Interface INativeCodeAnalyzerExtension<InsnType extends IInstruction>

Type Parameters:
InsnType - instruction type analyzed by this extension
All Known Subinterfaces:
INativeCodeAnalyzerExtensionsManager<InsnType>
All Known Implementing Classes:
AbstractAnalyzerExtension

@Ser public interface INativeCodeAnalyzerExtension<InsnType extends IInstruction>
Definition of analyzer extensions. A collection of heuristics to improve the generic code analyzer. Clients should inherit from AbstractAnalyzerExtension, which provides default implementations, and implement only the needed methods.
  • Method Details

    • initialize

      void initialize(INativeCodeAnalyzer<InsnType> analyzer)
      Implementors will receive a reference to the main analyzer. This method is not to be called by third-party code. It will be called by the GCA when the extension object (this object) is being initialized.
      Parameters:
      analyzer - analyzer using this extension
    • getUnit

      Get the code unit managing INativeCodeAnalyzer. This method allows extensions to easily retrieve properties of the managing code unit.

      Important: no guarantee is given on the unit's state; this method should be used preferably to retrieve static features (eg, unit's properties).

      Returns:
      code unit
    • getPrimitiveSizes

      IPrimitiveSizes getPrimitiveSizes(SubsystemType subsystemType, CompilerType compilerType)
      This method is called by the code unit to retrieve primitive type information.
      Parameters:
      subsystemType - optional hint
      compilerType - optional hint
      Returns:
      if null is returned, the code unit will attempt to generate the most appropriate primitive type sizes
    • typeManagerInitialized

      void typeManagerInitialized(ITypeManager typeman)
      This method is called by the code unit right after it created a type manager suitable for processing. This happens before the code analysis starts. Implementors are free to customize the type manager, e.g. by adding types that will be required when further processing takes place.
      Parameters:
      typeman - initialized type manager
    • skipSymbolProcessing

      boolean skipSymbolProcessing()
      This method is called by the code unit to determine whether the symbols provided by the container unit (e.g. ELF, PE, etc. if there is one) should be processed. Processing symbols may enqueue code pointers and data pointers for subsequent code analysis.
      Returns:
      true if symbol processing should be skipped
    • preprocessImage

      ChainedOperationResult<Boolean> preprocessImage(int passIndex)
      This method is called before an analysis pass is performed.
      Parameters:
      passIndex - 0-based analysis pass index
      Returns:
      true if some processing
    • postprocessImage

      ChainedOperationResult<Boolean> postprocessImage(int passIndex)
      This method is called after the standard analysis pass completes (i.e., before any optional advanced analysis takes place).
      Parameters:
      passIndex - 0-based analysis pass index
      Returns:
      true if some processing
    • sigMatchingPostProcess

      ChainedOperationResult<Boolean> sigMatchingPostProcess(int passIndex)
      This method is called after signatures matching was performed.
      Parameters:
      passIndex - 0-based analysis pass index
      Returns:
      true if some processing
    • getTrampolineTarget

      ChainedOperationResult<? extends Pointer> getTrampolineTarget(CFG<InsnType> routine)
      Get the location targeted by a trampoline routine -- a routine that only branches to another location --, if any.

      Note: the target might be indirectly retrieved (i.e. through a memory/register indirection).

      Parameters:
      routine - candidate trampoline routine
      Returns:
      a Pointer in ChainedOperationResult.getResult() for the target, null if none (i.e. routine is not a trampoline, or target could not be retrieved) (the return value is not the target itself, but a pointer to the target)
    • getPrologueLooking

      ChainedOperationResult<CodePointer> getPrologueLooking(long address, long addressMax)
      Determine if a given memory area looks like (could be) the beginning of a routine.
      Parameters:
      address - address to be examined (inclusive)
      addressMax - end address to be examined (exclusive)
      Returns:
      an CodePointer in ChainedOperationResult.getResult() if a prologue was found, possibly at a different address than address parameter, null if no prologue was found.
    • verifyGapRoutineCandidate

      ChainedOperationResult<Boolean> verifyGapRoutineCandidate(long address)
      Determine whether a candidate routine in a gap should be considered as valid.
      Parameters:
      address - routine start address
      Returns:
      true or false (no nulls)
    • getPossiblePaddingSize

      ChainedOperationResult<Long> getPossiblePaddingSize(long address, long addressMax)
      Determine if a given memory area looks like (could be) starting with padding, and provides the size of the padding looking area, if any.
      Parameters:
      address - address to be examined (inclusive)
      addressMax - end address to be examined (exclusive)
      Returns:
      a Long in ChainedOperationResult.getResult(); the size of the padding-like area from address, 0L if it does not look like padding
    • getProbableEntryPoints

      ChainedOperationResult<List<CodePointer>> getProbableEntryPoints(long address, long addressMax)
      Get the likely entry points in the given memory area. It might provide several entry points (with different modes) ordered by their likelihood.

      This method is not guaranteed to give the correct result, and should be considered an heuristic. It is intended to be used before the actual disassembly of code; it may provide a hint on the entry point to disassemble with.

