This patch changes how faults are passed between methods in an attempt to copy as few reference-counting pointer instances as possible. This should avoid unecessary copies being created, contributing to the increment/decrement of the reference counters.
336 lines
11 KiB
C++
336 lines
11 KiB
C++
/*
|
|
* Copyright (c) 2007 MIPS Technologies, Inc.
|
|
* 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.
|
|
*
|
|
* Authors: Korey Sewell
|
|
*
|
|
*/
|
|
|
|
#include "config/the_isa.hh"
|
|
#include "cpu/inorder/resources/fetch_seq_unit.hh"
|
|
#include "cpu/inorder/resource_pool.hh"
|
|
#include "debug/InOrderFetchSeq.hh"
|
|
#include "debug/InOrderStall.hh"
|
|
|
|
using namespace std;
|
|
using namespace TheISA;
|
|
using namespace ThePipeline;
|
|
|
|
FetchSeqUnit::FetchSeqUnit(std::string res_name, int res_id, int res_width,
|
|
Cycles res_latency, InOrderCPU *_cpu,
|
|
ThePipeline::Params *params)
|
|
: Resource(res_name, res_id, res_width, res_latency, _cpu),
|
|
instSize(sizeof(MachInst))
|
|
{
|
|
for (ThreadID tid = 0; tid < ThePipeline::MaxThreads; tid++) {
|
|
pcValid[tid] = false;
|
|
pcBlockStage[tid] = 0;
|
|
|
|
//@todo: Use CPU's squashSeqNum here instead of maintaining our own
|
|
// state
|
|
squashSeqNum[tid] = (InstSeqNum)-1;
|
|
lastSquashCycle[tid] = 0;
|
|
}
|
|
}
|
|
|
|
FetchSeqUnit::~FetchSeqUnit()
|
|
{
|
|
delete [] resourceEvent;
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::init()
|
|
{
|
|
resourceEvent = new FetchSeqEvent[width];
|
|
|
|
for (int i = 0; i < width; i++) {
|
|
reqs[i] = new ResourceRequest(this);
|
|
}
|
|
|
|
initSlots();
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::execute(int slot_num)
|
|
{
|
|
ResourceRequest* fs_req = reqs[slot_num];
|
|
DynInstPtr inst = fs_req->inst;
|
|
ThreadID tid = inst->readTid();
|
|
int stage_num = fs_req->getStageNum();
|
|
|
|
if (inst->fault != NoFault) {
|
|
DPRINTF(InOrderFetchSeq,
|
|
"[tid:%i]: [sn:%i]: Detected %s fault @ %x. Forwarding to "
|
|
"next stage.\n", tid, inst->seqNum, inst->fault->name(),
|
|
inst->pcState());
|
|
fs_req->done();
|
|
return;
|
|
}
|
|
|
|
switch (fs_req->cmd)
|
|
{
|
|
case AssignNextPC:
|
|
{
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Current PC is %s\n", tid,
|
|
pc[tid]);
|
|
|
|
if (pcValid[tid]) {
|
|
inst->pcState(pc[tid]);
|
|
inst->setMemAddr(pc[tid].instAddr());
|
|
|
|
// Advance to next PC (typically PC + 4)
|
|
pc[tid].advance();
|
|
|
|
inst->setSeqNum(cpu->getAndIncrementInstSeq(tid));
|
|
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Assigning [sn:%i] to "
|
|
"PC %s\n", tid, inst->seqNum, inst->pcState());
|
|
|
|
fs_req->done();
|
|
} else {
|
|
DPRINTF(InOrderStall, "STALL: [tid:%i]: NPC not valid\n", tid);
|
|
fs_req->done(false);
|
|
}
|
|
}
|
|
break;
|
|
|
|
case UpdateTargetPC:
|
|
{
|
|
assert(!inst->isCondDelaySlot() &&
|
|
"Not Handling Conditional Delay Slot");
|
|
|
|
if (inst->isControl()) {
|
|
if (inst->isReturn() && !inst->predTaken()) {
|
|
// If it's a return, then we must wait for resolved address.
