Log in

View Full Version : Cloning Sentinel Dongle


cah
August 13th, 2002, 13:13
To,
All Dongle Reverse Engineers


Is it possible to clone sentinel dongle?

How to read/edit a sentinel dongle?
knowledge base on this subject is highly appreciated.

Regards
Cah....

Morlac
August 13th, 2002, 14:51
Apparently, you can, to a certain extent that is.
Some cells are not readable. Like counters, passwords, and algorithm cells.
Counters are no issue.
The problem is getting the 2 overwrite passwords and the algorithm cells.

Those where the only cells I couldnt reveal in a sentinel dongle.

Morlac.

PS - I dont know about hardware cloning though.

cyberheg
August 13th, 2002, 18:16
Same problems arises with the SproQuery function which seems to be tied to the developer id and a 32 bit value (2 cells) in the memory.

// CyberHeg

cah
August 14th, 2002, 09:48
Thanx for your reply.
How to edit/read sentinel dongle? Any tools/utilities available?
How to find queries form dongle? If I have sentinel dongle.

cah...

cyberheg
August 14th, 2002, 11:57
There are plenty of tools availble, look at CrackZ mirror website.

// CyberHeg

scorpie
August 14th, 2002, 15:30
Yes, in general it would be difficult to read "Active Query Cell", although in some program there is a "leak" with this Cells as well (Overwrite function needs write and overwrites passwords). So, if there is an upgrade facility, there is a chance for "leaking".

My question with cloning is: "is there any possibility to program the SuperPro with any Id if let say we can have access to the dongle programmer or to the IC programmer in general ?"


Bye,
Scorpie

warior_jal
August 14th, 2002, 15:39
Hiya Cah!

contact me on my other mail ID, I am working on the same but with a different approach.

VoxQuietis
August 15th, 2002, 19:20
Hi there Dongle reversers,

Some comments on the (Super)Pro:

- first of all get the SDK and (if possible) a dongle to play with
- then as a little exercise, write a program to read out all the
readable dongle cells. This is nice to gain some overview on
the dongle.
- then launch a brute force attack on the write password (but be
careful and use an empty cell for doing that). The attack on
the write password will take no longer than a couple of minutes.
- a brute force attack on the overwrite password is feasible,
but it will take a couple of month ( 2..12 depending on the
amount of luck ;-) to go.
I had to do this on a specific dongle some time ago, and there
is nothing special about it.
- reprogramming the D/I or the S/N should not be possible (at least
if You access the dongle through the standard API/driver).
Most likely there is a backdoor, which allows to reprogram the
dongle. But a possible exploit would require a dump of the
PIC device, which would require to remove the coating around
the chip plus access to a PIC programmer with a wafer prober.
(no need to stress the importance of a dump of the PICs
memory. I guess any Sentinel guru would love to put the
dump into IDA...)
- logging the query-calls is possible, and therefore it is
obviously possible to program an emulator around the logged
data.
- the buffer used for the I/O to the dongle is encrypted. The
encryption/decryption routines can easily ripped out of the
driver or a client application. The encryption towards the
dongle is slightly different than the one towards the
client.
- if You want to change the behaviour of the dongle driver
towards a more convenient operation (why to plug a dongle
to the backside of the box, if the results are clear in
advance? ;-) You should write a filter driver, which
attaches to the dongle driver. This is - by the way - a
fundamental approach, which should be applicable to all
dongles.
- it is possible to direct a query to a readable data cell.
Maybe this could help in reversing the query function,
which I presume to be a tedious task anyway.

So for the time being, the Superpro is not compromised, even
though it is most likely a cassical example for "security by
obscurity".

best regards
- Vox.

PS: Although I promote the idea of free information for all
(and I freely admit that I hate dongles for practial reasons...)
I would not like to see dongle cracks based on the mentioned
principles spreading over the warez sites to all the lamers
of the internet. I will therefore be very careful in discussing
the details of the above mentioned topics. I expressively don't
want that the Sentinel stuff gets a victim of something like
the Blastsoft release with about a hundred of Flexlm seeds.
I hope that the readers of the forum do understand and respect
this point of view.

cah
August 17th, 2002, 05:10
Give me your mail ID
Cah...

