End of the dongle old aera ~ Dongles bye bye (re-uploade version October 1998) first part ~ second part "How a single +HCU Fravia can easily blow a whole commercial sector out of history" |
Advanced | |
|
||
That's it, nothing more to add... let's hope we get a new dongle aera to work on. Awesome essay. Frog's Print's incredible work should be printed (44 pages !) and sipped slowly, it's 'cracking for conoisseurs'... Fravia's vintage 1998 "grand reserve"! Bye bye to all those that wanted to do some quick bucks selling hardware protections that were NOT protections at all. This is good, nobody will mourn the disappearing of smoke-sellers and bogus protectors. Bye bye to all those that never cared to study assembly. This is good. Bye bye to all the creations of the poor programmers that blindly trusted commercial (and THEREFORE bogus) dongle protections to defend their valuable software instead of writing their own much more solide protections. Tsch Tsch. You had better read +ORC's students essays first... next time. And learn. And now don't come to the silly idea to blame Frog's Print... blame those dilettantes that have sold you smoke. Blame yourself. A little reality cracking would have done you some good, probably. Well, one never ends learning, does one? I would never have believed myself that (almost) all implementations of the dongle protections scheme were so stupid. Oh my, how low, how deep have the programmers fallen, since Micro$oft introduced its frilly-dizzy rattamazz operating system :-) Of course anyone of my readers could now just 'go shopping' and download, fetch or get hundred thousand dongle protected software programs, each one of them -incidentally- extremely expensive. DO NOT DO IT. You'll only stuff your already exploding harddisks to death for nothing and bore yourself to death. See, you don't need to do it: that poor software won't disappear, nor it will go away. It will always be there. It wont be dongle protected any more after this, oh no, no, no :-) Yet, probably it will be as poorly protected as before by some other charlatans, smoke-selling some other 'infallible' protection scheme... people never learn. Unless they study. That's why you are here, my good reader, to master a difficult but POWERFUL art: reversing. So don't steal the software that Frog's Print has now stripped naked. Look at those protection schemes, pathetically whining and sobbing under Frog's streng and mighty reversing blick. They wont be ever able to run away, the poor scared things. Of course -as it happens- this essay could be useful AGAINST THEFT.. for our underfunded public universities for instance, where (as I very well know) hardware dongles are regularly and phisycally stolen from their parallel ports (it's difficult to find a way to fest them there and students don't show any respect for public property any more, having being corrupted by the 'sacrality' of private gains and property)... so this essay may be a good reading, for all colleagues in the public sector, in order to protect against stupid vandalic theft their legally bought software as well :-) See: +crackers bring solutions for the good ones and ruin for the evil ones :-) |
Dongles Programmers Corner |
|
So, Dongles could be a very powerful and strong way to protect a soft...as long as the programmer willing to utilize one knows how to use it. Unfortunately, most of them don't (note that we are not talking about shareware programmers but professional guys! ) and are still using our "beloved" Test eax,eax//jnz Bad_Guy stupid routine, thinking that they don't have to worry about the protection because the dongle will take care of it (same old problem discussed so much often here -: ). Some of them use ridiculous $5 shareware tricks to protect $4.000 professional softs (see Vision XXL) and, worse, others just forgot to read the Dongle documentation manual shipped with their hardware key (see ABBCS31), you'll even find some of the most amazingly stupid protections I can't find any words to describe them (see DigiSHOW). Too bad...specially when those programs are available for free download on the Net.
Sometimes it may be useful to know which dongle is needed for the
program you want to crack (Aladdin's Hasp, Rainbow Technologies' Sentinel,
Dinkey, Everlock...) or, very important if you plan to use SoftICE
and some Breakpoints, the type of dongle (parallel or serial key, internal
or PCMCIA card, Network dongle...), sometimes you won't even have to care
about that. And, finally, in just few cases, you may not be able to crack
a program if you don't have the right dongle (but if you have the dongle
why should you crack it? -: ) or because the program is strongly encrypted
and/or uses too many anti-debugger and anti-w32dasm/IDAPro protections
(usually, those anti-crackers protections are not written by the programer
himself :- ).
For this, never forget to read carefully the program's doc or HLP
files, to visit it's company home page (specially the "installation help"
or "FAQ" pages) to collect the maximum infos about the dongle needed. You'll
find them most of the time, but sometimes you may only find that the program
needs a "green" or "red" dongle. So don't forget to monitor the program
installation on your Hard Drive : for all the programs of this essay, during
their installation I used CleanSweep, WinDelete and TechFacts, all 3 programs
running at the same time. Then check for new .Vxd, .Drv, .Sys and .Dll
files in you windows(/system) or the programs directories. Do the same
the first time you'll run the program because it may use/create and hide
a lot of "~x34_m-!.sys" or similar strange named files anywhere on your
HD. If you can identify the Dongle manufacturer, then visit its WWW/FTP
sites, newsgroups, read newspaper articles about it and visit programmers'
tools retailers companies sites like SOSdevelopers
(if you can understand French), download their softs/demos (SOSdevelopers
download section),
get their free catalogue where you'll find all products for programers
and the latest protections tools of the market (BTW, SOSdevelopers is probably
actually one of the rare place where you can still find SoftICE 3.01 and
all others Numega 14 days trial version products available for downloading
since Numaga lately removed their Download section from their WWW home
site -: ). And finally, if you plan to crack several dongle protected softs,
install and monitor the first one, crack it and then delete it before
installing the second one, third one... otherwise it'll be to messy and
complicated to monitor and crack.
Here is now, for instance, some infos I found at Aladdin
SoftWares (Hasp/NetHasp dongles). They wrote this to help programers
to write a "good" protection for their dongle (blue comments are mine of
course!):
The golden rule of HASP-based software protection is that the
protection system is only as secure as the routines that call the key and
decide whether to allow execution of the protected program. In other words,
since the hardware of HASP keys is too complicated to break
or duplicate, hackers will usually try to trace the protection code
and eliminate the protection routines.
