic::lifetime_end || ID == Intrinsic::assume)
       return ID;
     else
       return Intrinsic::not_intrinsic;

lib/Transforms/Vectorize/LoopVectorize.cpp

@@ -55,7 +55,9 @@
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/Analysis/AliasSetTracker.h"
+#include "llvm/Analysis/AssumptionTracker.h"
 #include "llvm/Analysis/BlockFrequencyInfo.h"
+#include "llvm/Analysis/CodeMetrics.h"
 #include "llvm/Analysis/LoopInfo.h"
 #include "llvm/Analysis/LoopIterator.h"
 #include "llvm/Analysis/LoopPass.h"
@@ -884,8 +886,12 @@ public:
                              LoopVectorizationLegality *Legal,
                              const TargetTransformInfo &TTI,
                              const DataLayout *DL, const TargetLibraryInfo *TLI,
-                             const Function *F, const LoopVectorizeHints *Hints)
+                             AssumptionTracker *AT, const Function *F,
-      : TheLoop(L), SE(SE), LI(LI), Legal(Legal), TTI(TTI), DL(DL), TLI(TLI), TheFunction(F), Hints(Hints) {}
+                             const LoopVectorizeHints *Hints)
+      : TheLoop(L), SE(SE), LI(LI), Legal(Legal), TTI(TTI), DL(DL), TLI(TLI),
+        AT(AT), TheFunction(F), Hints(Hints) {
+    CodeMetrics::collectEphemeralValues(L, AT, EphValues);
+  }
 
   /// Information about vectorization costs
   struct VectorizationFactor {
@@ -954,6 +960,9 @@ private:
                                    *TheFunction, DL, Message.str());
   }
 
+  /// Values used only by @llvm.assume calls.
+  SmallPtrSet<const Value *, 32> EphValues;
+
   /// The loop that we evaluate.
   Loop *TheLoop;
   /// Scev analysis.
@@ -968,6 +977,8 @@ private:
   const DataLayout *DL;
   /// Target Library Info.
   const TargetLibraryInfo *TLI;
+  /// Tracker for @llvm.assume.
+  AssumptionTracker *AT;
   const Function *TheFunction;
   // Loop Vectorize Hint.
   const LoopVectorizeHints *Hints;
@@ -1251,6 +1262,7 @@ struct LoopVectorize : public FunctionPass {
   BlockFrequencyInfo *BFI;
   TargetLibraryInfo *TLI;
   AliasAnalysis *AA;
+  AssumptionTracker *AT;
   bool DisableUnrolling;
   bool AlwaysVectorize;
 
@@ -1266,6 +1278,7 @@ struct LoopVectorize : public FunctionPass {
     BFI = &getAnalysis<BlockFrequencyInfo>();
     TLI = getAnalysisIfAvailable<TargetLibraryInfo>();
     AA = &getAnalysis<AliasAnalysis>();
+    AT = &getAnalysis<AssumptionTracker>();
 
     // Compute some weights outside of the loop over the loops. Compute this
     // using a BranchProbability to re-use its scaling math.
@@ -1384,7 +1397,8 @@ struct LoopVectorize : public FunctionPass {
     }
 
     // Use the cost model.
-    LoopVectorizationCostModel CM(L, SE, LI, &LVL, *TTI, DL, TLI, F, &Hints);
+    LoopVectorizationCostModel CM(L, SE, LI, &LVL, *TTI, DL, TLI, AT, F,
+                                  &Hints);
 
     // Check the function attributes to find out if this function should be
     // optimized for size.
@@ -1471,6 +1485,7 @@ struct LoopVectorize : public FunctionPass {
   }
 
   void getAnalysisUsage(AnalysisUsage &AU) const override {
+    AU.addRequired<AssumptionTracker>();
     AU.addRequiredID(LoopSimplifyID);
     AU.addRequiredID(LCSSAID);
     AU.addRequired<BlockFrequencyInfo>();
@@ -3361,6 +3376,7 @@ void InnerLoopVectorizer::vectorizeBlockInLoop(BasicBlock *BB, PhiVector *PV) {
       Intrinsic::ID ID = getIntrinsicIDForCall(CI, TLI);
       assert(ID && "Not an intrinsic call!");
       switch (ID) {
+      case Intrinsic::assume:
       case Intrinsic::lifetime_end:
       case Intrinsic::lifetime_start:
         scalarizeInstruction(it);
@@ -5457,6 +5473,10 @@ unsigned LoopVectorizationCostModel::getWidestType() {
     for (BasicBlock::iterator it = BB->begin(), e = BB->end(); it != e; ++it) {
       Type *T = it->getType();
 
