2005-05-27 05:30:12 +02:00
|
|
|
/*
|
2013-03-26 19:46:42 +01:00
|
|
|
* Copyright (c) 2011-2013 ARM Limited
|
2013-10-15 20:22:44 +02:00
|
|
|
* Copyright (c) 2013 Advanced Micro Devices, Inc.
|
2012-01-17 19:55:08 +01:00
|
|
|
* 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.
|
|
|
|
*
|
2005-05-27 05:30:12 +02:00
|
|
|
* Copyright (c) 2004-2005 The Regents of The University of Michigan
|
2011-02-07 07:14:17 +01:00
|
|
|
* Copyright (c) 2011 Regents of the University of California
|
2005-05-27 05:30:12 +02:00
|
|
|
* 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.
|
2006-06-01 01:26:56 +02:00
|
|
|
*
|
|
|
|
* Authors: Kevin Lim
|
2006-06-16 04:01:28 +02:00
|
|
|
* Korey Sewell
|
2011-02-07 07:14:17 +01:00
|
|
|
* Rick Strong
|
2005-05-27 05:30:12 +02:00
|
|
|
*/
|
|
|
|
|
2006-05-19 21:37:52 +02:00
|
|
|
#ifndef __CPU_O3_CPU_HH__
|
|
|
|
#define __CPU_O3_CPU_HH__
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
#include <iostream>
|
|
|
|
#include <list>
|
2006-04-23 00:26:48 +02:00
|
|
|
#include <queue>
|
|
|
|
#include <set>
|
2005-05-03 16:56:47 +02:00
|
|
|
#include <vector>
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-08-15 11:07:15 +02:00
|
|
|
#include "arch/types.hh"
|
2004-08-20 20:54:07 +02:00
|
|
|
#include "base/statistics.hh"
|
2009-09-23 17:34:21 +02:00
|
|
|
#include "config/the_isa.hh"
|
2005-06-05 02:50:10 +02:00
|
|
|
#include "cpu/o3/comm.hh"
|
|
|
|
#include "cpu/o3/cpu_policy.hh"
|
2006-04-23 00:26:48 +02:00
|
|
|
#include "cpu/o3/scoreboard.hh"
|
|
|
|
#include "cpu/o3/thread_state.hh"
|
2011-04-15 19:44:06 +02:00
|
|
|
#include "cpu/activity.hh"
|
|
|
|
#include "cpu/base.hh"
|
|
|
|
#include "cpu/simple_thread.hh"
|
|
|
|
#include "cpu/timebuf.hh"
|
2006-07-01 01:52:08 +02:00
|
|
|
//#include "cpu/o3/thread_context.hh"
|
2008-08-11 21:22:16 +02:00
|
|
|
#include "params/DerivO3CPU.hh"
|
2011-04-15 19:44:06 +02:00
|
|
|
#include "sim/process.hh"
|
2008-08-11 21:22:16 +02:00
|
|
|
|
2006-05-16 20:06:35 +02:00
|
|
|
template <class>
|
|
|
|
class Checker;
|
2006-06-06 23:32:21 +02:00
|
|
|
class ThreadContext;
|
2006-07-01 01:52:08 +02:00
|
|
|
template <class>
|
|
|
|
class O3ThreadContext;
|
2006-07-06 19:59:02 +02:00
|
|
|
|
|
|
|
class Checkpoint;
|
2006-06-03 00:15:20 +02:00
|
|
|
class MemObject;
|
2004-08-20 20:54:07 +02:00
|
|
|
class Process;
|
|
|
|
|
2012-01-31 18:05:52 +01:00
|
|
|
struct BaseCPUParams;
|
2008-08-11 21:22:16 +02:00
|
|
|
|
2006-06-16 23:08:47 +02:00
|
|
|
class BaseO3CPU : public BaseCPU
|
2004-08-20 20:54:07 +02:00
|
|
|
{
|
|
|
|
//Stuff that's pretty ISA independent will go here.
|
|
|
|
public:
|
2008-08-11 21:22:16 +02:00
|
|
|
BaseO3CPU(BaseCPUParams *params);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
void regStats();
|
2004-08-20 20:54:07 +02:00
|
|
|
};
|
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/**
|
|
|
|
* FullO3CPU class, has each of the stages (fetch through commit)
|
|
|
|
* within it, as well as all of the time buffers between stages. The
|
|
|
|
* tick() function for the CPU is defined here.
|
|
|
|
*/
|
2004-08-20 20:54:07 +02:00
|
|
|
template <class Impl>
|
2006-06-16 23:08:47 +02:00
|
|
|
class FullO3CPU : public BaseO3CPU
|
2004-08-20 20:54:07 +02:00
|
|
|
{
|
|
|
|
public:
|
2006-05-19 21:37:52 +02:00
|
|
|
// Typedefs from the Impl here.
|
2004-08-20 20:54:07 +02:00
|
|
|
typedef typename Impl::CPUPol CPUPolicy;
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename Impl::DynInstPtr DynInstPtr;
|
2007-04-04 21:38:59 +02:00
|
|
|
typedef typename Impl::O3CPU O3CPU;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2008-10-09 09:08:50 +02:00
|
|
|
typedef O3ThreadState<Impl> ImplState;
|
2006-04-23 00:26:48 +02:00
|
|
|
typedef O3ThreadState<Impl> Thread;
|
|
|
|
|
|
|
|
typedef typename std::list<DynInstPtr>::iterator ListIt;
|
|
|
|
|
2006-07-01 01:52:08 +02:00
|
|
|
friend class O3ThreadContext<Impl>;
|
2006-07-02 00:52:02 +02:00
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
public:
|
|
|
|
enum Status {
|
|
|
|
Running,
|
|
|
|
Idle,
|
|
|
|
Halted,
|
2006-05-04 17:36:20 +02:00
|
|
|
Blocked,
|
|
|
|
SwitchedOut
|
2004-08-20 20:54:07 +02:00
|
|
|
};
|
|
|
|
|
2009-04-09 07:21:27 +02:00
|
|
|
TheISA::TLB * itb;
|
|
|
|
TheISA::TLB * dtb;
|
2006-12-06 17:36:40 +01:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Overall CPU status. */
|
2004-08-20 20:54:07 +02:00
|
|
|
Status _status;
|
|
|
|
|
|
|
|
private:
|
2012-01-17 19:55:08 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* IcachePort class for instruction fetch.
|
|
|
|
*/
|
2013-03-26 19:46:42 +01:00
|
|
|
class IcachePort : public MasterPort
|
2012-01-17 19:55:08 +01:00
|
|
|
{
|
|
|
|
protected:
|
|
|
|
/** Pointer to fetch. */
|
|
|
|
DefaultFetch<Impl> *fetch;
|
|
|
|
|
|
|
|
public:
|
|
|
|
/** Default constructor. */
|
|
|
|
IcachePort(DefaultFetch<Impl> *_fetch, FullO3CPU<Impl>* _cpu)
|
2013-03-26 19:46:42 +01:00
|
|
|
: MasterPort(_cpu->name() + ".icache_port", _cpu), fetch(_fetch)
|
2012-01-17 19:55:08 +01:00
|
|
|
{ }
|
|
|
|
|
|
|
|
protected:
|
|
|
|
|
|
|
|
/** Timing version of receive. Handles setting fetch to the
|
|
|
|
* proper status to start fetching. */
|
MEM: Separate requests and responses for timing accesses
This patch moves send/recvTiming and send/recvTimingSnoop from the
Port base class to the MasterPort and SlavePort, and also splits them
into separate member functions for requests and responses:
send/recvTimingReq, send/recvTimingResp, and send/recvTimingSnoopReq,
send/recvTimingSnoopResp. A master port sends requests and receives
responses, and also receives snoop requests and sends snoop
responses. A slave port has the reciprocal behaviour as it receives
requests and sends responses, and sends snoop requests and receives
snoop responses.
For all MemObjects that have only master ports or slave ports (but not
both), e.g. a CPU, or a PIO device, this patch merely adds more
clarity to what kind of access is taking place. For example, a CPU
port used to call sendTiming, and will now call
sendTimingReq. Similarly, a response previously came back through
recvTiming, which is now recvTimingResp. For the modules that have
both master and slave ports, e.g. the bus, the behaviour was
previously relying on branches based on pkt->isRequest(), and this is
now replaced with a direct call to the apprioriate member function
depending on the type of access. Please note that send/recvRetry is
still shared by all the timing accessors and remains in the Port base
class for now (to maintain the current bus functionality and avoid
changing the statistics of all regressions).
The packet queue is split into a MasterPort and SlavePort version to
facilitate the use of the new timing accessors. All uses of the
PacketQueue are updated accordingly.
With this patch, the type of packet (request or response) is now well
defined for each type of access, and asserts on pkt->isRequest() and
pkt->isResponse() are now moved to the appropriate send member
functions. It is also worth noting that sendTimingSnoopReq no longer
returns a boolean, as the semantics do not alow snoop requests to be
rejected or stalled. All these assumptions are now excplicitly part of
the port interface itself.
2012-05-01 19:40:42 +02:00
|
|
|
virtual bool recvTimingResp(PacketPtr pkt);
|
|
|
|
virtual void recvTimingSnoopReq(PacketPtr pkt) { }
|
2012-01-17 19:55:08 +01:00
|
|
|
|
|
|
|
/** Handles doing a retry of a failed fetch. */
|
|
|
|
virtual void recvRetry();
|
|
|
|
};
|
|
|
|
|
|
|
|
/**
|
|
|
|
* DcachePort class for the load/store queue.
|
|
|
|
*/
|
2013-03-26 19:46:42 +01:00
|
|
|
class DcachePort : public MasterPort
|
2012-01-17 19:55:08 +01:00
|
|
|
{
|
|
|
|
protected:
|
|
|
|
|
|
|
|
/** Pointer to LSQ. */
|
|
|
|
LSQ<Impl> *lsq;
|
2014-11-06 12:42:22 +01:00
|
|
|
FullO3CPU<Impl> *cpu;
|
2012-01-17 19:55:08 +01:00
|
|
|
|
|
|
|
public:
|
|
|
|
/** Default constructor. */
|
|
|
|
DcachePort(LSQ<Impl> *_lsq, FullO3CPU<Impl>* _cpu)
|
2014-11-06 12:42:22 +01:00
|
|
|
: MasterPort(_cpu->name() + ".dcache_port", _cpu), lsq(_lsq),
|
|
|
|
cpu(_cpu)
|
2012-01-17 19:55:08 +01:00
|
|
|
{ }
|
|
|
|
|
|
|
|
protected:
|
|
|
|
|
|
|
|
/** Timing version of receive. Handles writing back and
|
|
|
|
* completing the load or store that has returned from
|
|
|
|
* memory. */
|
MEM: Separate requests and responses for timing accesses
This patch moves send/recvTiming and send/recvTimingSnoop from the
Port base class to the MasterPort and SlavePort, and also splits them
into separate member functions for requests and responses:
send/recvTimingReq, send/recvTimingResp, and send/recvTimingSnoopReq,
send/recvTimingSnoopResp. A master port sends requests and receives
responses, and also receives snoop requests and sends snoop
responses. A slave port has the reciprocal behaviour as it receives
requests and sends responses, and sends snoop requests and receives
snoop responses.
For all MemObjects that have only master ports or slave ports (but not
both), e.g. a CPU, or a PIO device, this patch merely adds more
clarity to what kind of access is taking place. For example, a CPU
port used to call sendTiming, and will now call
sendTimingReq. Similarly, a response previously came back through
recvTiming, which is now recvTimingResp. For the modules that have
both master and slave ports, e.g. the bus, the behaviour was
previously relying on branches based on pkt->isRequest(), and this is
now replaced with a direct call to the apprioriate member function
depending on the type of access. Please note that send/recvRetry is
still shared by all the timing accessors and remains in the Port base
class for now (to maintain the current bus functionality and avoid
changing the statistics of all regressions).