Emulating the Dongle by re-writing its main
routine is much funnier than boring tracing and patching (and sometimes
a lot easier, specially the Hasp protection routine! )
While we supply you with the best hardware and software protection
tools available, and while the HASP Envelope can by itself provide an excellent
protection for application programs, if possible, you should enhance
your protection system by implementing the Application Programming Interface
(API) as well.
If only they knew how to do it...
Please keep in mind that while no single technique can defeat
every hacker, the more traps and safeguards you use, the more difficult
it will be to break your software. In order to achieve maximum protection
with HASP, we suggest you implement at least some of the procedures outlined
below:
Issue as many calls as possible with many different Seed
Codes.
The more calls and expected responses you use, the more difficult
it will be to trace and remove them.
Boring yes, difficult not really.
Perform logical divisions of the checking procedure. Divide
the checking procedure, and if possible, spread it across your program
at far-away points.
A logical division would be: calling, checking, and reacting
routines, each situated in a different section of the program.
Hide parts of the checking code within a regular source code.
For example, the calling routine can be hidden within a general initialization
routine.
Easy to find with a BPIO 278/378
Use complicated mathematical expressions. In order to
confuse hackers, check for the Return Codes using complicated mathematical
expressions. For example, if you are expecting a Return Code which
equals 26387, incorporate an expression such as the one below to perform
the check:
If (ReturnCode1 -
7)/4 = (ExpectedReturnCode1 - 6602)/3 then...
To further enhance your level of security, follow the same procedure
throughout your software for any mathematical constants used in implementing
your HASP software protection solution.
Yes, and add a beautiful jnz_Bad_Guy at the
end...
Use delayed reactions. Do not react immediately to a checking
call. For example,do not test for a certain value and then issue an error
message if you do not like the result. If you do, the hacker will easily
understand what you are doing.
It is better to store the value in a variable A, check A's value
a few subroutine calls later, and set a new variable B to say whether or
not you like it. Several calls after that, test the value of B, and only
then issue the error message.
Force hackers to work backward to figure out how B got a certain
value and why the program fails because of that. The relation to the original
returned value should no longer be obvious.
See above
Perform check-sums. Perform a check-sum on the program
to find out if it has been
tampered with. The most effective means to do this is with the
HASP Pro Kit.
Otherwise, the simplest way to do so is to use the following
pattern:
Calculate check-sum.
Compare with correct
value.
If the two do not
match,
issue an error message;
otherwise, continue.
Unfortunately, this technique is vulnerable to several attacks.
Bad luck!
etc,etc...
After that, there are different ways to crack it:
1/Reverse engineering the dongle protection main routine in order
to emulate the dongle (see ABBCS31).
-This is very efficient specially if this routine is called dozens
or sometimes hundreds of times throughout the program (otherwise we would
have to patch all those cross-references!).
-With this method, you don't need to have the dongle driver loaded
(if you have 10 programs using different dongles, you'll need to have 10
different drivers always loaded).
2/Patching some Call, Test or jnz instructions.
Not the most elegant method but useful if:
-you don't know how the dongle protection routine works but you
quickly found the right Bad_Guy Flag.
-your program is to slow to start/run . Some overbloated EXE or
DLL will check for a dongle, dongle ID and Seed Code and, if they can't
find them, will check again and again before telling you that you don't
have the requested hardware key. For that reason, those programs may check
for the dongle for 15 or sometimes +30 seconds. This even happens sometimes
after you "emulated" the Dongle. A real waste of time! (see WellCAD).
3/Brute force approach:
Close to the above solution, this one may be quite useful for program that
may react strangely when they are modified. This method just force them to work
the way you want (see Ai Damage).
4/You may use your own tips/findings like searching with Hiew the
bytes sequence 0xE900000000
jmp Next_Instruction (see OmniMark).
In this essay, we'll just assume that you are familiar with cracking because I won't give a lot of comments about all usual protections routines/tricks... that don't have anything to see with the dongle protection nor will I spend time to explain how to crack it when it will be quite obvious (most of the time).
As said above, all programs cracked in this essay can be found and
download on their companies homepages (see links below).
You'll find plenty of dongle softs to download for free on the
Net. Just learn how to search and you'll see what I mean.
OK, now let's have some fun!
Hiew 5.66
SoftICE 3.22
W32Dasm v6.x and W32Dasm8.5
+IDA PRO v3.7
A taste of Zen Cracking
1) Micro-Cap V (Spectrum SoftWare) v2.01
900Kb
2) Vision XL/XXL v2.0
(Impuls) 9.5Mb
3) ABBCS31 v1.02
Klinkmann Automation Softs (DDE Servers) 420Kb
4) WellCAD v2.30 : Full
version 10.5Mb or Lite
version 2.5Mb (similar but without help files and sample data files)
5) Label Craft v6.01
(Axicon Softs) 1.2Mb
6) Ai Damage v1.4
(Ai Training Services Ltd) 185Kb (requires the DK2 DESkey dongle
drivers -dk2wn95.386, dk2win32.dll...- if not installed on your system,
you can download them here: Bdk2wn32.exe
720Kb - read the essay for more infos-: )
7) Electro Chemistry 12
Nov.1997 version (Sycopel Scientific Ltd) 531Kb.
8) IPLab v1.7
(Signal Analytics Corp.) 1.5Mb
9) DigiSHOW.vld v1.24
1.59Mb
A) Axon Engineer Pro v2.11e
2.09Mb
B) FTI/DOE v.2.3.1.4
2.78Mb
C) S-Tagger for FrameMaker v2.0
1.32Mb
D) XinTianMa C version
1.04Mb
E) OmniMark V3R1a
Win95 1.88Mb
Apparently, Micro-Cap V programmers are the only guys who understood
that the dongle will not do all the job. They tried to add some improvements
to the protection scheme:
-Fake flags (Bad or Good_Guy).