|
|
// The Predictor will mark a return a false as "not taken"
|
|
// if there is no RAS entry
|
|
DPRINTF(InOrderFetchSeq, "[tid:%d]: Setting block signal "
|
|
"for stage %i.\n",
|
|
tid, stage_num);
|
|
cpu->pipelineStage[stage_num]->
|
|
toPrevStages->stageBlock[stage_num][tid] = true;
|
|
pcValid[tid] = false;
|
|
pcBlockStage[tid] = stage_num;
|
|
} else if (inst->predTaken()) {
|
|
// Taken Control
|
|
inst->setSquashInfo(stage_num);
|
|
setupSquash(inst, stage_num, tid);
|
|
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i] Setting up squash to "
|
|
"start from stage %i, after [sn:%i].\n",
|
|
tid, stage_num, inst->squashSeqNum);
|
|
}
|
|
} else {
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: [sn:%i]: Ignoring branch "
|
|
"target update since then is not a control "
|
|
"instruction.\n", tid, inst->seqNum);
|
|
}
|
|
|
|
fs_req->done();
|
|
}
|
|
break;
|
|
|
|
default:
|
|
fatal("Unrecognized command to %s", resName);
|
|
}
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::squash(DynInstPtr inst, int squash_stage,
|
|
InstSeqNum squash_seq_num, ThreadID tid)
|
|
{
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Updating due to squash from %s (%s) "
|
|
"stage %i.\n", tid, inst->instName(), inst->pcState(),
|
|
squash_stage);
|
|
|
|
if (lastSquashCycle[tid] == curTick() &&
|
|
squashSeqNum[tid] <= squash_seq_num) {
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Ignoring squash from stage %i, "
|
|
"since there is an outstanding squash that is older.\n",
|
|
tid, squash_stage);
|
|
} else {
|
|
squashSeqNum[tid] = squash_seq_num;
|
|
lastSquashCycle[tid] = curTick();
|
|
|
|
if (inst->staticInst) {
|
|
if (inst->fault != NoFault) {
|
|
// A Trap Caused This Fault and will update the pc state
|
|
// when done trapping
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i] Blocking due to fault @ "
|
|
"[sn:%i].%s %s \n", tid, inst->seqNum,
|
|
inst->instName(), inst->pcState());
|
|
pcValid[tid] = false;
|
|
} else {
|
|
TheISA::PCState nextPC;
|
|
assert(inst->staticInst);
|
|
if (inst->isControl()) {
|
|
nextPC = inst->readPredTarg();
|
|
|
|
// If we are already fetching this PC then advance to next PC
|
|
// =======
|
|
// This should handle ISAs w/delay slots and annulled delay
|
|
// slots to figure out which is the next PC to fetch after
|
|
// a mispredict
|
|
DynInstPtr bdelay_inst = NULL;
|
|
ListIt bdelay_it;
|
|
if (inst->onInstList) {
|
|
bdelay_it = inst->getInstListIt();
|
|
bdelay_it++;
|
|
} else {
|
|
InstSeqNum branch_delay_num = inst->seqNum + 1;
|
|
bdelay_it = cpu->findInst(branch_delay_num, tid);
|
|
}
|
|
|
|
if (bdelay_it != cpu->instList[tid].end()) {
|
|
bdelay_inst = (*bdelay_it);
|
|
}
|
|
|
|
if (bdelay_inst) {
|
|
if (bdelay_inst->pc.instAddr() == nextPC.instAddr()) {
|
|
bdelay_inst->pc = nextPC;
|
|
advancePC(nextPC, inst->staticInst);
|
|
DPRINTF(InOrderFetchSeq, "Advanced PC to %s\n", nextPC);
|
|
}
|
|
}
|
|
} else {
|
|
nextPC = inst->pcState();
|
|
advancePC(nextPC, inst->staticInst);
|
|
}
|
|
|
|
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Setting PC to %s.\n",
|
|
tid, nextPC);
|
|
pc[tid] = nextPC;
|
|
|
|
// Unblock Any Stages Waiting for this information to be updated ...
