ECDH without IPv6

Martyr

here is way to use WiteG ECDSA library to implement ECDH

the WiteG random function should be replaced by mersenne twister

[php].686p
.mmx
.model flat,stdcall
option casemap:none
option prologue:none
option epilogue:none

include gfp.inc
include ecp.inc
include random.inc

includelib gfp.lib
includelib ecp.lib
includelib random.lib

extern KEY_BASEPOINT:DWORD

.data?
pubKey1 db 48 dup (?)
pubKey2 db 48 dup (?)
prvKey1 db 16 dup (?)
prvKey2 db 16 dup (?)
commonPointA db 48 dup (?)
commonPointB db 48 dup (?)

.code
start:
;A & B do:
invoke random, 0, 0
invoke set_N

;Alice do:
@: invoke random, offset prvKey1, 16
invoke fixmod, offset prvKey1
invoke comparezero, offset prvKey1
jz @B

;Bob do:
@: invoke random, offset prvKey2, 16
invoke fixmod, offset prvKey2
invoke comparezero, offset prvKey2
jz @B

;A & B do:
invoke set_P

;Alice do:
invoke ECP_Mul, offset prvKey1, offset KEY_BASEPOINT, offset pubKey1

;now Alice send pubKey1 to Bob

;Bob do:
invoke ECP_Mul, offset prvKey2, offset KEY_BASEPOINT, offset pubKey2
;now Bob send pubKey2 to Alice

;A & B receive pubKeys, they check if those pubKeys are valid points
;on common constant elliptic curve and if points arent in infinite

;Alice do:
mov esi, offset commonPointA
invoke ECP_Mul, offset prvKey1, offset pubKey2, esi

assume esi: ptr ECPOINT

lea esi, [esi].X

;Bob do:
mov edi, offset commonPointB
invoke ECP_Mul, offset prvKey2, offset pubKey1, edi

assume edi: ptr ECPOINT

lea edi, [edi].X

;established key is X coordinate, under esi for Alice, under edi for Bob
;both should have the same number

;its only our check
invoke compare, esi, edi

jz @we_have_THEKEY
xor eax, eax
ret

@we_have_THEKEY:

;both should have the same number
nop

;now they can use some KDF (key derivation function) to create key for symmetric encryption
nop
xor eax, eax
inc eax

ret
end start[/php]

i suppose the point of this is to show how an attacker could use ECDH to hide details of an attack...what would you do in this situation?

i'm sure everyone has their own way..but its relative to the problem.