Merge pull request #130 from Decompollaborate/develop

1.16.0

pyproject.toml

@@ -4,7 +4,7 @@
 [project]
 name = "spimdisasm"
 # Version should be synced with spimdisasm/__init__.py
-version = "1.15.4"
+version = "1.16.0"
 description = "MIPS disassembler"
 # license = "MIT"
 readme = "README.md"

spimdisasm/__init__.py

@@ -5,7 +5,7 @@
 
 from __future__ import annotations
 
-__version_info__ = (1, 15, 4)
+__version_info__: tuple[int, int, int] = (1, 16, 0)
 __version__ = ".".join(map(str, __version_info__))
 __author__ = "Decompollaborate"
 

spimdisasm/common/GlobalConfig.py

@@ -161,6 +161,10 @@ def AGGRESSIVE_STRING_GUESSER(self, value: bool) -> None:
 
     COMPILER: Compiler = Compiler.IDO
 
+    DETECT_REDUNDANT_FUNCTION_END: bool = False
+    """Tries to detect redundant and unreferenced functions ends and merge them together.
+    This option is ignored if the compiler is not set to IDO"""
+
     ENDIAN: InputEndian = InputEndian.BIG
     """Endian for input binary files"""
     ENDIAN_DATA: InputEndian|None = None
@@ -294,6 +298,7 @@ def addParametersToArgParse(self, parser: argparse.ArgumentParser):
         backendConfig.add_argument("--custom-suffix", help="Set a custom suffix for automatically generated symbols")
 
         backendConfig.add_argument("--compiler", help=f"Enables some tweaks for the selected compiler. Defaults to {self.COMPILER.name}", choices=compilerOptions)
+        backendConfig.add_argument("--detect-redundant-function-end", help=f"Tries to detect redundant and unreferenced function ends (jr $ra; nop), and merge it into the previous function. Currently it only is applied when the compiler is set to IDO. Defaults to {self.DETECT_REDUNDANT_FUNCTION_END}", action=Utils.BooleanOptionalAction)
 
         backendConfig.add_argument("--endian", help=f"Set the endianness of input files. Defaults to {self.ENDIAN.name.lower()}", choices=["big", "little", "middle"], default=self.ENDIAN.name.lower())
 
@@ -428,6 +433,9 @@ def parseArgs(self, args: argparse.Namespace):
         if args.compiler is not None:
             self.COMPILER = Compiler.fromStr(args.compiler)
 
+        if args.detect_redundant_function_end is not None:
+            self.DETECT_REDUNDANT_FUNCTION_END = args.detect_redundant_function_end
+
         if args.endian is not None:
             self.ENDIAN = InputEndian.fromStr(args.endian)
 

spimdisasm/mips/MipsFileBase.py

@@ -65,7 +65,10 @@ def checkAndCreateFirstSymbol(self) -> None:
         "Check if the very start of the file has a symbol and create it if it doesn't exist yet"
 
         currentVram = self.getVramOffset(0)
-        currentVrom = self.getVromOffset(0)
+        if self.sectionType == common.FileSectionType.Bss:
+            currentVrom = None
+        else:
+            currentVrom = self.getVromOffset(0)
         contextSym = self.getSymbol(currentVram, vromAddress=currentVrom, tryPlusOffset=False)
         if contextSym is None:
             contextSym = self.addSymbol(currentVram, sectionType=self.sectionType, isAutogenerated=True, symbolVrom=currentVrom)

spimdisasm/mips/sections/MipsSectionText.py

@@ -20,6 +20,7 @@ def __init__(self, context: common.Context, vromStart: int, vromEnd: int, vram:
         super().__init__(context, vromStart, vromEnd, vram, filename, common.Utils.bytesToWords(array_of_bytes, vromStart, vromEnd), common.FileSectionType.Text, segmentVromStart, overlayCategory)
 
         self.instrCat: rabbitizer.Enum = rabbitizer.InstrCategory.CPU
+        self.detectRedundantFunctionEnd: bool|None = None
 
 
     @property
@@ -40,6 +41,14 @@ def wordListToInstructions(wordList: list[int], currentVram: int|None, instrCat:
         return instrsList
 
 
+    def tryDetectRedundantFunctionEnd(self) -> bool:
+        if common.GlobalConfig.COMPILER != common.Compiler.IDO:
+            return False
+
+        if self.detectRedundantFunctionEnd is None:
+            return common.GlobalConfig.DETECT_REDUNDANT_FUNCTION_END
+        return self.detectRedundantFunctionEnd
+
     def _findFunctions(self, instrsList: list[rabbitizer.Instruction]):
         if len(instrsList) == 0:
             return [0], [False]
@@ -164,10 +173,35 @@ def _findFunctions(self, instrsList: list[rabbitizer.Instruction]):
                 if instructionOffset + 8 == currentInstructionStart + currentFunctionSym.getSize():
                     functionEnded = True
             else:
-                if not (farthestBranch > 0) and instr.isJump():
+                funcSymbol = self.getSymbol(currentVram + 8, vromAddress=currentVrom + 8, tryPlusOffset=False, checkGlobalSegment=False)
+                # If there's another function after this then the current function has ended
+                if funcSymbol is not None and funcSymbol.isTrustableFunction(self.instrCat == rabbitizer.InstrCategory.RSP):
+                    if funcSymbol.vromAddress is None or currentVrom + 8 == funcSymbol.vromAddress:
+                        functionEnded = True
+
+                if not functionEnded and not (farthestBranch > 0) and instr.isJump():
                     if instr.isReturn():
                         # Found a jr $ra and there are no branches outside of this function
-                        functionEnded = True
+                        if self.tryDetectRedundantFunctionEnd():
+                            # IDO -g, -g1 and -g2 can generate a redundant and unused `jr $ra; nop`. In normal conditions this would be detected
+                            # as its own separate empty function, which would cause issues on a decompilation project.
+                            # In other words, we try to detect the following pattern, and the last two instructions not being a function
+                            # already referenced or user-declared.
+                            # jr         $ra
+                            #  nop
+                            # jr         $ra
+                            #  nop
+                            redundantPatternDetected = False
+                            if index + 3 < nInstr:
+                                instr1 = instrsList[index+1]
+                                instr2 = instrsList[index+2]
+                                instr3 = instrsList[index+3]
+                                if funcSymbol is None and instr1.isNop() and instr2.isReturn() and instr3.isNop():
+                                    redundantPatternDetected = True
+                            if not redundantPatternDetected:
+                                functionEnded = True
+                        else:
+                            functionEnded = True
                     elif instr.isJumptableJump():
                         # Usually jumptables, ignore
                         pass
@@ -176,12 +210,6 @@ def _findFunctions(self, instrsList: list[rabbitizer.Instruction]):
                             # I don't remember the reasoning of this condition...
                             functionEnded = True
 
-                # If there's another function after this then the current function has ended
-                funcSymbol = self.getSymbol(currentVram + 8, vromAddress=currentVrom + 8, tryPlusOffset=False, checkGlobalSegment=False)
-                if funcSymbol is not None and funcSymbol.isTrustableFunction(self.instrCat == rabbitizer.InstrCategory.RSP):
-                    if funcSymbol.vromAddress is None or currentVrom + 8 == funcSymbol.vromAddress:
-                        functionEnded = True
-
             index += 1
             farthestBranch -= 4
             instructionOffset += 4