|
|
if (!pcValid[tid]) {
|
|
DPRINTF(InOrderFetchSeq, "[tid:%d]: Setting unblock signal "
|
|
"for stage %i.\n",
|
|
tid, pcBlockStage[tid]);
|
|
|
|
// Need to use "fromNextStages" instead of "toPrevStages"
|
|
// because the timebuffer will have already have advanced
|
|
// in the tick function and this squash function will happen after
|
|
// the tick
|
|
cpu->pipelineStage[pcBlockStage[tid]]->
|
|
fromNextStages->stageUnblock[pcBlockStage[tid]][tid] = true;
|
|
}
|
|
|
|
pcValid[tid] = true;
|
|
}
|
|
}
|
|
}
|
|
|
|
Resource::squash(inst, squash_stage, squash_seq_num, tid);
|
|
}
|
|
|
|
FetchSeqUnit::FetchSeqEvent::FetchSeqEvent()
|
|
: ResourceEvent()
|
|
{ }
|
|
|
|
void
|
|
FetchSeqUnit::FetchSeqEvent::process()
|
|
{
|
|
FetchSeqUnit* fs_res = dynamic_cast<FetchSeqUnit*>(resource);
|
|
assert(fs_res);
|
|
|
|
for (int i = 0; i < MaxThreads; i++) {
|
|
fs_res->pc[i] = fs_res->cpu->pcState(i);
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Setting PC: %s.\n",
|
|
fs_res->pc[i]);
|
|
|
|
fs_res->pcValid[i] = true;
|
|
}
|
|
}
|
|
|
|
|
|
void
|
|
FetchSeqUnit::activateThread(ThreadID tid)
|
|
{
|
|
pcValid[tid] = true;
|
|
|
|
pc[tid] = cpu->pcState(tid);
|
|
|
|
cpu->fetchPriorityList.push_back(tid);
|
|
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Reading PC: %s.\n",
|
|
tid, pc[tid]);
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::deactivateThread(ThreadID tid)
|
|
{
|
|
pcValid[tid] = false;
|
|
pcBlockStage[tid] = 0;
|
|
|
|
squashSeqNum[tid] = (InstSeqNum)-1;
|
|
lastSquashCycle[tid] = 0;
|
|
|
|
list<ThreadID>::iterator thread_it = find(cpu->fetchPriorityList.begin(),
|
|
cpu->fetchPriorityList.end(),
|
|
tid);
|
|
|
|
if (thread_it != cpu->fetchPriorityList.end())
|
|
cpu->fetchPriorityList.erase(thread_it);
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::suspendThread(ThreadID tid)
|
|
{
|
|
deactivateThread(tid);
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::trap(const Fault &fault, ThreadID tid, DynInstPtr inst)
|
|
{
|
|
pcValid[tid] = true;
|
|
pc[tid] = cpu->pcState(tid);
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Trap updating to PC: "
|
|
"%s.\n", tid, pc[tid]);
|
|
}
|
|
|
|
void
|
|
FetchSeqUnit::updateAfterContextSwitch(DynInstPtr inst, ThreadID tid)
|
|
{
|
|
pcValid[tid] = true;
|
|
|
|
if (cpu->thread[tid]->lastGradIsBranch) {
|
|
/** This function assumes that the instruction causing the context
|
|
* switch was right after the branch. Thus, if it's not, then
|
|
* we are updating incorrectly here
|
|
*/
|
|
assert(cpu->nextInstAddr(tid) == inst->instAddr());
|
|
pc[tid] = cpu->thread[tid]->lastBranchPC;
|
|
} else {
|
|
pc[tid] = inst->pcState();
|
|
}
|
|
assert(inst->staticInst);
|
|
advancePC(pc[tid], inst->staticInst);
|
|
|
|
DPRINTF(InOrderFetchSeq, "[tid:%i]: Updating PCs due to Context Switch."
|
|
"Assigning PC: %s.\n", tid, pc[tid]);
|
|
}
|