/*
+ * Copyright (c) 2011 ARM Limited
+ * All rights reserved.
+ *
+ * The license below extends only to copyright in the software and shall
+ * not be construed as granting a license to any other intellectual
+ * property including but not limited to intellectual property relating
+ * to a hardware implementation of the functionality of the software
+ * licensed hereunder. You may use the software subject to the license
+ * terms below provided that you ensure that this notice is replicated
+ * unmodified and in its entirety in all distributions of the software,
+ * modified or unmodified, in source code or in binary form.
+ *
* Copyright (c) 2004-2006 The Regents of The University of Michigan
+ * Copyright (c) 2009 The University of Edinburgh
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* Authors: Kevin Lim
+ * Timothy M. Jones
*/
#ifndef __CPU_BASE_DYN_INST_HH__
#include <bitset>
#include <list>
#include <string>
+#include <queue>
-#include "arch/faults.hh"
-#include "base/fast_alloc.hh"
+#include "arch/utility.hh"
#include "base/trace.hh"
-#include "config/full_system.hh"
+#include "config/the_isa.hh"
+#include "cpu/checker/cpu.hh"
#include "cpu/o3/comm.hh"
#include "cpu/exetrace.hh"
#include "cpu/inst_seq.hh"
#include "cpu/op_class.hh"
#include "cpu/static_inst.hh"
+#include "cpu/translation.hh"
#include "mem/packet.hh"
+#include "sim/byteswap.hh"
+#include "sim/fault_fwd.hh"
#include "sim/system.hh"
+#include "sim/tlb.hh"
/**
* @file
* Defines a dynamic instruction context.
*/
-// Forward declaration.
-class StaticInstPtr;
-
template <class Impl>
-class BaseDynInst : public FastAlloc, public RefCounted
+class BaseDynInst : public RefCounted
{
public:
// Typedef for the CPU.
// The DynInstPtr type.
typedef typename Impl::DynInstPtr DynInstPtr;
+ typedef RefCountingPtr<BaseDynInst<Impl> > BaseDynInstPtr;
// The list of instructions iterator type.
typedef typename std::list<DynInstPtr>::iterator ListIt;
enum {
- MaxInstSrcRegs = TheISA::MaxInstSrcRegs, /// Max source regs
- MaxInstDestRegs = TheISA::MaxInstDestRegs, /// Max dest regs
+ MaxInstSrcRegs = TheISA::MaxInstSrcRegs, /// Max source regs
+ MaxInstDestRegs = TheISA::MaxInstDestRegs /// Max dest regs
};
- /** The StaticInst used by this BaseDynInst. */
- StaticInstPtr staticInst;
-
- ////////////////////////////////////////////
- //
- // INSTRUCTION EXECUTION
- //
- ////////////////////////////////////////////
- /** InstRecord that tracks this instructions. */
- Trace::InstRecord *traceData;
-
- /**
- * Does a read to a given address.
- * @param addr The address to read.
- * @param data The read's data is written into this parameter.
- * @param flags The request's flags.
- * @return Returns any fault due to the read.
- */
- template <class T>
- Fault read(Addr addr, T &data, unsigned flags);
-
- /**
- * Does a write to a given address.
- * @param data The data to be written.
- * @param addr The address to write to.
- * @param flags The request's flags.
- * @param res The result of the write (for load locked/store conditionals).
- * @return Returns any fault due to the write.
- */
- template <class T>
- Fault write(T data, Addr addr, unsigned flags,
- uint64_t *res);
-
- void prefetch(Addr addr, unsigned flags);
- void writeHint(Addr addr, int size, unsigned flags);
- Fault copySrcTranslate(Addr src);
- Fault copy(Addr dest);
-
- /** @todo: Consider making this private. */
- public:
- /** The sequence number of the instruction. */
- InstSeqNum seqNum;
+ union Result {
+ uint64_t integer;
+ double dbl;
+ void set(uint64_t i) { integer = i; }
+ void set(double d) { dbl = d; }
+ void get(uint64_t& i) { i = integer; }
+ void get(double& d) { d = dbl; }
+ };
+ protected:
enum Status {
IqEntry, /// Instruction is in the IQ
RobEntry, /// Instruction is in the ROB
NumStatus
};
- /** The status of this BaseDynInst. Several bits can be set. */
- std::bitset<NumStatus> status;
-
- /** The thread this instruction is from. */
- short threadNumber;
+ enum Flags {
+ TranslationStarted,
+ TranslationCompleted,
+ PossibleLoadViolation,
+ HitExternalSnoop,
+ EffAddrValid,
+ RecordResult,
+ Predicate,
+ PredTaken,
+ /** Whether or not the effective address calculation is completed.
