esp-idf/tools/ldgen/generation.py
Renz Bagaporo ee0f8adaed ldgen: determinism in mapping rule order
This MR imposes some determinism in the mapping rule order in the output
file. For each section, the archives are arranged alphabetically
(ascending), and the mapping rules in each archive are arranged by
increasing specificity then alphabetically (ascending). The default
rules remain the very first rule for each section.
2020-05-08 20:48:43 +08:00

640 lines
26 KiB
Python

#
# Copyright 2018-2019 Espressif Systems (Shanghai) PTE LTD
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
import collections
import itertools
import os
import fnmatch
from fragments import Sections, Scheme, Mapping, Fragment
from pyparsing import Suppress, White, ParseException, Literal, Group, ZeroOrMore
from pyparsing import Word, OneOrMore, nums, alphanums, alphas, Optional, LineEnd, printables
from ldgen_common import LdGenFailure
class PlacementRule():
"""
Encapsulates a generated placement rule placed under a target
"""
DEFAULT_SPECIFICITY = 0
ARCHIVE_SPECIFICITY = 1
OBJECT_SPECIFICITY = 2
SYMBOL_SPECIFICITY = 3
class __container():
def __init__(self, content):
self.content = content
__metadata = collections.namedtuple("__metadata", "excludes expansions expanded")
def __init__(self, archive, obj, symbol, sections, target):
if archive == "*":
archive = None
if obj == "*":
obj = None
self.archive = archive
self.obj = obj
self.symbol = symbol
self.target = target
self.sections = dict()
self.specificity = 0
self.specificity += 1 if self.archive else 0
self.specificity += 1 if (self.obj and not self.obj == '*') else 0
self.specificity += 1 if self.symbol else 0
for section in sections:
section_data = Sections.get_section_data_from_entry(section, self.symbol)
if not self.symbol:
for s in section_data:
metadata = self.__metadata(self.__container([]), self.__container([]), self.__container(False))
self.sections[s] = metadata
else:
(section, expansion) = section_data
if expansion:
metadata = self.__metadata(self.__container([]), self.__container([expansion]), self.__container(True))
self.sections[section] = metadata
def get_section_names(self):
return self.sections.keys()
def add_exclusion(self, other, sections_infos=None):
# Utility functions for this method
def do_section_expansion(rule, section):
if section in rule.get_section_names():
sections_in_obj = sections_infos.get_obj_sections(rule.archive, rule.obj)
expansions = fnmatch.filter(sections_in_obj, section)
return expansions
def remove_section_expansions(rule, section, expansions):
existing_expansions = self.sections[section].expansions.content
self.sections[section].expansions.content = [e for e in existing_expansions if e not in expansions]
# Exit immediately if the exclusion to be added is more general than this rule.
if not other.is_more_specific_rule_of(self):
return
for section in self.get_sections_intersection(other):
if(other.specificity == PlacementRule.SYMBOL_SPECIFICITY):
# If this sections has not been expanded previously, expand now and keep track.
previously_expanded = self.sections[section].expanded.content
if not previously_expanded:
expansions = do_section_expansion(self, section)
if expansions:
self.sections[section].expansions.content = expansions
self.sections[section].expanded.content = True
previously_expanded = True
# Remove the sections corresponding to the symbol name
remove_section_expansions(self, section, other.sections[section].expansions.content)
# If it has been expanded previously but now the expansions list is empty,
# it means adding exclusions has exhausted the list. Remove the section entirely.
if previously_expanded and not self.sections[section].expanded.content:
del self.sections[section]
else:
