Add an LLVM IR version of code sinking. This uses the same simple algorithm

as MachineSink, but it isn't constrained by MachineInstr-level details.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@103257 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/LinkAllPasses.h

@@ -138,6 +138,7 @@ namespace {
       (void) llvm::createGEPSplitterPass();
       (void) llvm::createABCDPass();
       (void) llvm::createLintPass();
+      (void) llvm::createSinkingPass();
 
       (void)new llvm::IntervalPartition();
       (void)new llvm::FindUsedTypes();

include/llvm/Transforms/Scalar.h

@@ -332,6 +332,12 @@ FunctionPass *createGEPSplitterPass();
 //
 FunctionPass *createABCDPass();
 
+//===----------------------------------------------------------------------===//
+//
+// Sink - Code Sinking
+//
+FunctionPass *createSinkingPass();
+
 } // End llvm namespace
 
 #endif

lib/Transforms/Scalar/Sink.cpp

@@ -0,0 +1,267 @@
+//===-- Sink.cpp - Code Sinking -------------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This pass moves instructions into successor blocks, when possible, so that
+// they aren't executed on paths where their results aren't needed.
+//
+//===----------------------------------------------------------------------===//
+
+#define DEBUG_TYPE "sink"
+#include "llvm/Transforms/Scalar.h"
+#include "llvm/IntrinsicInst.h"
+#include "llvm/Analysis/Dominators.h"
+#include "llvm/Analysis/LoopInfo.h"
+#include "llvm/Analysis/AliasAnalysis.h"
+#include "llvm/Assembly/Writer.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/Support/CFG.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/raw_ostream.h"
+using namespace llvm;
+
+STATISTIC(NumSunk, "Number of instructions sunk");
+
+namespace {
+  class Sinking : public FunctionPass {
+    DominatorTree *DT;
+    LoopInfo *LI;
+    AliasAnalysis *AA;
+
+  public:
+    static char ID; // Pass identification
+    Sinking() : FunctionPass(&ID) {}
+    
+    virtual bool runOnFunction(Function &F);
+    
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AU.setPreservesCFG();
+      FunctionPass::getAnalysisUsage(AU);
+      AU.addRequired<AliasAnalysis>();
+      AU.addRequired<DominatorTree>();
+      AU.addRequired<LoopInfo>();
+      AU.addPreserved<DominatorTree>();
+      AU.addPreserved<LoopInfo>();
+    }
+  private:
+    bool ProcessBlock(BasicBlock &BB);
+    bool SinkInstruction(Instruction *I, SmallPtrSet<Instruction *, 8> &Stores);
+    bool AllUsesDominatedByBlock(Instruction *Inst, BasicBlock *BB) const;
+  };
+} // end anonymous namespace
+  
+char Sinking::ID = 0;
+static RegisterPass<Sinking>
+X("sink", "Code sinking");
+
+FunctionPass *llvm::createSinkingPass() { return new Sinking(); }
+
+/// AllUsesDominatedByBlock - Return true if all uses of the specified value
+/// occur in blocks dominated by the specified block.
+bool Sinking::AllUsesDominatedByBlock(Instruction *Inst, 
+                                      BasicBlock *BB) const {
+  // Ignoring debug uses is necessary so debug info doesn't affect the code.
+  // This may leave a referencing dbg_value in the original block, before
+  // the definition of the vreg.  Dwarf generator handles this although the
+  // user might not get the right info at runtime.
+  for (Value::use_iterator I = Inst->use_begin(),
+       E = Inst->use_end(); I != E; ++I) {
+    // Determine the block of the use.
+    Instruction *UseInst = cast<Instruction>(*I);
+    BasicBlock *UseBlock = UseInst->getParent();
+    if (PHINode *PN = dyn_cast<PHINode>(UseInst)) {
+      // PHI nodes use the operand in the predecessor block, not the block with
+      // the PHI.
