Fill delay slot with useful instructions. Modified from Sparc's version of delay

slot filler.

Patch by Reed Kotler at Mips Technologies.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@140825 91177308-0d34-0410-b5e6-96231b3b80d8

2 files changed, 216 insertions, 12 deletions

diff --git a/lib/Target/Mips/MipsDelaySlotFiller.cpp b/lib/Target/Mips/MipsDelaySlotFiller.cpp
index 1a4ef0cd7a..a6bd5f33f8 100644
--- a/lib/Target/Mips/MipsDelaySlotFiller.cpp
+++ b/lib/Target/Mips/MipsDelaySlotFiller.cpp
@@ -7,7 +7,7 @@
 //
 //===----------------------------------------------------------------------===//
 //
-// Simple pass to fills delay slots with NOPs.
+// Simple pass to fills delay slots with useful instructions.
 //
 //===----------------------------------------------------------------------===//
 
@@ -17,13 +17,23 @@
 #include "MipsTargetMachine.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetInstrInfo.h"
+#include "llvm/Target/TargetRegisterInfo.h"
+#include "llvm/ADT/SmallSet.h"
 #include "llvm/ADT/Statistic.h"
 
 using namespace llvm;
 
 STATISTIC(FilledSlots, "Number of delay slots filled");
 
+static cl::opt<bool> EnableDelaySlotFiller(
+  "enable-mips-delay-filler",
+  cl::init(false),
+  cl::desc("Fill the Mips delay slots with noop."),
+  cl::Hidden);
+
 namespace {
   struct Filler : public MachineFunctionPass {
 
@@ -47,29 +57,56 @@ namespace {
       return Changed;
     }
 
+    bool isDelayFiller(MachineBasicBlock &MBB,
+                       MachineBasicBlock::iterator candidate);
+
+    void insertCallUses(MachineBasicBlock::iterator MI,
+                        SmallSet<unsigned, 32>& RegDefs,
+                        SmallSet<unsigned, 32>& RegUses);
+
+    void insertDefsUses(MachineBasicBlock::iterator MI,
+                        SmallSet<unsigned, 32>& RegDefs,
+                        SmallSet<unsigned, 32>& RegUses);
+
+    bool IsRegInSet(SmallSet<unsigned, 32>& RegSet,
+                    unsigned Reg);
+
+    bool delayHasHazard(MachineBasicBlock::iterator candidate,
+                        bool &sawLoad, bool &sawStore,
+                        SmallSet<unsigned, 32> &RegDefs,
+                        SmallSet<unsigned, 32> &RegUses);
+
+    MachineBasicBlock::iterator
+    findDelayInstr(MachineBasicBlock &MBB, MachineBasicBlock::iterator slot);
+
+
   };
   char Filler::ID = 0;
 } // end of anonymous namespace
 
 /// runOnMachineBasicBlock - Fill in delay slots for the given basic block.
-/// Currently, we fill delay slots with NOPs. We assume there is only one
-/// delay slot per delayed instruction.
+/// We assume there is only one delay slot per delayed instruction.
 bool Filler::
-runOnMachineBasicBlock(MachineBasicBlock &MBB)
-{
+runOnMachineBasicBlock(MachineBasicBlock &MBB) {
   bool Changed = false;
-  for (MachineBasicBlock::iterator I = MBB.begin(); I != MBB.end(); ++I) {
-    const MCInstrDesc& MCid = I->getDesc();
-    if (MCid.hasDelaySlot()) {
+  for (MachineBasicBlock::iterator I = MBB.begin(); I != MBB.end(); ++I)
+    if (I->getDesc().hasDelaySlot()) {
+      MachineBasicBlock::iterator D = MBB.end();
       MachineBasicBlock::iterator J = I;
-      ++J;
-      BuildMI(MBB, J, I->getDebugLoc(), TII->get(Mips::NOP));
+
+      if (EnableDelaySlotFiller)
+        D = findDelayInstr(MBB, I);
+
       ++FilledSlots;
       Changed = true;
-    }
-  }
 
+      if (D == MBB.end())
+        BuildMI(MBB, ++J, I->getDebugLoc(), TII->get(Mips::NOP));
+      else
+        MBB.splice(++J, &MBB, D);
+     }
   return Changed;
+
 }
 
 /// createMipsDelaySlotFillerPass - Returns a pass that fills in delay
@@ -78,3 +115,155 @@ FunctionPass *llvm::createMipsDelaySlotFillerPass(MipsTargetMachine &tm) {
   return new Filler(tm);
 }
 
