gem5/src/SConscript

1350 lines
45 KiB
Python
Raw Normal View History

# -*- mode:python -*-
Many files: Update copyright dates and author list SConscript: arch/alpha/alpha_linux_process.cc: arch/alpha/alpha_linux_process.hh: arch/alpha/alpha_memory.cc: arch/alpha/alpha_memory.hh: arch/alpha/alpha_tru64_process.cc: arch/alpha/alpha_tru64_process.hh: arch/alpha/aout_machdep.h: arch/alpha/arguments.cc: arch/alpha/arguments.hh: arch/alpha/ev5.cc: arch/alpha/ev5.hh: arch/alpha/faults.cc: arch/alpha/faults.hh: arch/alpha/isa_desc: arch/alpha/isa_traits.hh: arch/alpha/osfpal.cc: arch/alpha/osfpal.hh: arch/alpha/pseudo_inst.cc: arch/alpha/pseudo_inst.hh: arch/alpha/vptr.hh: arch/alpha/vtophys.cc: arch/alpha/vtophys.hh: base/bitfield.hh: base/callback.hh: base/circlebuf.cc: base/circlebuf.hh: base/cprintf.cc: base/cprintf.hh: base/cprintf_formats.hh: base/crc.hh: base/date.cc: base/dbl_list.hh: base/endian.hh: base/fast_alloc.cc: base/fast_alloc.hh: base/fifo_buffer.cc: base/fifo_buffer.hh: base/hashmap.hh: base/hostinfo.cc: base/hostinfo.hh: base/hybrid_pred.cc: base/hybrid_pred.hh: base/inet.cc: base/inet.hh: base/inifile.cc: base/inifile.hh: base/intmath.cc: base/intmath.hh: base/match.cc: base/match.hh: base/misc.cc: base/misc.hh: base/mod_num.hh: base/mysql.cc: base/mysql.hh: base/output.cc: base/output.hh: base/pollevent.cc: base/pollevent.hh: base/predictor.hh: base/random.cc: base/random.hh: base/range.cc: base/range.hh: base/refcnt.hh: base/remote_gdb.cc: base/remote_gdb.hh: base/res_list.hh: base/sat_counter.cc: base/sat_counter.hh: base/sched_list.hh: base/socket.cc: base/socket.hh: base/statistics.cc: base/statistics.hh: base/compression/lzss_compression.cc: base/compression/lzss_compression.hh: base/compression/null_compression.hh: base/loader/aout_object.cc: base/loader/aout_object.hh: base/loader/ecoff_object.cc: base/loader/ecoff_object.hh: base/loader/elf_object.cc: base/loader/elf_object.hh: base/loader/object_file.cc: base/loader/object_file.hh: base/loader/symtab.cc: base/loader/symtab.hh: base/stats/events.cc: base/stats/events.hh: base/stats/flags.hh: base/stats/mysql.cc: base/stats/mysql.hh: base/stats/mysql_run.hh: base/stats/output.hh: base/stats/statdb.cc: base/stats/statdb.hh: base/stats/text.cc: base/stats/text.hh: base/stats/types.hh: base/stats/visit.cc: base/stats/visit.hh: base/str.cc: base/str.hh: base/time.cc: base/time.hh: base/timebuf.hh: base/trace.cc: base/trace.hh: base/userinfo.cc: base/userinfo.hh: build/SConstruct: cpu/base.cc: cpu/base.hh: cpu/base_dyn_inst.cc: cpu/base_dyn_inst.hh: cpu/exec_context.cc: cpu/exec_context.hh: cpu/exetrace.cc: cpu/exetrace.hh: cpu/inst_seq.hh: cpu/intr_control.cc: cpu/intr_control.hh: cpu/memtest/memtest.cc: cpu/pc_event.cc: cpu/pc_event.hh: cpu/smt.hh: cpu/static_inst.cc: cpu/static_inst.hh: cpu/memtest/memtest.hh: cpu/o3/sat_counter.cc: cpu/o3/sat_counter.hh: cpu/ozone/cpu.hh: cpu/simple/cpu.cc: cpu/simple/cpu.hh: cpu/trace/opt_cpu.cc: cpu/trace/opt_cpu.hh: cpu/trace/reader/ibm_reader.cc: cpu/trace/reader/ibm_reader.hh: cpu/trace/reader/itx_reader.cc: cpu/trace/reader/itx_reader.hh: cpu/trace/reader/m5_reader.cc: cpu/trace/reader/m5_reader.hh: cpu/trace/reader/mem_trace_reader.cc: cpu/trace/reader/mem_trace_reader.hh: cpu/trace/trace_cpu.cc: cpu/trace/trace_cpu.hh: dev/alpha_access.h: dev/alpha_console.cc: dev/alpha_console.hh: dev/baddev.cc: dev/baddev.hh: dev/disk_image.cc: dev/disk_image.hh: dev/etherbus.cc: dev/etherbus.hh: dev/etherdump.cc: dev/etherdump.hh: dev/etherint.cc: dev/etherint.hh: dev/etherlink.cc: dev/etherlink.hh: dev/etherpkt.cc: dev/etherpkt.hh: dev/ethertap.cc: dev/ethertap.hh: dev/ide_ctrl.cc: dev/ide_ctrl.hh: dev/ide_disk.cc: dev/ide_disk.hh: dev/io_device.cc: dev/io_device.hh: dev/ns_gige.cc: dev/ns_gige.hh: dev/ns_gige_reg.h: dev/pciconfigall.cc: dev/pciconfigall.hh: dev/pcidev.cc: dev/pcidev.hh: dev/pcireg.h: dev/pktfifo.cc: dev/pktfifo.hh: dev/platform.cc: dev/platform.hh: dev/simconsole.cc: dev/simconsole.hh: dev/simple_disk.cc: dev/simple_disk.hh: dev/sinic.cc: dev/sinic.hh: dev/sinicreg.hh: dev/tsunami.cc: dev/tsunami.hh: dev/tsunami_cchip.cc: dev/tsunami_cchip.hh: dev/tsunami_io.cc: dev/tsunami_io.hh: dev/tsunami_pchip.cc: dev/tsunami_pchip.hh: dev/tsunamireg.h: dev/uart.cc: dev/uart.hh: dev/uart8250.cc: dev/uart8250.hh: docs/stl.hh: encumbered/cpu/full/op_class.hh: kern/kernel_stats.cc: kern/kernel_stats.hh: kern/linux/linux.hh: kern/linux/linux_syscalls.cc: kern/linux/linux_syscalls.hh: kern/linux/linux_system.cc: kern/linux/linux_system.hh: kern/linux/linux_threadinfo.hh: kern/linux/printk.cc: kern/linux/printk.hh: kern/system_events.cc: kern/system_events.hh: kern/tru64/dump_mbuf.cc: kern/tru64/dump_mbuf.hh: kern/tru64/mbuf.hh: kern/tru64/printf.cc: kern/tru64/printf.hh: kern/tru64/tru64.hh: kern/tru64/tru64_events.cc: kern/tru64/tru64_events.hh: kern/tru64/tru64_syscalls.cc: kern/tru64/tru64_syscalls.hh: kern/tru64/tru64_system.cc: kern/tru64/tru64_system.hh: python/SConscript: python/m5/__init__.py: python/m5/config.py: python/m5/convert.py: python/m5/multidict.py: python/m5/smartdict.py: sim/async.hh: sim/builder.cc: sim/builder.hh: sim/debug.cc: sim/debug.hh: sim/eventq.cc: sim/eventq.hh: sim/host.hh: sim/main.cc: sim/param.cc: sim/param.hh: sim/process.cc: sim/process.hh: sim/root.cc: sim/serialize.cc: sim/serialize.hh: sim/sim_events.cc: sim/sim_events.hh: sim/sim_exit.hh: sim/sim_object.cc: sim/sim_object.hh: sim/startup.cc: sim/startup.hh: sim/stat_control.cc: sim/stat_control.hh: sim/stats.hh: sim/syscall_emul.cc: sim/syscall_emul.hh: sim/system.cc: sim/system.hh: test/bitvectest.cc: test/circletest.cc: test/cprintftest.cc: test/genini.py: test/initest.cc: test/lru_test.cc: test/nmtest.cc: test/offtest.cc: test/paramtest.cc: test/rangetest.cc: test/sized_test.cc: test/stattest.cc: test/strnumtest.cc: test/symtest.cc: test/tokentest.cc: test/tracetest.cc: util/ccdrv/devtime.c: util/m5/m5.c: util/oprofile-top.py: util/rundiff: util/m5/m5op.h: util/m5/m5op.s: util/stats/db.py: util/stats/dbinit.py: util/stats/display.py: util/stats/info.py: util/stats/print.py: util/stats/stats.py: util/tap/tap.cc: Update copyright dates and author list --HG-- extra : convert_revision : 0faba08fc0fc0146f1efb7f61e4b043c020ff9e4
2005-06-05 11:16:00 +02:00
# Copyright (c) 2004-2005 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.
#
# Authors: Nathan Binkert
import array
import bisect
import imp
import marshal
import os
import re
import sys
import zlib
from os.path import basename, dirname, exists, isdir, isfile, join as joinpath
import SCons
# This file defines how to build a particular configuration of gem5
# based on variable settings in the 'env' build environment.
Import('*')
# Children need to see the environment
Export('env')
build_env = [(opt, env[opt]) for opt in export_vars]
from m5.util import code_formatter, compareVersions
########################################################################
# Code for adding source files of various types
#
# When specifying a source file of some type, a set of guards can be
# specified for that file. When get() is used to find the files, if
# get specifies a set of filters, only files that match those filters
# will be accepted (unspecified filters on files are assumed to be
# false). Current filters are:
# main -- specifies the gem5 main() function
# skip_lib -- do not put this file into the gem5 library
# skip_no_python -- do not put this file into a no_python library
# as it embeds compiled Python
# <unittest> -- unit tests use filters based on the unit test name
#
# A parent can now be specified for a source file and default filter
# values will be retrieved recursively from parents (children override
# parents).
#
class SourceMeta(type):
'''Meta class for source files that keeps track of all files of a
particular type and has a get function for finding all functions
of a certain type that match a set of guards'''
def __init__(cls, name, bases, dict):
super(SourceMeta, cls).__init__(name, bases, dict)
cls.all = []
2016-01-19 20:28:22 +01:00
def get(cls, **guards):
'''Find all files that match the specified guards. If a source
file does not specify a flag, the default is False'''
for src in cls.all:
for flag,value in guards.iteritems():
# if the flag is found and has a different value, skip
# this file
if src.all_guards.get(flag, False) != value:
break
else:
yield src
class SourceFile(object):
'''Base object that encapsulates the notion of a source file.