-Several flags hidden at different locations (using different registers).
-Delayed reactions (calling, checking and reacting routines).
-Mathematical expressions for checking the RetCodes (though they
could improve this part).
-Several calls with different Seed Codes.
...
This is the first Hasp dongle of this essay. I hope you read Zafer's essay about the various hasp-dongle services because I will not explain this again. However, I will show you the main Hasp routine from Micro-Cap as it is just a little different from the one describe by Zafer, but it works just the same:
:0054141A 55
push ebp
:0054141B 89E5
mov ebp, esp
:0054141D 60
pushad
:0054141E 88C7
mov bh, al
; Service called
:00541420 52
push edx
:00541421 58
pop eax
:00541422 8B5508 mov edx, dword
ptr [ebp+08]
:00541425 8B750C mov esi, dword
ptr [ebp+0C]
:00541428 8B3E
mov edi, dword ptr [esi]
:0054142A 80FF32 cmp bh, 32
:0054142D 7205
jb 00541434
:0054142F 8B7518 mov esi, dword
ptr [ebp+18]
:00541432 8B06
mov eax, dword ptr [esi]
:00541434 8B7510 mov esi, dword
ptr [ebp+10]
:00541437 8B36
mov esi, dword ptr [esi]
:00541439 55
push ebp
:0054143A E8B1D00000 call 0054E4F0
; Haspreg
:0054143F 5D
pop ebp
:00541440 8B7D0C mov edi, dword
ptr [ebp+0C]
:00541443 8907
mov dword ptr [edi], eax ; RetCode1
:00541445 8B7D10 mov edi, dword
ptr [ebp+10]
:00541448 891F
mov dword ptr [edi], ebx ; RetCode2
:0054144A 8B7D14 mov edi, dword
ptr [ebp+14]
:0054144D 890F
mov dword ptr [edi], ecx ; RetCode3
:0054144F 8B7D18 mov edi, dword
ptr [ebp+18]
:00541452 8917
mov dword ptr [edi], edx ; RetCode3
:00541454 61
popad
:00541455 5D
pop ebp
:00541456 C21400 ret 0014
We will rewrite this routine in order to emulate the dongle because there will be too may occurences to patch in this soft.
Running Micro-Cap will display the message box: 'Security Key
missing. Replace the key'.
With SoftICE, Bpx the above routine and run Micro-Cap.
The first break occures here:
:00422C39 B801000000 mov eax, 00000001
; Service#1
...
:00422C5B E8BAE71100 call 0054141A
; IsHasp
:00422C60 8B45F0 mov eax, dword
ptr [ebp-10] ; Store RetCode1 in EAX
:00422C63 89EC
mov esp, ebp
:00422C65 5D
pop ebp
:00422C66 5A
pop edx
:00422C67 59
pop ecx
:00422C68 5B
pop ebx
:00422C69 C3
ret
The caller:
:00422AEC E82F010000 call 00422C20
:00422AF1 85C0
test eax, eax
; Check if OK
:00422AF3 752D
jne 00422B22
; Good_Guy jump
Of course, the RetCode1 should be equal to 1.
The second break:
:00422E8D B805000000 mov eax,
00000005
; Service#5
:00422E92 89542414
mov dword ptr [esp+14], edx
:00422E96 E87FE51100 call 0054141A
; HaspStatus
:00422E9B 8B442408
mov eax, dword ptr [esp+08] ;
Parallel Port # (RetCode3)
:00422E9F 8B7C2404
mov edi, dword ptr [esp+04] ;
Type of Hasp (Retcode2)
:00422EA3 A37CF65500 mov dword
ptr [0055F67C], eax
:00422EA8 83FF01
cmp edi, 00000001
; Is it a Memo-Hasp?
:00422EAB 740D
je 00422EBA
:00422EAD B8FCFFFFFF mov eax,
FFFFFFFC
; Bad_Guy
:00422EB2 83C410
add esp, 00000010
:00422EB5 5F
pop edi
:00422EB6 5A
pop edx
:00422EB7 59
pop ecx
:00422EB8 5B
pop ebx
:00422EB9 C3
ret
* Referenced by a (C)onditional Jump at Address:00422EAB
:00422EBA 31C0
xor eax, eax
; Good_Guy
:00422EBC 83C410
add esp, 00000010
:00422EBF 5F
pop edi
:00422EC0 5A
pop edx
:00422EC1 59
pop ecx
:00422EC2 5B
pop ebx
:00422EC3 C3
ret
No surprises yet: after checking if there
is a Hasp dongle connected, it checks its type (Memo-Hasp). It doesn't
care if it is a Memo-Hasp-1 or a Memo-Hasp-4 as Retcode1 is not verified.
Both
The 3rd break:
:00422C0E B802000000
mov eax, 00000002
; Service#2
:00422C13 E802E81100
call 0054141A
; HaspCode
:00422C18 5D
pop ebp
:00422C19 5F
pop edi
:00422C1A 5E
pop esi
:00422C1B C20400
ret 0004
The caller:
:00422B5B E864000000
call 00422BC4
:00422B60 8B1424
mov edx, dword ptr [esp] ;
Store RetCode1 in Edx
:00422B63 81EACA190000
sub edx, 000019CA
; Edx-0x19CA
:00422B69 B905000000
mov ecx, 00000005
:00422B6E 89D0
mov eax, edx
; Store in Eax
:00422B70 C1FA1F
sar edx, 1F
:00422B73 F7F9
idiv ecx
; Eax/5 and store result in Eax
:00422B75 3D982D0000
cmp eax, 00002D98
; Is it correct?
:00422B7A 7405
je 00422B81
; jmp over if OK
:00422B7C BE01000000
mov esi, 00000001
; Set Bad_Guy#1 Flag discreetly so that stupid
; crackers will never find it!
