gem5/src/python/m5/SimObject.py

1528 lines
56 KiB
Python
Raw Normal View History

# Copyright (c) 2012 ARM Limited
# All rights reserved.
#
# The license below extends only to copyright in the software and shall
# not be construed as granting a license to any other intellectual
# property including but not limited to intellectual property relating
# to a hardware implementation of the functionality of the software
# licensed hereunder. You may use the software subject to the license
# terms below provided that you ensure that this notice is replicated
# unmodified and in its entirety in all distributions of the software,
# modified or unmodified, in source code or in binary form.
#
# Copyright (c) 2004-2006 The Regents of The University of Michigan
# Copyright (c) 2010-20013 Advanced Micro Devices, Inc.
# Copyright (c) 2013 Mark D. Hill and David A. Wood
# 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: Steve Reinhardt
# Nathan Binkert
# Andreas Hansson
import sys
from types import FunctionType, MethodType, ModuleType
import m5
from m5.util import *
# Have to import params up top since Param is referenced on initial
# load (when SimObject class references Param to create a class
# variable, the 'name' param)...
from m5.params import *
# There are a few things we need that aren't in params.__all__ since
# normal users don't need them
from m5.params import ParamDesc, VectorParamDesc, \
isNullPointer, SimObjectVector, Port
from m5.proxy import *
from m5.proxy import isproxy
#####################################################################
#
# M5 Python Configuration Utility
#
# The basic idea is to write simple Python programs that build Python
# objects corresponding to M5 SimObjects for the desired simulation
# configuration. For now, the Python emits a .ini file that can be
# parsed by M5. In the future, some tighter integration between M5
# and the Python interpreter may allow bypassing the .ini file.
#
# Each SimObject class in M5 is represented by a Python class with the
# same name. The Python inheritance tree mirrors the M5 C++ tree
# (e.g., SimpleCPU derives from BaseCPU in both cases, and all
# SimObjects inherit from a single SimObject base class). To specify
# an instance of an M5 SimObject in a configuration, the user simply
# instantiates the corresponding Python object. The parameters for
# that SimObject are given by assigning to attributes of the Python
# object, either using keyword assignment in the constructor or in
# separate assignment statements. For example:
#
# cache = BaseCache(size='64KB')
# cache.hit_latency = 3
# cache.assoc = 8
#
# The magic lies in the mapping of the Python attributes for SimObject
# classes to the actual SimObject parameter specifications. This
# allows parameter validity checking in the Python code. Continuing
# the example above, the statements "cache.blurfl=3" or
# "cache.assoc='hello'" would both result in runtime errors in Python,
# since the BaseCache object has no 'blurfl' parameter and the 'assoc'
# parameter requires an integer, respectively. This magic is done
# primarily by overriding the special __setattr__ method that controls
# assignment to object attributes.
#
# Once a set of Python objects have been instantiated in a hierarchy,
# calling 'instantiate(obj)' (where obj is the root of the hierarchy)
# will generate a .ini file.
#
#####################################################################
# list of all SimObject classes
allClasses = {}
# dict to look up SimObjects based on path
instanceDict = {}
# Did any of the SimObjects lack a header file?
noCxxHeader = False
def public_value(key, value):
return key.startswith('_') or \
isinstance(value, (FunctionType, MethodType, ModuleType,
classmethod, type))
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 createCxxConfigDirectoryEntryFile(code, name, simobj, is_header):
entry_class = 'CxxConfigDirectoryEntry_%s' % name
param_class = '%sCxxConfigParams' % name
code('#include "params/%s.hh"' % name)
if not is_header:
for param in simobj._params.values():
if isSimObjectClass(param.ptype):
code('#include "%s"' % param.ptype._value_dict['cxx_header'])
code('#include "params/%s.hh"' % param.ptype.__name__)
else:
param.ptype.cxx_ini_predecls(code)
if is_header:
member_prefix = ''
end_of_decl = ';'
code('#include "sim/cxx_config.hh"')
code()
code('class ${param_class} : public CxxConfigParams,'
' public ${name}Params')
code('{')
code(' private:')
code.indent()
code('class DirectoryEntry : public CxxConfigDirectoryEntry')
code('{')
code(' public:')
code.indent()
code('DirectoryEntry();');
code()
code('CxxConfigParams *makeParamsObject() const')
code('{ return new ${param_class}; }')
code.dedent()
code('};')
code()
code.dedent()
code(' public:')
code.indent()
else:
member_prefix = '%s::' % param_class
end_of_decl = ''
code('#include "%s"' % simobj._value_dict['cxx_header'])
code('#include "base/str.hh"')
code('#include "cxx_config/${name}.hh"')
if simobj._ports.values() != []:
code('#include "mem/mem_object.hh"')
code('#include "mem/port.hh"')
code()
code('${member_prefix}DirectoryEntry::DirectoryEntry()');
code('{')
def cxx_bool(b):
return 'true' if b else 'false'
code.indent()
for param in simobj._params.values():
is_vector = isinstance(param, m5.params.VectorParamDesc)
is_simobj = issubclass(param.ptype, m5.SimObject.SimObject)
code('parameters["%s"] = new ParamDesc("%s", %s, %s);' %
(param.name, param.name, cxx_bool(is_vector),
cxx_bool(is_simobj)));
for port in simobj._ports.values():
is_vector = isinstance(port, m5.params.VectorPort)
is_master = port.role == 'MASTER'
code('ports["%s"] = new PortDesc("%s", %s, %s);' %
(port.name, port.name, cxx_bool(is_vector),
cxx_bool(is_master)))
code.dedent()
code('}')
code()
code('bool ${member_prefix}setSimObject(const std::string &name,')
code(' SimObject *simObject)${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('bool ret = true;')