+      unsigned Num = PHINode::getIncomingValueNumForOperand(I.getOperandNo());
+      UseBlock = PN->getIncomingBlock(Num);
+    }
+    // Check that it dominates.
+    if (!DT->dominates(BB, UseBlock))
+      return false;
+  }
+  return true;
+}
+
+bool Sinking::runOnFunction(Function &F) {
+  DT = &getAnalysis<DominatorTree>();
+  LI = &getAnalysis<LoopInfo>();
+  AA = &getAnalysis<AliasAnalysis>();
+
+  bool EverMadeChange = false;
+  
+  while (1) {
+    bool MadeChange = false;
+
+    // Process all basic blocks.
+    for (Function::iterator I = F.begin(), E = F.end(); 
+         I != E; ++I)
+      MadeChange |= ProcessBlock(*I);
+    
+    // If this iteration over the code changed anything, keep iterating.
+    if (!MadeChange) break;
+    EverMadeChange = true;
+  } 
+  return EverMadeChange;
+}
+
+bool Sinking::ProcessBlock(BasicBlock &BB) {
+  // Can't sink anything out of a block that has less than two successors.
+  if (BB.getTerminator()->getNumSuccessors() <= 1 || BB.empty()) return false;
+
+  // Don't bother sinking code out of unreachable blocks. In addition to being
+  // unprofitable, it can also lead to infinite looping, because in an unreachable
+  // loop there may be nowhere to stop.
+  if (!DT->isReachableFromEntry(&BB)) return false;
+
+  bool MadeChange = false;
+
+  // Walk the basic block bottom-up.  Remember if we saw a store.
+  BasicBlock::iterator I = BB.end();
+  --I;
+  bool ProcessedBegin = false;
+  SmallPtrSet<Instruction *, 8> Stores;
+  do {
+    Instruction *Inst = I;  // The instruction to sink.
+    
+    // Predecrement I (if it's not begin) so that it isn't invalidated by
+    // sinking.
+    ProcessedBegin = I == BB.begin();
+    if (!ProcessedBegin)
+      --I;
+
+    if (isa<DbgInfoIntrinsic>(Inst))
+      continue;
+
+    if (SinkInstruction(Inst, Stores))
+      ++NumSunk, MadeChange = true;
+    
+    // If we just processed the first instruction in the block, we're done.
+  } while (!ProcessedBegin);
+  
+  return MadeChange;
+}
+
+static bool isSafeToMove(Instruction *Inst, AliasAnalysis *AA,
+                         SmallPtrSet<Instruction *, 8> &Stores) {
+  if (LoadInst *L = dyn_cast<LoadInst>(Inst)) {
+    if (L->isVolatile()) return false;
+
+    Value *Ptr = L->getPointerOperand();
+    unsigned Size = AA->getTypeStoreSize(L->getType());
+    for (SmallPtrSet<Instruction *, 8>::iterator I = Stores.begin(),
+         E = Stores.end(); I != E; ++I)
+      if (AA->getModRefInfo(*I, Ptr, Size) & AliasAnalysis::Mod)
+        return false;
+  }
+
+  if (Inst->mayWriteToMemory()) {
+    Stores.insert(Inst);
+    return false;
+  }
+
+  return Inst->isSafeToSpeculativelyExecute();
+}
+
+/// SinkInstruction - Determine whether it is safe to sink the specified machine
+/// instruction out of its current block into a successor.
+bool Sinking::SinkInstruction(Instruction *Inst,
+                              SmallPtrSet<Instruction *, 8> &Stores) {
+  // Check if it's safe to move the instruction.
+  if (!isSafeToMove(Inst, AA, Stores))
+    return false;
+  
+  // FIXME: This should include support for sinking instructions within the
+  // block they are currently in to shorten the live ranges.  We often get
+  // instructions sunk into the top of a large block, but it would be better to
+  // also sink them down before their first use in the block.  This xform has to
+  // be careful not to *increase* register pressure though, e.g. sinking
+  // "x = y + z" down if it kills y and z would increase the live ranges of y
+  // and z and only shrink the live range of x.
+  
+  // Loop over all the operands of the specified instruction.  If there is
