sanchayanmaity/gem5
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
gem5/src/cpu/simple/base.cc

566 lines
15 KiB
C++
Raw Normal View History

Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
/*
CheckerCPU: Re-factor CheckerCPU to be compatible with current gem5 Brings the CheckerCPU back to life to allow FS and SE checking of the O3CPU. These changes have only been tested with the ARM ISA. Other ISAs potentially require modification.
2012-01-31 16:46:03 +01:00
* Copyright (c) 2010-2011 ARM Limited
CPU: Reset fetch offset after a exception
2010-06-02 19:58:12 +02:00
* 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.
*
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
* Copyright (c) 2002-2005 The Regents of The University of Michigan
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met: redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer;
* redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution;
* neither the name of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Updated Authors from bk prs info --HG-- extra : convert_revision : 77f475b156d81c03a2811818fa23593d5615c685
2006-06-01 01:26:56 +02:00
*
* Authors: Steve Reinhardt
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
*/
This one really needs to be arch/faults.hh --HG-- extra : convert_revision : aad1ee04ade9f4394c9ef0386f23d6f2ca373412
2006-10-28 10:44:05 +02:00
#include "arch/faults.hh"
includes: sort includes again
2009-05-17 23:34:52 +02:00
#include "arch/utility.hh"
includes: sort all includes
2011-04-15 19:44:06 +02:00
#include "base/loader/symtab.hh"
CPA: Add code to automatically record function symbols as CPU executes.
2009-02-27 01:29:17 +01:00
#include "base/cp_annotate.hh"
includes: sort includes again
2009-05-17 23:34:52 +02:00
#include "base/cprintf.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "base/inifile.hh"
#include "base/misc.hh"
#include "base/pollevent.hh"
#include "base/range.hh"
#include "base/trace.hh"
includes: sort includes again
2009-05-17 23:34:52 +02:00
#include "base/types.hh"
arch: nuke arch/isa_specific.hh and move stuff to generated config/the_isa.hh
2009-09-23 17:34:21 +02:00
#include "config/the_isa.hh"
CheckerCPU: Re-factor CheckerCPU to be compatible with current gem5 Brings the CheckerCPU back to life to allow FS and SE checking of the O3CPU. These changes have only been tested with the ARM ISA. Other ISAs potentially require modification.
2012-01-31 16:46:03 +01:00
#include "config/use_checker.hh"
includes: sort all includes
2011-04-15 19:44:06 +02:00
#include "cpu/simple/base.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "cpu/base.hh"
#include "cpu/exetrace.hh"
#include "cpu/profile.hh"
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
#include "cpu/simple_thread.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "cpu/smt.hh"
#include "cpu/static_inst.hh"
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
#include "cpu/thread_context.hh"
trace: reimplement the DTRACE function so it doesn't use a vector At the same time, rename the trace flags to debug flags since they have broader usage than simply tracing. This means that --trace-flags is now --debug-flags and --trace-help is now --debug-help
2011-04-15 19:44:32 +02:00
#include "debug/Decode.hh"
#include "debug/Fetch.hh"
#include "debug/Quiesce.hh"
refactor code for the packet, get rid of packet_impl.hh and call it packet_access.hh and fix the #includes so things compile right. --HG-- extra : convert_revision : d3626c9715b9f7e51bb3ab8d97e971fad4e0b724
2006-10-20 08:38:45 +02:00
#include "mem/packet.hh"
request: rename INST_READ to INST_FETCH.
2009-04-21 03:54:02 +02:00
#include "mem/request.hh"
includes: sort includes again
2009-05-17 23:34:52 +02:00
#include "params/BaseSimpleCPU.hh"
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
#include "sim/byteswap.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "sim/debug.hh"
#include "sim/sim_events.hh"
#include "sim/sim_object.hh"
#include "sim/stats.hh"
memory mode information now contained in system object States are now running, draining, or drained. memory state information moved into system object system parameter is not fs only for cpus Implement drain() support in devices Update for drain() call that returns number of times drain_event->process() will be called Break O3 CPU! No sense in putting in a hack change that kevin is going to remove in a few minutes i imagine src/cpu/simple/atomic.cc: src/cpu/simple/atomic.hh: Since se mode has a system, allow access to it Verify that the atomic cpu is connected to an atomic system on resume src/cpu/simple/base.cc: Since se mode has a system, allow access to it src/cpu/simple/timing.cc: src/cpu/simple/timing.hh: Update for new drain() call that returns number of times drain_event->process() will be called and memory state being moved into the system Since se mode has a system, allow access to it Verify that the timing cpu is connected to an timing system on resume src/dev/ide_disk.cc: src/dev/io_device.cc: src/dev/io_device.hh: src/dev/ns_gige.cc: src/dev/ns_gige.hh: src/dev/pcidev.cc: src/dev/pcidev.hh: src/dev/sinic.cc: src/dev/sinic.hh: Implement drain() support in devices src/python/m5/config.py: Allow drain to return number of times drain_event->process() will be called. Normally 0 or 1 but things like O3 cpu or devices with multiple ports may want to call it many times src/python/m5/objects/BaseCPU.py: move system parameter out of fs to everyone src/sim/sim_object.cc: src/sim/sim_object.hh: States are now running, draining, or drained. memory state information moved into system object src/sim/system.cc: src/sim/system.hh: memory mode information now contained in system object --HG-- extra : convert_revision : 1389c77e66ee6d9710bf77b4306fb47e107b21cf
2006-07-13 02:22:07 +02:00
#include "sim/system.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#if FULL_SYSTEM
Put kernel_stats back into arch. --HG-- rename : src/kern/alpha/idle_event.cc => src/arch/alpha/idle_event.cc rename : src/kern/alpha/idle_event.hh => src/arch/alpha/idle_event.hh rename : src/kern/alpha/kernel_stats.cc => src/arch/alpha/kernel_stats.cc rename : src/kern/alpha/kernel_stats.hh => src/arch/alpha/kernel_stats.hh rename : src/kern/sparc/kernel_stats.hh => src/arch/sparc/kernel_stats.hh rename : src/kern/base_kernel_stats.cc => src/kern/kernel_stats.cc rename : src/kern/base_kernel_stats.hh => src/kern/kernel_stats.hh extra : convert_revision : 42bd3e36b407edbd19b912c9218f4e5923a15966
2006-11-08 04:34:34 +01:00
#include "arch/kernel_stats.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "arch/stacktrace.hh"
Put kernel_stats back into arch. --HG-- rename : src/kern/alpha/idle_event.cc => src/arch/alpha/idle_event.cc rename : src/kern/alpha/idle_event.hh => src/arch/alpha/idle_event.hh rename : src/kern/alpha/kernel_stats.cc => src/arch/alpha/kernel_stats.cc rename : src/kern/alpha/kernel_stats.hh => src/arch/alpha/kernel_stats.hh rename : src/kern/sparc/kernel_stats.hh => src/arch/sparc/kernel_stats.hh rename : src/kern/base_kernel_stats.cc => src/kern/kernel_stats.cc rename : src/kern/base_kernel_stats.hh => src/kern/kernel_stats.hh extra : convert_revision : 42bd3e36b407edbd19b912c9218f4e5923a15966
2006-11-08 04:34:34 +01:00
#include "arch/tlb.hh"
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#include "arch/vtophys.hh"
#else // !FULL_SYSTEM
#include "mem/mem_object.hh"
#endif // FULL_SYSTEM
CheckerCPU: Re-factor CheckerCPU to be compatible with current gem5 Brings the CheckerCPU back to life to allow FS and SE checking of the O3CPU. These changes have only been tested with the ARM ISA. Other ISAs potentially require modification.
2012-01-31 16:46:03 +01:00
#if USE_CHECKER
#include "cpu/checker/cpu.hh"
#include "cpu/checker/thread_context.hh"
#endif
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
using namespace std;
using namespace TheISA;
params: Convert the CPU objects to use the auto generated param structs. A whole bunch of stuff has been converted to use the new params stuff, but the CPU wasn't one of them. While we're at it, make some things a bit more stylish. Most of the work was done by Gabe, I just cleaned stuff up a bit more at the end.