code()
code('if (false) {')
for param in simobj._params.values():
is_vector = isinstance(param, m5.params.VectorParamDesc)
is_simobj = issubclass(param.ptype, m5.SimObject.SimObject)
if is_simobj and not is_vector:
code('} else if (name == "${{param.name}}") {')
code.indent()
code('this->${{param.name}} = '
'dynamic_cast<${{param.ptype.cxx_type}}>(simObject);')
code('if (simObject && !this->${{param.name}})')
code(' ret = false;')
code.dedent()
code('} else {')
code(' ret = false;')
code('}')
code()
code('return ret;')
code.dedent()
code('}')
code()
code('bool ${member_prefix}setSimObjectVector('
'const std::string &name,')
code(' const std::vector<SimObject *> &simObjects)${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('bool ret = true;')
code()
code('if (false) {')
for param in simobj._params.values():
is_vector = isinstance(param, m5.params.VectorParamDesc)
is_simobj = issubclass(param.ptype, m5.SimObject.SimObject)
if is_simobj and is_vector:
code('} else if (name == "${{param.name}}") {')
code.indent()
code('this->${{param.name}}.clear();')
code('for (auto i = simObjects.begin(); '
'ret && i != simObjects.end(); i ++)')
code('{')
code.indent()
code('${{param.ptype.cxx_type}} object = '
'dynamic_cast<${{param.ptype.cxx_type}}>(*i);')
code('if (*i && !object)')
code(' ret = false;')
code('else')
code(' this->${{param.name}}.push_back(object);')
code.dedent()
code('}')
code.dedent()
code('} else {')
code(' ret = false;')
code('}')
code()
code('return ret;')
code.dedent()
code('}')
code()
code('void ${member_prefix}setName(const std::string &name_)'
'${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('this->name = name_;')
code('this->pyobj = NULL;')
code.dedent()
code('}')
if is_header:
code('const std::string &${member_prefix}getName()')
code('{ return this->name; }')
code()
code('bool ${member_prefix}setParam(const std::string &name,')
code(' const std::string &value, const Flags flags)${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('bool ret = true;')
code()
code('if (false) {')
for param in simobj._params.values():
is_vector = isinstance(param, m5.params.VectorParamDesc)
is_simobj = issubclass(param.ptype, m5.SimObject.SimObject)
if not is_simobj and not is_vector:
code('} else if (name == "${{param.name}}") {')
code.indent()
param.ptype.cxx_ini_parse(code,
'value', 'this->%s' % param.name, 'ret =')
code.dedent()
code('} else {')
code(' ret = false;')
code('}')
code()
code('return ret;')
code.dedent()
code('}')
code()
code('bool ${member_prefix}setParamVector('
'const std::string &name,')
code(' const std::vector<std::string> &values,')
code(' const Flags flags)${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('bool ret = true;')
code()
code('if (false) {')
for param in simobj._params.values():
is_vector = isinstance(param, m5.params.VectorParamDesc)
is_simobj = issubclass(param.ptype, m5.SimObject.SimObject)
if not is_simobj and is_vector:
code('} else if (name == "${{param.name}}") {')
code.indent()
code('${{param.name}}.clear();')
code('for (auto i = values.begin(); '
'ret && i != values.end(); i ++)')
code('{')
code.indent()
code('${{param.ptype.cxx_type}} elem;')
param.ptype.cxx_ini_parse(code,
'*i', 'elem', 'ret =')
code('if (ret)')
code(' this->${{param.name}}.push_back(elem);')
code.dedent()
code('}')
code.dedent()
code('} else {')
code(' ret = false;')
code('}')
code()
code('return ret;')
code.dedent()
code('}')
code()
code('bool ${member_prefix}setPortConnectionCount('
'const std::string &name,')
code(' unsigned int count)${end_of_decl}')
if not is_header:
code('{')
code.indent()
code('bool ret = true;')
code()
code('if (false)')
code(' ;')
for port in simobj._ports.values():
code('else if (name == "${{port.name}}")')
code(' this->port_${{port.name}}_connection_count = count;')
code('else')
code(' ret = false;')
code()
code('return ret;')
code.dedent()
code('}')
code()
code('SimObject *${member_prefix}simObjectCreate()${end_of_decl}')
if not is_header:
code('{')
if hasattr(simobj, 'abstract') and simobj.abstract:
code(' return NULL;')
else:
code(' return this->create();')
code('}')
if is_header:
code()
code('static CxxConfigDirectoryEntry'
' *${member_prefix}makeDirectoryEntry()')
code('{ return new DirectoryEntry; }')
if is_header:
code.dedent()
code('};')
# The metaclass for SimObject. This class controls how new classes
# that derive from SimObject are instantiated, and provides inherited
# class behavior (just like a class controls how instances of that
# class are instantiated, and provides inherited instance behavior).
class MetaSimObject(type):
# Attributes that can be set only at initialization time
init_keywords = { 'abstract' : bool,
'cxx_class' : str,
'cxx_type' : str,
'cxx_header' : str,
'type' : str,
'cxx_bases' : list }
# Attributes that can be set any time
keywords = { 'check' : FunctionType }
# __new__ is called before __init__, and is where the statements
# in the body of the class definition get loaded into the class's
# __dict__. We intercept this to filter out parameter & port assignments
# and only allow "private" attributes to be passed to the base
# __new__ (starting with underscore).
def __new__(mcls, name, bases, dict):
assert name not in allClasses, "SimObject %s already present" % name
# Copy "private" attributes, functions, and classes to the
# official dict. Everything else goes in _init_dict to be
# filtered in __init__.
cls_dict = {}
value_dict = {}
for key,val in dict.items():
if public_value(key, val):
cls_dict[key] = val
else:
# must be a param/port setting
value_dict[key] = val
if 'abstract' not in value_dict:
value_dict['abstract'] = False
if 'cxx_bases' not in value_dict:
value_dict['cxx_bases'] = []
cls_dict['_value_dict'] = value_dict
cls = super(MetaSimObject, mcls).__new__(mcls, name, bases, cls_dict)
if 'type' in value_dict:
allClasses[name] = cls
return cls
# subclass initialization
def __init__(cls, name, bases, dict):