# A rule section can have multiple rule sections excluded from it. Get the
# most specific rule from the list, and if an even more specific rule is found,
# replace it entirely. Otherwise, keep appending.
exclusions = self.sections[section].excludes
exclusions_list = exclusions.content if exclusions.content is not None else []
exclusions_to_remove = filter(lambda r: r.is_more_specific_rule_of(other), exclusions_list)
remaining_exclusions = [e for e in exclusions_list if e not in exclusions_to_remove]
remaining_exclusions.append(other)
self.sections[section].excludes.content = remaining_exclusions
def get_sections_intersection(self, other):
return set(self.sections.keys()).intersection(set(other.sections.keys()))
def is_more_specific_rule_of(self, other):
if (self.specificity <= other.specificity):
return False
# Compare archive, obj and target
for entity_index in range(1, other.specificity + 1):
if self[entity_index] != other[entity_index] and other[entity_index] is not None:
return False
return True
def maps_same_entities_as(self, other):
if self.specificity != other.specificity:
return False
# Compare archive, obj and target
for entity_index in range(1, other.specificity + 1):
if self[entity_index] != other[entity_index] and other[entity_index] is not None:
return False
return True
def __getitem__(self, key):
if key == PlacementRule.ARCHIVE_SPECIFICITY:
return self.archive
elif key == PlacementRule.OBJECT_SPECIFICITY:
return self.obj
elif key == PlacementRule.SYMBOL_SPECIFICITY:
return self.symbol
else:
return None
def __str__(self):
sorted_sections = sorted(self.get_section_names())
sections_string = list()
for section in sorted_sections:
exclusions = self.sections[section].excludes.content
exclusion_string = None
if exclusions:
exclusion_string = " ".join(map(lambda e: "*" + e.archive + (":" + e.obj + ".*" if e.obj else ""), exclusions))
exclusion_string = "EXCLUDE_FILE(" + exclusion_string + ")"
else:
exclusion_string = ""
section_string = None
exclusion_section_string = None
section_expansions = self.sections[section].expansions.content
section_expanded = self.sections[section].expanded.content
if section_expansions and section_expanded:
section_string = " ".join(section_expansions)
exclusion_section_string = section_string
else:
section_string = section
exclusion_section_string = exclusion_string + " " + section_string
sections_string.append(exclusion_section_string)
sections_string = " ".join(sections_string)
archive = str(self.archive) if self.archive else ""
obj = (str(self.obj) + (".*" if self.obj else "")) if self.obj else ""
# Handle output string generation based on information available
if self.specificity == PlacementRule.DEFAULT_SPECIFICITY:
rule_string = "*(%s)" % (sections_string)
elif self.specificity == PlacementRule.ARCHIVE_SPECIFICITY:
rule_string = "*%s:(%s)" % (archive, sections_string)
else:
rule_string = "*%s:%s(%s)" % (archive, obj, sections_string)
return rule_string
def __eq__(self, other):
if id(self) == id(other):
return True
def exclusions_set(exclusions):
exclusions_set = {(e.archive, e.obj, e.symbol, e.target) for e in exclusions}
return exclusions_set
if self.archive != other.archive:
return False
if self.obj != other.obj:
return False
if self.symbol != other.symbol:
return False
if set(self.sections.keys()) != set(other.sections.keys()):
return False
for (section, metadata) in self.sections.items():
self_meta = metadata
other_meta = other.sections[section]
if exclusions_set(self_meta.excludes.content) != exclusions_set(other_meta.excludes.content):
return False
if set(self_meta.expansions.content) != set(other_meta.expansions.content):
return False
return True
def __ne__(self, other):
return not self.__eq__(other)
def __iter__(self):
yield self.archive
yield self.obj
yield self.symbol
raise StopIteration
class GenerationModel:
"""
Implements generation of placement rules based on collected sections, scheme and mapping fragment.
"""
DEFAULT_SCHEME = "default"
def __init__(self):
self.schemes = {}
self.sections = {}
self.mappings = {}
def _add_mapping_rules(self, archive, obj, symbol, scheme_name, scheme_dict, rules):
# Use an ordinary dictionary to raise exception on non-existing keys
temp_dict = dict(scheme_dict)
sections_bucket = temp_dict[scheme_name]
for (target, sections) in sections_bucket.items():
section_entries = []
for section in sections:
section_entries.extend(section.entries)
rule = PlacementRule(archive, obj, symbol, section_entries, target)
if rule not in rules:
rules.append(rule)
def _build_scheme_dictionary(self):
scheme_dictionary = collections.defaultdict(dict)
# Collect sections into buckets based on target name
for scheme in self.schemes.values():
sections_bucket = collections.defaultdict(list)
for (sections_name, target_name) in scheme.entries:
# Get the sections under the bucket 'target_name'. If this bucket does not exist
# is is created automatically
sections_in_bucket = sections_bucket[target_name]
try:
sections = self.sections[sections_name]
except KeyError:
message = GenerationException.UNDEFINED_REFERENCE + " to sections '" + sections + "'."