2008-08-11 21:22:16 +02:00
BaseSimpleCPU::BaseSimpleCPU(BaseSimpleCPUParams *p)
Assign traceData to be NULL at BaseSimpleCPU constructor. Initialize a temporary variable for thread->readPC() at setupFetchRequest() to reduce function calls. exec tracing isn't needed for m5.fast binaries Moved MISCREG_GL, MISCREG_CWP, and MISCREG_TLB_DATA out of switch statement and use if blocks instead. src/arch/sparc/miscregfile.cc: Moved MISCREG_GL, MISCREG_CWP, and MISCREG_TLB_DATA out of switch statement and use if blocks instead. src/cpu/simple/base.cc: Assign traceData to be NULL at BaseSimpleCPU constructor. Initialize a temporary variable for thread->readPC() at setupFetchRequest() to reduce function calls. exec tracing isn't needed for m5.fast binaries --HG-- extra : convert_revision : 5dc92fff05c9bde994f1e0f1bb40e11c44eb72c6
2007-05-31 22:01:41 +02:00
: BaseCPU(p), traceData(NULL), thread(NULL), predecoder(NULL)
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
{
#if FULL_SYSTEM
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
thread = new SimpleThread(this, 0, p->system, p->itb, p->dtb);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#else
params: Convert the CPU objects to use the auto generated param structs. A whole bunch of stuff has been converted to use the new params stuff, but the CPU wasn't one of them. While we're at it, make some things a bit more stylish. Most of the work was done by Gabe, I just cleaned stuff up a bit more at the end.
2008-08-11 21:22:16 +02:00
thread = new SimpleThread(this, /* thread_num */ 0, p->workload[0],
Get rid of the unused get(Data|Inst)Asid and (inst|data)Asid functions.
2009-07-09 08:02:22 +02:00
p->itb, p->dtb);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#endif // !FULL_SYSTEM
Get rid of the Unallocated thread context state. Basically merge it in with Halted. Also had to get rid of a few other functions that called ThreadContext::deallocate(), including: - InOrderCPU's setThreadRescheduleCondition. - ThreadContext::exit(). This function was there to avoid terminating simulation when one thread out of a multi-thread workload exits, but we need to find a better (non-cpu-centric) way.
2009-04-15 22:13:47 +02:00
thread->setStatus(ThreadContext::Halted);
Fixes to get everything working again. src/cpu/simple/base.cc: Start XC's as suspended. --HG-- extra : convert_revision : f1b15a810696bee3db259a655438dce06fcb9d48
2006-06-05 04:13:42 +02:00
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
tc = thread->getTC();
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
CheckerCPU: Re-factor CheckerCPU to be compatible with current gem5 Brings the CheckerCPU back to life to allow FS and SE checking of the O3CPU. These changes have only been tested with the ARM ISA. Other ISAs potentially require modification.
2012-01-31 16:46:03 +01:00
#if USE_CHECKER
if (p->checker) {
BaseCPU *temp_checker = p->checker;
checker = dynamic_cast<CheckerCPU *>(temp_checker);
#if FULL_SYSTEM
checker->setSystem(p->system);
#endif
// Manipulate thread context
ThreadContext *cpu_tc = tc;
tc = new CheckerThreadContext<ThreadContext>(cpu_tc, this->checker);
} else {
checker = NULL;
}
#endif
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
numInst = 0;
startNumInst = 0;
numLoad = 0;
startNumLoad = 0;
lastIcacheStall = 0;
lastDcacheStall = 0;
Change ExecContext to ThreadContext. This is being renamed to differentiate between the interface used objects outside of the CPU, and the interface used by the ISA. ThreadContext is used by objects outside of the CPU and is specifically defined in thread_context.hh. ExecContext is more implicit, and is defined by files such as base_dyn_inst.hh or cpu/simple/base.hh. Further renames/reorganization will be coming shortly; what is currently CPUExecContext (the old ExecContext from m5) will be renamed to SimpleThread or something similar. src/arch/alpha/arguments.cc: src/arch/alpha/arguments.hh: src/arch/alpha/ev5.cc: src/arch/alpha/faults.cc: src/arch/alpha/faults.hh: src/arch/alpha/freebsd/system.cc: src/arch/alpha/freebsd/system.hh: src/arch/alpha/isa/branch.isa: src/arch/alpha/isa/decoder.isa: src/arch/alpha/isa/main.isa: src/arch/alpha/linux/process.cc: src/arch/alpha/linux/system.cc: src/arch/alpha/linux/system.hh: src/arch/alpha/linux/threadinfo.hh: src/arch/alpha/process.cc: src/arch/alpha/regfile.hh: src/arch/alpha/stacktrace.cc: src/arch/alpha/stacktrace.hh: src/arch/alpha/tlb.cc: src/arch/alpha/tlb.hh: src/arch/alpha/tru64/process.cc: src/arch/alpha/tru64/system.cc: src/arch/alpha/tru64/system.hh: src/arch/alpha/utility.hh: src/arch/alpha/vtophys.cc: src/arch/alpha/vtophys.hh: src/arch/mips/faults.cc: src/arch/mips/faults.hh: src/arch/mips/isa_traits.cc: src/arch/mips/isa_traits.hh: src/arch/mips/linux/process.cc: src/arch/mips/process.cc: src/arch/mips/regfile/float_regfile.hh: src/arch/mips/regfile/int_regfile.hh: src/arch/mips/regfile/misc_regfile.hh: src/arch/mips/regfile/regfile.hh: src/arch/mips/stacktrace.hh: src/arch/sparc/faults.cc: src/arch/sparc/faults.hh: src/arch/sparc/isa_traits.hh: src/arch/sparc/linux/process.cc: src/arch/sparc/linux/process.hh: src/arch/sparc/process.cc: src/arch/sparc/regfile.hh: src/arch/sparc/solaris/process.cc: src/arch/sparc/stacktrace.hh: src/arch/sparc/ua2005.cc: src/arch/sparc/utility.hh: src/arch/sparc/vtophys.cc: src/arch/sparc/vtophys.hh: src/base/remote_gdb.cc: src/base/remote_gdb.hh: src/cpu/base.cc: src/cpu/base.hh: src/cpu/base_dyn_inst.hh: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/exec_context.hh: src/cpu/cpu_exec_context.cc: src/cpu/cpu_exec_context.hh: src/cpu/cpuevent.cc: src/cpu/cpuevent.hh: src/cpu/exetrace.hh: src/cpu/intr_control.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/alpha_cpu.hh: src/cpu/o3/alpha_cpu_impl.hh: src/cpu/o3/alpha_dyn_inst_impl.hh: src/cpu/o3/commit.hh: src/cpu/o3/commit_impl.hh: src/cpu/o3/cpu.cc: src/cpu/o3/cpu.hh: src/cpu/o3/fetch_impl.hh: src/cpu/o3/regfile.hh: src/cpu/o3/thread_state.hh: src/cpu/ozone/back_end.hh: src/cpu/ozone/cpu.hh: src/cpu/ozone/cpu_impl.hh: src/cpu/ozone/front_end.hh: src/cpu/ozone/front_end_impl.hh: src/cpu/ozone/inorder_back_end.hh: src/cpu/ozone/lw_back_end.hh: src/cpu/ozone/lw_back_end_impl.hh: src/cpu/ozone/lw_lsq.hh: src/cpu/ozone/lw_lsq_impl.hh: src/cpu/ozone/thread_state.hh: src/cpu/pc_event.cc: src/cpu/pc_event.hh: src/cpu/profile.cc: src/cpu/profile.hh: src/cpu/quiesce_event.cc: src/cpu/quiesce_event.