:00422B81 817C24046DF60000 cmp dword
ptr [esp+04],F66D ; Check RetCode2
:00422B89 0F8566FFFFFF
jne 00422AF5
; Bad_Guy#2 jump
:00422B8F 817C24086CD60000 cmp dword
ptr [esp+08],D66C ; Check RetCode3
:00422B97 0F8558FFFFFF
jne 00422AF5
; Bad_Guy#2 jump
:00422B9D 817C240C973B0000 cmp dword
ptr [esp+0C],3B97 ; Check RetCode4
:00422BA5 0F854AFFFFFF
jne 00422AF5
; Bad_Guy#2 jump
:00422BAB EB80
jmp 00422B2D
; JUMP if OK and make him believe everything
; is fine...
...
...
:00422B2D 85F6
test esi, esi
; ...and check Bad_Guy#1 Flag!
:00422B2F 7507
jne 00422B38
; Bad_Guy_Jump.
RetCodes 2, 3 and 4 are displayed in the code (0xF66D, 0xD66C,
0x3B97). To check the RetCode1, the program uses simple mathematical expressions
from offset 422B63 to offset 422B75.
Finding the RetCode1 value is simple.
We have:
0x2D98 = (RetCode1 - 0x19CA) / 5
So:
RetCode1 = (0x2D98 * 5) + 0x19CA)
RetCode1 = 0xFDC2
You can see that if one of the 3 last RetCodes is wrong, it will
clearly send you on the Bad_Guy#2.
But it will discreetly set the Bag_guy#1 flag (mov esi,1) if the
first RetCode is wrong and will go ahead as if everything was fine before
checking it later at offset 422B2F.
The 4th break:
This time, the program will read a block from the dongle, and check if no error occured (RetCode3)
:00422CE2 B832000000 mov
eax, 00000032
; Service#32
:00422CE7 31D2
xor edx, edx
:00422CE9 E82CE71100
call 0054141A
; ReadBlock
:00422CEE 807D02EA
cmp byte ptr [ebp+02], EA ; Check correct
value
:00422CF2 741F
je 00422D13
; jmp if OK
:00422CF4 837D7A00
cmp dword ptr [ebp+7A], 0 ; Check RetCode3
(status Code)
:00422CF8 7519
jne 00422D13
; jmp if NOT correct as well
:00422CFA B801000000
mov eax, 00000001
; Bad_Guy flag
but make it looks
; like a Good_Guy Flag!
:00422CFF 8DA582000000 lea esp, dword
ptr [ebp+82]
..
:00422D0A C3
ret
; Back to caller
...
...
:00422D13 8B457A
mov eax, dword ptr [ebp+7A] ; store status code in
Eax
:00422D16 8DA582000000 lea esp, dword
ptr [ebp+82]
..
:00422D21 C3
ret
; Back to caller
...
...
:00422B22 E831030000
call 00422E58
:00422B27 89C6
mov esi, eax
; store in esi as well
:00422B29 85C0
test eax, eax
; Check if OK
:00422B2B 7418
je 00422B45
; je Good_Guy
:00422B2D 85F6
test esi, esi
; Check again to confuse...
:00422B2F 7507
jne 00422B38
; jne Bad_Guy
We could think that the mov eax,1 is our Good_Guy because
right above, it checks the RetCode3 (equal to 0x0 if no error) and jump
if an error was returned.
But it stores this RetCode into Eax (:422D13) and later will store
it into Esi too, and will check both registers to see if it's OK. It is
clear that, if the dongle was connected, the program should never reach
offset :422CF4.
0xEA is probably a byte calculated somewhere depending on the dongle
RetCode (it is not RetCode1, 2, 3 or 4). To crack this will have
to store 0xEa into ebp+2 and to ensure that the RetCode3 will be equal
to 0x0.
Now, Micro-Cap V seems to work fine but it still check for the dongle
depending on which key is pressed.
The 5th break:
Here, the program will check again Service#1 (IsHasp) but in another location.
The 6th break:
Now it calls again Service 2.
:00422C0E B802000000
mov eax, 00000002
:00422C13 E802E81100
call 0054141A
:00422C18 5D
pop ebp
:00422C19 5F
pop edi
:00422C1A 5E
pop esi
:00422C1B C20400
ret 0004
The Caller:
:00422D6A E855FEFFFF
call 00422BC4
:00422D6F 6681FE8203
cmp si, 0382
:00422D74 7527
jne 00422D9D
:00422D76 813C24C7710000
cmp dword ptr [esp], 000071C7
:00422D7D 75C8
jne 00422D47
:00422D7F 817C2404935B0000 cmp dword
ptr [esp+04], 00005B93
:00422D87 75BE
jne 00422D47
:00422D89 817C2408E1BF0000 cmp dword
ptr [esp+08], 0000BFE1
:00422D91 75B4
jne 00422D47
:00422D93 817C240C900D0000 cmp dword
ptr [esp+0C], 00000D90
:00422D9B 75AA
jne 00422D47
:00422D9D 6681FEE486
cmp si, 86E4
:00422DA2 752B
jne 00422DCF
:00422DA4 813C248BF60000
cmp dword ptr [esp], 0000F68B
:00422DAB 759A
jne 00422D47
:00422DAD 817C2404D3B40000 cmp dword
ptr [esp+04], 0000B4D3
:00422DB5 7590
jne 00422D47
:00422DB7 817C2408FE0E0000 cmp dword
ptr [esp+08], 00000EFE
:00422DBF 7586
jne 00422D47
:00422DC1 817C240CD43F0000 cmp dword
ptr [esp+0C], 00003FD4
:00422DC9 0F8578FFFFFF
jne 00422D47
:00422DCF 6681FE9907
cmp si, 0799
:00422DD4 7537
jne 00422E0D
:00422DD6 813C2411560000
cmp dword ptr [esp], 00005611
:00422DDD 0F8564FFFFFF
jne 00422D47
:00422DE3 817C2404BB6E0000 cmp dword
ptr [esp+04], 00006EBB
:00422DEB 0F8556FFFFFF
jne 00422D47
:00422DF1 817C2408925C0000 cmp dword
ptr [esp+08], 00005C92
:00422DF9 0F8548FFFFFF
jne 00422D47
:00422DFF 817C240C8EB60000 cmp dword
ptr [esp+0C], 0000B68E
:00422E07 0F853AFFFFFF
jne 00422D47
:00422E0D 6681FE3E05
cmp si, 053E
:00422E12 7537
jne 00422E4B
:00422E14 813C24EB6F0000
cmp dword ptr [esp], 00006FEB
:00422E1B 0F8526FFFFFF
jne 00422D47
:00422E21 817C2404945D0000 cmp dword
ptr [esp+04], 00005D94
:00422E29 0F8518FFFFFF
jne 00422D47
:00422E2F 817C24087D200000 cmp dword
ptr [esp+08], 0000207D
:00422E37 0F850AFFFFFF
jne 00422D47
:00422E3D 817C240CCEAE0000 cmp dword
ptr [esp+0C], 0000AECE
:00422E45 0F85FCFEFFFF
jne 00422D47
:00422E4B 31C0
xor eax, eax
; Good_Guy
:00422E4D 83C410
add esp, 00000010
:00422E50 5E
pop esi
:00422E51 5A
pop edx
:00422E52 59
pop ecx
:00422E53 5B
pop ebx
:00422E54 C3
ret
..
