Generate PPC early conditional returns

PowerPC has a conditional branch to the link register (return) instruction: BCLR.
This should be used any time when we'd otherwise have a conditional branch to a
return. This adds a small pass, PPCEarlyReturn, which runs just prior to the
branch selection pass (and, importantly, after block placement) to generate
these conditional returns when possible. It will also eliminate unconditional
branches to returns (these happen rarely; most of the time these have already
been tail duplicated by the time PPCEarlyReturn is invoked). This is a nice
optimization for small functions that do not maintain a stack frame.

llvm-svn: 179026

1 files changed, 148 insertions, 0 deletions

diff --git a/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp b/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
index e106f2201d8..fcf18e31f78 100644
--- a/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCInstrInfo.cpp
@@ -18,8 +18,10 @@
 #include "PPCInstrBuilder.h"
 #include "PPCMachineFunctionInfo.h"
 #include "PPCTargetMachine.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
@@ -896,3 +898,149 @@ unsigned PPCInstrInfo::GetInstSizeInBytes(const MachineInstr *MI) const {
     return 4; // PowerPC instructions are all 4 bytes
   }
 }
+
+#undef DEBUG_TYPE
+#define DEBUG_TYPE "ppc-early-ret"
+STATISTIC(NumBCLR, "Number of early conditional returns");
+STATISTIC(NumBLR,  "Number of early returns");
+
+namespace llvm {
+  void initializePPCEarlyReturnPass(PassRegistry&);
+}
+
+namespace {
+  // PPCEarlyReturn pass - For simple functions without epilogue code, move
+  // returns up, and create conditional returns, to avoid unnecessary
+  // branch-to-blr sequences.
+  struct PPCEarlyReturn : public MachineFunctionPass {
+    static char ID;
+    PPCEarlyReturn() : MachineFunctionPass(ID) {
+      initializePPCEarlyReturnPass(*PassRegistry::getPassRegistry());
+    }
+
+    const PPCTargetMachine *TM;
+    const PPCInstrInfo *TII;
+
+protected:
+    bool processBlock(MachineBasicBlock &LastMBB) {
+      bool Changed = false;
+
+      MachineBasicBlock::iterator I = LastMBB.begin();
+      I = LastMBB.SkipPHIsAndLabels(I);
+
+      // The block must be essentially empty except for the blr.
+      if (I == LastMBB.end() || I->getOpcode() != PPC::BLR ||
+          I != LastMBB.getLastNonDebugInstr())
+        return Changed;
+
+      SmallVector<MachineBasicBlock*, 8> PredToRemove;
+      for (MachineBasicBlock::pred_iterator PI = LastMBB.pred_begin(),
+           PIE = LastMBB.pred_end(); PI != PIE; ++PI) {
+        bool OtherReference = false, BlockChanged = false;
+        for (MachineBasicBlock::iterator J = (*PI)->begin();
+             J != (*PI)->end();) {
+          if (J->getOpcode() == PPC::B) {
+            if (J->getOperand(0).getMBB() == &LastMBB) {
+              // This is an unconditional branch to the return. Replace the
+	      // branch with a blr.
+              BuildMI(**PI, J, J->getDebugLoc(), TII->get(PPC::BLR));
+              MachineBasicBlock::iterator K = J++;
+              K->eraseFromParent();
+              BlockChanged = true;
+              ++NumBLR;
+              continue;
+            }
+          } else if (J->getOpcode() == PPC::BCC) {
+            if (J->getOperand(2).getMBB() == &LastMBB) {
+              // This is a conditional branch to the return. Replace the branch
+              // with a bclr.
+              BuildMI(**PI, J, J->getDebugLoc(), TII->get(PPC::BCLR))
+                .addImm(J->getOperand(0).getImm())
+                .addReg(J->getOperand(1).getReg());
+              MachineBasicBlock::iterator K = J++;
+              K->eraseFromParent();
+              BlockChanged = true;
+              ++NumBCLR;
+              continue;
+            }
+          } else if (J->isBranch()) {
+            if (J->isIndirectBranch()) {
+              if (LastMBB.hasAddressTaken())
+                OtherReference = true;
+            } else
+              for (unsigned i = 0; i < J->getNumOperands(); ++i)
+                if (J->getOperand(i).isMBB() &&
+                    J->getOperand(i).getMBB() == &LastMBB)
+                  OtherReference = true;
+          }
+
+          ++J;
+        }
+
+        if ((*PI)->canFallThrough() && (*PI)->isLayoutSuccessor(&LastMBB))
+          OtherReference = true;
+
+	// Predecessors are stored in a vector and can't be removed here.
+        if (!OtherReference && BlockChanged) {
+          PredToRemove.push_back(*PI);
+        }
+
+        if (BlockChanged)
+          Changed = true;
+      }
+
+      for (unsigned i = 0, ie = PredToRemove.size(); i != ie; ++i)
+        PredToRemove[i]->removeSuccessor(&LastMBB);
+
+      if (Changed && !LastMBB.hasAddressTaken()) {
+        // We now might be able to merge this blr-only block into its
+        // by-layout predecessor.
+        if (LastMBB.pred_size() == 1 &&
+            (*LastMBB.pred_begin())->isLayoutSuccessor(&LastMBB)) {
+          // Move the blr into the preceding block.
+          MachineBasicBlock &PrevMBB = **LastMBB.pred_begin();
+          PrevMBB.splice(PrevMBB.end(), &LastMBB, I);
+          PrevMBB.removeSuccessor(&LastMBB);
+        }
+
+        if (LastMBB.pred_empty())
+          LastMBB.eraseFromParent();
+      }
+
+      return Changed;
+    }
+
+public:
+    virtual bool runOnMachineFunction(MachineFunction &MF) {
+      TM = static_cast<const PPCTargetMachine *>(&MF.getTarget());
+      TII = TM->getInstrInfo();
+
+      bool Changed = false;
+
+      // If the function does not have at least two block, then there is
+      // nothing to do.
+      if (MF.size() < 2)
+        return Changed;
+
+      for (MachineFunction::iterator I = MF.begin(); I != MF.end();) {
+        MachineBasicBlock &B = *I++; 
+        if (processBlock(B))
+          Changed = true;
+      }
+
+      return Changed;
+    }
+
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      MachineFunctionPass::getAnalysisUsage(AU);
+    }
+  };
+}
+
+INITIALIZE_PASS(PPCEarlyReturn, DEBUG_TYPE,
+                "PowerPC Early-Return Creation", false, false)
+
+char PPCEarlyReturn::ID = 0;
+FunctionPass*
+llvm::createPPCEarlyReturnPass() { return new PPCEarlyReturn(); }
+