summaryrefslogtreecommitdiff
path: root/tools/libdol2asm
diff options
context:
space:
mode:
authorJulgodis <>2021-04-08 21:05:48 +0200
committerJulgodis <>2021-04-08 21:05:48 +0200
commit2661db6aaa98a5bbd17691cb438ef64752dbcdf4 (patch)
tree4e6d8499431192316bc2bae0cec8e6f92636b4b4 /tools/libdol2asm
parenta23a10639979ee10cbea5ffdadadc490d9e60227 (diff)
auto decompile more functions
Diffstat (limited to 'tools/libdol2asm')
-rw-r--r--tools/libdol2asm/data/base.py14
-rw-r--r--tools/libdol2asm/data/function/ret.py33
-rw-r--r--tools/libdol2asm/data/reference_array.py1
-rw-r--r--tools/libdol2asm/data/symbol.py9
-rw-r--r--tools/libdol2asm/generate_functions.py161
-rw-r--r--tools/libdol2asm/split_asm.py7
6 files changed, 191 insertions, 34 deletions
diff --git a/tools/libdol2asm/data/base.py b/tools/libdol2asm/data/base.py
index 6ceada5eb9..6878ca1341 100644
--- a/tools/libdol2asm/data/base.py
+++ b/tools/libdol2asm/data/base.py
@@ -84,6 +84,12 @@ class ArbitraryData(Symbol):
self.size // self.element_size,
self.padding // self.element_size)
+ def return_type(self):
+ array_type = self.array_type()
+ if isinstance(array_type, ArrayType) or isinstance(array_type, PaddingArrayType) or isinstance(array_type, ZeroArrayType):
+ return PointerType(array_type.base)
+ return array_type
+
def cpp_reference(self, accessor, addr):
name = self.declaration_name(forward=False, c_export=False,full_qualified_name=True)
if addr == self.addr:
@@ -92,6 +98,14 @@ class ArbitraryData(Symbol):
offset = addr - self.addr
return f"(((char*)&{name})+0x{offset:X})"
+ def cpp_load(self, accessor, addr):
+ name = self.declaration_name(forward=False, c_export=False,full_qualified_name=True)
+ if addr == self.addr:
+ return f"{name}"
+ else:
+ offset = addr - self.addr
+ return f"*(((char*)&{name})+0x{offset:X})"
+
def declaration_name(self, forward: bool,
c_export: bool,
full_qualified_name: bool):
diff --git a/tools/libdol2asm/data/function/ret.py b/tools/libdol2asm/data/function/ret.py
index bf15e02e9b..aed96c2315 100644
--- a/tools/libdol2asm/data/function/ret.py
+++ b/tools/libdol2asm/data/function/ret.py
@@ -6,16 +6,39 @@ from .base import *
@dataclass(eq=False)
class ReturnFunction(Function):
- return_value: str = None
-
- def export_return_value(self):
- return self.return_value
+ def export_return_value(self, symbol_table):
+ assert False
async def export_function_body(self, exporter, builder: AsyncBuilder):
- return_value = self.export_return_value()
+ return_value = self.export_return_value(exporter.gst)
await builder.write(f" {{")
if return_value:
await builder.write(f"\treturn {return_value};")
else:
await builder.write(f"\t/* empty function */")
await builder.write(f"}}")
+
+@dataclass(eq=False)
+class CustomReturnFunction(ReturnFunction):
+ return_value: str = None
+
+ def export_return_value(self, symbol_table):
+ return self.return_value
+
+@dataclass(eq=False)
+class SymbolReturnFunction(ReturnFunction):
+ symbol_addr: int = 0
+ load_or_reference: bool = True
+
+ def gather_references(self, context, valid_range):
+ self.references = set([ self.symbol_addr ])
+
+ def export_return_value(self, symbol_table):
+ symbol = symbol_table[-1, self.symbol_addr]
+ assert symbol
+ name = symbol.cpp_reference(self, self.symbol_addr)
+ type = PointerType(self.return_type)
+ dereference = ""
+ if self.load_or_reference:
+ dereference = "*"
+ return f"{dereference}({type.type()})({name})"
diff --git a/tools/libdol2asm/data/reference_array.py b/tools/libdol2asm/data/reference_array.py