This includes, the source node, target node, various manipulations
of those. A source file also specifies a set of guards which
describing which builds the source file applies to. A parent can
also be specified to get default guards from'''
__metaclass__ = SourceMeta
def __init__(self, source, parent=None, **guards):
self.guards = guards
self.parent = parent
tnode = source
if not isinstance(source, SCons.Node.FS.File):
tnode = File(source)
self.tnode = tnode
self.snode = tnode.srcnode()
for base in type(self).__mro__:
if issubclass(base, SourceFile):
base.all.append(self)
@property
def filename(self):
return str(self.tnode)
@property
def dirname(self):
return dirname(self.filename)
@property
def basename(self):
return basename(self.filename)
@property
def extname(self):
index = self.basename.rfind('.')
if index <= 0:
# dot files aren't extensions
return self.basename, None
return self.basename[:index], self.basename[index+1:]
@property
def all_guards(self):
'''find all guards for this object getting default values
recursively from its parents'''
guards = {}
if self.parent:
guards.update(self.parent.guards)
guards.update(self.guards)
return guards
def __lt__(self, other): return self.filename < other.filename
def __le__(self, other): return self.filename <= other.filename
def __gt__(self, other): return self.filename > other.filename
def __ge__(self, other): return self.filename >= other.filename
def __eq__(self, other): return self.filename == other.filename
def __ne__(self, other): return self.filename != other.filename
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
@staticmethod
def done():
def disabled(cls, name, *ignored):
raise RuntimeError("Additional SourceFile '%s'" % name,\
"declared, but targets deps are already fixed.")
SourceFile.__init__ = disabled
class Source(SourceFile):
'''Add a c/c++ source file to the build'''
def __init__(self, source, Werror=True, swig=False, **guards):
'''specify the source file, and any guards'''
super(Source, self).__init__(source, **guards)
self.Werror = Werror
self.swig = swig
class PySource(SourceFile):
'''Add a python source file to the named package'''
invalid_sym_char = re.compile('[^A-z0-9_]')
modules = {}
tnodes = {}
symnames = {}
def __init__(self, package, source, **guards):
'''specify the python package, the source file, and any guards'''
super(PySource, self).__init__(source, **guards)
modname,ext = self.extname
assert ext == 'py'
if package:
path = package.split('.')
else:
path = []
modpath = path[:]
if modname != '__init__':
modpath += [ modname ]
modpath = '.'.join(modpath)
arcpath = path + [ self.basename ]
abspath = self.snode.abspath
if not exists(abspath):
abspath = self.tnode.abspath
self.package = package
self.modname = modname
self.modpath = modpath
self.arcname = joinpath(*arcpath)
self.abspath = abspath
self.compiled = File(self.filename + 'c')
self.cpp = File(self.filename + '.cc')
self.symname = PySource.invalid_sym_char.sub('_', modpath)
PySource.modules[modpath] = self
PySource.tnodes[self.tnode] = self
PySource.symnames[self.symname] = self
class SimObject(PySource):
'''Add a SimObject python file as a python source object and add
it to a list of sim object modules'''
fixed = False
modnames = []
def __init__(self, source, **guards):
'''Specify the source file and any guards (automatically in
the m5.objects package)'''
super(SimObject, self).__init__('m5.objects', source, **guards)
if self.fixed:
raise AttributeError, "Too late to call SimObject now."
bisect.insort_right(SimObject.modnames, self.modname)
class SwigSource(SourceFile):
'''Add a swig file to build'''
def __init__(self, package, source, **guards):
'''Specify the python package, the source file, and any guards'''
super(SwigSource, self).__init__(source, skip_no_python=True, **guards)
modname,ext = self.extname
assert ext == 'i'
self.package = package
self.module = modname
cc_file = joinpath(self.dirname, modname + '_wrap.cc')
py_file = joinpath(self.dirname, modname + '.py')
self.cc_source = Source(cc_file, swig=True, parent=self, **guards)
self.py_source = PySource(package, py_file, parent=self, **guards)
scons: Add support for google protobuf building This patch enables the use of protobuf input files in the build process, thus allowing .proto files to be added to input. Each .proto file is compiled using the protoc tool and the newly created C++ source is added to the list of sources. The first location where the protobufs will be used is in the capturing and replay of memory traces, involving the communication monitor and the trace-generator state of the traffic generator. This will follow in the next patch. This patch does add a dependency on the availability of the BSD licensed protobuf library (and headers), and the protobuf compiler, protoc. These dependencies are checked in the SConstruct, similar to e.g. swig. The user can override the use of protoc from the PATH by specifying the PROTOC environment variable. Although the dependency on libprotobuf and protoc might seem like a big step, they add significant value to the project going forward. Execution traces and other types of traces could easily be added and parsers for C++ and Python are automatically generated. We could also envision using protobufs for the checkpoints, description of the traffic-generator behaviour etc. The sky is the limit. We could also use the GzipOutputStream from the protobuf library instead of the current GPL gzstream. Currently, only the C++ source and header is generated. Going forward we might want to add the Python output to support simple command-line tools for displaying and editing the traces.
2013-01-07 19:05:37 +01:00
class ProtoBuf(SourceFile):
'''Add a Protocol Buffer to build'''
def __init__(self, source, **guards):
'''Specify the source file, and any guards'''
super(ProtoBuf, self).__init__(source, **guards)
# Get the file name and the extension
modname,ext = self.extname
assert ext == 'proto'
# Currently, we stick to generating the C++ headers, so we
# only need to track the source and header.
self.cc_file = File(modname + '.pb.cc')
self.hh_file = File(modname + '.pb.h')
scons: Add support for google protobuf building This patch enables the use of protobuf input files in the build process, thus allowing .proto files to be added to input. Each .proto file is compiled using the protoc tool and the newly created C++ source is added to the list of sources. The first location where the protobufs will be used is in the capturing and replay of memory traces, involving the communication monitor and the trace-generator state of the traffic generator. This will follow in the next patch. This patch does add a dependency on the availability of the BSD licensed protobuf library (and headers), and the protobuf compiler, protoc. These dependencies are checked in the SConstruct, similar to e.g. swig. The user can override the use of protoc from the PATH by specifying the PROTOC environment variable. Although the dependency on libprotobuf and protoc might seem like a big step, they add significant value to the project going forward. Execution traces and other types of traces could easily be added and parsers for C++ and Python are automatically generated. We could also envision using protobufs for the checkpoints, description of the traffic-generator behaviour etc. The sky is the limit. We could also use the GzipOutputStream from the protobuf library instead of the current GPL gzstream. Currently, only the C++ source and header is generated. Going forward we might want to add the Python output to support simple command-line tools for displaying and editing the traces.
2013-01-07 19:05:37 +01:00
class UnitTest(object):
'''Create a UnitTest'''
all = []
2012-05-11 01:04:28 +02:00
def __init__(self, target, *sources, **kwargs):
'''Specify the target name and any sources. Sources that are
not SourceFiles are evalued with Source(). All files are
guarded with a guard of the same name as the UnitTest
target.'''
srcs = []
for src in sources:
if not isinstance(src, SourceFile):
src = Source(src, skip_lib=True)
src.guards[target] = True
srcs.append(src)
self.sources = srcs
self.target = target
2012-05-11 01:04:28 +02:00
self.main = kwargs.get('main', False)
UnitTest.all.append(self)
# Children should have access
Export('Source')
Export('PySource')
Export('SimObject')
Export('SwigSource')
scons: Add support for google protobuf building This patch enables the use of protobuf input files in the build process, thus allowing .proto files to be added to input. Each .proto file is compiled using the protoc tool and the newly created C++ source is added to the list of sources. The first location where the protobufs will be used is in the capturing and replay of memory traces, involving the communication monitor and the trace-generator state of the traffic generator. This will follow in the next patch. This patch does add a dependency on the availability of the BSD licensed protobuf library (and headers), and the protobuf compiler, protoc. These dependencies are checked in the SConstruct, similar to e.g. swig. The user can override the use of protoc from the PATH by specifying the PROTOC environment variable. Although the dependency on libprotobuf and protoc might seem like a big step, they add significant value to the project going forward. Execution traces and other types of traces could easily be added and parsers for C++ and Python are automatically generated. We could also envision using protobufs for the checkpoints, description of the traffic-generator behaviour etc. The sky is the limit. We could also use the GzipOutputStream from the protobuf library instead of the current GPL gzstream. Currently, only the C++ source and header is generated. Going forward we might want to add the Python output to support simple command-line tools for displaying and editing the traces.
2013-01-07 19:05:37 +01:00
Export('ProtoBuf')
Export('UnitTest')
########################################################################
#
# Debug Flags
#
debug_flags = {}
def DebugFlag(name, desc=None):
if name in debug_flags:
raise AttributeError, "Flag %s already specified" % name
debug_flags[name] = (name, (), desc)
def CompoundFlag(name, flags, desc=None):
if name in debug_flags:
raise AttributeError, "Flag %s already specified" % name
compound = tuple(flags)
debug_flags[name] = (name, compound, desc)
Export('DebugFlag')
Export('CompoundFlag')
########################################################################
#
# Set some compiler variables
#
Move main control from C++ into Python. User script now invokes initialization and simulation loop after building configuration. These functions are exported from C++ to Python using SWIG. SConstruct: Set up SWIG builder & scanner. Set up symlinking of source files into build directory (by not disabling the default behavior). configs/test/test.py: Rewrite to use new script-driven interface. Include a sample option. src/SConscript: Set up symlinking of source files into build directory (by not disabling the default behavior). Add SWIG-generated main_wrap.cc to source list. src/arch/SConscript: Set up symlinking of source files into build directory (by not disabling the default behavior). src/arch/alpha/ev5.cc: src/arch/alpha/isa/decoder.isa: src/cpu/o3/alpha_cpu_impl.hh: src/cpu/trace/opt_cpu.cc: src/cpu/trace/trace_cpu.cc: src/sim/pseudo_inst.cc: src/sim/root.cc: src/sim/serialize.cc: src/sim/syscall_emul.cc: SimExit() is now exitSimLoop(). src/cpu/base.cc: SimExitEvent is now SimLoopExitEvent src/python/SConscript: Add SWIG build command for main.i. Use python/m5 in build dir as source for zip archive... easy now with file duplication enabled. src/python/m5/__init__.py: - Move copyright notice back to C++ so we can print it right away, even for interactive sessions. - Get rid of argument parsing code; just provide default option descriptors for user script to call optparse with. - Don't clutter m5 namespace by sucking in all of m5.config and m5.objects. - Move instantiate() function here from config.py. src/python/m5/config.py: - Move instantiate() function to __init__.py. - Param.Foo deferred type lookups must use m5.objects namespace now (not m5). src/python/m5/objects/AlphaConsole.py: src/python/m5/objects/AlphaFullCPU.py: src/python/m5/objects/AlphaTLB.py: src/python/m5/objects/BadDevice.py: src/python/m5/objects/BaseCPU.py: src/python/m5/objects/BaseCache.py: src/python/m5/objects/Bridge.py: src/python/m5/objects/Bus.py: src/python/m5/objects/CoherenceProtocol.py: src/python/m5/objects/Device.py: src/python/m5/objects/DiskImage.py: src/python/m5/objects/Ethernet.py: src/python/m5/objects/Ide.py: src/python/m5/objects/IntrControl.py: src/python/m5/objects/MemObject.py: src/python/m5/objects/MemTest.py: src/python/m5/objects/Pci.py: src/python/m5/objects/PhysicalMemory.py: src/python/m5/objects/Platform.py: src/python/m5/objects/Process.py: src/python/m5/objects/Repl.py: src/python/m5/objects/Root.py: src/python/m5/objects/SimConsole.py: src/python/m5/objects/SimpleDisk.py: src/python/m5/objects/System.py: src/python/m5/objects/Tsunami.py: src/python/m5/objects/Uart.py: Fix up imports (m5 namespace no longer includes m5.config). src/sim/eventq.cc: src/sim/eventq.hh: Support for Python-called simulate() function: - Use IsExitEvent flag to signal events that want to exit the simulation loop gracefully (instead of calling exit() to terminate the process). - Modify interface to hand exit event object back to caller so it can be inspected for cause. src/sim/host.hh: Add MaxTick constant. src/sim/main.cc: Move copyright notice back to C++ so we can print it right away, even for interactive sessions. Use PYTHONPATH environment var to set module path (instead of clunky code injection method). Move main control from here into Python: - Separate initialization code and simulation loop into separate functions callable from Python. - Make Python interpreter invocation more pure (more like directly invoking interpreter). Add -i and -p flags (only options on binary itself; other options processed by Python). Import readline package when using interactive mode. src/sim/sim_events.cc: SimExitEvent is now SimLoopExitEvent, and uses IsSimExit flag to terminate loop (instead of exiting simulator process). src/sim/sim_events.hh: SimExitEvent is now SimLoopExitEvent, and uses IsSimExit flag to terminate loop (instead of exiting simulator process). Get rid of a few unused constructors. src/sim/sim_exit.hh: SimExit() is now exitSimLoop(). Get rid of unused functions. Add comments. --HG-- extra : convert_revision : 280b0d671516b25545a6f24cefa64a68319ff3d4
2006-06-10 05:01:31 +02:00
# Include file paths are rooted in this directory. SCons will
# automatically expand '.' to refer to both the source directory and
# the corresponding build directory to pick up generated include
# files.
env.Append(CPPPATH=Dir('.'))
