SimplifyCFG has stopped duplicating returns into predecessors to canonicalize IR

to have single return block (at least getting there) for optimizations. This
is general goodness but it would prevent some tailcall optimizations.
One specific case is code like this:
int f1(void);
int f2(void);
int f3(void);
int f4(void);
int f5(void);
int f6(void);
int foo(int x) {
  switch(x) {
  case 1: return f1();
  case 2: return f2();
  case 3: return f3();
  case 4: return f4();
  case 5: return f5();
  case 6: return f6();
  }
}

=>
LBB0_2:                                 ## %sw.bb
  callq   _f1
  popq    %rbp
  ret
LBB0_3:                                 ## %sw.bb1
  callq   _f2
  popq    %rbp
  ret
LBB0_4:                                 ## %sw.bb3
  callq   _f3
  popq    %rbp
  ret

This patch teaches codegenprep to duplicate returns when the return value
is a phi and where the phi operands are produced by tail calls followed by
an unconditional branch:

sw.bb7:                                           ; preds = %entry
  %call8 = tail call i32 @f5() nounwind
  br label %return
sw.bb9:                                           ; preds = %entry
  %call10 = tail call i32 @f6() nounwind
  br label %return
return:
  %retval.0 = phi i32 [ %call10, %sw.bb9 ], [ %call8, %sw.bb7 ], ... [ 0, %entry ]
  ret i32 %retval.0

This allows codegen to generate better code like this:

LBB0_2:                                 ## %sw.bb
        jmp     _f1                     ## TAILCALL
LBB0_3:                                 ## %sw.bb1
        jmp     _f2                     ## TAILCALL
LBB0_4:                                 ## %sw.bb3
        jmp     _f3                     ## TAILCALL

rdar://9147433

llvm-svn: 127953

5 files changed, 126 insertions, 4 deletions

diff --git a/llvm/lib/Target/ARM/ARMISelLowering.cpp b/llvm/lib/Target/ARM/ARMISelLowering.cpp
index 35a9bf78c2e..891ea62191f 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.cpp
+++ b/llvm/lib/Target/ARM/ARMISelLowering.cpp
@@ -1805,6 +1805,16 @@ bool ARMTargetLowering::isUsedByReturnOnly(SDNode *N) const {
   return HasRet;
 }
 
+bool ARMTargetLowering::mayBeEmittedAsTailCall(CallInst *CI) const {
+  if (!EnableARMTailCalls)
+    return false;
+
+  if (!CI->isTailCall())
+    return false;
+
+  return !Subtarget->isThumb1Only();
+}
+
 // ConstantPool, JumpTable, GlobalAddress, and ExternalSymbol are lowered as
 // their target counterpart wrapped in the ARMISD::Wrapper node. Suppose N is
 // one of the above mentioned nodes. It has to be wrapped because otherwise
diff --git a/llvm/lib/Target/ARM/ARMISelLowering.h b/llvm/lib/Target/ARM/ARMISelLowering.h
index 402e1c6c5da..e09c1dacfa8 100644
--- a/llvm/lib/Target/ARM/ARMISelLowering.h
+++ b/llvm/lib/Target/ARM/ARMISelLowering.h
@@ -457,6 +457,8 @@ namespace llvm {
 
     virtual bool isUsedByReturnOnly(SDNode *N) const;
 
+    virtual bool mayBeEmittedAsTailCall(CallInst *CI) const;
+
     SDValue getARMCmp(SDValue LHS, SDValue RHS, ISD::CondCode CC,
                       SDValue &ARMcc, SelectionDAG &DAG, DebugLoc dl) const;
     SDValue getVFPCmp(SDValue LHS, SDValue RHS,
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 576c879c80c..58acf4faac2 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -45,6 +45,7 @@
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/VectorExtras.h"
+#include "llvm/Support/CallSite.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/Dwarf.h"
 #include "llvm/Support/ErrorHandling.h"
@@ -1595,6 +1596,18 @@ static bool IsTailCallConvention(CallingConv::ID CC) {
   return (CC == CallingConv::Fast || CC == CallingConv::GHC);
 }
 
+bool X86TargetLowering::mayBeEmittedAsTailCall(CallInst *CI) const {
+  if (!CI->isTailCall())
+    return false;
+
+  CallSite CS(CI);
+  CallingConv::ID CalleeCC = CS.getCallingConv();
+  if (!IsTailCallConvention(CalleeCC) && CalleeCC != CallingConv::C)
+    return false;
+
+  return true;
+}
+
 /// FuncIsMadeTailCallSafe - Return true if the function is being made into
 /// a tailcall target by changing its ABI.
 static bool FuncIsMadeTailCallSafe(CallingConv::ID CC) {
diff --git a/llvm/lib/Target/X86/X86ISelLowering.h b/llvm/lib/Target/X86/X86ISelLowering.h
index 7c1b13a1ad5..63010573989 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.h
+++ b/llvm/lib/Target/X86/X86ISelLowering.h
@@ -843,6 +843,8 @@ namespace llvm {
 
     virtual bool isUsedByReturnOnly(SDNode *N) const;
 
+    virtual bool mayBeEmittedAsTailCall(CallInst *CI) const;
+
     virtual EVT
     getTypeForExtArgOrReturn(LLVMContext &Context, EVT VT,
                              ISD::NodeType ExtendKind) const;
diff --git a/llvm/lib/Transforms/Scalar/CodeGenPrepare.cpp b/llvm/lib/Transforms/Scalar/CodeGenPrepare.cpp
index f0babcccee0..eb80f5c64ac 100644
--- a/llvm/lib/Transforms/Scalar/CodeGenPrepare.cpp
+++ b/llvm/lib/Transforms/Scalar/CodeGenPrepare.cpp
@@ -47,16 +47,17 @@ using namespace llvm;
 using namespace llvm::PatternMatch;
 