Antipodean
August 20th, 2002, 13:05
>- reprogramming the D/I or the S/N should not be possible (at least
>if You access the dongle through the standard API/driver).
>Most likely there is a backdoor, which allows to reprogram the
>dongle. But a possible exploit would require a dump of the
>PIC device, which would require to remove the coating around
>the chip plus access to a PIC programmer with a wafer prober.
>(no need to stress the importance of a dump of the PICs
memory. I guess any Sentinel guru would love to put the
dump into IDA...)

Are you sure they use a PIC micro in them?

If they do, then the programming function is by a serial protocol through a pair of data pins which change their function by putting 13 volts on the MCLR pin. I doubt they will have set the chip to low voltage programming mode, but they will surely have turned on the code protect, which will stop reading out the code.

The MCLR pin and the data pins would all be accessable through the normal dongle pins, so there would be no special pads needed to program the device from a blank chip.

In short, if it is a PIC device, your chances of reading it out are very slim.

You will not need IDA to dump the code if you ever get it out. Just load the hex into the Microchip MPLAB, and it will provide a complete opcode listing )

scooterk
August 21st, 2002, 14:48
I thought there is something on the web showing a clone using an Atmel ASIC 93C46 using an external power source and assembled onto a pc board which is housed into a 25 pin connector resembling a dongle key.
My question is how is the actual chip on the key read and then copied?
via software or via watching the output from multi-trace scope patterns (when the key is accessed), or can one clip to the chip using a hardware emulator?
scooterk

Antipodean
August 22nd, 2002, 17:13
>I thought there is something on the web showing a
>clone using an Atmel ASIC 93C46 using an external
>power source and assembled onto a pc board which is
>housed into a 25 pin connector resembling a dongle key.
>My question is how is the actual chip on the key read
>and then copied? via software or via watching the
>output from multi-trace scope patterns (when the key is
>accessed), or can one clip to the chip using a hardware emulator?
>scooterk

I believe some of the very early dongles had exactly this chip in them, but done as a "chip on Board" where the bare chip is mounted on the PCB, and then covered with a dob of epoxy. The later dongles which can protect locations and do encryption have a micro in them, a bit like a smart card.

The 934x family of chips are serial interface eeproms, so the host PC clocks out the data under software control. New chips are easily programmed using a similar write sequence.

VoxQuietis
August 23rd, 2002, 16:43
Hi there reversers,

to my knowlegde there are only two dongle diagrams published so far. First the C-Plus dongle, which is the one with the EEPROM device mentioned above. I saw it first on the famous essay of Dr Fuhrball. The protocoll is known as 2-wire protocol. All the information is available, at it is both easy to dump the dongle and to clone it.
The second one is the Pro. It is build around a Catalyst CAT-702 chip (I'd be glad if someone could point me to a data sheet of that device). Here the diagram itself is of little interest, since the security is relying on the CAT device.
Wrt to Superpro I am far from shure, whether it is based on the PIC. Nevertheless Dr. Fuhrball made a comment pointing into that direction. On the other hand it seems perfectly matching, that one uses a PIC, since the development of a program is by far cheaper than the development of an ASIC - the ASIC results in NRE cost of at least 30.000 $, and I doubt, that there are that much dongles around, for the ASIC solution becoming cheaper than the PIC approach.

Bye
- Vox.

scooterk
August 23rd, 2002, 17:20
Thanks for the response on this.. I find the hardware/software aspect of making a clone to be an interesting project.
I found an interesting setup form Christian Scheurer using an Atmel AT90S8515 ( 8 bit RISC microcontroller) , however the board is quite large considering it uses a 40 pin layout .. but the article proved interesting. ( The original article is in German ).
scooterk

scorpie
August 24th, 2002, 16:21
Hello there,

Nice explanation VoxQuietis, Antipodean, Scooterk, etc.

Just to add concerning Superpro:

It uses only pins 6, 7, 11, and 18 (ground) on the male connector (which is connected to the Parallel port), and the old dongle use CAT 701. During programming, the female connector is fed by some signals from the programmer.

Although my German is not good (Deutsh ist schwer), kindly let me have the article (Scooterk ?).


