Add Target hook to duplicate machine instructions.

Some instructions refer to unique labels, and so cannot be trivially cloned
with CloneMachineInstr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@92873 91177308-0d34-0410-b5e6-96231b3b80d8

6 files changed, 151 insertions, 22 deletions

diff --git a/include/llvm/Target/TargetInstrInfo.h b/include/llvm/Target/TargetInstrInfo.h
index 1bcd6fd..f779801 100644
--- a/include/llvm/Target/TargetInstrInfo.h
+++ b/include/llvm/Target/TargetInstrInfo.h
@@ -232,6 +232,12 @@ public:
                              const MachineInstr *Orig,
                              const TargetRegisterInfo *TRI) const = 0;
 
+  /// duplicate - Create a duplicate of the Orig instruction in MF. This is like
+  /// MachineFunction::CloneMachineInstr(), but the target may update operands
+  /// that are required to be unique.
+  virtual MachineInstr *duplicate(MachineInstr *Orig,
+                                  MachineFunction &MF) const = 0;
+
   /// convertToThreeAddress - This method must be implemented by targets that
   /// set the M_CONVERTIBLE_TO_3_ADDR flag.  When this flag is set, the target
   /// may be able to convert a two-address instruction into one or more true
@@ -560,6 +566,8 @@ public:
                              unsigned DestReg, unsigned SubReg,
                              const MachineInstr *Orig,
                              const TargetRegisterInfo *TRI) const;
+  virtual MachineInstr *duplicate(MachineInstr *Orig,
+                                  MachineFunction &MF) const;
   virtual bool isIdentical(const MachineInstr *MI,
                            const MachineInstr *Other,
                            const MachineRegisterInfo *MRI) const;
diff --git a/lib/CodeGen/TailDuplication.cpp b/lib/CodeGen/TailDuplication.cpp
index fddbebd9..c99c74c 100644
--- a/lib/CodeGen/TailDuplication.cpp
+++ b/lib/CodeGen/TailDuplication.cpp
@@ -346,7 +346,7 @@ void TailDuplicatePass::DuplicateInstruction(MachineInstr *MI,
                                      MachineBasicBlock *PredBB,
                                      MachineFunction &MF,
                                      DenseMap<unsigned, unsigned> &LocalVRMap) {
-  MachineInstr *NewMI = MF.CloneMachineInstr(MI);
+  MachineInstr *NewMI = TII->duplicate(MI, MF);
   for (unsigned i = 0, e = NewMI->getNumOperands(); i != e; ++i) {
     MachineOperand &MO = NewMI->getOperand(i);
     if (!MO.isReg())
diff --git a/lib/CodeGen/TargetInstrInfoImpl.cpp b/lib/CodeGen/TargetInstrInfoImpl.cpp
index 393e315..a0fccab 100644
--- a/lib/CodeGen/TargetInstrInfoImpl.cpp
+++ b/lib/CodeGen/TargetInstrInfoImpl.cpp
@@ -150,6 +150,13 @@ void TargetInstrInfoImpl::reMaterialize(MachineBasicBlock &MBB,
   MBB.insert(I, MI);
 }
 
+MachineInstr *TargetInstrInfoImpl::duplicate(MachineInstr *Orig,
+                                             MachineFunction &MF) const {
+  assert(!Orig->getDesc().isNotDuplicable() &&
+         "Instruction cannot be duplicated");
+  return MF.CloneMachineInstr(Orig);
+}
+
 bool
 TargetInstrInfoImpl::isIdentical(const MachineInstr *MI,
                                  const MachineInstr *Other,
diff --git a/lib/Target/ARM/ARMBaseInstrInfo.cpp b/lib/Target/ARM/ARMBaseInstrInfo.cpp
index 7cfa097..969c4a4 100644
--- a/lib/Target/ARM/ARMBaseInstrInfo.cpp
+++ b/lib/Target/ARM/ARMBaseInstrInfo.cpp
@@ -938,6 +938,35 @@ ARMBaseInstrInfo::canFoldMemoryOperand(const MachineInstr *MI,
   return false;
 }
 