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: src/cpu/static_inst.cc: src/cpu/static_inst.hh: src/cpu/thread_state.hh: src/dev/alpha_console.cc: src/dev/ns_gige.cc: src/dev/sinic.cc: src/dev/tsunami_cchip.cc: src/kern/kernel_stats.cc: src/kern/kernel_stats.hh: src/kern/linux/events.cc: src/kern/linux/events.hh: src/kern/system_events.cc: src/kern/system_events.hh: src/kern/tru64/dump_mbuf.cc: src/kern/tru64/tru64.hh: src/kern/tru64/tru64_events.cc: src/kern/tru64/tru64_events.hh: src/mem/vport.cc: src/mem/vport.hh: src/sim/faults.cc: src/sim/faults.hh: src/sim/process.cc: src/sim/process.hh: src/sim/pseudo_inst.cc: src/sim/pseudo_inst.hh: src/sim/syscall_emul.cc: src/sim/syscall_emul.hh: src/sim/system.cc: src/cpu/thread_context.hh: src/sim/system.hh: src/sim/vptr.hh: Change ExecContext to ThreadContext. --HG-- rename : src/cpu/exec_context.hh => src/cpu/thread_context.hh extra : convert_revision : 108bb97d15a114a565a2a6a23faa554f4e2fd77e
2006-06-06 23:32:21 +02:00
threadContexts.push_back(tc);
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
CPU: Use the ThreadContext cpu id instead of the params cpu id in all cases. --HG-- extra : convert_revision : 6d025764682181b1f67df3b1d8d1d59099136df7
2007-10-18 19:15:08 +02:00
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
fetchOffset = 0;
stayAtPC = false;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
BaseSimpleCPU::~BaseSimpleCPU()
{
}
void
BaseSimpleCPU::deallocateContext(int thread_num)
{
// for now, these are equivalent
suspendContext(thread_num);
}
void
BaseSimpleCPU::haltContext(int thread_num)
{
// for now, these are equivalent
suspendContext(thread_num);
}
void
BaseSimpleCPU::regStats()
{
using namespace Stats;
BaseCPU::regStats();
numInsts
.name(name() + ".num_insts")
.desc("Number of instructions executed")
;
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
numIntAluAccesses
.name(name() + ".num_int_alu_accesses")
.desc("Number of integer alu accesses")
;
numFpAluAccesses
.name(name() + ".num_fp_alu_accesses")
.desc("Number of float alu accesses")
;
numCallsReturns
.name(name() + ".num_func_calls")
.desc("number of times a function call or return occured")
;
numCondCtrlInsts
.name(name() + ".num_conditional_control_insts")
.desc("number of instructions that are conditional controls")
;
numIntInsts
.name(name() + ".num_int_insts")
.desc("number of integer instructions")
;
numFpInsts
.name(name() + ".num_fp_insts")
.desc("number of float instructions")
;
numIntRegReads
.name(name() + ".num_int_register_reads")
.desc("number of times the integer registers were read")
;
numIntRegWrites
.name(name() + ".num_int_register_writes")
.desc("number of times the integer registers were written")
;
numFpRegReads
.name(name() + ".num_fp_register_reads")
.desc("number of times the floating registers were read")
;
numFpRegWrites
.name(name() + ".num_fp_register_writes")
.desc("number of times the floating registers were written")
;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
numMemRefs
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
.name(name()+".num_mem_refs")
.desc("number of memory refs")
;
numStoreInsts
.name(name() + ".num_store_insts")
.desc("Number of store instructions")
;
numLoadInsts
.name(name() + ".num_load_insts")
.desc("Number of load instructions")
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
;
notIdleFraction
.name(name() + ".not_idle_fraction")
.desc("Percentage of non-idle cycles")
;
idleFraction
.name(name() + ".idle_fraction")
.desc("Percentage of idle cycles")
;
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
numBusyCycles
.name(name() + ".num_busy_cycles")
.desc("Number of busy cycles")
;
numIdleCycles
.name(name()+".num_idle_cycles")
.desc("Number of idle cycles")
;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
icacheStallCycles
.name(name() + ".icache_stall_cycles")
.desc("ICache total stall cycles")
.prereq(icacheStallCycles)
;
dcacheStallCycles
.name(name() + ".dcache_stall_cycles")
.desc("DCache total stall cycles")
.prereq(dcacheStallCycles)
;
icacheRetryCycles
.name(name() + ".icache_retry_cycles")
.desc("ICache total retry cycles")
.prereq(icacheRetryCycles)
;
dcacheRetryCycles
.name(name() + ".dcache_retry_cycles")
.desc("DCache total retry cycles")
.prereq(dcacheRetryCycles)
;
idleFraction = constant(1.0) - notIdleFraction;
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
numIdleCycles = idleFraction * numCycles;
numBusyCycles = (notIdleFraction)*numCycles;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::resetStats()
{
Merge ktlim@zamp:./local/clean/o3-merge/m5 into zamp.eecs.umich.edu:/z/ktlim2/clean/o3-merge/newmem configs/boot/micro_memlat.rcS: configs/boot/micro_tlblat.rcS: src/arch/alpha/ev5.cc: src/arch/alpha/isa/decoder.isa: src/arch/alpha/isa_traits.hh: src/cpu/base.cc: src/cpu/base.hh: src/cpu/base_dyn_inst.hh: src/cpu/checker/cpu.hh: src/cpu/checker/cpu_impl.hh: src/cpu/o3/alpha/cpu_impl.hh: src/cpu/o3/alpha/params.hh: src/cpu/o3/checker_builder.cc: src/cpu/o3/commit_impl.hh: src/cpu/o3/cpu.cc: src/cpu/o3/decode_impl.hh: src/cpu/o3/fetch_impl.hh: src/cpu/o3/iew.hh: src/cpu/o3/iew_impl.hh: src/cpu/o3/inst_queue.hh: src/cpu/o3/lsq.hh: src/cpu/o3/lsq_impl.hh: src/cpu/o3/lsq_unit.hh: src/cpu/o3/lsq_unit_impl.hh: src/cpu/o3/regfile.hh: src/cpu/o3/rename_impl.hh: src/cpu/o3/thread_state.hh: src/cpu/ozone/checker_builder.cc: src/cpu/ozone/cpu.hh: src/cpu/ozone/cpu_impl.hh: src/cpu/ozone/front_end.hh: src/cpu/ozone/front_end_impl.hh: src/cpu/ozone/lw_back_end.hh: src/cpu/ozone/lw_back_end_impl.hh: src/cpu/ozone/lw_lsq.hh: src/cpu/ozone/lw_lsq_impl.hh: src/cpu/ozone/thread_state.hh: src/cpu/simple/base.cc: src/cpu/simple_thread.cc: src/cpu/simple_thread.hh: src/cpu/thread_state.hh: src/dev/ide_disk.cc: src/python/m5/objects/O3CPU.py: src/python/m5/objects/Root.py: src/python/m5/objects/System.py: src/sim/pseudo_inst.cc: src/sim/pseudo_inst.hh: src/sim/system.hh: util/m5/m5.c: Hand merge. --HG-- rename : arch/alpha/ev5.cc => src/arch/alpha/ev5.cc rename : arch/alpha/freebsd/system.cc => src/arch/alpha/freebsd/system.cc