[LV] Pick vector loop body as insert point for SCEV expansion.

[fhahn]

Summary

Currently the DomTree is not kept up to date for additional blocks
generated in the vector loop, for example when vectorizing with
predication. SCEVExpander relies on dominance checks when looking for
existing instructions to re-use and in some cases that can lead to the
expander picking instructions that do not actually dominate their insert
point (e.g. as in PR46525).

Unfortunately keeping the DT up-to-date is a bit tricky, because the CFG
is only patched up after generating code for a block. For now, we can
just use the vector loop header, as this ensures the inserted
instructions dominate all uses in the vector loop. There should be no
noticeable impact on the generated code, as other passes should sink
those instructions, if profitable.

Fixes PR46525.

[dmgreen]

Is the DT reliable enough to use for checking the block is in the loop? I see we might have to exclude the preheader and midblock. But if it's not uptodate, and only knows about split block, it might think the midblock it dominated by the vector body.

Maybe if LI->getLoopFor(LoopVectorBody == LI->getLoopFor(InsertBB) works, that may be better? It looks like LI is kept uptodate as blocks get split.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

2881

*choose

[fhahn]

In D83288#2135681, @dmgreen wrote:

Is the DT reliable enough to use for checking the block is in the loop? I see we might have to exclude the preheader and midblock. But if it's not uptodate, and only knows about split block, it might think the midblock it dominated by the vector body.

Maybe if LI->getLoopFor(LoopVectorBody == LI->getLoopFor(InsertBB) works, that may be better? It looks like LI is kept uptodate as blocks get split.

We are mainly want to avoid using LoopVectorBody as insert point for blocks not dominated by t. The new blocks will be unreachable initially and dominated by everything, so it should work fine.

That being said, LI is updated directly as you mentioned, so it is probably better to just use LI. Updated the patch, thanks!

[dmgreen]

Thanks. LGTM.

[Ayal]

Good catch!

The specific culprit for the pr is the call to SE.DT.dominates(EntInst, InsertPt) by SCEVExpander::FindValueInExprValueMap().

Would probably be better to keep DT up to date as we go along, due to SE.DT's dependence on it, instead of fixing it after code-gen via updateDominatorTree(); but there was some reason for doing it this way(?).

[fhahn]

In D83288#2146282, @Ayal wrote:

Good catch!

The specific culprit for the pr is the call to SE.DT.dominates(EntInst, InsertPt) by SCEVExpander::FindValueInExprValueMap().

Would probably be better to keep DT up to date as we go along, due to SE.DT's dependence on it, instead of fixing it after code-gen via updateDominatorTree(); but there was some reason for doing it this way(?).

Yes, ideally we would keep the DT up-to-date while vectorizing. It is a slightly bigger change though, mostly because we only patch up the CFG after vectorizing some blocks. I'll look into that as follow-up.

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

if (auto *CX = dyn_cast<ConstantInt>(X))
       if (CX->isOne())
         return Y;
     if (auto *CY = dyn_cast<ConstantInt>(Y))
       if (CY->isOne())
         return X;
     return B.CreateMul(X, Y);
   };
 