for extra_dir in extras_dir_list:
env.Append(CPPPATH=Dir(extra_dir))
# Workaround for bug in SCons version > 0.97d20071212
# Scons bug id: 2006 gem5 Bug id: 308
for root, dirs, files in os.walk(base_dir, topdown=True):
Dir(root[len(base_dir) + 1:])
########################################################################
#
# Walk the tree and execute all SConscripts in subdirectories
#
here = Dir('.').srcnode().abspath
for root, dirs, files in os.walk(base_dir, topdown=True):
if root == here:
# we don't want to recurse back into this SConscript
continue
if 'SConscript' in files:
build_dir = joinpath(env['BUILDDIR'], root[len(base_dir) + 1:])
SConscript(joinpath(root, 'SConscript'), variant_dir=build_dir)
for extra_dir in extras_dir_list:
prefix_len = len(dirname(extra_dir)) + 1
# Also add the corresponding build directory to pick up generated
# include files.
env.Append(CPPPATH=Dir(joinpath(env['BUILDDIR'], extra_dir[prefix_len:])))
for root, dirs, files in os.walk(extra_dir, topdown=True):
# if build lives in the extras directory, don't walk down it
if 'build' in dirs:
dirs.remove('build')
if 'SConscript' in files:
build_dir = joinpath(env['BUILDDIR'], root[prefix_len:])
SConscript(joinpath(root, 'SConscript'), variant_dir=build_dir)
for opt in export_vars:
Build options are set via a build_options file in the build directory instead of being inferred from the name of the build directory. Options are passed to C++ via config/*.hh files instead of via the command line. Build option flags are now always defined to 0 or 1, so checks must use '#if' rather than '#ifdef'. SConscript: MySQL detection moved to SConstruct. Add config/*.hh files (via ConfigFile builder). arch/alpha/alpha_memory.cc: arch/alpha/ev5.cc: arch/alpha/ev5.hh: arch/alpha/isa_traits.hh: base/fast_alloc.hh: base/statistics.cc: base/statistics.hh: base/stats/events.cc: base/stats/events.hh: cpu/base.cc: cpu/base.hh: cpu/base_dyn_inst.cc: cpu/base_dyn_inst.hh: cpu/exec_context.cc: cpu/exec_context.hh: cpu/o3/alpha_cpu.hh: cpu/o3/alpha_cpu_builder.cc: cpu/o3/alpha_cpu_impl.hh: cpu/o3/alpha_dyn_inst.hh: cpu/o3/alpha_dyn_inst_impl.hh: cpu/o3/alpha_params.hh: cpu/o3/commit_impl.hh: cpu/o3/cpu.cc: cpu/o3/cpu.hh: cpu/o3/fetch_impl.hh: cpu/o3/iew.hh: cpu/o3/iew_impl.hh: cpu/o3/regfile.hh: cpu/o3/rename_impl.hh: cpu/o3/rob_impl.hh: cpu/ozone/cpu.hh: cpu/pc_event.cc: cpu/simple/cpu.cc: cpu/simple/cpu.hh: sim/process.cc: sim/process.hh: Convert compile flags from def/undef to 0/1. Set via #include config/*.hh instead of command line. arch/alpha/isa_desc: Convert compile flags from def/undef to 0/1. Set via #include config/*.hh instead of command line. Revamp fenv.h support... most of the ugliness is hidden in base/fenv.hh now. base/mysql.hh: Fix typo in #ifndef guard. build/SConstruct: Build options are set via a build_options file in the build directory instead of being inferred from the name of the build directory. Options are passed to C++ via config/*.hh files instead of via the command line. python/SConscript: Generate m5_build_env directly from scons options instead of indirectly via CPPDEFINES. python/m5/convert.py: Allow '0' and '1' for booleans. Rewrite toBool to use dict. base/fenv.hh: Revamp <fenv.h> support to make it a compile option (so we can test w/o it even if it's present) and to make isa_desc cleaner. --HG-- extra : convert_revision : 8f97dc11185bef5e1865b3269c7341df8525c9ad
2005-08-30 19:18:54 +02:00
env.ConfigFile(opt)
def makeTheISA(source, target, env):
isas = [ src.get_contents() for src in source ]
target_isa = env['TARGET_ISA']
def define(isa):
return isa.upper() + '_ISA'
2016-01-19 20:28:22 +01:00
def namespace(isa):
2016-01-19 20:28:22 +01:00
return isa[0].upper() + isa[1:].lower() + 'ISA'
code = code_formatter()
code('''\
#ifndef __CONFIG_THE_ISA_HH__
#define __CONFIG_THE_ISA_HH__
''')
# create defines for the preprocessing and compile-time determination
for i,isa in enumerate(isas):
code('#define $0 $1', define(isa), i + 1)
code()
# create an enum for any run-time determination of the ISA, we
# reuse the same name as the namespaces
code('enum class Arch {')
for i,isa in enumerate(isas):
if i + 1 == len(isas):
code(' $0 = $1', namespace(isa), define(isa))
else:
code(' $0 = $1,', namespace(isa), define(isa))
code('};')
code('''
#define THE_ISA ${{define(target_isa)}}
#define TheISA ${{namespace(target_isa)}}
#define THE_ISA_STR "${{target_isa}}"
#endif // __CONFIG_THE_ISA_HH__''')
code.write(str(target[0]))
env.Command('config/the_isa.hh', map(Value, all_isa_list),
MakeAction(makeTheISA, Transform("CFG ISA", 0)))
2016-01-19 20:28:22 +01:00
def makeTheGPUISA(source, target, env):
isas = [ src.get_contents() for src in source ]
target_gpu_isa = env['TARGET_GPU_ISA']
def define(isa):
return isa.upper() + '_ISA'
def namespace(isa):
return isa[0].upper() + isa[1:].lower() + 'ISA'
code = code_formatter()
code('''\
#ifndef __CONFIG_THE_GPU_ISA_HH__
#define __CONFIG_THE_GPU_ISA_HH__
''')
# create defines for the preprocessing and compile-time determination
for i,isa in enumerate(isas):
code('#define $0 $1', define(isa), i + 1)
code()
# create an enum for any run-time determination of the ISA, we
# reuse the same name as the namespaces
code('enum class GPUArch {')
for i,isa in enumerate(isas):
if i + 1 == len(isas):
code(' $0 = $1', namespace(isa), define(isa))
else:
code(' $0 = $1,', namespace(isa), define(isa))
code('};')
code('''
#define THE_GPU_ISA ${{define(target_gpu_isa)}}
#define TheGpuISA ${{namespace(target_gpu_isa)}}
#define THE_GPU_ISA_STR "${{target_gpu_isa}}"
#endif // __CONFIG_THE_GPU_ISA_HH__''')
code.write(str(target[0]))
env.Command('config/the_gpu_isa.hh', map(Value, all_gpu_isa_list),
MakeAction(makeTheGPUISA, Transform("CFG ISA", 0)))
########################################################################
#
# Prevent any SimObjects from being added after this point, they
# should all have been added in the SConscripts above
#
SimObject.fixed = True
class DictImporter(object):
'''This importer takes a dictionary of arbitrary module names that
map to arbitrary filenames.'''
def __init__(self, modules):
self.modules = modules
self.installed = set()
def __del__(self):
self.unload()
def unload(self):
import sys
for module in self.installed:
del sys.modules[module]
self.installed = set()
def find_module(self, fullname, path):
if fullname == 'm5.defines':
return self
if fullname == 'm5.objects':
return self
if fullname.startswith('m5.internal'):
return None
source = self.modules.get(fullname, None)
if source is not None and fullname.startswith('m5.objects'):
return self
return None
def load_module(self, fullname):
mod = imp.new_module(fullname)
sys.modules[fullname] = mod
self.installed.add(fullname)
mod.__loader__ = self
if fullname == 'm5.objects':
mod.__path__ = fullname.split('.')
return mod
if fullname == 'm5.defines':
mod.__dict__['buildEnv'] = m5.util.SmartDict(build_env)
return mod
source = self.modules[fullname]
if source.modname == '__init__':
mod.__path__ = source.modpath
mod.__file__ = source.abspath
exec file(source.abspath, 'r') in mod.__dict__
return mod
import m5.SimObject
import m5.params
from m5.util import code_formatter
m5.SimObject.clear()
m5.params.clear()
# install the python importer so we can grab stuff from the source
# tree itself. We can't have SimObjects added after this point or
# else we won't know about them for the rest of the stuff.
importer = DictImporter(PySource.modules)
sys.meta_path[0:0] = [ importer ]
# import all sim objects so we can populate the all_objects list
# make sure that we're working with a list, then let's sort it
for modname in SimObject.modnames:
exec('from m5.objects import %s' % modname)
# we need to unload all of the currently imported modules so that they
# will be re-imported the next time the sconscript is run
importer.unload()
sys.meta_path.remove(importer)
sim_objects = m5.SimObject.allClasses
all_enums = m5.params.allEnums
if m5.SimObject.noCxxHeader:
print >> sys.stderr, \
"warning: At least one SimObject lacks a header specification. " \
"This can cause unexpected results in the generated SWIG " \
"wrappers."
