123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406 |
- /*
- * Copyright (c) 2020-2021, Itamar S. <itamar8910@gmail.com>
- *
- * SPDX-License-Identifier: BSD-2-Clause
- */
- #include "DwarfInfo.h"
- #include "AddressRanges.h"
- #include "AttributeValue.h"
- #include "CompilationUnit.h"
- #include <AK/ByteReader.h>
- #include <AK/LEB128.h>
- #include <AK/MemoryStream.h>
- #include <LibDebug/DebugInfo.h>
- namespace Debug::Dwarf {
- DwarfInfo::DwarfInfo(ELF::Image const& elf)
- : m_elf(elf)
- {
- m_debug_info_data = section_data(".debug_info"sv);
- m_abbreviation_data = section_data(".debug_abbrev"sv);
- m_debug_strings_data = section_data(".debug_str"sv);
- m_debug_line_data = section_data(".debug_line"sv);
- m_debug_line_strings_data = section_data(".debug_line_str"sv);
- m_debug_range_lists_data = section_data(".debug_rnglists"sv);
- m_debug_str_offsets_data = section_data(".debug_str_offsets"sv);
- m_debug_addr_data = section_data(".debug_addr"sv);
- m_debug_ranges_data = section_data(".debug_ranges"sv);
- populate_compilation_units().release_value_but_fixme_should_propagate_errors();
- }
- DwarfInfo::~DwarfInfo() = default;
- ReadonlyBytes DwarfInfo::section_data(StringView section_name) const
- {
- auto section = m_elf.lookup_section(section_name);
- if (!section.has_value())
- return {};
- return section->bytes();
- }
- ErrorOr<void> DwarfInfo::populate_compilation_units()
- {
- if (!m_debug_info_data.data())
- return {};
- FixedMemoryStream debug_info_stream { m_debug_info_data };
- FixedMemoryStream line_info_stream { m_debug_line_data };
- while (!debug_info_stream.is_eof()) {
- auto unit_offset = TRY(debug_info_stream.tell());
- auto compilation_unit_header = TRY(debug_info_stream.read_value<CompilationUnitHeader>());
- VERIFY(compilation_unit_header.common.version <= 5);
- VERIFY(compilation_unit_header.address_size() == sizeof(FlatPtr));
- u32 length_after_header = compilation_unit_header.length() - (compilation_unit_header.header_size() - offsetof(CompilationUnitHeader, common.version));
- auto line_program = TRY(LineProgram::create(*this, line_info_stream));
- // HACK: Clang generates line programs for embedded resource assembly files, but not compile units.
- // Meaning that for graphical applications, some line info data would be unread, triggering the assertion below.
- // As a fix, we don't create compilation units for line programs that come from resource files.
- #if defined(AK_COMPILER_CLANG)
- if (line_program->looks_like_embedded_resource()) {
- TRY(debug_info_stream.seek(unit_offset));
- } else
- #endif
- {
- m_compilation_units.append(make<CompilationUnit>(*this, unit_offset, compilation_unit_header, move(line_program)));
- TRY(debug_info_stream.discard(length_after_header));
- }
- }
- VERIFY(line_info_stream.is_eof());
- return {};
- }
- ErrorOr<AttributeValue> DwarfInfo::get_attribute_value(AttributeDataForm form, ssize_t implicit_const_value,
- SeekableStream& debug_info_stream, CompilationUnit const* unit) const
- {
- AttributeValue value;
- value.m_form = form;
- value.m_compilation_unit = unit;
- auto assign_raw_bytes_value = [&](size_t length) -> ErrorOr<void> {
- value.m_data.as_raw_bytes = { debug_info_data().offset_pointer(TRY(debug_info_stream.tell())), length };
- TRY(debug_info_stream.discard(length));
- return {};
- };
- switch (form) {
- case AttributeDataForm::StringPointer: {
- auto offset = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::String;
- auto strings_data = debug_strings_data();
- value.m_data.as_string = bit_cast<char const*>(strings_data.offset_pointer(offset));
- break;
- }
- case AttributeDataForm::Data1: {
- auto data = TRY(debug_info_stream.read_value<u8>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_unsigned = data;
- break;
- }
- case AttributeDataForm::Data2: {
- auto data = TRY(debug_info_stream.read_value<u16>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_signed = data;
- break;
- }
- case AttributeDataForm::Addr: {
- auto address = TRY(debug_info_stream.read_value<FlatPtr>());
- value.m_type = AttributeValue::Type::Address;
- value.m_data.as_addr = address;
- break;
- }
- case AttributeDataForm::SData: {
- i64 data = TRY(debug_info_stream.read_value<LEB128<i64>>());
- value.m_type = AttributeValue::Type::SignedNumber;
- value.m_data.as_signed = data;
- break;
- }
