Merge ktlim@zizzer:/bk/m5

into  zamp.eecs.umich.edu:/z/ktlim2/clean/m5-o3

--HG--
extra : convert_revision : 0a6140f3f5e7c454981c5aa2b221224f076e50eb

31 files changed, 553 insertions, 784 deletions

diff --git a/arch/alpha/ev5.cc b/arch/alpha/ev5.cc
index 019e83dd4..ad3a9ec4c 100644
--- a/arch/alpha/ev5.cc
+++ b/arch/alpha/ev5.cc
@@ -146,7 +146,7 @@ CPUExecContext::hwrei()
     setNextPC(readMiscReg(AlphaISA::IPR_EXC_ADDR));
 
     if (!misspeculating()) {
-        cpu->kernelStats->hwrei();
+        kernelStats->hwrei();
 
         cpu->checkInterrupts = true;
     }
@@ -336,7 +336,8 @@ AlphaISA::MiscRegFile::setIpr(int idx, uint64_t val, ExecContext *xc)
         // write entire quad w/ no side-effect
         old = ipr[idx];
         ipr[idx] = val;
-        xc->getCpuPtr()->kernelStats->context(old, val, xc);
+        if (xc->getKernelStats())
+            xc->getKernelStats()->context(old, val, xc);
         break;
 
       case AlphaISA::IPR_DTB_PTE:
@@ -363,14 +364,19 @@ AlphaISA::MiscRegFile::setIpr(int idx, uint64_t val, ExecContext *xc)
 
         // only write least significant five bits - interrupt level
         ipr[idx] = val & 0x1f;
-        xc->getCpuPtr()->kernelStats->swpipl(ipr[idx]);
+        if (xc->getKernelStats())
+            xc->getKernelStats()->swpipl(ipr[idx]);
         break;
 
       case AlphaISA::IPR_DTB_CM:
-        if (val & 0x18)
-            xc->getCpuPtr()->kernelStats->mode(Kernel::user, xc);
-        else
-            xc->getCpuPtr()->kernelStats->mode(Kernel::kernel, xc);
+        if (val & 0x18) {
+            if (xc->getKernelStats())
+                xc->getKernelStats()->mode(Kernel::user, xc);
+            else {
+                if (xc->getKernelStats())
+                    xc->getKernelStats()->mode(Kernel::kernel, xc);
+            }
+        }
 
       case AlphaISA::IPR_ICM:
         // only write two mode bits - processor mode
@@ -556,7 +562,7 @@ AlphaISA::MiscRegFile::copyIprs(ExecContext *xc)
 bool
 CPUExecContext::simPalCheck(int palFunc)
 {
-    cpu->kernelStats->callpal(palFunc, proxy);
+    kernelStats->callpal(palFunc, proxy);
 
     switch (palFunc) {
       case PAL::halt:
diff --git a/cpu/base.cc b/cpu/base.cc
index 74b679d5d..de03b9eab 100644
--- a/cpu/base.cc
+++ b/cpu/base.cc
@@ -45,10 +45,6 @@
 
 #include "base/trace.hh"
 
-#if FULL_SYSTEM
-#include "kern/kernel_stats.hh"
-#endif
-
 using namespace std;
 
 vector<BaseCPU *> BaseCPU::cpuList;
@@ -153,8 +149,6 @@ BaseCPU::BaseCPU(Params *p)
     profileEvent = NULL;
     if (params->profile)
         profileEvent = new ProfileEvent(this, params->profile);
-
-    kernelStats = new Kernel::Statistics(system);
 #endif
 
 }
@@ -175,10 +169,6 @@ BaseCPU::enableFunctionTrace()
 
 BaseCPU::~BaseCPU()
 {
-#if FULL_SYSTEM
-    if (kernelStats)
-        delete kernelStats;
-#endif
 }
 
 void
@@ -219,8 +209,6 @@ BaseCPU::regStats()
         execContexts[0]->regStats(name());
 
 #if FULL_SYSTEM
-    if (kernelStats)
-        kernelStats->regStats(name() + ".kern");
 #endif
 }
 
@@ -348,12 +336,6 @@ BaseCPU::serialize(std::ostream &os)
 {
     SERIALIZE_ARRAY(interrupts, TheISA::NumInterruptLevels);
     SERIALIZE_SCALAR(intstatus);
-
-#if FULL_SYSTEM
-    if (kernelStats)
-        kernelStats->serialize(os);
-#endif
-
 }
 
 void
@@ -361,11 +343,6 @@ BaseCPU::unserialize(Checkpoint *cp, const std::string &section)
 {
     UNSERIALIZE_ARRAY(interrupts, TheISA::NumInterruptLevels);
     UNSERIALIZE_SCALAR(intstatus);
-
-#if FULL_SYSTEM
-    if (kernelStats)
-        kernelStats->unserialize(cp, section);
-#endif
 }
 
 #endif // FULL_SYSTEM
diff --git a/cpu/base.hh b/cpu/base.hh
index 20166d7ee..dd776859d 100644
--- a/cpu/base.hh
+++ b/cpu/base.hh
@@ -38,14 +38,10 @@
 #include "sim/sim_object.hh"
 #include "arch/isa_traits.hh"
 
-#if FULL_SYSTEM
-class System;
-namespace Kernel { class Statistics; }
-#endif
-
 class BranchPred;
 class CheckerCPU;
 class ExecContext;
+class System;
 
 class BaseCPU : public SimObject
 {
@@ -237,10 +233,6 @@ class BaseCPU : public SimObject
   public:
     // Number of CPU cycles simulated
     Stats::Scalar<> numCycles;
-
-#if FULL_SYSTEM
-    Kernel::Statistics *kernelStats;
-#endif
 };
 
 #endif // __CPU_BASE_HH__
diff --git a/cpu/base_dyn_inst.hh b/cpu/base_dyn_inst.hh
index 9403faec3..388ea4a8d 100644
--- a/cpu/base_dyn_inst.hh
+++ b/cpu/base_dyn_inst.hh
@@ -450,13 +450,10 @@ class BaseDynInst : public FastAlloc, public RefCounted
         instResult.integer = val;
     }
 
-    //Push to .cc file.
     /** Records that one of the source registers is ready. */
     void markSrcRegReady();
 
-    /** Marks a specific register as ready.
-     *  @todo: Move this to .cc file.
-     */
+    /** Marks a specific register as ready. */
     void markSrcRegReady(RegIndex src_idx);
 
     /** Returns if a source register is ready. */
diff --git a/cpu/checker/cpu.cc b/cpu/checker/cpu.cc
index f76f1e063..08ab5d5c8 100644
--- a/cpu/checker/cpu.cc
+++ b/cpu/checker/cpu.cc
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002-2005 The Regents of The University of Michigan
+ * Copyright (c) 2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -26,41 +26,17 @@
  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-//#include <cmath>
-#include <cstdio>
-//#include <cstdlib>
-#include <iostream>
-#include <iomanip>
 #include <list>
-//#include <sstream>
 #include <string>
 
-//#include "base/cprintf.hh"
-//#include "base/inifile.hh"
-//#include "base/loader/symtab.hh"
-#include "base/misc.hh"
-//#include "base/pollevent.hh"
-//#include "base/range.hh"
 #include "base/refcnt.hh"
-//#include "base/stats/events.hh"
 #include "cpu/base.hh"
 #include "cpu/base_dyn_inst.hh"
 #include "cpu/checker/cpu.hh"
 #include "cpu/cpu_exec_context.hh"
 #include "cpu/exec_context.hh"
-//#include "cpu/exetrace.hh"
-//#include "cpu/profile.hh"
-#include "cpu/sampler/sampler.hh"
-//#include "cpu/smt.hh"
 #include "cpu/static_inst.hh"
-//#include "kern/kernel_stats.hh"
-#include "mem/base_mem.hh"
-#include "mem/mem_interface.hh"
 #include "sim/byteswap.hh"
-#include "sim/builder.hh"
-//#include "sim/debug.hh"
-//#include "sim/host.hh"
-//#include "sim/sim_events.hh"
 #include "sim/sim_object.hh"
 #include "sim/stats.hh"
 
@@ -72,15 +48,8 @@
 #include "cpu/ozone/simple_impl.hh"
 
 #if FULL_SYSTEM
-#include "base/remote_gdb.hh"
-#include "mem/functional/memory_control.hh"
-#include "mem/functional/physical.hh"
 #include "sim/system.hh"
-#include "arch/tlb.hh"
-#include "arch/stacktrace.hh"
 #include "arch/vtophys.hh"
-#else // !FULL_SYSTEM
-#include "mem/functional/functional.hh"
 #endif // FULL_SYSTEM
 
 using namespace std;
@@ -90,17 +59,6 @@ using namespace AlphaISA;
 void
 CheckerCPU::init()
 {
-/*
-    BaseCPU::init();
-#if FULL_SYSTEM
-    for (int i = 0; i < execContexts.size(); ++i) {
-        ExecContext *xc = execContexts[i];
-
-        // initialize CPU, including PC
-        TheISA::initCPU(xc, xc->readCpuId());
-    }
-#endif
-*/
 }
 
 CheckerCPU::CheckerCPU(Params *p)
@@ -151,6 +109,8 @@ CheckerCPU::setMemory(FunctionalMemory *mem)
         xcProxy = cpuXC->getProxy();
         execContexts.push_back(xcProxy);
         memReq->xc = xcProxy;
+        delete cpuXC->kernelStats;
+        cpuXC->kernelStats = NULL;
     }
 #endif
 }
@@ -168,6 +128,8 @@ CheckerCPU::setSystem(System *system)
         xcProxy = cpuXC->getProxy();
         execContexts.push_back(xcProxy);
         memReq->xc = xcProxy;
+        delete cpuXC->kernelStats;
+        cpuXC->kernelStats = NULL;
     }
 }
 #endif
@@ -197,82 +159,15 @@ CheckerCPU::unserialize(Checkpoint *cp, const string &section)
 Fault
 CheckerCPU::copySrcTranslate(Addr src)
 {
-    static bool no_warn = true;
-    int blk_size = 64;
-    // Only support block sizes of 64 atm.
-    assert(blk_size == 64);
-    int offset = src & (blk_size - 1);
-
-    // Make sure block doesn't span page
-    if (no_warn &&
-        (src & PageMask) != ((src + blk_size) & PageMask) &&
-        (src >> 40) != 0xfffffc) {
-        warn("Copied block source spans pages %x.", src);
-        no_warn = false;
-    }
-
-    memReq->reset(src & ~(blk_size - 1), blk_size);
-
-    // translate to physical address
-    Fault fault = cpuXC->translateDataReadReq(memReq);
-
-    if (fault == NoFault) {
-        cpuXC->copySrcAddr = src;
-        cpuXC->copySrcPhysAddr = memReq->paddr + offset;
-    } else {
-        assert(!fault->isAlignmentFault());
-
-        cpuXC->copySrcAddr = 0;
-        cpuXC->copySrcPhysAddr = 0;
-    }
-    return fault;
+    panic("Unimplemented!");
 }
 
 Fault
 CheckerCPU::copy(Addr dest)
 {
-    static bool no_warn = true;
-    int blk_size = 64;
-    // Only support block sizes of 64 atm.
-    assert(blk_size == 64);
-    uint8_t data[blk_size];
-    //assert(cpuXC->copySrcAddr);
-    int offset = dest & (blk_size - 1);
-
-    // Make sure block doesn't span page
-    if (no_warn &&
-        (dest & PageMask) != ((dest + blk_size) & PageMask) &&
-        (dest >> 40) != 0xfffffc) {
-        no_warn = false;
-        warn("Copied block destination spans pages %x. ", dest);
-    }
-
-    memReq->reset(dest & ~(blk_size -1), blk_size);
-    // translate to physical address
-    Fault fault = cpuXC->translateDataWriteReq(memReq);
-
-    if (fault == NoFault) {
-        Addr dest_addr = memReq->paddr + offset;
-        // Need to read straight from memory since we have more than 8 bytes.
-        memReq->paddr = cpuXC->copySrcPhysAddr;
-        cpuXC->mem->read(memReq, data);
-        memReq->paddr = dest_addr;
-        cpuXC->mem->write(memReq, data);
-        memReq->cmd = Copy;
-        memReq->completionEvent = NULL;
-        memReq->paddr = cpuXC->copySrcPhysAddr;
-        memReq->dest = dest_addr;
-        memReq->size = 64;
-        memReq->time = curTick;
-        memReq->flags &= ~INST_READ;
-    }
-    else
-        assert(!fault->isAlignmentFault());
-
-    return fault;
+    panic("Unimplemented!");
 }
 
