1 Commits

Author	SHA1	Message	Date
Dodji Seketeli	cf8eba68c3	Implement string interning for Libabigail ... This patch implements string interning optimization. One can read about the principles of this optimization at https://en.wikipedia.org/wiki/String_interning. The patch introduces an abigail::interned_string type, as well as an abigail::interned_string_pool type. Each environment type owns a string pool and strings are interned in that pool for all types and decls of that environments. The interned_string has methods to interact seemingly with std::string including a hashing function. Of course hashing and comparing interned_string is faster than for std::string. To enable ABI artifacts to intern strings, each constructor of ABI artifacts now takes the environment it's constructed in as parameter. From the environment, it can thus use the interned string pool. The patch then changes declaration names to be of type interned_string, and performs the necessary adjustments. The hash maps that hash strings coming from those declaration names are adjusted to hash interned_string. * include/Makefile.am: Add the new abg-interned-str.h file to source distribution. * include/abg-corpus.h (corpus::corpus): Re-arrange the order of * src/abg-corpus.cc (corpus::exported_decls_builder::priv::get_id): Return interned_string rather than std::string. (corpus::corpus): Re-arrange the order of parameters: take an environment as first parameter. parameters: take an environment as first parameter. * include/abg-dwarf-reader.h (lookup_symbol_from_elf) (lookup_public_function_symbol_from_elf): Likewise. * src/abg-dwarf-reader.cc (lookup_symbol_from_sysv_hash_tab) (lookup_symbol_from_gnu_hash_tab) (lookup_symbol_from_elf_hash_tab, lookup_symbol_from_symtab) (lookup_symbol_from_elf, lookup_public_function_symbol_from_elf) (lookup_public_variable_symbol_from_elf, lookup_symbol_from_elf) (lookup_public_function_symbol_from_elf): Take an environment as first parameter and adjust. (build_translation_unit_and_add_to_ir) (build_namespace_decl_and_add_to_ir, build_type_decl) (build_enum_type, finish_member_function_reading) (build_class_type_and_add_to_ir, build_function_type) (read_debug_info_into_corpus, read_corpus_from_elf): Adjust. * include/abg-fwd.h: Include abg-interned-str.h (get_type_name, get_function_type_name, get_method_type_name): Return a interned_string, rather than a std::string. * include/abg-interned-str.h: New declarations for interned strings and their pool. * include/abg-ir.h (environment::intern): Declare new method. (elf_symbol::{g,s}et_environment): Likewise. (type_or_decl_base::type_or_decl_base): Make the default constructor private. ({translation, type_or_decl_base}::set_environment) (set_environment_for_artifact): Take a const environment*. (elf_symbol::elf_symbol) (elf_symbol::create) (type_or_decl_base::type_or_decl_base) (translation::translation, decl_base::decl_base) (scope_decl::scope_decl, type_base::type_base) (type_decl::type_decl, scope_type_decl::scope_type_decl) (namespace_decl::namespace_decl) (enum_type_decl::enumerator::enumerator) (function_type::function_type, method_type::method_type) (template_decl::template_decl, function_tdecl::function_tdecl) (class_tdecl::class_tdecl, class_decl::class_decl): Take an environment. (type_or_decl_base::operator=) (enum_type_decl::enumerator::get_environment): Declare new method. (decl_base::{peek_qualified_name, peek_temporary_qualified_name, get_qualified_name, get_name, get_qualified_parent_name, get_linkage_name}, qualified_type_def::get_qualified_name) (reference_type_def::get_qualified_name) (array_type_def::get_qualified_name) (enum_type_decl::enumerator::{get_name, get_qualified_name}) ({var,function}_decl::get_id) (function_decl::parameter::{get_type_name, get_name_id}): Return an interned_string, rather than a std::string. (decl_base::{set_qualified_name, set_temporary_qualified_name, get_qualified_name, set_linkage_name}) (qualified_type_def::get_qualified_name) (reference_type_def::get_qualified_name) (array_type_def::get_qualified_name) (function_decl::parameter::get_qualified_name): Take an interned_string, rather than a std::string. (class_decl::member_{class,function}_template::member_{class,function}_template): Adjust. * src/abg-ir.cc (environment_setter::env_): Make this be a pointer to const environment. (environment_setter::visit_begin): Adjust. (interned_string_pool::priv): Define new type. (interned_string_pool::*): Define the method declared in abg-interned-str. h. (operator==, operator!=, operator+): Define operator for interned_string and std::string (operator<<): Define for interned_string. (translation_unit::priv::env_): Make this be a pointer to const environment. (translation_unit::priv::priv): Take a pointer to const environment. (elf_symbol::priv::env_): New data member. (elf_symbol::priv::priv): Adjust. Make an overoad take an environment. (translation_unit::{g,s}et_environment): Adjust. (interned_string_bool_map_type): New typedef. (environment::priv::classes_being_compared_): Make this hastable of string be a hashtable of interned_string. (environment::priv::string_pool_): New data member. (environment::{get_void_type_decl, get_variadic_parameter_type_decl}): Adjust. (type_or_decl_base::priv::env_): Make this be a pointer to const environment. (type_or_decl::base::priv::priv): Adjust. (type_or_decl_base::set_environment) (set_environment_for_artifact): Take a pointer to const