Bye,
Scorpie

cah
August 25th, 2002, 04:22
Hayya Scooterk, VoxQuietis, Antipodean & others,

Please attach or give url path for these articles/knowledge base.

Cah...

tgodd
August 25th, 2002, 06:44
As far as the PIC issue, it is not a PIC.
It is an ASIC, simular to an FPGA but a totally custom chip.

As far as the SproQuery it is not tied to the LockId at all.
It is in fact tied to the Write Password.

Have a close look at the Patterns it generates.
You can see the shifting that goes on.

hint, hint

It a big shift register and toggle.


That's about all I can tell you.

chao!!!

toteu
August 25th, 2002, 18:05
Also related to SSPRO shell (SPRO also)
We know that the values(responses from sproQuery) used to decrypt code and data sections are in range (0 -> 2^32).
If we have the dongle these are easy to get/log, if not we can bruteforce them in some cases.
What is interesting is that there are always 3 values(in 0->2^32 range) that decrypt the sections in the same way.
In other words beside sproQuery response, there are other two values that decrypt the section identically.


TOTEU

scooterk
August 26th, 2002, 14:35
Hi scorpie,
This is a sample of the documentation I have on the hasp project using the ATMEL chip. The documentation is written in German.
There are 80 pages to the article and due to size limitations of uploads I've attached the first 9 pages.
If you are interested in the rest of the atricle I can email it to you, (approx 691K in size)
scooterk

VoxQuietis
August 26th, 2002, 15:29
Hi Scorpie (and all RSEG readers)

The mentioned article ist here:

http://www.mountpoint.ch/unique/project/usph/index.html

Just out of curiosity I scaned over it, and I think, it
might be helpful, if I summarized it up: This guy acquired
a "Gesellenbrief" (i.e. a grade after a three year education
in a craftmanship) by doing a practical work, which was the
design of a dongle.

He started by a small assessment of possible attacks and
came to the conclusion, that a device using a microcontroller
running the dongle S/W with some crypto-stuff inside would
be a nice solution. He developed the uC program in assembler
(due to the lack of a C cross compiler, which was found to be
too expensive) plus a PC console application runing on all
versions of DOS from 3.0 to Win95 ;-) Getting it to work
on WinNT was found to be "possible in principle" (which is
a mere alternation of "not possible for me" :-P

Anyway the Atmel listings are there, and if someone
replaces the crystal by a crystal oscillator, then
one could use it as a platform for a dongle emulator -
whereas the problem of triggering the emulator needs
to be redesigned for each class of dongles.

On the other hand I need to stress, that the crucial
point of an emulator remains unsolved: the complete
reversing of the target dongles internal operation.
This is solved for the C-Plus and for (at least the older)
Hasp dongles, but not for the Pro and Superpro.

Moreover the H/W emulator suffers from the same shortcomings
than the H/W lock itself: it is far more effective to implement
the copy protection in S/W than in H/W: as a cracker I'd
rather program the emulator to run on the PC-CPU directly
than to reengineer the communication S/W plus the dongle core.

"You should not copy a faked hardware protection" (this shall
be my first commandment ;-)

Finally some stuff for the German speaking readers of
the board:

> Vergleicht man aber USP!H mit Software – only
> Kopierschutzsystemen, so bringt die Hardware
> doch einen erheblichen Sicherheitsgewinn, da
> diese schwer nachbaubar ist und vorhanden sein
> muss, wenn das Programm gestartet werden will.
> Dongle – Cracks gelten nach wie vor als die
> Herausforderung für Cracker. Denn sie verlangen
> eine eingehende Kenntnis des Computersystems sowie
> eine gewisse Erfahrung mit Hardware.

This could be ripped straight out of a Rainbow-ad.
And now the best goodie:

> Und das schaffen von den bekannten Crackern beispielsweise
> in der Schweiz wohl nur eine Handvoll.
(From the known crackers in Switzerland only a few are
capable to do that)

ROTFL - and within that spirit the usual dongle protection
is written. As Christian Scheuer mentioned he searched the
net for dongle stuff, I expect him to have crossed Crackz
site. But he clearly proved not to have unstood the message.