The packet queue is split into a MasterPort and SlavePort version to
facilitate the use of the new timing accessors. All uses of the
PacketQueue are updated accordingly.
With this patch, the type of packet (request or response) is now well
defined for each type of access, and asserts on pkt->isRequest() and
pkt->isResponse() are now moved to the appropriate send member
functions. It is also worth noting that sendTimingSnoopReq no longer
returns a boolean, as the semantics do not alow snoop requests to be
rejected or stalled. All these assumptions are now excplicitly part of
the port interface itself.
2012-05-01 19:40:42 +02:00
|
|
|
virtual bool recvTimingResp(PacketPtr pkt);
|
|
|
|
virtual void recvTimingSnoopReq(PacketPtr pkt);
|
2012-01-17 19:55:08 +01:00
|
|
|
|
2013-03-26 19:46:42 +01:00
|
|
|
virtual void recvFunctionalSnoop(PacketPtr pkt)
|
|
|
|
{
|
|
|
|
// @todo: Is there a need for potential invalidation here?
|
|
|
|
}
|
|
|
|
|
2012-01-17 19:55:08 +01:00
|
|
|
/** Handles doing a retry of the previous send. */
|
|
|
|
virtual void recvRetry();
|
|
|
|
|
|
|
|
/**
|
|
|
|
* As this CPU requires snooping to maintain the load store queue
|
|
|
|
* change the behaviour from the base CPU port.
|
|
|
|
*
|
2012-01-17 19:55:09 +01:00
|
|
|
* @return true since we have to snoop
|
2012-01-17 19:55:08 +01:00
|
|
|
*/
|
MEM: Introduce the master/slave port sub-classes in C++
This patch introduces the notion of a master and slave port in the C++
code, thus bringing the previous classification from the Python
classes into the corresponding simulation objects and memory objects.
The patch enables us to classify behaviours into the two bins and add
assumptions and enfore compliance, also simplifying the two
interfaces. As a starting point, isSnooping is confined to a master
port, and getAddrRanges to slave ports. More of these specilisations
are to come in later patches.
The getPort function is not getMasterPort and getSlavePort, and
returns a port reference rather than a pointer as NULL would never be
a valid return value. The default implementation of these two
functions is placed in MemObject, and calls fatal.
The one drawback with this specific patch is that it requires some
code duplication, e.g. QueuedPort becomes QueuedMasterPort and
QueuedSlavePort, and BusPort becomes BusMasterPort and BusSlavePort
(avoiding multiple inheritance). With the later introduction of the
port interfaces, moving the functionality outside the port itself, a
lot of the duplicated code will disappear again.
2012-03-30 15:40:11 +02:00
|
|
|
virtual bool isSnooping() const { return true; }
|
2012-01-17 19:55:08 +01:00
|
|
|
};
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
class TickEvent : public Event
|
|
|
|
{
|
|
|
|
private:
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Pointer to the CPU. */
|
2005-06-05 09:25:26 +02:00
|
|
|
FullO3CPU<Impl> *cpu;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
public:
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Constructs a tick event. */
|
2005-06-05 09:25:26 +02:00
|
|
|
TickEvent(FullO3CPU<Impl> *c);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Processes a tick event, calling tick() on the CPU. */
|
2004-08-20 20:54:07 +02:00
|
|
|
void process();
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Returns the description of the tick event. */
|
2008-02-06 22:32:40 +01:00
|
|
|
const char *description() const;
|
2004-08-20 20:54:07 +02:00
|
|
|
};
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** The tick event used for scheduling CPU ticks. */
|
2004-08-20 20:54:07 +02:00
|
|
|
TickEvent tickEvent;
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Schedule tick event, regardless of its current state. */
|
2012-08-28 20:30:33 +02:00
|
|
|
void scheduleTickEvent(Cycles delay)
|
2004-08-20 20:54:07 +02:00
|
|
|
{
|
|
|
|
if (tickEvent.squashed())
|
2012-08-28 20:30:31 +02:00
|
|
|
reschedule(tickEvent, clockEdge(delay));
|
2004-08-20 20:54:07 +02:00
|
|
|
else if (!tickEvent.scheduled())
|
2012-08-28 20:30:31 +02:00
|
|
|
schedule(tickEvent, clockEdge(delay));
|
2004-08-20 20:54:07 +02:00
|
|
|
}
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Unschedule tick event, regardless of its current state. */
|
2004-08-20 20:54:07 +02:00
|
|
|
void unscheduleTickEvent()
|
|
|
|
{
|
|
|
|
if (tickEvent.scheduled())
|
|
|
|
tickEvent.squash();
|
|
|
|
}
|
|
|
|
|
2013-01-07 19:05:46 +01:00
|
|
|
/**
|
|
|
|
* Check if the pipeline has drained and signal the DrainManager.
|
|
|
|
*
|
|
|
|
* This method checks if a drain has been requested and if the CPU
|
|
|
|
* has drained successfully (i.e., there are no instructions in
|
|
|
|
* the pipeline). If the CPU has drained, it deschedules the tick
|
|
|
|
* event and signals the drain manager.
|
|
|
|
*
|
|
|
|
* @return False if a drain hasn't been requested or the CPU
|
|
|
|
* hasn't drained, true otherwise.
|
|
|
|
*/
|
|
|
|
bool tryDrain();
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Perform sanity checks after a drain.
|
|
|
|
*
|
|
|
|
* This method is called from drain() when it has determined that
|
|
|
|
* the CPU is fully drained when gem5 is compiled with the NDEBUG
|
|
|
|
* macro undefined. The intention of this method is to do more
|
|
|
|
* extensive tests than the isDrained() method to weed out any
|
|
|
|
* draining bugs.
|
|
|
|
*/
|
|
|
|
void drainSanityCheck() const;
|
|
|
|
|
|
|
|
/** Check if a system is in a drained state. */
|
|
|
|
bool isDrained() const;
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
public:
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Constructs a CPU with the given parameters. */
|
2008-10-09 09:08:50 +02:00
|
|
|
FullO3CPU(DerivO3CPUParams *params);
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Destructor. */
|
2005-06-05 09:25:26 +02:00
|
|
|
~FullO3CPU();
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Registers statistics. */
|
2008-10-09 09:08:50 +02:00
|
|
|
void regStats();
|
Check in of various updates to the CPU. Mainly adds in stats, improves
branch prediction, and makes memory dependence work properly.
SConscript:
Added return address stack, tournament predictor.
cpu/base_cpu.cc:
Added debug break and print statements.
cpu/base_dyn_inst.cc:
cpu/base_dyn_inst.hh:
Comment out possibly unneeded variables.
cpu/beta_cpu/2bit_local_pred.cc:
2bit predictor no longer speculatively updates itself.
cpu/beta_cpu/alpha_dyn_inst.hh:
Comment formatting.
cpu/beta_cpu/alpha_full_cpu.hh:
Formatting
cpu/beta_cpu/alpha_full_cpu_builder.cc:
Added new parameters for branch predictors, and IQ parameters.
cpu/beta_cpu/alpha_full_cpu_impl.hh:
Register stats.
cpu/beta_cpu/alpha_params.hh:
Added parameters for IQ, branch predictors, and store sets.
cpu/beta_cpu/bpred_unit.cc:
Removed one class.
cpu/beta_cpu/bpred_unit.hh:
Add in RAS, stats. Changed branch predictor unit functionality
so that it holds a history of past branches so it can update, and also
hold a proper history of the RAS so it can be restored on branch
mispredicts.
cpu/beta_cpu/bpred_unit_impl.hh:
Added in stats, history of branches, RAS. Now bpred unit actually
modifies the instruction's predicted next PC.
cpu/beta_cpu/btb.cc:
Add in sanity checks.
cpu/beta_cpu/comm.hh:
Add in communication where needed, remove it where it's not.
cpu/beta_cpu/commit.hh:
cpu/beta_cpu/rename.hh:
cpu/beta_cpu/rename_impl.hh:
Add in stats.
cpu/beta_cpu/commit_impl.hh:
Stats, update what is sent back on branch mispredict.
cpu/beta_cpu/cpu_policy.hh:
Change the bpred unit being used.
cpu/beta_cpu/decode.hh:
cpu/beta_cpu/decode_impl.hh:
Stats.
cpu/beta_cpu/fetch.hh:
Stats, change squash so it can handle squashes from decode differently
than squashes from commit.
cpu/beta_cpu/fetch_impl.hh:
Add in stats. Change how a cache line is fetched. Update to work with
caches. Also have separate functions for different behavior if squash
is coming from decode vs commit.
cpu/beta_cpu/free_list.hh:
Remove some old comments.
cpu/beta_cpu/full_cpu.cc:
cpu/beta_cpu/full_cpu.hh:
Added function to remove instructions from back of instruction list
until a certain sequence number.
cpu/beta_cpu/iew.hh:
Stats, separate squashing behavior due to branches vs memory.
cpu/beta_cpu/iew_impl.hh:
Stats, separate squashing behavior for branches vs memory.
cpu/beta_cpu/inst_queue.cc:
Debug stuff
cpu/beta_cpu/inst_queue.hh:
Stats, change how mem dep unit works, debug stuff
cpu/beta_cpu/inst_queue_impl.hh:
Stats, change how mem dep unit works, debug stuff. Also add in
parameters that used to be hardcoded.
cpu/beta_cpu/mem_dep_unit.hh:
cpu/beta_cpu/mem_dep_unit_impl.hh:
Add in stats, change how memory dependence unit works. It now holds
the memory instructions that are waiting for their memory dependences
to resolve. It provides which instructions are ready directly to the
IQ.
cpu/beta_cpu/regfile.hh:
Fix up sanity checks.
cpu/beta_cpu/rename_map.cc:
Fix loop variable type.
cpu/beta_cpu/rob_impl.hh:
Remove intermediate DynInstPtr
cpu/beta_cpu/store_set.cc:
Add in debugging statements.
cpu/beta_cpu/store_set.hh:
Reorder function arguments to match the rest of the calls.
--HG--
extra : convert_revision : aabf9b1fecd1d743265dfc3b174d6159937c6f44
2004-10-22 00:02:36 +02:00
|
|
|
|
base: add support for probe points and common probes
The probe patch is motivated by the desire to move analytical and trace code
away from functional code. This is achieved by the probe interface which is
essentially a glorified observer model.
What this means to users:
* add a probe point and a "notify" call at the source of an "event"
* add an isolated module, that is being used to carry out *your* analysis (e.g. generate a trace)
* register that module as a probe listener
Note: an example is given for reference in src/cpu/o3/simple_trace.[hh|cc] and src/cpu/SimpleTrace.py
What is happening under the hood:
* every SimObject maintains has a ProbeManager.
* during initialization (src/python/m5/simulate.py) first regProbePoints and
the regProbeListeners is called on each SimObject. this hooks up the probe
point notify calls with the listeners.
FAQs:
Why did you develop probe points:
* to remove trace, stats gathering, analytical code out of the functional code.
* the belief that probes could be generically useful.
What is a probe point:
* a probe point is used to notify upon a given event (e.g. cpu commits an instruction)
What is a probe listener:
* a class that handles whatever the user wishes to do when they are notified
about an event.
What can be passed on notify:
* probe points are templates, and so the user can generate probes that pass any
type of argument (by const reference) to a listener.
What relationships can be generated (1:1, 1:N, N:M etc):
* there isn't a restriction. You can hook probe points and listeners up in a
1:1, 1:N, N:M relationship. They become useful when a number of modules
listen to the same probe points. The idea being that you can add a small
number of probes into the source code and develop a larger number of useful
analysis modules that use information passed by the probes.