# calls type.__init__()... I think that's a no-op, but leave
# it here just in case it's not.
super(MetaSimObject, cls).__init__(name, bases, dict)
# initialize required attributes
# class-only attributes
cls._params = multidict() # param descriptions
cls._ports = multidict() # port descriptions
# class or instance attributes
cls._values = multidict() # param values
cls._hr_values = multidict() # human readable param values
cls._children = multidict() # SimObject children
cls._port_refs = multidict() # port ref objects
cls._instantiated = False # really instantiated, cloned, or subclassed
# We don't support multiple inheritance of sim objects. If you want
# to, you must fix multidict to deal with it properly. Non sim-objects
# are ok, though
bTotal = 0
for c in bases:
if isinstance(c, MetaSimObject):
bTotal += 1
if bTotal > 1:
raise TypeError, "SimObjects do not support multiple inheritance"
base = bases[0]
# Set up general inheritance via multidicts. A subclass will
# inherit all its settings from the base class. The only time
# the following is not true is when we define the SimObject
# class itself (in which case the multidicts have no parent).
if isinstance(base, MetaSimObject):
cls._base = base
cls._params.parent = base._params
cls._ports.parent = base._ports
cls._values.parent = base._values
cls._hr_values.parent = base._hr_values
cls._children.parent = base._children
cls._port_refs.parent = base._port_refs
# mark base as having been subclassed
base._instantiated = True
else:
cls._base = None
# default keyword values
if 'type' in cls._value_dict:
if 'cxx_class' not in cls._value_dict:
cls._value_dict['cxx_class'] = cls._value_dict['type']
cls._value_dict['cxx_type'] = '%s *' % cls._value_dict['cxx_class']
if 'cxx_header' not in cls._value_dict:
global noCxxHeader
noCxxHeader = True
warn("No header file specified for SimObject: %s", name)
# Export methods are automatically inherited via C++, so we
# don't want the method declarations to get inherited on the
# python side (and thus end up getting repeated in the wrapped
# versions of derived classes). The code below basicallly
# suppresses inheritance by substituting in the base (null)
# versions of these methods unless a different version is
# explicitly supplied.
for method_name in ('export_methods', 'export_method_cxx_predecls',
'export_method_swig_predecls'):
if method_name not in cls.__dict__:
base_method = getattr(MetaSimObject, method_name)
m = MethodType(base_method, cls, MetaSimObject)
setattr(cls, method_name, m)
# Now process the _value_dict items. They could be defining
# new (or overriding existing) parameters or ports, setting
# class keywords (e.g., 'abstract'), or setting parameter
# values or port bindings. The first 3 can only be set when
# the class is defined, so we handle them here. The others
# can be set later too, so just emulate that by calling
# setattr().
for key,val in cls._value_dict.items():
# param descriptions
if isinstance(val, ParamDesc):
cls._new_param(key, val)
# port objects
elif isinstance(val, Port):
cls._new_port(key, val)
# init-time-only keywords
elif cls.init_keywords.has_key(key):
cls._set_keyword(key, val, cls.init_keywords[key])
# default: use normal path (ends up in __setattr__)
else:
setattr(cls, key, val)
def _set_keyword(cls, keyword, val, kwtype):
if not isinstance(val, kwtype):
raise TypeError, 'keyword %s has bad type %s (expecting %s)' % \
(keyword, type(val), kwtype)
if isinstance(val, FunctionType):
val = classmethod(val)
type.__setattr__(cls, keyword, val)
def _new_param(cls, name, pdesc):
# each param desc should be uniquely assigned to one variable
assert(not hasattr(pdesc, 'name'))
pdesc.name = name
cls._params[name] = pdesc
if hasattr(pdesc, 'default'):
cls._set_param(name, pdesc.default, pdesc)
def _set_param(cls, name, value, param):
assert(param.name == name)
try:
hr_value = value
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
value = param.convert(value)
except Exception, e:
msg = "%s\nError setting param %s.%s to %s\n" % \
(e, cls.__name__, name, value)
e.args = (msg, )
raise
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
cls._values[name] = value
# if param value is a SimObject, make it a child too, so that
# it gets cloned properly when the class is instantiated
if isSimObjectOrVector(value) and not value.has_parent():
cls._add_cls_child(name, value)
# update human-readable values of the param if it has a literal
# value and is not an object or proxy.
if not (isSimObjectOrVector(value) or\
isinstance(value, m5.proxy.BaseProxy)):
cls._hr_values[name] = hr_value
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
def _add_cls_child(cls, name, child):
# It's a little funky to have a class as a parent, but these
# objects should never be instantiated (only cloned, which
# clears the parent pointer), and this makes it clear that the
# object is not an orphan and can provide better error
# messages.
child.set_parent(cls, name)
cls._children[name] = child
def _new_port(cls, name, port):
# each port should be uniquely assigned to one variable
assert(not hasattr(port, 'name'))
port.name = name
cls._ports[name] = port
# same as _get_port_ref, effectively, but for classes
def _cls_get_port_ref(cls, attr):
# Return reference that can be assigned to another port
# via __setattr__. There is only ever one reference
# object per port, but we create them lazily here.
ref = cls._port_refs.get(attr)
if not ref:
ref = cls._ports[attr].makeRef(cls)
cls._port_refs[attr] = ref
return ref
# Set attribute (called on foo.attr = value when foo is an
# instance of class cls).
def __setattr__(cls, attr, value):
# normal processing for private attributes
if public_value(attr, value):
type.__setattr__(cls, attr, value)
return
if cls.keywords.has_key(attr):
cls._set_keyword(attr, value, cls.keywords[attr])
return
if cls._ports.has_key(attr):
cls._cls_get_port_ref(attr).connect(value)
return
if isSimObjectOrSequence(value) and cls._instantiated:
raise RuntimeError, \
"cannot set SimObject parameter '%s' after\n" \
" class %s has been instantiated or subclassed" \
% (attr, cls.__name__)
# check for param
param = cls._params.get(attr)
if param:
cls._set_param(attr, value, param)
return
if isSimObjectOrSequence(value):