rename : arch/alpha/isa/decoder.isa => src/arch/alpha/isa/decoder.isa rename : arch/alpha/isa/mem.isa => src/arch/alpha/isa/mem.isa rename : arch/alpha/isa_traits.hh => src/arch/alpha/isa_traits.hh rename : arch/alpha/linux/system.cc => src/arch/alpha/linux/system.cc rename : arch/alpha/system.cc => src/arch/alpha/system.cc rename : arch/alpha/tru64/system.cc => src/arch/alpha/tru64/system.cc rename : cpu/base.cc => src/cpu/base.cc rename : cpu/base.hh => src/cpu/base.hh rename : cpu/base_dyn_inst.hh => src/cpu/base_dyn_inst.hh rename : cpu/checker/cpu.hh => src/cpu/checker/cpu.hh rename : cpu/checker/cpu.cc => src/cpu/checker/cpu_impl.hh rename : cpu/o3/alpha_cpu_builder.cc => src/cpu/o3/alpha/cpu_builder.cc rename : cpu/checker/o3_cpu_builder.cc => src/cpu/o3/checker_builder.cc rename : cpu/o3/commit_impl.hh => src/cpu/o3/commit_impl.hh rename : cpu/o3/cpu.cc => src/cpu/o3/cpu.cc rename : cpu/o3/fetch_impl.hh => src/cpu/o3/fetch_impl.hh rename : cpu/o3/iew.hh => src/cpu/o3/iew.hh rename : cpu/o3/iew_impl.hh => src/cpu/o3/iew_impl.hh rename : cpu/o3/inst_queue.hh => src/cpu/o3/inst_queue.hh rename : cpu/o3/inst_queue_impl.hh => src/cpu/o3/inst_queue_impl.hh rename : cpu/o3/lsq_impl.hh => src/cpu/o3/lsq_impl.hh rename : cpu/o3/lsq_unit.hh => src/cpu/o3/lsq_unit.hh rename : cpu/o3/lsq_unit_impl.hh => src/cpu/o3/lsq_unit_impl.hh rename : cpu/o3/mem_dep_unit_impl.hh => src/cpu/o3/mem_dep_unit_impl.hh rename : cpu/o3/rename.hh => src/cpu/o3/rename.hh rename : cpu/o3/rename_impl.hh => src/cpu/o3/rename_impl.hh rename : cpu/o3/thread_state.hh => src/cpu/o3/thread_state.hh rename : cpu/o3/tournament_pred.cc => src/cpu/o3/tournament_pred.cc rename : cpu/o3/tournament_pred.hh => src/cpu/o3/tournament_pred.hh rename : cpu/checker/cpu_builder.cc => src/cpu/ozone/checker_builder.cc rename : cpu/ozone/cpu.hh => src/cpu/ozone/cpu.hh rename : cpu/ozone/cpu_builder.cc => src/cpu/ozone/cpu_builder.cc rename : cpu/ozone/cpu_impl.hh => src/cpu/ozone/cpu_impl.hh rename : cpu/ozone/front_end.hh => src/cpu/ozone/front_end.hh rename : cpu/ozone/front_end_impl.hh => src/cpu/ozone/front_end_impl.hh rename : cpu/ozone/inorder_back_end_impl.hh => src/cpu/ozone/inorder_back_end_impl.hh rename : cpu/ozone/inst_queue_impl.hh => src/cpu/ozone/inst_queue_impl.hh rename : cpu/ozone/lw_back_end.hh => src/cpu/ozone/lw_back_end.hh rename : cpu/ozone/lw_back_end_impl.hh => src/cpu/ozone/lw_back_end_impl.hh rename : cpu/ozone/lw_lsq.hh => src/cpu/ozone/lw_lsq.hh rename : cpu/ozone/lw_lsq_impl.hh => src/cpu/ozone/lw_lsq_impl.hh rename : cpu/ozone/simple_params.hh => src/cpu/ozone/simple_params.hh rename : cpu/ozone/thread_state.hh => src/cpu/ozone/thread_state.hh rename : cpu/simple/cpu.cc => src/cpu/simple/base.cc rename : cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : cpu/thread_state.hh => src/cpu/thread_state.hh rename : dev/ide_disk.hh => src/dev/ide_disk.hh rename : python/m5/objects/BaseCPU.py => src/python/m5/objects/BaseCPU.py rename : python/m5/objects/AlphaFullCPU.py => src/python/m5/objects/O3CPU.py rename : python/m5/objects/OzoneCPU.py => src/python/m5/objects/OzoneCPU.py rename : python/m5/objects/Root.py => src/python/m5/objects/Root.py rename : python/m5/objects/System.py => src/python/m5/objects/System.py rename : sim/eventq.hh => src/sim/eventq.hh rename : sim/pseudo_inst.cc => src/sim/pseudo_inst.cc rename : sim/pseudo_inst.hh => src/sim/pseudo_inst.hh rename : sim/serialize.cc => src/sim/serialize.cc rename : sim/stat_control.cc => src/sim/stat_control.cc rename : sim/stat_control.hh => src/sim/stat_control.hh rename : sim/system.hh => src/sim/system.hh extra : convert_revision : 135d90e43f6cea89f9460ba4e23f4b0b85886e7d
2006-10-01 05:43:23 +02:00
// startNumInst = numInst;
After a checkpoint (and thus a stats reset), the not_idle_fraction/notIdleFraction statistic is really wrong. The notIdleFraction statistic isn't updated when the statistics reset, probably because the cpu Status information was pulled into the atomic and timing cpus. This changeset pulls Status back into the BaseSimpleCPU object. Anyone care to comment on the odd naming of the Status instance? It shouldn't just be status because that is confusing with Port::Status, but _status seems a bit strage too.
2008-07-01 16:24:09 +02:00
notIdleFraction = (_status != Idle);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::serialize(ostream &os)
{
After a checkpoint (and thus a stats reset), the not_idle_fraction/notIdleFraction statistic is really wrong. The notIdleFraction statistic isn't updated when the statistics reset, probably because the cpu Status information was pulled into the atomic and timing cpus. This changeset pulls Status back into the BaseSimpleCPU object. Anyone care to comment on the odd naming of the Status instance? It shouldn't just be status because that is confusing with Port::Status, but _status seems a bit strage too.
2008-07-01 16:24:09 +02:00
SERIALIZE_ENUM(_status);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
BaseCPU::serialize(os);
Updates for serialization. As long as the tickEvent doesn't need to be serialized (I don't believe it does because we drain all CPUs prior to checkpointing), it should be feasible to start up from other CPU's checkpoints. src/cpu/simple/atomic.cc: src/cpu/simple/atomic.hh: src/cpu/simple/base.cc: src/cpu/simple/timing.cc: src/cpu/simple_thread.cc: Updates for serialization. --HG-- extra : convert_revision : 0f150de75d4bc833e4c9b83568e7fd22688d5727
2006-07-12 23:11:57 +02:00
// SERIALIZE_SCALAR(inst);
nameOut(os, csprintf("%s.xc.0", name()));
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
thread->serialize(os);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::unserialize(Checkpoint *cp, const string &section)
{
After a checkpoint (and thus a stats reset), the not_idle_fraction/notIdleFraction statistic is really wrong. The notIdleFraction statistic isn't updated when the statistics reset, probably because the cpu Status information was pulled into the atomic and timing cpus. This changeset pulls Status back into the BaseSimpleCPU object. Anyone care to comment on the odd naming of the Status instance? It shouldn't just be status because that is confusing with Port::Status, but _status seems a bit strage too.
