Merge pull request #131 from Decompollaborate/develop

1.16.2

Author: AngheloAlf

pyproject.toml

@@ -4,7 +4,7 @@
 [project]
 name = "spimdisasm"
 # Version should be synced with spimdisasm/__init__.py
-version = "1.16.0"
+version = "1.16.2"
 description = "MIPS disassembler"
 # license = "MIT"
 readme = "README.md"

spimdisasm/__init__.py

@@ -5,7 +5,7 @@
 
 from __future__ import annotations
 
-__version_info__: tuple[int, int, int] = (1, 16, 0)
+__version_info__: tuple[int, int, int] = (1, 16, 2)
 __version__ = ".".join(map(str, __version_info__))
 __author__ = "Decompollaborate"
 

spimdisasm/common/ContextSymbols.py

@@ -519,7 +519,7 @@ def getLabelMacro(self) -> str|None:
             if self.isStatic():
                 label += f"# static variable{GlobalConfig.LINE_ENDS}"
             if self.isAutogeneratedPad():
-                label += f"# Automatically generated pad{GlobalConfig.LINE_ENDS}"
+                label += f"# Automatically generated and unreferenced pad{GlobalConfig.LINE_ENDS}"
 
         currentType = self.getTypeSpecial()
         if currentType == SymbolSpecialType.jumptablelabel:

spimdisasm/common/GlobalConfig.py

@@ -244,6 +244,15 @@ def AGGRESSIVE_STRING_GUESSER(self, value: bool) -> None:
     PANIC_RANGE_CHECK: bool = False
     """Produce a fatal error if a range check fails instead of just printing a warning"""
 
+    CREATE_DATA_PADS: bool = True
+    """Create dummy and unreferenced data symbols after another symbol which has non-zero user-declared size.
+
+    The generated pad symbols may have non-zero data"""
+    CREATE_RODATA_PADS: bool = False
+    """Create dummy and unreferenced rodata symbols after another symbol which has non-zero user-declared size.
+
+    The generated pad symbols may have non-zero data"""
+
     QUIET: bool = False
     VERBOSE: bool = False
 
@@ -349,6 +358,9 @@ def addParametersToArgParse(self, parser: argparse.ArgumentParser):
 
         miscConfig.add_argument("--panic-range-check", help=f"Produce a fatal error if a range check fails instead of just printing a warning. Defaults to {self.PANIC_RANGE_CHECK}", action=Utils.BooleanOptionalAction)
 
+        miscConfig.add_argument("--create-data-pads", help=f"Create dummy and unreferenced data symbols after another symbol which has non-zero user-declared size.\nThe generated pad symbols may have non-zero data. Defaults to {self.CREATE_DATA_PADS}", action=Utils.BooleanOptionalAction)
+        miscConfig.add_argument("--create-rodata-pads", help=f"Create dummy and unreferenced rodata symbols after another symbol which has non-zero user-declared size.\nThe generated pad symbols may have non-zero data. Defaults to {self.CREATE_RODATA_PADS}", action=Utils.BooleanOptionalAction)
+
 
         verbosityConfig = parser.add_argument_group("Verbosity options")
 
@@ -517,6 +529,11 @@ def parseArgs(self, args: argparse.Namespace):
         if args.panic_range_check is not None:
             self.PANIC_RANGE_CHECK = args.panic_range_check
 
+        if args.create_data_pads is not None:
+            self.CREATE_DATA_PADS = args.create_data_pads
+        if args.create_rodata_pads is not None:
+            self.CREATE_RODATA_PADS = args.create_rodata_pads
+
 
         if args.verbose is not None:
             self.VERBOSE = args.verbose

spimdisasm/common/Relocation.py

@@ -11,6 +11,7 @@
 from .ContextSymbols import ContextSymbol
 from .FileSectionType import FileSectionType
 from .GlobalConfig import GlobalConfig
+from .GlobalConfig import Compiler
 
 
 class RelocType(enum.Enum):
@@ -149,7 +150,7 @@ class RelocationInfo:
     staticReference: RelocationStaticReference|None = None
     globalReloc: bool = False
 
