---------------------------------------------------------------------------
Funny tricks
---------------------------------------------------------------------------
Funny tricks. Xoring, Junking, Sliding
EXERCISE 01: [LARRY in search of the King]
Before the next step let's resume what you have learned in
the lessons 3-5, beginning with a very simple crack exercise
(again, we'll use the protection scheme of a game, for the
reasons explained in lesson 1): SEARCH FOR THE KING (Version
1.1.). This old "Larry" protection sequence, is a "paper
protection" primitive. It's a very widespread (and therefore easy
to find) program, and one of the first programs that instead of
asking meaningful passwords (which offer us the possibility to
immediately track them down in memory) asked for a random number
that the good buyer could find on the manual, whereby the bad
cracker could not. (Here you choose -with the mouse- one number
out of 5 possible for a "gadget" choosen at random). I don't need
any more to teach you how to find the relevant section of code
(-> see lesson 3). Once you find the protection, this is what you
get:
:protection_loop
:C922 8E0614A3 MOV ES,[A314]
...
:C952 50 0E PUSH AX & CS
:C954 E81BFF CALL C872 <- call protection scheme
:C957 5B POP BX twice
:C959 8B76FA MOV SI,[BP-06] <- prepare store_room
:C95C D1E6 SHL SI,1 <- final prepare
:C95E 8942FC MOV [BP+SI-04],AX <- store AX
:C961 837EFA00 CMP Word Ptr [BP-06],+00 <- good_guy?
:C965 75BB JNZ C922 <- loop, bad guy
:C967 8E0614A3 MOV ES,[A314]
:C96B 26F606BE3501 TEST Byte Ptr ES:[35BE],01 <- bad_guy?
:C971 74AF JZ C922 <- loop, bad guy
:C973 8B46FC MOV AX,[BP-04]... <- go on good guy
Let's see now the protection scheme called from :C954
:C872 55 PUSH BP
...
:C8F7 90 NOP
:C8F8 0E PUSH CS
:C8F9 E87234 CALL FD6E <- call user input
:C8FC 5B POP BX
:C8FD 5B POP BX
:C8FE 8B5E06 MOV BX,[BP+06]
:C901 D1E3 SHL BX,1
:C903 39872266 CMP [BX+6622],AX <- right answer?
:C907 7505 JNZ C90E <- no, beggar_off
:C909 B80100 MOV AX,0001 <- yes, AX=1
:C90C EB02 JMP C910
:C90E 2BC0 SUB AX,AX <- beggar_off with AX=0
:C910 8BE5 MOV SP,BP
:C912 5D POP BP
:C913 CB RETF <- back to main
Here follow 5 questions, please answer all of them:
1) Where in memory (in which locations) are stored the "right"
passnumbers? Where in memory is the SEGMENT of this
locations stored? How does the scheme get the OFFSET?
2) Would setting NOPs instructions at :C965 and :C971 crack?
Would it be a good idea?
3) Would changing :C907 to JZ crack? Would it be a good idea?
4) Would changing :C907 to JNZ C909 crack? Would it be a good
idea?
5) Write down (and try) at least 7 OTHER different patches to
crack this scheme in spades (without using any NOP!).
Uff! By now you should be able to do the above 5 exercises in
less than 15 minutes WITHOUT USING THE DEBUGGER! Just look at the
data above and find the right answers feeling them... (you 'll
now which one are the right one checking with your debugger...
score as many points as you like for each correct answer and sip
a good Martini-Wodka... do you know that the sequence should
ALWAYS be 1) Ice cubes 2) Martini Dry 3) Wodka Moskovskaja 4)
olive 5) lemon 6) Schweppes Indian tonic?