# If RHS is a SimObject, it's an implicit child assignment.
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
cls._add_cls_child(attr, coerceSimObjectOrVector(value))
return
# no valid assignment... raise exception
raise AttributeError, \
"Class %s has no parameter \'%s\'" % (cls.__name__, attr)
def __getattr__(cls, attr):
if attr == 'cxx_class_path':
return cls.cxx_class.split('::')
if attr == 'cxx_class_name':
return cls.cxx_class_path[-1]
if attr == 'cxx_namespaces':
return cls.cxx_class_path[:-1]
if cls._values.has_key(attr):
return cls._values[attr]
if cls._children.has_key(attr):
return cls._children[attr]
raise AttributeError, \
"object '%s' has no attribute '%s'" % (cls.__name__, attr)
def __str__(cls):
return cls.__name__
# See ParamValue.cxx_predecls for description.
def cxx_predecls(cls, code):
code('#include "params/$cls.hh"')
# See ParamValue.swig_predecls for description.
def swig_predecls(cls, code):
code('%import "python/m5/internal/param_$cls.i"')
# Hook for exporting additional C++ methods to Python via SWIG.
# Default is none, override using @classmethod in class definition.
def export_methods(cls, code):
pass
# Generate the code needed as a prerequisite for the C++ methods
# exported via export_methods() to be compiled in the _wrap.cc
# file. Typically generates one or more #include statements. If
# any methods are exported, typically at least the C++ header
# declaring the relevant SimObject class must be included.
def export_method_cxx_predecls(cls, code):
pass
# Generate the code needed as a prerequisite for the C++ methods
# exported via export_methods() to be processed by SWIG.
# Typically generates one or more %include or %import statements.
# If any methods are exported, typically at least the C++ header
# declaring the relevant SimObject class must be included.
def export_method_swig_predecls(cls, code):
pass
# Generate the declaration for this object for wrapping with SWIG.
# Generates code that goes into a SWIG .i file. Called from
# src/SConscript.
def swig_decl(cls, code):
class_path = cls.cxx_class.split('::')
classname = class_path[-1]
namespaces = class_path[:-1]
# The 'local' attribute restricts us to the params declared in
# the object itself, not including inherited params (which
# will also be inherited from the base class's param struct
# here). Sort the params based on their key
params = map(lambda (k, v): v, sorted(cls._params.local.items()))
ports = cls._ports.local
code('%module(package="m5.internal") param_$cls')
code()
code('%{')
code('#include "sim/sim_object.hh"')
code('#include "params/$cls.hh"')
for param in params:
param.cxx_predecls(code)
code('#include "${{cls.cxx_header}}"')
cls.export_method_cxx_predecls(code)
code('''\
/**
* This is a workaround for bug in swig. Prior to gcc 4.6.1 the STL
* headers like vector, string, etc. used to automatically pull in
* the cstddef header but starting with gcc 4.6.1 they no longer do.
* This leads to swig generated a file that does not compile so we
* explicitly include cstddef. Additionally, including version 2.0.4,
* swig uses ptrdiff_t without the std:: namespace prefix which is
* required with gcc 4.6.1. We explicitly provide access to it.
*/
#include <cstddef>
using std::ptrdiff_t;
''')
code('%}')
code()
for param in params:
param.swig_predecls(code)
cls.export_method_swig_predecls(code)
code()
if cls._base:
code('%import "python/m5/internal/param_${{cls._base}}.i"')
code()
for ns in namespaces:
code('namespace $ns {')
if namespaces:
code('// avoid name conflicts')
sep_string = '_COLONS_'
flat_name = sep_string.join(class_path)
code('%rename($flat_name) $classname;')
code()
code('// stop swig from creating/wrapping default ctor/dtor')
code('%nodefault $classname;')
code('class $classname')
if cls._base:
bases = [ cls._base.cxx_class ] + cls.cxx_bases
else:
bases = cls.cxx_bases
base_first = True
for base in bases:
if base_first:
code(' : public ${{base}}')
base_first = False
else:
code(' , public ${{base}}')
code('{')
code(' public:')
cls.export_methods(code)
code('};')
for ns in reversed(namespaces):
code('} // namespace $ns')
code()
code('%include "params/$cls.hh"')
# Generate the C++ declaration (.hh file) for this SimObject's
# param struct. Called from src/SConscript.
def cxx_param_decl(cls, code):
# The 'local' attribute restricts us to the params declared in
# the object itself, not including inherited params (which
# will also be inherited from the base class's param struct
# here). Sort the params based on their key
params = map(lambda (k, v): v, sorted(cls._params.local.items()))
ports = cls._ports.local
try:
ptypes = [p.ptype for p in params]
except:
print cls, p, p.ptype_str
print params
raise
class_path = cls._value_dict['cxx_class'].split('::')
code('''\
#ifndef __PARAMS__${cls}__
#define __PARAMS__${cls}__
''')