- case AttributeDataForm::UData: {
- u64 data = TRY(debug_info_stream.read_value<LEB128<u64>>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_unsigned = data;
- break;
- }
- case AttributeDataForm::SecOffset: {
- auto data = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::SecOffset;
- value.m_data.as_unsigned = data;
- break;
- }
- case AttributeDataForm::Data4: {
- auto data = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_unsigned = data;
- break;
- }
- case AttributeDataForm::Data8: {
- auto data = TRY(debug_info_stream.read_value<u64>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_unsigned = data;
- break;
- }
- case AttributeDataForm::Data16: {
- value.m_type = AttributeValue::Type::RawBytes;
- TRY(assign_raw_bytes_value(16));
- break;
- }
- case AttributeDataForm::Ref4: {
- auto data = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::DieReference;
- VERIFY(unit);
- value.m_data.as_unsigned = data + unit->offset();
- break;
- }
- case AttributeDataForm::RefUData: {
- auto data = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- value.m_type = AttributeValue::Type::DieReference;
- VERIFY(unit);
- value.m_data.as_unsigned = data + unit->offset();
- break;
- }
- case AttributeDataForm::FlagPresent: {
- value.m_type = AttributeValue::Type::Boolean;
- value.m_data.as_bool = true;
- break;
- }
- case AttributeDataForm::ExprLoc: {
- size_t length = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- value.m_type = AttributeValue::Type::DwarfExpression;
- TRY(assign_raw_bytes_value(length));
- break;
- }
- case AttributeDataForm::String: {
- u32 str_offset = TRY(debug_info_stream.tell());
- value.m_type = AttributeValue::Type::String;
- value.m_data.as_string = bit_cast<char const*>(debug_info_data().offset_pointer(str_offset));
- TRY(debug_info_stream.discard(strlen(value.m_data.as_string) + 1));
- break;
- }
- case AttributeDataForm::Block1: {
- value.m_type = AttributeValue::Type::RawBytes;
- auto length = TRY(debug_info_stream.read_value<u8>());
- TRY(assign_raw_bytes_value(length));
- break;
- }
- case AttributeDataForm::Block2: {
- value.m_type = AttributeValue::Type::RawBytes;
- auto length = TRY(debug_info_stream.read_value<u16>());
- TRY(assign_raw_bytes_value(length));
- break;
- }
- case AttributeDataForm::Block4: {
- value.m_type = AttributeValue::Type::RawBytes;
- auto length = TRY(debug_info_stream.read_value<u32>());
- TRY(assign_raw_bytes_value(length));
- break;
- }
- case AttributeDataForm::Block: {
- value.m_type = AttributeValue::Type::RawBytes;
- size_t length = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- TRY(assign_raw_bytes_value(length));
- break;
- }
- case AttributeDataForm::LineStrP: {
- auto offset = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::String;
- auto strings_data = debug_line_strings_data();
- value.m_data.as_string = bit_cast<char const*>(strings_data.offset_pointer(offset));
- break;
- }
- case AttributeDataForm::ImplicitConst: {
- /* Value is part of the abbreviation record. */
- value.m_type = AttributeValue::Type::SignedNumber;
- value.m_data.as_signed = implicit_const_value;
- break;
- }
- case AttributeDataForm::StrX1: {
- auto index = TRY(debug_info_stream.read_value<u8>());
- value.m_type = AttributeValue::Type::String;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::StrX2: {
- auto index = TRY(debug_info_stream.read_value<u16>());
- value.m_type = AttributeValue::Type::String;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::StrX4: {
- auto index = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::String;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::StrX: {
- size_t index = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- value.m_type = AttributeValue::Type::String;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::AddrX1: {
- auto index = TRY(debug_info_stream.read_value<u8>());
- value.m_type = AttributeValue::Type::Address;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::AddrX2: {
- auto index = TRY(debug_info_stream.read_value<u16>());
- value.m_type = AttributeValue::Type::Address;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::AddrX4: {
- auto index = TRY(debug_info_stream.read_value<u32>());
- value.m_type = AttributeValue::Type::Address;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::AddrX: {
