Merge pull request #246 from chaoticgd/bitfields

Improve support for bitfields and 128-bit values

CMakeLists.txt

@@ -46,6 +46,8 @@ add_library(ccc STATIC
 	src/ccc/elf_symtab.h
 	src/ccc/importer_flags.cpp
 	src/ccc/importer_flags.h
+	src/ccc/int128.cpp
+	src/ccc/int128.h
 	src/ccc/mdebug_analysis.cpp
 	src/ccc/mdebug_analysis.h
 	src/ccc/mdebug_importer.cpp
@@ -93,6 +95,7 @@ add_library(ccc_platform STATIC
 set(TEST_SOURCES
 	test/demangler_tests.cpp
 	test/collision_tests.cpp
+	test/ccc/int128_tests.cpp
 	test/ccc/mdebug_importer_tests.cpp
 	test/ccc/stabs_tests.cpp
 	test/ccc/symbol_database_tests.cpp

docs/JsonFormat.md

@@ -8,6 +8,7 @@ and `ast_json.cpp` for the code that currently defines the format.
 
 | Format Version | Release | Changes |
 | - | - | - |
+| 15 | v2.2 | The offset_bytes and bitfield_offset_bits properties of bitfield nodes now represent and are relative to the beginning of the storage unit respectively. |
 | 14 | v2.1 | Added stack_frame_size property for function symbols. |
 | 13 | v2.0 | Added size_bytes field to all nodes. Renamed data_type_handle property to just data_type (since it's not a handle). |
 | 12 | | Added format and application properties to root object. Added hash property to function symbols. |

docs/ProjectStructure.md

@@ -11,6 +11,7 @@
 - src/ccc/elf.cpp: Parses ELF files.
 - src/ccc/elf_symtab.cpp: Parses the ELF symbol table.
 - src/ccc/importer_flags.cpp: An enum and help information printing for importer configuration flags.
+- src/ccc/int128.cpp: 128-bit integer types.
 - src/ccc/mdebug_analysis.cpp: Accepts a stream of symbols and imports the data.
 - src/ccc/mdebug_importer.cpp: Top-level file for parsing .mdebug symbol tables.
 - src/ccc/mdebug_section.cpp: Parses the .mdebug binary format.

src/ccc/ast.cpp

@@ -42,6 +42,47 @@ std::pair<const Node*, const DataType*> Node::physical_type(const SymbolDatabase
 	return const_cast<Node*>(this)->physical_type(const_cast<SymbolDatabase&>(database), max_depth);
 }
 
+BuiltInClass BitField::storage_unit_type(const SymbolDatabase& database) const
+{
+	BuiltInClass result = BuiltInClass::VOID_TYPE;
+
+	if(underlying_type.get()) {
+		auto [type, symbol] = underlying_type->physical_type(database);
+		switch(type->descriptor) {
+			case BUILTIN:
+				result = type->as<BuiltIn>().bclass;
+				break;
+			case ENUM:
+				result = BuiltInClass::SIGNED_32;
+				break;
+			default:
+				break;
+		}
+	}
+
+	return result;
+}
+
+u128 BitField::unpack_unsigned(u128 storage_unit) const
+{
+	return (storage_unit >> bitfield_offset_bits) & ((u128(1) << size_bits) - 1);
+}
+
+s128 BitField::unpack_signed(u128 storage_unit) const
+{
+	return s128(storage_unit << (128 - (bitfield_offset_bits + size_bits))) >> (128 - size_bits);
+}
+
+u128 BitField::pack_unsigned(u128 bitfield) const
+{
+	return (bitfield & ((u128(1) << size_bits) - 1)) << bitfield_offset_bits;
+}
+
+u128 BitField::pack_signed(s128 bitfield) const
+{
+	return u128((bitfield & ((u128(1) << size_bits) - 1)) << (128 - size_bits)) >> (128 - bitfield_offset_bits - size_bits);
+}
+
 const char* member_function_modifier_to_string(MemberFunctionModifier modifier)
 {
 	switch(modifier) {

src/ccc/ast.h

@@ -3,6 +3,7 @@
 
 #pragma once
 
+#include "int128.h"
 #include "symbol_database.h"
 
 namespace ccc::ast {
@@ -97,14 +98,6 @@ struct Array : Node {
 	static const constexpr NodeDescriptor DESCRIPTOR = ARRAY;
 };
 
-struct BitField : Node {
-	s32 bitfield_offset_bits = -1; // Offset relative to the last byte (not the position of the underlying type!).
-	std::unique_ptr<Node> underlying_type;
-
-	BitField() : Node(DESCRIPTOR) {}
-	static const constexpr NodeDescriptor DESCRIPTOR = BITFIELD;
-};
-
 enum class BuiltInClass {
 	VOID_TYPE,
 	UNSIGNED_8, SIGNED_8, UNQUALIFIED_8, BOOL_8,
@@ -114,6 +107,20 @@ enum class BuiltInClass {
 	UNSIGNED_128, SIGNED_128, UNQUALIFIED_128, FLOAT_128
 };
 
+struct BitField : Node {
+	s32 bitfield_offset_bits = -1;
+	std::unique_ptr<Node> underlying_type;
+
+	BitField() : Node(DESCRIPTOR) {}
+	static const constexpr NodeDescriptor DESCRIPTOR = BITFIELD;
+
+	BuiltInClass storage_unit_type(const SymbolDatabase& database) const;
+	u128 unpack_unsigned(u128 storage_unit) const;
+	s128 unpack_signed(u128 storage_unit) const;
+	u128 pack_unsigned(u128 bitfield) const;
+	u128 pack_signed(s128 bitfield) const;
+};
+
 struct BuiltIn : Node {
 	BuiltInClass bclass = BuiltInClass::VOID_TYPE;
 

src/ccc/ccc.h

@@ -10,6 +10,7 @@
 #include "elf.h"
 #include "elf_symtab.h"
 #include "importer_flags.h"
+#include "int128.h"
 #include "mdebug_analysis.h"
 #include "mdebug_importer.h"
 #include "mdebug_section.h"