+ * @todo: Consider if this is necessary or not.
+ */
+ EACalcDone,
+ IsUncacheable,
+ ReqMade,
+ MemOpDone,
+ MaxFlags
+ };
- /** data address space ID, for loads & stores. */
- short asid;
+ public:
+ /** The sequence number of the instruction. */
+ InstSeqNum seqNum;
- /** How many source registers are ready. */
- unsigned readyRegs;
+ /** The StaticInst used by this BaseDynInst. */
+ StaticInstPtr staticInst;
/** Pointer to the Impl's CPU object. */
ImplCPU *cpu;
/** The kind of fault this instruction has generated. */
Fault fault;
- /** Pointer to the data for the memory access. */
- uint8_t *memData;
+ /** InstRecord that tracks this instructions. */
+ Trace::InstRecord *traceData;
- /** The effective virtual address (lds & stores only). */
- Addr effAddr;
+ protected:
+ /** The result of the instruction; assumes an instruction can have many
+ * destination registers.
+ */
+ std::queue<Result> instResult;
- /** Is the effective virtual address valid. */
- bool effAddrValid;
+ /** PC state for this instruction. */
+ TheISA::PCState pc;
- /** The effective physical address. */
- Addr physEffAddr;
+ /* An amalgamation of a lot of boolean values into one */
+ std::bitset<MaxFlags> instFlags;
- /** Effective virtual address for a copy source. */
- Addr copySrcEffAddr;
+ /** The status of this BaseDynInst. Several bits can be set. */
+ std::bitset<NumStatus> status;
- /** Effective physical address for a copy source. */
- Addr copySrcPhysEffAddr;
+ /** Whether or not the source register is ready.
+ * @todo: Not sure this should be here vs the derived class.
+ */
+ std::bitset<MaxInstSrcRegs> _readySrcRegIdx;
- /** The memory request flags (from translation). */
- unsigned memReqFlags;
+ public:
+ /** The thread this instruction is from. */
+ ThreadID threadNumber;
- union Result {
- uint64_t integer;
-// float fp;
- double dbl;
- };
+ /** Iterator pointing to this BaseDynInst in the list of all insts. */
+ ListIt instListIt;
- /** The result of the instruction; assumes for now that there's only one
- * destination register.
- */
- Result instResult;
+ ////////////////////// Branch Data ///////////////
+ /** Predicted PC state after this instruction. */
+ TheISA::PCState predPC;
- /** Records changes to result? */
- bool recordResult;
+ /** The Macroop if one exists */
+ StaticInstPtr macroop;
- /** PC of this instruction. */
- Addr PC;
+ /** How many source registers are ready. */
+ uint8_t readyRegs;
- protected:
- /** Next non-speculative PC. It is not filled in at fetch, but rather
- * once the target of the branch is truly known (either decode or
- * execute).
- */
- Addr nextPC;
+ public:
+ /////////////////////// Load Store Data //////////////////////
+ /** The effective virtual address (lds & stores only). */
+ Addr effAddr;
- /** Next non-speculative NPC. Target PC for Mips or Sparc. */
- Addr nextNPC;
+ /** The effective physical address. */
+ Addr physEffAddr;
- /** Predicted next PC. */
- Addr predPC;
+ /** The memory request flags (from translation). */
+ unsigned memReqFlags;
- /** Predicted next NPC. */
- Addr predNPC;
+ /** data address space ID, for loads & stores. */
+ short asid;
- /** If this is a branch that was predicted taken */
- bool predTaken;
+ /** The size of the request */
+ uint8_t effSize;
- public:
+ /** Pointer to the data for the memory access. */
+ uint8_t *memData;
- /** Count of total number of dynamic instructions. */
- static int instcount;
+ /** Load queue index. */
+ int16_t lqIdx;
-#ifdef DEBUG
- void dumpSNList();
-#endif
+ /** Store queue index. */
+ int16_t sqIdx;
- /** Whether or not the source register is ready.
- * @todo: Not sure this should be here vs the derived class.
+
+ /////////////////////// TLB Miss //////////////////////
+ /**
+ * Saved memory requests (needed when the DTB address translation is
+ * delayed due to a hw page table walk).
+ */
+ RequestPtr savedReq;
+ RequestPtr savedSreqLow;
+ RequestPtr savedSreqHigh;
+
+ /////////////////////// Checker //////////////////////
+ // Need a copy of main request pointer to verify on writes.
+ RequestPtr reqToVerify;
+
+ private:
+ /** Instruction effective address.
+ * @todo: Consider if this is necessary or not.