+/// Create a copy of a const pool value. Update CPI to the new index and return
+/// the label UID.
+static unsigned duplicateCPV(MachineFunction &MF, unsigned &CPI) {
+  MachineConstantPool *MCP = MF.getConstantPool();
+  ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
+
+  const MachineConstantPoolEntry &MCPE = MCP->getConstants()[CPI];
+  assert(MCPE.isMachineConstantPoolEntry() &&
+         "Expecting a machine constantpool entry!");
+  ARMConstantPoolValue *ACPV =
+    static_cast<ARMConstantPoolValue*>(MCPE.Val.MachineCPVal);
+
+  unsigned PCLabelId = AFI->createConstPoolEntryUId();
+  ARMConstantPoolValue *NewCPV = 0;
+  if (ACPV->isGlobalValue())
+    NewCPV = new ARMConstantPoolValue(ACPV->getGV(), PCLabelId,
+                                      ARMCP::CPValue, 4);
+  else if (ACPV->isExtSymbol())
+    NewCPV = new ARMConstantPoolValue(MF.getFunction()->getContext(),
+                                      ACPV->getSymbol(), PCLabelId, 4);
+  else if (ACPV->isBlockAddress())
+    NewCPV = new ARMConstantPoolValue(ACPV->getBlockAddress(), PCLabelId,
+                                      ARMCP::CPBlockAddress, 4);
+  else
+    llvm_unreachable("Unexpected ARM constantpool value type!!");
+  CPI = MCP->getConstantPoolIndex(NewCPV, MCPE.getAlignment());
+  return PCLabelId;
+}
+
 void ARMBaseInstrInfo::
 reMaterialize(MachineBasicBlock &MBB,
               MachineBasicBlock::iterator I,
@@ -960,28 +989,8 @@ reMaterialize(MachineBasicBlock &MBB,
   case ARM::tLDRpci_pic:
   case ARM::t2LDRpci_pic: {
     MachineFunction &MF = *MBB.getParent();
-    ARMFunctionInfo *AFI = MF.getInfo<ARMFunctionInfo>();
-    MachineConstantPool *MCP = MF.getConstantPool();
     unsigned CPI = Orig->getOperand(1).getIndex();
-    const MachineConstantPoolEntry &MCPE = MCP->getConstants()[CPI];
-    assert(MCPE.isMachineConstantPoolEntry() &&
-           "Expecting a machine constantpool entry!");
-    ARMConstantPoolValue *ACPV =
-      static_cast<ARMConstantPoolValue*>(MCPE.Val.MachineCPVal);
-    unsigned PCLabelId = AFI->createConstPoolEntryUId();
-    ARMConstantPoolValue *NewCPV = 0;
-    if (ACPV->isGlobalValue())
-      NewCPV = new ARMConstantPoolValue(ACPV->getGV(), PCLabelId,
-                                        ARMCP::CPValue, 4);
-    else if (ACPV->isExtSymbol())
-      NewCPV = new ARMConstantPoolValue(MF.getFunction()->getContext(),
-                                        ACPV->getSymbol(), PCLabelId, 4);
-    else if (ACPV->isBlockAddress())
-      NewCPV = new ARMConstantPoolValue(ACPV->getBlockAddress(), PCLabelId,
-                                        ARMCP::CPBlockAddress, 4);
-    else
-      llvm_unreachable("Unexpected ARM constantpool value type!!");
-    CPI = MCP->getConstantPoolIndex(NewCPV, MCPE.getAlignment());
+    unsigned PCLabelId = duplicateCPV(MF, CPI);
     MachineInstrBuilder MIB = BuildMI(MBB, I, Orig->getDebugLoc(), get(Opcode),
                                       DestReg)
       .addConstantPoolIndex(CPI).addImm(PCLabelId);
@@ -994,6 +1003,22 @@ reMaterialize(MachineBasicBlock &MBB,
   NewMI->getOperand(0).setSubReg(SubIdx);
 }
 
+MachineInstr *
+ARMBaseInstrInfo::duplicate(MachineInstr *Orig, MachineFunction &MF) const {
+  MachineInstr *MI = TargetInstrInfoImpl::duplicate(Orig, MF);
+  switch(Orig->getOpcode()) {
+  case ARM::tLDRpci_pic:
+  case ARM::t2LDRpci_pic: {
+    unsigned CPI = Orig->getOperand(1).getIndex();
+    unsigned PCLabelId = duplicateCPV(MF, CPI);
+    Orig->getOperand(1).setIndex(CPI);
+    Orig->getOperand(2).setImm(PCLabelId);
+    break;
+  }
+  }
+  return MI;
+}
+
 bool ARMBaseInstrInfo::isIdentical(const MachineInstr *MI0,
                                   const MachineInstr *MI1,
                                   const MachineRegisterInfo *MRI) const {
diff --git a/lib/Target/ARM/ARMBaseInstrInfo.h b/lib/Target/ARM/ARMBaseInstrInfo.h
index 78d9135..0d9d4a7 100644
--- a/lib/Target/ARM/ARMBaseInstrInfo.h
+++ b/lib/Target/ARM/ARMBaseInstrInfo.h
@@ -287,6 +287,8 @@ public:
                              const MachineInstr *Orig,
                              const TargetRegisterInfo *TRI) const;
 
