EICAR Commented

; ================
; == About this ==
; ================

; This is an assembly file to generate the EICAR AntiVirus Test File.
; Assembling this triggers your anti virus, so don't do it.
; The file itself is not dangerous and servers as a safe AV demo.

; The compiled assembly uses only instructions that result in the byte range 0x1F < x < 0x80
; To achieve this, it uses a lot of XOR instructions because MOV is outside of ASCII
; moving data to and from registers is done via PUSH/POP for the same reason.

; ================
; == Assembling ==
; ================

; Depending on your assembler, the settings differ. Flat assembler (FASM) doesn't has any of them.
; 1. Disable all optimizations
; 2. Set Target instruction set to 16 bit x86 (Real Mode)
; 3. Set OS Target to DOS

; ===============
; == Verifying ==
; ===============

; If properly done, the output can be opened in any text editor and should read exactly:
; X5O!P%@AP[4\PZX54(P^)7CC)7}$EICAR-STANDARD-ANTIVIRUS-TEST-FILE!$H+H*
; No other bytes should be in the file.

; =============
; == Running ==
; =============

; Your anti virus should trigger as soon as you assemble.
; To actually run the binary, you have to disable your anti virus.
; 64 bit Windows has no support for 16 bit executables anymore.
; To run under 64 bit Windows, use a DOS emulator like DOSBox

; =======================
; == DOS Memory Layout ==
; =======================

org 0100h   ; Make this a DOS .com executable
; The first 0xFF bytes are used by DOS to store information for the application.
; An example would be the command line string to extract the arguments from.
; This is the reason why .com files start at offset 0x100
; MS-DOS .com executables run under 16 bit mode and thus have to fit into 64K of memory.
; This binary makes usage of the stack to avoid the MOV instruction.
; The stack pointer is initialized by DOS to point to the end of memory (0xFFFE to fit a WORD).
; Stack, data and code share the same unprotected memory page,
; which means that if you are pushing too much onto the stack it will eventually overwrite code.
; It also means that the stack size is primarily determined by how much memory you use for code and data.
; There are at most 2 values on the stack and no memory for data storage is needed,
; therefore only 4 bytes of memory need to be free of the 64k page for this application.

; ===================
; == Start Of Code ==
; ===================

; Get 0x214F on the stack
pop  ax     ; AX = 0 because DOS initialized the stack for us already and the memory is set to zero.
xor  ax,214Fh   ; xor with zero is the same as using mov but falls into ASCII range
push ax
; Note: The reason we can try to get a value from an empty stack is because in x86 it wraps around

; Get address of instruction to be changed and move it from ax into bx
and  ax,4140h   ; AX = 0x0140 because 0x214F was still in there: 0x214F & 0x4140 = 0x140
push ax
pop  bx

; Get address location of printable string into dx. The DOS print call expects it to be there
xor  al,5Ch ; AX = 0x011C because we only use xor on the lower byte of the register (AL)
push ax
pop  dx

; Get the first value (0x214F) back from the stack, modify it, and put it into the SI register
pop  ax
xor  ax,2834h   ; AX = 0x097B, AH = 0x09 -> DOS INT21 Service: "Display String"
push ax
pop  si
; We don't have to worry about AL, because DOS INT21 only cares for AH.

; Self modify code: change instruction at 0x0140 (second last instruction of this code) to "int 21h"
sub  [bx],si    ; [0x0140] = 0x2B48 - 0x097B = 0x21CD -> 0x21 = "DOS print string call", 0xCD = "int instruction"
; This also overwrites the first byte of the last instruction

; Self modify code: change instruction at 0x0142 (last two bytes of last instruction) to "int 20h"
inc  bx
inc  bx     ; BX = 0x0142
sub  [bx],si    ; [0x0142] = 0x2A48 - 0x097B = 0x20CD -> 0x20 = DOS exit call, 0xCD = "int instruction"

; Jump over the string to the (meanwhile) changed Instruction at 0x0140.
jge  0140h  ; "jge" is the same as "jnl"
; Because of the subtraction done before this jump, the required flags (S+Z) are set as needed.
; This jump is always performed.

; EICAR string in the middle of the file.
; The '$' is used by the DOS print call to signal the end of a string, as opposed to the more common '\0' in C.
; The content of the string can be freely changed as long as the length stays the same.
; To keep the length, the string can be padded with '$'
; If you change the string you likely will no longer trigger any anti virus response.
DB   "EICAR-STANDARD-ANTIVIRUS-TEST-FILE!$"

; These two instructions are here to occupy 1+3 bytes
dec  ax
sub  cx,[bx+si+2Ah]