-// precise architected memory state accessor macros
 template <class T>
 Fault
 CheckerCPU::read(Addr addr, T &data, unsigned flags)
@@ -280,17 +175,15 @@ CheckerCPU::read(Addr addr, T &data, unsigned flags)
     memReq->reset(addr, sizeof(T), flags);
 
     // translate to physical address
-    // Should I probe the DTB?  Or should I just take the physical address
-    // and assume correct translation?
     translateDataReadReq(memReq);
 
-    // if we have a cache, do cache access too
     memReq->cmd = Read;
     memReq->completionEvent = NULL;
     memReq->time = curTick;
     memReq->flags &= ~INST_READ;
 
     if (!(memReq->flags & UNCACHEABLE)) {
+        // Access memory to see if we have the same data
         cpuXC->read(memReq, data);
     } else {
         // Assume the data is correct if it's an uncached access
@@ -350,29 +243,34 @@ CheckerCPU::write(T data, Addr addr, unsigned flags, uint64_t *res)
     // translate to physical address
     cpuXC->translateDataWriteReq(memReq);
 
-    if ((!(unverifiedReq->flags & LOCKED) ||
-        ((unverifiedReq->flags & LOCKED) &&
-         unverifiedReq->result == 1)) &&
-        !(unverifiedReq->flags & UNCACHEABLE)) {
-        // do functional access
-//        cpuXC->read(memReq, data);
-
-        memReq->cmd = Write;
-//    memcpy(memReq->data,(uint8_t *)&data,memReq->size);
-        T inst_data;
-        memcpy(&inst_data, unverifiedReq->data, sizeof(T));
+    // Can compare the write data and result only if it's cacheable,
+    // not a store conditional, or is a store conditional that
+    // succeeded.
+    // @todo: Verify that actual memory matches up with these values.
+    // Right now it only verifies that the instruction data is the
+    // same as what was in the request that got sent to memory; there
+    // is no verification that it is the same as what is in memory.
+    // This is because the LSQ would have to be snooped in the CPU to
+    // verify this data.
+    if (unverifiedReq &&
+        !(unverifiedReq->flags & UNCACHEABLE) &&
+        (!(unverifiedReq->flags & LOCKED) ||
+         ((unverifiedReq->flags & LOCKED) &&
+          unverifiedReq->result == 1))) {
+#if 0
+        memReq->cmd = Read;
         memReq->completionEvent = NULL;
         memReq->time = curTick;
         memReq->flags &= ~INST_READ;
+        cpuXC->read(memReq, inst_data);
+#endif
+        T inst_data;
+        memcpy(&inst_data, unverifiedReq->data, sizeof(T));
 
-        // Hard to verify this as the data writes back after the
-        // instruction commits.  May only be able to check that the
-        // value produced from execute() matches the value produced
-        // from the instruction's first execution.
         if (data != inst_data) {
-            warn("Store value does not match value in memory! "
+            warn("%lli: Store value does not match value in memory! "
                  "Instruction: %#x, memory: %#x",
-                 inst_data, data);
+                 curTick, inst_data, data);
             handleError();
         }
     }
@@ -436,19 +334,6 @@ CheckerCPU::dbg_vtophys(Addr addr)
 }
 #endif // FULL_SYSTEM
 
-#if FULL_SYSTEM
-void
-CheckerCPU::post_interrupt(int int_num, int index)
-{
-    BaseCPU::post_interrupt(int_num, index);
-
-    if (cpuXC->status() == ExecContext::Suspended) {
-                DPRINTF(IPI,"Suspended Processor awoke\n");
-        cpuXC->activate();
-    }
-}
-#endif // FULL_SYSTEM
-
 bool
 CheckerCPU::translateInstReq(MemReqPtr &req)
 {
@@ -466,15 +351,16 @@ CheckerCPU::translateDataReadReq(MemReqPtr &req)
     cpuXC->translateDataReadReq(req);
 
     if (req->vaddr != unverifiedReq->vaddr) {
-        warn("Request virtual addresses do not match! Inst: %#x, checker:"
-             " %#x",
-             unverifiedReq->vaddr, req->vaddr);
+        warn("%lli: Request virtual addresses do not match! Inst: %#x, "
+             "checker: %#x",
+             curTick, unverifiedReq->vaddr, req->vaddr);
+        handleError();
     }
     req->paddr = unverifiedReq->paddr;
 
     if (checkFlags(req)) {
-        warn("Request flags do not match! Inst: %#x, checker: %#x",
-             unverifiedReq->flags, req->flags);
+        warn("%lli: Request flags do not match! Inst: %#x, checker: %#x",
+             curTick, unverifiedReq->flags, req->flags);
         handleError();
     }
 }
@@ -485,15 +371,16 @@ CheckerCPU::translateDataWriteReq(MemReqPtr &req)
     cpuXC->translateDataWriteReq(req);
 
     if (req->vaddr != unverifiedReq->vaddr) {
-        warn("Request virtual addresses do not match! Inst: %#x, checker:"
-             " %#x",
-             unverifiedReq->vaddr, req->vaddr);
+        warn("%lli: Request virtual addresses do not match! Inst: %#x, "
+             "checker: %#x",
+             curTick, unverifiedReq->vaddr, req->vaddr);
+        handleError();
     }
     req->paddr = unverifiedReq->paddr;
 
     if (checkFlags(req)) {
-        warn("Request flags do not match! Inst: %#x, checker: %#x",
-             unverifiedReq->flags, req->flags);
+        warn("%lli: Request flags do not match! Inst: %#x, checker: %#x",
+             curTick, unverifiedReq->flags, req->flags);
         handleError();
     }
 }
@@ -512,13 +399,17 @@ CheckerCPU::checkFlags(MemReqPtr &req)
     }
 }
 
-/* start simulation, program loaded, processor precise state initialized */
 template <class DynInstPtr>
 void
 Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
 {
     DynInstPtr inst;
 
+    // Either check this instruction, or add it to a list of
+    // instructions waiting to be checked.  Instructions must be
+    // checked in program order, so if a store has committed yet not
+    // completed, there may be some instructions that are waiting
+    // behind it that have completed and must be checked.
     if (!instList.empty()) {
         if (youngestSN < completed_inst->seqNum) {
             DPRINTF(Checker, "Adding instruction [sn:%lli] PC:%#x to list.\n",
@@ -547,16 +438,17 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
                 inst = completed_inst;
                 youngestSN = completed_inst->seqNum;
             } else {
-//                panic("SN already seen yet the list is empty!");
                 return;
             }
         }
     }
 
+    // Try to check all instructions that are completed, ending if we
+    // run out of instructions to check or if an instruction is not
+    // yet completed.
     while (1) {
         DPRINTF(Checker, "Processing instruction [sn:%lli] PC:%#x.\n",
                 inst->seqNum, inst->readPC());
-//    verifyInst = completed_inst;
         unverifiedResult.integer = inst->readIntResult();
         unverifiedReq = inst->req;
         numCycles++;
@@ -569,15 +461,9 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
         cpuXC->setFloatRegDouble(ZeroReg, 0.0);
 #endif // TARGET_ALPHA
 
-        // Try to fetch an instruction
-
-        // set up memory request for instruction fetch
-#if FULL_SYSTEM
-#define IFETCH_FLAGS(pc)	((pc) & 1) ? PHYSICAL : 0
-#else
-#define IFETCH_FLAGS(pc)	0
-#endif
-
+        // Check if any recent PC changes match up with anything we
+        // expect to happen.  This is mostly to check if traps or
+        // PC-based events have occurred in both the checker and CPU.
         if (changedPC) {
             DPRINTF(Checker, "Changed PC recently to %#x\n",
                     cpuXC->readPC());
@@ -585,9 +471,9 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
                 if (newPC == cpuXC->readPC()) {
                     DPRINTF(Checker, "Changed PC matches expected PC\n");
                 } else {
-                    warn("Changed PC does not match expected PC, changed: %#x, "
-                         "expected: %#x",
-                         cpuXC->readPC(), newPC);
+                    warn("%lli: Changed PC does not match expected PC, "
+                         "changed: %#x, expected: %#x",
+                         curTick, cpuXC->readPC(), newPC);
                     handleError();
                 }
                 willChangePC = false;
@@ -600,6 +486,15 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
             changedNextPC = false;
         }
 
+        // Try to fetch the instruction
+
+#if FULL_SYSTEM
+#define IFETCH_FLAGS(pc)	((pc) & 1) ? PHYSICAL : 0
+#else
+#define IFETCH_FLAGS(pc)	0
+#endif
+
+        // set up memory request for instruction fetch
         memReq->cmd = Read;
         memReq->reset(cpuXC->readPC() & ~3, sizeof(uint32_t),
                       IFETCH_FLAGS(cpuXC->readPC()));
@@ -608,8 +503,13 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
 
         if (!succeeded) {
             if (inst->getFault() == NoFault) {
-                warn("Instruction PC %#x was not found in the ITB!",
-                     cpuXC->readPC());
+                // In this case the instruction was not a dummy
+                // instruction carrying an ITB fault.  In the single
+                // threaded case the ITB should still be able to
+                // translate this instruction; in the SMT case it's
+                // possible that its ITB entry was kicked out.
+                warn("%lli: Instruction PC %#x was not found in the ITB!",
+                     curTick, cpuXC->readPC());
                 handleError();
 
                 // go to the next instruction
@@ -618,20 +518,18 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
 
                 return;
             } else {
+                // The instruction is carrying an ITB fault.  Handle
+                // the fault and see if our results match the CPU on
+                // the next tick().
                 fault = inst->getFault();
             }
         }
 
         if (fault == NoFault) {
-//        fault = cpuXC->mem->read(memReq, machInst);
             cpuXC->mem->read(memReq, machInst);
 
-            // If we've got a valid instruction (i.e., no fault on instruction
-            // fetch), then execute it.
-
-        // keep an instruction count
+            // keep an instruction count
             numInst++;
-//	numInsts++;
 
             // decode the instruction
             machInst = gtoh(machInst);
@@ -639,7 +537,8 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
             // Checks both the machine instruction and the PC.
             validateInst(inst);
 
-            curStaticInst = StaticInst::decode(makeExtMI(machInst, cpuXC->readPC()));
+            curStaticInst = StaticInst::decode(makeExtMI(machInst,
+                                                         cpuXC->readPC()));
 
 #if FULL_SYSTEM
             cpuXC->setInst(machInst);
@@ -660,10 +559,6 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
             // Checks to make sure instrution results are correct.
             validateExecution(inst);
 
-//	if (curStaticInst->isMemRef()) {
-//	    numMemRefs++;
-//	}
-
             if (curStaticInst->isLoad()) {
                 ++numLoad;
             }
@@ -693,6 +588,9 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
         }
 
 #if FULL_SYSTEM
+        // @todo: Determine if these should happen only if the
+        // instruction hasn't faulted.  In the SimpleCPU case this may
+        // not be true, but in the O3 or Ozone case this may be true.
         Addr oldpc;
         int count = 0;
         do {
@@ -707,10 +605,12 @@ Checker<DynInstPtr>::tick(DynInstPtr &completed_inst)
         }
 #endif
 
-        // Checks PC, next PC.  Optionally can check all registers. (Or just those
+        // @todo:  Optionally can check all registers. (Or just those
         // that have been modified).
         validateState();
 
+        // Continue verifying instructions if there's another completed
+        // instruction waiting to be verified.
         if (instList.empty()) {
             break;
         } else if (instList.front()->isCompleted()) {
@@ -726,7 +626,6 @@ template <class DynInstPtr>
 void
 Checker<DynInstPtr>::switchOut(Sampler *s)
 {
-    sampler = s;
     instList.clear();
 }
 
@@ -734,15 +633,6 @@ template <class DynInstPtr>
 void
 Checker<DynInstPtr>::takeOverFrom(BaseCPU *oldCPU)
 {
-//    BaseCPU::takeOverFrom(oldCPU);
-
-    // if any of this CPU's ExecContexts are active, mark the CPU as
-    // running and schedule its tick event.
-/*
-    for (int i = 0; i < execContexts.size(); ++i) {
-        ExecContext *xc = execContexts[i];
-    }
-*/
 }
 
 template <class DynInstPtr>
@@ -750,20 +640,22 @@ void
 Checker<DynInstPtr>::validateInst(DynInstPtr &inst)
 {
     if (inst->readPC() != cpuXC->readPC()) {
-        warn("PCs do not match! Inst: %#x, checker: %#x",
-             inst->readPC(), cpuXC->readPC());
+        warn("%lli: PCs do not match! Inst: %#x, checker: %#x",
+             curTick, inst->readPC(), cpuXC->readPC());
         if (changedPC) {
-            warn("Changed PCs recently, may not be an error");
+            warn("%lli: Changed PCs recently, may not be an error",
+                 curTick);
         } else {
             handleError();
         }
     }
 
