|WikiProject Computer Security / Computing||(Rated Stub-class, Mid-importance)|
Description of what the program does:
%## = and ax, ## (remember that x86 uses little endian order!)
-## = sub ax, ##
P = push ax (search for "computer stack" or something)
X = pop ax
\ = pop sp
T = push sp
A = inc cx
1. The program calculates the value of "50 bytes after where the algorithm ends"
2. It sets the stack pointer to point to that location
3. It calculates a hexadecimal C3C3 and places it in the AX-cpu register. (Sort of like a very-very-fast variable)
4. It pushes (overwrites) "C3C3" to the stack
5. It jumps to the written code.
The reason why it has to calculate the values is that it is impossible to store this information in alphanumeric writeable ASCII. The values are simply way out of the reach for anything that you can write with an ordinary keyboard.
C3 means "RET" in x86 assembly language. It returns back to the previous routine/function by popping a new IP (Instruction Pointer) off the stack.
The reson why I choose to rewrite the code first after 50 bytes is because of the Prefetch Input Queue. Kind of complicated. The CPU reads its codes some cycles before it executes them, so, well, i had to be sure it wasnt allready loaded into the PIQ.
0912:0100 252121 AND AX,2121 ; %!! 0912:0103 254242 AND AX,4242 ; %BB .. null ax-register (register = very very fast variable hardcoded in the cpu) 0912:0106 2D747A SUB AX,7A74 0912:0109 2D2121 SUB AX,2121 0912:010C 2D2563 SUB AX,6325 ; zero minus large numbers equals 0x146 (the program starts at 0x100 + code + 50 bytes) 0912:010F 50 PUSH AX 0912:0110 5C POP SP ; ser the stack pointer to this location (*sp = 0x146 in C'ish) 0912:0111 252121 AND AX,2121 0912:0114 254242 AND AX,4242 ; null ax-register again 0912:0117 2D3D3C SUB AX,3C3D : 0 - 0x3D3C = 0xC3C3 (remember little endian byte order) 0912:011A 50 PUSH AX ; *sp = 0xC3C3 ... C3 = "RET" in x86 assembly, meaning "return from function" or "}" in Cish. 0912:011B 7528 JNZ 0145 ; "goto *sp", think of it as the last "}" in the main-function (...)
Links to alphanumeric/unicode shellcode articles
Even though the topics of these articles mention shellcode, the information applies to alphanumeric code in general. They are more relevant to alphanumeric code then they are to shellcode. SkyLined (talk) 08:46, 10 March 2008 (UTC)
Alphanumeric vs printable
This article states that alphanumeric code uses any printable character. This is not true; alphanumeric code uses only 0-9A-Za-z (hence the name alphanumeric). Printable code uses any printable character (charcodes 0x20-0x7F). SkyLined (talk) 08:46, 10 March 2008 (UTC)
I'm planning on rewriting and expanding this article. Here's what I think should be in it:
1) alphanumeric / printable code 1.1) Techniques 1.2.1) individual instructions create code, which is executed (rix, others?) 1.2.2) decoder loop decodes code, which is executed (Alpha2, others?) 1.2) uppercase/lowercase 2) Platform specific sections: 2.1) IA-32 2.1.1) The GetPC problem 2.1.2) uppercase/lowercase Other platforms? 3) Semi-alphanumeric/printable code: 3.1) toLowercase / toUppercase safe code 3.2) unicode safe code 3.3) Ascii art code
Removed BSCS Reference
I removed this:
- In BSCS the alphanumeric character have meaning: "a"-"z", "A"-"Z", "0"-"9". No special characters are included.
The last link "Shellcoding for Linux and Windows — Printable Shellcode: Explanation and tutorial" doesn't strike me as particularly relevant or informative. It also doesn't seem to cover anything that isn't already covered in more detail in the other links. Because a lot of the links are pointing to my site, I'm worried that I will appear biased towards my own links. Can somebody else confirm the link isn't useful and remove it? — SkyLined (talk) 23:22, 26 February 2011 (UTC)