..
:00422D47 B8B13A0000
mov eax, 00003AB1
; Bad_Guy
:00422D4C 83C410
add esp, 00000010
:00422D4F 5E
pop esi
:00422D50 5A
pop edx
:00422D51 59
pop ecx
:00422D52 5B
pop ebx
:00422D53 C3
ret
Depending on the value of Esi ( 0x0382, 0x86E4,
0x0799 or 0x053E) the program will check for different RetCodes (they all
are shown in the code) and if it's OK, will clear Eax (Good_Guy) otherwise
will make you jump to offset :422D47.
If Esi has a different value (this shouldn't
happen) the program will get to the Good_Guy as well.
To crack this, we will simply xor Ecx (on
the stack as it is saved there before the call to the Hasp routine) during
a call to Service 2 and therefore we will always reach our Good_Guy. Clearing
Ecx will not affect the call to Service 2 I described before this one as
it is unused (saved on the stack).
There is a last break when exiting the program:
The 7th break:
:00422F38 B806000000
mov eax, 00000006
:00422F3D EBDC
jmp 00422F1B
...
:00422F1B E8FAE41100
call 0054141A
:00422F20 8B45F8
mov eax, dword ptr [ebp-08]
:00422F23 85C0
test eax, eax
:00422F25 7418
je 00422F3F
Here, the program just check if any error occured.
There are still few xrefs to the main Hasp routine (services 1,
6...) but it doesn't matter as we will 'emulate' the dongle and this will
work for thoses xrefs as well. However, one of them is a call to service
0x29 (HaspID) at offset :422BF6 that apparently the program doesn't reach.
We will not crack it.
As there's not enough room in the main Hasp routine to write our
own one, we will do it mostly at offset :514E4F0 that is called by this
routine:
All changes are shown in red:
.0514141A: 55
push ebp
.0514141B: 89E5
mov ebp,esp
.0514141D: 60
pushad
.0514141E: 88C7
mov bh,al
; bh=Hasp Service called
.05141420: 80FF02
cmp bh,02
; Is it Service#2 ?
.05141423: 0F8510000000
jne .00141439
; jump over if not
.05141429: 66C745240000
mov w,[ebp][00024],00000 ;
otherwise clear Esi
.0514142F: E905000000
jmp .000141439
; Go ahead
.05141434: 8B7510
mov esi,[ebp][00010]
; unused
.05141437: 8B36
mov esi,[esi]
; unused
.05141439: 55
push ebp
; back to original code
.0514143A: E8B1D00000 call
.00514E4F0
; call our new routine
.0514143F: 5D
pop ebp
.05141440: 8B7D0C
mov edi,[ebp][0000C]
.05141443: 8907
mov [edi],eax
; RetCode1
.05141445: 8B7D10
mov edi,[ebp][00010]
.05141448: 891F
mov [edi],ebx
; RetCode2
.0514144A: 8B7D14
mov edi,[ebp][00014]
.0514144D: 890F
mov [edi],ecx
; RetCode3
.0514144F: 8B7D18
mov edi,[ebp][00018]
.05141452: 8917
mov [edi],edx
; RetCode4
.0514E4F0:
80FF01 cmp
bh,001
; Is it Service#1 ?
.0514E4F3: 7505
jne .00014E4FA
.0514E4F5: B801000000
mov eax,000000001
; RetCode1
.0514E4FA: 80FF02
cmp bh,002
; Is it Service#2 ?
.0514E4FD: 7519
jne .00014E518
.0514E4FF: B8C2FD0000
mov eax,00000FDC2
; RetCode1
.0514E504: BB6DF60000
mov ebx,00000F66D
; RetCode2
.0514E509: B96CD60000
mov ecx,00000D66C
; RetCode3
.0514E50E: BA973B0000
mov edx,000003B97
; RetCode4
.0514E513: E925000000
jmp .00014E53D
.0514E518: 80FF05
cmp bh,005
; Is it Service#5 ?
.0514E51B: 750C
jne .00014E529
.0514E51D: BB01000000
mov ebx,000000001
; RetCode2
.0514E522: B901000000
mov ecx,000000001
; RetCode3
.0514E527: EB14
jmps .00014E53D
.0514E529: 80FF06
cmp bh,006
; Is it Service#6 ?