Can you give examples:
* adding a probe point to the cpu's commit method allows you to build a trace
module (outputting assembler), you could re-use this to gather instruction
distribution (arithmetic, load/store, conditional, control flow) stats.
Why is the probe interface currently restricted to passing a const reference:
* the desire, initially at least, is to allow an interface to observe
functionality, but not to change functionality.
* of course this can be subverted by const-casting.
What is the performance impact of adding probes:
* when nothing is actively listening to the probes they should have a
relatively minor impact. Profiling has suggested even with a large number of
probes (60) the impact of them (when not active) is very minimal (<1%).
2014-01-24 22:29:30 +01:00
|
|
|
ProbePointArg<PacketPtr> *ppInstAccessComplete;
|
|
|
|
ProbePointArg<std::pair<DynInstPtr, PacketPtr> > *ppDataAccessComplete;
|
|
|
|
|
|
|
|
/** Register probe points. */
|
|
|
|
void regProbePoints();
|
|
|
|
|
2008-02-27 05:38:51 +01:00
|
|
|
void demapPage(Addr vaddr, uint64_t asn)
|
|
|
|
{
|
|
|
|
this->itb->demapPage(vaddr, asn);
|
|
|
|
this->dtb->demapPage(vaddr, asn);
|
|
|
|
}
|
|
|
|
|
|
|
|
void demapInstPage(Addr vaddr, uint64_t asn)
|
|
|
|
{
|
|
|
|
this->itb->demapPage(vaddr, asn);
|
|
|
|
}
|
|
|
|
|
|
|
|
void demapDataPage(Addr vaddr, uint64_t asn)
|
|
|
|
{
|
|
|
|
this->dtb->demapPage(vaddr, asn);
|
|
|
|
}
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Ticks CPU, calling tick() on each stage, and checking the overall
|
|
|
|
* activity to see if the CPU should deschedule itself.
|
|
|
|
*/
|
2005-05-19 07:28:25 +02:00
|
|
|
void tick();
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Initialize the CPU */
|
2005-05-19 07:28:25 +02:00
|
|
|
void init();
|
|
|
|
|
2013-01-07 19:05:44 +01:00
|
|
|
void startup();
|
|
|
|
|
2006-07-03 18:19:35 +02:00
|
|
|
/** Returns the Number of Active Threads in the CPU */
|
|
|
|
int numActiveThreads()
|
|
|
|
{ return activeThreads.size(); }
|
|
|
|
|
2006-07-02 00:52:02 +02:00
|
|
|
/** Add Thread to Active Threads List */
|
2009-05-26 18:23:13 +02:00
|
|
|
void activateThread(ThreadID tid);
|
2006-07-07 10:06:26 +02:00
|
|
|
|
|
|
|
/** Remove Thread from Active Threads List */
|
2009-05-26 18:23:13 +02:00
|
|
|
void deactivateThread(ThreadID tid);
|
2006-07-02 00:52:02 +02:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Setup CPU to insert a thread's context */
|
2009-05-26 18:23:13 +02:00
|
|
|
void insertThread(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Remove all of a thread's context from CPU */
|
2009-05-26 18:23:13 +02:00
|
|
|
void removeThread(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Count the Total Instructions Committed in the CPU. */
|
2012-02-12 23:07:39 +01:00
|
|
|
virtual Counter totalInsts() const;
|
|
|
|
|
|
|
|
/** Count the Total Ops (including micro ops) committed in the CPU. */
|
|
|
|
virtual Counter totalOps() const;
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Add Thread to Active Threads List. */
|
alpha,arm,mips,power,x86,cpu,sim: Cleanup activate/deactivate
activate(), suspend(), and halt() used on thread contexts had an optional
delay parameter. However this parameter was often ignored. Also, when used,
the delay was seemily arbitrarily set to 0 or 1 cycle (no other delays were
ever specified). This patch removes the delay parameter and 'Events'
associated with them across all ISAs and cores. Unused activate logic
is also removed.
2014-09-20 23:18:35 +02:00
|
|
|
void activateContext(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Remove Thread from Active Threads List */
|
2009-05-26 18:23:13 +02:00
|
|
|
void suspendContext(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Remove Thread from Active Threads List &&
|
|
|
|
* Remove Thread Context from CPU.
|
|
|
|
*/
|
2009-05-26 18:23:13 +02:00
|
|
|
void haltContext(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Update The Order In Which We Process Threads. */
|
|
|
|
void updateThreadPriority();
|
|
|
|
|
2013-01-07 19:05:46 +01:00
|
|
|
/** Is the CPU draining? */
|
|
|
|
bool isDraining() const { return getDrainState() == Drainable::Draining; }
|
|
|
|
|
2013-01-07 19:05:52 +01:00
|
|
|
void serializeThread(std::ostream &os, ThreadID tid);
|
2006-07-07 05:13:38 +02:00
|
|
|
|
2013-01-07 19:05:52 +01:00
|
|
|
void unserializeThread(Checkpoint *cp, const std::string §ion,
|
|
|
|
ThreadID tid);
|
2006-07-07 05:13:38 +02:00
|
|
|
|
|
|
|
public:
|
2008-10-09 09:08:50 +02:00
|
|
|
/** Executes a syscall.
|
|
|
|
* @todo: Determine if this needs to be virtual.
|
2006-04-23 00:26:48 +02:00
|
|
|
*/
|
2009-05-26 18:23:13 +02:00
|
|
|
void syscall(int64_t callnum, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-07-06 19:59:02 +02:00
|
|
|
/** Starts draining the CPU's pipeline of all instructions in
|
|
|
|
* order to stop all memory accesses. */
|
2012-11-02 17:32:01 +01:00
|
|
|
unsigned int drain(DrainManager *drain_manager);
|
2006-07-06 19:59:02 +02:00
|
|
|
|
|
|
|
/** Resumes execution after a drain. */
|
2012-11-02 17:32:01 +01:00
|
|
|
void drainResume();
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2013-01-07 19:05:46 +01:00
|
|
|
/**
|
|
|
|
* Commit has reached a safe point to drain a thread.
|
|
|
|
*
|
|
|
|
* Commit calls this method to inform the pipeline that it has
|
|
|
|
* reached a point where it is not executed microcode and is about
|
|
|
|
* to squash uncommitted instructions to fully drain the pipeline.
|
|
|
|
*/
|
|
|
|
void commitDrained(ThreadID tid);
|
2006-07-06 19:59:02 +02:00
|
|
|
|
|
|
|
/** Switches out this CPU. */
|
|
|
|
virtual void switchOut();
|
2006-05-16 20:06:35 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Takes over from another CPU. */
|
2006-07-06 19:59:02 +02:00
|
|
|
virtual void takeOverFrom(BaseCPU *oldCPU);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2013-02-15 23:40:08 +01:00
|
|
|
void verifyMemoryMode() const;
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
/** Get the current instruction sequence number, and increment it. */
|
2006-05-16 20:06:35 +02:00
|
|
|
InstSeqNum getAndIncrementInstSeq()
|
|
|
|
{ return globalSeqNum++; }
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2008-10-09 09:08:50 +02:00
|
|
|
/** Traps to handle given fault. */
|
2014-09-27 15:08:36 +02:00
|
|
|
void trap(const Fault &fault, ThreadID tid, const StaticInstPtr &inst);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
2008-10-20 22:22:59 +02:00
|
|
|
/** HW return from error interrupt. */
|
2009-05-26 18:23:13 +02:00
|
|
|
Fault hwrei(ThreadID tid);
|
2008-10-20 22:22:59 +02:00
|
|
|
|
2009-05-26 18:23:13 +02:00
|
|
|
bool simPalCheck(int palFunc, ThreadID tid);
|
2008-10-20 22:22:59 +02:00
|
|
|
|
2008-10-09 09:08:50 +02:00
|
|
|
/** Returns the Fault for any valid interrupt. */
|
|
|
|
Fault getInterrupts();
|
|
|
|
|
|
|
|
/** Processes any an interrupt fault. */
|
2014-09-19 16:35:18 +02:00
|
|
|
void processInterrupts(const Fault &interrupt);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
|
|
|
/** Halts the CPU. */
|
|
|
|
void halt() { panic("Halt not implemented!\n"); }
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
/** Check if this address is a valid instruction address. */
|
|
|
|
bool validInstAddr(Addr addr) { return true; }
|
|
|
|
|
|
|
|
/** Check if this address is a valid data address. */
|
|
|
|
bool validDataAddr(Addr addr) { return true; }
|
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Register accessors. Index refers to the physical register index. */
|
2008-10-09 09:08:50 +02:00
|
|
|
|
|
|
|
/** Reads a miscellaneous register. */
|
2015-02-16 09:33:28 +01:00
|
|
|
TheISA::MiscReg readMiscRegNoEffect(int misc_reg, ThreadID tid) const;
|
2008-10-09 09:08:50 +02:00
|
|
|
|
|
|
|
/** Reads a misc. register, including any side effects the read
|
|
|
|
* might have as defined by the architecture.
|
|
|
|
*/
|
2009-05-26 18:23:13 +02:00
|
|
|
TheISA::MiscReg readMiscReg(int misc_reg, ThreadID tid);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
|
|
|
/** Sets a miscellaneous register. */
|
2009-05-26 18:23:13 +02:00
|
|
|
void setMiscRegNoEffect(int misc_reg, const TheISA::MiscReg &val,
|
|
|
|
ThreadID tid);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
|
|
|
/** Sets a misc. register, including any side effects the write
|
|
|
|
* might have as defined by the architecture.
|
|
|
|
*/
|
|
|
|
void setMiscReg(int misc_reg, const TheISA::MiscReg &val,
|
2009-05-26 18:23:13 +02:00
|
|
|
ThreadID tid);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
uint64_t readIntReg(int reg_idx);
|
|
|
|
|
2006-12-06 17:36:40 +01:00
|
|
|
TheISA::FloatReg readFloatReg(int reg_idx);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-12-06 17:36:40 +01:00
|
|
|
TheISA::FloatRegBits readFloatRegBits(int reg_idx);
|
2006-03-14 21:55:00 +01:00
|
|
|
|
2013-10-15 20:22:44 +02:00
|
|
|
TheISA::CCReg readCCReg(int reg_idx);
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
void setIntReg(int reg_idx, uint64_t val);
|
|
|
|
|
2006-12-06 17:36:40 +01:00
|
|
|
void setFloatReg(int reg_idx, TheISA::FloatReg val);
|
2006-03-14 21:55:00 +01:00
|
|
|
|
2006-12-06 17:36:40 +01:00
|
|
|
void setFloatRegBits(int reg_idx, TheISA::FloatRegBits val);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2013-10-15 20:22:44 +02:00
|
|
|
void setCCReg(int reg_idx, TheISA::CCReg val);
|
|
|
|
|
2009-05-26 18:23:13 +02:00
|
|
|
uint64_t readArchIntReg(int reg_idx, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2009-07-09 08:02:20 +02:00
|
|
|
float readArchFloatReg(int reg_idx, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2009-05-26 18:23:13 +02:00
|
|
|
uint64_t readArchFloatRegInt(int reg_idx, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2013-10-15 20:22:44 +02:00
|
|
|
TheISA::CCReg readArchCCReg(int reg_idx, ThreadID tid);
|
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Architectural register accessors. Looks up in the commit
|
|
|
|
* rename table to obtain the true physical index of the
|
|
|
|
* architected register first, then accesses that physical
|
|
|
|
* register.
|
|
|
|
*/
|
2009-05-26 18:23:13 +02:00
|
|
|
void setArchIntReg(int reg_idx, uint64_t val, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2009-07-09 08:02:20 +02:00
|
|
|
void setArchFloatReg(int reg_idx, float val, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2009-05-26 18:23:13 +02:00
|
|
|
void setArchFloatRegInt(int reg_idx, uint64_t val, ThreadID tid);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2013-10-15 20:22:44 +02:00
|
|
|
void setArchCCReg(int reg_idx, TheISA::CCReg val, ThreadID tid);
|
|
|
|
|
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors.