-    if (static_cast<MachInst>(inst->staticInst->machInst) !=
-        machInst) {
-        warn("Binary instructions do not match! Inst: %#x, checker: %#x",
-             static_cast<MachInst>(inst->staticInst->machInst),
-             machInst);
+    MachInst mi = static_cast<MachInst>(inst->staticInst->machInst);
+
+    if (mi != machInst) {
+        warn("%lli: Binary instructions do not match! Inst: %#x, "
+             "checker: %#x",
+             curTick, mi, machInst);
         handleError();
     }
 }
@@ -773,10 +665,11 @@ void
 Checker<DynInstPtr>::validateExecution(DynInstPtr &inst)
 {
     if (inst->numDestRegs()) {
+        // @todo: Support more destination registers.
         if (inst->isUnverifiable()) {
-            // @todo: Support more destination registers.
-            // Grab the result from the instruction and write it to the
-            // register.
+            // Unverifiable instructions assume they were executed
+            // properly by the CPU. Grab the result from the
+            // instruction and write it to the register.
             RegIndex idx = inst->destRegIdx(0);
             if (idx < TheISA::FP_Base_DepTag) {
                 cpuXC->setIntReg(idx, inst->readIntResult());
@@ -786,16 +679,17 @@ Checker<DynInstPtr>::validateExecution(DynInstPtr &inst)
                 cpuXC->setMiscReg(idx, inst->readIntResult());
             }
         } else if (result.integer != inst->readIntResult()) {
-            warn("Instruction results do not match! (May not be integer results) "
-                 "Inst: %#x, checker: %#x",
-                 inst->readIntResult(), result.integer);
+            warn("%lli: Instruction results do not match! (Results may not "
+                 "actually be integers) Inst: %#x, checker: %#x",
+                 curTick, inst->readIntResult(), result.integer);
             handleError();
         }
     }
 
     if (inst->readNextPC() != cpuXC->readNextPC()) {
-        warn("Instruction next PCs do not match! Inst: %#x, checker: %#x",
-             inst->readNextPC(), cpuXC->readNextPC());
+        warn("%lli: Instruction next PCs do not match! Inst: %#x, "
+             "checker: %#x",
+             curTick, inst->readNextPC(), cpuXC->readNextPC());
         handleError();
     }
 
@@ -810,9 +704,10 @@ Checker<DynInstPtr>::validateExecution(DynInstPtr &inst)
 
         if (inst->xcBase()->readMiscReg(misc_reg_idx) !=
             cpuXC->readMiscReg(misc_reg_idx)) {
-            warn("Misc reg idx %i (side effect) does not match! Inst: %#x, "
-                 "checker: %#x",
-                 misc_reg_idx, inst->xcBase()->readMiscReg(misc_reg_idx),
+            warn("%lli: Misc reg idx %i (side effect) does not match! "
+                 "Inst: %#x, checker: %#x",
+                 curTick, misc_reg_idx,
+                 inst->xcBase()->readMiscReg(misc_reg_idx),
                  cpuXC->readMiscReg(misc_reg_idx));
             handleError();
         }
diff --git a/cpu/checker/cpu.hh b/cpu/checker/cpu.hh
index 678e888df..37fe59d95 100644
--- a/cpu/checker/cpu.hh
+++ b/cpu/checker/cpu.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2002-2005 The Regents of The University of Michigan
+ * Copyright (c) 2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -39,7 +39,6 @@
 #include "cpu/base_dyn_inst.hh"
 #include "cpu/cpu_exec_context.hh"
 #include "cpu/pc_event.hh"
-#include "cpu/sampler/sampler.hh"
 #include "cpu/static_inst.hh"
 #include "sim/eventq.hh"
 
@@ -63,6 +62,7 @@ class BaseDynInst;
 class ExecContext;
 class MemInterface;
 class Checkpoint;
+class Sampler;
 
 class CheckerCPU : public BaseCPU
 {
@@ -86,8 +86,6 @@ class CheckerCPU : public BaseCPU
     };
 
   public:
-    void post_interrupt(int int_num, int index);
-
     CheckerCPU(Params *p);
     virtual ~CheckerCPU();
 
@@ -111,8 +109,6 @@ class CheckerCPU : public BaseCPU
 
 #if FULL_SYSTEM
     Addr dbg_vtophys(Addr addr);
-
-    bool interval_stats;
 #endif
 
     union Result {
@@ -129,11 +125,6 @@ class CheckerCPU : public BaseCPU
     // Refcounted pointer to the one memory request.
     MemReqPtr memReq;
 
-    // Pointer to the sampler that is telling us to switchover.
-    // Used to signal the completion of the pipe drain and schedule
-    // the next switchover
-    Sampler *sampler;
-
     StaticInstPtr curStaticInst;
 
     // number of simulated instructions
@@ -284,6 +275,7 @@ class CheckerCPU : public BaseCPU
     bool simPalCheck(int palFunc) { return cpuXC->simPalCheck(palFunc); }
 #else
     // Assume that the normal CPU's call to syscall was successful.
+    // The checker's state would have already been updated by the syscall.
     void syscall() { }
 #endif
 
@@ -307,8 +299,6 @@ class CheckerCPU : public BaseCPU
     bool exitOnError;
 
     InstSeqNum youngestSN;
-//    std::map<Addr, uint64_t> storeBuff;
-//    typedef std::map<Addr, uint64_t>::iterator map_it;
 };
 
 template <class DynInstPtr>
diff --git a/cpu/checker/exec_context.hh b/cpu/checker/exec_context.hh
index 4843d1cf0..38784867d 100644
--- a/cpu/checker/exec_context.hh
+++ b/cpu/checker/exec_context.hh
@@ -1,3 +1,31 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
 #ifndef __CPU_CHECKER_EXEC_CONTEXT_HH__
 #define __CPU_CHECKER_EXEC_CONTEXT_HH__
 
@@ -6,6 +34,9 @@
 #include "cpu/exec_context.hh"
 
 class EndQuiesceEvent;
+namespace Kernel {
+    class Statistics;
+};
 
 template <class XC>
 class CheckerExecContext : public ExecContext
@@ -13,7 +44,8 @@ class CheckerExecContext : public ExecContext
   public:
     CheckerExecContext(XC *actual_xc,
                        CheckerCPU *checker_cpu)
-        : actualXC(actual_xc), checkerXC(checker_cpu->cpuXC), checkerCPU(checker_cpu)
+        : actualXC(actual_xc), checkerXC(checker_cpu->cpuXC),
+          checkerCPU(checker_cpu)
     { }
 
   private:
@@ -43,6 +75,8 @@ class CheckerExecContext : public ExecContext
     AlphaITB *getITBPtr() { return actualXC->getITBPtr(); }
 
     AlphaDTB *getDTBPtr() { return actualXC->getDTBPtr(); }
+
+    Kernel::Statistics *getKernelStats() { return actualXC->getKernelStats(); }
 #else
     Process *getProcessPtr() { return actualXC->getProcessPtr(); }
 #endif
@@ -50,8 +84,10 @@ class CheckerExecContext : public ExecContext
     Status status() const { return actualXC->status(); }
 
     void setStatus(Status new_status)
-    { actualXC->setStatus(new_status);
-    checkerXC->setStatus(new_status); }
+    {
+        actualXC->setStatus(new_status);
+        checkerXC->setStatus(new_status);
+    }
 
     /// Set the status to Active.  Optional delay indicates number of
     /// cycles to wait before beginning execution.
@@ -216,8 +252,6 @@ class CheckerExecContext : public ExecContext
         actualXC->setSyscallReturn(return_value);
     }
 
-//    void syscall() { actualXC->syscall(); }
-
     Counter readFuncExeInst() { return actualXC->readFuncExeInst(); }
 #endif
 };
diff --git a/cpu/cpu_exec_context.cc b/cpu/cpu_exec_context.cc
index 24de6d450..78ce058e8 100644
--- a/cpu/cpu_exec_context.cc
+++ b/cpu/cpu_exec_context.cc
@@ -188,7 +188,8 @@ CPUExecContext::serialize(ostream &os)
     if (quiesceEvent->scheduled())
         quiesceEndTick = quiesceEvent->when();
     SERIALIZE_SCALAR(quiesceEndTick);
-
+    if (kernelStats)
+        kernelStats->serialize(os);
 #endif
 }
 
@@ -207,6 +208,8 @@ CPUExecContext::unserialize(Checkpoint *cp, const std::string &section)
     UNSERIALIZE_SCALAR(quiesceEndTick);
     if (quiesceEndTick)
         quiesceEvent->schedule(quiesceEndTick);
+    if (kernelStats)
+        kernelStats->unserialize(cp, section);
 #endif
 }
 
@@ -275,6 +278,10 @@ CPUExecContext::halt()
 void
 CPUExecContext::regStats(const string &name)
 {
+#if FULL_SYSTEM
+    kernelStats = new Kernel::Statistics(system);
+    kernelStats->regStats(name + ".kern");
+#endif
 }
 
 void
diff --git a/cpu/cpu_exec_context.hh b/cpu/cpu_exec_context.hh
index cac006925..3d1428933 100644
--- a/cpu/cpu_exec_context.hh
+++ b/cpu/cpu_exec_context.hh
@@ -53,6 +53,10 @@ class FunctionProfile;
 class ProfileNode;
 class MemoryController;
 
+namespace Kernel {
+    class Statistics;
+};
+
 #else // !FULL_SYSTEM
 
 #include "sim/process.hh"
@@ -147,6 +151,9 @@ class CPUExecContext
 
     void profileSample();
 
+    Kernel::Statistics *getKernelStats() { return kernelStats; }
+
+    Kernel::Statistics *kernelStats;
 #else
     Process *process;
 
diff --git a/cpu/exec_context.hh b/cpu/exec_context.hh
index 7bd7d5682..e1f1016e5 100644
--- a/cpu/exec_context.hh
+++ b/cpu/exec_context.hh
@@ -48,6 +48,9 @@ class FunctionalMemory;
 class PhysicalMemory;
 class Process;
 class System;
+namespace Kernel {
+    class Statistics;
+};
 
 class ExecContext
 {
@@ -98,6 +101,8 @@ class ExecContext
     virtual AlphaITB *getITBPtr() = 0;
 
     virtual AlphaDTB * getDTBPtr() = 0;
+
+    virtual Kernel::Statistics *getKernelStats() = 0;
 #else
     virtual Process *getProcessPtr() = 0;
 #endif
@@ -243,6 +248,8 @@ class ProxyExecContext : public ExecContext
     AlphaITB *getITBPtr() { return actualXC->getITBPtr(); }
 
     AlphaDTB *getDTBPtr() { return actualXC->getDTBPtr(); }
+
+    Kernel::Statistics *getKernelStats() { return actualXC->getKernelStats(); }
 #else
     Process *getProcessPtr() { return actualXC->getProcessPtr(); }
 #endif
diff --git a/cpu/o3/alpha_cpu.hh b/cpu/o3/alpha_cpu.hh
index 78ad5f7d8..5c89e3462 100644
--- a/cpu/o3/alpha_cpu.hh
+++ b/cpu/o3/alpha_cpu.hh
@@ -35,6 +35,9 @@
 #include "sim/byteswap.hh"
 
 class EndQuiesceEvent;
+namespace Kernel {
+    class Statistics;
+};
 
 template <class Impl>
 class AlphaFullCPU : public FullO3CPU<Impl>
@@ -60,11 +63,6 @@ class AlphaFullCPU : public FullO3CPU<Impl>
 
         O3ThreadState<Impl> *thread;
 