+      // Ignore ephemeral values.
+      if (EphValues.count(it))
+        continue;
+
       // Only examine Loads, Stores and PHINodes.
       if (!isa<LoadInst>(it) && !isa<StoreInst>(it) && !isa<PHINode>(it))
         continue;
@@ -5719,6 +5739,10 @@ LoopVectorizationCostModel::calculateRegisterUsage() {
     // Ignore instructions that are never used within the loop.
     if (!Ends.count(I)) continue;
 
+    // Ignore ephemeral values.
+    if (EphValues.count(I))
+      continue;
+
     // Remove all of the instructions that end at this location.
     InstrList &List = TransposeEnds[i];
     for (unsigned int j=0, e = List.size(); j < e; ++j)
@@ -5759,6 +5783,10 @@ unsigned LoopVectorizationCostModel::expectedCost(unsigned VF) {
       if (isa<DbgInfoIntrinsic>(it))
         continue;
 
+      // Ignore ephemeral values.
+      if (EphValues.count(it))
+        continue;
+
       unsigned C = getInstructionCost(it, VF);
 
       // Check if we should override the cost.
@@ -6059,6 +6087,7 @@ static const char lv_name[] = "Loop Vectorization";
 INITIALIZE_PASS_BEGIN(LoopVectorize, LV_NAME, lv_name, false, false)
 INITIALIZE_AG_DEPENDENCY(TargetTransformInfo)
 INITIALIZE_AG_DEPENDENCY(AliasAnalysis)
+INITIALIZE_PASS_DEPENDENCY(AssumptionTracker)
 INITIALIZE_PASS_DEPENDENCY(BlockFrequencyInfo)
 INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass)
 INITIALIZE_PASS_DEPENDENCY(ScalarEvolution)

test/Transforms/LoopVectorize/X86/assume.ll

@@ -0,0 +1,100 @@
+; RUN: opt < %s  -loop-vectorize -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -S | FileCheck %s
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; Function Attrs: nounwind uwtable
+define void @test1(float* noalias nocapture %a, float* noalias nocapture readonly %b) #0 {
+entry:
+  br label %for.body
+
+; CHECK-LABEL: @test1
+; CHECK: vector.body:
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: for.body:
+; CHECK: ret void
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %arrayidx = getelementptr inbounds float* %b, i64 %indvars.iv
+  %0 = load float* %arrayidx, align 4
+  %cmp1 = fcmp ogt float %0, 1.000000e+02
+  tail call void @llvm.assume(i1 %cmp1)
+  %add = fadd float %0, 1.000000e+00
+  %arrayidx5 = getelementptr inbounds float* %a, i64 %indvars.iv
+  store float %add, float* %arrayidx5, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond = icmp eq i64 %indvars.iv, 1599
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+
+; Function Attrs: nounwind
+declare void @llvm.assume(i1) #1
+
+attributes #0 = { nounwind uwtable }
+attributes #1 = { nounwind }
+
+%struct.data = type { float*, float* }
+
+; Function Attrs: nounwind uwtable
+define void @test2(%struct.data* nocapture readonly %d) #0 {
+entry:
+  %b = getelementptr inbounds %struct.data* %d, i64 0, i32 1
+  %0 = load float** %b, align 8
+  %ptrint = ptrtoint float* %0 to i64
+  %maskedptr = and i64 %ptrint, 31
+  %maskcond = icmp eq i64 %maskedptr, 0
+  %a = getelementptr inbounds %struct.data* %d, i64 0, i32 0
+  %1 = load float** %a, align 8
+  %ptrint2 = ptrtoint float* %1 to i64
+  %maskedptr3 = and i64 %ptrint2, 31
+  %maskcond4 = icmp eq i64 %maskedptr3, 0
+  br label %for.body
+
+; CHECK-LABEL: @test2
+; CHECK: vector.body:
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: @llvm.assume
+; CHECK: for.body:
+; CHECK: ret void
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  tail call void @llvm.assume(i1 %maskcond)
+  %arrayidx = getelementptr inbounds float* %0, i64 %indvars.iv
+  %2 = load float* %arrayidx, align 4
+  %add = fadd float %2, 1.000000e+00
+  tail call void @llvm.assume(i1 %maskcond4)
+  %arrayidx5 = getelementptr inbounds float* %1, i64 %indvars.iv
+  store float %add, float* %arrayidx5, align 4
+  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
+  %exitcond = icmp eq i64 %indvars.iv, 1599
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret void
+}
+