[SystemZ] Improved debug dumping during post-RA scheduling.

Review: Ulrich Weigand
llvm-svn: 326878

lib/Target/SystemZ/SystemZHazardRecognizer.cpp

@@ -59,7 +59,7 @@ getNumDecoderSlots(SUnit *SU) const {
   return 1; // Normal instruction
 }
 
-unsigned SystemZHazardRecognizer::getCurrCycleIdx() {
+unsigned SystemZHazardRecognizer::getCurrCycleIdx() const {
   unsigned Idx = CurrGroupSize;
   if (GrpCount % 2)
     Idx += 3;
@@ -100,30 +100,30 @@ SystemZHazardRecognizer::fitsIntoCurrentGroup(SUnit *SU) const {
   return true;
 }
 
-void SystemZHazardRecognizer::nextGroup(bool DbgOutput) {
-  if (CurrGroupSize > 0) {
-    DEBUG(dumpCurrGroup("Completed decode group"));
-    DEBUG(CurGroupDbg = "";);
+void SystemZHazardRecognizer::nextGroup() {
+  if (CurrGroupSize == 0)
+    return;
 
-    GrpCount++;
+  DEBUG(dumpCurrGroup("Completed decode group"));
+  DEBUG(CurGroupDbg = "";);
 
-    // Reset counter for next group.
-    CurrGroupSize = 0;
+  GrpCount++;
 
-    // Decrease counters for execution units by one.
-    for (unsigned i = 0; i < SchedModel->getNumProcResourceKinds(); ++i)
-      if (ProcResourceCounters[i] > 0)
-        ProcResourceCounters[i]--;
+  // Reset counter for next group.
+  CurrGroupSize = 0;
 
-    // Clear CriticalResourceIdx if it is now below the threshold.
-    if (CriticalResourceIdx != UINT_MAX &&
-        (ProcResourceCounters[CriticalResourceIdx] <=
-         ProcResCostLim))
-      CriticalResourceIdx = UINT_MAX;
-  }
+  // Decrease counters for execution units by one.
+  for (unsigned i = 0; i < SchedModel->getNumProcResourceKinds(); ++i)
+    if (ProcResourceCounters[i] > 0)
+      ProcResourceCounters[i]--;
 
-  DEBUG(if (DbgOutput)
-          dumpProcResourceCounters(););
+  // Clear CriticalResourceIdx if it is now below the threshold.
+  if (CriticalResourceIdx != UINT_MAX &&
+      (ProcResourceCounters[CriticalResourceIdx] <=
+       ProcResCostLim))
+    CriticalResourceIdx = UINT_MAX;
+
+  DEBUG(dumpState(););
 }
 
 #ifndef NDEBUG // Debug output
@@ -163,7 +163,7 @@ void SystemZHazardRecognizer::dumpSU(SUnit *SU, raw_ostream &OS) const {
 }
 
 void SystemZHazardRecognizer::dumpCurrGroup(std::string Msg) const {
-  dbgs() << "+++ " << Msg;
+  dbgs() << "++ " << Msg;
   dbgs() << ": ";
 
   if (CurGroupDbg.empty())
@@ -188,15 +188,28 @@ void SystemZHazardRecognizer::dumpProcResourceCounters() const {
   if (!any)
     return;
 
-  dbgs() << "+++ Resource counters:\n";
+  dbgs() << "++ | Resource counters: ";
   for (unsigned i = 0; i < SchedModel->getNumProcResourceKinds(); ++i)
-    if (ProcResourceCounters[i] > 0) {
-      dbgs() << "+++ Extra schedule for execution unit "
-             << SchedModel->getProcResource(i)->Name
-             << ": " << ProcResourceCounters[i] << "\n";
-      any = true;
-    }
+    if (ProcResourceCounters[i] > 0)
+      dbgs() << SchedModel->getProcResource(i)->Name
+             << ":" << ProcResourceCounters[i] << " ";
+  dbgs() << "\n";
+
+  if (CriticalResourceIdx != UINT_MAX)
+    dbgs() << "++ | Critical resource: "
+           << SchedModel->getProcResource(CriticalResourceIdx)->Name
+           << "\n";
 }
+
+void SystemZHazardRecognizer::dumpState() const {
+  dumpCurrGroup("| Current decoder group");
+  dbgs() << "++ | Current cycle index: "
+         << getCurrCycleIdx() << "\n";
+  dumpProcResourceCounters();
+  if (LastFPdOpCycleIdx != UINT_MAX)
+    dbgs() << "++ | Last FPd cycle index: " << LastFPdOpCycleIdx << "\n";
+}
+
 #endif //NDEBUG
 