-        Tick lastActivate;
-        Tick lastSuspend;
-
-        EndQuiesceEvent *quiesceEvent;
-
         virtual BaseCPU *getCpuPtr() { return cpu; }
 
         virtual void setCpuId(int id) { cpu->cpu_id = id; }
@@ -81,6 +79,9 @@ class AlphaFullCPU : public FullO3CPU<Impl>
         virtual AlphaITB *getITBPtr() { return cpu->itb; }
 
         virtual AlphaDTB * getDTBPtr() { return cpu->dtb; }
+
+        virtual Kernel::Statistics *getKernelStats()
+        { return thread->kernelStats; }
 #else
         virtual Process *getProcessPtr() { return thread->process; }
 #endif
diff --git a/cpu/o3/alpha_cpu_impl.hh b/cpu/o3/alpha_cpu_impl.hh
index 58b2b3548..91cd3d9e6 100644
--- a/cpu/o3/alpha_cpu_impl.hh
+++ b/cpu/o3/alpha_cpu_impl.hh
@@ -31,7 +31,6 @@
 #include "base/statistics.hh"
 #include "base/timebuf.hh"
 #include "cpu/checker/exec_context.hh"
-#include "cpu/quiesce_event.hh"
 #include "mem/mem_interface.hh"
 #include "sim/sim_events.hh"
 #include "sim/stats.hh"
@@ -44,6 +43,8 @@
 #if FULL_SYSTEM
 #include "arch/alpha/osfpal.hh"
 #include "arch/isa_traits.hh"
+#include "cpu/quiesce_event.hh"
+#include "kern/kernel_stats.hh"
 #endif
 
 using namespace TheISA;
@@ -101,11 +102,12 @@ AlphaFullCPU<Impl>::AlphaFullCPU(Params *params)
         alpha_xc_proxy->cpu = this;
         alpha_xc_proxy->thread = this->thread[i];
 
-        alpha_xc_proxy->quiesceEvent =
+#if FULL_SYSTEM
+        this->thread[i]->quiesceEvent =
             new EndQuiesceEvent(xc_proxy);
-        alpha_xc_proxy->lastActivate = 0;
-        alpha_xc_proxy->lastSuspend = 0;
-
+        this->thread[i]->lastActivate = 0;
+        this->thread[i]->lastSuspend = 0;
+#endif
         this->thread[i]->xcProxy = xc_proxy;
 
         this->execContexts.push_back(xc_proxy);
@@ -181,6 +183,9 @@ AlphaFullCPU<Impl>::AlphaXC::takeOverFrom(ExecContext *old_context)
     if (thread->quiesceEvent) {
         thread->quiesceEvent->xc = this;
     }
+
+    // Transfer kernel stats from one CPU to the other.
+    thread->kernelStats = old_context->getKernelStats();
 //    storeCondFailures = 0;
     cpu->lockFlag = false;
 #endif
@@ -200,7 +205,9 @@ AlphaFullCPU<Impl>::AlphaXC::activate(int delay)
     if (thread->status() == ExecContext::Active)
         return;
 
-    lastActivate = curTick;
+#if FULL_SYSTEM
+    thread->lastActivate = curTick;
+#endif
 
     if (thread->status() == ExecContext::Unallocated) {
         cpu->activateWhenReady(thread->tid);
@@ -222,8 +229,10 @@ AlphaFullCPU<Impl>::AlphaXC::suspend()
     if (thread->status() == ExecContext::Suspended)
         return;
 
-    lastActivate = curTick;
-    lastSuspend = curTick;
+#if FULL_SYSTEM
+    thread->lastActivate = curTick;
+    thread->lastSuspend = curTick;
+#endif
 /*
 #if FULL_SYSTEM
     // Don't change the status from active if there are pending interrupts
@@ -266,38 +275,55 @@ AlphaFullCPU<Impl>::AlphaXC::halt()
 template <class Impl>
 void
 AlphaFullCPU<Impl>::AlphaXC::regStats(const std::string &name)
-{}
+{
+#if FULL_SYSTEM
+    thread->kernelStats = new Kernel::Statistics(cpu->system);
+    thread->kernelStats->regStats(name + ".kern");
+#endif
+}
 
 template <class Impl>
 void
 AlphaFullCPU<Impl>::AlphaXC::serialize(std::ostream &os)
-{}
+{
+#if FULL_SYSTEM
+    if (thread->kernelStats)
+        thread->kernelStats->serialize(os);
+#endif
+
+}
 
 template <class Impl>
 void
 AlphaFullCPU<Impl>::AlphaXC::unserialize(Checkpoint *cp, const std::string &section)
-{}
+{
+#if FULL_SYSTEM
+    if (thread->kernelStats)
+        thread->kernelStats->unserialize(cp, section);
+#endif
+
+}
 
 #if FULL_SYSTEM
 template <class Impl>
 EndQuiesceEvent *
 AlphaFullCPU<Impl>::AlphaXC::getQuiesceEvent()
 {
-    return quiesceEvent;
+    return thread->quiesceEvent;
 }
 
 template <class Impl>
 Tick
 AlphaFullCPU<Impl>::AlphaXC::readLastActivate()
 {
-    return lastActivate;
+    return thread->lastActivate;
 }
 
 template <class Impl>
 Tick
 AlphaFullCPU<Impl>::AlphaXC::readLastSuspend()
 {
-    return lastSuspend;
+    return thread->lastSuspend;
 }
 
 template <class Impl>
@@ -595,7 +621,7 @@ AlphaFullCPU<Impl>::hwrei(unsigned tid)
     // Need to clear the lock flag upon returning from an interrupt.
     this->lockFlag = false;
 
-    this->kernelStats->hwrei();
+    this->thread[tid]->kernelStats->hwrei();
 
     this->checkInterrupts = true;
 
@@ -607,9 +633,9 @@ template <class Impl>
 bool
 AlphaFullCPU<Impl>::simPalCheck(int palFunc, unsigned tid)
 {
-    if (this->kernelStats)
-        this->kernelStats->callpal(palFunc,
-                                   this->execContexts[tid]);
+    if (this->thread[tid]->kernelStats)
+        this->thread[tid]->kernelStats->callpal(palFunc,
+                                                this->execContexts[tid]);
 
     switch (palFunc) {
       case PAL::halt:
diff --git a/cpu/o3/iew_impl.hh b/cpu/o3/iew_impl.hh
index cf28f2efc..b0137d7fc 100644
--- a/cpu/o3/iew_impl.hh
+++ b/cpu/o3/iew_impl.hh
@@ -1232,13 +1232,14 @@ DefaultIEW<Impl>::executeInsts()
 #endif
 
     // Execute/writeback any instructions that are available.
+    int insts_to_execute = fromIssue->size;
     int inst_num = 0;
-    for ( ; inst_num < issueWidth && fromIssue->insts[inst_num];
+    for (; inst_num < insts_to_execute;
           ++inst_num) {
 
         DPRINTF(IEW, "Execute: Executing instructions from IQ.\n");
 
-        DynInstPtr inst = fromIssue->insts[inst_num];
+        DynInstPtr inst = instQueue.getInstToExecute();
 
         DPRINTF(IEW, "Execute: Processing PC %#x, [tid:%i] [sn:%i].\n",
                 inst->readPC(), inst->threadNumber,inst->seqNum);
diff --git a/cpu/o3/inst_queue.hh b/cpu/o3/inst_queue.hh
index 6bdf4ddc2..518de73d9 100644
--- a/cpu/o3/inst_queue.hh
+++ b/cpu/o3/inst_queue.hh
@@ -171,6 +171,8 @@ class InstructionQueue
      */
     void insertBarrier(DynInstPtr &barr_inst);
 
+    DynInstPtr getInstToExecute();
+
     /**
      * Records the instruction as the producer of a register without
      * adding it to the rest of the IQ.
@@ -272,6 +274,8 @@ class InstructionQueue
     /** List of all the instructions in the IQ (some of which may be issued). */
     std::list<DynInstPtr> instList[Impl::MaxThreads];
 
+    std::list<DynInstPtr> instsToExecute;
+
     /**
      * Struct for comparing entries to be added to the priority queue.  This
      * gives reverse ordering to the instructions in terms of sequence
diff --git a/cpu/o3/inst_queue_impl.hh b/cpu/o3/inst_queue_impl.hh
index 71541b4f8..f1dc4e01f 100644
--- a/cpu/o3/inst_queue_impl.hh
+++ b/cpu/o3/inst_queue_impl.hh
@@ -589,6 +589,16 @@ InstructionQueue<Impl>::insertBarrier(DynInstPtr &barr_inst)
 }
 
 template <class Impl>
+typename Impl::DynInstPtr
+InstructionQueue<Impl>::getInstToExecute()
+{
+    assert(!instsToExecute.empty());
+    DynInstPtr inst = instsToExecute.front();
+    instsToExecute.pop_front();
+    return inst;
+}
+
+template <class Impl>
 void
 InstructionQueue<Impl>::addToOrderList(OpClass op_class)
 {
@@ -662,9 +672,11 @@ InstructionQueue<Impl>::processFUCompletion(DynInstPtr &inst, int fu_idx)
     // @todo: This could break if there's multiple multi-cycle ops
     // finishing on this cycle.  Maybe implement something like
     // instToCommit in iew_impl.hh.
-    int &size = issueToExecuteQueue->access(0)->size;
+    issueToExecuteQueue->access(0)->size++;
+    instsToExecute.push_back(inst);
+//    int &size = issueToExecuteQueue->access(0)->size;
 
-    issueToExecuteQueue->access(0)->insts[size++] = inst;
+//    issueToExecuteQueue->access(0)->insts[size++] = inst;
 }
 
 // @todo: Figure out a better way to remove the squashed items from the
@@ -690,9 +702,8 @@ InstructionQueue<Impl>::scheduleReadyInsts()
     ListOrderIt order_it = listOrder.begin();
     ListOrderIt order_end_it = listOrder.end();
     int total_issued = 0;
-    int exec_queue_slot = i2e_info->size;
 
-    while (exec_queue_slot < totalWidth && total_issued < totalWidth &&
+    while (total_issued < totalWidth &&
            order_it != order_end_it) {
         OpClass op_class = (*order_it).queueType;
 
@@ -733,8 +744,9 @@ InstructionQueue<Impl>::scheduleReadyInsts()
 
         if (idx == -2 || idx != -1) {
             if (op_latency == 1) {
-                i2e_info->insts[exec_queue_slot++] = issuing_inst;
+//                i2e_info->insts[exec_queue_slot++] = issuing_inst;
                 i2e_info->size++;
+                instsToExecute.push_back(issuing_inst);
 
                 // Add the FU onto the list of FU's to be freed next
                 // cycle if we used one.
diff --git a/cpu/ozone/cpu.hh b/cpu/ozone/cpu.hh
index 7e12e75e5..5af2b02b2 100644
--- a/cpu/ozone/cpu.hh
+++ b/cpu/ozone/cpu.hh
@@ -57,6 +57,10 @@ class Sampler;
 class RemoteGDB;
 class GDBListener;
 
+namespace Kernel {
+    class Statistics;
+};
+
 #else
 
 class Process;
@@ -116,6 +120,8 @@ class OzoneCPU : public BaseCPU
         AlphaITB *getITBPtr() { return cpu->itb; }
 
         AlphaDTB * getDTBPtr() { return cpu->dtb; }
+
+        Kernel::Statistics *getKernelStats() { return thread->kernelStats; }
 #else
         Process *getProcessPtr() { return thread->process; }
 #endif
@@ -238,14 +244,7 @@ class OzoneCPU : public BaseCPU
 
   private:
     OzoneThreadState<Impl> thread;
-/*
-    // Squash event for when the XC needs to squash all inflight instructions.
-    struct XCSquashEvent : public Event
-    {
-        void process();
-        const char *description();
-    };
-*/
+
   public:
     // main simulation loop (one cycle)
     void tick();
@@ -288,7 +287,6 @@ class OzoneCPU : public BaseCPU
     void trace_data(T data);
 
   public:
-    //
     enum Status {
         Running,
         Idle,
@@ -325,8 +323,6 @@ class OzoneCPU : public BaseCPU
 
     int readCpuId() { return cpuId; }
 
-//    FunctionalMemory *getMemPtr() { return mem; }
-
     int cpuId;
 
     void switchOut(Sampler *sampler);
@@ -369,8 +365,6 @@ class OzoneCPU : public BaseCPU
     Status status() const { return _status; }
     void setStatus(Status new_status) { _status = new_status; }
 