+  // anything we can't handle, bail out.
+  BasicBlock *ParentBlock = Inst->getParent();
+  
+  // SuccToSinkTo - This is the successor to sink this instruction to, once we
+  // decide.
+  BasicBlock *SuccToSinkTo = 0;
+  
+  // FIXME: This picks a successor to sink into based on having one
+  // successor that dominates all the uses.  However, there are cases where
+  // sinking can happen but where the sink point isn't a successor.  For
+  // example:
+  //   x = computation
+  //   if () {} else {}
+  //   use x
+  // the instruction could be sunk over the whole diamond for the 
+  // if/then/else (or loop, etc), allowing it to be sunk into other blocks
+  // after that.
+  
+  // Instructions can only be sunk if all their uses are in blocks
+  // dominated by one of the successors.
+  // Look at all the successors and decide which one
+  // we should sink to.
+  for (succ_iterator SI = succ_begin(ParentBlock),
+       E = succ_end(ParentBlock); SI != E; ++SI) {
+    if (AllUsesDominatedByBlock(Inst, *SI)) {
+      SuccToSinkTo = *SI;
+      break;
+    }
+  }
+      
+  // If we couldn't find a block to sink to, ignore this instruction.
+  if (SuccToSinkTo == 0)
+    return false;
+  
+  // It is not possible to sink an instruction into its own block.  This can
+  // happen with loops.
+  if (Inst->getParent() == SuccToSinkTo)
+    return false;
+  
+  DEBUG(dbgs() << "Sink instr " << *Inst);
+  DEBUG(dbgs() << "to block ";
+        WriteAsOperand(dbgs(), SuccToSinkTo, false));
+  
+  // If the block has multiple predecessors, this would introduce computation on
+  // a path that it doesn't already exist.  We could split the critical edge,
+  // but for now we just punt.
+  // FIXME: Split critical edges if not backedges.
+  if (SuccToSinkTo->getUniquePredecessor() != ParentBlock) {
+    // We cannot sink a load across a critical edge - there may be stores in
+    // other code paths.
+    if (!Inst->isSafeToSpeculativelyExecute()) {
+      DEBUG(dbgs() << " *** PUNTING: Wont sink load along critical edge.\n");
+      return false;
+    }
+
+    // We don't want to sink across a critical edge if we don't dominate the
+    // successor. We could be introducing calculations to new code paths.
+    if (!DT->dominates(ParentBlock, SuccToSinkTo)) {
+      DEBUG(dbgs() << " *** PUNTING: Critical edge found\n");
+      return false;
+    }
+
+    // Don't sink instructions into a loop.
+    if (LI->isLoopHeader(SuccToSinkTo)) {
+      DEBUG(dbgs() << " *** PUNTING: Loop header found\n");
+      return false;
+    }
+
+    // Otherwise we are OK with sinking along a critical edge.
+    DEBUG(dbgs() << "Sinking along critical edge.\n");
+  }
+  
+  // Determine where to insert into.  Skip phi nodes.
+  BasicBlock::iterator InsertPos = SuccToSinkTo->begin();
+  while (InsertPos != SuccToSinkTo->end() && isa<PHINode>(InsertPos))
+    ++InsertPos;
+  
+  // Move the instruction.
+  Inst->moveBefore(InsertPos);
+  return true;
+}

test/Transforms/Sink/basic.ll

@@ -0,0 +1,22 @@
+; RUN: opt < %s -sink -S | FileCheck %s
+
+@A = external global i32
+@B = external global i32
+
+; Sink should sink the load past the store (which doesn't overlap) into
+; the block that uses it.
+
+;      CHECK: @foo
+;      CHECK: true:
+; CHECK-NEXT: %l = load i32* @A
+; CHECK-NEXT: ret i32 %l
+
+define i32 @foo(i1 %z) {
+  %l = load i32* @A
+  store i32 0, i32* @B
+  br i1 %z, label %true, label %false
+true:
+  ret i32 %l
+false:
+  ret i32 0
+}

test/Transforms/Sink/dg.exp

@@ -0,0 +1,3 @@
+load_lib llvm.exp
+
+RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll,c,cpp}]]