 void SystemZHazardRecognizer::clearProcResCounters() {
@@ -213,25 +226,25 @@ static inline bool isBranchRetTrap(MachineInstr *MI) {
 void SystemZHazardRecognizer::
 EmitInstruction(SUnit *SU) {
   const MCSchedClassDesc *SC = getSchedClass(SU);
-  DEBUG( dumpCurrGroup("Decode group before emission"););
+  DEBUG(dbgs() << "++ HazardRecognizer emitting "; dumpSU(SU, dbgs());
+        dbgs() << "\n";);
+  DEBUG(dumpCurrGroup("Decode group before emission"););
 
   // If scheduling an SU that must begin a new decoder group, move on
   // to next group.
   if (!fitsIntoCurrentGroup(SU))
     nextGroup();
 
-  DEBUG( dbgs() << "+++ HazardRecognizer emitting "; dumpSU(SU, dbgs());
-         dbgs() << "\n";
-         raw_string_ostream cgd(CurGroupDbg);
-         if (CurGroupDbg.length())
-           cgd << ", ";
-         dumpSU(SU, cgd););
+  DEBUG(raw_string_ostream cgd(CurGroupDbg);
+        if (CurGroupDbg.length())
+          cgd << ", ";
+        dumpSU(SU, cgd););
 
   LastEmittedMI = SU->getInstr();
 
   // After returning from a call, we don't know much about the state.
   if (SU->isCall) {
-    DEBUG (dbgs() << "+++ Clearing state after call.\n";);
+    DEBUG(dbgs() << "++ Clearing state after call.\n";);
     clearProcResCounters();
     LastFPdOpCycleIdx = UINT_MAX;
     CurrGroupSize += getNumDecoderSlots(SU);
@@ -256,9 +269,9 @@ EmitInstruction(SUnit *SU) {
          (PI->ProcResourceIdx != CriticalResourceIdx &&
           CurrCounter >
           ProcResourceCounters[CriticalResourceIdx]))) {
-      DEBUG( dbgs() << "+++ New critical resource: "
-             << SchedModel->getProcResource(PI->ProcResourceIdx)->Name
-             << "\n";);
+      DEBUG(dbgs() << "++ New critical resource: "
+            << SchedModel->getProcResource(PI->ProcResourceIdx)->Name
+            << "\n";);
       CriticalResourceIdx = PI->ProcResourceIdx;
     }
   }
@@ -266,8 +279,8 @@ EmitInstruction(SUnit *SU) {
   // Make note of an instruction that uses a blocking resource (FPd).
   if (SU->isUnbuffered) {
     LastFPdOpCycleIdx = getCurrCycleIdx();
-    DEBUG (dbgs() << "+++ Last FPd cycle index: "
-           << LastFPdOpCycleIdx << "\n";);
+    DEBUG(dbgs() << "++ Last FPd cycle index: "
+          << LastFPdOpCycleIdx << "\n";);
   }
 
   bool GroupEndingBranch =
@@ -376,7 +389,7 @@ void SystemZHazardRecognizer::emitInstruction(MachineInstr *MI,
   EmitInstruction(&SU);
 
   if (TakenBranch && CurrGroupSize > 0)
-    nextGroup(false /*DbgOutput*/);
+    nextGroup();
 
   assert ((!MI->isTerminator() || isBranchRetTrap(MI)) &&
           "Scheduler: unhandled terminator!");
@@ -386,7 +399,7 @@ void SystemZHazardRecognizer::
 copyState(SystemZHazardRecognizer *Incoming) {
   // Current decoder group
   CurrGroupSize = Incoming->CurrGroupSize;
-  DEBUG (CurGroupDbg = Incoming->CurGroupDbg;);
+  DEBUG(CurGroupDbg = Incoming->CurGroupDbg;);
 
   // Processor resources
   ProcResourceCounters = Incoming->ProcResourceCounters;

lib/Target/SystemZ/SystemZHazardRecognizer.h

@@ -76,7 +76,7 @@ class SystemZHazardRecognizer : public ScheduleHazardRecognizer {
   /// Two decoder groups per cycle are formed (for z13), meaning 2x3
   /// instructions. This function returns a number between 0 and 5,
   /// representing the current decoder slot of the current cycle.
-  unsigned getCurrCycleIdx();
+  unsigned getCurrCycleIdx() const;
 