Best regards,
- Vox.

PS: I have an open Sentinel Pro here on my desk. It features
a CAT-702. If someone minds, I'll upload a picture of it.

scooterk
August 26th, 2002, 18:52
Hi,
I'm glad you posted the info. I wasn't sure if I should post the Url, but a quick search in Google would have given the website.
I would like to see the pix of the 702 though .. if you could be so kind
I still have an interest in the hardware side of this.. hopefully some more tid bits of info will happen by.
It is like a forum that I visited on reprogramming the prom chip for my GM automobile. A little hardware info, and some software implementation... allows to to tweak the chip for better performance.
scooterk

tgodd
August 26th, 2002, 20:46
The remarks given were not necissarily true.
As for the hardware in the pro & superpro not being
already reversed. The Pro was done back in the 80's.
As for the super pro there are 2 generations of superpro.

The first gen. was cracked in 93'.
And since then Rainbow has enhanced the superpro by adding
an enhanced algo circuit. They did this to remove the linearity
of the original algo unit.

The super pro algo was really an alteratrion on the pro.
the pro was a straight data in data out algo unit.
The super pro enhanced on this to accomodate memory cells
which could alter the effects of the algo unit.

The new gen of superpro enhanced algo has not thus far been cracked.

scorpie
August 27th, 2002, 13:50
Hello VoxQuietis,

Thank you for the info about the article(s), and it seems to me that it is a good article(s).



To Scooterk,

Thank for a partial info on the subject.


To Tgodd,

Where can I find the article(s) about the SentinelPro/SuperPro have been "reversed engineer".



I believe the old Pro (with CAT 702) just deploys a PROM, so it is unlikely that it can be reprogrammed, while SuperPro should contain a kind of EEPROM (I agree with VoxQuietis that it is unlikely that Rainbow uses ASIC).


Bye,
scorpie

tgodd
August 27th, 2002, 13:57
Actually the pro is only about 7 or 8 flipflops, depending on the model.

The Cat is a Catalyst number.
Catalyst make Custom Asics.

It is most likely that in manufacturing the ID, Write Password and overwrite passwords are fused. This means that once programed for an OEM they can not be reprogrammed.

Keep in mind that this is not a processor which can be
'glitched' into submition, such as a PIC.

And as far as Tutorials; there are non on the pro and spro
internals.

A company called safekey (formerly SafeSoft) cracked those back in the 80's and early 90's, and sell emulators for these lock devices.

scorpie
August 27th, 2002, 14:39
Hello Tgodd,

What kind of Flip Flop can "store" information without power ? There is no Id, Write Password, etc. on the SentinelPro. There is only Family Code and "algorithm" (programmed by Rainbow or their distributors).

Bye,
Scorpie

tgodd
August 27th, 2002, 15:18
The sentinel pro has no memory only flipflops.
connections between the flipflops are fuses which are
blown or not blown which define the algo.
What distinguishes one Pro from another are the clock, data, and
turn on bits on the parrallel port.

The sentinel superpro on the other hand does have memory.
eeprom. When the encrypt command is sent to the lock device the latches are preloaded with data from the memory cells.

Believe me I know what it is I speek of.


I was coding lock emulators 14 years ago.

Just a note.

The earliest Sentinel Pro locks were about four to five inches in length.

Used to use discreet logic.

scooterk
August 27th, 2002, 15:39
Hi
I'm curious about this.. If a custom ASIC is used then, when you
program a hardlock for instance.. the keys are not preprogrammed, but rather programmed from the board plugged into the computer. My question is .. are the keys blank ASIC's or are they EEproms? (because they can be reprogrammed?)
And if so could you make you own keys.. (hardware) or are the chips proprietary?
scooterk

tgodd
August 27th, 2002, 21:35
When you design a custom asic you can have it design with
eeprom memory included on the die.

You can also design it so that certain memory cells are fused
so that once written they can no longer be reprogrammed.
(this was done with the super pro)

So Rainbow when they receive the manufactured lock devices,
will program in the First eight bytes of memory which contains
the Lock ID, Write Password, and the 2 Overwrite Passwords, Etc.
and these cells once preprogrammed by Rainbow can no longer
be reprogrammed.