# Find param types that need to be explicitly wrapped with swig.
# These will be recognized because the ParamDesc will have a
# swig_decl() method. Most param types are based on types that don't
# need this, either because they're based on native types (like Int)
# or because they're SimObjects (which get swigged independently).
# For now the only things handled here are VectorParam types.
params_to_swig = {}
for name,obj in sorted(sim_objects.iteritems()):
for param in obj._params.local.values():
# load the ptype attribute now because it depends on the
# current version of SimObject.allClasses, but when scons
# actually uses the value, all versions of
# SimObject.allClasses will have been loaded
param.ptype
if not hasattr(param, 'swig_decl'):
continue
pname = param.ptype_str
if pname not in params_to_swig:
params_to_swig[pname] = param
########################################################################
#
# calculate extra dependencies
#
module_depends = ["m5", "m5.SimObject", "m5.params"]
depends = [ PySource.modules[dep].snode for dep in module_depends ]
depends.sort(key = lambda x: x.name)
########################################################################
#
# Commands for the basic automatically generated python files
#
# Generate Python file containing a dict specifying the current
# buildEnv flags.
def makeDefinesPyFile(target, source, env):
build_env = source[0].get_contents()
code = code_formatter()
code("""
import m5.internal
import m5.util
buildEnv = m5.util.SmartDict($build_env)
compileDate = m5.internal.core.compileDate
_globals = globals()
for key,val in m5.internal.core.__dict__.iteritems():
if key.startswith('flag_'):
flag = key[5:]
_globals[flag] = val
del _globals
""")
code.write(target[0].abspath)
defines_info = Value(build_env)
2009-01-19 18:59:13 +01:00
# Generate a file with all of the compile options in it
env.Command('python/m5/defines.py', defines_info,
MakeAction(makeDefinesPyFile, Transform("DEFINES", 0)))
2009-01-19 18:59:13 +01:00
PySource('m5', 'python/m5/defines.py')
# Generate python file containing info about the M5 source code
def makeInfoPyFile(target, source, env):
code = code_formatter()
for src in source:
data = ''.join(file(src.srcnode().abspath, 'r').xreadlines())
code('$src = ${{repr(data)}}')
code.write(str(target[0]))
2009-01-19 18:59:13 +01:00
# Generate a file that wraps the basic top level files
env.Command('python/m5/info.py',
[ '#/COPYING', '#/LICENSE', '#/README', ],
MakeAction(makeInfoPyFile, Transform("INFO")))
2009-01-19 18:59:13 +01:00
PySource('m5', 'python/m5/info.py')
########################################################################
#
# Create all of the SimObject param headers and enum headers
#
def createSimObjectParamStruct(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
obj = sim_objects[name]
code = code_formatter()
obj.cxx_param_decl(code)
code.write(target[0].abspath)
config: Add the ability to read a config file using C++ and Python This patch adds the ability to load in config.ini files generated from gem5 into another instance of gem5 built without Python configuration support. The intended use case is for configuring gem5 when it is a library embedded in another simulation system. A parallel config file reader is also provided purely in Python to demonstrate the approach taken and to provided similar functionality for as-yet-unknown use models. The Python configuration file reader can read both .ini and .json files. C++ configuration file reading: A command line option has been added for scons to enable C++ configuration file reading: --with-cxx-config There is an example in util/cxx_config that shows C++ configuration in action. util/cxx_config/README explains how to build the example. Configuration is achieved by the object CxxConfigManager. It handles reading object descriptions from a CxxConfigFileBase object which wraps a config file reader. The wrapper class CxxIniFile is provided which wraps an IniFile for reading .ini files. Reading .json files from C++ would be possible with a similar wrapper and a JSON parser. After reading object descriptions, CxxConfigManager creates SimObjectParam-derived objects from the classes in the (generated with this patch) directory build/ARCH/cxx_config CxxConfigManager can then build SimObjects from those SimObjectParams (in an order dictated by the SimObject-value parameters on other objects) and bind ports of the produced SimObjects. A minimal set of instantiate-replacing member functions are provided by CxxConfigManager and few of the member functions of SimObject (such as drain) are extended onto CxxConfigManager. Python configuration file reading (configs/example/read_config.py): A Python version of the reader is also supplied with a similar interface to CxxConfigFileBase (In Python: ConfigFile) to config file readers. The Python config file reading will handle both .ini and .json files. The object construction strategy is slightly different in Python from the C++ reader as you need to avoid objects prematurely becoming the children of other objects when setting parameters. Port binding also needs to be strictly in the same port-index order as the original instantiation.
2014-10-16 11:49:37 +02:00
def createSimObjectCxxConfig(is_header):
def body(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
obj = sim_objects[name]
code = code_formatter()
obj.cxx_config_param_file(code, is_header)
code.write(target[0].abspath)
return body
def createParamSwigWrapper(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
param = params_to_swig[name]
code = code_formatter()
param.swig_decl(code)
code.write(target[0].abspath)
def createEnumStrings(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
obj = all_enums[name]
code = code_formatter()
obj.cxx_def(code)
code.write(target[0].abspath)
def createEnumDecls(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
obj = all_enums[name]
code = code_formatter()
obj.cxx_decl(code)
code.write(target[0].abspath)
def createEnumSwigWrapper(target, source, env):
assert len(target) == 1 and len(source) == 1
name = str(source[0].get_contents())
obj = all_enums[name]
code = code_formatter()
obj.swig_decl(code)
code.write(target[0].abspath)
def createSimObjectSwigWrapper(target, source, env):
name = source[0].get_contents()
obj = sim_objects[name]
code = code_formatter()
obj.swig_decl(code)
code.write(target[0].abspath)
# dummy target for generated code
# we start out with all the Source files so they get copied to build/*/ also.
SWIG = env.Dummy('swig', [s.tnode for s in Source.get()])
# Generate all of the SimObject param C++ struct header files
params_hh_files = []
for name,simobj in sorted(sim_objects.iteritems()):
py_source = PySource.modules[simobj.__module__]
extra_deps = [ py_source.tnode ]
hh_file = File('params/%s.hh' % name)
params_hh_files.append(hh_file)
env.Command(hh_file, Value(name),
MakeAction(createSimObjectParamStruct, Transform("SO PARAM")))
env.Depends(hh_file, depends + extra_deps)
env.Depends(SWIG, hh_file)
config: Add the ability to read a config file using C++ and Python This patch adds the ability to load in config.ini files generated from gem5 into another instance of gem5 built without Python configuration support. The intended use case is for configuring gem5 when it is a library embedded in another simulation system. A parallel config file reader is also provided purely in Python to demonstrate the approach taken and to provided similar functionality for as-yet-unknown use models. The Python configuration file reader can read both .ini and .json files. C++ configuration file reading: A command line option has been added for scons to enable C++ configuration file reading: --with-cxx-config There is an example in util/cxx_config that shows C++ configuration in action. util/cxx_config/README explains how to build the example. Configuration is achieved by the object CxxConfigManager. It handles reading object descriptions from a CxxConfigFileBase object which wraps a config file reader. The wrapper class CxxIniFile is provided which wraps an IniFile for reading .ini files. Reading .json files from C++ would be possible with a similar wrapper and a JSON parser. After reading object descriptions, CxxConfigManager creates SimObjectParam-derived objects from the classes in the (generated with this patch) directory build/ARCH/cxx_config CxxConfigManager can then build SimObjects from those SimObjectParams (in an order dictated by the SimObject-value parameters on other objects) and bind ports of the produced SimObjects. A minimal set of instantiate-replacing member functions are provided by CxxConfigManager and few of the member functions of SimObject (such as drain) are extended onto CxxConfigManager. Python configuration file reading (configs/example/read_config.py): A Python version of the reader is also supplied with a similar interface to CxxConfigFileBase (In Python: ConfigFile) to config file readers. The Python config file reading will handle both .ini and .json files. The object construction strategy is slightly different in Python from the C++ reader as you need to avoid objects prematurely becoming the children of other objects when setting parameters. Port binding also needs to be strictly in the same port-index order as the original instantiation.
2014-10-16 11:49:37 +02:00
# C++ parameter description files
if GetOption('with_cxx_config'):
for name,simobj in sorted(sim_objects.iteritems()):
py_source = PySource.modules[simobj.__module__]
extra_deps = [ py_source.tnode ]
cxx_config_hh_file = File('cxx_config/%s.hh' % name)
cxx_config_cc_file = File('cxx_config/%s.cc' % name)
env.Command(cxx_config_hh_file, Value(name),
MakeAction(createSimObjectCxxConfig(True),
Transform("CXXCPRHH")))
env.Command(cxx_config_cc_file, Value(name),
MakeAction(createSimObjectCxxConfig(False),
Transform("CXXCPRCC")))
env.Depends(cxx_config_hh_file, depends + extra_deps +
[File('params/%s.hh' % name), File('sim/cxx_config.hh')])
env.Depends(cxx_config_cc_file, depends + extra_deps +
[cxx_config_hh_file])
Source(cxx_config_cc_file)
cxx_config_init_cc_file = File('cxx_config/init.cc')
def createCxxConfigInitCC(target, source, env):
assert len(target) == 1 and len(source) == 1
code = code_formatter()
for name,simobj in sorted(sim_objects.iteritems()):
if not hasattr(simobj, 'abstract') or not simobj.abstract:
code('#include "cxx_config/${name}.hh"')
code()
code('void cxxConfigInit()')
code('{')
code.indent()
for name,simobj in sorted(sim_objects.iteritems()):
not_abstract = not hasattr(simobj, 'abstract') or \
not simobj.abstract
if not_abstract and 'type' in simobj.__dict__:
code('cxx_config_directory["${name}"] = '
'${name}CxxConfigParams::makeDirectoryEntry();')
code.dedent()
code('}')
code.write(target[0].abspath)
py_source = PySource.modules[simobj.__module__]
extra_deps = [ py_source.tnode ]
env.Command(cxx_config_init_cc_file, Value(name),
MakeAction(createCxxConfigInitCC, Transform("CXXCINIT")))
cxx_param_hh_files = ["cxx_config/%s.hh" % simobj
for name,simobj in sorted(sim_objects.iteritems())
config: Add the ability to read a config file using C++ and Python This patch adds the ability to load in config.ini files generated from gem5 into another instance of gem5 built without Python configuration support. The intended use case is for configuring gem5 when it is a library embedded in another simulation system. A parallel config file reader is also provided purely in Python to demonstrate the approach taken and to provided similar functionality for as-yet-unknown use models. The Python configuration file reader can read both .ini and .json files. C++ configuration file reading: A command line option has been added for scons to enable C++ configuration file reading: --with-cxx-config There is an example in util/cxx_config that shows C++ configuration in action. util/cxx_config/README explains how to build the example. Configuration is achieved by the object CxxConfigManager. It handles reading object descriptions from a CxxConfigFileBase object which wraps a config file reader. The wrapper class CxxIniFile is provided which wraps an IniFile for reading .ini files. Reading .json files from C++ would be possible with a similar wrapper and a JSON parser. After reading object descriptions, CxxConfigManager creates SimObjectParam-derived objects from the classes in the (generated with this patch) directory build/ARCH/cxx_config CxxConfigManager can then build SimObjects from those SimObjectParams (in an order dictated by the SimObject-value parameters on other objects) and bind ports of the produced SimObjects. A minimal set of instantiate-replacing member functions are provided by CxxConfigManager and few of the member functions of SimObject (such as drain) are extended onto CxxConfigManager. Python configuration file reading (configs/example/read_config.py): A Python version of the reader is also supplied with a similar interface to CxxConfigFileBase (In Python: ConfigFile) to config file readers. The Python config file reading will handle both .ini and .json files. The object construction strategy is slightly different in Python from the C++ reader as you need to avoid objects prematurely becoming the children of other objects when setting parameters. Port binding also needs to be strictly in the same port-index order as the original instantiation.