   /// LastFPdOpCycleIdx stores the numbeer returned by getCurrCycleIdx()
   /// when a stalling operation is scheduled (which uses the FPd resource).
@@ -88,7 +88,7 @@ class SystemZHazardRecognizer : public ScheduleHazardRecognizer {
   unsigned getCurrGroupSize() {return CurrGroupSize;};
 
   /// Start next decoder group.
-  void nextGroup(bool DbgOutput = true);
+  void nextGroup();
 
   /// Clear all counters for processor resources.
   void clearProcResCounters();
@@ -145,6 +145,7 @@ public:
   void dumpSU(SUnit *SU, raw_ostream &OS) const;
   void dumpCurrGroup(std::string Msg = "") const;
   void dumpProcResourceCounters() const;
+  void dumpState() const;
 #endif
 
   MachineBasicBlock::iterator getLastEmittedMI() { return LastEmittedMI; }

lib/Target/SystemZ/SystemZMachineScheduler.cpp

@@ -71,19 +71,24 @@ advanceTo(MachineBasicBlock::iterator NextBegin) {
   }
 }
 
+void SystemZPostRASchedStrategy::initialize(ScheduleDAGMI *dag) {
+  DEBUG(HazardRec->dumpState(););
+}
+
 void SystemZPostRASchedStrategy::enterMBB(MachineBasicBlock *NextMBB) {
   assert ((SchedStates.find(NextMBB) == SchedStates.end()) &&
           "Entering MBB twice?");
-  DEBUG(dbgs() << "+++ Entering " << printMBBReference(*NextMBB));
+  DEBUG(dbgs() << "** Entering " << printMBBReference(*NextMBB));
 
   MBB = NextMBB;
+
   /// Create a HazardRec for MBB, save it in SchedStates and set HazardRec to
   /// point to it.
   HazardRec = SchedStates[MBB] = new SystemZHazardRecognizer(TII, &SchedModel);
-  DEBUG (const MachineLoop *Loop = MLI->getLoopFor(MBB);
-         if(Loop && Loop->getHeader() == MBB)
-           dbgs() << " (Loop header)";
-         dbgs() << ":\n";);
+  DEBUG(const MachineLoop *Loop = MLI->getLoopFor(MBB);
+        if(Loop && Loop->getHeader() == MBB)
+          dbgs() << " (Loop header)";
+        dbgs() << ":\n";);
 
   // Try to take over the state from a single predecessor, if it has been
   // scheduled. If this is not possible, we are done.
@@ -93,16 +98,17 @@ void SystemZPostRASchedStrategy::enterMBB(MachineBasicBlock *NextMBB) {
       SchedStates.find(SinglePredMBB) == SchedStates.end())
     return;
 
-  DEBUG(dbgs() << "+++ Continued scheduling from "
-               << printMBBReference(*SinglePredMBB) << "\n";);
+  DEBUG(dbgs() << "** Continued scheduling from "
+        << printMBBReference(*SinglePredMBB) << "\n";);
 
   HazardRec->copyState(SchedStates[SinglePredMBB]);
+  DEBUG(HazardRec->dumpState(););
 
   // Emit incoming terminator(s). Be optimistic and assume that branch
   // prediction will generally do "the right thing".
   for (MachineBasicBlock::iterator I = SinglePredMBB->getFirstTerminator();
        I != SinglePredMBB->end(); I++) {
-    DEBUG (dbgs() << "+++ Emitting incoming branch: "; I->dump(););
+    DEBUG(dbgs() << "** Emitting incoming branch: "; I->dump(););
     bool TakenBranch = (I->isBranch() &&
       (TII->getBranchInfo(*I).Target->isReg() || // Relative branch
        TII->getBranchInfo(*I).Target->getMBB() == MBB));
@@ -113,7 +119,7 @@ void SystemZPostRASchedStrategy::enterMBB(MachineBasicBlock *NextMBB) {
 }
 
 void SystemZPostRASchedStrategy::leaveMBB() {
-  DEBUG(dbgs() << "+++ Leaving " << printMBBReference(*MBB) << "\n";);
+  DEBUG(dbgs() << "** Leaving " << printMBBReference(*MBB) << "\n";);
 