This change is a low level and pervasive reorganization of how PCs are managed
in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about,
the PC and the NPC, and the lsb of the PC signaled whether or not you were in
PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next
micropc, x86 and ARM introduced variable length instruction sets, and ARM
started to keep track of mode bits in the PC. Each CPU model handled PCs in
its own custom way that needed to be updated individually to handle the new
dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack,
the complexity could be hidden in the ISA at the ISA implementation's expense.
Areas like the branch predictor hadn't been updated to handle branch delay
slots or micropcs, and it turns out that had introduced a significant (10s of
percent) performance bug in SPARC and to a lesser extend MIPS. Rather than
perpetuate the problem by reworking O3 again to handle the PC features needed
by x86, this change was introduced to rework PC handling in a more modular,
transparent, and hopefully efficient way.
PC type:
Rather than having the superset of all possible elements of PC state declared
in each of the CPU models, each ISA defines its own PCState type which has
exactly the elements it needs. A cross product of canned PCState classes are
defined in the new "generic" ISA directory for ISAs with/without delay slots
and microcode. These are either typedef-ed or subclassed by each ISA. To read
or write this structure through a *Context, you use the new pcState() accessor
which reads or writes depending on whether it has an argument. If you just
want the address of the current or next instruction or the current micro PC,
you can get those through read-only accessors on either the PCState type or
the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the
move away from readPC. That name is ambiguous since it's not clear whether or
not it should be the actual address to fetch from, or if it should have extra
bits in it like the PAL mode bit. Each class is free to define its own
functions to get at whatever values it needs however it needs to to be used in
ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the
PC and into a separate field like ARM.
These types can be reset to a particular pc (where npc = pc +
sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as
appropriate), printed, serialized, and compared. There is a branching()
function which encapsulates code in the CPU models that checked if an
instruction branched or not. Exactly what that means in the context of branch
delay slots which can skip an instruction when not taken is ambiguous, and
ideally this function and its uses can be eliminated. PCStates also generally
know how to advance themselves in various ways depending on if they point at
an instruction, a microop, or the last microop of a macroop. More on that
later.
Ideally, accessing all the PCs at once when setting them will improve
performance of M5 even though more data needs to be moved around. This is
because often all the PCs need to be manipulated together, and by getting them
all at once you avoid multiple function calls. Also, the PCs of a particular
thread will have spatial locality in the cache. Previously they were grouped
by element in arrays which spread out accesses.
Advancing the PC:
The PCs were previously managed entirely by the CPU which had to know about PC
semantics, try to figure out which dimension to increment the PC in, what to
set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction
with the PC type itself. Because most of the information about how to
increment the PC (mainly what type of instruction it refers to) is contained
in the instruction object, a new advancePC virtual function was added to the
StaticInst class. Subclasses provide an implementation that moves around the
right element of the PC with a minimal amount of decision making. In ISAs like
Alpha, the instructions always simply assign NPC to PC without having to worry
about micropcs, nnpcs, etc. The added cost of a virtual function call should
be outweighed by not having to figure out as much about what to do with the
PCs and mucking around with the extra elements.
One drawback of making the StaticInsts advance the PC is that you have to
actually have one to advance the PC. This would, superficially, seem to
require decoding an instruction before fetch could advance. This is, as far as
I can tell, realistic. fetch would advance through memory addresses, not PCs,
perhaps predicting new memory addresses using existing ones. More
sophisticated decisions about control flow would be made later on, after the
instruction was decoded, and handed back to fetch. If branching needs to
happen, some amount of decoding needs to happen to see that it's a branch,
what the target is, etc. This could get a little more complicated if that gets
done by the predecoder, but I'm choosing to ignore that for now.
Variable length instructions:
To handle variable length instructions in x86 and ARM, the predecoder now
takes in the current PC by reference to the getExtMachInst function. It can
modify the PC however it needs to (by setting NPC to be the PC + instruction
length, for instance). This could be improved since the CPU doesn't know if
the PC was modified and always has to write it back.
ISA parser:
To support the new API, all PC related operand types were removed from the
parser and replaced with a PCState type. There are two warts on this
implementation. First, as with all the other operand types, the PCState still
has to have a valid operand type even though it doesn't use it. Second, using
syntax like PCS.npc(target) doesn't work for two reasons, this looks like the
syntax for operand type overriding, and the parser can't figure out if you're
reading or writing. Instructions that use the PCS operand (which I've
consistently called it) need to first read it into a local variable,
manipulate it, and then write it back out.
Return address stack:
The return address stack needed a little extra help because, in the presence
of branch delay slots, it has to merge together elements of the return PC and
the call PC. To handle that, a buildRetPC utility function was added. There
are basically only two versions in all the ISAs, but it didn't seem short
enough to put into the generic ISA directory. Also, the branch predictor code
in O3 and InOrder were adjusted so that they always store the PC of the actual
call instruction in the RAS, not the next PC. If the call instruction is a
microop, the next PC refers to the next microop in the same macroop which is
probably not desirable. The buildRetPC function advances the PC intelligently
to the next macroop (in an ISA specific way) so that that case works.
Change in stats:
There were no change in stats except in MIPS and SPARC in the O3 model. MIPS
runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could
likely be improved further by setting call/return instruction flags and taking
advantage of the RAS.
TODO:
Add != operators to the PCState classes, defined trivially to be !(a==b).
Smooth out places where PCs are split apart, passed around, and put back
together later. I think this might happen in SPARC's fault code. Add ISA
specific constructors that allow setting PC elements without calling a bunch
of accessors. Try to eliminate the need for the branching() function. Factor
out Alpha's PAL mode pc bit into a separate flag field, and eliminate places
where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
|
|
|
/** Sets the commit PC state of a specific thread. */
|
|
|
|
void pcState(const TheISA::PCState &newPCState, ThreadID tid);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors.
This change is a low level and pervasive reorganization of how PCs are managed
in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about,
the PC and the NPC, and the lsb of the PC signaled whether or not you were in
PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next
micropc, x86 and ARM introduced variable length instruction sets, and ARM
started to keep track of mode bits in the PC. Each CPU model handled PCs in
its own custom way that needed to be updated individually to handle the new
dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack,
the complexity could be hidden in the ISA at the ISA implementation's expense.
Areas like the branch predictor hadn't been updated to handle branch delay
slots or micropcs, and it turns out that had introduced a significant (10s of
percent) performance bug in SPARC and to a lesser extend MIPS. Rather than
perpetuate the problem by reworking O3 again to handle the PC features needed
by x86, this change was introduced to rework PC handling in a more modular,
transparent, and hopefully efficient way.
PC type:
Rather than having the superset of all possible elements of PC state declared
in each of the CPU models, each ISA defines its own PCState type which has
exactly the elements it needs. A cross product of canned PCState classes are
defined in the new "generic" ISA directory for ISAs with/without delay slots
and microcode. These are either typedef-ed or subclassed by each ISA. To read
or write this structure through a *Context, you use the new pcState() accessor
which reads or writes depending on whether it has an argument. If you just
want the address of the current or next instruction or the current micro PC,
you can get those through read-only accessors on either the PCState type or
the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the
move away from readPC. That name is ambiguous since it's not clear whether or
not it should be the actual address to fetch from, or if it should have extra
bits in it like the PAL mode bit. Each class is free to define its own
functions to get at whatever values it needs however it needs to to be used in
ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the
PC and into a separate field like ARM.
These types can be reset to a particular pc (where npc = pc +
sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as
appropriate), printed, serialized, and compared. There is a branching()
function which encapsulates code in the CPU models that checked if an
instruction branched or not. Exactly what that means in the context of branch
delay slots which can skip an instruction when not taken is ambiguous, and
ideally this function and its uses can be eliminated. PCStates also generally
know how to advance themselves in various ways depending on if they point at
an instruction, a microop, or the last microop of a macroop. More on that
later.
Ideally, accessing all the PCs at once when setting them will improve
performance of M5 even though more data needs to be moved around. This is
because often all the PCs need to be manipulated together, and by getting them
all at once you avoid multiple function calls. Also, the PCs of a particular
thread will have spatial locality in the cache. Previously they were grouped
by element in arrays which spread out accesses.
Advancing the PC:
The PCs were previously managed entirely by the CPU which had to know about PC
semantics, try to figure out which dimension to increment the PC in, what to
set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction
with the PC type itself. Because most of the information about how to
increment the PC (mainly what type of instruction it refers to) is contained
in the instruction object, a new advancePC virtual function was added to the
StaticInst class. Subclasses provide an implementation that moves around the
right element of the PC with a minimal amount of decision making. In ISAs like
Alpha, the instructions always simply assign NPC to PC without having to worry
about micropcs, nnpcs, etc. The added cost of a virtual function call should
be outweighed by not having to figure out as much about what to do with the
PCs and mucking around with the extra elements.
One drawback of making the StaticInsts advance the PC is that you have to
actually have one to advance the PC. This would, superficially, seem to
require decoding an instruction before fetch could advance. This is, as far as
I can tell, realistic. fetch would advance through memory addresses, not PCs,
perhaps predicting new memory addresses using existing ones. More
sophisticated decisions about control flow would be made later on, after the
instruction was decoded, and handed back to fetch. If branching needs to
happen, some amount of decoding needs to happen to see that it's a branch,
what the target is, etc. This could get a little more complicated if that gets
done by the predecoder, but I'm choosing to ignore that for now.
Variable length instructions:
To handle variable length instructions in x86 and ARM, the predecoder now
takes in the current PC by reference to the getExtMachInst function. It can
modify the PC however it needs to (by setting NPC to be the PC + instruction
length, for instance). This could be improved since the CPU doesn't know if
the PC was modified and always has to write it back.
ISA parser:
To support the new API, all PC related operand types were removed from the
parser and replaced with a PCState type. There are two warts on this
implementation. First, as with all the other operand types, the PCState still
has to have a valid operand type even though it doesn't use it. Second, using
syntax like PCS.npc(target) doesn't work for two reasons, this looks like the
syntax for operand type overriding, and the parser can't figure out if you're
reading or writing. Instructions that use the PCS operand (which I've
consistently called it) need to first read it into a local variable,
manipulate it, and then write it back out.
Return address stack:
The return address stack needed a little extra help because, in the presence
of branch delay slots, it has to merge together elements of the return PC and
the call PC. To handle that, a buildRetPC utility function was added. There
are basically only two versions in all the ISAs, but it didn't seem short
enough to put into the generic ISA directory. Also, the branch predictor code
in O3 and InOrder were adjusted so that they always store the PC of the actual
call instruction in the RAS, not the next PC. If the call instruction is a
microop, the next PC refers to the next microop in the same macroop which is
probably not desirable. The buildRetPC function advances the PC intelligently
to the next macroop (in an ISA specific way) so that that case works.
Change in stats:
There were no change in stats except in MIPS and SPARC in the O3 model. MIPS
runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could
likely be improved further by setting call/return instruction flags and taking
advantage of the RAS.
TODO:
Add != operators to the PCState classes, defined trivially to be !(a==b).
Smooth out places where PCs are split apart, passed around, and put back
together later. I think this might happen in SPARC's fault code. Add ISA
specific constructors that allow setting PC elements without calling a bunch
of accessors. Try to eliminate the need for the branching() function. Factor
out Alpha's PAL mode pc bit into a separate flag field, and eliminate places
where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
|
|
|
/** Reads the commit PC state of a specific thread. */
|
|
|
|
TheISA::PCState pcState(ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors.