*/
- bool _readySrcRegIdx[MaxInstSrcRegs];
+ Addr instEffAddr;
protected:
/** Flattened register index of the destination registers of this
*/
TheISA::RegIndex _flatDestRegIdx[TheISA::MaxInstDestRegs];
- /** Flattened register index of the source registers of this
- * instruction.
- */
- TheISA::RegIndex _flatSrcRegIdx[TheISA::MaxInstSrcRegs];
-
/** Physical register index of the destination registers of this
* instruction.
*/
*/
PhysRegIndex _prevDestRegIdx[TheISA::MaxInstDestRegs];
+
+ public:
+ /** Records changes to result? */
+ void recordResult(bool f) { instFlags[RecordResult] = f; }
+
+ /** Is the effective virtual address valid. */
+ bool effAddrValid() const { return instFlags[EffAddrValid]; }
+
+ /** Whether or not the memory operation is done. */
+ bool memOpDone() const { return instFlags[MemOpDone]; }
+ void memOpDone(bool f) { instFlags[MemOpDone] = f; }
+
+
+ ////////////////////////////////////////////
+ //
+ // INSTRUCTION EXECUTION
+ //
+ ////////////////////////////////////////////
+
+ void demapPage(Addr vaddr, uint64_t asn)
+ {
+ cpu->demapPage(vaddr, asn);
+ }
+ void demapInstPage(Addr vaddr, uint64_t asn)
+ {
+ cpu->demapPage(vaddr, asn);
+ }
+ void demapDataPage(Addr vaddr, uint64_t asn)
+ {
+ cpu->demapPage(vaddr, asn);
+ }
+
+ Fault readMem(Addr addr, uint8_t *data, unsigned size, unsigned flags);
+
+ Fault writeMem(uint8_t *data, unsigned size,
+ Addr addr, unsigned flags, uint64_t *res);
+
+ /** Splits a request in two if it crosses a dcache block. */
+ void splitRequest(RequestPtr req, RequestPtr &sreqLow,
+ RequestPtr &sreqHigh);
+
+ /** Initiate a DTB address translation. */
+ void initiateTranslation(RequestPtr req, RequestPtr sreqLow,
+ RequestPtr sreqHigh, uint64_t *res,
+ BaseTLB::Mode mode);
+
+ /** Finish a DTB address translation. */
+ void finishTranslation(WholeTranslationState *state);
+
+ /** True if the DTB address translation has started. */
+ bool translationStarted() const { return instFlags[TranslationStarted]; }
+ void translationStarted(bool f) { instFlags[TranslationStarted] = f; }
+
+ /** True if the DTB address translation has completed. */
+ bool translationCompleted() const { return instFlags[TranslationCompleted]; }
+ void translationCompleted(bool f) { instFlags[TranslationCompleted] = f; }
+
+ /** True if this address was found to match a previous load and they issued
+ * out of order. If that happend, then it's only a problem if an incoming
+ * snoop invalidate modifies the line, in which case we need to squash.
+ * If nothing modified the line the order doesn't matter.
+ */
+ bool possibleLoadViolation() const { return instFlags[PossibleLoadViolation]; }
+ void possibleLoadViolation(bool f) { instFlags[PossibleLoadViolation] = f; }
+
+ /** True if the address hit a external snoop while sitting in the LSQ.
+ * If this is true and a older instruction sees it, this instruction must
+ * reexecute
+ */
+ bool hitExternalSnoop() const { return instFlags[HitExternalSnoop]; }
+ void hitExternalSnoop(bool f) { instFlags[HitExternalSnoop] = f; }
+
+ /**
+ * Returns true if the DTB address translation is being delayed due to a hw
+ * page table walk.
+ */
+ bool isTranslationDelayed() const
+ {
+ return (translationStarted() && !translationCompleted());
+ }
+
public:
+#ifdef DEBUG
+ void dumpSNList();
+#endif
/** Returns the physical register index of the i'th destination
* register.
/** Returns the physical register index of the i'th source register. */
PhysRegIndex renamedSrcRegIdx(int idx) const
{
+ assert(TheISA::MaxInstSrcRegs > idx);
return _srcRegIdx[idx];
}
return _flatDestRegIdx[idx];
}
- /** Returns the flattened register index of the i'th source register */
- TheISA::RegIndex flattenedSrcRegIdx(int idx) const
- {
- return _flatSrcRegIdx[idx];
- }
-
/** Returns the physical register index of the previous physical register
* that remapped to the same logical register index.
*/
_srcRegIdx[idx] = renamed_src;
}
- /** Flattens a source architectural register index into a logical index.