2008-07-01 16:24:09 +02:00
UNSERIALIZE_ENUM(_status);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
BaseCPU::unserialize(cp, section);
Updates for serialization. As long as the tickEvent doesn't need to be serialized (I don't believe it does because we drain all CPUs prior to checkpointing), it should be feasible to start up from other CPU's checkpoints. src/cpu/simple/atomic.cc: src/cpu/simple/atomic.hh: src/cpu/simple/base.cc: src/cpu/simple/timing.cc: src/cpu/simple_thread.cc: Updates for serialization. --HG-- extra : convert_revision : 0f150de75d4bc833e4c9b83568e7fd22688d5727
2006-07-12 23:11:57 +02:00
// UNSERIALIZE_SCALAR(inst);
thread->unserialize(cp, csprintf("%s.xc.0", section));
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
types: add a type for thread IDs and try to use it everywhere
2009-05-26 18:23:13 +02:00
change_thread_state(ThreadID tid, int activate, int priority)
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
{
}
#if FULL_SYSTEM
Addr
BaseSimpleCPU::dbg_vtophys(Addr addr)
{
Change ExecContext to ThreadContext. This is being renamed to differentiate between the interface used objects outside of the CPU, and the interface used by the ISA. ThreadContext is used by objects outside of the CPU and is specifically defined in thread_context.hh. ExecContext is more implicit, and is defined by files such as base_dyn_inst.hh or cpu/simple/base.hh. Further renames/reorganization will be coming shortly; what is currently CPUExecContext (the old ExecContext from m5) will be renamed to SimpleThread or something similar. src/arch/alpha/arguments.cc: src/arch/alpha/arguments.hh: src/arch/alpha/ev5.cc: src/arch/alpha/faults.cc: src/arch/alpha/faults.hh: src/arch/alpha/freebsd/system.cc: src/arch/alpha/freebsd/system.hh: src/arch/alpha/isa/branch.isa: src/arch/alpha/isa/decoder.isa: src/arch/alpha/isa/main.isa: src/arch/alpha/linux/process.cc: src/arch/alpha/linux/system.cc: src/arch/alpha/linux/system.hh: src/arch/alpha/linux/threadinfo.hh: src/arch/alpha/process.cc: src/arch/alpha/regfile.hh: src/arch/alpha/stacktrace.cc: src/arch/alpha/stacktrace.hh: src/arch/alpha/tlb.cc: src/arch/alpha/tlb.hh: src/arch/alpha/tru64/process.cc: src/arch/alpha/tru64/system.cc: src/arch/alpha/tru64/system.hh: src/arch/alpha/utility.hh: src/arch/alpha/vtophys.cc: src/arch/alpha/vtophys.hh: src/arch/mips/faults.cc: src/arch/mips/faults.hh: src/arch/mips/isa_traits.cc: src/arch/mips/isa_traits.hh: src/arch/mips/linux/process.cc: src/arch/mips/process.cc: src/arch/mips/regfile/float_regfile.hh: src/arch/mips/regfile/int_regfile.hh: src/arch/mips/regfile/misc_regfile.hh: src/arch/mips/regfile/regfile.hh: src/arch/mips/stacktrace.hh: src/arch/sparc/faults.cc: src/arch/sparc/faults.hh: src/arch/sparc/isa_traits.hh: src/arch/sparc/linux/process.cc: src/arch/sparc/linux/process.hh: src/arch/sparc/process.cc: src/arch/sparc/regfile.hh: src/arch/sparc/solaris/process.cc: src/arch/sparc/stacktrace.hh: src/arch/sparc/ua2005.cc: src/arch/sparc/utility.hh: src/arch/sparc/vtophys.cc: src/arch/sparc/vtophys.hh: src/base/remote_gdb.cc: src/base/remote_gdb.hh: src/cpu/base.cc: src/cpu/base.hh: src/cpu/base_dyn_inst.hh: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/exec_context.hh: src/cpu/cpu_exec_context.cc: src/cpu/cpu_exec_context.hh: src/cpu/cpuevent.cc: src/cpu/cpuevent.hh: src/cpu/exetrace.hh: src/cpu/intr_control.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/alpha_cpu.hh: src/cpu/o3/alpha_cpu_impl.hh: src/cpu/o3/alpha_dyn_inst_impl.hh: src/cpu/o3/commit.hh: src/cpu/o3/commit_impl.hh: src/cpu/o3/cpu.cc: src/cpu/o3/cpu.hh: src/cpu/o3/fetch_impl.hh: src/cpu/o3/regfile.hh: src/cpu/o3/thread_state.hh: src/cpu/ozone/back_end.hh: src/cpu/ozone/cpu.hh: src/cpu/ozone/cpu_impl.hh: src/cpu/ozone/front_end.hh: src/cpu/ozone/front_end_impl.hh: src/cpu/ozone/inorder_back_end.hh: src/cpu/ozone/lw_back_end.hh: src/cpu/ozone/lw_back_end_impl.hh: src/cpu/ozone/lw_lsq.hh: src/cpu/ozone/lw_lsq_impl.hh: src/cpu/ozone/thread_state.hh: src/cpu/pc_event.cc: src/cpu/pc_event.hh: src/cpu/profile.cc: src/cpu/profile.hh: src/cpu/quiesce_event.cc: src/cpu/quiesce_event.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: src/cpu/static_inst.cc: src/cpu/static_inst.hh: src/cpu/thread_state.hh: src/dev/alpha_console.cc: src/dev/ns_gige.cc: src/dev/sinic.cc: src/dev/tsunami_cchip.cc: src/kern/kernel_stats.cc: src/kern/kernel_stats.hh: src/kern/linux/events.cc: src/kern/linux/events.hh: src/kern/system_events.cc: src/kern/system_events.hh: src/kern/tru64/dump_mbuf.cc: src/kern/tru64/tru64.hh: src/kern/tru64/tru64_events.cc: src/kern/tru64/tru64_events.hh: src/mem/vport.cc: src/mem/vport.hh: src/sim/faults.cc: src/sim/faults.hh: src/sim/process.cc: src/sim/process.hh: src/sim/pseudo_inst.cc: src/sim/pseudo_inst.hh: src/sim/syscall_emul.cc: src/sim/syscall_emul.hh: src/sim/system.cc: src/cpu/thread_context.hh: src/sim/system.hh: src/sim/vptr.hh: Change ExecContext to ThreadContext. --HG-- rename : src/cpu/exec_context.hh => src/cpu/thread_context.hh extra : convert_revision : 108bb97d15a114a565a2a6a23faa554f4e2fd77e
2006-06-06 23:32:21 +02:00
return vtophys(tc, addr);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
#endif // FULL_SYSTEM
#if FULL_SYSTEM
void
cpu: provide a wakeup mechanism that can be used to pull CPUs out of sleep. Make interrupts use the new wakeup method, and pull all of the interrupt stuff into the cpu base class so that only the wakeup code needs to be updated. I tried to make wakeup, wakeCPU, and the various other mechanisms for waking and sleeping a little more sane, but I couldn't understand why the statistics were changing the way they were. Maybe we'll try again some day.
2009-01-24 16:27:21 +01:00
BaseSimpleCPU::wakeup()
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
{
cpu: provide a wakeup mechanism that can be used to pull CPUs out of sleep. Make interrupts use the new wakeup method, and pull all of the interrupt stuff into the cpu base class so that only the wakeup code needs to be updated. I tried to make wakeup, wakeCPU, and the various other mechanisms for waking and sleeping a little more sane, but I couldn't understand why the statistics were changing the way they were. Maybe we'll try again some day.
2009-01-24 16:27:21 +01:00
if (thread->status() != ThreadContext::Suspended)
return;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
cpu: provide a wakeup mechanism that can be used to pull CPUs out of sleep. Make interrupts use the new wakeup method, and pull all of the interrupt stuff into the cpu base class so that only the wakeup code needs to be updated. I tried to make wakeup, wakeCPU, and the various other mechanisms for waking and sleeping a little more sane, but I couldn't understand why the statistics were changing the way they were. Maybe we'll try again some day.
2009-01-24 16:27:21 +01:00
DPRINTF(Quiesce,"Suspended Processor awoke\n");
thread->activate();
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
#endif // FULL_SYSTEM
void
BaseSimpleCPU::checkForInterrupts()
{
#if FULL_SYSTEM
style: Use the correct m5 style for things relating to interrupts.
2008-10-21 16:12:53 +02:00
if (checkInterrupts(tc)) {
Turn Interrupts objects into SimObjects. Also, move local APIC state into x86's Interrupts object.
2008-10-12 18:09:56 +02:00
Fault interrupt = interrupts->getInterrupt(tc);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
Add a new file which describes an ISA's interrupt handling mechanism. It records when interrupts are requested, and returns an interrupt to execute if the --HG-- extra : convert_revision : c535000a6a170caefd441687b60f940513d29739
2006-11-03 08:25:39 +01:00
if (interrupt != NoFault) {
CPU: Reset fetch offset after a exception
2010-06-02 19:58:12 +02:00
fetchOffset = 0;
Turn Interrupts objects into SimObjects. Also, move local APIC state into x86's Interrupts object.