+  // Get a suitable insert point for SCEV expansion. For blocks in the vector
+  // loop, chose the end of the vector loop header (=LoopVectorBody), because
+  // the DomTree is not kept up-to-date for additional blocks generated in the
+  // vector loop. By using the header as insertion point, we guarantee that the
+  // expanded instructions dominate all their uses.
+  auto GetInsertPoint = [this, &B]() {
+    BasicBlock *InsertBB = B.GetInsertPoint()->getParent();
+    if (InsertBB != LoopVectorBody && DT->dominates(LoopVectorBody, InsertBB))
+      return LoopVectorBody->getTerminator();
+    return &*B.GetInsertPoint();
+  };
   switch (ID.getKind()) {
   case InductionDescriptor::IK_IntInduction: {
     assert(Index->getType() == StartValue->getType() &&
            "Index type does not match StartValue type");
     if (ID.getConstIntStepValue() && ID.getConstIntStepValue()->isMinusOne())
       return B.CreateSub(StartValue, Index);
     auto *Offset = CreateMul(
-        Index, Exp.expandCodeFor(Step, Index->getType(), &*B.GetInsertPoint()));
+        Index, Exp.expandCodeFor(Step, Index->getType(), GetInsertPoint()));
     return CreateAdd(StartValue, Offset);
   }
   case InductionDescriptor::IK_PtrInduction: {
     assert(isa<SCEVConstant>(Step) &&
            "Expected constant step for pointer induction");
     return B.CreateGEP(
         StartValue->getType()->getPointerElementType(), StartValue,
-        CreateMul(Index, Exp.expandCodeFor(Step, Index->getType(),
-                                           &*B.GetInsertPoint())));
+        CreateMul(Index,
+                  Exp.expandCodeFor(Step, Index->getType(), GetInsertPoint())));
   }
   case InductionDescriptor::IK_FpInduction: {
     assert(Step->getType()->isFloatingPointTy() && "Expected FP Step value");
     auto InductionBinOp = ID.getInductionBinOp();
     assert(InductionBinOp &&
            (InductionBinOp->getOpcode() == Instruction::FAdd ||
             InductionBinOp->getOpcode() == Instruction::FSub) &&
            "Original bin op should be defined for FP induction");