+  MachineInstr *duplicate(MachineInstr *Orig, MachineFunction &MF) const;
+
   virtual bool isIdentical(const MachineInstr *MI, const MachineInstr *Other,
                            const MachineRegisterInfo *MRI) const;
 };
diff --git a/test/CodeGen/Thumb2/2010-01-06-TailDuplicateLabels.ll b/test/CodeGen/Thumb2/2010-01-06-TailDuplicateLabels.ll
new file mode 100644
index 0000000..a632715
--- /dev/null
+++ b/test/CodeGen/Thumb2/2010-01-06-TailDuplicateLabels.ll
@@ -0,0 +1,87 @@
+; RUN: llc -relocation-model=pic -pre-regalloc-taildup < %s | grep {:$} | sort | uniq -d | count 0
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32"
+target triple = "thumbv7-apple-darwin10"
+
+%struct.PlatformMutex = type { i32, [40 x i8] }
+%struct.SpinLock = type { %struct.PlatformMutex }
+%"struct.WTF::TCMalloc_ThreadCache" = type { i32, %struct._opaque_pthread_t*, i8, [68 x %"struct.WTF::TCMalloc_ThreadCache_FreeList"], i32, i32, %"struct.WTF::TCMalloc_ThreadCache"*, %"struct.WTF::TCMalloc_ThreadCache"* }
+%"struct.WTF::TCMalloc_ThreadCache_FreeList" = type { i8*, i16, i16 }
+%struct.__darwin_pthread_handler_rec = type { void (i8*)*, i8*, %struct.__darwin_pthread_handler_rec* }
+%struct._opaque_pthread_t = type { i32, %struct.__darwin_pthread_handler_rec*, [596 x i8] }
+
+@_ZN3WTFL8heap_keyE = internal global i32 0       ; <i32*> [#uses=1]
+@_ZN3WTFL10tsd_initedE.b = internal global i1 false ; <i1*> [#uses=2]
+@_ZN3WTFL13pageheap_lockE = internal global %struct.SpinLock { %struct.PlatformMutex { i32 850045863, [40 x i8] zeroinitializer } } ; <%struct.SpinLock*> [#uses=1]
+@_ZN3WTFL12thread_heapsE = internal global %"struct.WTF::TCMalloc_ThreadCache"* null ; <%"struct.WTF::TCMalloc_ThreadCache"**> [#uses=1]
+@llvm.used = appending global [1 x i8*] [i8* bitcast (%"struct.WTF::TCMalloc_ThreadCache"* ()* @_ZN3WTF20TCMalloc_ThreadCache22CreateCacheIfNecessaryEv to i8*)], section "llvm.metadata" ; <[1 x i8*]*> [#uses=0]
+
+define arm_apcscc %"struct.WTF::TCMalloc_ThreadCache"* @_ZN3WTF20TCMalloc_ThreadCache22CreateCacheIfNecessaryEv() nounwind {
+entry:
+  %0 = tail call arm_apcscc  i32 @pthread_mutex_lock(%struct.PlatformMutex* getelementptr inbounds (%struct.SpinLock* @_ZN3WTFL13pageheap_lockE, i32 0, i32 0)) nounwind
+  %.b24 = load i1* @_ZN3WTFL10tsd_initedE.b, align 4 ; <i1> [#uses=1]
+  br i1 %.b24, label %bb5, label %bb6
+
+bb5:                                              ; preds = %entry
+  %1 = tail call arm_apcscc  %struct._opaque_pthread_t* @pthread_self() nounwind
+  br label %bb6
+
+bb6:                                              ; preds = %bb5, %entry
+  %me.0 = phi %struct._opaque_pthread_t* [ %1, %bb5 ], [ null, %entry ] ; <%struct._opaque_pthread_t*> [#uses=2]
+  br label %bb11
+
+bb7:                                              ; preds = %bb11
+  %2 = getelementptr inbounds %"struct.WTF::TCMalloc_ThreadCache"* %h.0, i32 0, i32 1