# A forward class declaration is sufficient since we are just
# declaring a pointer.
for ns in class_path[:-1]:
code('namespace $ns {')
code('class $0;', class_path[-1])
for ns in reversed(class_path[:-1]):
code('} // namespace $ns')
code()
# The base SimObject has a couple of params that get
# automatically set from Python without being declared through
# the normal Param mechanism; we slip them in here (needed
# predecls now, actual declarations below)
if cls == SimObject:
code('''
#ifndef PY_VERSION
struct PyObject;
#endif
#include <string>
''')
for param in params:
param.cxx_predecls(code)
for port in ports.itervalues():
port.cxx_predecls(code)
code()
if cls._base:
code('#include "params/${{cls._base.type}}.hh"')
code()
for ptype in ptypes:
if issubclass(ptype, Enum):
code('#include "enums/${{ptype.__name__}}.hh"')
code()
# now generate the actual param struct
code("struct ${cls}Params")
if cls._base:
code(" : public ${{cls._base.type}}Params")
code("{")
if not hasattr(cls, 'abstract') or not cls.abstract:
if 'type' in cls.__dict__:
code(" ${{cls.cxx_type}} create();")
code.indent()
2008-06-18 21:07:15 +02:00
if cls == SimObject:
code('''
SimObjectParams() {}
virtual ~SimObjectParams() {}
std::string name;
PyObject *pyobj;
''')
for param in params:
param.cxx_decl(code)
for port in ports.itervalues():
port.cxx_decl(code)
code.dedent()
code('};')
code()
code('#endif // __PARAMS__${cls}__')
return code
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
# Generate the C++ declaration/definition files for this SimObject's
# param struct to allow C++ initialisation
def cxx_config_param_file(cls, code, is_header):
createCxxConfigDirectoryEntryFile(code, cls.__name__, cls, is_header)
return code
# This *temporary* definition is required to support calls from the
# SimObject class definition to the MetaSimObject methods (in
# particular _set_param, which gets called for parameters with default
# values defined on the SimObject class itself). It will get
# overridden by the permanent definition (which requires that
# SimObject be defined) lower in this file.
def isSimObjectOrVector(value):
return False
# This class holds information about each simobject parameter
# that should be displayed on the command line for use in the
# configuration system.
class ParamInfo(object):
def __init__(self, type, desc, type_str, example, default_val, access_str):
self.type = type
self.desc = desc
self.type_str = type_str
self.example_str = example
self.default_val = default_val
# The string representation used to access this param through python.
# The method to access this parameter presented on the command line may
# be different, so this needs to be stored for later use.
self.access_str = access_str
self.created = True
# Make it so we can only set attributes at initialization time
# and effectively make this a const object.
def __setattr__(self, name, value):
if not "created" in self.__dict__:
self.__dict__[name] = value
# The SimObject class is the root of the special hierarchy. Most of
# the code in this class deals with the configuration hierarchy itself
# (parent/child node relationships).
class SimObject(object):
# Specify metaclass. Any class inheriting from SimObject will
# get this metaclass.
__metaclass__ = MetaSimObject
type = 'SimObject'
abstract = True
cxx_header = "sim/sim_object.hh"
cxx_bases = [ "Drainable", "Serializable" ]
eventq_index = Param.UInt32(Parent.eventq_index, "Event Queue Index")
@classmethod
def export_method_swig_predecls(cls, code):
code('''
%include <std_string.i>
%import "python/swig/drain.i"
%import "python/swig/serialize.i"
''')
@classmethod
def export_methods(cls, code):
code('''
void init();
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
void loadState(CheckpointIn &cp);
void initState();
void memInvalidate();
void memWriteback();
void regStats();
void resetStats();
base: add support for probe points and common probes The probe patch is motivated by the desire to move analytical and trace code away from functional code. This is achieved by the probe interface which is essentially a glorified observer model. What this means to users: * add a probe point and a "notify" call at the source of an "event" * add an isolated module, that is being used to carry out *your* analysis (e.g. generate a trace) * register that module as a probe listener Note: an example is given for reference in src/cpu/o3/simple_trace.[hh|cc] and src/cpu/SimpleTrace.py What is happening under the hood: * every SimObject maintains has a ProbeManager. * during initialization (src/python/m5/simulate.py) first regProbePoints and the regProbeListeners is called on each SimObject. this hooks up the probe point notify calls with the listeners. FAQs: Why did you develop probe points: * to remove trace, stats gathering, analytical code out of the functional code. * the belief that probes could be generically useful. What is a probe point: * a probe point is used to notify upon a given event (e.g. cpu commits an instruction) What is a probe listener: * a class that handles whatever the user wishes to do when they are notified about an event. What can be passed on notify: * probe points are templates, and so the user can generate probes that pass any type of argument (by const reference) to a listener. What relationships can be generated (1:1, 1:N, N:M etc): * there isn't a restriction. You can hook probe points and listeners up in a 1:1, 1:N, N:M relationship. They become useful when a number of modules listen to the same probe points. The idea being that you can add a small number of probes into the source code and develop a larger number of useful analysis modules that use information passed by the probes. Can you give examples: * adding a probe point to the cpu's commit method allows you to build a trace module (outputting assembler), you could re-use this to gather instruction distribution (arithmetic, load/store, conditional, control flow) stats. Why is the probe interface currently restricted to passing a const reference: * the desire, initially at least, is to allow an interface to observe functionality, but not to change functionality. * of course this can be subverted by const-casting. What is the performance impact of adding probes: * when nothing is actively listening to the probes they should have a relatively minor impact. Profiling has suggested even with a large number of probes (60) the impact of them (when not active) is very minimal (<1%).
2014-01-24 22:29:30 +01:00
void regProbePoints();
void regProbeListeners();
void startup();
''')
# Returns a dict of all the option strings that can be
# generated as command line options for this simobject instance
# by tracing all reachable params in the top level instance and
# any children it contains.
def enumerateParams(self, flags_dict = {},
cmd_line_str = "", access_str = ""):
if hasattr(self, "_paramEnumed"):
print "Cycle detected enumerating params"
else:
self._paramEnumed = True
# Scan the children first to pick up all the objects in this SimObj
for keys in self._children:
child = self._children[keys]
next_cmdline_str = cmd_line_str + keys
next_access_str = access_str + keys
if not isSimObjectVector(child):
next_cmdline_str = next_cmdline_str + "."
next_access_str = next_access_str + "."