index 1b05b92342..d5d5fdf918 100644
--- a/tools/libdol2asm/data/reference_array.py
+++ b/tools/libdol2asm/data/reference_array.py
@@ -33,6 +33,7 @@ class ReferenceArray(ArbitraryData):
def values(self):
return Integer.u32_from(self.data)
+
def gather_references(self, context, valid_range):
self.references = [ addr for addr in self.values() if addr in valid_range ]
diff --git a/tools/libdol2asm/data/symbol.py b/tools/libdol2asm/data/symbol.py
index 5a708fb361..9a75d16bb6 100644
--- a/tools/libdol2asm/data/symbol.py
+++ b/tools/libdol2asm/data/symbol.py
@@ -139,6 +139,12 @@ class Symbol:
f"invalid reference addr 0x{addr:08X} for {type(self).__name__}\n{self}")
return f"&{self.identifier.label}"
+ def cpp_load(self, accessor, addr):
+ if addr != self.addr:
+ raise Dol2ZelException(
+ f"invalid reference addr 0x{addr:08X} for {type(self).__name__}\n{self}")
+ return f"{self.identifier.label}"
+
def asm_reference(self, addr):
if addr != self.addr:
return None
@@ -147,6 +153,9 @@ class Symbol:
def gather_references(self, context, valid_range):
pass
+ def get_reference_information(self, context, symbol_table):
+ pass
+
def types(self):
return set()
diff --git a/tools/libdol2asm/generate_functions.py b/tools/libdol2asm/generate_functions.py
index 8df6a6c8e9..90fd2354e7 100644
--- a/tools/libdol2asm/generate_functions.py
+++ b/tools/libdol2asm/generate_functions.py
@@ -1,6 +1,10 @@
import struct
+from capstone import *
+from capstone.ppc import *
+
from . import linker_map
+from . import disassemble
from .types import *
from .data import *
@@ -159,6 +163,61 @@ def is_load_global_function(data: bytearray) -> Tuple[bool, int, str, int]:
return False, None, None, None
+# TODO: @!game move
+R2_ADDR = 0x80459A00
+R13_ADDR = 0x80458580
+
+RETURN_SYMBOL_LOAD_INSTS = {
+ PPC_INS_LWZ,
+ PPC_INS_LHZ,
+ PPC_INS_LBZ,
+}
+
+RETURN_SYMBOL_TYPE = {
+ PPC_INS_LWZ: U32,
+ PPC_INS_LHZ: U16,
+ PPC_INS_LBZ: U8,
+}
+
+def decompile_return_symbol_function(symbol, block, insts, symbol_table) -> Function:
+ if len(insts) != 2:
+ return None
+
+ load = insts[0]
+ ret = insts[1]
+
+ if ret.id != PPC_INS_BLR:
+ return None
+ if not load.id in RETURN_SYMBOL_LOAD_INSTS:
+ return None
+
+ address = 0
+ mem_base = load.operands[1].mem.base
+ mem_disp = load.operands[1].mem.disp
+ if mem_base == PPC_REG_R2:
+ address = R2_ADDR + disassemble.sign_extend_16(mem_disp)
+ elif mem_base == PPC_REG_R13:
+ address = R13_ADDR + disassemble.sign_extend_16(mem_disp)
+ else:
+ return None
+
+ return_symbol = symbol_table[-1, address]
+ if not return_symbol:
+ return None
+
+ # structure does not work :(
+ if isinstance(return_symbol, Structure):
+ return None
+
+ return SymbolReturnFunction(
+ symbol.identifier,
+ addr=symbol.addr,
+ size=symbol.size,
+ padding=symbol.padding,
+ alignment=0,
+ return_type=RETURN_SYMBOL_TYPE[load.id],
+ symbol_addr=address)
+
def from_group(section: Section, group: List[linker_map.Symbol]) -> Function:
"""
@@ -169,35 +228,38 @@ def from_group(section: Section, group: List[linker_map.Symbol]) -> Function:
if len(group) == 1:
block = group[0]
data = section.get_data(block.start, block.end)
- if is_return_function(data):
- return [ReturnFunction(
- Identifier("func", block.start, block.name),
- addr=block.addr,
- size=block.size,
- padding=block.padding,
- alignment=0,
- return_type=VOID)]
-
- if is_return_integer_function(data):
- integer_value = get_short_value(data)