.0514E52C: 740D
je .00014E53B
.0514E52E: 80FF32
cmp bh,032
; Is it Service#32 ?
.0514E531: 0F8504000000
jne .00014E53B
.0514E537: C6451CEA
mov b,[ebp][0001C],0EA ;
store our 0xEA
.0514E53B: 33C9
xor ecx,ecx
; clear RetCode3 (NO error)
.0514E53D: C3
retn
; back to Hasp main routine
.0514E53E: 8D3534FC5600
lea esi,[00056FC34]
; will never reach this part.
Everything works quite well but it may be safer not to forget the
call to Service 0x29 that we didn't patch, just in case the program would
reach it one of these days...
Vision XL/XXL v2.0
-Description:
High performance image processing and image analysis system.
-OS:
Win95/NT
-Protection:
Dongle (parallel or serial key). Driver: Hardlock.vxd (c:\windows\system).
-Misc.:
Run as a DEMO if no dongle detected (Save, save as, copy disabled,
macros disabled after 30 program runs...).
Described as "seeing the invisible", this is a very sophisticated and extremely expensive soft. For those reasons, we could only expect a strong or at least, a good protection. Unfortunately, its protection is really stupid, and then, the most amazing is that we won't even have to care about the dongle to crack it!
During its installation, the program added the file Hardlock.vxd and told us that we need to restart Windows before running it (to load the dongle driver!).
Put a BPIO -H 378 with SoftICE and then run VisionXXL.
Nothing happens! The program doesn't seem to check
the parallel port.
After a quick check at the huge documentation (25Mb is MS-Word format!),
I found this:
"The security key can be reversed to be used
on a serial interface port...".
I read that there are some lines to add to
your Autoexec.bat as well (SET HL_SEARCH=2f8s,...where 2f8 denotes
the IO address of the serial port,and s denotes the serial port. The ellipsis
... is used to indicate that you can check several ports one after the
other.).
OK, but let's just forget the dongle for now!
Re-run the program. You'll get the big nagscreen/bitmap
telling you that the program is running in demo mode.
As the "Save" feature is disabled, press
it in the target's toolbar. The following message box pops up:
"Sorry, this function is not available in
demo mode"
Put a BPX MessageBoxA and press "Save" again.
SoftICE pops. Press <F11>, click 'OK'
and press <F12> 3 times. You'll land here:
.000CA432: E899820600
call .0001326D0 ;
Check if Full or Demo mode
.000CA437: 85C0
test eax,eax
.000CA439: 740A
je .0000CA445
; Good_Guy_Jump
.000CA43B: E840EBF4FF
call .000018F80 ;
Our MessageBoxA
.000CA440: E975040000
jmp .0000CA8BA ;
Bad_Guy_Exit
.000CA445: E8127D0B00
call .00018215C ;
Good_Guy_Go_Ahead
Those who are familiar with "disabled features"
cracking won't need to read my comments to understand this: this is a stupid
typical protection used so much often for this kind of program.
Let's have a look at the .0001326D0 routine:
001326D0: 8B81E8000000 mov
eax,[ecx][0000000E8]
001326D6: 85C0
test eax,eax
001326D8: 740D
je .0001326E7
001326DA: 8B4804
mov ecx,[eax][00004]
001326DD: 85C9
test ecx,ecx
001326DF: 7406
je .0001326E7
001326E1: B801000000
mov eax,000000001 ;
Bad_Guy_Flag
001326E6: C3
retn
001326E7: 33C0
xor eax,eax
; Good_Guy_Flag
001326E9: C3
retn
This routine is called as soon as you run
the program (to decide if it's a Demo or full working version) and each
time you'll press or try to use one of the disabled features.
Obviously, patching the first test eax,eax
or changing the mov eax,1 will be enough to crack this dongle protection.
ABBCS31 v1.02
-Description:
DDE (Dynamic Data Exchange) Server allowing other Windows application
programs to access the data from the ABB Procontic CS31 automation system.
-OS:
Win95/NT
-Protection:
Dongle (parallel key). Driver: Hasp95.vxd (c:\windows\system).
-Misc.:
Run as a time-limited DEMO if no dongle detected (will only run 1 hour).
At the end of the installation setup, we get a "reboot to activate HASP driver!" message. So we know where we are and where we going! The program added the file C:\Windows\System\Hasp95.vxd and the following line in the Registry: 'HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\VxD\HASP95'.
With HIEW5.66, we can quickly find the Hasp main routine located
in the Abbcs31.exe file (193Kb).
It starts at offset :00001AC9. Now we can BPX it with SoftICE (don't
forget to add 0x400000 to the offset from Hiew to SoftICE ):
BPX 401AC9
Running ABBCS32, we'll get 3 breaks. Callers are:
:401C9D
:401D0C
:401D8A
Here is the code (again from Hiew):
.00001C97: A1C40A4200
mov eax,[000420AC4] ;
eax=0x05
.00001C9C: 50
push eax
.00001C9D: E827FEFFFF
call .000001AC9
; Hasp routine service #5
.00001CA2: 83C424
add esp,024
.00001CA5: 8B45CC
mov eax,[ebp][-0034]
.00001CA8: A3A0204200
mov [0004220A0],eax
.00001CAD: 837DD000
cmp d,[ebp][-0030],0 ;
Is it a Hasp-3 (RetCode2=0)?
.00001CB1: 0F850C000000
jne .000001CC3
.00001CB7: C745C401000000
mov d,[ebp][-003C],1 ;
Yes: Good_Guy_Flag
.00001CBE: E907000000
jmp .000001CCA
.00001CC3: C745C400000000
mov d,[ebp][-003C],0 ;
Bad_Guy_Flag
.00001CCA: 837DC400
cmp d,[ebp][-003C],0 ;
Is it OK?
