[VPlan] Replace VPBBs with VPIRBBs during skeleton creation (NFC).

Move replacement of VPBBs for vector preheader, middle block and scalar
preheader from VPlan::execute to skeleton creation, which actually
creates the IR basic blocks.

For now, the vector preheader can only be replaced after
prepareToExecute as it may create new instructions in the vector
preheader.

Author: fhahn

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

@@ -2603,6 +2603,21 @@ BasicBlock *InnerLoopVectorizer::emitMemRuntimeChecks(BasicBlock *Bypass) {
   return MemCheckBlock;
 }
 
+/// Replace \p VPBB with a VPIRBasicBlock wrapping \p IRBB. All recipes from \p
+/// VPBB are moved to the end of the newly created VPIRBasicBlock. VPBB must
+/// have a single predecessor, which is rewired to the new VPIRBasicBlock. All
+/// successors of VPBB, if any, are rewired to the new VPIRBasicBlock.
+static void replaceVPBBWithIRVPBB(VPBasicBlock *VPBB, BasicBlock *IRBB) {
+  VPIRBasicBlock *IRVPBB = VPBB->getPlan()->createVPIRBasicBlock(IRBB);
+  for (auto &R : make_early_inc_range(*VPBB)) {
+    assert(!R.isPhi() && "Tried to move phi recipe to end of block");
+    R.moveBefore(*IRVPBB, IRVPBB->end());
+  }
+
+  VPBlockUtils::reassociateBlocks(VPBB, IRVPBB);
+  // VPBB is now dead and will be cleaned up when the plan gets destroyed.
+}
+
 void InnerLoopVectorizer::createVectorLoopSkeleton(StringRef Prefix) {
   LoopVectorPreHeader = OrigLoop->getLoopPreheader();
   assert(LoopVectorPreHeader && "Invalid loop structure");
@@ -2613,9 +2628,11 @@ void InnerLoopVectorizer::createVectorLoopSkeleton(StringRef Prefix) {
   LoopMiddleBlock =
       SplitBlock(LoopVectorPreHeader, LoopVectorPreHeader->getTerminator(), DT,
                  LI, nullptr, Twine(Prefix) + "middle.block");
+  replaceVPBBWithIRVPBB(Plan.getMiddleBlock(), LoopMiddleBlock);
   LoopScalarPreHeader =
       SplitBlock(LoopMiddleBlock, LoopMiddleBlock->getTerminator(), DT, LI,
                  nullptr, Twine(Prefix) + "scalar.ph");
+  replaceVPBBWithIRVPBB(Plan.getScalarPreheader(), LoopScalarPreHeader);
 }
 
 /// Return the expanded step for \p ID using \p ExpandedSCEVs to look up SCEV
@@ -7757,6 +7774,7 @@ DenseMap<const SCEV *, Value *> LoopVectorizationPlanner::executePlan(
   BestVPlan.prepareToExecute(
       ILV.getTripCount(),
       ILV.getOrCreateVectorTripCount(ILV.LoopVectorPreHeader), State);
+  replaceVPBBWithIRVPBB(BestVPlan.getVectorPreheader(), State.CFG.PrevBB);
 
   BestVPlan.execute(&State);
 

llvm/lib/Transforms/Vectorize/VPlan.cpp

@@ -949,47 +949,20 @@ void VPlan::prepareToExecute(Value *TripCountV, Value *VectorTripCountV,
   }
 }
 
-/// Replace \p VPBB with a VPIRBasicBlock wrapping \p IRBB. All recipes from \p
-/// VPBB are moved to the end of the newly created VPIRBasicBlock. VPBB must
-/// have a single predecessor, which is rewired to the new VPIRBasicBlock. All
-/// successors of VPBB, if any, are rewired to the new VPIRBasicBlock.
-static void replaceVPBBWithIRVPBB(VPBasicBlock *VPBB, BasicBlock *IRBB) {
-  VPIRBasicBlock *IRVPBB = VPBB->getPlan()->createVPIRBasicBlock(IRBB);
-  for (auto &R : make_early_inc_range(*VPBB)) {
-    assert(!R.isPhi() && "Tried to move phi recipe to end of block");
-    R.moveBefore(*IRVPBB, IRVPBB->end());
-  }
-
-  VPBlockUtils::reassociateBlocks(VPBB, IRVPBB);
-  // VPBB is now dead and will be cleaned up when the plan gets destroyed.
-}
-
 /// Generate the code inside the preheader and body of the vectorized loop.
 /// Assumes a single pre-header basic-block was created for this. Introduce
 /// additional basic-blocks as needed, and fill them all.
 void VPlan::execute(VPTransformState *State) {
   // Initialize CFG state.
   State->CFG.PrevVPBB = nullptr;
   State->CFG.ExitBB = State->CFG.PrevBB->getSingleSuccessor();
-  BasicBlock *VectorPreHeader = State->CFG.PrevBB;
-  State->Builder.SetInsertPoint(VectorPreHeader->getTerminator());
 
   // Disconnect VectorPreHeader from ExitBB in both the CFG and DT.
+  BasicBlock *VectorPreHeader = State->CFG.PrevBB;
   cast<BranchInst>(VectorPreHeader->getTerminator())->setSuccessor(0, nullptr);
   State->CFG.DTU.applyUpdates(
       {{DominatorTree::Delete, VectorPreHeader, State->CFG.ExitBB}});
 
-  // Replace regular VPBB's for the vector preheader, middle and scalar
-  // preheader blocks with VPIRBasicBlocks wrapping their IR blocks. The IR
-  // blocks are created during skeleton creation, so we can only create the
-  // VPIRBasicBlocks now during VPlan execution rather than earlier during VPlan
-  // construction.
-  BasicBlock *MiddleBB = State->CFG.ExitBB;
-  BasicBlock *ScalarPh = MiddleBB->getSingleSuccessor();
-  replaceVPBBWithIRVPBB(getVectorPreheader(), VectorPreHeader);
-  replaceVPBBWithIRVPBB(getMiddleBlock(), MiddleBB);
-  replaceVPBBWithIRVPBB(getScalarPreheader(), ScalarPh);
-
   LLVM_DEBUG(dbgs() << "Executing best plan with VF=" << State->VF
                     << ", UF=" << getUF() << '\n');
   setName("Final VPlan");
@@ -998,6 +971,8 @@ void VPlan::execute(VPTransformState *State) {
   // Disconnect the middle block from its single successor (the scalar loop
   // header) in both the CFG and DT. The branch will be recreated during VPlan
   // execution.
+  BasicBlock *MiddleBB = State->CFG.ExitBB;
+  BasicBlock *ScalarPh = MiddleBB->getSingleSuccessor();
   auto *BrInst = new UnreachableInst(MiddleBB->getContext());
   BrInst->insertBefore(MiddleBB->getTerminator());
   MiddleBB->getTerminator()->eraseFromParent();