2014-10-16 11:49:37 +02:00
if not hasattr(simobj, 'abstract') or not simobj.abstract]
Depends(cxx_config_init_cc_file, cxx_param_hh_files +
[File('sim/cxx_config.hh')])
Source(cxx_config_init_cc_file)
# Generate any needed param SWIG wrapper files
2008-06-14 21:10:50 +02:00
params_i_files = []
for name,param in sorted(params_to_swig.iteritems()):
i_file = File('python/m5/internal/%s.i' % (param.swig_module_name()))
2008-06-14 21:10:50 +02:00
params_i_files.append(i_file)
env.Command(i_file, Value(name),
MakeAction(createParamSwigWrapper, Transform("SW PARAM")))
env.Depends(i_file, depends)
env.Depends(SWIG, i_file)
SwigSource('m5.internal', i_file)
# Generate all enum header files
for name,enum in sorted(all_enums.iteritems()):
py_source = PySource.modules[enum.__module__]
extra_deps = [ py_source.tnode ]
cc_file = File('enums/%s.cc' % name)
env.Command(cc_file, Value(name),
MakeAction(createEnumStrings, Transform("ENUM STR")))
env.Depends(cc_file, depends + extra_deps)
env.Depends(SWIG, cc_file)
Source(cc_file)
hh_file = File('enums/%s.hh' % name)
env.Command(hh_file, Value(name),
MakeAction(createEnumDecls, Transform("ENUMDECL")))
env.Depends(hh_file, depends + extra_deps)
env.Depends(SWIG, hh_file)
i_file = File('python/m5/internal/enum_%s.i' % name)
env.Command(i_file, Value(name),
MakeAction(createEnumSwigWrapper, Transform("ENUMSWIG")))
env.Depends(i_file, depends + extra_deps)
env.Depends(SWIG, i_file)
SwigSource('m5.internal', i_file)
# Generate SimObject SWIG wrapper files
for name,simobj in sorted(sim_objects.iteritems()):
py_source = PySource.modules[simobj.__module__]
extra_deps = [ py_source.tnode ]
i_file = File('python/m5/internal/param_%s.i' % name)
env.Command(i_file, Value(name),
MakeAction(createSimObjectSwigWrapper, Transform("SO SWIG")))
env.Depends(i_file, depends + extra_deps)
SwigSource('m5.internal', i_file)
# Generate the main swig init file
def makeEmbeddedSwigInit(package):
def body(target, source, env):
assert len(target) == 1 and len(source) == 1
code = code_formatter()
module = source[0].get_contents()
# Provide the full context so that the swig-generated call to
# Py_InitModule ends up placing the embedded module in the
# right package.
context = str(package) + "._" + str(module)
code('''\
#include "sim/init.hh"
extern "C" {
void init_${module}();
}
EmbeddedSwig embed_swig_${module}(init_${module}, "${context}");
''')
code.write(str(target[0]))
return body
2016-01-19 20:28:22 +01:00
# Build all swig modules
for swig in SwigSource.all:
env.Command([swig.cc_source.tnode, swig.py_source.tnode], swig.tnode,
MakeAction('$SWIG $SWIGFLAGS -outdir ${TARGETS[1].dir} '
'-o ${TARGETS[0]} $SOURCES', Transform("SWIG")))
cc_file = str(swig.tnode)
init_file = '%s/%s_init.cc' % (dirname(cc_file), basename(cc_file))
env.Command(init_file, Value(swig.module),
MakeAction(makeEmbeddedSwigInit(swig.package),
Transform("EMBED SW")))
env.Depends(SWIG, init_file)
Source(init_file, **swig.guards)
scons: Add support for google protobuf building This patch enables the use of protobuf input files in the build process, thus allowing .proto files to be added to input. Each .proto file is compiled using the protoc tool and the newly created C++ source is added to the list of sources. The first location where the protobufs will be used is in the capturing and replay of memory traces, involving the communication monitor and the trace-generator state of the traffic generator. This will follow in the next patch. This patch does add a dependency on the availability of the BSD licensed protobuf library (and headers), and the protobuf compiler, protoc. These dependencies are checked in the SConstruct, similar to e.g. swig. The user can override the use of protoc from the PATH by specifying the PROTOC environment variable. Although the dependency on libprotobuf and protoc might seem like a big step, they add significant value to the project going forward. Execution traces and other types of traces could easily be added and parsers for C++ and Python are automatically generated. We could also envision using protobufs for the checkpoints, description of the traffic-generator behaviour etc. The sky is the limit. We could also use the GzipOutputStream from the protobuf library instead of the current GPL gzstream. Currently, only the C++ source and header is generated. Going forward we might want to add the Python output to support simple command-line tools for displaying and editing the traces.
2013-01-07 19:05:37 +01:00
# Build all protocol buffers if we have got protoc and protobuf available
if env['HAVE_PROTOBUF']:
for proto in ProtoBuf.all:
# Use both the source and header as the target, and the .proto
# file as the source. When executing the protoc compiler, also
# specify the proto_path to avoid having the generated files
# include the path.
env.Command([proto.cc_file, proto.hh_file], proto.tnode,
MakeAction('$PROTOC --cpp_out ${TARGET.dir} '
'--proto_path ${SOURCE.dir} $SOURCE',
Transform("PROTOC")))
env.Depends(SWIG, [proto.cc_file, proto.hh_file])
scons: Add support for google protobuf building This patch enables the use of protobuf input files in the build process, thus allowing .proto files to be added to input. Each .proto file is compiled using the protoc tool and the newly created C++ source is added to the list of sources. The first location where the protobufs will be used is in the capturing and replay of memory traces, involving the communication monitor and the trace-generator state of the traffic generator. This will follow in the next patch. This patch does add a dependency on the availability of the BSD licensed protobuf library (and headers), and the protobuf compiler, protoc. These dependencies are checked in the SConstruct, similar to e.g. swig. The user can override the use of protoc from the PATH by specifying the PROTOC environment variable. Although the dependency on libprotobuf and protoc might seem like a big step, they add significant value to the project going forward. Execution traces and other types of traces could easily be added and parsers for C++ and Python are automatically generated. We could also envision using protobufs for the checkpoints, description of the traffic-generator behaviour etc. The sky is the limit. We could also use the GzipOutputStream from the protobuf library instead of the current GPL gzstream. Currently, only the C++ source and header is generated. Going forward we might want to add the Python output to support simple command-line tools for displaying and editing the traces.
2013-01-07 19:05:37 +01:00
# Add the C++ source file
Source(proto.cc_file, **proto.guards)
elif ProtoBuf.all:
print 'Got protobuf to build, but lacks support!'
Exit(1)
#
# Handle debug flags
#
def makeDebugFlagCC(target, source, env):
assert(len(target) == 1 and len(source) == 1)
code = code_formatter()
# delay definition of CompoundFlags until after all the definition
# of all constituent SimpleFlags
comp_code = code_formatter()
# file header
code('''
/*
* DO NOT EDIT THIS FILE! Automatically generated by SCons.
*/
#include "base/debug.hh"
namespace Debug {
''')
for name, flag in sorted(source[0].read().iteritems()):
n, compound, desc = flag
assert n == name
if not compound:
code('SimpleFlag $name("$name", "$desc");')
else:
comp_code('CompoundFlag $name("$name", "$desc",')
comp_code.indent()
last = len(compound) - 1
for i,flag in enumerate(compound):
if i != last:
comp_code('&$flag,')
else:
comp_code('&$flag);')
comp_code.dedent()
code.append(comp_code)
code()
code('} // namespace Debug')
code.write(str(target[0]))
def makeDebugFlagHH(target, source, env):
assert(len(target) == 1 and len(source) == 1)
val = eval(source[0].get_contents())
name, compound, desc = val
code = code_formatter()
# file header boilerplate
code('''\
/*
* DO NOT EDIT THIS FILE! Automatically generated by SCons.
*/
#ifndef __DEBUG_${name}_HH__
#define __DEBUG_${name}_HH__
namespace Debug {
''')
if compound:
code('class CompoundFlag;')
code('class SimpleFlag;')
if compound:
code('extern CompoundFlag $name;')
for flag in compound:
code('extern SimpleFlag $flag;')
else:
code('extern SimpleFlag $name;')
code('''
}
#endif // __DEBUG_${name}_HH__
''')
code.write(str(target[0]))
for name,flag in sorted(debug_flags.iteritems()):
n, compound, desc = flag
assert n == name
hh_file = 'debug/%s.hh' % name
env.Command(hh_file, Value(flag),
MakeAction(makeDebugFlagHH, Transform("TRACING", 0)))
env.Depends(SWIG, hh_file)
env.Command('debug/flags.cc', Value(debug_flags),
MakeAction(makeDebugFlagCC, Transform("TRACING", 0)))
env.Depends(SWIG, 'debug/flags.cc')
Source('debug/flags.cc')
# version tags
tags = \
env.Command('sim/tags.cc', None,
MakeAction('util/cpt_upgrader.py --get-cc-file > $TARGET',
Transform("VER TAGS")))
env.AlwaysBuild(tags)
# Embed python files. All .py files that have been indicated by a
# PySource() call in a SConscript need to be embedded into the M5
# library. To do that, we compile the file to byte code, marshal the
# byte code, compress it, and then generate a c++ file that
# inserts the result into an array.
def embedPyFile(target, source, env):
def c_str(string):
if string is None:
return "0"
return '"%s"' % string
'''Action function to compile a .py into a code object, marshal
it, compress it, and stick it into an asm file so the code appears
as just bytes with a label in the data section'''
src = file(str(source[0]), 'r').read()
pysource = PySource.tnodes[source[0]]
compiled = compile(src, pysource.abspath, 'exec')
marshalled = marshal.dumps(compiled)
compressed = zlib.compress(marshalled)
data = compressed
sym = pysource.symname
code = code_formatter()
code('''\
#include "sim/init.hh"
namespace {
clang/gcc: Fix compilation issues with clang 3.0 and gcc 4.6 This patch addresses a number of minor issues that cause problems when compiling with clang >= 3.0 and gcc >= 4.6. Most importantly, it avoids using the deprecated ext/hash_map and instead uses unordered_map (and similarly so for the hash_set). To make use of the new STL containers, g++ and clang has to be invoked with "-std=c++0x", and this is now added for all gcc versions >= 4.6, and for clang >= 3.0. For gcc >= 4.3 and <= 4.5 and clang <= 3.0 we use the tr1 unordered_map to avoid the deprecation warning. The addition of c++0x in turn causes a few problems, as the compiler is more stringent and adds a number of new warnings. Below, the most important issues are enumerated: 1) the use of namespaces is more strict, e.g. for isnan, and all headers opening the entire namespace std are now fixed. 2) another other issue caused by the more stringent compiler is the narrowing of the embedded python, which used to be a char array, and is now unsigned char since there were values larger than 128. 3) a particularly odd issue that arose with the new c++0x behaviour is found in range.hh, where the operator< causes gcc to complain about the template type parsing (the "<" is interpreted as the beginning of a template argument), and the problem seems to be related to the begin/end members introduced for the range-type iteration, which is a new feature in c++11. As a minor update, this patch also fixes the build flags for the clang debug target that used to be shared with gcc and incorrectly use "-ggdb".