2008-10-12 18:09:56 +02:00
interrupts->updateIntrInfo(tc);
Add a new file which describes an ISA's interrupt handling mechanism. It records when interrupts are requested, and returns an interrupt to execute if the --HG-- extra : convert_revision : c535000a6a170caefd441687b60f940513d29739
2006-11-03 08:25:39 +01:00
interrupt->invoke(tc);
ARM: Implement support for the IT instruction and the ITSTATE bits of CPSR.
2010-06-02 19:58:16 +02:00
predecoder.reset();
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
}
#endif
}
CPU: Implement translateTiming which defers to translateAtomic, and convert the timing simple CPU to use it.
2009-02-25 19:16:15 +01:00
void
Minor further cleanup & commenting of Packet class. src/cpu/simple/atomic.cc: Make common ifetch setup based on Request rather than Packet. Packet::reset() no longer a separate function. sendAtomic() returns latency, not absolute tick. src/cpu/simple/atomic.hh: sendAtomic returns latency, not absolute tick. src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Make common ifetch setup based on Request rather than Packet. src/dev/alpha_console.cc: src/dev/ide_ctrl.cc: src/dev/io_device.cc: src/dev/isa_fake.cc: src/dev/ns_gige.cc: src/dev/pciconfigall.cc: src/dev/sinic.cc: src/dev/tsunami_cchip.cc: src/dev/tsunami_io.cc: src/dev/tsunami_pchip.cc: src/dev/uart8250.cc: src/mem/physical.cc: Get rid of redundant Packet time field. src/mem/packet.cc: Eliminate reset() method. src/mem/packet.hh: Fold reset() function into reinitFromRequest()... it was only ever called together with that function. Get rid of redundant time field. Cleanup/add comments. src/mem/port.hh: Document in comment that sendAtomic returns latency, not absolute tick. --HG-- extra : convert_revision : 0252f1a294043ca3ed58f437232ad24fc0733e0c
2006-05-31 04:30:42 +02:00
BaseSimpleCPU::setupFetchRequest(Request *req)
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +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
Addr instAddr = thread->instAddr();
Assign traceData to be NULL at BaseSimpleCPU constructor. Initialize a temporary variable for thread->readPC() at setupFetchRequest() to reduce function calls. exec tracing isn't needed for m5.fast binaries Moved MISCREG_GL, MISCREG_CWP, and MISCREG_TLB_DATA out of switch statement and use if blocks instead. src/arch/sparc/miscregfile.cc: Moved MISCREG_GL, MISCREG_CWP, and MISCREG_TLB_DATA out of switch statement and use if blocks instead. src/cpu/simple/base.cc: Assign traceData to be NULL at BaseSimpleCPU constructor. Initialize a temporary variable for thread->readPC() at setupFetchRequest() to reduce function calls. exec tracing isn't needed for m5.fast binaries --HG-- extra : convert_revision : 5dc92fff05c9bde994f1e0f1bb40e11c44eb72c6
2007-05-31 22:01:41 +02:00
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
// set up memory request for instruction fetch
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
DPRINTF(Fetch, "Fetch: PC:%08p\n", instAddr);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +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
Addr fetchPC = (instAddr & PCMask) + fetchOffset;
req->setVirt(0, fetchPC, sizeof(MachInst), Request::INST_FETCH, instAddr);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::preExecute()
{
// maintain $r0 semantics
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
thread->setIntReg(ZeroReg, 0);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#if THE_ISA == ALPHA_ISA
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
thread->setFloatReg(ZeroReg, 0.0);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#endif // ALPHA_ISA
// check for instruction-count-based events
comInstEventQueue[0]->serviceEvents(numInst);
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
system->instEventQueue.serviceEvents(system->totalNumInsts);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
// decode the instruction
inst = gtoh(inst);
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +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
TheISA::PCState pcState = thread->pcState();
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +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
if (isRomMicroPC(pcState.microPC())) {
CPU: Make the highest order bit in the micro pc determine if it's combinational or from the ROM.
2008-10-13 01:59:55 +02:00
stayAtPC = false;
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
curStaticInst = microcodeRom.fetchMicroop(pcState.microPC(),
curMacroStaticInst);
CPU: Make the highest order bit in the micro pc determine if it's combinational or from the ROM.
2008-10-13 01:59:55 +02:00
} else if (!curMacroStaticInst) {
//We're not in the middle of a macro instruction
Replaced makeExtMI with predecode. Removed the getOpcode function from StaticInst which only made sense for Alpha. Started implementing the x86 predecoder. --HG-- extra : convert_revision : a13ea257c8943ef25e9bc573024a99abacf4a70d
2007-03-13 17:13:21 +01:00
StaticInstPtr instPtr = NULL;
//Predecode, ie bundle up an ExtMachInst
Make the predecoder an object with it's own switched header file. Start adding predecoding functionality to x86. src/arch/SConscript: src/arch/alpha/utility.hh: src/arch/mips/utility.hh: src/arch/sparc/utility.hh: src/cpu/base.hh: src/cpu/o3/fetch.hh: src/cpu/o3/fetch_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/static_inst.hh: src/arch/alpha/predecoder.hh: src/arch/mips/predecoder.hh: src/arch/sparc/predecoder.hh: Make the predecoder an object with it's own switched header file. --HG-- extra : convert_revision : 77206e29089130e86b97164c30022a062699ba86
2007-03-15 03:47:42 +01:00
//This should go away once the constructor can be set up properly
predecoder.setTC(thread->getTC());
//If more fetch data is needed, pass it in.
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 fetchPC = (pcState.instAddr() & PCMask) + fetchOffset;
Make branches work by repopulating the predecoder every time through. This is probably fine as far as the predecoder goes, but the simple cpu might want to not refetch something it already has. That reintroduces the self modifying code problem though. --HG-- extra : convert_revision : 802197e65f8dc1ad657c6b346091e03cb563b0c0
2007-06-19 20:17:34 +02:00
//if(predecoder.needMoreBytes())
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
predecoder.moreBytes(pcState, fetchPC, inst);
Make branches work by repopulating the predecoder every time through. This is probably fine as far as the predecoder goes, but the simple cpu might want to not refetch something it already has. That reintroduces the self modifying code problem though. --HG-- extra : convert_revision : 802197e65f8dc1ad657c6b346091e03cb563b0c0
2007-06-19 20:17:34 +02:00
//else
// predecoder.process();
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
//If an instruction is ready, decode it. Otherwise, we'll have to
//fetch beyond the MachInst at the current pc.
if (predecoder.extMachInstReady()) {
stayAtPC = false;
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
ExtMachInst machInst = predecoder.getExtMachInst(pcState);
thread->pcState(pcState);
Decode: Pull instruction decoding out of the StaticInst class into its own. This change pulls the instruction decoding machinery (including caches) out of the StaticInst class and puts it into its own class. This has a few intrinsic benefits. First, the StaticInst code, which has gotten to be quite large, gets simpler. Second, the code that handles decode caching is now separated out into its own component and can be looked at in isolation, making it easier to understand. I took the opportunity to restructure the code a bit which will hopefully also help. Beyond that, this change also lays some ground work for each ISA to have its own, potentially stateful decode object. We'd be able to include less contextualizing information in the ExtMachInst objects since that context would be applied at the decoder. Also, the decoder could "know" ahead of time that all the instructions it's going to see are going to be, for instance, 64 bit mode, and it will have one less thing to check when it decodes them. Because the decode caching mechanism has been separated out, it's now possible to have multiple caches which correspond to different types of decoding context. Having one cache for each element of the cross product of different configurations may become prohibitive, so it may be desirable to clear out the cache when relatively static state changes and not to have one for each setting. Because the decode function is no longer universally accessible as a static member of the StaticInst class, a new function was added to the ThreadContexts that returns the applicable decode object.