-    // Not sure what an activate() call on the CPU's proxy XC would mean...
-
     virtual void activateContext(int thread_num, int delay);
     virtual void suspendContext(int thread_num);
     virtual void deallocateContext(int thread_num);
@@ -384,7 +378,6 @@ class OzoneCPU : public BaseCPU
   public:
     Counter numInst;
     Counter startNumInst;
-//    Stats::Scalar<> numInsts;
 
     virtual Counter totalInstructions() const
     {
@@ -392,9 +385,6 @@ class OzoneCPU : public BaseCPU
     }
 
   private:
-    // number of simulated memory references
-//    Stats::Scalar<> numMemRefs;
-
     // number of simulated loads
     Counter numLoad;
     Counter startNumLoad;
@@ -472,7 +462,6 @@ class OzoneCPU : public BaseCPU
     template <class T>
     Fault read(MemReqPtr &req, T &data)
     {
-//	panic("CPU READ NOT IMPLEMENTED W/NEW MEMORY\n");
 #if 0
 #if FULL_SYSTEM && defined(TARGET_ALPHA)
         if (req->flags & LOCKED) {
@@ -483,7 +472,6 @@ class OzoneCPU : public BaseCPU
 #endif
         Fault error;
         if (req->flags & LOCKED) {
-//            lockAddr = req->paddr;
             lockAddrList.insert(req->paddr);
             lockFlag = true;
         }
@@ -558,7 +546,7 @@ class OzoneCPU : public BaseCPU
             if (req->flags & UNCACHEABLE) {
                 req->result = 2;
             } else {
-                if (this->lockFlag/* && this->lockAddr == req->paddr*/) {
+                if (this->lockFlag) {
                     if (lockAddrList.find(req->paddr) !=
                         lockAddrList.end()) {
                         req->result = 1;
diff --git a/cpu/ozone/cpu_impl.hh b/cpu/ozone/cpu_impl.hh
index 031b4b145..5675da3a8 100644
--- a/cpu/ozone/cpu_impl.hh
+++ b/cpu/ozone/cpu_impl.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005 The Regents of The University of Michigan
+ * Copyright (c) 2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -26,8 +26,8 @@
  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include <cstdio>
-#include <cstdlib>
+//#include <cstdio>
+//#include <cstdlib>
 
 #include "arch/isa_traits.hh" // For MachInst
 #include "base/trace.hh"
@@ -39,7 +39,7 @@
 #include "cpu/ozone/cpu.hh"
 #include "cpu/quiesce_event.hh"
 #include "cpu/static_inst.hh"
-#include "mem/base_mem.hh"
+//#include "mem/base_mem.hh"
 #include "mem/mem_interface.hh"
 #include "sim/sim_object.hh"
 #include "sim/stats.hh"
@@ -50,7 +50,7 @@
 #include "arch/alpha/tlb.hh"
 #include "arch/vtophys.hh"
 #include "base/callback.hh"
-#include "base/remote_gdb.hh"
+//#include "base/remote_gdb.hh"
 #include "cpu/profile.hh"
 #include "kern/kernel_stats.hh"
 #include "mem/functional/memory_control.hh"
@@ -94,80 +94,26 @@ OzoneCPU<Impl>::TickEvent::description()
 {
     return "OzoneCPU tick event";
 }
-/*
-template <class Impl>
-OzoneCPU<Impl>::ICacheCompletionEvent::ICacheCompletionEvent(OzoneCPU *_cpu)
-    : Event(&mainEventQueue),
-      cpu(_cpu)
-{
-}
-
-template <class Impl>
-void
-OzoneCPU<Impl>::ICacheCompletionEvent::process()
-{
-    cpu->processICacheCompletion();
-}
-
-template <class Impl>
-const char *
-OzoneCPU<Impl>::ICacheCompletionEvent::description()
-{
-    return "OzoneCPU I-cache completion event";
-}
-
-template <class Impl>
-OzoneCPU<Impl>::DCacheCompletionEvent::
-DCacheCompletionEvent(OzoneCPU *_cpu,
-                      DynInstPtr &_inst,
-                      DCacheCompEventIt &_dcceIt)
-    : Event(&mainEventQueue),
-      cpu(_cpu),
-      inst(_inst),
-      dcceIt(_dcceIt)
-{
-    this->setFlags(Event::AutoDelete);
-}
-
-template <class Impl>
-void
-OzoneCPU<Impl>::DCacheCompletionEvent::process()
-{
-    inst->setCompleted();
-
-    // Maybe remove the EA from the list of addrs?
-    cpu->eaList.clearAddr(inst->seqNum, inst->getEA());
-    cpu->dCacheCompList.erase(this->dcceIt);
-}
 
 template <class Impl>
-const char *
-OzoneCPU<Impl>::DCacheCompletionEvent::description()
-{
-    return "OzoneCPU D-cache completion event";
-}
-*/
-template <class Impl>
 OzoneCPU<Impl>::OzoneCPU(Params *p)
 #if FULL_SYSTEM
-    : BaseCPU(p), thread(this, 0, p->mem), tickEvent(this, p->width), mem(p->mem),
+    : BaseCPU(p), thread(this, 0, p->mem), tickEvent(this, p->width),
+      mem(p->mem),
 #else
     : BaseCPU(p), thread(this, 0, p->workload[0], 0), tickEvent(this, p->width),
       mem(p->workload[0]->getMemory()),
 #endif
       comm(5, 5)
 {
-    if (p->checker) {
-        BaseCPU *temp_checker = p->checker;
-        checker = dynamic_cast<Checker<DynInstPtr> *>(temp_checker);
-    } else {
-        checker = NULL;
-    }
     frontEnd = new FrontEnd(p);
     backEnd = new BackEnd(p);
 
     _status = Idle;
-    if (checker) {
+
+    if (p->checker) {
+        BaseCPU *temp_checker = p->checker;
+        checker = dynamic_cast<Checker<DynInstPtr> *>(temp_checker);
         checker->setMemory(mem);
 #if FULL_SYSTEM
         checker->setSystem(p->system);
@@ -176,19 +122,18 @@ OzoneCPU<Impl>::OzoneCPU(Params *p)
         thread.xcProxy = checkerXC;
         xcProxy = checkerXC;
     } else {
+        checker = NULL;
         thread.xcProxy = &ozoneXC;
         xcProxy = &ozoneXC;
     }
 
-    thread.inSyscall = false;
-
     ozoneXC.cpu = this;
     ozoneXC.thread = &thread;
 
+    thread.inSyscall = false;
+
     thread.setStatus(ExecContext::Suspended);
 #if FULL_SYSTEM
-//    xc = new ExecContext(this, 0, p->system, p->itb, p->dtb, p->mem);
-
     /***** All thread state stuff *****/
     thread.cpu = this;
     thread.tid = 0;
@@ -217,31 +162,15 @@ OzoneCPU<Impl>::OzoneCPU(Params *p)
     thread.profileNode = &dummyNode;
     thread.profilePC = 3;
 #else
-//    xc = new ExecContext(this, /* thread_num */ 0, p->workload[0], /* asid */ 0);
     thread.cpu = this;
     thread.tid = 0;
     thread.process = p->workload[0];
-//    thread.mem = thread.process->getMemory();
     thread.asid = 0;
 #endif // !FULL_SYSTEM
-/*
-    icacheInterface = p->icache_interface;
-    dcacheInterface = p->dcache_interface;
-
-    cacheMemReq = new MemReq();
-    cacheMemReq->xc = xc;
-    cacheMemReq->asid = 0;
-    cacheMemReq->data = new uint8_t[64];
-*/
+
     numInst = 0;
     startNumInst = 0;
-/*    numLoad = 0;
-    startNumLoad = 0;
-    lastIcacheStall = 0;
-    lastDcacheStall = 0;
 
-    issueWidth = p->issueWidth;
-*/
     execContexts.push_back(xcProxy);
 
     frontEnd->setCPU(this);
@@ -286,47 +215,7 @@ template <class Impl>
 OzoneCPU<Impl>::~OzoneCPU()
 {
 }
-/*
-template <class Impl>
-void
-OzoneCPU<Impl>::copyFromXC()
-{
-    for (int i = 0; i < TheISA::TotalNumRegs; ++i) {
-        if (i < TheISA::NumIntRegs) {
-            renameTable[i]->setIntResult(xc->readIntReg(i));
-        } else if (i < TheISA::NumFloatRegs) {
-            renameTable[i]->setDoubleResult(xc->readFloatRegDouble(i));
-        }
-    }
-
-    DPRINTF(OzoneCPU, "Func Exe inst is: %i\n", xc->func_exe_inst);
-    backEnd->funcExeInst = xc->func_exe_inst;
-//    PC = xc->readPC();
-//    nextPC = xc->regs.npc;
-}
-
-template <class Impl>
-void
-OzoneCPU<Impl>::copyToXC()
-{
-    for (int i = 0; i < TheISA::TotalNumRegs; ++i) {
-        if (i < TheISA::NumIntRegs) {
-            xc->setIntReg(i, renameTable[i]->readIntResult());
-        } else if (i < TheISA::NumFloatRegs) {
-            xc->setFloatRegDouble(i, renameTable[i]->readDoubleResult());
-        }
-    }
-
-    this->xc->regs.miscRegs.fpcr = this->regFile.miscRegs[tid].fpcr;
-    this->xc->regs.miscRegs.uniq = this->regFile.miscRegs[tid].uniq;
-    this->xc->regs.miscRegs.lock_flag = this->regFile.miscRegs[tid].lock_flag;
-    this->xc->regs.miscRegs.lock_addr = this->regFile.miscRegs[tid].lock_addr;
 
-    xc->func_exe_inst = backEnd->funcExeInst;
-    xc->regs.pc = PC;
-    xc->regs.npc = nextPC;
-}
-*/
 template <class Impl>
 void
 OzoneCPU<Impl>::switchOut(Sampler *_sampler)
@@ -394,7 +283,6 @@ OzoneCPU<Impl>::activateContext(int thread_num, int delay)
 {
     // Eventually change this in SMT.
     assert(thread_num == 0);
-//    assert(xcProxy);
 
     assert(_status == Idle);
     notIdleFraction++;
@@ -410,8 +298,8 @@ OzoneCPU<Impl>::suspendContext(int thread_num)
 {
     // Eventually change this in SMT.
     assert(thread_num == 0);
-//    assert(xcProxy);
-    // @todo: Figure out how to initially set the status properly so this is running.
+    // @todo: Figure out how to initially set the status properly so
+    // this is running.
 //    assert(_status == Running);
     notIdleFraction--;
     unscheduleTickEvent();
@@ -486,14 +374,7 @@ void
 OzoneCPU<Impl>::init()
 {
     BaseCPU::init();
-/*
-    copyFromXC();
 
-    // ALso copy over PC/nextPC.  This isn't normally copied in "copyFromXC()"
-    // so that the XC doesn't mess up the PC when returning from a syscall.
-    PC = xc->readPC();
-    nextPC = xc->regs.npc;
-*/
     // Mark this as in syscall so it won't need to squash
     thread.inSyscall = true;
 #if FULL_SYSTEM
@@ -514,8 +395,6 @@ template <class Impl>
 void
 OzoneCPU<Impl>::serialize(std::ostream &os)
 {
-    // At this point, all DCacheCompEvents should be processed.
-
     BaseCPU::serialize(os);
     SERIALIZE_ENUM(_status);
     nameOut(os, csprintf("%s.xc", name()));
@@ -631,31 +510,7 @@ OzoneCPU<Impl>::dbg_vtophys(Addr addr)
     return vtophys(xcProxy, addr);
 }
 #endif // FULL_SYSTEM
-/*
-template <class Impl>
-void
-OzoneCPU<Impl>::processICacheCompletion()
-{
-    switch (status()) {
-      case IcacheMiss:
-        DPRINTF(OzoneCPU, "OzoneCPU: Finished Icache miss.\n");
-
-        icacheStallCycles += curTick - lastIcacheStall;
-        _status = IcacheMissComplete;
-        cacheBlkValid = true;
-//	scheduleTickEvent(1);
-        break;
-      case SwitchedOut:
-        // If this CPU has been switched out due to sampling/warm-up,
-        // ignore any further status changes (e.g., due to cache
-        // misses outstanding at the time of the switch).
-        return;
-      default:
-        panic("OzoneCPU::processICacheCompletion: bad state");
-        break;
-    }
-}
-*/
+
 #if FULL_SYSTEM
 template <class Impl>
 void
@@ -663,7 +518,6 @@ OzoneCPU<Impl>::post_interrupt(int int_num, int index)
 {
     BaseCPU::post_interrupt(int_num, index);
 
-//    if (thread._status == ExecContext::Suspended) {
     if (_status == Idle) {
         DPRINTF(IPI,"Suspended Processor awoke\n");
 //	thread.activate();
@@ -690,9 +544,6 @@ OzoneCPU<Impl>::tick()
     frontEnd->tick();
     backEnd->tick();
 
-    // Do this here?  For now the front end will control the PC.
-//    PC = nextPC;
-
     // check for instruction-count-based events
     comInstEventQueue[0]->serviceEvents(numInst);
 