Let's now come to the subject of this lesson:
-----> [Xoring] (Simple encryption methods)
One easy way to encrypt data is the XOR method. XOR is a bit
manipulation instruction that can be used in order to cipher and
decipher data with the same key:
Byte to encrypt key result
FF XOR A1 5E
5E XOR A1 FF
As you can see XOR offers a very easy way to encrypt or to
decrypt data, for instance using the following routine:
encrypt_decrypt:
mov bx, offset_where_encryption/decryption_starts
xor_loop:
mov ah, [bx] <- get current byte
xor ah, encrypt_value <- engage/disengage xor
mov [bx], ah <- back where you got it
inc bx <- ahead one byte
cmp bx, offset_start_+_size <- are we done?
jle xor_loop <- no, then next cycle
ret <- back where we came from
The encrypt_value can be always the same (fixed) or chosen at
random, for instance using INT_21, service 2Ch (get current time)
and choosing as encrypt_value the value reported in DL (but
remembering to discard the eventual value 0, coz otherwise it
would not xor anything at all!)
random_value:
mov ah,2Ch
int 21h
cmp dl,0
je random_value
mov encrypt_value,dl
The problem with XORing (and with many other encryption
methods), is that the part of the code that calls the encryption
routine cannot be itself encrypted. You'll somewhere have, "in
clear" the encryption key.
The protectionist do at times their best to hide the
decrypting routine, here are some common methods:
-----> JUNK FILLING, SLIDING KEYS AND MUTATING DECRYPTORS
These are the more common protection method for the small
decryption part of the program code. This methods, originally
devised to fool signature virus scanners, have been pinched from
the polymorphic virus engines of our fellows viriwriters, and are
still in use for many simple decryption protection schemes. For
parts of the following many thanks go to the [Black Baron], it's
a real pity that so many potential good crackers dedicate so much
time to useless (and pretty repetitive) virus writing instead of
helping in our work. This said, virus studying is VERY important
for crackers coz the code of the viri is
* ULTRAPROTECTED
* TIGHT AND EFFECTIVE
* CLOAKED AND CONCEALED.
Let's show as example of the abovementioned protection tactics
the following ultra-simple decryptor:
MOV SI,jumbled_data ;Point to the jumbled data
MOV CX,10 ;Ten bytes to decrypt
mn_loop: XOR BYTE PTR [SI],44 ;XOR (un_scramble!) a byte
INC SI ;Next byte
LOOP mn_loop ;Loop the 9 other bytes
This small program will XOR the ten bytes at the location pointed
to by SI with the value 44. Providing the ten bytes were XORed
with 44 prior to running this decryptor the ten bytes will be
restored to their original state.
In this very simple case the "key" is the value 44. But there are
several tricks involving keys, the simplest one being the use of
a "sliding" key: a key that will be increased, or decreased, or
multiplied, or bit-shifted, or whatever, at every pass of the
loop.
A possible protection can also create a true "Polymorph"
decryptor, a whole decryptor ROUTINE that looks completely
different on each generation. The trick is to pepper totally
random amounts of totally random instructions, including JUMPS
and CALLS, that DO NOT AFFECT the registers that are used for the
decryption. Also this kind of protection oft uses a different
main decryptor (possibly from a selection of pre-coded ones) and
oft alters on each generation also all the registers that the
decryptor uses, invariably making sure that the JUNK code that
it generates doesn't destroy any of the registers used by the
real decryptor! So, with these rules in mind, here is our simple
decryptor again:
MOV DX,10 ;Real part of the decryptor!
MOV SI,1234 ;junk
AND AX,[SI+1234] ;junk
CLD ;junk
MOV DI,jumbled_data ;Real part of the decryptor!
TEST [SI+1234],BL ;junk
OR AL,CL ;junk
mn_loop: ADD SI,SI ;junk instr, but real loop!
XOR AX,1234 ;junk
XOR BYTE PTR [DI],44 ;Real part of the decryptor!
SUB SI,123 ;junk
INC DI ;Real part of the decryptor!
TEST DX,1234 ;junk
AND AL,[BP+1234] ;junk
DEC DX ;Real part of the decryptor!
NOP ;junk
XOR AX,DX ;junk
SBB AX,[SI+1234] ;junk
AND DX,DX ;Real part of the decryptor!
JNZ mn_loop ;Real part of the decryptor!
