o3: fix tick used for renaming and issue with range selection

Fixes the tick used from rename:
- previously this gathered the tick on leaving rename which was always 1 less
  than the dispatch. This conflated the decode ticks when back pressure built
  in the pipeline.
- now picks up tick on entry.

Added --store_completions flag:
- will additionally display the store completion tail in the viewer.
- this highlights periods when large numbers of stores are outstanding (>16 LSQ
  blocking)

Allows selection by tick range (previously this caused an infinite loop)

src/cpu/o3/commit_impl.hh

@@ -60,6 +60,7 @@
 #include "debug/CommitRate.hh"
 #include "debug/Drain.hh"
 #include "debug/ExecFaulting.hh"
+#include "debug/O3PipeView.hh"
 #include "params/DerivO3CPU.hh"
 #include "sim/faults.hh"
 #include "sim/full_system.hh"
@@ -1267,7 +1268,9 @@ DefaultCommit<Impl>::commitHead(DynInstPtr &head_inst, unsigned inst_num)
     rob->retireHead(tid);
 
 #if TRACING_ON
-    head_inst->commitTick = curTick() - head_inst->fetchTick;
+    if (DTRACE(O3PipeView)) {
+        head_inst->commitTick = curTick() - head_inst->fetchTick;
+    }
 #endif
 
     // If this was a store, record it for this cycle.

src/cpu/o3/decode_impl.hh

@@ -47,6 +47,7 @@
 #include "cpu/inst_seq.hh"
 #include "debug/Activity.hh"
 #include "debug/Decode.hh"
+#include "debug/O3PipeView.hh"
 #include "params/DerivO3CPU.hh"
 #include "sim/full_system.hh"
 
@@ -714,7 +715,9 @@ DefaultDecode<Impl>::decodeInsts(ThreadID tid)
         --insts_available;
 
 #if TRACING_ON
-        inst->decodeTick = curTick() - inst->fetchTick;
+        if (DTRACE(O3PipeView)) {
+            inst->decodeTick = curTick() - inst->fetchTick;
+        }
 #endif
 
         // Ensure that if it was predicted as a branch, it really is a

src/cpu/o3/dyn_inst.hh

@@ -132,6 +132,7 @@ class BaseO3DynInst : public BaseDynInst<Impl>
     int32_t issueTick;
     int32_t completeTick;
     int32_t commitTick;
+    int32_t storeTick;
 #endif
 
     /** Reads a misc. register, including any side-effects the read

src/cpu/o3/dyn_inst_impl.hh

@@ -66,26 +66,35 @@ BaseO3DynInst<Impl>::BaseO3DynInst(StaticInstPtr _staticInst,
 template <class Impl>BaseO3DynInst<Impl>::~BaseO3DynInst()
 {
 #if TRACING_ON
-    Tick val, fetch = this->fetchTick;
-    // Print info needed by the pipeline activity viewer.
-    DPRINTFR(O3PipeView, "O3PipeView:fetch:%llu:0x%08llx:%d:%llu:%s\n",
-             fetch,
-             this->instAddr(),
-             this->microPC(),
-             this->seqNum,
-             this->staticInst->disassemble(this->instAddr()));
-    val = (this->decodeTick == -1) ? 0 : fetch + this->decodeTick;
-    DPRINTFR(O3PipeView, "O3PipeView:decode:%llu\n", val);
-    val = (this->renameTick == -1) ? 0 : fetch + this->renameTick;
-    DPRINTFR(O3PipeView, "O3PipeView:rename:%llu\n", val);
-    val = (this->dispatchTick == -1) ? 0 : fetch + this->dispatchTick;
-    DPRINTFR(O3PipeView, "O3PipeView:dispatch:%llu\n", val);
-    val = (this->issueTick == -1) ? 0 : fetch + this->issueTick;
-    DPRINTFR(O3PipeView, "O3PipeView:issue:%llu\n", val);
-    val = (this->completeTick == -1) ? 0 : fetch + this->completeTick;
-    DPRINTFR(O3PipeView, "O3PipeView:complete:%llu\n", val);
-    val = (this->commitTick == -1) ? 0 : fetch + this->commitTick;
-    DPRINTFR(O3PipeView, "O3PipeView:retire:%llu\n", val);
+    if (DTRACE(O3PipeView)) {
+        Tick fetch = this->fetchTick;
+        // fetchTick can be -1 if the instruction fetched outside the trace window.
+        if (fetch != -1) {
+            Tick val;
+            // Print info needed by the pipeline activity viewer.
+            DPRINTFR(O3PipeView, "O3PipeView:fetch:%llu:0x%08llx:%d:%llu:%s\n",
+                     fetch,
+                     this->instAddr(),
+                     this->microPC(),
+                     this->seqNum,
+                     this->staticInst->disassemble(this->instAddr()));
+
+            val = (this->decodeTick == -1) ? 0 : fetch + this->decodeTick;
+            DPRINTFR(O3PipeView, "O3PipeView:decode:%llu\n", val);
+            val = (this->renameTick == -1) ? 0 : fetch + this->renameTick;
+            DPRINTFR(O3PipeView, "O3PipeView:rename:%llu\n", val);
+            val = (this->dispatchTick == -1) ? 0 : fetch + this->dispatchTick;
+            DPRINTFR(O3PipeView, "O3PipeView:dispatch:%llu\n", val);
+            val = (this->issueTick == -1) ? 0 : fetch + this->issueTick;
+            DPRINTFR(O3PipeView, "O3PipeView:issue:%llu\n", val);
+            val = (this->completeTick == -1) ? 0 : fetch + this->completeTick;
+            DPRINTFR(O3PipeView, "O3PipeView:complete:%llu\n", val);
+            val = (this->commitTick == -1) ? 0 : fetch + this->commitTick;
+
+            Tick valS = (this->storeTick == -1) ? 0 : fetch + this->storeTick;
+            DPRINTFR(O3PipeView, "O3PipeView:retire:%llu:store:%llu\n", val, valS);
+        }
+    }
 #endif
 };
 