- */
- void flattenSrcReg(int idx, TheISA::RegIndex flattened_src)
- {
- _flatSrcRegIdx[idx] = flattened_src;
- }
-
/** Flattens a destination architectural register index into a logical
* index.
*/
{
_flatDestRegIdx[idx] = flattened_dest;
}
-
/** BaseDynInst constructor given a binary instruction.
- * @param inst The binary instruction.
- * @param PC The PC of the instruction.
- * @param pred_PC The predicted next PC.
- * @param pred_NPC The predicted next NPC.
+ * @param staticInst A StaticInstPtr to the underlying instruction.
+ * @param pc The PC state for the instruction.
+ * @param predPC The predicted next PC state for the instruction.
* @param seq_num The sequence number of the instruction.
* @param cpu Pointer to the instruction's CPU.
*/
- BaseDynInst(TheISA::ExtMachInst inst, Addr PC, Addr NPC,
- Addr pred_PC, Addr pred_NPC,
- InstSeqNum seq_num, ImplCPU *cpu);
+ BaseDynInst(StaticInstPtr staticInst, StaticInstPtr macroop,
+ TheISA::PCState pc, TheISA::PCState predPC,
+ InstSeqNum seq_num, ImplCPU *cpu);
/** BaseDynInst constructor given a StaticInst pointer.
* @param _staticInst The StaticInst for this BaseDynInst.
*/
- BaseDynInst(StaticInstPtr &_staticInst);
+ BaseDynInst(StaticInstPtr staticInst, StaticInstPtr macroop);
/** BaseDynInst destructor. */
~BaseDynInst();
void dump(std::string &outstring);
/** Read this CPU's ID. */
- int readCpuId() { return cpu->readCpuId(); }
+ int cpuId() { return cpu->cpuId(); }
+
+ /** Read this CPU's data requestor ID */
+ MasterID masterId() { return cpu->dataMasterId(); }
+
+ /** Read this context's system-wide ID **/
+ int contextId() { return thread->contextId(); }
/** Returns the fault type. */
Fault getFault() { return fault; }
*/
bool doneTargCalc() { return false; }
- /** Returns the next PC. This could be the speculative next PC if it is
- * called prior to the actual branch target being calculated.
- */
- Addr readNextPC() { return nextPC; }
-
- /** Returns the next NPC. This could be the speculative next NPC if it is
- * called prior to the actual branch target being calculated.
- */
- Addr readNextNPC()
- {
-#if ISA_HAS_DELAY_SLOT
- return nextNPC;
-#else
- return nextPC + sizeof(TheISA::MachInst);
-#endif
- }
-
/** Set the predicted target of this current instruction. */
- void setPredTarg(Addr predicted_PC, Addr predicted_NPC)
+ void setPredTarg(const TheISA::PCState &_predPC)
{
- predPC = predicted_PC;
- predNPC = predicted_NPC;
+ predPC = _predPC;
}
+ const TheISA::PCState &readPredTarg() { return predPC; }
+
/** Returns the predicted PC immediately after the branch. */
- Addr readPredPC() { return predPC; }
+ Addr predInstAddr() { return predPC.instAddr(); }
/** Returns the predicted PC two instructions after the branch */
- Addr readPredNPC() { return predNPC; }
+ Addr predNextInstAddr() { return predPC.nextInstAddr(); }
+
+ /** Returns the predicted micro PC after the branch */
+ Addr predMicroPC() { return predPC.microPC(); }
/** Returns whether the instruction was predicted taken or not. */
bool readPredTaken()
{
- return predTaken;
+ return instFlags[PredTaken];
}
void setPredTaken(bool predicted_taken)
{
- predTaken = predicted_taken;
+ instFlags[PredTaken] = predicted_taken;
}
/** Returns whether the instruction mispredicted. */
bool mispredicted()
{
- return readPredPC() != readNextPC() ||
- readPredNPC() != readNextNPC();
+ TheISA::PCState tempPC = pc;
+ TheISA::advancePC(tempPC, staticInst);
+ return !(tempPC == predPC);
}
//
// Instruction types. Forward checks to StaticInst object.