As you should be able to see, quite a mess! But still executable
code. It is essential that any junk code generated by the
Polymorph protection is executable, as it is going to be peppered
throughout the decryptor. Note, in this example, that some of the
junk instructions use registers that are actually used in the
decryptor! This is fine, providing the values in these
registers aren't destroyed. Also note, that now we have random
registers and random instructions on each generation. So, a
Polymorph protection Engine can be summed up into three major
parts:
1 .. The random number generator.
2 .. The junk code generator.
3 .. The decryptor generator.
There are other discrete parts but these three are the ones where
most of the work goes on!
How does it all work? Well a good protection would
* choose a random selection of registers to use for the
decryptor and leave the remaining registers as "junk" registers
for the junk code generator.
* choose one of the compressed pre-coded decryptors.
* go into a loop generating the real decryptor, peppered with
junk code.
From the protectionist's point of view, the advantages of this
kind of method are mainly:
* the casual cracker will have to sweat to find the decryptor.
* the casual cracker will not be able to prepare a "patch" for
the lamers, unless he locates and patches the generators, (that
may be compressed) coz otherwise the decryptor will vary every
time.
To defeat this kind of protection you need a little "zen" feeling
and a moderate knowledge of assembler language... some of the
junk instructions "feel" quite singular when you look at them
(->see lesson B). Besides, you (now) know what may be going on
and memory breakpoints will immediately trigger on decryption...
the road is open and the rest is easy (->see lessons 3-5).
-----> Starting point number magic
For example, say the encrypted code started at address 10h, the
following could be used to index this address:
MOV SI,10h ;Start address
MOV AL,[SI] ;Index from initial address
But sometimes you'll instead find something like the following,
again based on the encrypted code starting at address 10h:
MOV DI,0BFAAh ;Indirect start address
MOV AL,[DI+4066h) ;4066h + 0BFAAh = 10010h (and FFFF = 10h)!!
The possible combinations are obviously infinite.
[BIG KEYS] (Complicated encryption methods)
Prime number factoring is the encryption used to protect
sensible data and very expensive applications. Obviously for few
digit keys the decoding is much easier than for, say, 129 or 250
digit keys. Nevertheless you can crack those huge encryption too,
using distributed processing of quadratic sieve equations (which
is far superior for cracking purpose to the sequential processing
methods) in order to break the key into prime numbers. To teach
you how to do this sort of "high" cracking is a little outside
the scope of my tutorial: you'll have to write a specific short
dedicated program, linking together more or less half a thousand
PC for a couple of hours, for a 250 bit key, this kind of things
have been done quite often on Internet, were you can also find
many sites that do untangle the mysteries (and vagaries) of such
techniques.
As References I would advocate the works of Lai Xueejia, those
swiss guys can crack *everything*. Begin with the following:
Xuejia Lai, James Massey, Sean Murphy, "Markov Ciphers and
Differential Cryptanalysis", Advances in Cryptology,
Eurocrypt 1991.
Xuejia Lai, "On the Design and Security of Block Ciphers",
Institute for Signal and Information Processing,
ETH-Zentrum, Zurich, Switzerland, 1992
Factoring and primality testing is obviously very important for
this kind of crack. The most comprehensive work I know of is:
(300 pages with lengthy bibliography!)
W. Bosma & M. van der Hulst
Primality Testing with Cyclotomy
Thesis, University of Amsterdam Press.
A very good old book you can incorporate in your probes to build
very effective crack programs (not only for BBS accesses :=) is
*the* "pomerance" catalog:
Pomerance, Selfridge, & Wagstaff Jr.
The pseudoprimes to 25*10^9
Math. Comp. Vol 35 1980 pp. 1003-1026
Anyway... make a good search with Lykos, and visit the relevant
sites... if encryption really interests you, you'll be back in
two or three (or thirty) years and you'll resume cracking with
deeper erudite knowledge.