And As for getting blanks of these IC's, It's a no, as Catalyst
will only make these Dies for Rainbow.

The SuperPro has (not including the memory) approx. 10000+
logic gates.

Not an easy device to emulated in hardware.
Unless you are skilled at designing State-Machines.

tgodd
August 27th, 2002, 21:42
Oh yes.

And as for the Hardlock The Algo unit there is 2 16bit selectors and 6 2bit selectors and about 8 or 12 flip flops.

The Hardlock die does not include the memory on the die.
However the die does include eeprom memory cells which program at a high voltage which are for the purposes of containing the encryption seeds.

The device can be programmed with the programming board, if you know how to manipulated the board to use diffent developers codes.

scorpie
August 28th, 2002, 07:58
Hello tgodd,

I am afraid you are mistaken again.

The SuperPro can be blanked again provided you can access to the programmer. You can assign a new ID, Password, and Overwrite Passwords to the blanked Superpro (of course with the Essence file you have). Once, I bought these dongles, and by mistake I got a couple of dongles with different ID. The distributor just reprogrammed them with my ID. This contradict with your statement that the SuperPro can not be reprogrammed.


Bye,
scorpie

tgodd
August 28th, 2002, 16:02
Are you sure it was a SuperPro lock device and not a Hardlock.

I continually see people posting about alladin hardlock.
Alladin only bought Hardlock.
Harlock was developed in germany by F.A.S.T


Also the activator was originally developed by a company called
Sofware Security Inc. and was later purchased by rainbow.


Most of you newbies haven't even seen a lock 4-5 inches in length, which is how big the original sentinel pro was.


If the SuperPro can be reprogrammed, then why isn't anybody doing it. Proffesionally as a lock replacement service.


Are you also going to dispute the fact that CAT on the rainbow chip is not a catalyst number?


Might I suggest getting a-hold of the largest magnifying glass you could find and have a look at the chip.


And not discounting the possibility that the superpro's first eight words of memory may not in fact be fused, as far as I know Rainbow is the only company which can reprogram the Lock ID.
They do not release (as far as I know) any devioces which allow a superpro's first eight bytes to be reprogrammed.


And do not kid yourselves people, the lock device industry is a muli-million dollar business. The development cost to develop some of these locks is in the millions of dollars.
The only reason I state this is because I have read someplace refered by many posts that the lock device industry can not possibly be that large.... It's HUGE.


I've been doing this far too long to know otherwise.

VoxQuietis
August 28th, 2002, 18:56
Dear tgodd,

please calm down. Newbies might generate a lot of obsfucation just by the lack of experience. I learned quite a little bit on the Pro/SPro during this thread.

W.r.t. the mixing of all those different dongles You are right,
that one should better be exact, since exactness is one of the
most important things to practise in S/W rev-eng.

I got in touch with the SPro about three or four years ago, i.e. I never saw the old-style Spro. The S/W protected by this little beast was in the order of 100k $. Nevertheless the implementation was poor (Flexlm plus some own onraments :-) Anyway I had the dongle free for a brute force attack on the overwrite password, which did succeed after little more than two month. Moreover I developed a filter driver (WinNT, it is a great OS) to emulate the dongle plus a logger consisting of a kernel mode driver and a Perl script. It allows me to run a log with the dongle and then compile the emulator with the data extracted by the Perl script :-)

It lacks of course a model for the query answer. Up to now this proved to be nothing more than a little bit unconvenient. Anyway I would be a huge step to have a generic solution. I will have a look on the stuff when being back from holydays.

Do You think it would be feasible, to separate a model of the query processing in a data shifting operation plus a (a priori unkown) boolean function. Then it might be feasible to run a set of (carefully selected) queries on an actual dongle to derive a table with sufficient information to synthesize the function by the means of a VHDL synthesis tool. While this procedure needed to be applied to each algorithm it nevertheless would be a major step towards the generic emulator.

W.r.t. the Pro CAT-701/2: Our sysadmin gave me an old Pro device as a gift, since he knows about my addiction. The first thing I did was cracking it up, and it contains a CAT-702 labeled chip. I'll put a photograph on the net in the next few days (despite the fact, that there is not much special on it).