//
- bool isNop() const { return staticInst->isNop(); }
- bool isMemRef() const { return staticInst->isMemRef(); }
- bool isLoad() const { return staticInst->isLoad(); }
- bool isStore() const { return staticInst->isStore(); }
+ bool isNop() const { return staticInst->isNop(); }
+ bool isMemRef() const { return staticInst->isMemRef(); }
+ bool isLoad() const { return staticInst->isLoad(); }
+ bool isStore() const { return staticInst->isStore(); }
bool isStoreConditional() const
{ return staticInst->isStoreConditional(); }
bool isInstPrefetch() const { return staticInst->isInstPrefetch(); }
bool isDataPrefetch() const { return staticInst->isDataPrefetch(); }
- bool isCopy() const { return staticInst->isCopy(); }
- bool isInteger() const { return staticInst->isInteger(); }
- bool isFloating() const { return staticInst->isFloating(); }
- bool isControl() const { return staticInst->isControl(); }
- bool isCall() const { return staticInst->isCall(); }
- bool isReturn() const { return staticInst->isReturn(); }
- bool isDirectCtrl() const { return staticInst->isDirectCtrl(); }
+ bool isInteger() const { return staticInst->isInteger(); }
+ bool isFloating() const { return staticInst->isFloating(); }
+ bool isControl() const { return staticInst->isControl(); }
+ bool isCall() const { return staticInst->isCall(); }
+ bool isReturn() const { return staticInst->isReturn(); }
+ bool isDirectCtrl() const { return staticInst->isDirectCtrl(); }
bool isIndirectCtrl() const { return staticInst->isIndirectCtrl(); }
- bool isCondCtrl() const { return staticInst->isCondCtrl(); }
- bool isUncondCtrl() const { return staticInst->isUncondCtrl(); }
+ bool isCondCtrl() const { return staticInst->isCondCtrl(); }
+ bool isUncondCtrl() const { return staticInst->isUncondCtrl(); }
bool isCondDelaySlot() const { return staticInst->isCondDelaySlot(); }
bool isThreadSync() const { return staticInst->isThreadSync(); }
bool isSerializing() const { return staticInst->isSerializing(); }
{ return staticInst->isSerializeBefore() || status[SerializeBefore]; }
bool isSerializeAfter() const
{ return staticInst->isSerializeAfter() || status[SerializeAfter]; }
+ bool isSquashAfter() const { return staticInst->isSquashAfter(); }
bool isMemBarrier() const { return staticInst->isMemBarrier(); }
bool isWriteBarrier() const { return staticInst->isWriteBarrier(); }
bool isNonSpeculative() const { return staticInst->isNonSpeculative(); }
bool isQuiesce() const { return staticInst->isQuiesce(); }
bool isIprAccess() const { return staticInst->isIprAccess(); }
bool isUnverifiable() const { return staticInst->isUnverifiable(); }
+ bool isSyscall() const { return staticInst->isSyscall(); }
+ bool isMacroop() const { return staticInst->isMacroop(); }
+ bool isMicroop() const { return staticInst->isMicroop(); }
+ bool isDelayedCommit() const { return staticInst->isDelayedCommit(); }
+ bool isLastMicroop() const { return staticInst->isLastMicroop(); }
+ bool isFirstMicroop() const { return staticInst->isFirstMicroop(); }
+ bool isMicroBranch() const { return staticInst->isMicroBranch(); }
/** Temporarily sets this instruction as a serialize before instruction. */
void setSerializeBefore() { status.set(SerializeBefore); }
OpClass opClass() const { return staticInst->opClass(); }
/** Returns the branch target address. */
- Addr branchTarget() const { return staticInst->branchTarget(PC); }
+ TheISA::PCState branchTarget() const
+ { return staticInst->branchTarget(pc); }
/** Returns the number of source registers. */
- int8_t numSrcRegs() const { return staticInst->numSrcRegs(); }
+ int8_t numSrcRegs() const { return staticInst->numSrcRegs(); }
/** Returns the number of destination registers. */
int8_t numDestRegs() const { return staticInst->numDestRegs(); }
/** Returns the logical register index of the i'th source register. */
RegIndex srcRegIdx(int i) const { return staticInst->srcRegIdx(i); }