[PATENTED PROTECTION SYSTEMS]
The study of the patented enciphering methods is also *quite*
interesting for our aims :=) Here are some interesting patents,
if you want to walk these paths get the complete texts:
[BEST] USPat 4168396 to Best discloses a microprocessor
for executing enciphered programs. Computer programs which have
been enciphered during manufacture to deter the execution of the
programs in unauthorized computers, must be decrypted before
execution. The disclosed microprocessor deciphers and executes
an enciphered program one instruction at a time, instead of on
a continuous basis, through a combination of substitutions,
transpositions, and exclusive OR additions, in which the address
of each instruction is combined with the instruction. Each unit
may use a unique set of substitutions so that a program which can
be executed on one microprocessor cannot be run on any other
microprocessor. Further, Best cannot accommodate a mixture of
encrypted and plain text programs.
[JOHNSTONE] USPat 4120030 to Johnstone describes a
computer in which the data portion of instructions are scrambled
and in which the data is of necessity stored in a separate
memory. There is no disclosure of operating with instructions
which are completely encrypted with both the operation code and
the data address portion being unreadable without a corresponding
key kernel.
[TWINPROGS] USPat 4183085 describes a technique for
protecting software by providing two separate program storages.
The first program storage is a secure storage and the second
program storage is a free storage. Security logic is provided to
check whether an output instruction has originated in the secure
store and to prevent operation of an output unit which receives
output instructions from the free storage. This makes it
difficult to produce information by loading a program into free
storage.
[AUTHENTICATOR] USPat 3996449 entitled "Operating System
Authenticator," discloses a technique for authenticating the
validity of a plain text program read into a computer, by
exclusive OR'ing the plain text of the program with a key to
generate a code word which must be a standard recognizable code
word which is successfully compared with a standard corresponding
code word stored in the computer. If there is a successful
compare, then the plain text program is considered to be
authenticated and is allowed to run, otherwise the program
is not allowed to run.
ELEMENTS OF [PGP] CRACKING
In order to try to crack PGP, you need to understand how these
public/private keys systems work. Cracking PGP seems extremely
difficult, though... I have a special dedicated "attack" computer
that runs 24 hours on 24 only to this aim and yet have only begun
to see the light at the famous other end of the tunnel. It's
hard, but good crackers never resign! We'll see... I publish here
the following only in the hope that somebody else will one day
be able to help...
In the public key cryptosystems, like PGP, each user has an
associated encryption key E=(e,n) and decryption key D=(d,n),
wherein the encryption keys for all users are available in a
public file, while the decryption keys for the users are only
known to the respective users. In order to maintain a high level
of security a user's decoding key is not determinable in a
practical manner from that user's encoding (public) key. Normally
in such systems, since
e.multidot.d.ident.1 (mod(1 cm((p-1),(q-1)))),
(where "1 cm((p-1),(q-1))" is the least common multiple of the
numbers p-1 and q-1)
d can be determined from e provided p and q are also known.
Accordingly, the security of the system is dependent upon the
ability to determine p and q which are the prime factors of n.
By selecting p and q to be large primes, the resultant composite
number n is also large, and correspondingly difficult to factor.
For example, using known computer-implemented factorization
methods, on the order of 10.sup.9 years is required to factor a
200 digit long number. Thus, as a practical matter, although a
user's encryption key E=(e,n) is public, the prime factors p and
q of n are effectively hidden from anyone due to the enormous
difficulty in factoring n. These aspects are described more fully
in the abundant publications on digital signatures and Public-Key
Cryptosystems. Most public/private systems relies on a message-
digest algorithm.
A message-digest algorithm maps a message of arbitrary length
to a "digest" of fixed length, and has three properties:
Computing the digest is easy, finding a message with a given
digest "inversion" is hard, and finding two messages with the
same digest "collision" is also hard. Message-digest algorithms
have many applications, not only digital signatures and message
authentication. RSA Data Security's MD5 message-digest algorithm,
developed by Ron Rivest, maps a message to a 128-bit message
digest. Computing the digest of a one-megabyte message takes as
little as a second. While no message-digest algorithm can yet
be secure, MD5 is believed to be at least as good as any other
that maps to a 128-bit digest.