flags_dict = child.enumerateParams(flags_dict,
next_cmdline_str,
next_access_str)
# Go through the simple params in the simobject in this level
# of the simobject hierarchy and save information about the
# parameter to be used for generating and processing command line
# options to the simulator to set these parameters.
for keys,values in self._params.items():
if values.isCmdLineSettable():
type_str = ''
ex_str = values.example_str()
ptype = None
if isinstance(values, VectorParamDesc):
type_str = 'Vector_%s' % values.ptype_str
ptype = values
else:
type_str = '%s' % values.ptype_str
ptype = values.ptype
if keys in self._hr_values\
and keys in self._values\
and not isinstance(self._values[keys], m5.proxy.BaseProxy):
cmd_str = cmd_line_str + keys
acc_str = access_str + keys
flags_dict[cmd_str] = ParamInfo(ptype,
self._params[keys].desc, type_str, ex_str,
values.pretty_print(self._hr_values[keys]),
acc_str)
elif not keys in self._hr_values\
and not keys in self._values:
# Empty param
cmd_str = cmd_line_str + keys
acc_str = access_str + keys
flags_dict[cmd_str] = ParamInfo(ptype,
self._params[keys].desc,
type_str, ex_str, '', acc_str)
return flags_dict
# Initialize new instance. For objects with SimObject-valued
# children, we need to recursively clone the classes represented
# by those param values as well in a consistent "deep copy"-style
# fashion. That is, we want to make sure that each instance is
# cloned only once, and that if there are multiple references to
# the same original object, we end up with the corresponding
# cloned references all pointing to the same cloned instance.
def __init__(self, **kwargs):
ancestor = kwargs.get('_ancestor')
memo_dict = kwargs.get('_memo')
if memo_dict is None:
# prepare to memoize any recursively instantiated objects
memo_dict = {}
elif ancestor:
# memoize me now to avoid problems with recursive calls
memo_dict[ancestor] = self
if not ancestor:
ancestor = self.__class__
ancestor._instantiated = True
# initialize required attributes
self._parent = None
self._name = None
self._ccObject = None # pointer to C++ object
self._ccParams = None
self._instantiated = False # really "cloned"
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
# Clone children specified at class level. No need for a
# multidict here since we will be cloning everything.
# Do children before parameter values so that children that
# are also param values get cloned properly.
self._children = {}
for key,val in ancestor._children.iteritems():
self.add_child(key, val(_memo=memo_dict))
# Inherit parameter values from class using multidict so
# individual value settings can be overridden but we still
# inherit late changes to non-overridden class values.
self._values = multidict(ancestor._values)
self._hr_values = multidict(ancestor._hr_values)
# clone SimObject-valued parameters
for key,val in ancestor._values.iteritems():
val = tryAsSimObjectOrVector(val)
if val is not None:
self._values[key] = val(_memo=memo_dict)
# clone port references. no need to use a multidict here
# since we will be creating new references for all ports.
self._port_refs = {}
for key,val in ancestor._port_refs.iteritems():
self._port_refs[key] = val.clone(self, memo_dict)
# apply attribute assignments from keyword args, if any
for key,val in kwargs.iteritems():
setattr(self, key, val)
# "Clone" the current instance by creating another instance of
# this instance's class, but that inherits its parameter values
# and port mappings from the current instance. If we're in a
# "deep copy" recursive clone, check the _memo dict to see if
# we've already cloned this instance.
def __call__(self, **kwargs):
memo_dict = kwargs.get('_memo')
if memo_dict is None:
# no memo_dict: must be top-level clone operation.
# this is only allowed at the root of a hierarchy
if self._parent:
raise RuntimeError, "attempt to clone object %s " \
"not at the root of a tree (parent = %s)" \
% (self, self._parent)
# create a new dict and use that.
memo_dict = {}
kwargs['_memo'] = memo_dict
elif memo_dict.has_key(self):
# clone already done & memoized
return memo_dict[self]
return self.__class__(_ancestor = self, **kwargs)
def _get_port_ref(self, attr):
# Return reference that can be assigned to another port
# via __setattr__. There is only ever one reference
# object per port, but we create them lazily here.
ref = self._port_refs.get(attr)
if ref == None:
ref = self._ports[attr].makeRef(self)
self._port_refs[attr] = ref
return ref
def __getattr__(self, attr):
if self._ports.has_key(attr):
return self._get_port_ref(attr)
if self._values.has_key(attr):
return self._values[attr]
if self._children.has_key(attr):
return self._children[attr]
# If the attribute exists on the C++ object, transparently
# forward the reference there. This is typically used for
# SWIG-wrapped methods such as init(), regStats(),
# resetStats(), startup(), drain(), and
# resume().
if self._ccObject and hasattr(self._ccObject, attr):
return getattr(self._ccObject, attr)
err_string = "object '%s' has no attribute '%s'" \
% (self.__class__.__name__, attr)
if not self._ccObject:
err_string += "\n (C++ object is not yet constructed," \
" so wrapped C++ methods are unavailable.)"
raise AttributeError, err_string
# Set attribute (called on foo.attr = value when foo is an
# instance of class cls).
def __setattr__(self, attr, value):
# normal processing for private attributes
if attr.startswith('_'):
object.__setattr__(self, attr, value)
return
if self._ports.has_key(attr):
# set up port connection
self._get_port_ref(attr).connect(value)
return
param = self._params.get(attr)
if param:
try:
hr_value = value
value = param.convert(value)
except Exception, e:
msg = "%s\nError setting param %s.%s to %s\n" % \
(e, self.__class__.__name__, attr, value)
e.args = (msg, )
raise
self._values[attr] = value
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
# implicitly parent unparented objects assigned as params
if isSimObjectOrVector(value) and not value.has_parent():
self.add_child(attr, value)
# set the human-readable value dict if this is a param
# with a literal value and is not being set as an object
# or proxy.
if not (isSimObjectOrVector(value) or\
isinstance(value, m5.proxy.BaseProxy)):
self._hr_values[attr] = hr_value
return
# if RHS is a SimObject, it's an implicit child assignment
if isSimObjectOrSequence(value):
self.add_child(attr, value)
return
# no valid assignment... raise exception
raise AttributeError, "Class %s has no parameter %s" \
% (self.__class__.__name__, attr)
# this hack allows tacking a '[0]' onto parameters that may or may
# not be vectors, and always getting the first element (e.g. cpus)
def __getitem__(self, key):
if key == 0:
return self
raise IndexError, "Non-zero index '%s' to SimObject" % key
# this hack allows us to iterate over a SimObject that may
# not be a vector, so we can call a loop over it and get just one
# element.
def __len__(self):
return 1
# Also implemented by SimObjectVector
def clear_parent(self, old_parent):
assert self._parent is old_parent
self._parent = None
# Also implemented by SimObjectVector
def set_parent(self, parent, name):
self._parent = parent
self._name = name
# Return parent object of this SimObject, not implemented by SimObjectVector
# because the elements in a SimObjectVector may not share the same parent
def get_parent(self):
return self._parent
# Also implemented by SimObjectVector
def get_name(self):
return self._name
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
# Also implemented by SimObjectVector
def has_parent(self):