- /** Returns the result of an integer instruction. */
- uint64_t readIntResult() { return instResult.integer; }
+ /** Pops a result off the instResult queue */
+ template <class T>
+ void popResult(T& t)
+ {
+ if (!instResult.empty()) {
+ instResult.front().get(t);
+ instResult.pop();
+ }
+ }
- /** Returns the result of a floating point instruction. */
- float readFloatResult() { return (float)instResult.dbl; }
+ /** Read the most recent result stored by this instruction */
+ template <class T>
+ void readResult(T& t)
+ {
+ instResult.back().get(t);
+ }
- /** Returns the result of a floating point (double) instruction. */
- double readDoubleResult() { return instResult.dbl; }
+ /** Pushes a result onto the instResult queue */
+ template <class T>
+ void setResult(T t)
+ {
+ if (instFlags[RecordResult]) {
+ Result instRes;
+ instRes.set(t);
+ instResult.push(instRes);
+ }
+ }
/** Records an integer register being set to a value. */
void setIntRegOperand(const StaticInst *si, int idx, uint64_t val)
{
- if (recordResult)
- instResult.integer = val;
+ setResult<uint64_t>(val);
}
/** Records an fp register being set to a value. */
void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val,
int width)
{
- if (recordResult) {
- if (width == 32)
- instResult.dbl = (double)val;
- else if (width == 64)
- instResult.dbl = val;
- else
- panic("Unsupported width!");
+ if (width == 32 || width == 64) {
+ setResult<double>(val);
+ } else {
+ panic("Unsupported width!");
}
}
/** Records an fp register being set to a value. */
void setFloatRegOperand(const StaticInst *si, int idx, FloatReg val)
{
- if (recordResult)
- instResult.dbl = (double)val;
+ setResult<double>(val);
}
/** Records an fp register being set to an integer value. */
void setFloatRegOperandBits(const StaticInst *si, int idx, uint64_t val,
int width)
{
- if (recordResult)
- instResult.integer = val;
+ setResult<uint64_t>(val);
}
/** Records an fp register being set to an integer value. */
void setFloatRegOperandBits(const StaticInst *si, int idx, uint64_t val)
{
- if (recordResult)
- instResult.integer = val;
+ setResult<uint64_t>(val);
}
/** Records that one of the source registers is ready. */
/** Returns whether or not this instruction is squashed in the ROB. */
bool isSquashedInROB() const { return status[SquashedInROB]; }
+ /** Read the PC state of this instruction. */
+ const TheISA::PCState pcState() const { return pc; }
+
+ /** Set the PC state of this instruction. */
+ const void pcState(const TheISA::PCState &val) { pc = val; }
+
/** Read the PC of this instruction. */
- const Addr readPC() const { return PC; }
+ const Addr instAddr() const { return pc.instAddr(); }
+
+ /** Read the PC of the next instruction. */
+ const Addr nextInstAddr() const { return pc.nextInstAddr(); }
- /** Set the next PC of this instruction (its actual target). */
- void setNextPC(uint64_t val)
+ /**Read the micro PC of this instruction. */
+ const Addr microPC() const { return pc.microPC(); }
+
+ bool readPredicate()
{
- nextPC = val;
+ return instFlags[Predicate];
}
- /** Set the next NPC of this instruction (the target in Mips or Sparc).*/
- void setNextNPC(uint64_t val)
+ void setPredicate(bool val)
{
- nextNPC = val;
+ instFlags[Predicate] = val;
+
+ if (traceData) {
+ traceData->setPredicate(val);
+ }
}
/** Sets the ASID. */
void setASID(short addr_space_id) { asid = addr_space_id; }
/** Sets the thread id. */
- void setTid(unsigned tid) { threadNumber = tid; }
+ void setTid(ThreadID tid) { threadNumber = tid; }
/** Sets the pointer to the thread state. */
void setThreadState(ImplState *state) { thread = state; }
/** Returns the thread context. */
ThreadContext *tcBase() { return thread->getTC(); }
- private:
- /** Instruction effective address.
- * @todo: Consider if this is necessary or not.
- */
- Addr instEffAddr;
-
- /** Whether or not the effective address calculation is completed.
- * @todo: Consider if this is necessary or not.