This change is a low level and pervasive reorganization of how PCs are managed
in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about,
the PC and the NPC, and the lsb of the PC signaled whether or not you were in
PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next
micropc, x86 and ARM introduced variable length instruction sets, and ARM
started to keep track of mode bits in the PC. Each CPU model handled PCs in
its own custom way that needed to be updated individually to handle the new
dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack,
the complexity could be hidden in the ISA at the ISA implementation's expense.
Areas like the branch predictor hadn't been updated to handle branch delay
slots or micropcs, and it turns out that had introduced a significant (10s of
percent) performance bug in SPARC and to a lesser extend MIPS. Rather than
perpetuate the problem by reworking O3 again to handle the PC features needed
by x86, this change was introduced to rework PC handling in a more modular,
transparent, and hopefully efficient way.
PC type:
Rather than having the superset of all possible elements of PC state declared
in each of the CPU models, each ISA defines its own PCState type which has
exactly the elements it needs. A cross product of canned PCState classes are
defined in the new "generic" ISA directory for ISAs with/without delay slots
and microcode. These are either typedef-ed or subclassed by each ISA. To read
or write this structure through a *Context, you use the new pcState() accessor
which reads or writes depending on whether it has an argument. If you just
want the address of the current or next instruction or the current micro PC,
you can get those through read-only accessors on either the PCState type or
the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the
move away from readPC. That name is ambiguous since it's not clear whether or
not it should be the actual address to fetch from, or if it should have extra
bits in it like the PAL mode bit. Each class is free to define its own
functions to get at whatever values it needs however it needs to to be used in
ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the
PC and into a separate field like ARM.
These types can be reset to a particular pc (where npc = pc +
sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as
appropriate), printed, serialized, and compared. There is a branching()
function which encapsulates code in the CPU models that checked if an
instruction branched or not. Exactly what that means in the context of branch
delay slots which can skip an instruction when not taken is ambiguous, and
ideally this function and its uses can be eliminated. PCStates also generally
know how to advance themselves in various ways depending on if they point at
an instruction, a microop, or the last microop of a macroop. More on that
later.
Ideally, accessing all the PCs at once when setting them will improve
performance of M5 even though more data needs to be moved around. This is
because often all the PCs need to be manipulated together, and by getting them
all at once you avoid multiple function calls. Also, the PCs of a particular
thread will have spatial locality in the cache. Previously they were grouped
by element in arrays which spread out accesses.
Advancing the PC:
The PCs were previously managed entirely by the CPU which had to know about PC
semantics, try to figure out which dimension to increment the PC in, what to
set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction
with the PC type itself. Because most of the information about how to
increment the PC (mainly what type of instruction it refers to) is contained
in the instruction object, a new advancePC virtual function was added to the
StaticInst class. Subclasses provide an implementation that moves around the
right element of the PC with a minimal amount of decision making. In ISAs like
Alpha, the instructions always simply assign NPC to PC without having to worry
about micropcs, nnpcs, etc. The added cost of a virtual function call should
be outweighed by not having to figure out as much about what to do with the
PCs and mucking around with the extra elements.
One drawback of making the StaticInsts advance the PC is that you have to
actually have one to advance the PC. This would, superficially, seem to
require decoding an instruction before fetch could advance. This is, as far as
I can tell, realistic. fetch would advance through memory addresses, not PCs,
perhaps predicting new memory addresses using existing ones. More
sophisticated decisions about control flow would be made later on, after the
instruction was decoded, and handed back to fetch. If branching needs to
happen, some amount of decoding needs to happen to see that it's a branch,
what the target is, etc. This could get a little more complicated if that gets
done by the predecoder, but I'm choosing to ignore that for now.
Variable length instructions:
To handle variable length instructions in x86 and ARM, the predecoder now
takes in the current PC by reference to the getExtMachInst function. It can
modify the PC however it needs to (by setting NPC to be the PC + instruction
length, for instance). This could be improved since the CPU doesn't know if
the PC was modified and always has to write it back.
ISA parser:
To support the new API, all PC related operand types were removed from the
parser and replaced with a PCState type. There are two warts on this
implementation. First, as with all the other operand types, the PCState still
has to have a valid operand type even though it doesn't use it. Second, using
syntax like PCS.npc(target) doesn't work for two reasons, this looks like the
syntax for operand type overriding, and the parser can't figure out if you're
reading or writing. Instructions that use the PCS operand (which I've
consistently called it) need to first read it into a local variable,
manipulate it, and then write it back out.
Return address stack:
The return address stack needed a little extra help because, in the presence
of branch delay slots, it has to merge together elements of the return PC and
the call PC. To handle that, a buildRetPC utility function was added. There
are basically only two versions in all the ISAs, but it didn't seem short
enough to put into the generic ISA directory. Also, the branch predictor code
in O3 and InOrder were adjusted so that they always store the PC of the actual
call instruction in the RAS, not the next PC. If the call instruction is a
microop, the next PC refers to the next microop in the same macroop which is
probably not desirable. The buildRetPC function advances the PC intelligently
to the next macroop (in an ISA specific way) so that that case works.
Change in stats:
There were no change in stats except in MIPS and SPARC in the O3 model. MIPS
runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could
likely be improved further by setting call/return instruction flags and taking
advantage of the RAS.
TODO:
Add != operators to the PCState classes, defined trivially to be !(a==b).
Smooth out places where PCs are split apart, passed around, and put back
together later. I think this might happen in SPARC's fault code. Add ISA
specific constructors that allow setting PC elements without calling a bunch
of accessors. Try to eliminate the need for the branching() function. Factor
out Alpha's PAL mode pc bit into a separate flag field, and eliminate places
where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
|
|
|
/** Reads the commit PC of a specific thread. */
|
|
|
|
Addr instAddr(ThreadID tid);
|
2007-04-14 19:13:18 +02:00
|
|
|
|
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors.
This change is a low level and pervasive reorganization of how PCs are managed
in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about,
the PC and the NPC, and the lsb of the PC signaled whether or not you were in
PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next
micropc, x86 and ARM introduced variable length instruction sets, and ARM
started to keep track of mode bits in the PC. Each CPU model handled PCs in
its own custom way that needed to be updated individually to handle the new
dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack,
the complexity could be hidden in the ISA at the ISA implementation's expense.
Areas like the branch predictor hadn't been updated to handle branch delay
slots or micropcs, and it turns out that had introduced a significant (10s of
percent) performance bug in SPARC and to a lesser extend MIPS. Rather than
perpetuate the problem by reworking O3 again to handle the PC features needed
by x86, this change was introduced to rework PC handling in a more modular,
transparent, and hopefully efficient way.
PC type:
Rather than having the superset of all possible elements of PC state declared
in each of the CPU models, each ISA defines its own PCState type which has
exactly the elements it needs. A cross product of canned PCState classes are
defined in the new "generic" ISA directory for ISAs with/without delay slots
and microcode. These are either typedef-ed or subclassed by each ISA. To read
or write this structure through a *Context, you use the new pcState() accessor
which reads or writes depending on whether it has an argument. If you just
want the address of the current or next instruction or the current micro PC,
you can get those through read-only accessors on either the PCState type or
the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the
move away from readPC. That name is ambiguous since it's not clear whether or
not it should be the actual address to fetch from, or if it should have extra
bits in it like the PAL mode bit. Each class is free to define its own
functions to get at whatever values it needs however it needs to to be used in
ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the
PC and into a separate field like ARM.
These types can be reset to a particular pc (where npc = pc +
sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as
appropriate), printed, serialized, and compared. There is a branching()
function which encapsulates code in the CPU models that checked if an
instruction branched or not. Exactly what that means in the context of branch
delay slots which can skip an instruction when not taken is ambiguous, and
ideally this function and its uses can be eliminated. PCStates also generally
know how to advance themselves in various ways depending on if they point at
an instruction, a microop, or the last microop of a macroop. More on that
later.
Ideally, accessing all the PCs at once when setting them will improve
performance of M5 even though more data needs to be moved around. This is
because often all the PCs need to be manipulated together, and by getting them
all at once you avoid multiple function calls. Also, the PCs of a particular
thread will have spatial locality in the cache. Previously they were grouped
by element in arrays which spread out accesses.
Advancing the PC:
The PCs were previously managed entirely by the CPU which had to know about PC
semantics, try to figure out which dimension to increment the PC in, what to
set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction
with the PC type itself. Because most of the information about how to
increment the PC (mainly what type of instruction it refers to) is contained
in the instruction object, a new advancePC virtual function was added to the
StaticInst class. Subclasses provide an implementation that moves around the
right element of the PC with a minimal amount of decision making. In ISAs like
Alpha, the instructions always simply assign NPC to PC without having to worry
about micropcs, nnpcs, etc. The added cost of a virtual function call should
be outweighed by not having to figure out as much about what to do with the
PCs and mucking around with the extra elements.
One drawback of making the StaticInsts advance the PC is that you have to
actually have one to advance the PC. This would, superficially, seem to
require decoding an instruction before fetch could advance. This is, as far as
I can tell, realistic. fetch would advance through memory addresses, not PCs,
perhaps predicting new memory addresses using existing ones. More
sophisticated decisions about control flow would be made later on, after the
instruction was decoded, and handed back to fetch. If branching needs to
happen, some amount of decoding needs to happen to see that it's a branch,
what the target is, etc. This could get a little more complicated if that gets
done by the predecoder, but I'm choosing to ignore that for now.
Variable length instructions:
To handle variable length instructions in x86 and ARM, the predecoder now
takes in the current PC by reference to the getExtMachInst function. It can
modify the PC however it needs to (by setting NPC to be the PC + instruction
length, for instance). This could be improved since the CPU doesn't know if
the PC was modified and always has to write it back.
ISA parser:
To support the new API, all PC related operand types were removed from the
parser and replaced with a PCState type. There are two warts on this
implementation. First, as with all the other operand types, the PCState still
has to have a valid operand type even though it doesn't use it. Second, using
syntax like PCS.npc(target) doesn't work for two reasons, this looks like the
syntax for operand type overriding, and the parser can't figure out if you're
reading or writing. Instructions that use the PCS operand (which I've
consistently called it) need to first read it into a local variable,
manipulate it, and then write it back out.
Return address stack:
The return address stack needed a little extra help because, in the presence
of branch delay slots, it has to merge together elements of the return PC and
the call PC. To handle that, a buildRetPC utility function was added. There
are basically only two versions in all the ISAs, but it didn't seem short
enough to put into the generic ISA directory. Also, the branch predictor code
in O3 and InOrder were adjusted so that they always store the PC of the actual
call instruction in the RAS, not the next PC. If the call instruction is a
microop, the next PC refers to the next microop in the same macroop which is
probably not desirable. The buildRetPC function advances the PC intelligently
to the next macroop (in an ISA specific way) so that that case works.
Change in stats:
There were no change in stats except in MIPS and SPARC in the O3 model. MIPS
runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could
likely be improved further by setting call/return instruction flags and taking
advantage of the RAS.
TODO:
Add != operators to the PCState classes, defined trivially to be !(a==b).
Smooth out places where PCs are split apart, passed around, and put back
together later. I think this might happen in SPARC's fault code. Add ISA
specific constructors that allow setting PC elements without calling a bunch
of accessors. Try to eliminate the need for the branching() function. Factor
out Alpha's PAL mode pc bit into a separate flag field, and eliminate places
where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
|
|
|
/** Reads the commit micro PC of a specific thread. */
|
|
|
|
MicroPC microPC(ThreadID tid);
|
2007-04-14 19:13:18 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Reads the next PC of a specific thread. */
|
ISA,CPU,etc: Create an ISA defined PC type that abstracts out ISA behaviors.
