Merge zizzer:/bk/sparcfs

into  zed.eecs.umich.edu:/z/hsul/work/m5/newmem

--HG--
extra : convert_revision : 92a865a90a7c3e251ed1443f79640f761b359c1d

6 files changed, 90 insertions, 70 deletions

diff --git a/src/cpu/o3/alpha/dyn_inst.hh b/src/cpu/o3/alpha/dyn_inst.hh
index 31df8ff78..49cc5a201 100644
--- a/src/cpu/o3/alpha/dyn_inst.hh
+++ b/src/cpu/o3/alpha/dyn_inst.hh
@@ -163,27 +163,28 @@ class AlphaDynInst : public BaseDynInst<Impl>
     // storage (which is pretty hard to imagine they would have reason
     // to do).
 
-    uint64_t readIntReg(const StaticInst *si, int idx)
+    uint64_t readIntRegOperand(const StaticInst *si, int idx)
     {
         return this->cpu->readIntReg(_srcRegIdx[idx]);
     }
 
-    FloatReg readFloatReg(const StaticInst *si, int idx, int width)
+    FloatReg readFloatRegOperand(const StaticInst *si, int idx, int width)
     {
         return this->cpu->readFloatReg(_srcRegIdx[idx], width);
     }
 
-    FloatReg readFloatReg(const StaticInst *si, int idx)
+    FloatReg readFloatRegOperand(const StaticInst *si, int idx)
     {
         return this->cpu->readFloatReg(_srcRegIdx[idx]);
     }
 
-    FloatRegBits readFloatRegBits(const StaticInst *si, int idx, int width)
+    FloatRegBits readFloatRegOperandBits(const StaticInst *si, int idx,
+                                         int width)
     {
         return this->cpu->readFloatRegBits(_srcRegIdx[idx], width);
     }
 
-    FloatRegBits readFloatRegBits(const StaticInst *si, int idx)
+    FloatRegBits readFloatRegOperandBits(const StaticInst *si, int idx)
     {
         return this->cpu->readFloatRegBits(_srcRegIdx[idx]);
     }
@@ -191,35 +192,37 @@ class AlphaDynInst : public BaseDynInst<Impl>
     /** @todo: Make results into arrays so they can handle multiple dest
      *  registers.
      */
-    void setIntReg(const StaticInst *si, int idx, uint64_t val)
+    void setIntRegOperand(const StaticInst *si, int idx, uint64_t val)
     {
         this->cpu->setIntReg(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setIntReg(si, idx, val);
+        BaseDynInst<Impl>::setIntRegOperand(si, idx, val);
     }
 
-    void setFloatReg(const StaticInst *si, int idx, FloatReg val, int width)
+    void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val,
+                            int width)
     {
         this->cpu->setFloatReg(_destRegIdx[idx], val, width);
-        BaseDynInst<Impl>::setFloatReg(si, idx, val, width);
+        BaseDynInst<Impl>::setFloatRegOperand(si, idx, val, width);
     }
 
-    void setFloatReg(const StaticInst *si, int idx, FloatReg val)
+    void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val)
     {
         this->cpu->setFloatReg(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setFloatReg(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperand(si, idx, val);
     }
 
-    void setFloatRegBits(const StaticInst *si, int idx,
-            FloatRegBits val, int width)
+    void setFloatRegOperandBits(const StaticInst *si, int idx,
+                                FloatRegBits val, int width)
     {
         this->cpu->setFloatRegBits(_destRegIdx[idx], val, width);
-        BaseDynInst<Impl>::setFloatRegBits(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperandBits(si, idx, val);
     }
 
-    void setFloatRegBits(const StaticInst *si, int idx, FloatRegBits val)
+    void setFloatRegOperandBits(const StaticInst *si, int idx,
+                                FloatRegBits val)
     {
         this->cpu->setFloatRegBits(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setFloatRegBits(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperandBits(si, idx, val);
     }
 