return self._parent is not None
# clear out child with given name. This code is not likely to be exercised.
# See comment in add_child.
def clear_child(self, name):
child = self._children[name]
child.clear_parent(self)
del self._children[name]
# Add a new child to this object.
def add_child(self, name, child):
child = coerceSimObjectOrVector(child)
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
if child.has_parent():
warn("add_child('%s'): child '%s' already has parent", name,
child.get_name())
if self._children.has_key(name):
# This code path had an undiscovered bug that would make it fail
# at runtime. It had been here for a long time and was only
# exposed by a buggy script. Changes here will probably not be
# exercised without specialized testing.
self.clear_child(name)
child.set_parent(self, name)
self._children[name] = child
# Take SimObject-valued parameters that haven't been explicitly
# assigned as children and make them children of the object that
# they were assigned to as a parameter value. This guarantees
# that when we instantiate all the parameter objects we're still
# inside the configuration hierarchy.
def adoptOrphanParams(self):
for key,val in self._values.iteritems():
if not isSimObjectVector(val) and isSimObjectSequence(val):
# need to convert raw SimObject sequences to
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
# SimObjectVector class so we can call has_parent()
val = SimObjectVector(val)
self._values[key] = val
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
if isSimObjectOrVector(val) and not val.has_parent():
warn("%s adopting orphan SimObject param '%s'", self, key)
self.add_child(key, val)
def path(self):
if not self._parent:
config: reinstate implicit parenting on parameter assignment Last summer's big rewrite of the initialization code (in particular cset 6efc3672733b) got rid of the implicit parenting that used to occur when an unparented SimObject was assigned as a parameter value to another SimObject. The idea was that the new adoptOrphanParams() step would catch these anyway so it was unnecessary. Unfortunately it turns out that adoptOrphanParams() has some inherent instability in that the parent that does the adoption depends on the config tree traversal order. Even making this order deterministic (e.g., by traversing children in alphabetical order) can introduce unwanted and unexpected hierarchy changes between similar configs (e.g., when adding a switch_cpu in place of a cpu), causing problems when trying to restore checkpoints across similar configs. The hierarchy created by implicit parenting is more stable and more controllable, so this patch turns that behavior back on. This patch also cleans up some long-standing holes regarding parenting of SimObjects that are created in class definitions (either in the body of the class, or as default parameters). To avoid breaking some existing config files, this necessitated changing the error on reparenting children to a warning. This change fixes another bug where attempting to print the prior error message would fail on reparenting SimObjectVectors because they lack a _parent attribute. Some further issues with SimObjectVectors were cleaned up by getting rid of the get_parent() call (which could cause errors with some SimObjectVectors where there was no single parent to return) with has_parent() (since all the uses of get_parent() were just boolean tests anyway). Finally, since the adoptOrphanParam() step turned out to be so problematic, we now issue a warning when it actually has to do an adoption. Future cleanup of config files will get rid of current warnings.
2011-05-23 23:29:08 +02:00
return '<orphan %s>' % self.__class__
ppath = self._parent.path()
if ppath == 'root':
return self._name
return ppath + "." + self._name
def __str__(self):
return self.path()
def config_value(self):
return self.path()
def ini_str(self):
return self.path()
def find_any(self, ptype):
if isinstance(self, ptype):
return self, True
found_obj = None
for child in self._children.itervalues():
visited = False
if hasattr(child, '_visited'):
visited = getattr(child, '_visited')
if isinstance(child, ptype) and not visited:
if found_obj != None and child != found_obj:
raise AttributeError, \
'parent.any matched more than one: %s %s' % \
(found_obj.path, child.path)
found_obj = child
# search param space
for pname,pdesc in self._params.iteritems():
if issubclass(pdesc.ptype, ptype):
match_obj = self._values[pname]
if found_obj != None and found_obj != match_obj:
raise AttributeError, \
'parent.any matched more than one: %s and %s' % (found_obj.path, match_obj.path)
found_obj = match_obj
return found_obj, found_obj != None
def find_all(self, ptype):
all = {}
# search children
for child in self._children.itervalues():
# a child could be a list, so ensure we visit each item
if isinstance(child, list):
children = child
else:
children = [child]
for child in children:
if isinstance(child, ptype) and not isproxy(child) and \
not isNullPointer(child):
all[child] = True
if isSimObject(child):
# also add results from the child itself
child_all, done = child.find_all(ptype)
all.update(dict(zip(child_all, [done] * len(child_all))))
# search param space
for pname,pdesc in self._params.iteritems():
if issubclass(pdesc.ptype, ptype):
match_obj = self._values[pname]
if not isproxy(match_obj) and not isNullPointer(match_obj):
all[match_obj] = True
# Also make sure to sort the keys based on the objects' path to
# ensure that the order is the same on all hosts
return sorted(all.keys(), key = lambda o: o.path()), True
def unproxy(self, base):
return self
def unproxyParams(self):
for param in self._params.iterkeys():
value = self._values.get(param)
if value != None and isproxy(value):
try:
value = value.unproxy(self)
except:
print "Error in unproxying param '%s' of %s" % \
(param, self.path())
raise
setattr(self, param, value)