2012-04-14 11:43:31 +02:00
const uint8_t data_${sym}[] = {
''')
code.indent()
step = 16
for i in xrange(0, len(data), step):
x = array.array('B', data[i:i+step])
code(''.join('%d,' % d for d in x))
code.dedent()
2016-01-19 20:28:22 +01:00
code('''};
EmbeddedPython embedded_${sym}(
${{c_str(pysource.arcname)}},
${{c_str(pysource.abspath)}},
${{c_str(pysource.modpath)}},
data_${sym},
${{len(data)}},
${{len(marshalled)}});
} // anonymous namespace
''')
code.write(str(target[0]))
for source in PySource.all:
env.Command(source.cpp, source.tnode,
MakeAction(embedPyFile, Transform("EMBED PY")))
env.Depends(SWIG, source.cpp)
Source(source.cpp, skip_no_python=True)
########################################################################
#
# Define binaries. Each different build type (debug, opt, etc.) gets
# a slightly different build environment.
#
# List of constructed environments to pass back to SConstruct
date_source = Source('base/date.cc', skip_lib=True)
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
# Capture this directory for the closure makeEnv, otherwise when it is
# called, it won't know what directory it should use.
variant_dir = Dir('.').path
def variant(*path):
return os.path.join(variant_dir, *path)
def variantd(*path):
return variant(*path)+'/'
# Function to create a new build environment as clone of current
# environment 'env' with modified object suffix and optional stripped
# binary. Additional keyword arguments are appended to corresponding
# build environment vars.
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
def makeEnv(env, label, objsfx, strip = False, **kwargs):
# SCons doesn't know to append a library suffix when there is a '.' in the
# name. Use '_' instead.
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
libname = variant('gem5_' + label)
exename = variant('gem5.' + label)
secondary_exename = variant('m5.' + label)
new_env = env.Clone(OBJSUFFIX=objsfx, SHOBJSUFFIX=objsfx + 's')
new_env.Label = label
new_env.Append(**kwargs)
swig_env = new_env.Clone()
# Both gcc and clang have issues with unused labels and values in
# the SWIG generated code
swig_env.Append(CCFLAGS=['-Wno-unused-label', '-Wno-unused-value'])
if env['GCC']:
# Depending on the SWIG version, we also need to supress
# warnings about uninitialized variables and missing field
# initializers.
swig_env.Append(CCFLAGS=['-Wno-uninitialized',
'-Wno-missing-field-initializers',
'-Wno-unused-but-set-variable',
'-Wno-maybe-uninitialized',
'-Wno-type-limits'])
# The address sanitizer is available for gcc >= 4.8
if GetOption('with_asan'):
if GetOption('with_ubsan') and \
compareVersions(env['GCC_VERSION'], '4.9') >= 0:
new_env.Append(CCFLAGS=['-fsanitize=address,undefined',
'-fno-omit-frame-pointer'])
new_env.Append(LINKFLAGS='-fsanitize=address,undefined')
else:
new_env.Append(CCFLAGS=['-fsanitize=address',
'-fno-omit-frame-pointer'])
new_env.Append(LINKFLAGS='-fsanitize=address')
# Only gcc >= 4.9 supports UBSan, so check both the version
# and the command-line option before adding the compiler and
# linker flags.
elif GetOption('with_ubsan') and \
compareVersions(env['GCC_VERSION'], '4.9') >= 0:
new_env.Append(CCFLAGS='-fsanitize=undefined')
new_env.Append(LINKFLAGS='-fsanitize=undefined')
if env['CLANG']:
swig_env.Append(CCFLAGS=['-Wno-sometimes-uninitialized',
'-Wno-deprecated-register',
'-Wno-tautological-compare'])
# We require clang >= 3.1, so there is no need to check any
# versions here.
if GetOption('with_ubsan'):
if GetOption('with_asan'):
new_env.Append(CCFLAGS=['-fsanitize=address,undefined',
'-fno-omit-frame-pointer'])
new_env.Append(LINKFLAGS='-fsanitize=address,undefined')
else:
new_env.Append(CCFLAGS='-fsanitize=undefined')
new_env.Append(LINKFLAGS='-fsanitize=undefined')
elif GetOption('with_asan'):
new_env.Append(CCFLAGS=['-fsanitize=address',
'-fno-omit-frame-pointer'])
new_env.Append(LINKFLAGS='-fsanitize=address')
werror_env = new_env.Clone()
# Treat warnings as errors but white list some warnings that we
# want to allow (e.g., deprecation warnings).
werror_env.Append(CCFLAGS=['-Werror',
'-Wno-error=deprecated-declarations',
'-Wno-error=deprecated',
])
def make_obj(source, static, extra_deps = None):
'''This function adds the specified source to the correct
build environment, and returns the corresponding SCons Object
nodes'''
if source.swig:
env = swig_env
elif source.Werror:
env = werror_env
else:
env = new_env
if static:
obj = env.StaticObject(source.tnode)
else:
obj = env.SharedObject(source.tnode)
if extra_deps:
env.Depends(obj, extra_deps)
return obj
lib_guards = {'main': False, 'skip_lib': False}
# Without Python, leave out all SWIG and Python content from the
# library builds. The option doesn't affect gem5 built as a program
if GetOption('without_python'):
lib_guards['skip_no_python'] = False
static_objs = [ make_obj(s, True) for s in Source.get(**lib_guards) ]
shared_objs = [ make_obj(s, False) for s in Source.get(**lib_guards) ]
static_date = make_obj(date_source, static=True, extra_deps=static_objs)
static_objs.append(static_date)
shared_date = make_obj(date_source, static=False, extra_deps=shared_objs)
shared_objs.append(shared_date)
# First make a library of everything but main() so other programs can
# link against m5.
static_lib = new_env.StaticLibrary(libname, static_objs)
shared_lib = new_env.SharedLibrary(libname, shared_objs)
# Now link a stub with main() and the static library.
main_objs = [ make_obj(s, True) for s in Source.get(main=True) ]
for test in UnitTest.all:
flags = { test.target : True }
test_sources = Source.get(**flags)
test_objs = [ make_obj(s, static=True) for s in test_sources ]
2012-05-11 01:04:28 +02:00
if test.main:
test_objs += main_objs
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
path = variant('unittest/%s.%s' % (test.target, label))
new_env.Program(path, test_objs + static_objs)
progname = exename
if strip:
progname += '.unstripped'
targets = new_env.Program(progname, main_objs + static_objs)
if strip:
if sys.platform == 'sunos5':
cmd = 'cp $SOURCE $TARGET; strip $TARGET'
else:
cmd = 'strip $SOURCE -o $TARGET'
targets = new_env.Command(exename, progname,
MakeAction(cmd, Transform("STRIP")))
new_env.Command(secondary_exename, exename,
MakeAction('ln $SOURCE $TARGET', Transform("HARDLINK")))
new_env.M5Binary = targets[0]
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
return new_env
# Start out with the compiler flags common to all compilers,
# i.e. they all use -g for opt and -g -pg for prof
ccflags = {'debug' : [], 'opt' : ['-g'], 'fast' : [], 'prof' : ['-g', '-pg'],
'perf' : ['-g']}
# Start out with the linker flags common to all linkers, i.e. -pg for
# prof, and -lprofiler for perf. The -lprofile flag is surrounded by
# no-as-needed and as-needed as the binutils linker is too clever and
# simply doesn't link to the library otherwise.
ldflags = {'debug' : [], 'opt' : [], 'fast' : [], 'prof' : ['-pg'],
'perf' : ['-Wl,--no-as-needed', '-lprofiler', '-Wl,--as-needed']}
gcc: Enable Link-Time Optimization for gcc >= 4.6 This patch adds Link-Time Optimization when building the fast target using gcc >= 4.6, and adds a scons flag to disable it (-no-lto). No check is performed to guarantee that the linker supports LTO and use of the linker plugin, so the user has to ensure that binutils GNU ld >= 2.21 or the gold linker is available. Typically, if gcc >= 4.6 is available, the latter should not be a problem. Currently the LTO option is only useful for gcc >= 4.6, due to the limited support on clang and earlier versions of gcc. The intention is to also add support for clang once the LTO integration matures. The same number of jobs is used for the parallel phase of LTO as the jobs specified on the scons command line, using the -flto=n flag that was introduced with gcc 4.6. The gold linker also supports concurrent and incremental linking, but this is not used at this point. The compilation and linking time is increased by almost 50% on average, although ARM seems to be particularly demanding with an increase of almost 100%. Also beware when using this as gcc uses a tremendous amount of memory and temp space in the process. You have been warned. After some careful consideration, and plenty discussions, the flag is only added to the fast target, and the warning that was issued in an earlier version of this patch is now removed. Similarly, the flag used to enable LTO, now the default is to use it, and the flag has been modified to disable LTO. The rationale behind this decision is that opt is used for development, whereas fast is only used for long runs, e.g. regressions or more elaborate experiments where the additional compile and link time is amortized by a much larger run time. When it comes to the return on investment, the regression seems to be roughly 15% faster with LTO. For a bit more detail, I ran twolf on ARM.fast, with three repeated runs, and they all finish within 42 minutes (+- 25 seconds) without LTO and 31 minutes (+- 25 seconds) with LTO, i.e. LTO gives an impressive >25% speed-up for this case. Without LTO (ARM.fast twolf) real 42m37.632s user 42m34.448s sys 0m0.390s real 41m51.793s user 41m50.384s sys 0m0.131s real 41m45.491s user 41m39.791s sys 0m0.139s With LTO (ARM.fast twolf) real 30m33.588s user 30m5.701s sys 0m0.141s real 31m27.791s user 31m24.674s sys 0m0.111s real 31m25.500s user 31m16.731s sys 0m0.106s
2012-09-14 18:13:22 +02:00
# For Link Time Optimization, the optimisation flags used to compile
# individual files are decoupled from those used at link time
# (i.e. you can compile with -O3 and perform LTO with -O0), so we need
# to also update the linker flags based on the target.
clang/gcc: Fix compilation issues with clang 3.0 and gcc 4.6 This patch addresses a number of minor issues that cause problems when compiling with clang >= 3.0 and gcc >= 4.6. Most importantly, it avoids using the deprecated ext/hash_map and instead uses unordered_map (and similarly so for the hash_set). To make use of the new STL containers, g++ and clang has to be invoked with "-std=c++0x", and this is now added for all gcc versions >= 4.6, and for clang >= 3.0. For gcc >= 4.3 and <= 4.5 and clang <= 3.0 we use the tr1 unordered_map to avoid the deprecation warning. The addition of c++0x in turn causes a few problems, as the compiler is more stringent and adds a number of new warnings. Below, the most important issues are enumerated: 1) the use of namespaces is more strict, e.g. for isnan, and all headers opening the entire namespace std are now fixed. 2) another other issue caused by the more stringent compiler is the narrowing of the embedded python, which used to be a char array, and is now unsigned char since there were values larger than 128. 3) a particularly odd issue that arose with the new c++0x behaviour is found in range.hh, where the operator< causes gcc to complain about the template type parsing (the "<" is interpreted as the beginning of a template argument), and the problem seems to be related to the begin/end members introduced for the range-type iteration, which is a new feature in c++11. As a minor update, this patch also fixes the build flags for the clang debug target that used to be shared with gcc and incorrectly use "-ggdb".