-    def getName(self) -> str:
+    def getName(self, isSplittedSymbol: bool=False) -> str:
         if isinstance(self.symbol, ContextSymbol):
             name = self.symbol.getName()
         else:
@@ -158,18 +159,27 @@ def getName(self) -> str:
         if self.addend == 0:
             return name
 
-        if self.relocType == RelocType.MIPS_LO16:
-            if self.addend < -0x8000:
-                return f"{name} - (0x{-self.addend:X} & 0xFFFF)"
-            if self.addend > 0x7FFF:
-                return f"{name} + (0x{self.addend:X} & 0xFFFF)"
+        # IDO always compiles assembly using modern GAS, for both whole files
+        # and splitted symbols. Using the workaround for addends outside the
+        # range of an s16 has different behavior in modern GAS.
+        if GlobalConfig.COMPILER == Compiler.IDO:
+            if self.addend < 0:
+                return f"{name} - 0x{-self.addend:X}"
+            return f"{name} + 0x{self.addend:X}"
+
+        if isSplittedSymbol:
+            if self.relocType == RelocType.MIPS_LO16:
+                if self.addend < -0x8000:
+                    return f"{name} - (0x{-self.addend:X} & 0xFFFF)"
+                if self.addend > 0x7FFF:
+                    return f"{name} + (0x{self.addend:X} & 0xFFFF)"
 
         if self.addend < 0:
             return f"{name} - 0x{-self.addend:X}"
         return f"{name} + 0x{self.addend:X}"
 
-    def getNameWithReloc(self) -> str:
-        name = self.getName()
+    def getNameWithReloc(self, isSplittedSymbol: bool=False) -> str:
+        name = self.getName(isSplittedSymbol=isSplittedSymbol)
 
         percentRel = _percentRel.get(self.relocType)
         if percentRel is not None:
@@ -185,8 +195,8 @@ def getNameWithReloc(self) -> str:
 
         return name
 
-    def getInlineStr(self) -> str:
-        output = f"    # {self.relocType.name} '{self.getName()}'"
+    def getInlineStr(self, isSplittedSymbol: bool=False) -> str:
+        output = f"    # {self.relocType.name} '{self.getName(isSplittedSymbol=isSplittedSymbol)}'"
         if self.staticReference is not None:
             output += f" (static)"
         if self.globalReloc:

spimdisasm/common/Utils.py

@@ -222,15 +222,20 @@ def getMaybeBooleyFromMaybeStr(booley: str|None) -> bool|None:
     0x17,
     0x18,
     0x19,
-    0x1A,
+    # 0x1A, # Some games use this char as an escape sequence
     # 0x1B, # VT escape sequences
     0x1C,
     0x1D,
     0x1E,
     0x1F,
 }
 
-escapeCharactersSpecialCases = {0x1B, 0x8C, 0x8D}
+escapeCharactersSpecialCases = {
+    0x1A,
+    0x1B,
+    0x8C,
+    0x8D,
+}
 
 def decodeBytesToStrings(buf: bytes, offset: int, stringEncoding: str, terminator: int=0) -> tuple[list[str], int]:
     result = []

spimdisasm/elf32/__init__.py

@@ -28,7 +28,7 @@
 from .Elf32Rels import Elf32Rels as Elf32Rels
 from .Elf32Rels import Elf32RelEntry as Elf32RelEntry
 