# Unproxy ports in sorted order so that 'append' operations on
# vector ports are done in a deterministic fashion.
port_names = self._ports.keys()
port_names.sort()
for port_name in port_names:
port = self._port_refs.get(port_name)
if port != None:
port.unproxy(self)
def print_ini(self, ini_file):
print >>ini_file, '[' + self.path() + ']' # .ini section header
instanceDict[self.path()] = self
if hasattr(self, 'type'):
print >>ini_file, 'type=%s' % self.type
if len(self._children.keys()):
print >>ini_file, 'children=%s' % \
' '.join(self._children[n].get_name() \
for n in sorted(self._children.keys()))
for param in sorted(self._params.keys()):
value = self._values.get(param)
if value != None:
print >>ini_file, '%s=%s' % (param,
self._values[param].ini_str())
for port_name in sorted(self._ports.keys()):
port = self._port_refs.get(port_name, None)
if port != None:
print >>ini_file, '%s=%s' % (port_name, port.ini_str())
print >>ini_file # blank line between objects
# generate a tree of dictionaries expressing all the parameters in the
# instantiated system for use by scripts that want to do power, thermal
# visualization, and other similar tasks
def get_config_as_dict(self):
d = attrdict()
if hasattr(self, 'type'):
d.type = self.type
if hasattr(self, 'cxx_class'):
d.cxx_class = self.cxx_class
# Add the name and path of this object to be able to link to
# the stats
d.name = self.get_name()
d.path = self.path()
for param in sorted(self._params.keys()):
value = self._values.get(param)
if value != None:
d[param] = value.config_value()
for n in sorted(self._children.keys()):
child = self._children[n]
# Use the name of the attribute (and not get_name()) as
# the key in the JSON dictionary to capture the hierarchy
# in the Python code that assembled this system
d[n] = child.get_config_as_dict()
for port_name in sorted(self._ports.keys()):
port = self._port_refs.get(port_name, None)
if port != None:
# Represent each port with a dictionary containing the
# prominent attributes
d[port_name] = port.get_config_as_dict()
return d
def getCCParams(self):
if self._ccParams:
return self._ccParams
cc_params_struct = getattr(m5.internal.params, '%sParams' % self.type)
cc_params = cc_params_struct()
2008-06-18 21:07:15 +02:00
cc_params.pyobj = self
cc_params.name = str(self)
param_names = self._params.keys()
param_names.sort()
for param in param_names:
value = self._values.get(param)
if value is None:
fatal("%s.%s without default or user set value",
self.path(), param)
value = value.getValue()
if isinstance(self._params[param], VectorParamDesc):
assert isinstance(value, list)
vec = getattr(cc_params, param)
assert not len(vec)
for v in value:
vec.append(v)
else:
setattr(cc_params, param, value)
port_names = self._ports.keys()
port_names.sort()
for port_name in port_names:
port = self._port_refs.get(port_name, None)
if port != None:
port_count = len(port)
else:
port_count = 0
setattr(cc_params, 'port_' + port_name + '_connection_count',
port_count)
self._ccParams = cc_params
return self._ccParams
# Get C++ object corresponding to this object, calling C++ if
# necessary to construct it. Does *not* recursively create
# children.
def getCCObject(self):
if not self._ccObject:
# Make sure this object is in the configuration hierarchy
if not self._parent and not isRoot(self):
raise RuntimeError, "Attempt to instantiate orphan node"
# Cycles in the configuration hierarchy are not supported. This
# will catch the resulting recursion and stop.
self._ccObject = -1
if not self.abstract:
params = self.getCCParams()
self._ccObject = params.create()
elif self._ccObject == -1:
raise RuntimeError, "%s: Cycle found in configuration hierarchy." \
% self.path()
return self._ccObject
def descendants(self):
yield self
# The order of the dict is implementation dependent, so sort
# it based on the key (name) to ensure the order is the same
# on all hosts
for (name, child) in sorted(self._children.iteritems()):
for obj in child.descendants():
yield obj
# Call C++ to create C++ object corresponding to this object
def createCCObject(self):
self.getCCParams()
self.getCCObject() # force creation
def getValue(self):
return self.getCCObject()
# Create C++ port connections corresponding to the connections in
# _port_refs
def connectPorts(self):
# Sort the ports based on their attribute name to ensure the
# order is the same on all hosts
for (attr, portRef) in sorted(self._port_refs.iteritems()):
portRef.ccConnect()
# Function to provide to C++ so it can look up instances based on paths
def resolveSimObject(name):
obj = instanceDict[name]
return obj.getCCObject()
def isSimObject(value):
return isinstance(value, SimObject)
def isSimObjectClass(value):
return issubclass(value, SimObject)
def isSimObjectVector(value):
return isinstance(value, SimObjectVector)
def isSimObjectSequence(value):
if not isinstance(value, (list, tuple)) or len(value) == 0:
return False
for val in value:
if not isNullPointer(val) and not isSimObject(val):
return False
return True
def isSimObjectOrSequence(value):
return isSimObject(value) or isSimObjectSequence(value)
def isRoot(obj):
from m5.objects import Root
return obj and obj is Root.getInstance()
def isSimObjectOrVector(value):
return isSimObject(value) or isSimObjectVector(value)
def tryAsSimObjectOrVector(value):
if isSimObjectOrVector(value):
return value
if isSimObjectSequence(value):
return SimObjectVector(value)
return None
def coerceSimObjectOrVector(value):
value = tryAsSimObjectOrVector(value)
if value is None:
raise TypeError, "SimObject or SimObjectVector expected"
return value
baseClasses = allClasses.copy()
baseInstances = instanceDict.copy()
def clear():
global allClasses, instanceDict, noCxxHeader
allClasses = baseClasses.copy()
instanceDict = baseInstances.copy()
noCxxHeader = False
# __all__ defines the list of symbols that get exported when
# 'from config import *' is invoked. Try to keep this reasonably
# short to avoid polluting other namespaces.
__all__ = [ 'SimObject' ]