This change is a low level and pervasive reorganization of how PCs are managed
in M5. Back when Alpha was the only ISA, there were only 2 PCs to worry about,
the PC and the NPC, and the lsb of the PC signaled whether or not you were in
PAL mode. As other ISAs were added, we had to add an NNPC, micro PC and next
micropc, x86 and ARM introduced variable length instruction sets, and ARM
started to keep track of mode bits in the PC. Each CPU model handled PCs in
its own custom way that needed to be updated individually to handle the new
dimensions of variability, or, in the case of ARMs mode-bit-in-the-pc hack,
the complexity could be hidden in the ISA at the ISA implementation's expense.
Areas like the branch predictor hadn't been updated to handle branch delay
slots or micropcs, and it turns out that had introduced a significant (10s of
percent) performance bug in SPARC and to a lesser extend MIPS. Rather than
perpetuate the problem by reworking O3 again to handle the PC features needed
by x86, this change was introduced to rework PC handling in a more modular,
transparent, and hopefully efficient way.
PC type:
Rather than having the superset of all possible elements of PC state declared
in each of the CPU models, each ISA defines its own PCState type which has
exactly the elements it needs. A cross product of canned PCState classes are
defined in the new "generic" ISA directory for ISAs with/without delay slots
and microcode. These are either typedef-ed or subclassed by each ISA. To read
or write this structure through a *Context, you use the new pcState() accessor
which reads or writes depending on whether it has an argument. If you just
want the address of the current or next instruction or the current micro PC,
you can get those through read-only accessors on either the PCState type or
the *Contexts. These are instAddr(), nextInstAddr(), and microPC(). Note the
move away from readPC. That name is ambiguous since it's not clear whether or
not it should be the actual address to fetch from, or if it should have extra
bits in it like the PAL mode bit. Each class is free to define its own
functions to get at whatever values it needs however it needs to to be used in
ISA specific code. Eventually Alpha's PAL mode bit could be moved out of the
PC and into a separate field like ARM.
These types can be reset to a particular pc (where npc = pc +
sizeof(MachInst), nnpc = npc + sizeof(MachInst), upc = 0, nupc = 1 as
appropriate), printed, serialized, and compared. There is a branching()
function which encapsulates code in the CPU models that checked if an
instruction branched or not. Exactly what that means in the context of branch
delay slots which can skip an instruction when not taken is ambiguous, and
ideally this function and its uses can be eliminated. PCStates also generally
know how to advance themselves in various ways depending on if they point at
an instruction, a microop, or the last microop of a macroop. More on that
later.
Ideally, accessing all the PCs at once when setting them will improve
performance of M5 even though more data needs to be moved around. This is
because often all the PCs need to be manipulated together, and by getting them
all at once you avoid multiple function calls. Also, the PCs of a particular
thread will have spatial locality in the cache. Previously they were grouped
by element in arrays which spread out accesses.
Advancing the PC:
The PCs were previously managed entirely by the CPU which had to know about PC
semantics, try to figure out which dimension to increment the PC in, what to
set NPC/NNPC, etc. These decisions are best left to the ISA in conjunction
with the PC type itself. Because most of the information about how to
increment the PC (mainly what type of instruction it refers to) is contained
in the instruction object, a new advancePC virtual function was added to the
StaticInst class. Subclasses provide an implementation that moves around the
right element of the PC with a minimal amount of decision making. In ISAs like
Alpha, the instructions always simply assign NPC to PC without having to worry
about micropcs, nnpcs, etc. The added cost of a virtual function call should
be outweighed by not having to figure out as much about what to do with the
PCs and mucking around with the extra elements.
One drawback of making the StaticInsts advance the PC is that you have to
actually have one to advance the PC. This would, superficially, seem to
require decoding an instruction before fetch could advance. This is, as far as
I can tell, realistic. fetch would advance through memory addresses, not PCs,
perhaps predicting new memory addresses using existing ones. More
sophisticated decisions about control flow would be made later on, after the
instruction was decoded, and handed back to fetch. If branching needs to
happen, some amount of decoding needs to happen to see that it's a branch,
what the target is, etc. This could get a little more complicated if that gets
done by the predecoder, but I'm choosing to ignore that for now.
Variable length instructions:
To handle variable length instructions in x86 and ARM, the predecoder now
takes in the current PC by reference to the getExtMachInst function. It can
modify the PC however it needs to (by setting NPC to be the PC + instruction
length, for instance). This could be improved since the CPU doesn't know if
the PC was modified and always has to write it back.
ISA parser:
To support the new API, all PC related operand types were removed from the
parser and replaced with a PCState type. There are two warts on this
implementation. First, as with all the other operand types, the PCState still
has to have a valid operand type even though it doesn't use it. Second, using
syntax like PCS.npc(target) doesn't work for two reasons, this looks like the
syntax for operand type overriding, and the parser can't figure out if you're
reading or writing. Instructions that use the PCS operand (which I've
consistently called it) need to first read it into a local variable,
manipulate it, and then write it back out.
Return address stack:
The return address stack needed a little extra help because, in the presence
of branch delay slots, it has to merge together elements of the return PC and
the call PC. To handle that, a buildRetPC utility function was added. There
are basically only two versions in all the ISAs, but it didn't seem short
enough to put into the generic ISA directory. Also, the branch predictor code
in O3 and InOrder were adjusted so that they always store the PC of the actual
call instruction in the RAS, not the next PC. If the call instruction is a
microop, the next PC refers to the next microop in the same macroop which is
probably not desirable. The buildRetPC function advances the PC intelligently
to the next macroop (in an ISA specific way) so that that case works.
Change in stats:
There were no change in stats except in MIPS and SPARC in the O3 model. MIPS
runs in about 9% fewer ticks. SPARC runs with 30%-50% fewer ticks, which could
likely be improved further by setting call/return instruction flags and taking
advantage of the RAS.
TODO:
Add != operators to the PCState classes, defined trivially to be !(a==b).
Smooth out places where PCs are split apart, passed around, and put back
together later. I think this might happen in SPARC's fault code. Add ISA
specific constructors that allow setting PC elements without calling a bunch
of accessors. Try to eliminate the need for the branching() function. Factor
out Alpha's PAL mode pc bit into a separate flag field, and eliminate places
where it's blindly masked out or tested in the PC.
2010-10-31 08:07:20 +01:00
|
|
|
Addr nextInstAddr(ThreadID tid);
|
2006-06-16 04:01:28 +02:00
|
|
|
|
2008-10-09 09:08:50 +02:00
|
|
|
/** Initiates a squash of all in-flight instructions for a given
|
|
|
|
* thread. The source of the squash is an external update of
|
|
|
|
* state through the TC.
|
|
|
|
*/
|
2009-05-26 18:23:13 +02:00
|
|
|
void squashFromTC(ThreadID tid);
|
2008-10-09 09:08:50 +02:00
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
/** Function to add instruction onto the head of the list of the
|
|
|
|
* instructions. Used when new instructions are fetched.
|
|
|
|
*/
|
2006-04-23 00:26:48 +02:00
|
|
|
ListIt addInst(DynInstPtr &inst);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
/** Function to tell the CPU that an instruction has completed. */
|
2012-02-12 23:07:39 +01:00
|
|
|
void instDone(ThreadID tid, DynInstPtr &inst);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-05-19 21:37:52 +02:00
|
|
|
/** Remove an instruction from the front end of the list. There's
|
|
|
|
* no restriction on location of the instruction.
|
2004-08-20 20:54:07 +02:00
|
|
|
*/
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
void removeFrontInst(DynInstPtr &inst);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-07-23 19:39:42 +02:00
|
|
|
/** Remove all instructions that are not currently in the ROB.
|
|
|
|
* There's also an option to not squash delay slot instructions.*/
|
2009-05-26 18:23:13 +02:00
|
|
|
void removeInstsNotInROB(ThreadID tid);
|
2004-08-20 20:54:07 +02:00
|
|
|
|
Check in of various updates to the CPU. Mainly adds in stats, improves
branch prediction, and makes memory dependence work properly.
SConscript:
Added return address stack, tournament predictor.
cpu/base_cpu.cc:
Added debug break and print statements.
cpu/base_dyn_inst.cc:
cpu/base_dyn_inst.hh:
Comment out possibly unneeded variables.
cpu/beta_cpu/2bit_local_pred.cc:
2bit predictor no longer speculatively updates itself.
cpu/beta_cpu/alpha_dyn_inst.hh:
Comment formatting.
cpu/beta_cpu/alpha_full_cpu.hh:
Formatting
cpu/beta_cpu/alpha_full_cpu_builder.cc:
Added new parameters for branch predictors, and IQ parameters.
cpu/beta_cpu/alpha_full_cpu_impl.hh:
Register stats.
cpu/beta_cpu/alpha_params.hh:
Added parameters for IQ, branch predictors, and store sets.
cpu/beta_cpu/bpred_unit.cc:
Removed one class.
cpu/beta_cpu/bpred_unit.hh:
Add in RAS, stats. Changed branch predictor unit functionality
so that it holds a history of past branches so it can update, and also
hold a proper history of the RAS so it can be restored on branch
mispredicts.
cpu/beta_cpu/bpred_unit_impl.hh:
Added in stats, history of branches, RAS. Now bpred unit actually
modifies the instruction's predicted next PC.
cpu/beta_cpu/btb.cc:
Add in sanity checks.
cpu/beta_cpu/comm.hh:
Add in communication where needed, remove it where it's not.
cpu/beta_cpu/commit.hh:
cpu/beta_cpu/rename.hh:
cpu/beta_cpu/rename_impl.hh:
Add in stats.
cpu/beta_cpu/commit_impl.hh:
Stats, update what is sent back on branch mispredict.
cpu/beta_cpu/cpu_policy.hh:
Change the bpred unit being used.
cpu/beta_cpu/decode.hh:
cpu/beta_cpu/decode_impl.hh:
Stats.
cpu/beta_cpu/fetch.hh:
Stats, change squash so it can handle squashes from decode differently
than squashes from commit.
cpu/beta_cpu/fetch_impl.hh:
Add in stats. Change how a cache line is fetched. Update to work with
caches. Also have separate functions for different behavior if squash
is coming from decode vs commit.
cpu/beta_cpu/free_list.hh:
Remove some old comments.
cpu/beta_cpu/full_cpu.cc:
cpu/beta_cpu/full_cpu.hh:
Added function to remove instructions from back of instruction list
until a certain sequence number.
cpu/beta_cpu/iew.hh:
Stats, separate squashing behavior due to branches vs memory.
cpu/beta_cpu/iew_impl.hh:
Stats, separate squashing behavior for branches vs memory.
cpu/beta_cpu/inst_queue.cc:
Debug stuff
cpu/beta_cpu/inst_queue.hh:
Stats, change how mem dep unit works, debug stuff
cpu/beta_cpu/inst_queue_impl.hh:
Stats, change how mem dep unit works, debug stuff. Also add in
parameters that used to be hardcoded.
cpu/beta_cpu/mem_dep_unit.hh:
cpu/beta_cpu/mem_dep_unit_impl.hh:
Add in stats, change how memory dependence unit works. It now holds
the memory instructions that are waiting for their memory dependences
to resolve. It provides which instructions are ready directly to the
IQ.
cpu/beta_cpu/regfile.hh:
Fix up sanity checks.
cpu/beta_cpu/rename_map.cc:
Fix loop variable type.
cpu/beta_cpu/rob_impl.hh:
Remove intermediate DynInstPtr
cpu/beta_cpu/store_set.cc:
Add in debugging statements.
cpu/beta_cpu/store_set.hh:
Reorder function arguments to match the rest of the calls.
