English English  Deutsch Deutsch

home  products  pe explorer  feature tour

PE Explorer Disassembler

Win32 PE Disassembler: Dig into Executables

The PE Explorer Disassembler is designed to be easy to use compared with other disassemblers. To that end, some of the functionality found in other products has been left out in order to keep the process simple and fast. While as powerful as the more expensive, dedicated disassemblers, PE Explorer focuses on ease of use, clarity and navigation.

The main Disassembler window

Name List to the right provides a list of labeled addresses (including conditional and unconditional branching destinations, function prologues, named data, and string references) by the disassembler, with the entry point clearly indicated. Labels can be renamed by pressing N (Edit | Rename Label).

The Strings tab provides a list of detected strings; you can further manipulate strings detection by using the toolbar, using menu items (Edit | Mark as String/Pascal String/Long Pascal String/Unicode), or pressing SAL, or U to activate each of them.

The lower left tabs View 1View 2View 3, and View 4 (F6F7F8, and F9) provide persistent disassemble views that are independent of the main view and are swappable.

Code can be manually marked in the assembly listing by pressing 'C.' Dwords and offsets can be marked by pressing D and O, respectively.

Comments can be entered by pressing ;.

The Unprocessed data tab displays some blocks of data that do not have a reference to a procedure.

The main disassembly view is towards the top-left. A nice feature in this view is the provision for an immediate adjustment of the space between each assembly line (Ins and Del) and the number of opcodes per line (Shift + Ins and Shift + Del).

Navigation is really simple. Branching addresses can be navigated by selecting the relevant line and pressing Enter. For instructions with a second operand destination address, press Ctrl + Enter. Going back to a previous address requires pressing Esc, and to visit a particular address, you have press Ctrl + G and type the address in the hexadecimal format.

Subroutines that might have references can be listed in a pop-up window by selecting the starting address of the procedure and pressing R (Search | References). The list can then be traversed by double-clicking on each listed address.

We tried to achieve most of the power of IDA Pro, while requiring less skill or knowledge, by automating more of the disassembly process. We just made a good disassembler at a reasonable price. If your daily work involves reverse engineering of software and exploiting code, source code reviews, testing and evaluation of vulnerabilities, PE Explorer will save you hours of time and it's easy to use!

The PE Explorer disassembler assumes that some manual editing of the reproduced code will be needed. To facilitate additional hand coding, however, the disassembler utilizes a qualitative algorithm designed to reconstruct the assembly language source code of target binary win32 PE files (EXE, DLL, OCX) with the highest degree of accuracy possible.

I've been using the PE Explorer for a while, and am very impressed with the latest version's functions -- especially the Disassembler.
Conrad Herrmann,
Zone Labs, Inc.

One of the reasons that I bought PE Explorer was the Disassembler. Good stuff.

Gerald Beuchelt,
Sun Microsystems, Inc.

After using your tool vs. other shareware products I felt that I was getting the best deal for my $ instead of paying $400 for Ida Pro.
David Burlingame,
Intel Corporation

More user testimonials

The Disassembly Process

The Disassembler (Ctrl + MTools | Disassembler) opens in its own window and overlaps the main interface, which can be toggled back anytime. The disassembler supports the most common Intel x86 instruction sets and extensions (MMX, 3D Now!, SSE, SSE2 and SSE3).

Before the disassembly process, the Options window (View | Disassembler Options) displays the following options:

Options window

The checked Auto Rescan option and Auto Rescan count value are fine at default values, but for complicated binaries, they may require more passes. The number of displayed opcodes can be set to a default value.

Once you pressed Start Now, the disassembly process begins by identifying the compiler used to build the target file. Forehand knowledge of how a compiler puts files together improves the guesswork involved in determining the data allocation patterns within the target file. Moreover, given this information, identifying most of the objects, procedures, variables, types etc. of the target file can be achieved with a very high degree of accuracy.

Only various Borland compilers are currently identified. The disassembler will decompile files built with other compilers too. At this time, however, it will only display specifically identified internal items for files compiled with Borland/CodeGear compilers. During the disassembly process the Processing Info window displays the following information:

Processing Info window

The fundamental challenge in disassembling compiled files is to correctly interpret the examined data. Separating code from data can be extremely difficult, especially when the initial code includes countermeasures intended to hinder disassembly. In order to meet this challenge, we developed a customized model, specifically designed to reduce incorrect data type identifications.