raise GenerationException(message, scheme)
sections_in_bucket.append(sections)
scheme_dictionary[scheme.name] = sections_bucket
# Search for and raise exception on first instance of sections mapped to multiple targets
for (scheme_name, sections_bucket) in scheme_dictionary.items():
for sections_a, sections_b in itertools.combinations(sections_bucket.values(), 2):
set_a = set()
set_b = set()
for sections in sections_a:
set_a.update(sections.entries)
for sections in sections_b:
set_b.update(sections.entries)
intersection = set_a.intersection(set_b)
# If the intersection is a non-empty set, it means sections are mapped to multiple
# targets. Raise exception.
if intersection:
scheme = self.schemes[scheme_name]
message = "Sections " + str(intersection) + " mapped to multiple targets."
raise GenerationException(message, scheme)
return scheme_dictionary
def generate_rules(self, sections_infos):
scheme_dictionary = self._build_scheme_dictionary()
# Generate default rules
default_rules = list()
self._add_mapping_rules(None, None, None, GenerationModel.DEFAULT_SCHEME, scheme_dictionary, default_rules)
all_mapping_rules = collections.defaultdict(list)
# Generate rules based on mapping fragments
for mapping in self.mappings.values():
archive = mapping.archive
mapping_rules = all_mapping_rules[archive]
for (obj, symbol, scheme_name) in mapping.entries:
try:
if not (obj == Mapping.MAPPING_ALL_OBJECTS and symbol is None and
scheme_name == GenerationModel.DEFAULT_SCHEME):
self._add_mapping_rules(archive, obj, symbol, scheme_name, scheme_dictionary, mapping_rules)
except KeyError:
message = GenerationException.UNDEFINED_REFERENCE + " to scheme '" + scheme_name + "'."
raise GenerationException(message, mapping)
# Detect rule conflicts
for mapping_rules in all_mapping_rules.items():
self._detect_conflicts(mapping_rules)
# Add exclusions
for mapping_rules in all_mapping_rules.values():
self._create_exclusions(mapping_rules, default_rules, sections_infos)
placement_rules = collections.defaultdict(list)
# Add the default rules grouped by target
for default_rule in default_rules:
existing_rules = placement_rules[default_rule.target]
if default_rule.get_section_names():
existing_rules.append(default_rule)
archives = sorted(all_mapping_rules.keys())
for archive in archives:
# Add the mapping rules grouped by target
mapping_rules = sorted(all_mapping_rules[archive], key=lambda m: (m.specificity, str(m)))
for mapping_rule in mapping_rules:
existing_rules = placement_rules[mapping_rule.target]
if mapping_rule.get_section_names():
existing_rules.append(mapping_rule)
return placement_rules
def _detect_conflicts(self, rules):
(archive, rules_list) = rules
for specificity in range(0, PlacementRule.OBJECT_SPECIFICITY + 1):
rules_with_specificity = filter(lambda r: r.specificity == specificity, rules_list)
for rule_a, rule_b in itertools.combinations(rules_with_specificity, 2):
intersections = rule_a.get_sections_intersection(rule_b)
if intersections and rule_a.maps_same_entities_as(rule_b):
rules_string = str([str(rule_a), str(rule_b)])
message = "Rules " + rules_string + " map sections " + str(list(intersections)) + " into multiple targets."
