[SROA] Fix the loop exit placement to be prior to indexing the splits

[oota-llvm.git] / lib / Transforms / Scalar / TailRecursionElimination.cpp

diff --git a/lib/Transforms/Scalar/TailRecursionElimination.cpp b/lib/Transforms/Scalar/TailRecursionElimination.cpp
index 946af800776f2e64680a180259b4428fa9a446ed..f3c3e3054b60a8d09ce21bd8f8b0439e3a1786a0 100644 (file)
--- a/lib/Transforms/Scalar/TailRecursionElimination.cpp
+++ b/lib/Transforms/Scalar/TailRecursionElimination.cpp
@@ -63,7 +63,9 @@
 #include "llvm/IR/CFG.h"
 #include "llvm/IR/CallSite.h"
 #include "llvm/IR/Constants.h"
+#include "llvm/IR/DataLayout.h"
 #include "llvm/IR/DerivedTypes.h"
+#include "llvm/IR/DiagnosticInfo.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/Instructions.h"
 #include "llvm/IR/IntrinsicInst.h"
@@ -85,6 +87,7 @@ STATISTIC(NumAccumAdded, "Number of accumulators introduced");
 namespace {
   struct TailCallElim : public FunctionPass {
     const TargetTransformInfo *TTI;
+    const DataLayout *DL;
 
     static char ID; // Pass identification, replacement for typeid
     TailCallElim() : FunctionPass(ID) {
@@ -156,6 +159,8 @@ bool TailCallElim::runOnFunction(Function &F) {
   if (skipOptnoneFunction(F))
     return false;
 
+  DL = F.getParent()->getDataLayout();
+
   bool AllCallsAreTailCalls = false;
   bool Modified = markTails(F, AllCallsAreTailCalls);
   if (AllCallsAreTailCalls)
@@ -174,7 +179,7 @@ struct AllocaDerivedValueTracker {
 
     auto AddUsesToWorklist = [&](Value *V) {
       for (auto &U : V->uses()) {
-        if (!Visited.insert(&U))
+        if (!Visited.insert(&U).second)
           continue;
         Worklist.push_back(&U);
       }
@@ -226,12 +231,10 @@ struct AllocaDerivedValueTracker {
   }
 
   void callUsesLocalStack(CallSite CS, bool IsNocapture) {
-    // Add it to the list of alloca users. If it's already there, skip further
-    // processing.
-    if (!AllocaUsers.insert(CS.getInstruction()))
-      return;
+    // Add it to the list of alloca users.
+    AllocaUsers.insert(CS.getInstruction());
 
-    // If it's nocapture then it can't capture the alloca.
+    // If it's nocapture then it can't capture this alloca.
     if (IsNocapture)
       return;
 
@@ -318,8 +321,8 @@ bool TailCallElim::markTails(Function &F, bool &AllCallsAreTailCalls) {
           break;
         }
         if (SafeToTail) {
-          F.getContext().emitOptimizationRemark(
-              "tailcallelim", F, CI->getDebugLoc(),
+          emitOptimizationRemark(
+              F.getContext(), "tailcallelim", F, CI->getDebugLoc(),
               "marked this readnone call a tail call candidate");
           CI->setTailCall();
           Modified = true;
@@ -365,9 +368,9 @@ bool TailCallElim::markTails(Function &F, bool &AllCallsAreTailCalls) {
     if (Visited[CI->getParent()] != ESCAPED) {
       // If the escape point was part way through the block, calls after the
       // escape point wouldn't have been put into DeferredTails.
-      F.getContext().emitOptimizationRemark(
-          "tailcallelim", F, CI->getDebugLoc(),
-          "marked this call a tail call candidate");
+      emitOptimizationRemark(F.getContext(), "tailcallelim", F,
+                             CI->getDebugLoc(),
+                             "marked this call a tail call candidate");
       CI->setTailCall();
       Modified = true;
     } else {
@@ -401,18 +404,28 @@ bool TailCallElim::runTRE(Function &F) {
   // alloca' is changed from being a static alloca to being a dynamic alloca.
   // Until this is resolved, disable this transformation if that would ever
   // happen.  This bug is PR962.
+  SmallVector<BasicBlock*, 8> BBToErase;
   for (Function::iterator BB = F.begin(), E = F.end(); BB != E; ++BB) {
     if (ReturnInst *Ret = dyn_cast<ReturnInst>(BB->getTerminator())) {
       bool Change = ProcessReturningBlock(Ret, OldEntry, TailCallsAreMarkedTail,
                                           ArgumentPHIs, !CanTRETailMarkedCall);
-      if (!Change && BB->getFirstNonPHIOrDbg() == Ret)
+      if (!Change && BB->getFirstNonPHIOrDbg() == Ret) {
         Change = FoldReturnAndProcessPred(BB, Ret, OldEntry,
                                           TailCallsAreMarkedTail, ArgumentPHIs,
                                           !CanTRETailMarkedCall);
+        // FoldReturnAndProcessPred may have emptied some BB. Remember to
+        // erase them.
+        if (Change && BB->empty())
+          BBToErase.push_back(BB);
+
+      }
       MadeChange |= Change;
     }
   }
 