As a final gift, I'll tell you that PGP relies on MD5 for a
secure one-way hash function. For PGP this is troublesome, to say
the least, coz an approximate relation exists between any four
consecutive additive constants. This means that one of the design
principles behind MD4 (and MD5), namely to design a collision
resistant function, is not satisfied. You can construct two
chaining variables (that only differ in the most significant bit
of every word) and a single message block that yield the same
hashcode. The attack takes a few minutes on a PC. From here you
should start, as I did.
[DOS 4GW] cracking - This is only a very provisory part of this
tutorial. DOS 4GW cracking will be much better described as soon
as [Lost soul] sends his stuff, if he ever does. For (parts of)
the following I thank [The Interrupt].
Most applications of every OS, and also of DOS 4GW, are
written in C language, coz as you'll have already learned or,
either, you'll learn, only C allows you to get the "guts" of a
program, almost approaching the effectiveness of assembler
language.
C is therefore the LANGUAGE OF CHOICE for crackers, when you
prepare your tools and do not directly use assembler routines.
Besides... you'll be able to find VERY GOOD books about C for
next to nothing in the second hand bookshops. All the lusers are
throwing money away in spades buying huge, coloured and
absolutely useless books on unproductive "bloated" languages like
Visual basic, C++ and Delphy. Good C new books are now rare
(books on assembler language have always been) and can be found
almost exclusively on the second hand market. Find them, buy
them, read them, use them for your/our aims. You can find a lot
of C tutorials and of C material on the Web, by all means DO IT!
Be a conscientious cracker... learn C! It's cheap, lean, mean and
very productive (and creative) :=)
Back to the point: most stuff is written in C and therefore
you need to find the "main" sub-routine inside the asm. With
DOS/4GW programs, search the exe file for "90 90 90 90", almost
always it'll be at the start of the compiled code. Now search for
an INT_21 executed with 4C in AH, the exec to dos code (if you
cannot "BPINT 21 AH=4C" with your tool, then search for the
sequence "b4 4c cd 21". This is the equivalent to [mov AH,4C &
int 21]: it's the most direct call, but as you'll have already
learned, there are half a dozen ways to put 4C in AX, try them
all in the order of their frequency).
A few bytes above the INT_21 service 4C, you'll find the
call to the "main" subroutine: "E8 xx xx". Now place a "CC" byte
a few bytes above the call in the exe and run the exe under a
debugger. When the computer tries to execute the instruction
you'll be throw back in the debugger coz the "CC" byte acts as
INT_01 instruction. Then proceed as usual.
[THE "STEGONATED" PASSWORD HIDEOUT]
A last, very nice trick should be explained to every wannabe
cracker, coz it would be embarrassing to search for passwords or
protection routines that (apparently) are not there. They may be
hidden INSIDE a picture (or a *.waw file for that matter). This
is steganography, a method of disguising messages within other
media.
Depending on how many shades of grey or hues of colour you want
to have, a pixel can be expressed using 8. 16, 32 or even more
bits. If the least significant bit is changed. the shade of the
pixel is altered only one-256th, one-65,OOOth or even less. No
human eye could tell the difference.
What the protectionist does, is hijack the least significant
bit in each pixel of a picture. It uses that bit to store one bit
of a protection, or of a password (or of a file, or of a secret
message). Because digitized pictures have lots of pixels, it's
possible to store lots of data in a single picture. A simple
algorithm will transfer them to the relevant parts of the program
when it needs be, and there we'll intercept them. You'll need to
learn very well the zen-cracking techniques to smell this kind
of stuff though (-> see lesson B).
Well, that's it for this lesson, reader. Not all lessons of my
tutorial are on the Web.
You 'll obtain the OTHER missing lessons IF AND ONLY IF you
mail me back (via anon.penet.fi) with some tricks of the trade
I may not know that YOU discovered. Mostly I'll actually know
them already, but if they are really new you'll be given full
credit, and even if they are not, should I judge that you
"rediscovered" them with your work, or that you actually did good
work on them, I'll send you the remaining lessons nevertheless.
Your suggestions and critics on the whole crap I wrote are also
welcomed.
E-mail me
andrew.nile@gmail.com
No comments:
Post a Comment