courgette/assembly_program.h

   1 // Copyright (c) 2011 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #ifndef COURGETTE_ASSEMBLY_PROGRAM_H_
   6 #define COURGETTE_ASSEMBLY_PROGRAM_H_
   7
   8 #include <map>
   9 #include <set>
  10 #include <vector>
  11
  12 #include "base/basictypes.h"
  13 #include "base/memory/scoped_ptr.h"
  14
  15 #include "courgette/disassembler.h"
  16 #include "courgette/memory_allocator.h"
  17
  18 namespace courgette {
  19
  20 class EncodedProgram;
  21 class Instruction;
  22
  23 typedef NoThrowBuffer<Instruction*> InstructionVector;
  24
  25 // A Label is a symbolic reference to an address.  Unlike a conventional
  26 // assembly language, we always know the address.  The address will later be
  27 // stored in a table and the Label will be replaced with the index into the
  28 // table.
  29 //
  30 // TODO(sra): Make fields private and add setters and getters.
  31 class Label {
  32  public:
  33   static const int kNoIndex = -1;
  34   Label() : rva_(0), index_(kNoIndex) {}
  35   explicit Label(RVA rva) : rva_(rva), index_(kNoIndex) {}
  36
  37   RVA rva_;    // Address referred to by the label.
  38   int index_;  // Index of address in address table, kNoIndex until assigned.
  39 };
  40
  41 typedef std::map<RVA, Label*> RVAToLabel;
  42
  43 // An AssemblyProgram is the result of disassembling an executable file.
  44 //
  45 // * The disassembler creates labels in the AssemblyProgram and emits
  46 //   'Instructions'.
  47 // * The disassembler then calls DefaultAssignIndexes to assign
  48 //   addresses to positions in the address tables.
  49 // * [Optional step]
  50 // * At this point the AssemblyProgram can be converted into an
  51 //   EncodedProgram and serialized to an output stream.
  52 // * Later, the EncodedProgram can be deserialized and assembled into
  53 //   the original file.
  54 //
  55 // The optional step is to modify the AssemblyProgram.  One form of modification
  56 // is to assign indexes in such a way as to make the EncodedProgram for this
  57 // AssemblyProgram look more like the EncodedProgram for some other
  58 // AssemblyProgram.  The modification process should call UnassignIndexes, do
  59 // its own assignment, and then call AssignRemainingIndexes to ensure all
  60 // indexes are assigned.
  61 //
  62 class AssemblyProgram {
  63  public:
  64   AssemblyProgram();
  65   ~AssemblyProgram();
  66
  67   void set_image_base(uint64 image_base) { image_base_ = image_base; }
  68
  69   // Instructions will be assembled in the order they are emitted.
  70
  71   // Generates an entire base relocation table.
  72   CheckBool EmitPeRelocsInstruction() WARN_UNUSED_RESULT;
  73
  74   // Generates an ELF style relocation table.
  75   CheckBool EmitElfRelocationInstruction() WARN_UNUSED_RESULT;
  76
  77   // Following instruction will be assembled at address 'rva'.
  78   CheckBool EmitOriginInstruction(RVA rva) WARN_UNUSED_RESULT;
  79
  80   // Generates a single byte of data or machine instruction.
  81   CheckBool EmitByteInstruction(uint8 byte) WARN_UNUSED_RESULT;
  82
  83   // Generates 4-byte relative reference to address of 'label'.
  84   CheckBool EmitRel32(Label* label) WARN_UNUSED_RESULT;
  85
  86   // Generates 4-byte absolute reference to address of 'label'.
  87   CheckBool EmitAbs32(Label* label) WARN_UNUSED_RESULT;
  88
  89   // Looks up a label or creates a new one.  Might return NULL.
  90   Label* FindOrMakeAbs32Label(RVA rva);
  91
  92   // Looks up a label or creates a new one.  Might return NULL.
  93   Label* FindOrMakeRel32Label(RVA rva);
  94
  95   void DefaultAssignIndexes();
  96   void UnassignIndexes();
  97   void AssignRemainingIndexes();
  98
  99   EncodedProgram* Encode() const;
 100
 101   // Accessor for instruction list.
 102   const InstructionVector& instructions() const {
 103     return instructions_;
 104   }
 105
 106   // Returns the label if the instruction contains and absolute address,
 107   // otherwise returns NULL.
 108   Label* InstructionAbs32Label(const Instruction* instruction) const;
 109
 110   // Returns the label if the instruction contains and rel32 offset,
 111   // otherwise returns NULL.
 112   Label* InstructionRel32Label(const Instruction* instruction) const;
 113
 114  private:
 115   CheckBool Emit(Instruction* instruction) WARN_UNUSED_RESULT;
 116
 117   // Looks up a label or creates a new one.  Might return NULL.
 118   Label* FindLabel(RVA rva, RVAToLabel* labels);
 119
 120   // Helper methods for the public versions.
 121   static void UnassignIndexes(RVAToLabel* labels);
 122   static void DefaultAssignIndexes(RVAToLabel* labels);
 123   static void AssignRemainingIndexes(RVAToLabel* labels);
 124
 125   // Sharing instructions that emit a single byte saves a lot of space.
 126   Instruction* GetByteInstruction(uint8 byte);
 127   scoped_array<Instruction*> byte_instruction_cache_;
 128
 129   uint64 image_base_;  // Desired or mandated base address of image.
 130
 131   InstructionVector instructions_;  // All the instructions in program.
 132
 133   // These are lookup maps to find the label associated with a given address.
 134   // We have separate label spaces for addresses referenced by rel32 labels and
 135   // abs32 labels.  This is somewhat arbitrary.
 136   RVAToLabel rel32_labels_;
 137   RVAToLabel abs32_labels_;
 138
 139   DISALLOW_COPY_AND_ASSIGN(AssemblyProgram);
 140 };
 141
 142 }  // namespace courgette
 143 #endif  // COURGETTE_ASSEMBLY_PROGRAM_H_