 STATISTIC(NumBlocksElim, "Number of blocks eliminated");
-STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
-STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
+STATISTIC(NumPHIsElim,   "Number of trivial PHIs eliminated");
+STATISTIC(NumGEPsElim,   "Number of GEPs converted to casts");
 STATISTIC(NumCmpUses, "Number of uses of Cmp expressions replaced with uses of "
                       "sunken Cmps");
 STATISTIC(NumCastUses, "Number of uses of Cast expressions replaced with uses "
                        "of sunken Casts");
 STATISTIC(NumMemoryInsts, "Number of memory instructions whose address "
                           "computations were sunk");
-STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
-STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
+STATISTIC(NumExtsMoved,  "Number of [s|z]ext instructions combined with loads");
+STATISTIC(NumExtUses,    "Number of uses of [s|z]ext instructions optimized");
+STATISTIC(NumRetsDup,    "Number of return instructions duplicated");
 
 static cl::opt<bool> DisableBranchOpts(
   "disable-cgp-branch-opts", cl::Hidden, cl::init(false),
@@ -104,6 +105,7 @@ namespace {
     bool OptimizeCallInst(CallInst *CI);
     bool MoveExtToFormExtLoad(Instruction *I);
     bool OptimizeExtUses(Instruction *I);
+    bool DupRetToEnableTailCallOpts(ReturnInst *RI);
   };
 }
 
@@ -547,6 +549,96 @@ bool CodeGenPrepare::OptimizeCallInst(CallInst *CI) {
   return Simplifier.fold(CI, TD);
 }
 
+/// DupRetToEnableTailCallOpts - Look for opportunities to duplicate return
+/// instructions to the predecessor to enable tail call optimizations. The
+/// case it is currently looking for is:
+/// bb0:
+///   %tmp0 = tail call i32 @f0()
+///   br label %return
+/// bb1:
+///   %tmp1 = tail call i32 @f1()
+///   br label %return
+/// bb2:
+///   %tmp2 = tail call i32 @f2()
+///   br label %return
+/// return:
+///   %retval = phi i32 [ %tmp0, %bb0 ], [ %tmp1, %bb1 ], [ %tmp2, %bb2 ]
+///   ret i32 %retval
+///
+/// =>
+///
+/// bb0:
+///   %tmp0 = tail call i32 @f0()
+///   ret i32 %tmp0
+/// bb1:
+///   %tmp1 = tail call i32 @f1()
+///   ret i32 %tmp1
+/// bb2:
+///   %tmp2 = tail call i32 @f2()
+///   ret i32 %tmp2
+///
+bool CodeGenPrepare::DupRetToEnableTailCallOpts(ReturnInst *RI) {
+  Value *V = RI->getReturnValue();
+  if (!V)
+    return false;
+
+  if (PHINode *PN = dyn_cast<PHINode>(V)) {
+    BasicBlock *BB = RI->getParent();
+    if (PN->getParent() != BB)
+      return false;
+
+    // It's not safe to eliminate the sign / zero extension of the return value.
+    // See llvm::isInTailCallPosition().
+    const Function *F = BB->getParent();
+    unsigned CallerRetAttr = F->getAttributes().getRetAttributes();
+    if ((CallerRetAttr & Attribute::ZExt) || (CallerRetAttr & Attribute::SExt))
+      return false;
+
+    // Make sure there are no instructions between PHI and return.
+    BasicBlock::iterator BI = PN;
+    do { ++BI; } while (isa<DbgInfoIntrinsic>(BI));
+    if (&*BI != RI)
+      return false;
+
+    /// Only dup the ReturnInst if the CallInst is likely to be emitted as a
+    /// tail call.
+    SmallVector<CallInst*, 4> TailCalls;
+    for (unsigned I = 0, E = PN->getNumIncomingValues(); I != E; ++I) {
+      CallInst *CI = dyn_cast<CallInst>(PN->getIncomingValue(I));
+      if (CI && TLI->mayBeEmittedAsTailCall(CI))
+        TailCalls.push_back(CI);
+    }
+
+    bool Changed = false;
+    for (unsigned i = 0, e = TailCalls.size(); i != e; ++i) {
+      CallInst *CI = TailCalls[i];
+      CallSite CS(CI);
+
+      // Conservatively require the attributes of the call to match those of
+      // the return. Ignore noalias because it doesn't affect the call sequence.
+      unsigned CalleeRetAttr = CS.getAttributes().getRetAttributes();
+      if ((CalleeRetAttr ^ CallerRetAttr) & ~Attribute::NoAlias)
+        continue;
+
+      // Make sure the call instruction is followed by an unconditional branch
+      // to the return block.
+      BasicBlock *CallBB = CI->getParent();
+      BranchInst *BI = dyn_cast<BranchInst>(CallBB->getTerminator());
+      if (!BI || !BI->isUnconditional() || BI->getSuccessor(0) != BB)
+        continue;
+
+      // Duplicate the return into CallBB.
+      (void)FoldReturnIntoUncondBranch(RI, BB, CallBB);
+      Changed = true;
+      ++NumRetsDup;
+    }
+
+    return Changed;
+  }
+
+  return false;
+}
+
 //===----------------------------------------------------------------------===//
 // Memory Optimization
 //===----------------------------------------------------------------------===//
@@ -970,6 +1062,9 @@ bool CodeGenPrepare::OptimizeInst(Instruction *I) {
   if (CallInst *CI = dyn_cast<CallInst>(I))
     return OptimizeCallInst(CI);
 
+  if (ReturnInst *RI = dyn_cast<ReturnInst>(I))
+    return DupRetToEnableTailCallOpts(RI);
+
   return false;
 }