   // Advance to first terminator. The successor block will handle terminators
   // dependent on CFG layout (T/NT branch etc).
@@ -159,14 +165,14 @@ SUnit *SystemZPostRASchedStrategy::pickNode(bool &IsTopNode) {
 
   // If only one choice, return it.
   if (Available.size() == 1) {
-    DEBUG (dbgs() << "+++ Only one: ";
-           HazardRec->dumpSU(*Available.begin(), dbgs()); dbgs() << "\n";);
+    DEBUG(dbgs() << "** Only one: ";
+          HazardRec->dumpSU(*Available.begin(), dbgs()); dbgs() << "\n";);
     return *Available.begin();
   }
 
   // All nodes that are possible to schedule are stored by in the
   // Available set.
-  DEBUG(dbgs() << "+++ Available: "; Available.dump(*HazardRec););
+  DEBUG(dbgs() << "** Available: "; Available.dump(*HazardRec););
 
   Candidate Best;
   for (auto *SU : Available) {
@@ -177,15 +183,13 @@ SUnit *SystemZPostRASchedStrategy::pickNode(bool &IsTopNode) {
     // Remeber which SU is the best candidate.
     if (Best.SU == nullptr || c < Best) {
       Best = c;
-      DEBUG(dbgs() << "+++ Best sofar: ";
-            HazardRec->dumpSU(Best.SU, dbgs());
-            if (Best.GroupingCost != 0)
-              dbgs() << "\tGrouping cost:" << Best.GroupingCost;
-            if (Best.ResourcesCost != 0)
-              dbgs() << " Resource cost:" << Best.ResourcesCost;
-            dbgs() << " Height:" << Best.SU->getHeight();
-            dbgs() << "\n";);
-    }
+      DEBUG(dbgs() << "** Best so far: ";);
+    } else
+      DEBUG(dbgs() << "** Tried      : ";);
+    DEBUG(HazardRec->dumpSU(c.SU, dbgs());
+          c.dumpCosts();
+          dbgs() << " Height:" << c.SU->getHeight();
+          dbgs() << "\n";);
 
     // Once we know we have seen all SUs that affect grouping or use unbuffered
     // resources, we can stop iterating if Best looks good.
@@ -206,7 +210,7 @@ Candidate(SUnit *SU_, SystemZHazardRecognizer &HazardRec) : Candidate() {
   // if it would fit naturally into the schedule.
   GroupingCost = HazardRec.groupingCost(SU);
 
-    // Check the resources cost for this SU.
+  // Check the resources cost for this SU.
   ResourcesCost = HazardRec.resourcesCost(SU);
 }
 
@@ -239,7 +243,12 @@ operator<(const Candidate &other) {
 }
 
 void SystemZPostRASchedStrategy::schedNode(SUnit *SU, bool IsTopNode) {
-  DEBUG(dbgs() << "+++ Scheduling SU(" << SU->NodeNum << ")\n";);
+  DEBUG(dbgs() << "** Scheduling SU(" << SU->NodeNum << ") ";
+        if (Available.size() == 1)
+          dbgs() << "(only one) ";
+        Candidate c(SU, *HazardRec);
+        c.dumpCosts();
+        dbgs() << "\n";);
 
   // Remove SU from Available set and update HazardRec.
   Available.erase(SU);

lib/Target/SystemZ/SystemZMachineScheduler.h

@@ -58,6 +58,15 @@ class SystemZPostRASchedStrategy : public MachineSchedStrategy {
     bool noCost() const {
       return (GroupingCost <= 0 && !ResourcesCost);
     }
+
+#ifndef NDEBUG
+    void dumpCosts() {
+      if (GroupingCost != 0)
+        dbgs() << "  Grouping cost:" << GroupingCost;
+      if (ResourcesCost != 0)
+        dbgs() << "  Resource cost:" << ResourcesCost;
+    }
+#endif
   };
 
   // A sorter for the Available set that makes sure that SUs are considered
@@ -119,7 +128,7 @@ public:
   // transferrred over scheduling boundaries.
   bool doMBBSchedRegionsTopDown() const override { return true; }
 
-  void initialize(ScheduleDAGMI *dag) override {}
+  void initialize(ScheduleDAGMI *dag) override;
 
   /// Tell the strategy that MBB is about to be processed.
   void enterMBB(MachineBasicBlock *NextMBB) override;