     /** Returns the physical register index of the i'th destination
diff --git a/src/cpu/o3/commit_impl.hh b/src/cpu/o3/commit_impl.hh
index e72679710..f400d757b 100644
--- a/src/cpu/o3/commit_impl.hh
+++ b/src/cpu/o3/commit_impl.hh
@@ -748,6 +748,7 @@ DefaultCommit<Impl>::commit()
                 }
             } else {
                 bdelay_done_seq_num = squashed_inst;
+                squash_bdelay_slot = true;
             }
 #endif
 
@@ -1196,16 +1197,16 @@ DefaultCommit<Impl>::getInsts()
              rename_idx < fromRename->size;
              rename_idx++) {
             DynInstPtr inst = fromRename->insts[rename_idx];
-            int tid = inst->threadNumber;
 
             if (!inst->isSquashed()) {
                 DPRINTF(Commit, "Inserting PC %#x [sn:%i] [tid:%i] into ",
-                        "skidBuffer.\n", inst->readPC(), inst->seqNum, tid);
+                        "skidBuffer.\n", inst->readPC(), inst->seqNum,
+                        inst->threadNumber);
                 skidBuffer.push(inst);
             } else {
                 DPRINTF(Commit, "Instruction PC %#x [sn:%i] [tid:%i] was "
                         "squashed, skipping.\n",
-                        inst->readPC(), inst->seqNum, tid);
+                        inst->readPC(), inst->seqNum, inst->threadNumber);
             }
         }
     }
diff --git a/src/cpu/o3/fetch_impl.hh b/src/cpu/o3/fetch_impl.hh
index 63d22b293..622259495 100644
--- a/src/cpu/o3/fetch_impl.hh
+++ b/src/cpu/o3/fetch_impl.hh
@@ -151,36 +151,6 @@ DefaultFetch<Impl>::DefaultFetch(Params *params)
               " RoundRobin,LSQcount,IQcount}\n");
     }
 
-    // Size of cache block.
-    cacheBlkSize = 64;
-
-    // Create mask to get rid of offset bits.
-    cacheBlkMask = (cacheBlkSize - 1);
-
-    for (int tid=0; tid < numThreads; tid++) {
-
-        fetchStatus[tid] = Running;
-
-        priorityList.push_back(tid);
-
-        memReq[tid] = NULL;
-
-        // Create space to store a cache line.
-        cacheData[tid] = new uint8_t[cacheBlkSize];
-        cacheDataPC[tid] = 0;
-        cacheDataValid[tid] = false;
-
-        delaySlotInfo[tid].branchSeqNum = -1;
-        delaySlotInfo[tid].numInsts = 0;
-        delaySlotInfo[tid].targetAddr = 0;
-        delaySlotInfo[tid].targetReady = false;
-
-        stalls[tid].decode = false;
-        stalls[tid].rename = false;
-        stalls[tid].iew = false;
-        stalls[tid].commit = false;
-    }
-
     // Get the size of an instruction.
     instSize = sizeof(TheISA::MachInst);
 }
@@ -353,6 +323,36 @@ DefaultFetch<Impl>::initStage()
         nextNPC[tid] = cpu->readNextNPC(tid);
 #endif
     }
+
+    // Size of cache block.
+    cacheBlkSize = icachePort->peerBlockSize();
+
+    // Create mask to get rid of offset bits.
+    cacheBlkMask = (cacheBlkSize - 1);
+
+    for (int tid=0; tid < numThreads; tid++) {
+
+        fetchStatus[tid] = Running;
+
+        priorityList.push_back(tid);
+
+        memReq[tid] = NULL;
+
+        // Create space to store a cache line.
+        cacheData[tid] = new uint8_t[cacheBlkSize];
+        cacheDataPC[tid] = 0;
+        cacheDataValid[tid] = false;
+
+        delaySlotInfo[tid].branchSeqNum = -1;
+        delaySlotInfo[tid].numInsts = 0;
+        delaySlotInfo[tid].targetAddr = 0;
+        delaySlotInfo[tid].targetReady = false;
+
+        stalls[tid].decode = false;
+        stalls[tid].rename = false;
+        stalls[tid].iew = false;
+        stalls[tid].commit = false;
+    }
 }
 
 template<class Impl>
@@ -1139,6 +1139,8 @@ DefaultFetch<Impl>::fetch(bool &status_change)
             ext_inst = TheISA::makeExtMI(inst, fetch_PC);
 #elif THE_ISA == SPARC_ISA
             ext_inst = TheISA::makeExtMI(inst, cpu->thread[tid]->getTC());
+#elif THE_ISA == MIPS_ISA
+            ext_inst = TheISA::makeExtMI(inst, cpu->thread[tid]->getTC());
 #endif
 