@@ -742,11 +593,13 @@ OzoneCPU<Impl>::setSyscallReturn(SyscallReturn return_value, int tid)
     if (return_value.successful()) {
         // no error
         thread.renameTable[SyscallSuccessReg]->setIntResult(0);
-        thread.renameTable[ReturnValueReg]->setIntResult(return_value.value());
+        thread.renameTable[ReturnValueReg]->setIntResult(
+            return_value.value());
     } else {
         // got an error, return details
         thread.renameTable[SyscallSuccessReg]->setIntResult((IntReg) -1);
-        thread.renameTable[ReturnValueReg]->setIntResult(-return_value.value());
+        thread.renameTable[ReturnValueReg]->setIntResult(
+            -return_value.value());
     }
 }
 #else
@@ -756,15 +609,10 @@ OzoneCPU<Impl>::hwrei()
 {
     // Need to move this to ISA code
     // May also need to make this per thread
-/*
-    if (!inPalMode())
-        return new UnimplementedOpcodeFault;
 
-    thread.setNextPC(thread.readMiscReg(AlphaISA::IPR_EXC_ADDR));
-*/
     lockFlag = false;
     lockAddrList.clear();
-    kernelStats->hwrei();
+    thread.kernelStats->hwrei();
 
     checkInterrupts = true;
 
@@ -835,7 +683,7 @@ OzoneCPU<Impl>::simPalCheck(int palFunc)
 {
     // Need to move this to ISA code
     // May also need to make this per thread
-    this->kernelStats->callpal(palFunc, xcProxy);
+    thread.kernelStats->callpal(palFunc, xcProxy);
 
     switch (palFunc) {
       case PAL::halt:
@@ -874,7 +722,6 @@ template <class Impl>
 void
 OzoneCPU<Impl>::OzoneXC::setStatus(Status new_status)
 {
-//    cpu->_status = new_status;
     thread->_status = new_status;
 }
 
@@ -932,6 +779,7 @@ OzoneCPU<Impl>::OzoneXC::takeOverFrom(ExecContext *old_context)
     setStatus(old_context->status());
     copyArchRegs(old_context);
     setCpuId(old_context->readCpuId());
+
 #if !FULL_SYSTEM
     setFuncExeInst(old_context->readFuncExeInst());
 #else
@@ -944,6 +792,8 @@ OzoneCPU<Impl>::OzoneXC::takeOverFrom(ExecContext *old_context)
     if (thread->quiesceEvent) {
         thread->quiesceEvent->xc = this;
     }
+
+    thread->kernelStats = old_context->getKernelStats();
 //    storeCondFailures = 0;
     cpu->lockFlag = false;
 #endif
@@ -954,7 +804,12 @@ OzoneCPU<Impl>::OzoneXC::takeOverFrom(ExecContext *old_context)
 template <class Impl>
 void
 OzoneCPU<Impl>::OzoneXC::regStats(const std::string &name)
-{ }
+{
+#if FULL_SYSTEM
+    thread->kernelStats = new Kernel::Statistics(cpu->system);
+    thread->kernelStats->regStats(name + ".kern");
+#endif
+}
 
 template <class Impl>
 void
diff --git a/cpu/ozone/dyn_inst.hh b/cpu/ozone/dyn_inst.hh
index f251c28ea..5d48bb361 100644
--- a/cpu/ozone/dyn_inst.hh
+++ b/cpu/ozone/dyn_inst.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005 The Regents of The University of Michigan
+ * Copyright (c) 2005-2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -52,8 +52,6 @@ class OzoneDynInst : public BaseDynInst<Impl>
     // Typedef for DynInstPtr.  This is really just a RefCountingPtr<OoODynInst>.
     typedef typename Impl::DynInstPtr DynInstPtr;
 
-//    typedef typename Impl::BranchPred::BPredInfo BPredInfo;
-
     typedef TheISA::ExtMachInst ExtMachInst;
     typedef TheISA::MachInst MachInst;
     typedef TheISA::MiscReg MiscReg;
@@ -107,12 +105,6 @@ class OzoneDynInst : public BaseDynInst<Impl>
     // up.  In the future, you only really need a counter.
     bool memDepReady() { return srcMemInsts.empty(); }
 
-//    void setBPredInfo(const BPredInfo &bp_info) { bpInfo = bp_info; }
-
-//    BPredInfo &getBPredInfo() { return bpInfo; }
-
-//    OzoneXC *thread;
-
   private:
     void initInstPtrs();
 
@@ -133,20 +125,12 @@ class OzoneDynInst : public BaseDynInst<Impl>
      */
     DynInstPtr prevDestInst[MaxInstSrcRegs];
 
-//    BPredInfo bpInfo;
-
   public:
 
     Fault initiateAcc();
 
     Fault completeAcc();
-/*
-    template <class T>
-    Fault read(Addr addr, T &data, unsigned flags);
 
-    template <class T>
-    Fault write(T data, Addr addr, unsigned flags, uint64_t *res);
-*/
     // The register accessor methods provide the index of the
     // instruction's operand (e.g., 0 or 1), not the architectural
     // register index, to simplify the implementation of register
@@ -244,38 +228,4 @@ class OzoneDynInst : public BaseDynInst<Impl>
     bool iqItValid;
 };
 
-/*
-template<class Impl>
-template<class T>
-inline Fault
-OzoneDynInst<Impl>::read(Addr addr, T &data, unsigned flags)
-{
-    Fault fault = this->cpu->read(addr, data, flags, this);
-
-    if (this->traceData) {
-        this->traceData->setAddr(addr);
-        this->traceData->setData(data);
-    }
-
-    return fault;
-}
-
-template<class Impl>
-template<class T>
-inline Fault
-OzoneDynInst<Impl>::write(T data, Addr addr, unsigned flags, uint64_t *res)
-{
-    Fault fault = this->cpu->write(data, addr, flags, res, this);
-
-    this->storeSize = sizeof(T);
-    this->storeData = data;
-
-    if (this->traceData) {
-        this->traceData->setAddr(addr);
-        this->traceData->setData(data);
-    }
-
-    return fault;
-}
-*/
 #endif // __CPU_OZONE_DYN_INST_HH__
diff --git a/cpu/ozone/dyn_inst_impl.hh b/cpu/ozone/dyn_inst_impl.hh
index a7e4460a1..f891ec515 100644
--- a/cpu/ozone/dyn_inst_impl.hh
+++ b/cpu/ozone/dyn_inst_impl.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005 The Regents of The University of Michigan
+ * Copyright (c) 2005-2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -266,12 +266,7 @@ OzoneDynInst<Impl>::hwrei()
     this->setNextPC(this->thread->readMiscReg(AlphaISA::IPR_EXC_ADDR));
 
     this->cpu->hwrei();
-/*
-    this->cpu->kernelStats->hwrei();
 
-    this->cpu->checkInterrupts = true;
-    this->cpu->lockFlag = false;
-*/
     // FIXME: XXX check for interrupts? XXX
     return NoFault;
 }
diff --git a/cpu/ozone/front_end.hh b/cpu/ozone/front_end.hh
index 326f7d2c9..dd382491f 100644
--- a/cpu/ozone/front_end.hh
+++ b/cpu/ozone/front_end.hh
@@ -1,14 +1,39 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
 #ifndef __CPU_OZONE_FRONT_END_HH__
 #define __CPU_OZONE_FRONT_END_HH__
 
 #include <deque>
 
-//#include "cpu/ozone/cpu.hh"
 #include "cpu/inst_seq.hh"
 #include "cpu/o3/bpred_unit.hh"
 #include "cpu/ozone/rename_table.hh"
-//#include "cpu/ozone/thread_state.hh"
 #include "mem/mem_req.hh"
 #include "sim/eventq.hh"
 #include "sim/stats.hh"
@@ -132,11 +157,6 @@ class FrontEnd
 
     typedef typename Impl::BranchPred BranchPred;
 
-    // Typedef for semi-opaque type that holds any information the branch
-    // predictor needs to update itself.  Only two fields are used outside of
-    // branch predictor, nextPC and isTaken.
-//    typedef typename BranchPred::BPredInfo BPredInfo;
-
     BranchPred branchPred;
 
     class ICacheCompletionEvent : public Event
diff --git a/cpu/ozone/front_end_impl.hh b/cpu/ozone/front_end_impl.hh
index cd57aeef4..15adae9b4 100644
--- a/cpu/ozone/front_end_impl.hh
+++ b/cpu/ozone/front_end_impl.hh
@@ -1,3 +1,30 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
 #include "arch/faults.hh"
 #include "arch/isa_traits.hh"
@@ -26,14 +53,6 @@ FrontEnd<Impl>::FrontEnd(Params *params)
 
     status = Idle;
 
-    // Setup branch predictor.
-
-    // Setup Memory Request
-/*
-    memReq = new MemReq();
-    memReq->asid = 0;
-    memReq->data = new uint8_t[64];
-*/
     memReq = NULL;
     // Size of cache block.
     cacheBlkSize = icacheInterface ? icacheInterface->getBlockSize() : 64;
@@ -77,7 +96,6 @@ void
 FrontEnd<Impl>::setXC(ExecContext *xc_ptr)
 {
     xc = xc_ptr;
-//    memReq->xc = xc;
 }
 
 template <class Impl>
@@ -321,7 +339,6 @@ FrontEnd<Impl>::tick()
             break;
         }
 
-        // if (generalizeFetch) {
         processInst(inst);
 
         if (status == SerializeBlocked) {
@@ -333,11 +350,6 @@ FrontEnd<Impl>::tick()
         instBuffer.push_back(inst);
         ++instBufferSize;
         ++num_inst;
-        // } else {
-        // fetch(num_inst);
-        // decode(num_inst);
-        // rename(num_inst);
-        // }
 
 #if FULL_SYSTEM
         if (inst->isQuiesce()) {
@@ -402,10 +414,6 @@ FrontEnd<Impl>::fetchCacheLine()
     // Translate the instruction request.
     fault = cpu->translateInstReq(memReq);
 
-    // In the case of faults, the fetch stage may need to stall and wait
-    // on what caused the fetch (ITB or Icache miss).
-//    assert(fault == NoFault);
-
     // Now do the timing access to see whether or not the instruction
     // exists within the cache.
     if (icacheInterface && fault == NoFault) {
@@ -466,7 +474,6 @@ FrontEnd<Impl>::processInst(DynInstPtr &inst)
 
     Addr inst_PC = inst->readPC();
 
-//    BPredInfo bp_info = branchPred.lookup(inst_PC);
     if (!inst->isControl()) {
         inst->setPredTarg(inst->readNextPC());
     } else {
@@ -482,7 +489,6 @@ FrontEnd<Impl>::processInst(DynInstPtr &inst)
             "%#x\n", inst->seqNum, inst_PC, next_PC);
 
 //    inst->setNextPC(next_PC);
-//    inst->setBPredInfo(bp_info);
 
     // Not sure where I should set this
     PC = next_PC;
@@ -535,7 +541,7 @@ void
 FrontEnd<Impl>::handleFault(Fault &fault)
 {
     DPRINTF(FE, "Fault at fetch, telling commit\n");
-//    backEnd->fetchFault(fault);
+
     // We're blocked on the back end until it handles this fault.
     status = TrapPending;
 
@@ -586,9 +592,6 @@ FrontEnd<Impl>::squash(const InstSeqNum &squash_num, const Addr &next_PC,
         instBuffer.pop_back();
         --instBufferSize;
 
-        // Fix up branch predictor if necessary.
-//        branchPred.undo(inst->getBPredInfo());
-
         freeRegs+= inst->numDestRegs();
     }
 
@@ -607,7 +610,6 @@ FrontEnd<Impl>::squash(const InstSeqNum &squash_num, const Addr &next_PC,
     // Clear the icache miss if it's outstanding.
     if (status == IcacheMissStall && icacheInterface) {
         DPRINTF(FE, "Squashing outstanding Icache miss.\n");
-//        icacheInterface->squash(0);
         memReq = NULL;
     }
 
@@ -693,17 +695,9 @@ template <class Impl>
 bool
 FrontEnd<Impl>::updateStatus()
 {
-//    bool rename_block = freeRegs <= 0;
     bool serialize_block = !backEnd->robEmpty() || instBufferSize;
     bool be_block = cpu->decoupledFrontEnd ? false : backEnd->isBlocked();
     bool ret_val = false;
-/*
-  // Should already be handled through addFreeRegs function
-    if (status == RenameBlocked && !rename_block) {
-        status = Running;
-        ret_val = true;
-    }
-*/
 
     if (status == SerializeBlocked && !serialize_block) {
         status = SerializeComplete;
@@ -753,10 +747,6 @@ FrontEnd<Impl>::getInstFromCacheline()
 
