Add two additional hazard recognizer functions

This adds two additional functions to the hazard recognizer interface. These
are optional (in the sense that the default implementations preserve the
current behavior), and used by the post-RA scheduler. Upcoming commits will use
this functionality in order to improve dispatch-group formation on the POWER7
and related cores. Dispatch groups are an odd construct: sometimes we need to
insert nops to force a new one to start (for performance reasons), and some
instructions need to appear in certain positions within a group, but the groups
are not fundamentally cycle based (they can contain instructions with data
dependencies with non-trivial latencies).

Motivation:

unsigned PreEmitNoops(SUnit *) - Used to force the post-RA scheduler to insert
nops to force a new dispatch group to begin. We already have a NoopHazard, and
this is also still needed. However, NoopHazard only causes a nop to be inserted
if there are no other available instructions, and so is not always sufficient.
The number of nops to insert depends on state that only the hazard recognizer
has, so a general callback is necessary.

bool ShouldPreferAnother(SUnit *) - Used to avoid scheduling instructions that
would start a new dispatch group when others are available that could be part
of the current dispatch group. In this case, we don't want to issue nops,
because the non-preferred instruction will implicitly start a new dispatch
group regardless.

Although the motivation for these functions is driven by the PowerPC backend,
they are completely general.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@197084 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/CodeGen/ScheduleHazardRecognizer.h

@@ -70,6 +70,22 @@ public:
   /// emitted, to advance the hazard state.
   virtual void EmitInstruction(SUnit *) {}
 
+  /// PreEmitNoops - This callback is invoked prior to emitting an instruction.
+  /// It should return the number of noops to emit prior to the provided
+  /// instruction.
+  /// Note: This is only used during PostRA scheduling. EmitNoop is not called
+  /// for these noops.
+  virtual unsigned PreEmitNoops(SUnit *) {
+    return 0;
+  }
+
+  /// ShouldPreferAnother - This callback may be invoked if getHazardType
+  /// returns NoHazard. If, even though there is no hazard, it would be better to
+  /// schedule another available instruction, this callback should return true.
+  virtual bool ShouldPreferAnother(SUnit *) {
+    return false;
+  }
+
   /// AdvanceCycle - This callback is invoked whenever the next top-down
   /// instruction to be scheduled cannot issue in the current cycle, either
   /// because of latency or resource conflicts.  This should increment the

lib/CodeGen/PostRASchedulerList.cpp

@@ -192,6 +192,7 @@ namespace {
     bool ToggleKillFlag(MachineInstr *MI, MachineOperand &MO);
 
     void dumpSchedule() const;
+    void emitNoop(unsigned CurCycle);
   };
 }
 
@@ -630,6 +631,14 @@ void SchedulePostRATDList::ScheduleNodeTopDown(SUnit *SU, unsigned CurCycle) {
   AvailableQueue.scheduledNode(SU);
 }
 
+/// emitNoop - Add a noop to the current instruction sequence.
+void SchedulePostRATDList::emitNoop(unsigned CurCycle) {
+  DEBUG(dbgs() << "*** Emitting noop in cycle " << CurCycle << '\n');
+  HazardRec->EmitNoop();
+  Sequence.push_back(0);   // NULL here means noop
+  ++NumNoops;
+}
+
 /// ListScheduleTopDown - The main loop of list scheduling for top-down
 /// schedulers.
 void SchedulePostRATDList::ListScheduleTopDown() {
@@ -678,7 +687,7 @@ void SchedulePostRATDList::ListScheduleTopDown() {
 
     DEBUG(dbgs() << "\n*** Examining Available\n"; AvailableQueue.dump(this));
 
-    SUnit *FoundSUnit = 0;
+    SUnit *FoundSUnit = 0, *NotPreferredSUnit = 0;
     bool HasNoopHazards = false;
     while (!AvailableQueue.empty()) {
       SUnit *CurSUnit = AvailableQueue.pop();
@@ -686,8 +695,19 @@ void SchedulePostRATDList::ListScheduleTopDown() {
       ScheduleHazardRecognizer::HazardType HT =
         HazardRec->getHazardType(CurSUnit, 0/*no stalls*/);
       if (HT == ScheduleHazardRecognizer::NoHazard) {
-        FoundSUnit = CurSUnit;
-        break;
+        if (HazardRec->ShouldPreferAnother(CurSUnit)) {
+          if (!NotPreferredSUnit) {
+	    // If this is the first non-preferred node for this cycle, then
+	    // record it and continue searching for a preferred node. If this
+	    // is not the first non-preferred node, then treat it as though
+	    // there had been a hazard.
+            NotPreferredSUnit = CurSUnit;
+            continue;
+          }
+        } else {
+          FoundSUnit = CurSUnit;
+          break;
+        }
       }
 
       // Remember if this is a noop hazard.
@@ -696,6 +716,20 @@ void SchedulePostRATDList::ListScheduleTopDown() {
       NotReady.push_back(CurSUnit);
     }
 
+    // If we have a non-preferred node, push it back onto the available list.
+    // If we did not find a preferred node, then schedule this first
+    // non-preferred node.
+    if (NotPreferredSUnit) {
+      if (!FoundSUnit) {
+        DEBUG(dbgs() << "*** Will schedule a non-preferred instruction...\n");
+        FoundSUnit = NotPreferredSUnit;
+      } else {
+        AvailableQueue.push(NotPreferredSUnit);
+      }
+
+      NotPreferredSUnit = 0;
+    }
+
     // Add the nodes that aren't ready back onto the available list.
     if (!NotReady.empty()) {
       AvailableQueue.push_all(NotReady);
@@ -704,6 +738,11 @@ void SchedulePostRATDList::ListScheduleTopDown() {
 
     // If we found a node to schedule...
     if (FoundSUnit) {
+      // If we need to emit noops prior to this instruction, then do so.
+      unsigned NumPreNoops = HazardRec->PreEmitNoops(FoundSUnit);
+      for (unsigned i = 0; i != NumPreNoops; ++i)
+        emitNoop(CurCycle);
+
       // ... schedule the node...
       ScheduleNodeTopDown(FoundSUnit, CurCycle);
       HazardRec->EmitInstruction(FoundSUnit);
@@ -728,10 +767,7 @@ void SchedulePostRATDList::ListScheduleTopDown() {
         // Otherwise, we have no instructions to issue and we have instructions
         // that will fault if we don't do this right.  This is the case for
         // processors without pipeline interlocks and other cases.
-        DEBUG(dbgs() << "*** Emitting noop in cycle " << CurCycle << '\n');
-        HazardRec->EmitNoop();
-        Sequence.push_back(0);   // NULL here means noop
-        ++NumNoops;
+        emitNoop(CurCycle);
       }
 
       ++CurCycle;