- if integer_value == 0:
- value = "false"
- type = BOOL
- elif integer_value == 1:
- value = "true"
- type = BOOL
- else:
- value = f"{integer_value}"
- type = S32
-
- return [ReturnFunction(
- Identifier("func", block.start, block.name),
- addr=block.addr,
- size=block.size,
- padding=block.padding,
- alignment=0,
- return_type=type,
- return_value=value)]
+ if len(data) >= 4 and len(data) < 16:
+ insts = list(disassemble.cs.disasm(data, block.start))
+
+ if is_return_function(data):
+ return [CustomReturnFunction(
+ Identifier("func", block.start, block.name),
+ addr=block.addr,
+ size=block.size,
+ padding=block.padding,
+ alignment=0,
+ return_type=VOID)]
+
+ if is_return_integer_function(data):
+ integer_value = get_short_value(data)
+ if integer_value == 0:
+ value = "false"
+ type = BOOL
+ elif integer_value == 1:
+ value = "true"
+ type = BOOL
+ else:
+ value = f"{integer_value}"
+ type = S32
+
+ return [CustomReturnFunction(
+ Identifier("func", block.start, block.name),
+ addr=block.addr,
+ size=block.size,
+ padding=block.padding,
+ alignment=0,
+ return_type=type,
+ return_value=value)]
first = group[0]
if first.size <= 0:
@@ -208,3 +270,44 @@ def from_group(section: Section, group: List[linker_map.Symbol]) -> Function:
# the function was not decompilable
return [ASMFunction.create(section, group)]
+
+def decompile_symbol(context, section, symbol, symbol_table, add_list, remove_list):
+ if not isinstance(symbol, ASMFunction):
+ return symbol
+ if len(symbol.blocks) > 1:
+ return symbol
+
+ block = symbol.blocks[0]
+ data = symbol.data[block.start-symbol.start:block.end-symbol.start]
+ if len(data) <= 0 or len(data) > 16:
+ return symbol
+
+ insts = list(disassemble.cs.disasm(data, block.start))
+ function = decompile_return_symbol_function(symbol, block, insts, symbol_table)
+ if function:
+ function.set_mlts(symbol._module, symbol._library, symbol._translation_unit, symbol._section)
+ assert function.addr == symbol.addr
+ assert function.size == symbol.size
+ add_list.add(function)
+ remove_list.discard(symbol)
+ return function
+
+ #if symbol.size == 8:
+ # context.debug(f"{symbol.addr:08X} {symbol.name}")
+
+ return symbol
+
+def decompile(context, libraries, symbol_table):
+ remove_list = set()
+ add_list = set()
+ for lib in libraries:
+ for tu in lib.translation_units.values():
+ for section in tu.sections.values():
+ if section.name != ".text" and section.name != ".init":
+ continue
+ symbols = []
+ for symbol in section.symbols:
+ symbols.append(decompile_symbol(context, section, symbol, symbol_table, add_list, remove_list))
+ section.symbols = symbols
+ return add_list, remove_list
+ \ No newline at end of file
diff --git a/tools/libdol2asm/split_asm.py b/tools/libdol2asm/split_asm.py
index 3b09ca3a9e..48621034b5 100644
--- a/tools/libdol2asm/split_asm.py
+++ b/tools/libdol2asm/split_asm.py
@@ -300,6 +300,12 @@ class Dol2AsmSplitter:
for symbol in add_list:
self.symbol_table.add_symbol(symbol)
+ add_list, remove_list = generate_functions.decompile(self.context, libs, self.symbol_table)
+ for symbol in remove_list:
+ self.symbol_table.remove_symbol(symbol)
+ for symbol in add_list:
+ self.symbol_table.add_symbol(symbol)
+
def name_symbols(self):
print(f"{self.step_count:2} Naming")
self.step_count += 1
@@ -658,6 +664,7 @@ class Dol2AsmSplitter:
self.search_binary(cache)
+ cache = True
start_time = time.time()
cache_path = Path("build/full_cache_xx.dump")
if cache and cache_path.exists():