      Parameters:
      address - start address, inclusive
      addressMax - end address, exclusive
      Returns:
      a list of Integer representing entry points in ChainedOperationResult.getResult(), ordered from the most probable to the least probable. The list will be empty if none could be determined
    • determineRoutineStackPointerDelta

      ChainedOperationResult<Integer> determineRoutineStackPointerDelta(CFG<InsnType> routine)
      Determine (heuristically) the stack pointer delta generated by the execution of a routine.
      Parameters:
      routine - routine CFG
      Returns:
      an Integer in ChainedOperationResult.getResult(), null if the routine failed to determine the stack pointer delta
    • determinePotentialPointers

      ChainedOperationResult<Boolean> determinePotentialPointers(long address, InsnType insn, List<Pointer> values)
      Determine pointers (addresses to data or code) contained in the provided instruction, to let the code analyzer enqueue them for additional analysis. This method is called by the INativeCodeAnalyzer on non breaking flow instructions before it tries to determine such additions itself.
      Parameters:
      address - address of the instruction
      insn - instruction to be examined
      values - output collection of pointers
      Returns:
      a Boolean in ChainedOperationResult.getResult(); false to let the code analyzer proceed with its own (generic) pointer discovery algorithm; true to instruct it to skip it
    • determinePotentialPointersInProtoBlock

      ChainedOperationResult<Boolean> determinePotentialPointersInProtoBlock(IBasicBlockSkeleton<InsnType> blk, List<PointerLocation> values)
      Determine additional pointers to be analyzed (addresses to data or code) that may be inferred from the analysis of the provided basic block skeleton (proto-basic block). This method is called by the INativeCodeAnalyzer after a basic block is preliminary terminated.
      Parameters:
      blk - preliminary basic block to be examined
      values - output collection of pointers
      Returns:
      N/A
    • shouldForceRoutineEnd

      ChainedOperationResult<Boolean> shouldForceRoutineEnd(long address, InsnType insn)
      Determine if the address is the last instruction of a routine. If true, the basic block containing the provided instruction will be closed, i.e. it will be as if the instruction breaks the flow to nothing.

      When working with delay slots, this method will indicate routine end for the jump instruction that precedes the delay slot(s), not for the delay slot.

      Parameters:
      address - instruction address
      insn - parsed instruction at the provided address
      Returns:
      a Boolean in ChainedOperationResult.getResult(); true if the address should be considered a routine termination, false otherwise
    • isNonReturningRoutine

      ChainedOperationResult<Boolean> isNonReturningRoutine(INativeMethodItem routine)
      Determine if the given routine is non-returning.
      Parameters:
      routine - routine item
      Returns:
      a Boolean in ChainedOperationResult.getResult(); true if the routine does not return, false otherwise, null when no determination could be made
    • getPreferredAdvancedAnalysisStage

      ChainedOperationResult<NativeDecompilationStage> getPreferredAdvancedAnalysisStage(INativeMethodItem routine)
      Determine which NativeDecompilationStage should be used by INativeCodeAdvancedAnalyzer for the given routine.

      The default implementation returns NativeDecompilationStage.SIMULATION

      Parameters:
      routine - routine item
      Returns:
      NativeDecompilationStage in ChainedOperationResult.getResult()
    • isCandidateSwitchDispatcher

      ChainedOperationResult<Boolean> isCandidateSwitchDispatcher(long branchInstructionAddress, InsnType branchInstruction, IBasicBlockSkeleton<InsnType> branchInstructionBlock)
      Determine (heuristically) if the provided branching instruction (jump/call/...) could be the dispatcher of a switch-like statement.

      This method is intended to be a fast check, and candidates will then be processed by determineSwitchInformation(long, IBasicBlockSkeleton, List).

      Parameters:
      branchInstructionAddress - branch instruction address
      branchInstruction - branch instruction
      branchInstructionBlock - basic block containing the branch instruction
      Returns:
      a boolean in ChainedOperationResult.getResult(); true if the instruction might correspond to a switch-like statement, false otherwise
    • determineSwitchInformation

      ChainedOperationResult<SwitchInformation> determineSwitchInformation(long branchInstructionAddress, IBasicBlockSkeleton<InsnType> branchInstructionBlock, List<IBasicBlockSkeleton<InsnType>> knownBlocks)
      Determine (heuristically) the information related to a switch-like statement, i.e. the list of reachable CodePointer (case handlers) and the jump table addresses (if any).

      This method works on not-yet-built routines, i.e. using IBasicBlockSkeleton.

      Parameters:
      branchInstructionAddress - branch instruction address
      branchInstructionBlock - basic block containing the branch instruction
      knownBlocks - known proto-basic blocks
      Returns:
      SwitchInformation in ChainedOperationResult.getResult() (might be empty), null if the routine failed
    • getPreferredBreakingFlow

      ChainedOperationResult<IFlowInformation> getPreferredBreakingFlow(long address, InsnType insn)
      Provide alternate/custom breaking-flow information. If no custom information is to be provided, the implementation should return a null object. Upon a null return, it is the responsibility of the caller to retrieve the standard breaking-flow information (IInstruction.getBreakingFlow(long)).
      Parameters:
      address - instruction address
      insn - parsed instruction
      Returns:
      a custom flow-information object or null
    • customizeInstructionItem

      ChainedOperationResult<Boolean> customizeInstructionItem(INativeInstructionItem item)
      Customize an instruction item just after its creation by the code analyzer.
      Parameters:
      item - instruction item
      Returns:
      true if the item was customized