+MachineBasicBlock::iterator
+Filler::findDelayInstr(MachineBasicBlock &MBB,
+                       MachineBasicBlock::iterator slot) {
+  SmallSet<unsigned, 32> RegDefs;
+  SmallSet<unsigned, 32> RegUses;
+  bool sawLoad = false;
+  bool sawStore = false;
+
+  MachineBasicBlock::iterator I = slot;
+
+  // Call's delay filler can def some of call's uses.
+  if (slot->getDesc().isCall())
+    insertCallUses(slot, RegDefs, RegUses);
+  else
+    insertDefsUses(slot, RegDefs, RegUses);
+
+  bool done = false;
+
+  while (!done) {
+    done = (I == MBB.begin());
+
+    if (!done)
+      --I;
+
+    // skip debug value
+    if (I->isDebugValue())
+      continue;
+
+    if (I->hasUnmodeledSideEffects()
+        || I->isInlineAsm()
+        || I->isLabel()
+        || I->getDesc().hasDelaySlot()
+        || isDelayFiller(MBB, I)
+        || I->getDesc().isPseudo()
+        //
+        // Should not allow:
+        // ERET, DERET or WAIT, PAUSE. Need to add these to instruction
+        // list. TBD.
+        )
+      break;
+
+    if (delayHasHazard(I, sawLoad, sawStore, RegDefs, RegUses)) {
+      insertDefsUses(I, RegDefs, RegUses);
+      continue;
+    }
+
+    return I;
+  }
+  return MBB.end();
+}
+
+bool Filler::delayHasHazard(MachineBasicBlock::iterator candidate,
+                            bool &sawLoad,
+                            bool &sawStore,
+                            SmallSet<unsigned, 32> &RegDefs,
+                            SmallSet<unsigned, 32> &RegUses) {
+  if (candidate->isImplicitDef() || candidate->isKill())
+    return true;
+
+  if (candidate->getDesc().mayLoad()) {
+    sawLoad = true;
+    if (sawStore)
+      return true;
+  }
+
+  if (candidate->getDesc().mayStore()) {
+    if (sawStore)
+      return true;
+    sawStore = true;
+    if (sawLoad)
+      return true;
+  }
+
+  for (unsigned i = 0, e = candidate->getNumOperands(); i!= e; ++i) {
+    const MachineOperand &MO = candidate->getOperand(i);
+    if (!MO.isReg())
+      continue; // skip
+
+    unsigned Reg = MO.getReg();
+
+    if (MO.isDef()) {
+      // check whether Reg is defined or used before delay slot.
+      if (IsRegInSet(RegDefs, Reg) || IsRegInSet(RegUses, Reg))
+        return true;
+    }
+    if (MO.isUse()) {
+      // check whether Reg is defined before delay slot.
+      if (IsRegInSet(RegDefs, Reg))
+        return true;
+    }
+  }
+  return false;
+}
+
+void Filler::insertCallUses(MachineBasicBlock::iterator MI,
+                            SmallSet<unsigned, 32>& RegDefs,
+                            SmallSet<unsigned, 32>& RegUses) {
+  switch(MI->getOpcode()) {
+  default: llvm_unreachable("Unknown opcode.");
+  case Mips::JAL:
+    RegDefs.insert(31);
+    break;
+  case Mips::JALR:
+    assert(MI->getNumOperands() >= 1);
+    const MachineOperand &Reg = MI->getOperand(0);
+    assert(Reg.isReg() && "JALR first operand is not a register.");
+    RegUses.insert(Reg.getReg());
+    RegDefs.insert(31);
+    break;
+  }
+}
+
+// Insert Defs and Uses of MI into the sets RegDefs and RegUses.
+void Filler::insertDefsUses(MachineBasicBlock::iterator MI,
+                            SmallSet<unsigned, 32>& RegDefs,
+                            SmallSet<unsigned, 32>& RegUses) {
+  for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
+    const MachineOperand &MO = MI->getOperand(i);
+    if (!MO.isReg())
+      continue;
+
+    unsigned Reg = MO.getReg();
+    if (Reg == 0)
+      continue;
+    if (MO.isDef())
+      RegDefs.insert(Reg);
+    if (MO.isUse())
+      RegUses.insert(Reg);
+  }
+}
+
+//returns true if the Reg or its alias is in the RegSet.
+bool Filler::IsRegInSet(SmallSet<unsigned, 32>& RegSet, unsigned Reg) {
+  if (RegSet.count(Reg))
+    return true;
+  // check Aliased Registers
+  for (const unsigned *Alias = TM.getRegisterInfo()->getAliasSet(Reg);
+       *Alias; ++Alias)
+    if (RegSet.count(*Alias))
+      return true;
+
+  return false;
+}
+
+// return true if the candidate is a delay filler.
+bool Filler::isDelayFiller(MachineBasicBlock &MBB,
+                           MachineBasicBlock::iterator candidate) {
+  if (candidate == MBB.begin())
+    return false;
+  const MCInstrDesc &prevdesc = (--candidate)->getDesc();
+  return prevdesc.hasDelaySlot();
+}
diff --git a/test/CodeGen/Mips/brdelayslot.ll b/test/CodeGen/Mips/brdelayslot.ll
new file mode 100644
index 0000000000..b266ce61a8
--- /dev/null
+++ b/test/CodeGen/Mips/brdelayslot.ll
@@ -0,0 +1,15 @@
+; RUN: llc -march=mipsel  -enable-mips-delay-filler < %s | FileCheck %s
+
+define void @foo1() nounwind {
+entry:
+; CHECK:      jalr 
+; CHECK-NOT:  nop 
+; CHECK:      jr 
+; CHECK-NOT:  nop
+; CHECK:      .end
+
+  tail call void @foo2(i32 3) nounwind
+  ret void
+}
+
+declare void @foo2(i32)