--HG--
extra : convert_revision : aabf9b1fecd1d743265dfc3b174d6159937c6f44
2004-10-22 00:02:36 +02:00
|
|
|
/** Remove all instructions younger than the given sequence number. */
|
2009-05-26 18:23:13 +02:00
|
|
|
void removeInstsUntil(const InstSeqNum &seq_num, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Removes the instruction pointed to by the iterator. */
|
2009-05-26 18:23:13 +02:00
|
|
|
inline void squashInstIt(const ListIt &instIt, ThreadID tid);
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Cleans up all instructions on the remove list. */
|
2006-04-23 00:26:48 +02:00
|
|
|
void cleanUpRemovedInsts();
|
Check in of various updates to the CPU. Mainly adds in stats, improves
branch prediction, and makes memory dependence work properly.
SConscript:
Added return address stack, tournament predictor.
cpu/base_cpu.cc:
Added debug break and print statements.
cpu/base_dyn_inst.cc:
cpu/base_dyn_inst.hh:
Comment out possibly unneeded variables.
cpu/beta_cpu/2bit_local_pred.cc:
2bit predictor no longer speculatively updates itself.
cpu/beta_cpu/alpha_dyn_inst.hh:
Comment formatting.
cpu/beta_cpu/alpha_full_cpu.hh:
Formatting
cpu/beta_cpu/alpha_full_cpu_builder.cc:
Added new parameters for branch predictors, and IQ parameters.
cpu/beta_cpu/alpha_full_cpu_impl.hh:
Register stats.
cpu/beta_cpu/alpha_params.hh:
Added parameters for IQ, branch predictors, and store sets.
cpu/beta_cpu/bpred_unit.cc:
Removed one class.
cpu/beta_cpu/bpred_unit.hh:
Add in RAS, stats. Changed branch predictor unit functionality
so that it holds a history of past branches so it can update, and also
hold a proper history of the RAS so it can be restored on branch
mispredicts.
cpu/beta_cpu/bpred_unit_impl.hh:
Added in stats, history of branches, RAS. Now bpred unit actually
modifies the instruction's predicted next PC.
cpu/beta_cpu/btb.cc:
Add in sanity checks.
cpu/beta_cpu/comm.hh:
Add in communication where needed, remove it where it's not.
cpu/beta_cpu/commit.hh:
cpu/beta_cpu/rename.hh:
cpu/beta_cpu/rename_impl.hh:
Add in stats.
cpu/beta_cpu/commit_impl.hh:
Stats, update what is sent back on branch mispredict.
cpu/beta_cpu/cpu_policy.hh:
Change the bpred unit being used.
cpu/beta_cpu/decode.hh:
cpu/beta_cpu/decode_impl.hh:
Stats.
cpu/beta_cpu/fetch.hh:
Stats, change squash so it can handle squashes from decode differently
than squashes from commit.
cpu/beta_cpu/fetch_impl.hh:
Add in stats. Change how a cache line is fetched. Update to work with
caches. Also have separate functions for different behavior if squash
is coming from decode vs commit.
cpu/beta_cpu/free_list.hh:
Remove some old comments.
cpu/beta_cpu/full_cpu.cc:
cpu/beta_cpu/full_cpu.hh:
Added function to remove instructions from back of instruction list
until a certain sequence number.
cpu/beta_cpu/iew.hh:
Stats, separate squashing behavior due to branches vs memory.
cpu/beta_cpu/iew_impl.hh:
Stats, separate squashing behavior for branches vs memory.
cpu/beta_cpu/inst_queue.cc:
Debug stuff
cpu/beta_cpu/inst_queue.hh:
Stats, change how mem dep unit works, debug stuff
cpu/beta_cpu/inst_queue_impl.hh:
Stats, change how mem dep unit works, debug stuff. Also add in
parameters that used to be hardcoded.
cpu/beta_cpu/mem_dep_unit.hh:
cpu/beta_cpu/mem_dep_unit_impl.hh:
Add in stats, change how memory dependence unit works. It now holds
the memory instructions that are waiting for their memory dependences
to resolve. It provides which instructions are ready directly to the
IQ.
cpu/beta_cpu/regfile.hh:
Fix up sanity checks.
cpu/beta_cpu/rename_map.cc:
Fix loop variable type.
cpu/beta_cpu/rob_impl.hh:
Remove intermediate DynInstPtr
cpu/beta_cpu/store_set.cc:
Add in debugging statements.
cpu/beta_cpu/store_set.hh:
Reorder function arguments to match the rest of the calls.
--HG--
extra : convert_revision : aabf9b1fecd1d743265dfc3b174d6159937c6f44
2004-10-22 00:02:36 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Debug function to print all instructions on the list. */
|
2004-08-20 20:54:07 +02:00
|
|
|
void dumpInsts();
|
|
|
|
|
|
|
|
public:
|
2008-11-10 20:51:18 +01:00
|
|
|
#ifndef NDEBUG
|
|
|
|
/** Count of total number of dynamic instructions in flight. */
|
|
|
|
int instcount;
|
|
|
|
#endif
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
/** List of all the instructions in flight. */
|
2006-04-23 00:26:48 +02:00
|
|
|
std::list<DynInstPtr> instList;
|
|
|
|
|
|
|
|
/** List of all the instructions that will be removed at the end of this
|
|
|
|
* cycle.
|
|
|
|
*/
|
|
|
|
std::queue<ListIt> removeList;
|
|
|
|
|
2006-05-19 21:37:52 +02:00
|
|
|
#ifdef DEBUG
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Debug structure to keep track of the sequence numbers still in
|
|
|
|
* flight.
|
|
|
|
*/
|
2006-04-23 00:26:48 +02:00
|
|
|
std::set<InstSeqNum> snList;
|
2006-05-19 21:37:52 +02:00
|
|
|
#endif
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-05-19 21:37:52 +02:00
|
|
|
/** Records if instructions need to be removed this cycle due to
|
|
|
|
* being retired or squashed.
|
2006-04-23 00:26:48 +02:00
|
|
|
*/
|
|
|
|
bool removeInstsThisCycle;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
protected:
|
|
|
|
/** The fetch stage. */
|
|
|
|
typename CPUPolicy::Fetch fetch;
|
|
|
|
|
|
|
|
/** The decode stage. */
|
|
|
|
typename CPUPolicy::Decode decode;
|
|
|
|
|
|
|
|
/** The dispatch stage. */
|
|
|
|
typename CPUPolicy::Rename rename;
|
|
|
|
|
|
|
|
/** The issue/execute/writeback stages. */
|
|
|
|
typename CPUPolicy::IEW iew;
|
|
|
|
|
|
|
|
/** The commit stage. */
|
|
|
|
typename CPUPolicy::Commit commit;
|
|
|
|
|
|
|
|
/** The register file. */
|
2013-10-15 20:22:44 +02:00
|
|
|
PhysRegFile regFile;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
/** The free list. */
|
|
|
|
typename CPUPolicy::FreeList freeList;
|
|
|
|
|
|
|
|
/** The rename map. */
|
2006-04-23 00:26:48 +02:00
|
|
|
typename CPUPolicy::RenameMap renameMap[Impl::MaxThreads];
|
|
|
|
|
|
|
|
/** The commit rename map. */
|
|
|
|
typename CPUPolicy::RenameMap commitRenameMap[Impl::MaxThreads];
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
/** The re-order buffer. */
|
|
|
|
typename CPUPolicy::ROB rob;
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Active Threads List */
|
2009-05-26 18:23:13 +02:00
|
|
|
std::list<ThreadID> activeThreads;
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Integer Register Scoreboard */
|
|
|
|
Scoreboard scoreboard;
|
|
|
|
|
2013-01-07 19:05:35 +01:00
|
|
|
std::vector<TheISA::ISA *> isa;
|
2009-07-09 08:02:20 +02:00
|
|
|
|
2012-01-17 19:55:08 +01:00
|
|
|
/** Instruction port. Note that it has to appear after the fetch stage. */
|
|
|
|
IcachePort icachePort;
|
|
|
|
|
|
|
|
/** Data port. Note that it has to appear after the iew stages */
|
|
|
|
DcachePort dcachePort;
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
public:
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Enum to give each stage a specific index, so when calling
|
|
|
|
* activateStage() or deactivateStage(), they can specify which stage
|
|
|
|
* is being activated/deactivated.
|
|
|
|
*/
|
|
|
|
enum StageIdx {
|
|
|
|
FetchIdx,
|
|
|
|
DecodeIdx,
|
|
|
|
RenameIdx,
|
|
|
|
IEWIdx,
|
|
|
|
CommitIdx,
|
|
|
|
NumStages };
|
|
|
|
|
2004-08-20 20:54:07 +02:00
|
|
|
/** Typedefs from the Impl to get the structs that each of the
|
|
|
|
* time buffers should use.
|
|
|
|
*/
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename CPUPolicy::TimeStruct TimeStruct;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename CPUPolicy::FetchStruct FetchStruct;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename CPUPolicy::DecodeStruct DecodeStruct;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename CPUPolicy::RenameStruct RenameStruct;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
Update to make multiple instruction issue and different latencies work.
Also change to ref counted DynInst.
SConscript:
Add branch predictor, BTB, load store queue, and storesets.
arch/isa_parser.py:
Specify the template parameter for AlphaDynInst
base/traceflags.py:
Add load store queue, store set, and mem dependence unit to the
list of trace flags.
cpu/base_dyn_inst.cc:
Change formating, add in debug statement.
cpu/base_dyn_inst.hh:
Change DynInst to be RefCounted, add flag to clear whether or not this
instruction can commit. This is likely to be removed in the future.
cpu/beta_cpu/alpha_dyn_inst.cc:
AlphaDynInst has been changed to be templated, so now this CC file
is just used to force instantiations of AlphaDynInst.
cpu/beta_cpu/alpha_dyn_inst.hh:
Changed AlphaDynInst to be templated on Impl. Removed some unnecessary
functions.
cpu/beta_cpu/alpha_full_cpu.cc:
AlphaFullCPU has been changed to be templated, so this CC file is now
just used to force instantation of AlphaFullCPU.
cpu/beta_cpu/alpha_full_cpu.hh:
Change AlphaFullCPU to be templated on Impl.
cpu/beta_cpu/alpha_impl.hh:
Update it to reflect AlphaDynInst and AlphaFullCPU being templated
on Impl. Also removed time buffers from here, as they are really
a part of the CPU and are thus in the CPU policy now.
cpu/beta_cpu/alpha_params.hh:
Make AlphaSimpleParams inherit from the BaseFullCPU so that it doesn't
need to specifically declare any parameters that are already in the
BaseFullCPU.
cpu/beta_cpu/comm.hh:
Changed the structure of the time buffer communication structs. Now
they include the size of the packet of instructions it is sending.
Added some parameters to the backwards communication struct, mainly
for squashing.
cpu/beta_cpu/commit.hh:
Update typenames to reflect change in location of time buffer structs.
Update DynInst to DynInstPtr (it is refcounted now).
cpu/beta_cpu/commit_impl.hh:
Formatting changes mainly. Also sends back proper information
on branch mispredicts so that the bpred unit can update itself.
Updated behavior for non-speculative instructions (stores, any
other non-spec instructions): once they reach the head of the ROB,
the ROB signals back to the IQ that it can go ahead and issue the
non-speculative instruction. The instruction itself is updated so that
commit won't try to commit it again until it is done executing.
cpu/beta_cpu/cpu_policy.hh:
Added branch prediction unit, mem dependence prediction unit, load
store queue. Moved time buffer structs from AlphaSimpleImpl to here.
cpu/beta_cpu/decode.hh:
Changed typedefs to reflect change in location of time buffer structs
and also the change from DynInst to ref counted DynInstPtr.
cpu/beta_cpu/decode_impl.hh:
Continues to buffer instructions even while unblocking now. Changed
how it loops through groups of instructions so it can properly block
during the middle of a group of instructions.
cpu/beta_cpu/fetch.hh:
Changed typedefs to reflect change in location of time buffer structs
and the change to ref counted DynInsts. Also added in branch
brediction unit.
cpu/beta_cpu/fetch_impl.hh:
Add in branch prediction. Changed how fetch checks inputs and its
current state to make for easier logic.
cpu/beta_cpu/free_list.cc:
Changed int regs and float regs to logically use one flat namespace.