Although the customized modeling performed by the PE Explorer Disassembler does increase processing time, the result is a dramatic reduction of incorrect opcode translations. We think you will agree that that the extra time needed to achieve this high level of accuracy is justly compensated for by the time saved when hand correcting the output.

After the compiler has been identified, the disassembler searches the target file for a relocation table. If a relocation table is found, the information is used to detect the absolute offsets to the various content items in the file (for example: mov eax, offset L0041F46A, jmp L004A49FE). In turn, this information is used to form the Jumps list and the Rets list. The Jumps list stores the addresses to positively identified instructions and the Rets list stores the addresses to tentatively identified instructions. These addresses are then arranged according to the degree of certainty regarding their identification.

As the information from the Jumps list and the Rets list accumulates, the decoding process begins to emulate the execution of the code found within the target file. When the primary branch of the decoding process finishes executing, the remaining information from the Jumps list and the Rets list is processed. Information accumulated in the Jumps list is evaluated first, followed by the information in the Rets list, until both lists are empty. If the repeated scanning option is selected, this step is repeated until the disassembler has resolved any outstanding references.

The decoding process concludes by generating labels for any items that remain unidentified and setting the output positions for all of the various items found in the target file.

After all processing has been completed, the disassembler displays the resulting source code for the target file. This output can be manually edited or saved to disk for future reference.

Executable File Forensics: Search for Text Strings within an EXE

Searching through the strings within an executable file could be a way to get hints about the functionality of a program. With PE Explorer, you can try to fish out the strings that are present in the program, like messages, something connecting to a URL, or copies of file that might be present at a specific location, etc. When you are looking for leads in the form of text, the output of strings found in the executable gives you a good knowledge of what some of the functions and subroutines called by this binary are.

The disassembler pulls ASCII text strings out the data portion of the file. Unlike the various strings utilities that search and extract the text strings from a file, PE Explorer is much more accurate and detailed in extracting valid strings out from specified memory locations instead of searching.

Finding ASCII text strings in EXE

Search for Delphi VCL Obiects

The Delphi VCL object model is designed in such a way that we think it will be possible to reproduce the original assembly language source code perfectly. At Heaventools, improving the disassembler is an ongoing part of our PE Explorer development efforts.

Borland VCL objects

Known Limitations

PE Explorer does not decompile code. It disassembles code, which is the task of converting machine code into assembler, but it does not generate C or C++ code from the disassembled output. Which is a task of great difficulty. Besides, in most cases, there is nothing in the executable that indicates what programming language was used to generate the program.

Disassembling files larger than 1 Mb in size can take several minutes depending on the capabilities of your system. Generally, each byte of a target file requires 40 bytes of memory for processing. For example, a 1 Mb file would require 40 Mb of processing memory, a 2 Mb file — 80 Mb and so on.

At this time, features in the disassembly implementation do not allow for producing source code that could be recompiled as is. File sections in which the physical size equals 0 and the data is located behind the boundaries of the physical size of the section cannot be accurately translated. The disassembler marks these items [DB Count DUP (??)] and does not place labels inside these areas. These kinds of sections arise because programmers often request memory and then fail to manage it properly, relying entirely on the operating system or the compiler to release the memory resources. For example (in Pascal):

Var MyData: Array [WORD] of Byte;

In this case, the space for storing the variable MyData will not be allocated physically in the data section, but the virtual size of the data section containing it will be increased by a WORD value. It also happens that variables declared in this fashion will result in lost megabytes of virtual space for the containing data section.

At present, due to the features of the internal data structures in PE Explorer and for the reasons cited above, mismanaged memory allocations are excluded from disassembly. Otherwise, the memory expenditure required to process target files could grow astronomically. Currently, each byte of incoming data requires 30-40 bytes of memory to process.


Feature Tour  
 prev | next 



PE Explorer

View Screenshots

Download a 30 day trial version of PE Explorer Buy the Full Version