Table of Contents for
The IDA Pro Book, 2nd Edition

Version ebook / Retour

Cover image for bash Cookbook, 2nd Edition The IDA Pro Book, 2nd Edition by Chris Eagle Published by No Starch Press, 2011
  1. Cover
  2. The IDA Pro Book
  3. PRAISE FOR THE FIRST EDITION OF THE IDA PRO BOOK
  4. Acknowledgments
  5. Introduction
  6. I. Introduction to IDA
  7. 1. Introduction to Disassembly
  8. The What of Disassembly
  9. The Why of Disassembly
  10. The How of Disassembly
  11. Summary
  12. 2. Reversing and Disassembly Tools
  13. Summary Tools
  14. Deep Inspection Tools
  15. Summary
  16. 3. IDA Pro Background
  17. Obtaining IDA Pro
  18. IDA Support Resources
  19. Your IDA Installation
  20. Thoughts on IDA’s User Interface
  21. Summary
  22. II. Basic IDA Usage
  23. 4. Getting Started with IDA
  24. IDA Database Files
  25. Introduction to the IDA Desktop
  26. Desktop Behavior During Initial Analysis
  27. IDA Desktop Tips and Tricks
  28. Reporting Bugs
  29. Summary
  30. 5. IDA Data Displays
  31. Secondary IDA Displays
  32. Tertiary IDA Displays
  33. Summary
  34. 6. Disassembly Navigation
  35. Stack Frames
  36. Searching the Database
  37. Summary
  38. 7. Disassembly Manipulation
  39. Commenting in IDA
  40. Basic Code Transformations
  41. Basic Data Transformations
  42. Summary
  43. 8. Datatypes and Data Structures
  44. Creating IDA Structures
  45. Using Structure Templates
  46. Importing New Structures
  47. Using Standard Structures
  48. IDA TIL Files
  49. C++ Reversing Primer
  50. Summary
  51. 9. Cross-References and Graphing
  52. IDA Graphing
  53. Summary
  54. 10. The Many Faces of IDA
  55. Using IDA’s Batch Mode
  56. Summary
  57. III. Advanced IDA Usage
  58. 11. Customizing IDA
  59. Additional IDA Configuration Options
  60. Summary
  61. 12. Library Recognition Using FLIRT Signatures
  62. Applying FLIRT Signatures
  63. Creating FLIRT Signature Files
  64. Summary
  65. 13. Extending IDA’s Knowledge
  66. Augmenting Predefined Comments with loadint
  67. Summary
  68. 14. Patching Binaries and Other IDA Limitations
  69. IDA Output Files and Patch Generation
  70. Summary
  71. IV. Extending IDA’s Capabilities
  72. 15. IDA Scripting
  73. The IDC Language
  74. Associating IDC Scripts with Hotkeys
  75. Useful IDC Functions
  76. IDC Scripting Examples
  77. IDAPython
  78. IDAPython Scripting Examples
  79. Summary
  80. 16. The IDA Software Development Kit
  81. The IDA Application Programming Interface
  82. Summary
  83. 17. The IDA Plug-in Architecture
  84. Building Your Plug-ins
  85. Installing Plug-ins
  86. Configuring Plug-ins
  87. Extending IDC
  88. Plug-in User Interface Options
  89. Scripted Plug-ins
  90. Summary
  91. 18. Binary Files and IDA Loader Modules
  92. Manually Loading a Windows PE File
  93. IDA Loader Modules
  94. Writing an IDA Loader Using the SDK
  95. Alternative Loader Strategies
  96. Writing a Scripted Loader
  97. Summary
  98. 19. IDA Processor Modules
  99. The Python Interpreter
  100. Writing a Processor Module Using the SDK
  101. Building Processor Modules
  102. Customizing Existing Processors
  103. Processor Module Architecture
  104. Scripting a Processor Module
  105. Summary
  106. V. Real-World Applications
  107. 20. Compiler Personalities
  108. RTTI Implementations
  109. Locating main
  110. Debug vs. Release Binaries
  111. Alternative Calling Conventions
  112. Summary
  113. 21. Obfuscated Code Analysis
  114. Anti–Dynamic Analysis Techniques
  115. Static De-obfuscation of Binaries Using IDA
  116. Virtual Machine-Based Obfuscation
  117. Summary
  118. 22. Vulnerability Analysis
  119. After-the-Fact Vulnerability Discovery with IDA
  120. IDA and the Exploit-Development Process
  121. Analyzing Shellcode
  122. Summary
  123. 23. Real-World IDA Plug-ins
  124. IDAPython
  125. collabREate
  126. ida-x86emu
  127. Class Informer
  128. MyNav
  129. IdaPdf
  130. Summary
  131. VI. The IDA Debugger
  132. 24. The IDA Debugger
  133. Basic Debugger Displays
  134. Process Control
  135. Automating Debugger Tasks
  136. Summary
  137. 25. Disassembler/Debugger Integration
  138. IDA Databases and the IDA Debugger
  139. Debugging Obfuscated Code
  140. IdaStealth
  141. Dealing with Exceptions
  142. Summary
  143. 26. Additional Debugger Features
  144. Debugging with Bochs
  145. Appcall
  146. Summary
  147. A. Using IDA Freeware 5.0
  148. Using IDA Freeware
  149. B. IDC/SDK Cross-Reference
  150. Index
  151. About the Author