- */
- bool eaCalcDone;
-
- /** Is this instruction's memory access uncacheable. */
- bool isUncacheable;
-
- /** Has this instruction generated a memory request. */
- bool reqMade;
-
public:
/** Sets the effective address. */
- void setEA(Addr &ea) { instEffAddr = ea; eaCalcDone = true; }
+ void setEA(Addr &ea) { instEffAddr = ea; instFlags[EACalcDone] = true; }
/** Returns the effective address. */
const Addr &getEA() const { return instEffAddr; }
/** Returns whether or not the eff. addr. calculation has been completed. */
- bool doneEACalc() { return eaCalcDone; }
+ bool doneEACalc() { return instFlags[EACalcDone]; }
/** Returns whether or not the eff. addr. source registers are ready. */
bool eaSrcsReady();
- /** Whether or not the memory operation is done. */
- bool memOpDone;
-
/** Is this instruction's memory access uncacheable. */
- bool uncacheable() { return isUncacheable; }
+ bool uncacheable() { return instFlags[IsUncacheable]; }
/** Has this instruction generated a memory request. */
- bool hasRequest() { return reqMade; }
-
- public:
- /** Load queue index. */
- int16_t lqIdx;
-
- /** Store queue index. */
- int16_t sqIdx;
-
- /** Iterator pointing to this BaseDynInst in the list of all insts. */
- ListIt instListIt;
+ bool hasRequest() { return instFlags[ReqMade]; }
/** Returns iterator to this instruction in the list of all insts. */
ListIt &getInstListIt() { return instListIt; }
};
template<class Impl>
-template<class T>
-inline Fault
-BaseDynInst<Impl>::read(Addr addr, T &data, unsigned flags)
+Fault
+BaseDynInst<Impl>::readMem(Addr addr, uint8_t *data,
+ unsigned size, unsigned flags)
{
- reqMade = true;
- Request *req = new Request();
- req->setVirt(asid, addr, sizeof(T), flags, this->PC);
- req->setThreadContext(thread->readCpuId(), threadNumber);
+ instFlags[ReqMade] = true;
+ Request *req = NULL;
+ Request *sreqLow = NULL;
+ Request *sreqHigh = NULL;
+
+ if (instFlags[ReqMade] && translationStarted()) {
+ req = savedReq;
+ sreqLow = savedSreqLow;
+ sreqHigh = savedSreqHigh;
+ } else {
+ req = new Request(asid, addr, size, flags, masterId(), this->pc.instAddr(),
+ thread->contextId(), threadNumber);
- if ((req->getVaddr() & (TheISA::VMPageSize - 1)) + req->getSize() >
- TheISA::VMPageSize) {
- delete req;
- return TheISA::genAlignmentFault();
+ // Only split the request if the ISA supports unaligned accesses.
+ if (TheISA::HasUnalignedMemAcc) {
+ splitRequest(req, sreqLow, sreqHigh);
+ }
+ initiateTranslation(req, sreqLow, sreqHigh, NULL, BaseTLB::Read);
}
- fault = cpu->translateDataReadReq(req, thread);
-
- if (req->isUncacheable())
- isUncacheable = true;
-
- if (fault == NoFault) {
- effAddr = req->getVaddr();
- effAddrValid = true;
- physEffAddr = req->getPaddr();
- memReqFlags = req->getFlags();
-
-#if 0
- if (cpu->system->memctrl->badaddr(physEffAddr)) {
- fault = TheISA::genMachineCheckFault();
- data = (T)-1;
- this->setExecuted();
+ if (translationCompleted()) {
+ if (fault == NoFault) {
+ effAddr = req->getVaddr();
+ effSize = size;
+ instFlags[EffAddrValid] = true;
+
+ if (cpu->checker) {
+ if (reqToVerify != NULL) {
+ delete reqToVerify;
+ }
+ reqToVerify = new Request(*req);
+ }
+ fault = cpu->read(req, sreqLow, sreqHigh, data, lqIdx);
} else {
- fault = cpu->read(req, data, lqIdx);
+ // Commit will have to clean up whatever happened. Set this
+ // instruction as executed.
+ this->setExecuted();
}
-#else
- fault = cpu->read(req, data, lqIdx);
-#endif
- } else {
- // Return a fixed value to keep simulation deterministic even
- // along misspeculated paths.
- data = (T)-1;
- // Commit will have to clean up whatever happened. Set this
- // instruction as executed.
- this->setExecuted();
- delete req;
+ if (fault != NoFault) {
+ // Return a fixed value to keep simulation deterministic even
+ // along misspeculated paths.
+ if (data)
+ bzero(data, size);
+ }
}
if (traceData) {
traceData->setAddr(addr);
- traceData->setData(data);
}
return fault;
}
template<class Impl>
-template<class T>
-inline Fault
-BaseDynInst<Impl>::write(T data, Addr addr, unsigned flags, uint64_t *res)
+Fault
+BaseDynInst<Impl>::writeMem(uint8_t *data, unsigned size,
+ Addr addr, unsigned flags, uint64_t *res)
{
if (traceData) {
traceData->setAddr(addr);
- traceData->setData(data);
}
- reqMade = true;
- Request *req = new Request();
- req->setVirt(asid, addr, sizeof(T), flags, this->PC);
- req->setThreadContext(thread->readCpuId(), threadNumber);
+ instFlags[ReqMade] = true;
+ Request *req = NULL;
+ Request *sreqLow = NULL;
+ Request *sreqHigh = NULL;
- if ((req->getVaddr() & (TheISA::VMPageSize - 1)) + req->getSize() >
- TheISA::VMPageSize) {
- delete req;
- return TheISA::genAlignmentFault();
+ if (instFlags[ReqMade] && translationStarted()) {
+ req = savedReq;
+ sreqLow = savedSreqLow;
+ sreqHigh = savedSreqHigh;
+ } else {
+ req = new Request(asid, addr, size, flags, masterId(), this->pc.instAddr(),
+ thread->contextId(), threadNumber);
+
+ // Only split the request if the ISA supports unaligned accesses.