environment. (elf_symbol::{g,s}et_environment, environment::intern) (type_or_decl_base::operator=): Define new methods. (decl_base::priv::{name_, qualified_parent_name_, temporary_qualified_name_, qualified_name_, linkage_name_}): Make these data member be of tpe interned_string. (decl_base::priv::priv): Make this take an environment. Adjust. (decl_base::{peek_qualified_name, peek_temporary_qualified_name, get_linkage_name, get_qualified_parent_name, get_name, get_qualified_name}, get_type_name, get_function_type_name) (get_method_type_name, get_node_name) (qualified_type_def::get_qualified_name) (pointer_type_def::get_qualified_name) (array_type_def::get_qualified_name) (enum_type_decl::enumerator::get_qualified_name) (var_decl::get_id, function_decl::get_id) (function_decl::parameter::get_{name_id, type_name}): Return an interned_string. (decl_base::{set_qualified_name, set_temporary_qualified_name}) (qualified_type_def::get_qualified_name) (pointer_type_def::get_qualified_name) (reference_type_def::get_qualified_name) (array_type_def::get_qualified_name) (function_decl::parameter::get_qualified_name): Take an interned_string. (decl_base::{set_name, set_linkage_name}): Intern the std::string passed in parameter. (equals): In the overload for decl_base, adjust for a little speed optimization that is justified by profiling. (pointer_type_def::priv::{internal_qualified_name_, temp_internal_qualified_name_}): Make these data member be interned_string. (enum_type_decl::enumerator::priv::env_): New data member. (enum_type_decl::enumerator::priv::{name_, qualified_name}): Make these data member be of type interned_string. (enum_type_decl::enumerator::get_environment): New method. (enum_type_decl::enumerator::priv::priv) Adjust. (typedef_decl::operator==): Implement a little speed optimization. (var_decl::priv::nake_type_): New data member. (var_decl::priv::id_): Make this data member be of type interned_string. (equals): In the overload for var_decl, function_type, function_decl, adjust for the use of interned_string. (function_decl::priv::id_): Make this be of type interned_string. (scope_decl::{add_member_decl, insert_member_decl}) (lookup_function_type_in_translation_unit) (synthesize_type_from_translation_unit, lookup_node_in_scope) (lookup_type_in_scope, scope_decl::scope_decl) (qualified_type_def::qualified_type_def) (qualified_type_def::get_qualified_name) (pointer_type_def::pointer_type_def) (reference_type_def::reference_type_def) (array_type_def::array_type_def, array_type_def::append_subrange) (array_type_def::get_qualified_name) (enum_type_decl::enum_type_decl) (enum_type_decl::enumerator::get_qualified_name) (enum_type_decl::enumerator::set_name) (typedef_decl::typedef_decl, var_decl::var_decl) (function_type::function_type, method_type::method_type) (function_decl::function_decl) (function_decl::parameter::parameter) (class_decl::priv::comparison_started) (class_decl::add_base_specifier) (class_decl::base_spec::base_spec) (class_decl::method_decl::method_decl) (type_tparameter::type_tparameter) (non_type_tparameter::non_type_tparameter) (template_tparameter::template_tparameter) (type_composition::type_composition) (function_tdecl::function_tdecl, class_tdecl::class_tdecl) (qualified_name_setter::do_update): Adjust. (translation_unit::translation_unit, elf_symbol::elf_symbol) (elf_symbol::create, type_or_decl_base::type_or_decl_base) (decl_base::decl_base, type_base::type_base) (type_decl::type_decl, scope_type_decl::scope_type_decl) (namespace_decl::namespace_decl) (enum_type_decl::enumerator::enumerator, class_decl::class_decl) (template_decl::template_decl, function_tdecl::function_tdecl) (class_tdecl::class_tdecl): Take an environment. * src/abg-comparison.cc (function_suppression::suppresses_function): Adjust. * src/abg-reader.cc (read_translation_unit) (read_corpus_from_input, build_namespace_decl, build_elf_symbol) (build_function_parameter, build_function_decl, build_type_decl) (build_function_type, build_enum_type_decl, build_enum_type_decl) (build_class_decl, build_function_tdecl, build_class_tdecl) (read_corpus_from_native_xml): Likewise. * src/abg-writer.cc (id_manager::m_cur_id): Make this mutable. (id_manager::m_env): New data member. (id_manager::id_manager): Adjust. (id_manager::get_environment): New method. (id_manager::{get_id, get_id_with_prefix}): Return an interned_string. (type_ptr_map): Make this be a hash map of type_base* -> interned_string, rather a type_base* -> string. (write_context::m_env): New data member. (write_context::m_type_id_map): Make this data member be mutable. (write_context::m_emitted_type_id_map): Make this be a hash map of interned_string -> bool, rather than string -> bool. (write_context::write_context): Take an environment and adjust. (write_context::get_environment): New method. (write_context::get_id_manager): New const overload. (write_context::get_id_for_type): Return an interned_string; adjust. (write_context::{record_type_id_as_emitted, record_type_as_referenced}): Adjust. (write_context::type_id_is_emitted): Take an interned_string. (write_context::{type_is_emitted, record_decl_only_type_as_emitted}): Adjust. (write_translation_unit, write_corpus_to_native_xml, dump): Adjust. * tools/abisym.cc (main): Adjust. * tests/data/test-read-write/test22.xml: Adjust. * tests/data/test-read-write/test23.xml: Adjust. * tests/data/test-read-write/test26.xml: Adjust. Signed-off-by: Dodji Seketeli <dodji@redhat.com>	2016-02-24 15:13:20 +01:00