             // Create a new DynInst from the instruction fetched.
diff --git a/src/cpu/o3/iew_impl.hh b/src/cpu/o3/iew_impl.hh
index ba5260fe2..76047b295 100644
--- a/src/cpu/o3/iew_impl.hh
+++ b/src/cpu/o3/iew_impl.hh
@@ -514,6 +514,7 @@ DefaultIEW<Impl>::squashDueToMemOrder(DynInstPtr &inst, unsigned tid)
     toCommit->squash[tid] = true;
     toCommit->squashedSeqNum[tid] = inst->seqNum;
     toCommit->nextPC[tid] = inst->readNextPC();
+    toCommit->branchMispredict[tid] = false;
 
     toCommit->includeSquashInst[tid] = false;
 
@@ -530,6 +531,7 @@ DefaultIEW<Impl>::squashDueToMemBlocked(DynInstPtr &inst, unsigned tid)
     toCommit->squash[tid] = true;
     toCommit->squashedSeqNum[tid] = inst->seqNum;
     toCommit->nextPC[tid] = inst->readPC();
+    toCommit->branchMispredict[tid] = false;
 
     // Must include the broadcasted SN in the squash.
     toCommit->includeSquashInst[tid] = true;
@@ -1291,7 +1293,8 @@ DefaultIEW<Impl>::executeInsts()
                 } else if (fault != NoFault) {
                     // If the instruction faulted, then we need to send it along to commit
                     // without the instruction completing.
-                    DPRINTF(IEW, "Store has fault! [sn:%lli]\n", inst->seqNum);
+                    DPRINTF(IEW, "Store has fault %s! [sn:%lli]\n",
+                            fault->name(), inst->seqNum);
 
                     // Send this instruction to commit, also make sure iew stage
                     // realizes there is activity.
@@ -1328,7 +1331,8 @@ DefaultIEW<Impl>::executeInsts()
         // instruction first, so the branch resolution order will be correct.
         unsigned tid = inst->threadNumber;
 
-        if (!fetchRedirect[tid]) {
+        if (!fetchRedirect[tid] ||
+            toCommit->squashedSeqNum[tid] > inst->seqNum) {
 
             if (inst->mispredicted()) {
                 fetchRedirect[tid] = true;
@@ -1350,8 +1354,6 @@ DefaultIEW<Impl>::executeInsts()
                     predictedNotTakenIncorrect++;
                 }
             } else if (ldstQueue.violation(tid)) {
-                fetchRedirect[tid] = true;
-
                 // If there was an ordering violation, then get the
                 // DynInst that caused the violation.  Note that this
                 // clears the violation signal.
@@ -1362,6 +1364,14 @@ DefaultIEW<Impl>::executeInsts()
                         "%#x, inst PC: %#x.  Addr is: %#x.\n",
                         violator->readPC(), inst->readPC(), inst->physEffAddr);
 
+                // Ensure the violating instruction is older than
+                // current squash
+                if (fetchRedirect[tid] &&
+                    violator->seqNum >= toCommit->squashedSeqNum[tid])
+                    continue;
+
+                fetchRedirect[tid] = true;
+
                 // Tell the instruction queue that a violation has occured.
                 instQueue.violation(inst, violator);
 