Future change will be moving them to a single scoreboard to conserve
space.
cpu/beta_cpu/free_list.hh:
Mostly debugging statements. Might be removed for performance in future.
cpu/beta_cpu/full_cpu.cc:
Added in some debugging statements. Updated BaseFullCPU to take
a params object.
cpu/beta_cpu/full_cpu.hh:
Added params class within BaseCPU that other param classes will be
able to inherit from. Updated typedefs to reflect change in location
of time buffer structs and ref counted DynInst.
cpu/beta_cpu/iew.hh:
Updated typedefs to reflect change in location of time buffer structs
and use of ref counted DynInsts.
cpu/beta_cpu/iew_impl.hh:
Added in load store queue, updated iew to be able to execute non-
speculative instructions, instead of having them execute in commit.
cpu/beta_cpu/inst_queue.hh:
Updated change to ref counted DynInsts. Changed inst queue to hold
non-speculative instructions as well, which are issued only when
commit signals backwards that a nonspeculative instruction is at
the head of the ROB.
cpu/beta_cpu/inst_queue_impl.hh:
Updated to allow for non-speculative instructions to be in the inst
queue. Also added some debug functions.
cpu/beta_cpu/regfile.hh:
Added debugging statements, changed formatting.
cpu/beta_cpu/rename.hh:
Updated typedefs, added some functions to clean up code.
cpu/beta_cpu/rename_impl.hh:
Moved some code into functions to make it easier to read.
cpu/beta_cpu/rename_map.cc:
Changed int and float reg behavior to use a single flat namespace. In
the future, the rename maps can be combined to a single rename map to
save space.
cpu/beta_cpu/rename_map.hh:
Added destructor.
cpu/beta_cpu/rob.hh:
Updated it with change from DynInst to ref counted DynInst.
cpu/beta_cpu/rob_impl.hh:
Formatting, updated to use ref counted DynInst.
cpu/static_inst.hh:
Updated forward declaration for AlphaDynInst now that it is templated.
--HG--
extra : convert_revision : 1045f240ee9b6a4bd368e1806aca029ebbdc6dd3
2004-09-23 20:06:03 +02:00
|
|
|
typedef typename CPUPolicy::IEWStruct IEWStruct;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
|
|
|
/** The main time buffer to do backwards communication. */
|
|
|
|
TimeBuffer<TimeStruct> timeBuffer;
|
|
|
|
|
|
|
|
/** The fetch stage's instruction queue. */
|
|
|
|
TimeBuffer<FetchStruct> fetchQueue;
|
|
|
|
|
|
|
|
/** The decode stage's instruction queue. */
|
|
|
|
TimeBuffer<DecodeStruct> decodeQueue;
|
|
|
|
|
|
|
|
/** The rename stage's instruction queue. */
|
|
|
|
TimeBuffer<RenameStruct> renameQueue;
|
|
|
|
|
|
|
|
/** The IEW stage's instruction queue. */
|
|
|
|
TimeBuffer<IEWStruct> iewQueue;
|
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
private:
|
|
|
|
/** The activity recorder; used to tell if the CPU has any
|
|
|
|
* activity remaining or if it can go to idle and deschedule
|
|
|
|
* itself.
|
|
|
|
*/
|
2006-05-19 21:37:52 +02:00
|
|
|
ActivityRecorder activityRec;
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
public:
|
|
|
|
/** Records that there was time buffer activity this cycle. */
|
2006-05-19 21:37:52 +02:00
|
|
|
void activityThisCycle() { activityRec.activity(); }
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Changes a stage's status to active within the activity recorder. */
|
2006-05-19 21:37:52 +02:00
|
|
|
void activateStage(const StageIdx idx)
|
|
|
|
{ activityRec.activateStage(idx); }
|
2006-03-04 21:18:40 +01:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Changes a stage's status to inactive within the activity recorder. */
|
2006-05-19 21:37:52 +02:00
|
|
|
void deactivateStage(const StageIdx idx)
|
|
|
|
{ activityRec.deactivateStage(idx); }
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Wakes the CPU, rescheduling the CPU if it's not already active. */
|
|
|
|
void wakeCPU();
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2009-01-24 16:27:21 +01:00
|
|
|
virtual void wakeup();
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Gets a free thread id. Use if thread ids change across system. */
|
2009-05-26 18:23:13 +02:00
|
|
|
ThreadID getFreeTid();
|
2006-03-04 21:18:40 +01:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
public:
|
2006-06-06 23:32:21 +02:00
|
|
|
/** Returns a pointer to a thread context. */
|
2009-05-26 18:23:13 +02:00
|
|
|
ThreadContext *
|
|
|
|
tcBase(ThreadID tid)
|
2006-03-04 21:18:40 +01:00
|
|
|
{
|
2006-06-06 23:32:21 +02:00
|
|
|
return thread[tid]->getTC();
|
2005-05-03 16:56:47 +02:00
|
|
|
}
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** The global sequence number counter. */
|
2006-09-01 02:51:30 +02:00
|
|
|
InstSeqNum globalSeqNum;//[Impl::MaxThreads];
|
2004-08-20 20:54:07 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Pointer to the checker, which can dynamically verify
|
|
|
|
* instruction results at run time. This can be set to NULL if it
|
|
|
|
* is not being used.
|
|
|
|
*/
|
2012-01-31 16:46:03 +01:00
|
|
|
Checker<Impl> *checker;
|
2006-05-16 20:06:35 +02:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Pointer to the system. */
|
2004-08-20 20:54:07 +02:00
|
|
|
System *system;
|
|
|
|
|
2012-11-02 17:32:01 +01:00
|
|
|
/** DrainManager to notify when draining has completed. */
|
|
|
|
DrainManager *drainManager;
|
2006-07-06 19:59:02 +02:00
|
|
|
|
2006-05-31 17:45:02 +02:00
|
|
|
/** Pointers to all of the threads in the CPU. */
|
2006-05-16 20:06:35 +02:00
|
|
|
std::vector<Thread *> thread;
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Threads Scheduled to Enter CPU */
|
|
|
|
std::list<int> cpuWaitList;
|
|
|
|
|
|
|
|
/** The cycle that the CPU was last running, used for statistics. */
|
2012-08-28 20:30:33 +02:00
|
|
|
Cycles lastRunningCycle;
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2006-07-02 00:52:02 +02:00
|
|
|
/** The cycle that the CPU was last activated by a new thread*/
|
|
|
|
Tick lastActivatedCycle;
|
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Mapping for system thread id to cpu id */
|
2009-05-26 18:23:13 +02:00
|
|
|
std::map<ThreadID, unsigned> threadMap;
|
2006-04-23 00:26:48 +02:00
|
|
|
|
|
|
|
/** Available thread ids in the cpu*/
|
2009-05-26 18:23:13 +02:00
|
|
|
std::vector<ThreadID> tids;
|
2006-04-23 00:26:48 +02:00
|
|
|
|
2008-10-09 09:08:50 +02:00
|
|
|
/** CPU read function, forwards read to LSQ. */
|
2010-02-12 20:53:20 +01:00
|
|
|
Fault read(RequestPtr &req, RequestPtr &sreqLow, RequestPtr &sreqHigh,
|
2010-08-13 15:16:02 +02:00
|
|
|
uint8_t *data, int load_idx)
|
2008-10-09 09:08:50 +02:00
|
|
|
{
|
2010-02-12 20:53:20 +01:00
|
|
|
return this->iew.ldstQueue.read(req, sreqLow, sreqHigh,
|
|
|
|
data, load_idx);
|
2008-10-09 09:08:50 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/** CPU write function, forwards write to LSQ. */
|
2010-02-12 20:53:20 +01:00
|
|
|
Fault write(RequestPtr &req, RequestPtr &sreqLow, RequestPtr &sreqHigh,
|
2010-08-13 15:16:02 +02:00
|
|
|
uint8_t *data, int store_idx)
|
2008-10-09 09:08:50 +02:00
|
|
|
{
|
2010-02-12 20:53:20 +01:00
|
|
|
return this->iew.ldstQueue.write(req, sreqLow, sreqHigh,
|
|
|
|
data, store_idx);
|
2008-10-09 09:08:50 +02:00
|
|
|
}
|
|
|
|
|
2012-01-17 19:55:08 +01:00
|
|
|
/** Used by the fetch unit to get a hold of the instruction port. */
|
2013-03-26 19:46:42 +01:00
|
|
|
virtual MasterPort &getInstPort() { return icachePort; }
|
2012-01-17 19:55:08 +01:00
|
|
|
|
2010-02-12 20:53:20 +01:00
|
|
|
/** Get the dcache port (used to find block size for translations). */
|
2013-03-26 19:46:42 +01:00
|
|
|
virtual MasterPort &getDataPort() { return dcachePort; }
|
2010-02-12 20:53:20 +01:00
|
|
|
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Stat for total number of times the CPU is descheduled. */
|
2009-03-06 04:09:53 +01:00
|
|
|
Stats::Scalar timesIdled;
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Stat for total number of cycles the CPU spends descheduled. */
|
2009-03-06 04:09:53 +01:00
|
|
|
Stats::Scalar idleCycles;
|
2011-12-01 09:15:22 +01:00
|
|
|
/** Stat for total number of cycles the CPU spends descheduled due to a
|
|
|
|
* quiesce operation or waiting for an interrupt. */
|
|
|
|
Stats::Scalar quiesceCycles;
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Stat for the number of committed instructions per thread. */
|
2009-03-06 04:09:53 +01:00
|
|
|
Stats::Vector committedInsts;
|
2012-02-12 23:07:39 +01:00
|
|
|
/** Stat for the number of committed ops (including micro ops) per thread. */
|
|
|
|
Stats::Vector committedOps;
|
2006-04-23 00:26:48 +02:00
|
|
|
/** Stat for the CPI per thread. */
|
|
|
|
Stats::Formula cpi;
|
|
|
|
/** Stat for the total CPI. */
|
|
|
|
Stats::Formula totalCpi;
|
|
|
|
/** Stat for the IPC per thread. */
|
|
|
|
Stats::Formula ipc;
|
|
|
|
/** Stat for the total IPC. */
|
|
|
|
Stats::Formula totalIpc;
|
2011-02-07 07:14:17 +01:00
|
|
|
|
|
|
|
//number of integer register file accesses
|
|
|
|
Stats::Scalar intRegfileReads;
|
|
|
|
Stats::Scalar intRegfileWrites;
|
|
|
|
//number of float register file accesses
|
|
|
|
Stats::Scalar fpRegfileReads;
|
|
|
|
Stats::Scalar fpRegfileWrites;
|
2013-10-15 20:22:44 +02:00
|
|
|
//number of CC register file accesses
|
|
|
|
Stats::Scalar ccRegfileReads;
|
|
|
|
Stats::Scalar ccRegfileWrites;
|
2011-02-07 07:14:17 +01:00
|
|
|
//number of misc
|
|
|
|
Stats::Scalar miscRegfileReads;
|
|
|
|
Stats::Scalar miscRegfileWrites;
|
2004-08-20 20:54:07 +02:00
|
|
|
};
|
|
|
|
|
2006-05-19 21:37:52 +02:00
|
|
|
#endif // __CPU_O3_CPU_HH__
|