-from .Elf32File import Elf32File
+from .Elf32File import Elf32File as Elf32File
 
 # To avoid breaking backwards compatibility
 from ..common.Relocation import RelocType as Elf32Relocs

spimdisasm/mips/FuncRodataEntry.py

@@ -29,7 +29,7 @@ def writeToFile(self, f: TextIO, writeFunction: bool=True):
             # Write the rdata
             f.write(f".section .rodata{common.GlobalConfig.LINE_ENDS}")
             for sym in self.rodataSyms:
-                f.write(sym.disassemble(migrate=True, useGlobalLabel=True))
+                f.write(sym.disassemble(migrate=True, useGlobalLabel=True, isSplittedSymbol=True))
                 f.write(common.GlobalConfig.LINE_ENDS)
 
         if len(self.lateRodataSyms) > 0:
@@ -48,7 +48,7 @@ def writeToFile(self, f: TextIO, writeFunction: bool=True):
                     align = 8
                 f.write(f".late_rodata_alignment {align}{common.GlobalConfig.LINE_ENDS}")
             for sym in self.lateRodataSyms:
-                f.write(sym.disassemble(migrate=True, useGlobalLabel=True))
+                f.write(sym.disassemble(migrate=True, useGlobalLabel=True, isSplittedSymbol=True))
                 f.write(common.GlobalConfig.LINE_ENDS)
 
         if self.function is not None:
@@ -57,7 +57,7 @@ def writeToFile(self, f: TextIO, writeFunction: bool=True):
 
             if writeFunction:
                 # Write the function itself
-                f.write(self.function.disassemble(migrate=self.hasRodataSyms()))
+                f.write(self.function.disassemble(migrate=self.hasRodataSyms(), isSplittedSymbol=True))
 
     @staticmethod
     def getEntryForFuncFromSection(func: symbols.SymbolFunction|None, rodataSection: sections.SectionRodata|None) -> FunctionRodataEntry:

spimdisasm/mips/MipsFileBase.py

@@ -163,7 +163,7 @@ def disassemble(self, migrate: bool=False, useGlobalLabel: bool=True) -> str:
             output += self.getSpimdisasmVersionString()
 
         for i, sym in enumerate(self.symbolList):
-            output += sym.disassemble(useGlobalLabel=useGlobalLabel)
+            output += sym.disassemble(migrate=migrate, useGlobalLabel=useGlobalLabel, isSplittedSymbol=False)
             if i + 1 < len(self.symbolList):
                 output += common.GlobalConfig.LINE_ENDS
         return output

spimdisasm/mips/sections/MipsSectionData.py

@@ -38,6 +38,14 @@ def analyze(self):
                 symbolList.append((localOffset, contextSym))
                 contextSym.isMaybeString = self._stringGuesser(contextSym, localOffset)
                 contextSym.isMaybePascalString = self._pascalStringGuesser(contextSym, localOffset)
+
+                if common.GlobalConfig.CREATE_DATA_PADS and contextSym.hasUserDeclaredSize():
+                    symDeclaredSize = contextSym.getSize()
+                    if symDeclaredSize > 0 and localOffset + symDeclaredSize < self.sizew * 4:
+                        # Try to respect the user-declared size for this symbol
+                        extraContextSym = self.addSymbol(contextSym.vram+symDeclaredSize, sectionType=self.sectionType, isAutogenerated=True, symbolVrom=currentVrom+symDeclaredSize)
+                        extraContextSym.isAutoCreatedPad = True
+
             elif self.popPointerInDataReference(currentVram) is not None:
                 contextSym = self.addSymbol(currentVram, sectionType=self.sectionType, isAutogenerated=True)
                 contextSym.isMaybeString = self._stringGuesser(contextSym, localOffset)

spimdisasm/mips/sections/MipsSectionRodata.py

@@ -101,6 +101,13 @@ def analyze(self):
                 self.symbolsVRams.add(currentVram)
                 symbolList.append((localOffset, currentVram))
 
+                if common.GlobalConfig.CREATE_RODATA_PADS and contextSym.hasUserDeclaredSize():
+                    symDeclaredSize = contextSym.getSize()
+                    if symDeclaredSize > 0 and localOffset + symDeclaredSize < self.sizew * 4:
+                        # Try to respect the user-declared size for this symbol
+                        extraContextSym = self.addSymbol(contextSym.vram+symDeclaredSize, sectionType=self.sectionType, isAutogenerated=True, symbolVrom=currentVrom+symDeclaredSize)
+                        extraContextSym.isAutoCreatedPad = True
+
             localOffset += 4
 
         previousSymbolWasLateRodata = False