Alternative Calling Conventions

In Chapter 6, we discussed the most common calling conventions utilized in C and C++ code. While adherence to a published calling convention is crucial when attempting to interface one compiled module to another, nothing prohibits the use of custom calling conventions by functions within a single module. This is commonly seen in highly optimized functions that are not designed to be called from outside the module in which they reside.

The following code represents the first four lines of a function that uses a nonstandard calling convention:

.text:000158AC sub_158AC       proc near
  .text:000158AC
 .text:000158AC arg_0           = dword ptr  4
  .text:000158AC
  .text:000158AC                 push    [esp+arg_0]
  .text:000158B0              mov     edx, [eax+118h]
  .text:000158B6                 push    eax
  .text:000158B7              movzx   ecx, cl
  .text:000158BA                 mov     cl, [edx+ecx+0A0h]

According to IDA’s analysis, only one argument exists in the function’s stack frame. However, upon closer inspection of the code, you can see that both the EAX register and the CL register are used without any initialization taking place within the function. The only possible conclusion is that both EAX and CL are expected to be initialized by the caller. Therefore, you should view this function as a three-argument function rather than a single-argument function, and you must take special care when calling it to ensure that the three arguments are all in their proper places.

IDA allows you to specify custom calling conventions for any function by setting the function’s “type.” This is done by entering the function’s prototype via the Edit ▸ Functions ▸ Set function type menu option and using IDA’s __usercall calling convention. Figure 20-1 shows the resulting dialog used to set the type for sub_158AC in the preceding example.

Designating a function as __usercall

Figure 20-1. Designating a function as __usercall

For clarity, the declaration is shown again here:

int __usercall sub_158AC<eax>(struc_1 *<eax>, unsigned __int8 index<cl>, int)

Here the IDA keyword __usercall is used in place of one of the standard calling conventions such as __cdecl or __stdcall. The use of __usercall requires us to tell IDA the name of the register used to hold the function’s return value by appending the register name to the name of the function (yielding sub_158AC<eax> in this case). If the function returns no value, the return register may be omitted. Within the parameter list, each register-based parameter must also be annotated by appending the corresponding register name to the parameter’s data type. After the function’s type has been set, IDA propagates parameter information to calling functions, which results in improved commenting of function call sequences as shown in the following listing:

.text:00014B9F       lea     eax, [ebp+var_218] ; struc_1 *
.text:00014BA5       mov     cl, 1           ; index
.text:00014BA7       push    edx             ; int
.text:00014BA8        call    sub_158AC

Here it is clear that IDA recognizes that EAX will hold the first argument to the function , CL will hold the second argument , and the third argument will placed on the stack .

To demonstrate that calling conventions can vary widely even with a single executable, a second example using a custom calling convention is taken from the same binary file and shown here:

.text:0001669E sub_1669E       proc near
  .text:0001669E
 .text:0001669E arg_0           = byte ptr  4
  .text:0001669E
  .text:0001669E              mov     eax, [esi+18h]
  .text:000166A1                 add     eax, 684h
  .text:000166A6                 cmp     [esp+arg_0], 0

Here again, IDA has indicated that the function accesses only one argument within the stack frame. Closer inspection makes it quite clear that the ESI register is also expected to be initialized prior to calling this function. This example demonstrates that even with the same binary file, the registers chosen to hold register-based arguments may vary from function to function.

The lesson to be learned here is to make certain that you understand how each register used in a function is initialized. If a function makes use of a register prior to initializing that register, then the register is being used to pass a parameter. Please refer to Chapter 6 for a review of which registers are used by various compilers and common calling conventions.