+  %3 = load %struct._opaque_pthread_t** %2, align 4
+  %4 = tail call arm_apcscc  i32 @pthread_equal(%struct._opaque_pthread_t* %3, %struct._opaque_pthread_t* %me.0) nounwind
+  %5 = icmp eq i32 %4, 0
+  br i1 %5, label %bb10, label %bb14
+
+bb10:                                             ; preds = %bb7
+  %6 = getelementptr inbounds %"struct.WTF::TCMalloc_ThreadCache"* %h.0, i32 0, i32 6
+  br label %bb11
+
+bb11:                                             ; preds = %bb10, %bb6
+  %h.0.in = phi %"struct.WTF::TCMalloc_ThreadCache"** [ @_ZN3WTFL12thread_heapsE, %bb6 ], [ %6, %bb10 ] ; <%"struct.WTF::TCMalloc_ThreadCache"**> [#uses=1]
+  %h.0 = load %"struct.WTF::TCMalloc_ThreadCache"** %h.0.in, align 4 ; <%"struct.WTF::TCMalloc_ThreadCache"*> [#uses=4]
+  %7 = icmp eq %"struct.WTF::TCMalloc_ThreadCache"* %h.0, null
+  br i1 %7, label %bb13, label %bb7
+
+bb13:                                             ; preds = %bb11
+  %8 = tail call arm_apcscc  %"struct.WTF::TCMalloc_ThreadCache"* @_ZN3WTF20TCMalloc_ThreadCache7NewHeapEP17_opaque_pthread_t(%struct._opaque_pthread_t* %me.0) nounwind
+  br label %bb14
+
+bb14:                                             ; preds = %bb13, %bb7
+  %heap.1 = phi %"struct.WTF::TCMalloc_ThreadCache"* [ %8, %bb13 ], [ %h.0, %bb7 ] ; <%"struct.WTF::TCMalloc_ThreadCache"*> [#uses=4]
+  %9 = tail call arm_apcscc  i32 @pthread_mutex_unlock(%struct.PlatformMutex* getelementptr inbounds (%struct.SpinLock* @_ZN3WTFL13pageheap_lockE, i32 0, i32 0)) nounwind
+  %10 = getelementptr inbounds %"struct.WTF::TCMalloc_ThreadCache"* %heap.1, i32 0, i32 2
+  %11 = load i8* %10, align 4
+  %toBool15not = icmp eq i8 %11, 0                ; <i1> [#uses=1]
+  br i1 %toBool15not, label %bb19, label %bb22
+
+bb19:                                             ; preds = %bb14
+  %.b = load i1* @_ZN3WTFL10tsd_initedE.b, align 4 ; <i1> [#uses=1]
+  br i1 %.b, label %bb21, label %bb22
+
+bb21:                                             ; preds = %bb19
+  store i8 1, i8* %10, align 4
+  %12 = load i32* @_ZN3WTFL8heap_keyE, align 4
+  %13 = bitcast %"struct.WTF::TCMalloc_ThreadCache"* %heap.1 to i8*
+  %14 = tail call arm_apcscc  i32 @pthread_setspecific(i32 %12, i8* %13) nounwind
+  ret %"struct.WTF::TCMalloc_ThreadCache"* %heap.1
+
+bb22:                                             ; preds = %bb19, %bb14
+  ret %"struct.WTF::TCMalloc_ThreadCache"* %heap.1
+}
+
+declare arm_apcscc i32 @pthread_mutex_lock(%struct.PlatformMutex*)
+
+declare arm_apcscc i32 @pthread_mutex_unlock(%struct.PlatformMutex*)
+
+declare hidden arm_apcscc %"struct.WTF::TCMalloc_ThreadCache"* @_ZN3WTF20TCMalloc_ThreadCache7NewHeapEP17_opaque_pthread_t(%struct._opaque_pthread_t*) nounwind
+
+declare arm_apcscc i32 @pthread_setspecific(i32, i8*)
+
+declare arm_apcscc %struct._opaque_pthread_t* @pthread_self()
+
+declare arm_apcscc i32 @pthread_equal(%struct._opaque_pthread_t*, %struct._opaque_pthread_t*)
+