+  for (auto BB: BBToErase)
+    BB->eraseFromParent();
+
   // If we eliminated any tail recursions, it's possible that we inserted some
   // silly PHI nodes which just merge an initial value (the incoming operand)
   // with themselves.  Check to see if we did and clean up our mess if so.  This
@@ -451,7 +464,7 @@ bool TailCallElim::CanMoveAboveCall(Instruction *I, CallInst *CI) {
       // being loaded from.
       if (CI->mayWriteToMemory() ||
           !isSafeToLoadUnconditionally(L->getPointerOperand(), L,
-                                       L->getAlignment()))
+                                       L->getAlignment(), DL))
         return false;
     }
   }
@@ -678,9 +691,8 @@ bool TailCallElim::EliminateRecursiveTailCall(CallInst *CI, ReturnInst *Ret,
   BasicBlock *BB = Ret->getParent();
   Function *F = BB->getParent();
 
-  F->getContext().emitOptimizationRemark(
-      "tailcallelim", *F, CI->getDebugLoc(),
-      "transforming tail recursion to loop");
+  emitOptimizationRemark(F->getContext(), "tailcallelim", *F, CI->getDebugLoc(),
+                         "transforming tail recursion to loop");
 
   // OK! We can transform this tail call.  If this is the first one found,
   // create the new entry block, allowing us to branch back to the old entry.
@@ -821,8 +833,20 @@ bool TailCallElim::FoldReturnAndProcessPred(BasicBlock *BB,
     if (CallInst *CI = FindTRECandidate(BI, CannotTailCallElimCallsMarkedTail)){
       DEBUG(dbgs() << "FOLDING: " << *BB
             << "INTO UNCOND BRANCH PRED: " << *Pred);
-      EliminateRecursiveTailCall(CI, FoldReturnIntoUncondBranch(Ret, BB, Pred),
-                                 OldEntry, TailCallsAreMarkedTail, ArgumentPHIs,
+      ReturnInst *RI = FoldReturnIntoUncondBranch(Ret, BB, Pred);
+
+      // Cleanup: if all predecessors of BB have been eliminated by
+      // FoldReturnIntoUncondBranch, we would like to delete it, but we
+      // can not just nuke it as it is being used as an iterator by our caller.
+      // Just empty it, and the caller will erase it when it is safe to do so.
+      // It is important to empty it, because the ret instruction in there is
+      // still using a value which EliminateRecursiveTailCall will attempt
+      // to remove.
+      if (!BB->hasAddressTaken() && pred_begin(BB) == pred_end(BB))
+        BB->getInstList().clear();
+
+      EliminateRecursiveTailCall(CI, RI, OldEntry, TailCallsAreMarkedTail,
+                                 ArgumentPHIs,
                                  CannotTailCallElimCallsMarkedTail);
       ++NumRetDuped;
       Change = true;