5 files changed, 37 insertions, 14 deletions
diff --git a/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp b/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp
index 98dd34fe16..0728cef437 100644
--- a/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp
+++ b/lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp
@@ -1256,8 +1256,10 @@ bool src_ls_rr_sort::operator()(const SUnit *left, const SUnit *right) const {
 }
 
 bool hybrid_ls_rr_sort::operator()(const SUnit *left, const SUnit *right) const{
-  bool LStall = SPQ->getCurCycle() < left->getHeight();
-  bool RStall = SPQ->getCurCycle() < right->getHeight();
+  bool LStall = left->SchedulingPref == Sched::Latency &&
+    SPQ->getCurCycle() < left->getHeight();
+  bool RStall = right->SchedulingPref == Sched::Latency &&
+    SPQ->getCurCycle() < right->getHeight();
   // If scheduling one of the node will cause a pipeline stall, delay it.
   // If scheduling either one of the node will cause a pipeline stall, sort them
   // according to their height.
diff --git a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
index f765409d9e..3185c88b82 100644
--- a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
+++ b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.cpp
@@ -19,6 +19,7 @@
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetInstrInfo.h"
+#include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 #include "llvm/Target/TargetSubtarget.h"
 #include "llvm/ADT/DenseMap.h"
@@ -44,6 +45,24 @@ void ScheduleDAGSDNodes::Run(SelectionDAG *dag, MachineBasicBlock *bb,
   ScheduleDAG::Run(bb, insertPos);
 }
 
+/// NewSUnit - Creates a new SUnit and return a ptr to it.
+///
+SUnit *ScheduleDAGSDNodes::NewSUnit(SDNode *N) {
+#ifndef NDEBUG
+  const SUnit *Addr = 0;
+  if (!SUnits.empty())
+    Addr = &SUnits[0];
+#endif
+  SUnits.push_back(SUnit(N, (unsigned)SUnits.size()));
+  assert((Addr == 0 || Addr == &SUnits[0]) &&
+         "SUnits std::vector reallocated on the fly!");
+  SUnits.back().OrigNode = &SUnits.back();
+  SUnit *SU = &SUnits.back();
+  const TargetLowering &TLI = DAG->getTargetLoweringInfo();
+  SU->SchedulingPref = TLI.getSchedulingPreference(N);
+  return SU;
+}
+
 SUnit *ScheduleDAGSDNodes::Clone(SUnit *Old) {
   SUnit *SU = NewSUnit(Old->getNode());
   SU->OrigNode = Old->OrigNode;
@@ -52,6 +71,7 @@ SUnit *ScheduleDAGSDNodes::Clone(SUnit *Old) {
   SU->isCommutable = Old->isCommutable;
   SU->hasPhysRegDefs = Old->hasPhysRegDefs;
   SU->hasPhysRegClobbers = Old->hasPhysRegClobbers;
+  SU->SchedulingPref = Old->SchedulingPref;
   Old->isCloned = true;
   return SU;
 }
diff --git a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.h b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.h
index 42c58c105b..e8714ba832 100644
--- a/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.h
+++ b/lib/CodeGen/SelectionDAG/ScheduleDAGSDNodes.h
@@ -66,18 +66,7 @@ namespace llvm {
 
     /// NewSUnit - Creates a new SUnit and return a ptr to it.
     ///
-    SUnit *NewSUnit(SDNode *N) {
-#ifndef NDEBUG
-      const SUnit *Addr = 0;
-      if (!SUnits.empty())
-        Addr = &SUnits[0];
-#endif
-      SUnits.push_back(SUnit(N, (unsigned)SUnits.size()));
-      assert((Addr == 0 || Addr == &SUnits[0]) &&
-             "SUnits std::vector reallocated on the fly!");
-      SUnits.back().OrigNode = &SUnits.back();
-      return &SUnits.back();
-    }
+    SUnit *NewSUnit(SDNode *N);
 
     /// Clone - Creates a clone of the specified SUnit. It does not copy the
     /// predecessors / successors info nor the temporary scheduling states.
diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp
index 48f3bbfac9..340e887044 100644
--- a/lib/Target/ARM/ARMISelLowering.cpp
+++ b/lib/Target/ARM/ARMISelLowering.cpp
@@ -466,6 +466,7 @@ ARMTargetLowering::ARMTargetLowering(TargetMachine &TM)
   setTargetDAGCombine(ISD::MUL);
 
   setStackPointerRegisterToSaveRestore(ARM::SP);
+
   setSchedulingPreference(Sched::RegPressure);
 
   // FIXME: If-converter should use instruction latency to determine
@@ -600,6 +601,15 @@ unsigned ARMTargetLowering::getFunctionAlignment(const Function *F) const {
   return getTargetMachine().getSubtarget<ARMSubtarget>().isThumb() ? 0 : 1;
 }
 
+Sched::Preference ARMTargetLowering::getSchedulingPreference(SDNode *N) const {
+  for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) {
+    EVT VT = N->getValueType(i);
+    if (VT.isFloatingPoint() || VT.isVector())
+      return Sched::Latency;
+  }
+  return Sched::RegPressure;
+}
+
 //===----------------------------------------------------------------------===//
 // Lowering Code
 //===----------------------------------------------------------------------===//
diff --git a/lib/Target/ARM/ARMISelLowering.h b/lib/Target/ARM/ARMISelLowering.h
index 54dfb429c2..6a03d10fe4 100644
--- a/lib/Target/ARM/ARMISelLowering.h
+++ b/lib/Target/ARM/ARMISelLowering.h
@@ -247,6 +247,8 @@ namespace llvm {
     /// getFunctionAlignment - Return the Log2 alignment of this function.
     virtual unsigned getFunctionAlignment(const Function *F) const;
 
+    Sched::Preference getSchedulingPreference(SDNode *N) const;
+
     bool isShuffleMaskLegal(const SmallVectorImpl<int> &M, EVT VT) const;
     bool isOffsetFoldingLegal(const GlobalAddressSDNode *GA) const;