diff --git a/src/cpu/o3/lsq_unit_impl.hh b/src/cpu/o3/lsq_unit_impl.hh
index 4facea9f9..3b84d3411 100644
--- a/src/cpu/o3/lsq_unit_impl.hh
+++ b/src/cpu/o3/lsq_unit_impl.hh
@@ -418,7 +418,8 @@ LSQUnit<Impl>::executeLoad(DynInstPtr &inst)
         // realizes there is activity.
         // Mark it as executed unless it is an uncached load that
         // needs to hit the head of commit.
-        if (!(inst->req->isUncacheable()) || inst->isAtCommit()) {
+        if (!(inst->req && inst->req->isUncacheable()) ||
+            inst->isAtCommit()) {
             inst->setExecuted();
         }
         iewStage->instToCommit(inst);
diff --git a/src/cpu/o3/mips/dyn_inst.hh b/src/cpu/o3/mips/dyn_inst.hh
index 9e95b2bfb..833371e00 100755
--- a/src/cpu/o3/mips/dyn_inst.hh
+++ b/src/cpu/o3/mips/dyn_inst.hh
@@ -156,27 +156,28 @@ class MipsDynInst : public BaseDynInst<Impl>
     // storage (which is pretty hard to imagine they would have reason
     // to do).
 
-    uint64_t readIntReg(const StaticInst *si, int idx)
+    uint64_t readIntRegOperand(const StaticInst *si, int idx)
     {
         return this->cpu->readIntReg(_srcRegIdx[idx]);
     }
 
-    FloatReg readFloatReg(const StaticInst *si, int idx, int width)
+    FloatReg readFloatRegOperand(const StaticInst *si, int idx, int width)
     {
         return this->cpu->readFloatReg(_srcRegIdx[idx], width);
     }
 
-    FloatReg readFloatReg(const StaticInst *si, int idx)
+    FloatReg readFloatRegOperand(const StaticInst *si, int idx)
     {
         return this->cpu->readFloatReg(_srcRegIdx[idx]);
     }
 
-    FloatRegBits readFloatRegBits(const StaticInst *si, int idx, int width)
+    FloatRegBits readFloatRegOperandBits(const StaticInst *si, int idx,
+                                         int width)
     {
         return this->cpu->readFloatRegBits(_srcRegIdx[idx], width);
     }
 
-    FloatRegBits readFloatRegBits(const StaticInst *si, int idx)
+    FloatRegBits readFloatRegOperandBits(const StaticInst *si, int idx)
     {
         return this->cpu->readFloatRegBits(_srcRegIdx[idx]);
     }
@@ -184,35 +185,37 @@ class MipsDynInst : public BaseDynInst<Impl>
     /** @todo: Make results into arrays so they can handle multiple dest
      *  registers.
      */
-    void setIntReg(const StaticInst *si, int idx, uint64_t val)
+    void setIntRegOperand(const StaticInst *si, int idx, uint64_t val)
     {
         this->cpu->setIntReg(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setIntReg(si, idx, val);
+        BaseDynInst<Impl>::setIntRegOperand(si, idx, val);
     }
 
-    void setFloatReg(const StaticInst *si, int idx, FloatReg val, int width)
+    void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val,
+                            int width)
     {
         this->cpu->setFloatReg(_destRegIdx[idx], val, width);
-        BaseDynInst<Impl>::setFloatReg(si, idx, val, width);
+        BaseDynInst<Impl>::setFloatRegOperand(si, idx, val, width);
     }
 
-    void setFloatReg(const StaticInst *si, int idx, FloatReg val)
+    void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val)
     {
         this->cpu->setFloatReg(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setFloatReg(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperand(si, idx, val);
     }
 
-    void setFloatRegBits(const StaticInst *si, int idx,
-            FloatRegBits val, int width)
+    void setFloatRegOperandBits(const StaticInst *si, int idx,
+                                FloatRegBits val, int width)
     {
         this->cpu->setFloatRegBits(_destRegIdx[idx], val, width);
-        BaseDynInst<Impl>::setFloatRegBits(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperandBits(si, idx, val);
     }
 
-    void setFloatRegBits(const StaticInst *si, int idx, FloatRegBits val)
+    void setFloatRegOperandBits(const StaticInst *si, int idx,
+                                FloatRegBits val)
     {
         this->cpu->setFloatRegBits(_destRegIdx[idx], val);
-        BaseDynInst<Impl>::setFloatRegBits(si, idx, val);
+        BaseDynInst<Impl>::setFloatRegOperandBits(si, idx, val);
     }
 
     /** Returns the physical register index of the i'th destination