2012-04-14 11:43:31 +02:00
if env['GCC']:
make our code a little more standards compliant pretty close to compiling w/ suns compiler briefly: add dummy return after panic()/fatal() split out flags by compiler vendor include cstring and cmath where appropriate use std namespace for string ops SConstruct: Add code to detect compiler and choose cflags based on detected compiler Fix zlib check to work with suncc src/SConscript: split out flags by compiler vendor src/arch/sparc/isa/decoder.isa: use correct namespace for sqrt src/arch/sparc/isa/formats/basic.isa: add dummy return around panic src/arch/sparc/isa/formats/integerop.isa: use correct namespace for stringops src/arch/sparc/isa/includes.isa: include cstring and cmath where appropriate src/arch/sparc/isa_traits.hh: remove dangling comma src/arch/sparc/system.cc: dummy return to make sun cc front end happy src/arch/sparc/tlb.cc: src/base/compression/lzss_compression.cc: use std namespace for string ops src/arch/sparc/utility.hh: no reason to say something is unsigned unsigned int src/base/compression/null_compression.hh: dummy returns to for suncc front end src/base/cprintf.hh: use standard variadic argument syntax instead of gnuc specefic renaming src/base/hashmap.hh: don't need to define hash for suncc src/base/hostinfo.cc: need stdio.h for sprintf src/base/loader/object_file.cc: munmap is in std namespace not null src/base/misc.hh: use M5 generic noreturn macros use standard variadic macro __VA_ARGS__ src/base/pollevent.cc: we need file.h for file flags src/base/random.cc: mess with include files to make suncc happy src/base/remote_gdb.cc: malloc memory for function instead of having a non-constant in an array size src/base/statistics.hh: use std namespace for floor src/base/stats/text.cc: include math.h for rint (cmath won't work) src/base/time.cc: use suncc version of ctime_r src/base/time.hh: change macro to work with both gcc and suncc src/base/timebuf.hh: include cstring from memset and use std:: src/base/trace.hh: change variadic macros to be normal format src/cpu/SConscript: add dummy returns where appropriate src/cpu/activity.cc: include cstring for memset src/cpu/exetrace.hh: include cstring fro memcpy src/cpu/simple/base.hh: add dummy return for panic src/dev/baddev.cc: src/dev/pciconfigall.cc: src/dev/platform.cc: src/dev/sparc/t1000.cc: add dummy return where appropriate src/dev/ide_atareg.h: make define work for both gnuc and suncc src/dev/io_device.hh: add dummy returns where approirate src/dev/pcidev.hh: src/mem/cache/cache_impl.hh: src/mem/cache/miss/blocking_buffer.cc: src/mem/cache/tags/lru.hh: src/mem/cache/tags/split.hh: src/mem/cache/tags/split_lifo.hh: src/mem/cache/tags/split_lru.hh: src/mem/dram.cc: src/mem/packet.cc: src/mem/port.cc: include cstring for string ops src/dev/sparc/mm_disk.cc: add dummy return where appropriate include cstring for string ops src/mem/cache/miss/blocking_buffer.hh: src/mem/port.hh: Add dummy return where appropriate src/mem/cache/tags/iic.cc: cast hastSets to double for log() call src/mem/physical.cc: cast pmemAddr to char* for munmap src/sim/byteswap.hh: make define work for suncc and gnuc --HG-- extra : convert_revision : ef8a1f1064e43b6c39838a85c01aee4f795497bd
2007-01-27 00:48:51 +01:00
if sys.platform == 'sunos5':
ccflags['debug'] += ['-gstabs+']
make our code a little more standards compliant pretty close to compiling w/ suns compiler briefly: add dummy return after panic()/fatal() split out flags by compiler vendor include cstring and cmath where appropriate use std namespace for string ops SConstruct: Add code to detect compiler and choose cflags based on detected compiler Fix zlib check to work with suncc src/SConscript: split out flags by compiler vendor src/arch/sparc/isa/decoder.isa: use correct namespace for sqrt src/arch/sparc/isa/formats/basic.isa: add dummy return around panic src/arch/sparc/isa/formats/integerop.isa: use correct namespace for stringops src/arch/sparc/isa/includes.isa: include cstring and cmath where appropriate src/arch/sparc/isa_traits.hh: remove dangling comma src/arch/sparc/system.cc: dummy return to make sun cc front end happy src/arch/sparc/tlb.cc: src/base/compression/lzss_compression.cc: use std namespace for string ops src/arch/sparc/utility.hh: no reason to say something is unsigned unsigned int src/base/compression/null_compression.hh: dummy returns to for suncc front end src/base/cprintf.hh: use standard variadic argument syntax instead of gnuc specefic renaming src/base/hashmap.hh: don't need to define hash for suncc src/base/hostinfo.cc: need stdio.h for sprintf src/base/loader/object_file.cc: munmap is in std namespace not null src/base/misc.hh: use M5 generic noreturn macros use standard variadic macro __VA_ARGS__ src/base/pollevent.cc: we need file.h for file flags src/base/random.cc: mess with include files to make suncc happy src/base/remote_gdb.cc: malloc memory for function instead of having a non-constant in an array size src/base/statistics.hh: use std namespace for floor src/base/stats/text.cc: include math.h for rint (cmath won't work) src/base/time.cc: use suncc version of ctime_r src/base/time.hh: change macro to work with both gcc and suncc src/base/timebuf.hh: include cstring from memset and use std:: src/base/trace.hh: change variadic macros to be normal format src/cpu/SConscript: add dummy returns where appropriate src/cpu/activity.cc: include cstring for memset src/cpu/exetrace.hh: include cstring fro memcpy src/cpu/simple/base.hh: add dummy return for panic src/dev/baddev.cc: src/dev/pciconfigall.cc: src/dev/platform.cc: src/dev/sparc/t1000.cc: add dummy return where appropriate src/dev/ide_atareg.h: make define work for both gnuc and suncc src/dev/io_device.hh: add dummy returns where approirate src/dev/pcidev.hh: src/mem/cache/cache_impl.hh: src/mem/cache/miss/blocking_buffer.cc: src/mem/cache/tags/lru.hh: src/mem/cache/tags/split.hh: src/mem/cache/tags/split_lifo.hh: src/mem/cache/tags/split_lru.hh: src/mem/dram.cc: src/mem/packet.cc: src/mem/port.cc: include cstring for string ops src/dev/sparc/mm_disk.cc: add dummy return where appropriate include cstring for string ops src/mem/cache/miss/blocking_buffer.hh: src/mem/port.hh: Add dummy return where appropriate src/mem/cache/tags/iic.cc: cast hastSets to double for log() call src/mem/physical.cc: cast pmemAddr to char* for munmap src/sim/byteswap.hh: make define work for suncc and gnuc --HG-- extra : convert_revision : ef8a1f1064e43b6c39838a85c01aee4f795497bd
2007-01-27 00:48:51 +01:00
else:
ccflags['debug'] += ['-ggdb3']
ldflags['debug'] += ['-O0']
gcc: Enable Link-Time Optimization for gcc >= 4.6 This patch adds Link-Time Optimization when building the fast target using gcc >= 4.6, and adds a scons flag to disable it (-no-lto). No check is performed to guarantee that the linker supports LTO and use of the linker plugin, so the user has to ensure that binutils GNU ld >= 2.21 or the gold linker is available. Typically, if gcc >= 4.6 is available, the latter should not be a problem. Currently the LTO option is only useful for gcc >= 4.6, due to the limited support on clang and earlier versions of gcc. The intention is to also add support for clang once the LTO integration matures. The same number of jobs is used for the parallel phase of LTO as the jobs specified on the scons command line, using the -flto=n flag that was introduced with gcc 4.6. The gold linker also supports concurrent and incremental linking, but this is not used at this point. The compilation and linking time is increased by almost 50% on average, although ARM seems to be particularly demanding with an increase of almost 100%. Also beware when using this as gcc uses a tremendous amount of memory and temp space in the process. You have been warned. After some careful consideration, and plenty discussions, the flag is only added to the fast target, and the warning that was issued in an earlier version of this patch is now removed. Similarly, the flag used to enable LTO, now the default is to use it, and the flag has been modified to disable LTO. The rationale behind this decision is that opt is used for development, whereas fast is only used for long runs, e.g. regressions or more elaborate experiments where the additional compile and link time is amortized by a much larger run time. When it comes to the return on investment, the regression seems to be roughly 15% faster with LTO. For a bit more detail, I ran twolf on ARM.fast, with three repeated runs, and they all finish within 42 minutes (+- 25 seconds) without LTO and 31 minutes (+- 25 seconds) with LTO, i.e. LTO gives an impressive >25% speed-up for this case. Without LTO (ARM.fast twolf) real 42m37.632s user 42m34.448s sys 0m0.390s real 41m51.793s user 41m50.384s sys 0m0.131s real 41m45.491s user 41m39.791s sys 0m0.139s With LTO (ARM.fast twolf) real 30m33.588s user 30m5.701s sys 0m0.141s real 31m27.791s user 31m24.674s sys 0m0.111s real 31m25.500s user 31m16.731s sys 0m0.106s
2012-09-14 18:13:22 +02:00
# opt, fast, prof and perf all share the same cc flags, also add
# the optimization to the ldflags as LTO defers the optimization
# to link time
for target in ['opt', 'fast', 'prof', 'perf']:
ccflags[target] += ['-O3']
gcc: Enable Link-Time Optimization for gcc >= 4.6 This patch adds Link-Time Optimization when building the fast target using gcc >= 4.6, and adds a scons flag to disable it (-no-lto). No check is performed to guarantee that the linker supports LTO and use of the linker plugin, so the user has to ensure that binutils GNU ld >= 2.21 or the gold linker is available. Typically, if gcc >= 4.6 is available, the latter should not be a problem. Currently the LTO option is only useful for gcc >= 4.6, due to the limited support on clang and earlier versions of gcc. The intention is to also add support for clang once the LTO integration matures. The same number of jobs is used for the parallel phase of LTO as the jobs specified on the scons command line, using the -flto=n flag that was introduced with gcc 4.6. The gold linker also supports concurrent and incremental linking, but this is not used at this point. The compilation and linking time is increased by almost 50% on average, although ARM seems to be particularly demanding with an increase of almost 100%. Also beware when using this as gcc uses a tremendous amount of memory and temp space in the process. You have been warned. After some careful consideration, and plenty discussions, the flag is only added to the fast target, and the warning that was issued in an earlier version of this patch is now removed. Similarly, the flag used to enable LTO, now the default is to use it, and the flag has been modified to disable LTO. The rationale behind this decision is that opt is used for development, whereas fast is only used for long runs, e.g. regressions or more elaborate experiments where the additional compile and link time is amortized by a much larger run time. When it comes to the return on investment, the regression seems to be roughly 15% faster with LTO. For a bit more detail, I ran twolf on ARM.fast, with three repeated runs, and they all finish within 42 minutes (+- 25 seconds) without LTO and 31 minutes (+- 25 seconds) with LTO, i.e. LTO gives an impressive >25% speed-up for this case. Without LTO (ARM.fast twolf) real 42m37.632s user 42m34.448s sys 0m0.390s real 41m51.793s user 41m50.384s sys 0m0.131s real 41m45.491s user 41m39.791s sys 0m0.139s With LTO (ARM.fast twolf) real 30m33.588s user 30m5.701s sys 0m0.141s real 31m27.791s user 31m24.674s sys 0m0.111s real 31m25.500s user 31m16.731s sys 0m0.106s
2012-09-14 18:13:22 +02:00
ldflags[target] += ['-O3']
ccflags['fast'] += env['LTO_CCFLAGS']
ldflags['fast'] += env['LTO_LDFLAGS']
clang/gcc: Fix compilation issues with clang 3.0 and gcc 4.6 This patch addresses a number of minor issues that cause problems when compiling with clang >= 3.0 and gcc >= 4.6. Most importantly, it avoids using the deprecated ext/hash_map and instead uses unordered_map (and similarly so for the hash_set). To make use of the new STL containers, g++ and clang has to be invoked with "-std=c++0x", and this is now added for all gcc versions >= 4.6, and for clang >= 3.0. For gcc >= 4.3 and <= 4.5 and clang <= 3.0 we use the tr1 unordered_map to avoid the deprecation warning. The addition of c++0x in turn causes a few problems, as the compiler is more stringent and adds a number of new warnings. Below, the most important issues are enumerated: 1) the use of namespaces is more strict, e.g. for isnan, and all headers opening the entire namespace std are now fixed. 2) another other issue caused by the more stringent compiler is the narrowing of the embedded python, which used to be a char array, and is now unsigned char since there were values larger than 128. 3) a particularly odd issue that arose with the new c++0x behaviour is found in range.hh, where the operator< causes gcc to complain about the template type parsing (the "<" is interpreted as the beginning of a template argument), and the problem seems to be related to the begin/end members introduced for the range-type iteration, which is a new feature in c++11. As a minor update, this patch also fixes the build flags for the clang debug target that used to be shared with gcc and incorrectly use "-ggdb".