raise GenerationException(message)
def _create_extra_rules(self, rules):
# This function generates extra rules for symbol specific rules. The reason for generating extra rules is to isolate,
# as much as possible, rules that require expansion. Particularly, object specific extra rules are generated.
rules_to_process = sorted(rules, key=lambda r: r.specificity)
symbol_specific_rules = list(filter(lambda r: r.specificity == PlacementRule.SYMBOL_SPECIFICITY, rules_to_process))
extra_rules = dict()
for symbol_specific_rule in symbol_specific_rules:
extra_rule_candidate = {s: None for s in symbol_specific_rule.get_section_names()}
super_rules = filter(lambda r: symbol_specific_rule.is_more_specific_rule_of(r), rules_to_process)
# Take a look at the existing rules that are more general than the current symbol-specific rule.
# Only generate an extra rule if there is no existing object specific rule for that section
for super_rule in super_rules:
intersections = symbol_specific_rule.get_sections_intersection(super_rule)
for intersection in intersections:
if super_rule.specificity != PlacementRule.OBJECT_SPECIFICITY:
extra_rule_candidate[intersection] = super_rule
else:
extra_rule_candidate[intersection] = None
# Generate the extra rules for the symbol specific rule section, keeping track of the generated extra rules
for (section, section_rule) in extra_rule_candidate.items():
if section_rule:
extra_rule = None
extra_rules_key = (symbol_specific_rule.archive, symbol_specific_rule.obj, section_rule.target)
try:
extra_rule = extra_rules[extra_rules_key]
if section not in extra_rule.get_section_names():
new_rule = PlacementRule(extra_rule.archive, extra_rule.obj, extra_rule.symbol,
list(extra_rule.get_section_names()) + [section], extra_rule.target)
extra_rules[extra_rules_key] = new_rule
except KeyError:
extra_rule = PlacementRule(symbol_specific_rule.archive, symbol_specific_rule.obj, None, [section], section_rule.target)
extra_rules[extra_rules_key] = extra_rule
return extra_rules.values()
def _create_exclusions(self, mapping_rules, default_rules, sections_info):
rules = list(default_rules)
rules.extend(mapping_rules)
extra_rules = self._create_extra_rules(rules)
mapping_rules.extend(extra_rules)
rules.extend(extra_rules)
# Sort the rules by means of how specific they are. Sort by specificity from lowest to highest
# * -> lib:* -> lib:obj -> lib:obj:symbol
sorted_rules = sorted(rules, key=lambda r: r.specificity)
# Now that the rules have been sorted, loop through each rule, and then loop
# through rules below it (higher indeces), adding exclusions whenever appropriate.
for general_rule in sorted_rules:
for specific_rule in reversed(sorted_rules):
if (specific_rule.specificity > general_rule.specificity and
specific_rule.specificity != PlacementRule.SYMBOL_SPECIFICITY) or \
(specific_rule.specificity == PlacementRule.SYMBOL_SPECIFICITY and
general_rule.specificity == PlacementRule.OBJECT_SPECIFICITY):
general_rule.add_exclusion(specific_rule, sections_info)
def add_fragments_from_file(self, fragment_file):
for fragment in fragment_file.fragments:
dict_to_append_to = None
if isinstance(fragment, Mapping) and fragment.deprecated and fragment.name in self.mappings.keys():
self.mappings[fragment.name].entries |= fragment.entries
else:
if isinstance(fragment, Scheme):
dict_to_append_to = self.schemes
elif isinstance(fragment, Sections):
dict_to_append_to = self.sections
else:
dict_to_append_to = self.mappings
# Raise exception when the fragment of the same type is already in the stored fragments
if fragment.name in dict_to_append_to.keys():
stored = dict_to_append_to[fragment.name].path
new = fragment.path
message = "Duplicate definition of fragment '%s' found in %s and %s." % (fragment.name, stored, new)
raise GenerationException(message)
dict_to_append_to[fragment.name] = fragment
class TemplateModel:
"""
Encapsulates a linker script template file. Finds marker syntax and handles replacement to generate the
final output.