     // PC of inst is not in this cache block
     if (PC >= (cacheBlkPC + cacheBlkSize) || PC < cacheBlkPC || !cacheBlkValid) {
-//        DPRINTF(OoOCPU, "OoOCPU: PC is not in this cache block\n");
-//        DPRINTF(OoOCPU, "OoOCPU: PC: %#x, cacheBlkPC: %#x, cacheBlkValid: %i",
-//                PC, cacheBlkPC, cacheBlkValid);
-//        panic("Instruction not in cache line or cache line invalid!");
         return NULL;
     }
 
diff --git a/cpu/ozone/lw_back_end.hh b/cpu/ozone/lw_back_end.hh
index 770b66ad5..1c03ffb73 100644
--- a/cpu/ozone/lw_back_end.hh
+++ b/cpu/ozone/lw_back_end.hh
@@ -1,3 +1,30 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
 #ifndef __CPU_OZONE_LW_BACK_END_HH__
 #define __CPU_OZONE_LW_BACK_END_HH__
@@ -238,10 +265,6 @@ class LWBackEnd
     Counter funcExeInst;
 
   private:
-//    typedef typename Impl::InstQueue InstQueue;
-
-//    InstQueue IQ;
-
     typedef typename Impl::LdstQueue LdstQueue;
 
     LdstQueue LSQ;
@@ -342,8 +365,6 @@ class LWBackEnd
 
     bool exactFullStall;
 
-//    bool fetchRedirect[Impl::MaxThreads];
-
     // number of cycles stalled for D-cache misses
 /*    Stats::Scalar<> dcacheStallCycles;
       Counter lastDcacheStall;
@@ -438,43 +459,6 @@ template <class T>
 Fault
 LWBackEnd<Impl>::read(MemReqPtr &req, T &data, int load_idx)
 {
-/*    memReq->reset(addr, sizeof(T), flags);
-
-    // translate to physical address
-    Fault fault = cpu->translateDataReadReq(memReq);
-
-    // if we have a cache, do cache access too
-    if (fault == NoFault && dcacheInterface) {
-        memReq->cmd = Read;
-        memReq->completionEvent = NULL;
-        memReq->time = curTick;
-        memReq->flags &= ~INST_READ;
-        MemAccessResult result = dcacheInterface->access(memReq);
-
-        // Ugly hack to get an event scheduled *only* if the access is
-        // a miss.  We really should add first-class support for this
-        // at some point.
-        if (result != MA_HIT && dcacheInterface->doEvents()) {
-            // Fix this hack for keeping funcExeInst correct with loads that
-            // are executed twice.
-            --funcExeInst;
-
-            memReq->completionEvent = &cacheCompletionEvent;
-            lastDcacheStall = curTick;
-//	    unscheduleTickEvent();
-//	    status = DcacheMissStall;
-            DPRINTF(OzoneCPU, "Dcache miss stall!\n");
-        } else {
-            // do functional access
-            fault = thread->mem->read(memReq, data);
-
-        }
-    }
-*/
-/*
-    if (!dcacheInterface && (memReq->flags & UNCACHEABLE))
-        recordEvent("Uncached Read");
-*/
     return LSQ.read(req, data, load_idx);
 }
 
@@ -483,39 +467,6 @@ template <class T>
 Fault
 LWBackEnd<Impl>::write(MemReqPtr &req, T &data, int store_idx)
 {
-/*
-    memReq->reset(addr, sizeof(T), flags);
-
-    // translate to physical address
-    Fault fault = cpu->translateDataWriteReq(memReq);
-
-    if (fault == NoFault && dcacheInterface) {
-        memReq->cmd = Write;
-        memcpy(memReq->data,(uint8_t *)&data,memReq->size);
-        memReq->completionEvent = NULL;
-        memReq->time = curTick;
-        memReq->flags &= ~INST_READ;
-        MemAccessResult result = dcacheInterface->access(memReq);
-
-        // Ugly hack to get an event scheduled *only* if the access is
-        // a miss.  We really should add first-class support for this
-        // at some point.
-        if (result != MA_HIT && dcacheInterface->doEvents()) {
-            memReq->completionEvent = &cacheCompletionEvent;
-            lastDcacheStall = curTick;
-//	    unscheduleTickEvent();
-//	    status = DcacheMissStall;
-            DPRINTF(OzoneCPU, "Dcache miss stall!\n");
-        }
-    }
-
-    if (res && (fault == NoFault))
-        *res = memReq->result;
-        */
-/*
-    if (!dcacheInterface && (memReq->flags & UNCACHEABLE))
-        recordEvent("Uncached Write");
-*/
     return LSQ.write(req, data, store_idx);
 }
 
diff --git a/cpu/ozone/lw_back_end_impl.hh b/cpu/ozone/lw_back_end_impl.hh
index db0872e52..881d6e6b1 100644
--- a/cpu/ozone/lw_back_end_impl.hh
+++ b/cpu/ozone/lw_back_end_impl.hh
@@ -1,7 +1,34 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
-#include "encumbered/cpu/full/op_class.hh"
 #include "cpu/checker/cpu.hh"
 #include "cpu/ozone/lw_back_end.hh"
+#include "encumbered/cpu/full/op_class.hh"
 
 template <class Impl>
 void
@@ -194,7 +221,6 @@ LWBackEnd<Impl>::LWBackEnd(Params *params)
     switchedOut = false;
     switchPending = false;
 
-//    IQ.setBE(this);
     LSQ.setBE(this);
 
     // Setup IQ and LSQ with their parameters here.
@@ -202,8 +228,6 @@ LWBackEnd<Impl>::LWBackEnd(Params *params)
 
     instsToExecute = i2e.getWire(-1);
 
-//    IQ.setIssueExecQueue(&i2e);
-
     dispatchWidth = params->dispatchWidth ? params->dispatchWidth : width;
     issueWidth = params->issueWidth ? params->issueWidth : width;
     wbWidth = params->wbWidth ? params->wbWidth : width;
@@ -538,8 +562,6 @@ LWBackEnd<Impl>::regStats()
         .desc("ROB Occupancy per cycle")
         .flags(total | cdf)
         ;
-
-//    IQ.regStats();
 }
 
 template <class Impl>
@@ -652,17 +674,7 @@ LWBackEnd<Impl>::tick()
         squashFromTrap();
     } else if (xcSquash) {
         squashFromXC();
-    } /*else if (fetchHasFault && robEmpty() && frontEnd->isEmpty() && !LSQ.hasStoresToWB()) {
-        DPRINTF(BE, "ROB and front end empty, handling fetch fault\n");
-        Fault fetch_fault = frontEnd->getFault();
-        if (fetch_fault == NoFault) {
-            DPRINTF(BE, "Fetch no longer has a fault, cancelling out.\n");
-            fetchHasFault = false;
-        } else {
-            handleFault(fetch_fault);
-            fetchHasFault = false;
-        }
-        }*/
+    }
 #endif
 
     if (dispatchStatus != Blocked) {
@@ -773,7 +785,8 @@ LWBackEnd<Impl>::dispatchInsts()
                     inst->iqItValid = true;
                     waitingInsts++;
                 } else {
-                    DPRINTF(BE, "Instruction [sn:%lli] ready, addding to exeList.\n",
+                    DPRINTF(BE, "Instruction [sn:%lli] ready, addding to "
+                            "exeList.\n",
                             inst->seqNum);
                     exeList.push(inst);
                 }
@@ -784,7 +797,8 @@ LWBackEnd<Impl>::dispatchInsts()
                 inst->setExecuted();
                 inst->setCanCommit();
             } else {
-                DPRINTF(BE, "Instruction [sn:%lli] ready, addding to exeList.\n",
+                DPRINTF(BE, "Instruction [sn:%lli] ready, addding to "
+                        "exeList.\n",
                         inst->seqNum);
                 exeList.push(inst);
             }
@@ -993,7 +1007,7 @@ LWBackEnd<Impl>::instToCommit(DynInstPtr &inst)
 
     writeback_count[0]++;
 }
-
+#if 0
 template <class Impl>
 void
 LWBackEnd<Impl>::writebackInsts()
@@ -1040,7 +1054,7 @@ LWBackEnd<Impl>::writebackInsts()
     consumer_inst[0]+= consumer_insts;
     writeback_count[0]+= inst_num;
 }
-
+#endif
 template <class Impl>
 bool
 LWBackEnd<Impl>::commitInst(int inst_num)
@@ -1219,15 +1233,15 @@ LWBackEnd<Impl>::commitInst(int inst_num)
 
     --numInsts;
     ++thread->funcExeInst;
-    // Maybe move this to where the fault is handled; if the fault is handled,
-    // don't try to set this myself as the fault will set it.  If not, then
-    // I set thread->PC = thread->nextPC and thread->nextPC = thread->nextPC + 4.
+    // Maybe move this to where the fault is handled; if the fault is
+    // handled, don't try to set this myself as the fault will set it.
+    // If not, then I set thread->PC = thread->nextPC and
+    // thread->nextPC = thread->nextPC + 4.
     thread->setPC(thread->readNextPC());
     thread->setNextPC(thread->readNextPC() + sizeof(TheISA::MachInst));
     updateComInstStats(inst);
 
     // Write the done sequence number here.
-//    LSQ.commitLoads(inst->seqNum);
     toIEW->doneSeqNum = inst->seqNum;
     lastCommitCycle = curTick;
 
@@ -1357,7 +1371,8 @@ LWBackEnd<Impl>::squash(const InstSeqNum &sn)
     }
 
     while (memBarrier && memBarrier->seqNum > sn) {
-        DPRINTF(BE, "[sn:%lli] Memory barrier squashed (or previously squashed)\n", memBarrier->seqNum);
+        DPRINTF(BE, "[sn:%lli] Memory barrier squashed (or previously "
+                "squashed)\n", memBarrier->seqNum);
         memBarrier->clearMemDependents();
         if (memBarrier->memDepReady()) {
             DPRINTF(BE, "No previous barrier\n");
diff --git a/cpu/ozone/lw_lsq.hh b/cpu/ozone/lw_lsq.hh
index 042610324..6fe343b42 100644
--- a/cpu/ozone/lw_lsq.hh
+++ b/cpu/ozone/lw_lsq.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2004-2005 The Regents of The University of Michigan
+ * Copyright (c) 2004-2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -138,7 +138,6 @@ class OzoneLWLSQ {
     /** Executes a load instruction. */
     Fault executeLoad(DynInstPtr &inst);
 
-//    Fault executeLoad(int lq_idx);
     /** Executes a store instruction. */
     Fault executeStore(DynInstPtr &inst);
 
@@ -304,10 +303,8 @@ class OzoneLWLSQ {
     Status _status;
 
     /** The store queue. */
-//    std::vector<SQEntry> storeQueue;
     std::list<SQEntry> storeQueue;
     /** The load queue. */
-//    std::vector<DynInstPtr> loadQueue;
     std::list<DynInstPtr> loadQueue;
 
     typedef typename std::list<SQEntry>::iterator SQIt;
@@ -365,7 +362,6 @@ class OzoneLWLSQ {
      */
     InstSeqNum stallingStoreIsn;
     /** The index of the above store. */
-//    int stallingLoadIdx;
     LQIt stallingLoad;
 
     /** Whether or not a load is blocked due to the memory system.  It is
@@ -398,8 +394,6 @@ class OzoneLWLSQ {
     template <class T>
     Fault write(MemReqPtr &req, T &data, int store_idx);
 
-    /** Returns the index of the head load instruction. */
-//    int getLoadHead() { return loadHead; }
     /** Returns the sequence number of the head load instruction. */
     InstSeqNum getLoadHeadSeqNum()
     {
@@ -411,8 +405,6 @@ class OzoneLWLSQ {
 