2011-09-09 11:30:01 +02:00
instPtr = thread->decoder.decode(machInst, pcState.instAddr());
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
} else {
stayAtPC = true;
fetchOffset += sizeof(MachInst);
}
Replaced makeExtMI with predecode. Removed the getOpcode function from StaticInst which only made sense for Alpha. Started implementing the x86 predecoder. --HG-- extra : convert_revision : a13ea257c8943ef25e9bc573024a99abacf4a70d
2007-03-13 17:13:21 +01:00
//If we decoded an instruction and it's microcoded, start pulling
//out micro ops
Make microOp vs microop and macroOp vs macroop capitilization consistent. src/arch/x86/isa/macroop.isa: Make microOp vs microop and macroOp vs macroop capitilization consistent. Also fill out the emulation environment handling a little more, and use an object to pass around output code. src/arch/x86/isa/microops/base.isa: Make microOp vs microop and macroOp vs macroop capitilization consistent. Also adjust python to C++ bool translation. --HG-- extra : convert_revision : 6f4bacfa334c42732c845f9a7f211cbefc73f96f
2007-06-12 18:21:47 +02:00
if (instPtr && instPtr->isMacroop()) {
Started implementing microcode. --HG-- extra : convert_revision : 51df0454085e13df023efd8a0c0a12f9756c4690
2006-10-16 03:04:14 +02:00
curMacroStaticInst = instPtr;
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
curStaticInst = curMacroStaticInst->fetchMicroop(pcState.microPC());
Fix up microcode support. src/arch/sparc/isa/formats/blockmem.isa: Several small and medium bug fixes. src/cpu/simple/base.cc: Fixed a few compiler errors and made sure the next micro pc is set to 1 to prevent the first microop from executing twice. Also fixed a fetching bug. src/cpu/thread_state.cc: Made sure the microPC and nextMicroPC are initialized properly. --HG-- extra : convert_revision : a0fc8aa18d1ade916f17c557181a793c6108a8af
2006-10-16 21:56:46 +02:00
} else {
curStaticInst = instPtr;
Started implementing microcode. --HG-- extra : convert_revision : 51df0454085e13df023efd8a0c0a12f9756c4690
2006-10-16 03:04:14 +02:00
}
} else {
//Read the next micro op from the macro op
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
curStaticInst = curMacroStaticInst->fetchMicroop(pcState.microPC());
Started implementing microcode. --HG-- extra : convert_revision : 51df0454085e13df023efd8a0c0a12f9756c4690
2006-10-16 03:04:14 +02:00
}
Replaced makeExtMI with predecode. Removed the getOpcode function from StaticInst which only made sense for Alpha. Started implementing the x86 predecoder. --HG-- extra : convert_revision : a13ea257c8943ef25e9bc573024a99abacf4a70d
2007-03-13 17:13:21 +01:00
//If we decoded an instruction this "tick", record information about it.
if(curStaticInst)
{
don't be so aggressive with the tracing on #if --HG-- extra : convert_revision : 8ee88bff8010dcb7a412f6a6b49d40fad1c0bb68
2007-06-04 21:53:04 +02:00
#if TRACING_ON
Replace curTick global variable with accessor functions. This step makes it easy to replace the accessor functions (which still access a global variable) with ones that access per-thread curTick values.
2011-01-08 06:50:29 +01:00
traceData = tracer->getInstRecord(curTick(), tc,
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
curStaticInst, thread->pcState(), curMacroStaticInst);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
Replaced makeExtMI with predecode. Removed the getOpcode function from StaticInst which only made sense for Alpha. Started implementing the x86 predecoder. --HG-- extra : convert_revision : a13ea257c8943ef25e9bc573024a99abacf4a70d
2007-03-13 17:13:21 +01:00
DPRINTF(Decode,"Decode: Decoded %s instruction: 0x%x\n",
curStaticInst->getName(), curStaticInst->machInst);
don't be so aggressive with the tracing on #if --HG-- extra : convert_revision : 8ee88bff8010dcb7a412f6a6b49d40fad1c0bb68
2007-06-04 21:53:04 +02:00
#endif // TRACING_ON
Replaced makeExtMI with predecode. Removed the getOpcode function from StaticInst which only made sense for Alpha. Started implementing the x86 predecoder. --HG-- extra : convert_revision : a13ea257c8943ef25e9bc573024a99abacf4a70d
2007-03-13 17:13:21 +01:00
}
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::postExecute()
{
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
assert(curStaticInst);
TheISA::PCState pc = tc->pcState();
Addr instAddr = pc.instAddr();
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
#if FULL_SYSTEM
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
if (thread->profile) {
Move the check to see if you're in user mode into the isa directory. --HG-- extra : convert_revision : b5b7cdf4a5e5e54228c592093516bf18d0f7dbe6
2006-11-08 19:55:48 +01:00
bool usermode = TheISA::inUserMode(tc);
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
thread->profilePC = usermode ? 1 : instAddr;
X86: Get X86_FS to compile. --HG-- extra : convert_revision : fb973bcf13648876d5691231845dd47a2be50f01
2007-09-25 02:39:56 +02:00
ProfileNode *node = thread->profile->consume(tc, curStaticInst);
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
if (node)
Reorganization/renaming of CPUExecContext. Now it is called SimpleThread in order to clear up the confusion due to the many ExecContexts. It also derives from a common ThreadState object, which holds various state common to threads across CPU models. Following with the previous check-in, ExecContext now refers only to the interface provided to the ISA in order to access CPU state. ThreadContext refers to the interface provided to all objects outside the CPU in order to access thread state. SimpleThread provides all thread state and the interface to access it, and is suitable for simple execution models such as the SimpleCPU. src/SConscript: Include thread state file. src/arch/alpha/ev5.cc: src/cpu/checker/cpu.cc: src/cpu/checker/cpu.hh: src/cpu/checker/thread_context.hh: src/cpu/memtest/memtest.cc: src/cpu/memtest/memtest.hh: src/cpu/o3/cpu.cc: src/cpu/ozone/cpu_impl.hh: src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: src/cpu/simple/base.hh: src/cpu/simple/timing.cc: Rename CPUExecContext to SimpleThread. src/cpu/base_dyn_inst.hh: Make thread member variables protected.. src/cpu/o3/alpha_cpu.hh: src/cpu/o3/cpu.hh: Make various members of ThreadState protected. src/cpu/o3/alpha_cpu_impl.hh: Push generation of TranslatingPort into the CPU itself. Make various members of ThreadState protected. src/cpu/o3/thread_state.hh: Pull a lot of common code into the base ThreadState class. src/cpu/ozone/thread_state.hh: Rename CPUExecContext to SimpleThread, move a lot of common code into base ThreadState class. src/cpu/thread_state.hh: Push a lot of common code into base ThreadState class. This goes along with renaming CPUExecContext to SimpleThread, and making it derive from ThreadState. src/cpu/simple_thread.cc: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. This helps push a lot of common code/state into a single class that can be used by all CPUs. src/cpu/simple_thread.hh: Rename CPUExecContext to SimpleThread, make it derive from ThreadState. src/kern/system_events.cc: Rename cpu_exec_context to thread_context. src/sim/process.hh: Remove unused forward declaration. --HG-- rename : src/cpu/cpu_exec_context.cc => src/cpu/simple_thread.cc rename : src/cpu/cpu_exec_context.hh => src/cpu/simple_thread.hh extra : convert_revision : 2ed617aa80b64016cb9270f75352607cca032733
2006-06-07 21:29:53 +02:00
thread->profileNode = node;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
#endif
if (curStaticInst->isMemRef()) {
numMemRefs++;
}
if (curStaticInst->isLoad()) {
++numLoad;
comLoadEventQueue[0]->serviceEvents(numLoad);
}
CPA: Add code to automatically record function symbols as CPU executes.
2009-02-27 01:29:17 +01:00
if (CPA::available()) {
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
CPA::cpa()->swAutoBegin(tc, pc.nextInstAddr());
CPA: Add code to automatically record function symbols as CPU executes.