"""
Marker = collections.namedtuple("Marker", "target indent rules")
def __init__(self, template_file):
self.members = []
self.file = os.path.realpath(template_file.name)
self._generate_members(template_file)
def _generate_members(self, template_file):
lines = template_file.readlines()
target = Fragment.IDENTIFIER
reference = Suppress("mapping") + Suppress("[") + target.setResultsName("target") + Suppress("]")
pattern = White(" \t").setResultsName("indent") + reference
# Find the markers in the template file line by line. If line does not match marker grammar,
# set it as a literal to be copied as is to the output file.
for line in lines:
try:
parsed = pattern.parseString(line)
indent = parsed.indent
target = parsed.target
marker = TemplateModel.Marker(target, indent, [])
self.members.append(marker)
except ParseException:
# Does not match marker syntax
self.members.append(line)
def fill(self, mapping_rules):
for member in self.members:
target = None
try:
target = member.target
rules = member.rules
del rules[:]
rules.extend(mapping_rules[target])
except KeyError:
message = GenerationException.UNDEFINED_REFERENCE + " to target '" + target + "'."
raise GenerationException(message)
except AttributeError:
pass
def write(self, output_file):
# Add information that this is a generated file.
output_file.write("/* Automatically generated file; DO NOT EDIT */\n")
output_file.write("/* Espressif IoT Development Framework Linker Script */\n")
output_file.write("/* Generated from: %s */\n" % self.file)
output_file.write("\n")
# Do the text replacement
for member in self.members:
try:
indent = member.indent
rules = member.rules
for rule in rules:
generated_line = "".join([indent, str(rule), '\n'])
output_file.write(generated_line)
except AttributeError:
output_file.write(member)
class GenerationException(LdGenFailure):
"""
Exception for linker script generation failures such as undefined references/ failure to
evaluate conditions, duplicate mappings, etc.
"""
UNDEFINED_REFERENCE = "Undefined reference"
def __init__(self, message, fragment=None):
self.fragment = fragment
self.message = message
def __str__(self):
if self.fragment:
return "%s\nIn fragment '%s' defined in '%s'." % (self.message, self.fragment.name, self.fragment.path)
else:
return self.message
class SectionsInfo(dict):
"""
Encapsulates an output of objdump. Contains information about the static library sections
and names
"""
__info = collections.namedtuple("__info", "filename content")
def __init__(self):
self.sections = dict()
def add_sections_info(self, sections_info_dump):
first_line = sections_info_dump.readline()
archive_path = (Literal("In archive").suppress() +
# trim the last character from archive_path, :
Word(printables + " ").setResultsName("archive_path").setParseAction(lambda t: t[0][:-1]) +
LineEnd())
parser = archive_path
results = None
try:
results = parser.parseString(first_line)
except ParseException as p:
raise ParseException("Parsing sections info for library " + sections_info_dump.name + " failed. " + p.message)
archive = os.path.basename(results.archive_path)
self.sections[archive] = SectionsInfo.__info(sections_info_dump.name, sections_info_dump.read())
def _get_infos_from_file(self, info):
# Object file line: '{object}: file format elf32-xtensa-le'
object = Fragment.ENTITY.setResultsName("object") + Literal(":").suppress() + Literal("file format elf32-xtensa-le").suppress()
# Sections table
header = Suppress(Literal("Sections:") + Literal("Idx") + Literal("Name") + Literal("Size") + Literal("VMA") +
Literal("LMA") + Literal("File off") + Literal("Algn"))
entry = Word(nums).suppress() + Fragment.ENTITY + Suppress(OneOrMore(Word(alphanums, exact=8)) +
Word(nums + "*") + ZeroOrMore(Word(alphas.upper()) +
Optional(Literal(","))))
# Content is object file line + sections table
content = Group(object + header + Group(ZeroOrMore(entry)).setResultsName("sections"))
parser = Group(ZeroOrMore(content)).setResultsName("contents")
sections_info_text = info.content
results = None
try:
results = parser.parseString(sections_info_text)
except ParseException as p:
raise ParseException("Unable to parse section info file " + info.filename + ". " + p.message)
return results
def get_obj_sections(self, archive, obj):
stored = self.sections[archive]
# Parse the contents of the sections file
if not isinstance(stored, dict):
parsed = self._get_infos_from_file(stored)
stored = dict()
for content in parsed.contents:
sections = list(map(lambda s: s, content.sections))
stored[content.object] = sections
self.sections[archive] = stored
for obj_key in stored.keys():
if obj_key == obj + ".o" or obj_key == obj + ".c.obj":
return stored[obj_key]