     }
 
-    /** Returns the index of the head store instruction. */
-//    int getStoreHead() { return storeHead; }
     /** Returns the sequence number of the head store instruction. */
     InstSeqNum getStoreHeadSeqNum()
     {
@@ -604,12 +596,7 @@ OzoneLWLSQ<Impl>::read(MemReqPtr &req, T &data, int load_idx)
         DPRINTF(OzoneLSQ, "D-cache: PC:%#x reading from paddr:%#x "
                 "vaddr:%#x flags:%i\n",
                 inst->readPC(), req->paddr, req->vaddr, req->flags);
-/*
-        Addr debug_addr = ULL(0xfffffc0000be81a8);
-        if (req->vaddr == debug_addr) {
-            debug_break();
-        }
-*/
+
         assert(!req->completionEvent);
         req->completionEvent =
             new typename BackEnd::LdWritebackEvent(inst, be);
@@ -631,9 +618,6 @@ OzoneLWLSQ<Impl>::read(MemReqPtr &req, T &data, int load_idx)
             _status = DcacheMissStall;
 
         } else {
-//            DPRINTF(Activity, "Activity: ld accessing mem hit [sn:%lli]\n",
-//                    inst->seqNum);
-
             DPRINTF(OzoneLSQ, "D-cache hit!\n");
         }
     } else {
@@ -664,12 +648,7 @@ OzoneLWLSQ<Impl>::write(MemReqPtr &req, T &data, int store_idx)
     assert(!req->data);
     req->data = new uint8_t[64];
     memcpy(req->data, (uint8_t *)&(*sq_it).data, req->size);
-/*
-    Addr debug_addr = ULL(0xfffffc0000be81a8);
-    if (req->vaddr == debug_addr) {
-        debug_break();
-    }
-*/
+
     // This function only writes the data to the store queue, so no fault
     // can happen here.
     return NoFault;
diff --git a/cpu/ozone/lw_lsq_impl.hh b/cpu/ozone/lw_lsq_impl.hh
index fdf6bff07..2f85a0396 100644
--- a/cpu/ozone/lw_lsq_impl.hh
+++ b/cpu/ozone/lw_lsq_impl.hh
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2004-2005 The Regents of The University of Michigan
+ * Copyright (c) 2004-2006 The Regents of The University of Michigan
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -104,12 +104,6 @@ OzoneLWLSQ<Impl>::init(Params *params, unsigned maxLQEntries,
         SQIndices.push(i);
     }
 
-    // May want to initialize these entries to NULL
-
-//    loadHead = loadTail = 0;
-
-//    storeHead = storeWBIdx = storeTail = 0;
-
     usedPorts = 0;
     cachePorts = params->cachePorts;
 
@@ -197,8 +191,6 @@ OzoneLWLSQ<Impl>::insert(DynInstPtr &inst)
     } else {
         insertStore(inst);
     }
-
-//    inst->setInLSQ();
 }
 
 template <class Impl>
@@ -569,12 +561,9 @@ OzoneLWLSQ<Impl>::writebackStores()
             }
 
             if (result != MA_HIT && dcacheInterface->doEvents()) {
-//                Event *wb = NULL;
                 store_event->miss = true;
                 typename BackEnd::LdWritebackEvent *wb = NULL;
                 if (req->flags & LOCKED) {
-                    // Stx_C does not generate a system port transaction.
-//                    req->result=1;
                     wb = new typename BackEnd::LdWritebackEvent(inst,
                                                             be);
                     store_event->wbEvent = wb;
@@ -585,8 +574,6 @@ OzoneLWLSQ<Impl>::writebackStores()
 //                DPRINTF(Activity, "Active st accessing mem miss [sn:%lli]\n",
 //                        inst->seqNum);
 
-                // Will stores need their own kind of writeback events?
-                // Do stores even need writeback events?
                 be->addDcacheMiss(inst);
 
                 lastDcacheStall = curTick;
@@ -604,20 +591,16 @@ OzoneLWLSQ<Impl>::writebackStores()
 //                        inst->seqNum);
 
                 if (req->flags & LOCKED) {
-                    // Stx_C does not generate a system port transaction.
-/*                    if (req->flags & UNCACHEABLE) {
-                        req->result = 2;
-                    } else {
-                        req->result = 1;
-                    }
-*/
+                    // Stx_C does not generate a system port
+                    // transaction in the 21264, but that might be
+                    // hard to accomplish in this model.
+
                     typename BackEnd::LdWritebackEvent *wb =
                         new typename BackEnd::LdWritebackEvent(inst,
                                                                be);
                     store_event->wbEvent = wb;
                 }
                 sq_it--;
-//                completeStore(inst->sqIdx);
             }
         } else {
             panic("Must HAVE DCACHE!!!!!\n");
@@ -780,7 +763,7 @@ OzoneLWLSQ<Impl>::completeStore(int store_idx)
     SQIndices.push(inst->sqIdx);
     storeQueue.erase(sq_it);
     --stores;
-//    assert(!inst->isCompleted());
+
     inst->setCompleted();
     if (cpu->checker) {
         cpu->checker->tick(inst);
@@ -791,7 +774,6 @@ template <class Impl>
 void
 OzoneLWLSQ<Impl>::switchOut()
 {
-//    assert(loads == 0);
     assert(storesToWB == 0);
     switchedOut = true;
     SQIt sq_it = --(storeQueue.end());
@@ -804,8 +786,6 @@ OzoneLWLSQ<Impl>::switchOut()
 
         if ((*sq_it).size == 0 && !(*sq_it).completed) {
             sq_it--;
-//            completeStore(inst->sqIdx);
-
             continue;
         }
 
@@ -817,7 +797,8 @@ OzoneLWLSQ<Impl>::switchOut()
             continue;
         } else if ((*sq_it).req->flags & LOCKED) {
             sq_it--;
-            assert(!(*sq_it).canWB || ((*sq_it).canWB && (*sq_it).req->flags & LOCKED));
+            assert(!(*sq_it).canWB ||
+                   ((*sq_it).canWB && (*sq_it).req->flags & LOCKED));
             continue;
         }
 
@@ -886,12 +867,6 @@ OzoneLWLSQ<Impl>::takeOverFrom(ExecContext *old_xc)
         SQIndices.push(i);
     }
 
-    // May want to initialize these entries to NULL
-
-//    loadHead = loadTail = 0;
-
-//    storeHead = storeWBIdx = storeTail = 0;
-
     usedPorts = 0;
 
     loadFaultInst = storeFaultInst = memDepViolator = NULL;
diff --git a/cpu/ozone/rename_table.hh b/cpu/ozone/rename_table.hh
index afbf6ff32..6ee23b21b 100644
--- a/cpu/ozone/rename_table.hh
+++ b/cpu/ozone/rename_table.hh
@@ -1,3 +1,31 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
 #ifndef __CPU_OZONE_RENAME_TABLE_HH__
 #define __CPU_OZONE_RENAME_TABLE_HH__
 
diff --git a/cpu/ozone/thread_state.hh b/cpu/ozone/thread_state.hh
index 269fc6459..c86c3a720 100644
--- a/cpu/ozone/thread_state.hh
+++ b/cpu/ozone/thread_state.hh
@@ -1,3 +1,30 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
 #ifndef __CPU_OZONE_THREAD_STATE_HH__
 #define __CPU_OZONE_THREAD_STATE_HH__
@@ -62,19 +89,14 @@ struct OzoneThreadState : public ThreadState {
 
     void setStatus(Status new_status) { _status = new_status; }
 
-    RenameTable<Impl> renameTable; // Should I include backend and frontend
-    // tables here?  For the ozone CPU, maybe, for the new full CPU, probably
-    // not...you wouldn't want threads just accessing the backend/frontend
-    // rename tables.
-    Addr PC; // What should these be set to?  Probably the committed ones.
+    RenameTable<Impl> renameTable;
+    Addr PC;
     Addr nextPC;
 
-    // Current instruction?
+    // Current instruction
     TheISA::MachInst inst;
 
     TheISA::RegFile regs;
-    // Front end?  Back end?
-//    MemReqPtr memReq;
 
     typename Impl::FullCPU *cpu;
 
diff --git a/cpu/simple/cpu.cc b/cpu/simple/cpu.cc
index 07f9d0dad..c03945ffa 100644
--- a/cpu/simple/cpu.cc
+++ b/cpu/simple/cpu.cc
@@ -782,7 +782,7 @@ SimpleCPU::tick()
 
 #if FULL_SYSTEM
         if (system->kernelBinning->fnbin) {
-            assert(kernelStats);
+            assert(cpuXC->getKernelStats());
             system->kernelBinning->execute(xcProxy, inst);
         }
 
diff --git a/cpu/thread_state.hh b/cpu/thread_state.hh
index e8381b9d3..e09cb12fd 100644
--- a/cpu/thread_state.hh
+++ b/cpu/thread_state.hh
@@ -1,3 +1,30 @@
+/*
+ * Copyright (c) 2006 The Regents of The University of Michigan
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
 
 #ifndef __CPU_THREAD_STATE_HH__
 #define __CPU_THREAD_STATE_HH__
@@ -8,11 +35,20 @@
 class EndQuiesceEvent;
 class FunctionProfile;
 class ProfileNode;
+namespace Kernel {
+    class Statistics;
+};
 #else
-class Process;
 class FunctionalMemory;
+class Process;
 #endif
 
+/**
+ *  Struct for holding general thread state that is needed across CPU
+ *  models.  This includes things such as pointers to the process,
+ *  memory, quiesce events, and certain stats.  This can be expanded
+ *  to hold more thread-specific stats within it.
+ */
 struct ThreadState {
 #if FULL_SYSTEM
     ThreadState(int _cpuId, int _tid, FunctionalMemory *_mem)
@@ -55,6 +91,7 @@ struct ThreadState {
 
     EndQuiesceEvent *quiesceEvent;
 
+    Kernel::Statistics *kernelStats;
 #else
     Process *process;
 
diff --git a/kern/system_events.cc b/kern/system_events.cc
index 9b9861497..221eb228d 100644
--- a/kern/system_events.cc
+++ b/kern/system_events.cc
@@ -67,15 +67,17 @@ FnEvent::process(ExecContext *xc)
 void
 IdleStartEvent::process(ExecContext *xc)
 {
-    xc->getCpuPtr()->kernelStats->setIdleProcess(
-        xc->readMiscReg(AlphaISA::IPR_PALtemp23), xc);
+    if (xc->getKernelStats())
+        xc->getKernelStats()->setIdleProcess(
+            xc->readMiscReg(AlphaISA::IPR_PALtemp23), xc);
     remove();
 }
 
 void
 InterruptStartEvent::process(ExecContext *xc)
 {
-    xc->getCpuPtr()->kernelStats->mode(Kernel::interrupt, xc);
+    if (xc->getKernelStats())
+        xc->getKernelStats()->mode(Kernel::interrupt, xc);
 }
 
 void
@@ -83,5 +85,6 @@ InterruptEndEvent::process(ExecContext *xc)
 {
     // We go back to kernel, if we are user, inside the rti
     // pal code we will get switched to user because of the ICM write
-    xc->getCpuPtr()->kernelStats->mode(Kernel::kernel, xc);
+    if (xc->getKernelStats())
+        xc->getKernelStats()->mode(Kernel::kernel, xc);
 }
diff --git a/sim/pseudo_inst.cc b/sim/pseudo_inst.cc
index 4d9541b58..0c20a6a53 100644
--- a/sim/pseudo_inst.cc
+++ b/sim/pseudo_inst.cc
@@ -65,7 +65,8 @@ namespace AlphaPseudo
     void
     arm(ExecContext *xc)
     {
-        xc->getCpuPtr()->kernelStats->arm();
+        if (xc->getKernelStats())
+            xc->getKernelStats()->arm();
     }
 
     void
@@ -75,7 +76,8 @@ namespace AlphaPseudo
             return;
 
         xc->suspend();
-        xc->getCpuPtr()->kernelStats->quiesce();
+        if (xc->getKernelStats())
+            xc->getKernelStats()->arm();
     }
 
     void
@@ -92,7 +94,8 @@ namespace AlphaPseudo
             quiesceEvent->schedule(curTick + Clock::Int::ns * ns);
 
         xc->suspend();
-        xc->getCpuPtr()->kernelStats->quiesce();
+        if (xc->getKernelStats())
+            xc->getKernelStats()->quiesce();
     }
 
     void
@@ -111,7 +114,8 @@ namespace AlphaPseudo
                                    xc->getCpuPtr()->cycles(cycles));
 
         xc->suspend();
-        xc->getCpuPtr()->kernelStats->quiesce();
+        if (xc->getKernelStats())
+            xc->getKernelStats()->quiesce();
     }
 
     uint64_t
@@ -123,7 +127,8 @@ namespace AlphaPseudo
     void
     ivlb(ExecContext *xc)
     {
-        xc->getCpuPtr()->kernelStats->ivlb();
+        if (xc->getKernelStats())
+            xc->getKernelStats()->ivlb();
     }
 
     void