2009-02-27 01:29:17 +01:00
}
mcpat: Adds McPAT performance counters Updated patches from Rick Strong's set that modify performance counters for McPAT
2011-02-07 07:14:17 +01:00
/* Power model statistics */
//integer alu accesses
if (curStaticInst->isInteger()){
numIntAluAccesses++;
numIntInsts++;
}
//float alu accesses
if (curStaticInst->isFloating()){
numFpAluAccesses++;
numFpInsts++;
}
//number of function calls/returns to get window accesses
if (curStaticInst->isCall() || curStaticInst->isReturn()){
numCallsReturns++;
}
//the number of branch predictions that will be made
if (curStaticInst->isCondCtrl()){
numCondCtrlInsts++;
}
//result bus acceses
if (curStaticInst->isLoad()){
numLoadInsts++;
}
if (curStaticInst->isStore()){
numStoreInsts++;
}
/* End power model statistics */
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
traceFunctions(instAddr);
Fixes for TimingSimpleCPU under full system. Now boots Alpha Linux! src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: Move traceData->finalize() into postExecute(). src/cpu/simple/timing.cc: Fixes for full system. Now boots Alpha Linux! - Handle ifetch faults, suspend/resume. - Delete memory request & packet objects on response. - Don't try to do split memory accesses on prefetch references (ISA description doesn't support this). src/cpu/simple/timing.hh: Minor reorganization of internal methods. --HG-- extra : convert_revision : 59e3ee5e4cb53c424ebdbe2e504d97e88c08a978
2006-05-26 20:33:43 +02:00
if (traceData) {
Clean up tracing stuff more, get rid of the trace log since its not all that useful. Fix a few bugs with python/C++ integration. --HG-- extra : convert_revision : a706512f7dc8b0c88f1ff96fe35ab8fbf9548b78
2007-02-11 00:14:50 +01:00
traceData->dump();
delete traceData;
traceData = NULL;
Fixes for TimingSimpleCPU under full system. Now boots Alpha Linux! src/cpu/simple/atomic.cc: src/cpu/simple/base.cc: Move traceData->finalize() into postExecute(). src/cpu/simple/timing.cc: Fixes for full system. Now boots Alpha Linux! - Handle ifetch faults, suspend/resume. - Delete memory request & packet objects on response. - Don't try to do split memory accesses on prefetch references (ISA description doesn't support this). src/cpu/simple/timing.hh: Minor reorganization of internal methods. --HG-- extra : convert_revision : 59e3ee5e4cb53c424ebdbe2e504d97e88c08a978
2006-05-26 20:33:43 +02:00
}
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
void
BaseSimpleCPU::advancePC(Fault fault)
{
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
//Since we're moving to a new pc, zero out the offset
fetchOffset = 0;
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
if (fault != NoFault) {
In the case that we generate a fault (e.g. a tlb miss) on a microcoded instruction set curMacroStaticInst to null This way we'll jump immediately to the handler --HG-- extra : convert_revision : 36218d3a5c2342337e66e1229ea2219533efd41e
2007-01-17 01:12:33 +01:00
curMacroStaticInst = StaticInst::nullStaticInstPtr;
Faults: Pass the StaticInst involved, if any, to a Fault's invoke method. Also move the "Fault" reference counted pointer type into a separate file, sim/fault.hh. It would be better to name this less similarly to sim/faults.hh to reduce confusion, but fault.hh matches the name of the type. We could change Fault to FaultPtr to match other pointer types, and then changing the name of the file would make more sense.
2010-09-14 04:26:03 +02:00
fault->invoke(tc, curStaticInst);
ARM: Implement support for the IT instruction and the ITSTATE bits of CPSR.
2010-06-02 19:58:16 +02:00
predecoder.reset();
Changes to make simple cpu handle pcs appropriately for x86 --HG-- extra : convert_revision : cf68886d53301e0a63705247bd7d66b2ff08ea84
2007-05-18 19:42:50 +02:00
} else {
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
if (curStaticInst) {
if (curStaticInst->isLastMicroop())
curMacroStaticInst = StaticInst::nullStaticInstPtr;
TheISA::PCState pcState = thread->pcState();
TheISA::advancePC(pcState, curStaticInst);
thread->pcState(pcState);
Started implementing microcode. --HG-- extra : convert_revision : 51df0454085e13df023efd8a0c0a12f9756c4690
2006-10-16 03:04:14 +02:00
}
Split SimpleCPU into two different models, AtomicSimpleCPU and TimingSimpleCPU, which use atomic and timing memory accesses respectively. Common code is factored into the BaseSimpleCPU class. AtomicSimpleCPU includes an option (simulate_stalls) to add delays based on the estimated latency reported by the atomic accesses. Plain old "SimpleCPU" is gone; I have not updated all the config files (just test/test.py). Also fixes to get timing accesses working in new memory model and to get split-phase memory instruction definitions working with new memory model as well. arch/alpha/isa/main.isa: Need to include packet_impl.h for functions that use Packet objects. arch/alpha/isa/mem.isa: Change completeAcc() methods to take Packet object pointers. Also split out StoreCond template for completeAcc(), since that's the only one that needs write_result and we get an unused variable warning if we always have it in there. build/SConstruct: Update list of recognized CPU model names. configs/test/test.py: Change SimpleCPU to AtomicSimpleCPU. cpu/SConscript: Define sources for new CPU models. Add split memory access methods to CPU model signatures. cpu/cpu_models.py: cpu/static_inst.hh: Define new CPU models. cpu/simple/base.cc: cpu/simple/base.hh: Factor out pieces specific to Atomic or Timing models. mem/bus.cc: Bus needs to be able to route timing packets based on explicit dest so responses can get back to requester. Set dest to Packet::Broadcast to indicate that dest should be derived from address. Also set packet src field based on port from which packet is sent. mem/bus.hh: Set packet src field based on port from which packet is sent. mem/packet.hh: Define Broadcast destination address to indicate that packet should be routed based on address. mem/physical.cc: Set packet dest on response so packet is routed back to requester properly. mem/port.cc: Flag blob packets as Broadcast. python/m5/objects/PhysicalMemory.py: Change default latency to be 1 cycle. --HG-- rename : cpu/simple/cpu.cc => cpu/simple/base.cc rename : cpu/simple/cpu.hh => cpu/simple/base.hh extra : convert_revision : e9646af6406a20c8c605087936dc4683375c2132
2006-05-16 23:36:50 +02:00
}
}
Get MIPS simple regression working. Take out unecessary functions "setShadowSet", "CacheOp" --HG-- extra : convert_revision : a9ae8a7e62c27c2db16fd3cfa7a7f0bf5f0bf8ea
2007-11-15 09:10:41 +01:00
/*Fault
Add in files from merge-bare-iron, get them compiling in FS and SE mode --HG-- extra : convert_revision : d4e19afda897bc3797868b40469ce2ec7ec7d251
2007-11-13 22:58:16 +01:00
BaseSimpleCPU::CacheOp(uint8_t Op, Addr EffAddr)
{
// translate to physical address
Fault fault = NoFault;
int CacheID = Op & 0x3; // Lower 3 bits identify Cache
int CacheOP = Op >> 2; // Upper 3 bits identify Cache Operation
if(CacheID > 1)
{
warn("CacheOps not implemented for secondary/tertiary caches\n");
}
else
{
switch(CacheOP)
{ // Fill Packet Type
case 0: warn("Invalidate Cache Op\n");
break;
case 1: warn("Index Load Tag Cache Op\n");
break;
case 2: warn("Index Store Tag Cache Op\n");
break;
case 4: warn("Hit Invalidate Cache Op\n");
break;
case 5: warn("Fill/Hit Writeback Invalidate Cache Op\n");
break;
case 6: warn("Hit Writeback\n");
break;
case 7: warn("Fetch & Lock Cache Op\n");
break;
default: warn("Unimplemented Cache Op\n");
}
}
return fault;
Get MIPS simple regression working. Take out unecessary functions "setShadowSet", "CacheOp" --HG-- extra : convert_revision : a9ae8a7e62c27c2db16fd3cfa7a7f0bf5f0bf8ea
2007-11-15 09:10:41 +01:00
}*/
Reference in a new issue Copy permalink