llvm/test/Transforms/LoopVectorize/pr46525-expander-insertpoint.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -loop-vectorize -force-vector-width=2 -S -prefer-predicate-over-epilog %s | FileCheck %s
+
+
+; Test case for PR46525. There are two candidates to pick for
+; `udiv i64 %y, %add` when expanding SCEV expressions. Make sure we pick %div,
+; which dominates the vector loop.
+
+define void @test(i16 %x, i64 %y, i32* %ptr) {
+; CHECK-LABEL: @test(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[CONV19:%.*]] = sext i16 [[X:%.*]] to i64
+; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[CONV19]], 492802768830814067
+; CHECK-NEXT:    br label [[LOOP_PREHEADER:%.*]]
+; CHECK:       loop.preheader:
+; CHECK-NEXT:    [[DIV:%.*]] = udiv i64 [[Y:%.*]], [[ADD]]
+; CHECK-NEXT:    [[INC:%.*]] = add i64 [[DIV]], 1
+; CHECK-NEXT:    [[TMP0:%.*]] = add nuw nsw i64 [[DIV]], 4
+; CHECK-NEXT:    [[TMP1:%.*]] = udiv i64 [[TMP0]], [[INC]]
+; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1
+; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; CHECK:       vector.ph:
+; CHECK-NEXT:    [[N_RND_UP:%.*]] = add i64 [[TMP2]], 1
+; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], 2
+; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
+; CHECK-NEXT:    [[IND_END:%.*]] = mul i64 [[N_VEC]], [[INC]]
+; CHECK-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = sub i64 [[TMP2]], 1
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> undef, i64 [[TRIP_COUNT_MINUS_1]], i32 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
+; CHECK:       vector.body:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE6:%.*]] ]
+; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], [[INC]]
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x i64> undef, i64 [[OFFSET_IDX]], i32 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT1]], <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <2 x i64> undef, i64 [[INC]], i32 0
+; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <2 x i64> [[DOTSPLATINSERT]], <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP3:%.*]] = mul <2 x i64> <i64 0, i64 1>, [[DOTSPLAT]]
+; CHECK-NEXT:    [[INDUCTION:%.*]] = add <2 x i64> [[BROADCAST_SPLAT2]], [[TMP3]]
+; CHECK-NEXT:    [[TMP4:%.*]] = mul i64 0, [[INC]]
+; CHECK-NEXT:    [[TMP5:%.*]] = add i64 [[OFFSET_IDX]], [[TMP4]]
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT3:%.*]] = insertelement <2 x i64> undef, i64 [[INDEX]], i32 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT4:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT3]], <2 x i64> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[VEC_IV:%.*]] = add <2 x i64> [[BROADCAST_SPLAT4]], <i64 0, i64 1>
+; CHECK-NEXT:    [[TMP6:%.*]] = icmp ule <2 x i64> [[VEC_IV]], [[BROADCAST_SPLAT]]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i1> [[TMP6]], i32 0
+; CHECK-NEXT:    br i1 [[TMP7]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
+; CHECK:       pred.store.if:
+; CHECK-NEXT:    store i32 0, i32* [[PTR:%.*]], align 4
+; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE]]
+; CHECK:       pred.store.continue:
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i1> [[TMP6]], i32 1
+; CHECK-NEXT:    br i1 [[TMP8]], label [[PRED_STORE_IF5:%.*]], label [[PRED_STORE_CONTINUE6]]
+; CHECK:       pred.store.if5:
+; CHECK-NEXT:    store i32 0, i32* [[PTR]], align 4
+; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE6]]
+; CHECK:       pred.store.continue6:
+; CHECK-NEXT:    [[OFFSET_IDX7:%.*]] = mul i64 [[INDEX]], [[INC]]
+; CHECK-NEXT:    [[TMP9:%.*]] = trunc i64 [[OFFSET_IDX7]] to i8
+; CHECK-NEXT:    [[TMP10:%.*]] = trunc i64 [[INC]] to i8
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT8:%.*]] = insertelement <2 x i8> undef, i8 [[TMP9]], i32 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT9:%.*]] = shufflevector <2 x i8> [[BROADCAST_SPLATINSERT8]], <2 x i8> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[DOTSPLATINSERT10:%.*]] = insertelement <2 x i8> undef, i8 [[TMP10]], i32 0
+; CHECK-NEXT:    [[DOTSPLAT11:%.*]] = shufflevector <2 x i8> [[DOTSPLATINSERT10]], <2 x i8> undef, <2 x i32> zeroinitializer
+; CHECK-NEXT:    [[TMP11:%.*]] = mul <2 x i8> <i8 0, i8 1>, [[DOTSPLAT11]]
+; CHECK-NEXT:    [[INDUCTION12:%.*]] = add <2 x i8> [[BROADCAST_SPLAT9]], [[TMP11]]
+; CHECK-NEXT:    [[TMP12:%.*]] = mul i8 0, [[TMP10]]
+; CHECK-NEXT:    [[TMP13:%.*]] = add i8 [[TMP9]], [[TMP12]]
+; CHECK-NEXT:    [[TMP14:%.*]] = add i8 [[TMP13]], 1
+; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 2
+; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop !0
+;
+entry:
+  %conv19 = sext i16 %x to i64
+  %add = add i64 %conv19, 492802768830814067
+  br label %loop.preheader
+
+loop.preheader:
+  %div = udiv i64 %y, %add
+  %inc = add i64 %div, 1
+  br label %loop
+
+loop:
+  %iv = phi i64 [ %iv.next, %loop ], [ 0, %loop.preheader ]
+  store i32 0, i32* %ptr, align 4
+  %v2 = trunc i64 %iv to i8
+  %v3 = add i8 %v2, 1
+  %cmp15 = icmp slt i8 %v3, 5
+  %iv.next = add i64 %iv, %inc
+  br i1 %cmp15, label %loop, label %loop.exit
+
+loop.exit:
+  %div.1 = udiv i64 %y, %add
+  %v1 = add i64 %div.1, 1
+  br label %loop.2
+
+loop.2:
+  %iv.1 = phi i64 [ %iv.next.1, %loop.2 ], [ 0, %loop.exit ]
+  %iv.next.1 = add i64 %iv.1, %v1
+  call void @use(i64 %iv.next.1)
+  %ec = icmp ult i64 %iv.next.1, 200
+  br i1 %ec, label %loop.2, label %loop.2.exit
+
+loop.2.exit:
+  %c = call i1 @cond()
+  br i1 %c, label %loop.preheader, label %exit
+
+exit:
+  ret void
+}
+
+declare void @use(i64)
+declare i1 @cond()