.00001CCE: 0F8446000000
je .000001D1A
; No, send_Him_Away...
.00001CD4: C705C40A420001000000 mov d,[000420AC4],1
; otherwise Go_Ahead
...
.00001D06: A1C40A4200
mov eax,[000420AC4] ;
eax=0x01
.00001D0B: 50
push eax
.00001D0C: E8B8FDFFFF
call .000001AC9
; Hasp routine service #1
.00001D11: 83C424
add esp,024
.00001D14: 8B45D4
mov eax,[ebp][-002C]
.00001D17: 8945C4
mov [ebp][-003C],eax
.00001D1A: 837DC400
cmp d,[ebp][-003C],0 ;
Is it a Hasp Dongle (RetCode1=1)?
.00001D1E: 0F84E7000000
je .000001E0B
; Bad_Guy_Jump...
.00001D24: C705C40A420002000000 mov d,[000420AC4],2
; otherwise Go_Ahead
...
.00001D84: A1C40A4200
mov eax,[000420AC4] ;
eax=0x02
.00001D89: 50
push eax
.00001D8A: E83AFDFFFF
call .000001AC9
; Hasp routine service #2
.00001D8F: 83C424
add esp,024
.00001D92: C705C00A420000000000 mov d,[000420AC0],0
.00001D9C: A1AC204200
mov eax,[0004220AC]
.00001DA1: C1E004
shl eax,004
.00001DA4: 8B4DD4
mov ecx,[ebp][-002C]
.00001DA7: 398818204200
cmp [eax][000422018],ecx; Is RetCode1=0x7A17
?
.00001DAD: 0F8551000000
jne .000001E04
; Bad_Guy_Jump
.00001DB3: A1AC204200
mov eax,[0004220AC]
.00001DB8: C1E004
shl eax,004
.00001DBB: 8B4DD0
mov ecx,[ebp][-0030]
.00001DBE: 39881C204200
cmp [eax][00042201C],ecx; Is RetCode2=0x768A
?
.00001DC4: 0F853A000000
jne .000001E04
; Bad_Guy_Jump
.00001DCA: A1AC204200
mov eax,[0004220AC]
.00001DCF: C1E004
shl eax,004
.00001DD2: 8B4DCC
mov ecx,[ebp][-0034]
.00001DD5: 398820204200
cmp [eax][000422020],ecx; Is RetCode3=0x554C
?
.00001DDB: 0F8523000000
jne .000001E04
; Bad_Guy_Jump
.00001DE1: A1AC204200
mov eax,[0004220AC]
.00001DE6: C1E004
shl eax,004
.00001DE9: 8B4DC8
mov ecx,[ebp][-0038]
.00001DEC: 398824204200
cmp [eax][000422024],ecx; Is RetCode4=0xE6EB
?
.00001DF2: 0F850C000000
jne .000001E04
; Bad_Guy_Jump
.00001DF8: C745C401000000
mov d,[ebp][-003C],1 ;
Good_Guy_Flag
.00001DFF: E907000000
jmp .000001E0B
.00001E04: C745C400000000
mov d,[ebp][-003C],0 ;
Bad_Guy_Flag
.00001E0B: 837DC400
cmp d,[ebp][-003C],0
.00001E0F: 0F8515000000
jne .000001E2A
; Dongle_OK
.00001E15: 837DC002
cmp d,[ebp][-0040],2 ;
No_Dongle
This routine is INCREDIBLY STUPID!!
As I said at the very beginning of this essay, the programmer just
forgot to read the Hasp documentation shipped with his dongle:
Usually, you first check if there is a Hasp dongle connected to
the parallel port (Service #1) and then, if there is one, the type of Hasp
(Service #5).
We can see here that the program check the type of Hasp before
verifying if there is a dongle connected.
And so?
The only problem is that we can see that ABBCS32 is expecting a
Hasp-3 dongle (RetCode2=0). But, after a call to Service #5, if you don't
have any dongle connected the RetCode2 will be equal to... 0x00 as
well!
This just means that if you have the correct dongle or nothing
at all, the program will just assume that you have the right Hasp-3! Unbelievable...
Fortunately, it checks Service #1 later.
There are a lot of different ways to crack this crap (patching the
Bad_Guy_Flag...) but let's emulate the dongle by reverse engineering the
main Hasp routine again:
Offsets changed are shown in red :
.00001AC9: 55 push
ebp
.00001ACA: 8BEC mov
ebp,esp
.00001ACC: 50 push
eax
.00001ACD: 53 push
ebx
.00001ACE: 51 push
ecx
.00001ACF: 52 push
edx
.00001AD0: 57 push
edi
.00001AD1: 56 push
esi
.00001AD2: 8B751C mov
esi,[ebp][0001C]
.00001AD5: 8B3E mov
edi,[esi]
.00001AD7: B900000000 mov
ecx,000000000 ; Clear ecx.
.00001ADC: 8B5D08
mov ebx,[ebp][00008] ; bl=Service
called.
.00001ADF: 80FB02
cmp bl,002
; Is it Service #2?
.00001AE2: 7512
jne .000001AF6
.00001AE4: 66B8177A mov
ax,07A17 ;
correct RetCode#1
.00001AE8: 66BB8A76 mov
bx,0768A ;
correct RetCode#2
.00001AEC: 66B94C55 mov
cx,0554C ;
correct RetCode#3
.00001AF0: 66BAEBE6 mov
dx,0E6EB ;
correct RetCode#4
.00001AF4: EB0F
jmps .000001B05
.00001AF6: 80FB01
cmp bl,001
; Is it Service #1?
.00001AF9: 7506
jne .000001B01
.00001AFB: 66B80100 mov
ax,00001 ;
correct RetCode#1
.00001AFF: EB04
jmps .000001B05
.00001B01: 33DB
xor ebx,ebx
; Xor unused registers
.00001B03: 33C0
xor eax,eax
; just in case.