@@ -119,6 +128,7 @@ BaseO3DynInst<Impl>::initVars()
     issueTick = -1;
     completeTick = -1;
     commitTick = -1;
+    storeTick = -1;
 #endif
 }
 

src/cpu/o3/fetch_impl.hh

@@ -60,6 +60,7 @@
 #include "debug/Activity.hh"
 #include "debug/Drain.hh"
 #include "debug/Fetch.hh"
+#include "debug/O3PipeView.hh"
 #include "mem/packet.hh"
 #include "params/DerivO3CPU.hh"
 #include "sim/byteswap.hh"
@@ -1313,7 +1314,9 @@ DefaultFetch<Impl>::fetch(bool &status_change)
             numInst++;
 
 #if TRACING_ON
-            instruction->fetchTick = curTick();
+            if (DTRACE(O3PipeView)) {
+                instruction->fetchTick = curTick();
+            }
 #endif
 
             nextPC = thisPC;

src/cpu/o3/iew_impl.hh

@@ -56,6 +56,7 @@
 #include "debug/Decode.hh"
 #include "debug/Drain.hh"
 #include "debug/IEW.hh"
+#include "debug/O3PipeView.hh"
 #include "params/DerivO3CPU.hh"
 
 using namespace std;
@@ -1611,7 +1612,9 @@ DefaultIEW<Impl>::updateExeInstStats(DynInstPtr &inst)
     iewExecutedInsts++;
 
 #if TRACING_ON
-    inst->completeTick = curTick() - inst->fetchTick;
+    if (DTRACE(O3PipeView)) {
+        inst->completeTick = curTick() - inst->fetchTick;
+    }
 #endif
 
     //

src/cpu/o3/lsq_unit_impl.hh

@@ -51,6 +51,7 @@
 #include "debug/Activity.hh"
 #include "debug/IEW.hh"
 #include "debug/LSQUnit.hh"
+#include "debug/O3PipeView.hh"
 #include "mem/packet.hh"
 #include "mem/request.hh"
 
@@ -1137,6 +1138,13 @@ LSQUnit<Impl>::completeStore(int store_idx)
             "idx:%i\n",
             storeQueue[store_idx].inst->seqNum, store_idx, storeHead);
 
+#if TRACING_ON
+    if (DTRACE(O3PipeView)) {
+        storeQueue[store_idx].inst->storeTick =
+            curTick() - storeQueue[store_idx].inst->fetchTick;
+    }
+#endif
+
     if (isStalled() &&
         storeQueue[store_idx].inst->seqNum == stallingStoreIsn) {
         DPRINTF(LSQUnit, "Unstalling, stalling store [sn:%lli] "

src/cpu/o3/rename_impl.hh

@@ -49,6 +49,7 @@
 #include "cpu/o3/rename.hh"
 #include "debug/Activity.hh"
 #include "debug/Rename.hh"
+#include "debug/O3PipeView.hh"
 #include "params/DerivO3CPU.hh"
 
 using namespace std;
@@ -658,9 +659,6 @@ DefaultRename<Impl>::renameInsts(ThreadID tid)
 
         ++renamed_insts;
 
-#if TRACING_ON
-        inst->renameTick = curTick() - inst->fetchTick;
-#endif
 
         // Put instruction in rename queue.
         toIEW->insts[toIEWIndex] = inst;
@@ -736,6 +734,11 @@ DefaultRename<Impl>::sortInsts()
     for (int i = 0; i < insts_from_decode; ++i) {
         DynInstPtr inst = fromDecode->insts[i];
         insts[inst->threadNumber].push_back(inst);
+#if TRACING_ON
+        if (DTRACE(O3PipeView)) {
+            inst->renameTick = curTick() - inst->fetchTick;
+        }
+#endif
     }
 }