- size_t index = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- value.m_type = AttributeValue::Type::Address;
- value.m_data.as_unsigned = index;
- break;
- }
- case AttributeDataForm::LocListX:
- case AttributeDataForm::RngListX: {
- size_t index = TRY(debug_info_stream.read_value<LEB128<size_t>>());
- value.m_type = AttributeValue::Type::UnsignedNumber;
- value.m_data.as_unsigned = index;
- break;
- }
- default:
- dbgln("Unimplemented AttributeDataForm: {}", to_underlying(form));
- VERIFY_NOT_REACHED();
- }
- return value;
- }
- ErrorOr<void> DwarfInfo::build_cached_dies() const
- {
- auto insert_to_cache = [this](DIE const& die, DIERange const& range) {
- m_cached_dies_by_range.insert(range.start_address, DIEAndRange { die, range });
- m_cached_dies_by_offset.insert(die.offset(), die);
- };
- auto get_ranges_of_die = [this](DIE const& die) -> ErrorOr<Vector<DIERange>> {
- auto ranges = TRY(die.get_attribute(Attribute::Ranges));
- if (ranges.has_value()) {
- size_t offset;
- if (ranges->form() == AttributeDataForm::SecOffset) {
- offset = ranges->as_unsigned();
- } else {
- auto index = ranges->as_unsigned();
- auto base = TRY(die.compilation_unit().range_lists_base());
- // FIXME: This assumes that the format is DWARf32
- auto offsets = debug_range_lists_data().slice(base);
- offset = ByteReader::load32(offsets.offset_pointer(index * sizeof(u32))) + base;
- }
- Vector<DIERange> entries;
- if (die.compilation_unit().dwarf_version() == 5) {
- auto range_lists_stream = TRY(try_make<FixedMemoryStream>(debug_range_lists_data()));
- TRY(range_lists_stream->seek(offset));
- AddressRangesV5 address_ranges(move(range_lists_stream), die.compilation_unit());
- TRY(address_ranges.for_each_range([&entries](auto range) {
- entries.empend(range.start, range.end);
- }));
- } else {
- auto ranges_stream = TRY(try_make<FixedMemoryStream>(debug_ranges_data()));
- TRY(ranges_stream->seek(offset));
- AddressRangesV4 address_ranges(move(ranges_stream), die.compilation_unit());
- TRY(address_ranges.for_each_range([&entries](auto range) {
- entries.empend(range.start, range.end);
- }));
- }
- return entries;
- }
- auto start = TRY(die.get_attribute(Attribute::LowPc));
- auto end = TRY(die.get_attribute(Attribute::HighPc));
- if (!start.has_value() || !end.has_value())
- return Vector<DIERange> {};
- VERIFY(start->type() == Dwarf::AttributeValue::Type::Address);
- // DW_AT_high_pc attribute can have different meanings depending on the attribute form.
- // (Dwarf version 5, section 2.17.2).
- uint32_t range_end = 0;
- if (end->form() == Dwarf::AttributeDataForm::Addr)
- range_end = TRY(end->as_addr());
- else
- range_end = TRY(start->as_addr()) + end->as_unsigned();
- return Vector<DIERange> { DIERange { TRY(start.value().as_addr()), range_end } };
- };
- // If we simply use a lambda, type deduction fails because it's used recursively.
- Function<ErrorOr<void>(DIE const& die)> insert_to_cache_recursively;
- insert_to_cache_recursively = [&](DIE const& die) -> ErrorOr<void> {
- if (die.offset() == 0 || die.parent_offset().has_value()) {
- auto ranges = TRY(get_ranges_of_die(die));
- for (auto& range : ranges) {
- insert_to_cache(die, range);
- }
- }
- TRY(die.for_each_child([&](DIE const& child) -> ErrorOr<void> {
- if (!child.is_null()) {
- TRY(insert_to_cache_recursively(child));
- }
- return {};
- }));
- return {};
- };
- TRY(for_each_compilation_unit([&](CompilationUnit const& compilation_unit) -> ErrorOr<void> {
- TRY(insert_to_cache_recursively(compilation_unit.root_die()));
- return {};
- }));
- m_built_cached_dies = true;
- return {};
- }
- ErrorOr<Optional<DIE>> DwarfInfo::get_die_at_address(FlatPtr address) const
- {
- if (!m_built_cached_dies)
- TRY(build_cached_dies());
- auto iter = m_cached_dies_by_range.find_largest_not_above_iterator(address);
- while (!iter.is_end() && !iter.is_begin() && iter->range.end_address < address) {
- --iter;
- }
- if (iter.is_end())
- return Optional<DIE> {};
- if (iter->range.start_address > address || iter->range.end_address < address) {
- return Optional<DIE> {};
- }
- return iter->die;
- }
- ErrorOr<Optional<DIE>> DwarfInfo::get_cached_die_at_offset(FlatPtr offset) const
- {
- if (!m_built_cached_dies)
- TRY(build_cached_dies());
- auto* die = m_cached_dies_by_offset.find(offset);
- if (!die)
- return Optional<DIE> {};
- return *die;
- }
- }
|