.00001B05: 8B7D1C mov
edi,[ebp][0001C] ; Back to original code.
.00001B08: 8907 mov
[edi],eax
.00001B0A: 8B7D20 mov
edi,[ebp][00020]
.00001B0D: 891F mov
[edi],ebx
.00001B0F: 8B7D24 mov
edi,[ebp][00024]
.00001B12: 890F mov
[edi],ecx
.00001B14: 8B7D28 mov
edi,[ebp][00028]
.00001B17: 8917 mov
[edi],edx
.00001B19: 5E pop
esi
.00001B1A: 5F pop
edi
.00001B1B: 5A pop
edx
.00001B1C: 59 pop
ecx
.00001B1D: 5B pop
ebx
.00001B1E: 58 pop
eax
.00001B1F: 5D pop
ebp
.00001B20: C3 retn
Abbcs31 is cracked.
Searching for the main Hasp routine, we easily find it at offset
:44D528 in the main program (wellcad.exe 1.42Mb). It has 2 X-refs and then,
the 14 X-refs (called several times) calling the Hasp services.
Here are two of them:
:0049ECFE 6A01
push 00000001
; Service #1
:0049ED00 E87B080000
call 0049F580
; Is_It_A_Hasp_Dongle ?
:0049ED05 8B442408
mov eax, dword ptr [esp+08]
:0049ED09 8D4C2410
lea ecx, dword ptr [esp+10]
:0049ED0D 48
dec eax
:0049ED0E 8D54240C
lea edx, dword ptr [esp+0C]
:0049ED12 83F801
cmp eax, 00000001
:0049ED15 8D442414
lea eax, dword ptr [esp+14]
:0049ED19 1BFF
sbb edi, edi
:0049ED1B 50
push eax
:0049ED1C 8D44240C
lea eax, dword ptr [esp+0C]
:0049ED20 51
push ecx
:0049ED21 52
push edx
:0049ED22 8BCE
mov ecx, esi
:0049ED24 F7DF
neg edi
:0049ED26 50
push eax
:0049ED27 6868200000
push 00002068
:0049ED2C 6A02
push 00000002
; Service # 2
:0049ED2E E84D080000
call 0049F580
; Get_Hasp_Code
:0049ED33 817C2408A5F70000 cmp dword ptr [esp+08], 0000F7A5;Is
RetCode1 OK ?
:0049ED3B 7529
jne 0049ED66
; Bad_Guy jump
:0049ED3D 817C240C8E420000 cmp dword ptr [esp+0C], 0000428E;Is
RetCode2 OK ?
:0049ED45 751F
jne 0049ED66
; Bad_Guy jump
:0049ED47 817C2410E54F0000 cmp dword ptr [esp+10], 00004FE5;Is
RetCode3 OK ?
:0049ED4F 7515
jne 0049ED66
; Bad_guy jump
:0049ED51 817C24144E470000 cmp dword ptr [esp+14], 0000474E;Is
RetCode4 OK ?
:0049ED59 750B
jne 0049ED66
; Bad_Guy jump
:0049ED5B 83E701
and edi, 00000001
:0049ED5E 8BC7
mov eax, edi
; Set Good_Guy flag (eax=0x01)
:0049ED60 5F
pop edi
:0049ED61 5E
pop esi
:0049ED62 83C410
add esp, 00000010
:0049ED65 C3
ret
* Referenced by a (C)onditional:0049ED3B(C), :0049ED45(C), :0049ED4F(C),
:0049ED59(C)
:0049ED66 33FF
xor edi, edi
:0049ED68 8BC7
mov eax, edi
; Bad_Guy (clear eax)
:0049ED6A 5F
pop edi
:0049ED6B 5E
pop esi
:0049ED6C 83C410
add esp, 00000010
:0049ED6F C3
ret
Let's crack it:
:0049ED66 66B80100
mov ax,00001
Now if you run the program, it will ask you for your name and company
name to type in. Then, it will take about 12 seconds on a P233 before WellCAD
main window appears on your screen. Let's solve this problem.
As there a too many checks and comparisons occuring even before
the calls to the dongle, with SoftICE, BPX the main Hasp routine (:44D528)
and try to find some interesting cross references that we'll lead us to
the fatal(s) jnz_Bad_Guy.
Without any problem, we find several of them but only 2 seem very
interesting:
:004C01E7 E8684A0200 MFC40.MFC40:NoName0713
:004C01EC 8D8BC8000000 lea ecx, dword
ptr [ebx+000000C8]
:004C01F2 E829E6FDFF
call 0049E820
; Verify_Dongle (+ Misc routines)
:004C01F7 85C0
test eax, eax
; Is_Dongle_OK ?
:004C01F9 0F8455FDFFFF je 004BFF54
; Bad_Guy jump
:004C01FF 68C8030000
push 000003C8
; Good_Guy go ahead
...
...
:004C025F E8E4490200
MFC40.MFC40:NoName0711
:004C0264 8D8BC8000000
lea ecx, dword ptr [ebx+000000C8]
:004C026A E881F0FDFF
call 0049F2F0
; Verify_Dongle (+ Misc routines)
:004C026F 85C0
test eax, eax
; Is_Dongle_OK ?
:004C0271 0F84DDFCFFFF
je 004BFF54
; Bad_Guy jump
:004C0277 8D4DC8
lea ecx, dword ptr [ebp-38] ;
Good_Guy go ahead
Just get rid of the two Verify_Dongle calls and replace them with mov eax,1.
This will considerabely speed up the program (just 1 little second
to start) and the crack is done.
OK, it seems very easy to crack but it takes longer that you may
think. The protection is not too bad: a lot of calls to sub-routines that
will call other sub-routines etc.. and then comparisons in one of them
with the conditional jump in another one... It may fool or confuse you
during a Live approach, but could not resist a good Dead listing.