+ if (TheISA::HasUnalignedMemAcc) {
+ splitRequest(req, sreqLow, sreqHigh);
+ }
+ initiateTranslation(req, sreqLow, sreqHigh, res, BaseTLB::Write);
+ }
+
+ if (fault == NoFault && translationCompleted()) {
+ effAddr = req->getVaddr();
+ effSize = size;
+ instFlags[EffAddrValid] = true;
+
+ if (cpu->checker) {
+ if (reqToVerify != NULL) {
+ delete reqToVerify;
+ }
+ reqToVerify = new Request(*req);
+ }
+ fault = cpu->write(req, sreqLow, sreqHigh, data, sqIdx);
}
- fault = cpu->translateDataWriteReq(req, thread);
+ return fault;
+}
- if (req->isUncacheable())
- isUncacheable = true;
+template<class Impl>
+inline void
+BaseDynInst<Impl>::splitRequest(RequestPtr req, RequestPtr &sreqLow,
+ RequestPtr &sreqHigh)
+{
+ // Check to see if the request crosses the next level block boundary.
+ unsigned block_size = cpu->getDataPort().peerBlockSize();
+ Addr addr = req->getVaddr();
+ Addr split_addr = roundDown(addr + req->getSize() - 1, block_size);
+ assert(split_addr <= addr || split_addr - addr < block_size);
+
+ // Spans two blocks.
+ if (split_addr > addr) {
+ req->splitOnVaddr(split_addr, sreqLow, sreqHigh);
+ }
+}
+
+template<class Impl>
+inline void
+BaseDynInst<Impl>::initiateTranslation(RequestPtr req, RequestPtr sreqLow,
+ RequestPtr sreqHigh, uint64_t *res,
+ BaseTLB::Mode mode)
+{
+ translationStarted(true);
+
+ if (!TheISA::HasUnalignedMemAcc || sreqLow == NULL) {
+ WholeTranslationState *state =
+ new WholeTranslationState(req, NULL, res, mode);
+
+ // One translation if the request isn't split.
+ DataTranslation<BaseDynInstPtr> *trans =
+ new DataTranslation<BaseDynInstPtr>(this, state);
+ cpu->dtb->translateTiming(req, thread->getTC(), trans, mode);
+ if (!translationCompleted()) {
+ // Save memory requests.
+ savedReq = state->mainReq;
+ savedSreqLow = state->sreqLow;
+ savedSreqHigh = state->sreqHigh;
+ }
+ } else {
+ WholeTranslationState *state =
+ new WholeTranslationState(req, sreqLow, sreqHigh, NULL, res, mode);
+
+ // Two translations when the request is split.
+ DataTranslation<BaseDynInstPtr> *stransLow =
+ new DataTranslation<BaseDynInstPtr>(this, state, 0);
+ DataTranslation<BaseDynInstPtr> *stransHigh =
+ new DataTranslation<BaseDynInstPtr>(this, state, 1);
+
+ cpu->dtb->translateTiming(sreqLow, thread->getTC(), stransLow, mode);
+ cpu->dtb->translateTiming(sreqHigh, thread->getTC(), stransHigh, mode);
+ if (!translationCompleted()) {
+ // Save memory requests.
+ savedReq = state->mainReq;
+ savedSreqLow = state->sreqLow;
+ savedSreqHigh = state->sreqHigh;
+ }
+ }
+}
+
+template<class Impl>
+inline void
+BaseDynInst<Impl>::finishTranslation(WholeTranslationState *state)
+{
+ fault = state->getFault();
+
+ instFlags[IsUncacheable] = state->isUncacheable();
if (fault == NoFault) {
- effAddr = req->getVaddr();
- effAddrValid = true;
- physEffAddr = req->getPaddr();
- memReqFlags = req->getFlags();
-#if 0
- if (cpu->system->memctrl->badaddr(physEffAddr)) {
- fault = TheISA::genMachineCheckFault();
- } else {
- fault = cpu->write(req, data, sqIdx);
+ physEffAddr = state->getPaddr();
+ memReqFlags = state->getFlags();
+
+ if (state->mainReq->isCondSwap()) {
+ assert(state->res);
+ state->mainReq->setExtraData(*state->res);
}
-#else
- fault = cpu->write(req, data, sqIdx);
-#endif
+
} else {
- delete req;
+ state->deleteReqs();
}
+ delete state;
- return fault;
+ translationCompleted(true);
}
#endif // __CPU_BASE_DYN_INST_HH__