2012-04-14 11:43:31 +02:00
elif env['CLANG']:
ccflags['debug'] += ['-g', '-O0']
# opt, fast, prof and perf all share the same cc flags
for target in ['opt', 'fast', 'prof', 'perf']:
ccflags[target] += ['-O3']
else:
make our code a little more standards compliant pretty close to compiling w/ suns compiler briefly: add dummy return after panic()/fatal() split out flags by compiler vendor include cstring and cmath where appropriate use std namespace for string ops SConstruct: Add code to detect compiler and choose cflags based on detected compiler Fix zlib check to work with suncc src/SConscript: split out flags by compiler vendor src/arch/sparc/isa/decoder.isa: use correct namespace for sqrt src/arch/sparc/isa/formats/basic.isa: add dummy return around panic src/arch/sparc/isa/formats/integerop.isa: use correct namespace for stringops src/arch/sparc/isa/includes.isa: include cstring and cmath where appropriate src/arch/sparc/isa_traits.hh: remove dangling comma src/arch/sparc/system.cc: dummy return to make sun cc front end happy src/arch/sparc/tlb.cc: src/base/compression/lzss_compression.cc: use std namespace for string ops src/arch/sparc/utility.hh: no reason to say something is unsigned unsigned int src/base/compression/null_compression.hh: dummy returns to for suncc front end src/base/cprintf.hh: use standard variadic argument syntax instead of gnuc specefic renaming src/base/hashmap.hh: don't need to define hash for suncc src/base/hostinfo.cc: need stdio.h for sprintf src/base/loader/object_file.cc: munmap is in std namespace not null src/base/misc.hh: use M5 generic noreturn macros use standard variadic macro __VA_ARGS__ src/base/pollevent.cc: we need file.h for file flags src/base/random.cc: mess with include files to make suncc happy src/base/remote_gdb.cc: malloc memory for function instead of having a non-constant in an array size src/base/statistics.hh: use std namespace for floor src/base/stats/text.cc: include math.h for rint (cmath won't work) src/base/time.cc: use suncc version of ctime_r src/base/time.hh: change macro to work with both gcc and suncc src/base/timebuf.hh: include cstring from memset and use std:: src/base/trace.hh: change variadic macros to be normal format src/cpu/SConscript: add dummy returns where appropriate src/cpu/activity.cc: include cstring for memset src/cpu/exetrace.hh: include cstring fro memcpy src/cpu/simple/base.hh: add dummy return for panic src/dev/baddev.cc: src/dev/pciconfigall.cc: src/dev/platform.cc: src/dev/sparc/t1000.cc: add dummy return where appropriate src/dev/ide_atareg.h: make define work for both gnuc and suncc src/dev/io_device.hh: add dummy returns where approirate src/dev/pcidev.hh: src/mem/cache/cache_impl.hh: src/mem/cache/miss/blocking_buffer.cc: src/mem/cache/tags/lru.hh: src/mem/cache/tags/split.hh: src/mem/cache/tags/split_lifo.hh: src/mem/cache/tags/split_lru.hh: src/mem/dram.cc: src/mem/packet.cc: src/mem/port.cc: include cstring for string ops src/dev/sparc/mm_disk.cc: add dummy return where appropriate include cstring for string ops src/mem/cache/miss/blocking_buffer.hh: src/mem/port.hh: Add dummy return where appropriate src/mem/cache/tags/iic.cc: cast hastSets to double for log() call src/mem/physical.cc: cast pmemAddr to char* for munmap src/sim/byteswap.hh: make define work for suncc and gnuc --HG-- extra : convert_revision : ef8a1f1064e43b6c39838a85c01aee4f795497bd
2007-01-27 00:48:51 +01:00
print 'Unknown compiler, please fix compiler options'
Exit(1)
# To speed things up, we only instantiate the build environments we
# need. We try to identify the needed environment for each target; if
# we can't, we fall back on instantiating all the environments just to
# be safe.
target_types = ['debug', 'opt', 'fast', 'prof', 'perf']
obj2target = {'do': 'debug', 'o': 'opt', 'fo': 'fast', 'po': 'prof',
'gpo' : 'perf'}
def identifyTarget(t):
ext = t.split('.')[-1]
if ext in target_types:
return ext
if obj2target.has_key(ext):
return obj2target[ext]
match = re.search(r'/tests/([^/]+)/', t)
if match and match.group(1) in target_types:
return match.group(1)
return 'all'
needed_envs = [identifyTarget(target) for target in BUILD_TARGETS]
if 'all' in needed_envs:
needed_envs += target_types
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
def makeEnvirons(target, source, env):
# cause any later Source() calls to be fatal, as a diagnostic.
Source.done()
envList = []
# Debug binary
if 'debug' in needed_envs:
envList.append(
makeEnv(env, 'debug', '.do',
CCFLAGS = Split(ccflags['debug']),
CPPDEFINES = ['DEBUG', 'TRACING_ON=1'],
LINKFLAGS = Split(ldflags['debug'])))
# Optimized binary
if 'opt' in needed_envs:
envList.append(
makeEnv(env, 'opt', '.o',
CCFLAGS = Split(ccflags['opt']),
CPPDEFINES = ['TRACING_ON=1'],
LINKFLAGS = Split(ldflags['opt'])))
# "Fast" binary
if 'fast' in needed_envs:
envList.append(
makeEnv(env, 'fast', '.fo', strip = True,
CCFLAGS = Split(ccflags['fast']),
CPPDEFINES = ['NDEBUG', 'TRACING_ON=0'],
LINKFLAGS = Split(ldflags['fast'])))
# Profiled binary using gprof
if 'prof' in needed_envs:
envList.append(
makeEnv(env, 'prof', '.po',
CCFLAGS = Split(ccflags['prof']),
CPPDEFINES = ['NDEBUG', 'TRACING_ON=0'],
LINKFLAGS = Split(ldflags['prof'])))
# Profiled binary using google-pprof
if 'perf' in needed_envs:
envList.append(
makeEnv(env, 'perf', '.gpo',
CCFLAGS = Split(ccflags['perf']),
CPPDEFINES = ['NDEBUG', 'TRACING_ON=0'],
LINKFLAGS = Split(ldflags['perf'])))
# Set up the regression tests for each build.
for e in envList:
SConscript(os.path.join(env.root.abspath, 'tests', 'SConscript'),
arch: teach ISA parser how to split code across files This patch encompasses several interrelated and interdependent changes to the ISA generation step. The end goal is to reduce the size of the generated compilation units for instruction execution and decoding so that batch compilation can proceed with all CPUs active without exhausting physical memory. The ISA parser (src/arch/isa_parser.py) has been improved so that it can accept 'split [output_type];' directives at the top level of the grammar and 'split(output_type)' python calls within 'exec {{ ... }}' blocks. This has the effect of "splitting" the files into smaller compilation units. I use air-quotes around "splitting" because the files themselves are not split, but preprocessing directives are inserted to have the same effect. Architecturally, the ISA parser has had some changes in how it works. In general, it emits code sooner. It doesn't generate per-CPU files, and instead defers to the C preprocessor to create the duplicate copies for each CPU type. Likewise there are more files emitted and the C preprocessor does more substitution that used to be done by the ISA parser. Finally, the build system (SCons) needs to be able to cope with a dynamic list of source files coming out of the ISA parser. The changes to the SCons{cript,truct} files support this. In broad strokes, the targets requested on the command line are hidden from SCons until all the build dependencies are determined, otherwise it would try, realize it can't reach the goal, and terminate in failure. Since build steps (i.e. running the ISA parser) must be taken to determine the file list, several new build stages have been inserted at the very start of the build. First, the build dependencies from the ISA parser will be emitted to arch/$ISA/generated/inc.d, which is then read by a new SCons builder to finalize the dependencies. (Once inc.d exists, the ISA parser will not need to be run to complete this step.) Once the dependencies are known, the 'Environments' are made by the makeEnv() function. This function used to be called before the build began but now happens during the build. It is easy to see that this step is quite slow; this is a known issue and it's important to realize that it was already slow, but there was no obvious cause to attribute it to since nothing was displayed to the terminal. Since new steps that used to be performed serially are now in a potentially-parallel build phase, the pathname handling in the SCons scripts has been tightened up to deal with chdir() race conditions. In general, pathnames are computed earlier and more likely to be stored, passed around, and processed as absolute paths rather than relative paths. In the end, some of these issues had to be fixed by inserting serializing dependencies in the build. Minor note: For the null ISA, we just provide a dummy inc.d so SCons is never compelled to try to generate it. While it seems slightly wrong to have anything in src/arch/*/generated (i.e. a non-generated 'generated' file), it's by far the simplest solution.
2014-05-10 00:58:47 +02:00
variant_dir = variantd('tests', e.Label),
exports = { 'env' : e }, duplicate = False)
# The MakeEnvirons Builder defers the full dependency collection until
# after processing the ISA definition (due to dynamically generated
# source files). Add this dependency to all targets so they will wait
# until the environments are completely set up. Otherwise, a second
# process (e.g. -j2 or higher) will try to compile the requested target,
# not know how, and fail.
env.Append(BUILDERS = {'MakeEnvirons' :
Builder(action=MakeAction(makeEnvirons,
Transform("ENVIRONS", 1)))})
isa_target = env['PHONY_BASE'] + '-deps'
environs = env['PHONY_BASE'] + '-environs'
env.Depends('#all-deps', isa_target)
env.Depends('#all-environs', environs)
env.ScanISA(isa_target, File('arch/%s/generated/inc.d' % env['TARGET_ISA']))
envSetup = env.MakeEnvirons(environs, isa_target)
# make sure no -deps targets occur before all ISAs are complete
env.Depends(isa_target, '#all-isas')
# likewise for -environs targets and all the -deps targets
env.Depends(environs, '#all-deps')