Files
gem5/src/cpu/base_dyn_inst_impl.hh
Gabe Black 344ea0330a cpu: Replace fixed sized arrays in the O3 inst with variable arrays.
The only way to allocate fixed sized arrays which will definitely be big
enough for all source/destination registers for a given instruction is
to track the maximum number of each at compile time, and then size the
arrays appropriately. That creates a point of centralization which
prevents breaking up decoder and instruction definitions into more
modular pieces, and if multiple ISAs are ever built at once, would
require coordination between all ISAs, and wasting memory for most of
them.

The dynamic allocation overhead is minimized by allocating the storage
for all variable arrays in one chunk, and then placing the arrays there
using placement new. There is still some overhead, although less than it
might be otherwise.

Change-Id: Id2c42869cba944deb97da01ca9e0e70186e22532
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/38384
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Maintainer: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
2021-02-05 03:04:56 +00:00

271 lines
7.2 KiB
C++

/*
* Copyright (c) 2011, 2019 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
* 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_BASE_DYN_INST_IMPL_HH__
#define __CPU_BASE_DYN_INST_IMPL_HH__
#include <iostream>
#include <set>
#include <sstream>
#include <string>
#include "base/cprintf.hh"
#include "base/trace.hh"
#include "config/the_isa.hh"
#include "cpu/base_dyn_inst.hh"
#include "cpu/exetrace.hh"
#include "debug/DynInst.hh"
#include "debug/IQ.hh"
#include "mem/request.hh"
#include "sim/faults.hh"
template <class Impl>
BaseDynInst<Impl>::BaseDynInst(const StaticInstPtr &_staticInst,
const StaticInstPtr &_macroop,
TheISA::PCState _pc, TheISA::PCState _predPC,
InstSeqNum seq_num, ImplCPU *cpu)
: staticInst(_staticInst), cpu(cpu),
thread(nullptr),
traceData(nullptr),
regs(staticInst->numSrcRegs(), staticInst->numDestRegs()),
macroop(_macroop),
memData(nullptr),
savedReq(nullptr),
reqToVerify(nullptr)
{
seqNum = seq_num;
pc = _pc;
predPC = _predPC;
initVars();
}
template <class Impl>
BaseDynInst<Impl>::BaseDynInst(const StaticInstPtr &_staticInst,
const StaticInstPtr &_macroop)
: staticInst(_staticInst), traceData(NULL),
regs(staticInst->numSrcRegs(), staticInst->numDestRegs()),
macroop(_macroop)
{
seqNum = 0;
initVars();
}
template <class Impl>
void
BaseDynInst<Impl>::initVars()
{
memData = NULL;
effAddr = 0;
physEffAddr = 0;
readyRegs = 0;
memReqFlags = 0;
// hardware transactional memory
htmUid = -1;
htmDepth = 0;
status.reset();
instFlags.reset();
instFlags[RecordResult] = true;
instFlags[Predicate] = true;
instFlags[MemAccPredicate] = true;
lqIdx = -1;
sqIdx = -1;
// Eventually make this a parameter.
threadNumber = 0;
// Initialize the fault to be NoFault.
fault = NoFault;
#ifndef NDEBUG
++cpu->instcount;
if (cpu->instcount > 1500) {
#ifdef DEBUG
cpu->dumpInsts();
dumpSNList();
#endif
assert(cpu->instcount <= 1500);
}
DPRINTF(DynInst,
"DynInst: [sn:%lli] Instruction created. Instcount for %s = %i\n",
seqNum, cpu->name(), cpu->instcount);
#endif
#ifdef DEBUG
cpu->snList.insert(seqNum);
#endif
}
template <class Impl>
BaseDynInst<Impl>::~BaseDynInst()
{
if (memData) {
delete [] memData;
}
if (traceData) {
delete traceData;
}
fault = NoFault;
#ifndef NDEBUG
--cpu->instcount;
DPRINTF(DynInst,
"DynInst: [sn:%lli] Instruction destroyed. Instcount for %s = %i\n",
seqNum, cpu->name(), cpu->instcount);
#endif
#ifdef DEBUG
cpu->snList.erase(seqNum);
#endif
}
#ifdef DEBUG
template <class Impl>
void
BaseDynInst<Impl>::dumpSNList()
{
std::set<InstSeqNum>::iterator sn_it = cpu->snList.begin();
int count = 0;
while (sn_it != cpu->snList.end()) {
cprintf("%i: [sn:%lli] not destroyed\n", count, (*sn_it));
count++;
sn_it++;
}
}
#endif
template <class Impl>
void
BaseDynInst<Impl>::dump()
{
cprintf("T%d : %#08d `", threadNumber, pc.instAddr());
std::cout << staticInst->disassemble(pc.instAddr());
cprintf("'\n");
}
template <class Impl>
void
BaseDynInst<Impl>::dump(std::string &outstring)
{
std::ostringstream s;
s << "T" << threadNumber << " : 0x" << pc.instAddr() << " "
<< staticInst->disassemble(pc.instAddr());
outstring = s.str();
}
template <class Impl>
void
BaseDynInst<Impl>::markSrcRegReady()
{
DPRINTF(IQ, "[sn:%lli] has %d ready out of %d sources. RTI %d)\n",
seqNum, readyRegs+1, numSrcRegs(), readyToIssue());
if (++readyRegs == numSrcRegs()) {
setCanIssue();
}
}
template <class Impl>
void
BaseDynInst<Impl>::markSrcRegReady(RegIndex src_idx)
{
regs.readySrcIdx(src_idx, true);
markSrcRegReady();
}
template <class Impl>
bool
BaseDynInst<Impl>::eaSrcsReady() const
{
// For now I am assuming that src registers 1..n-1 are the ones that the
// EA calc depends on. (i.e. src reg 0 is the source of the data to be
// stored)
for (int i = 1; i < numSrcRegs(); ++i) {
if (!regs.readySrcIdx(i))
return false;
}
return true;
}
template <class Impl>
void
BaseDynInst<Impl>::setSquashed()
{
status.set(Squashed);
if (!isPinnedRegsRenamed() || isPinnedRegsSquashDone())
return;
// This inst has been renamed already so it may go through rename
// again (e.g. if the squash is due to memory access order violation).
// Reset the write counters for all pinned destination register to ensure
// that they are in a consistent state for a possible re-rename. This also
// ensures that dest regs will be pinned to the same phys register if
// re-rename happens.
for (int idx = 0; idx < numDestRegs(); idx++) {
PhysRegIdPtr phys_dest_reg = regs.renamedDestIdx(idx);
if (phys_dest_reg->isPinned()) {
phys_dest_reg->incrNumPinnedWrites();
if (isPinnedRegsWritten())
phys_dest_reg->incrNumPinnedWritesToComplete();
}
}
setPinnedRegsSquashDone();
}
#endif//__CPU_BASE_DYN_INST_IMPL_HH__