Actually I am trying to extend my logger to the iButton parallel port dongle, but this thing is a mean beast. It floods You with hundreds of driver calls to derive a simple serial number. Gladly enough the driver is small (only 7 k), thus complete revering of the driver is feasible (and the approach I selected). They do sinful stuff: the store information on I/O processing into the device extension. I wonder, whether this is a safe approach for multi dongle / multi CPU machines. Anyway I don't have results up to now.

Best regards
- Vox.

tgodd
August 28th, 2002, 20:38
There is currently a company that sells a generic emulator for the superpro. Unfortunately that emulator does not currently work for the enhanced algo.

SPRO points to remember:
Algocells are two memory cells starting on an even address.
The data in the second cell is masked with 03fffh.
to activate an algo cell the data in the odd cell is masked with
08000h and to for the algo into enhanced mode the data in the odd cell is masked with 04000h.

And all new Overwrite2 passwords have the 04000h masked on.

This renders SafeKeys reader useless on any of the Newer SPros.

From my own analysis of the algo unit, it's response is based on the write password and the cell pair data. And through statistical analysis a table can be constructed to evaluate the cell contents. Once one has the cell contents one can determine the write password through an emulation of the algo unit.

I beleive that the enhanced mode throws in another variable (not sure which variable that is as yet), as well as another stage of flipflops of which the responses can not be seen they only have an effect on the Original stage. This throws the statistical analysis out the window.

Even with a super computer I calculate that it would take approx. 5 years of calculation to generically determine the Cell data on an enhanced mode algo cell pair.

I doubt that the cell pairs will be readable or read reversable.

tgodd
August 28th, 2002, 20:45
Problem with the iButton is it is sooooo timing sensitive.

VoxQuietis
August 28th, 2002, 23:10
Dear tgodd,

You say, the algo response is dependent from the write password. OK. Do You think of the dongles 16 bit write password, which resides in cell #3?

I think, this can be bruteforced within 2 minutes. Or do I miss the understanding a conceptually important point within Your explanation of how the query algorithm works?

Best regards,
- Vox.

PS: I got the best optimism, that my iButton emulator will be _extremely_ timing independent ;-)

tgodd
August 29th, 2002, 00:42
Using brute force yes.

You can in fact get the write password.
But it is a destructive method, as you have to attempt to write memory to do it.

What happens when you have a SPRO which has ALL of it's cells written as hidden or Algo.

You can Query cell 4 (Overwrite passwords) and determine
the Overwrites as well as the write password from a Query.

But in order to do this you need a working knowledge of the inner workings of the SuperPro and it only works on the older versions of the SPRO.

scorpie
August 30th, 2002, 08:36
Dear VoxQuietis,

I am curious about the Overwrite passwords "retrieval", so I have a question for you. To read the Overwrite passwords, do you treat them as 32-bits ? Write Password has been solved by the one who has the "base-ball" web-site for obsfucation (I think you know who he is) without destroying the overwrite password at all (you are absolutely right again that on the average it takes two minutes to read it).

Do you mind if I send you a sort of "personal e-mail", as I do not like to expose some "sensitive information" ?


Regards,
scorpie.




To tgodd,

I have found a biggggggg magnifier as recommended, namely a smallllll round window on the Hardlock Cover (as well as Hardlock TWIN). I can not find the magifier on the SuperPro which is reprogrammable by Rainbow Distributor who have access to the programmer.


Regards,
scorpie

Regards,
scorpie

VoxQuietis
August 30th, 2002, 11:37
Hi scorpie, (and tgodd)

bruteforcing requires the dongle to have an unused cell,
which had been the case for all the Superpro's I have
seen up to now (about five). I expect this to be the
general rule rather than the exception.

Anyway, You just run a write within a loop, until You
caught it. (if You drop me an email address -> voxquietis
@gmx.de. I'll send You the full source. there is nothing
special about it, it is just to boring to be posted in
full length)


: writeData = 0;
: cellAddress = 0x8;
: accessCode = 0;
:
: for (id = 0x0; guess < 0xffff; guess++) {
:
: writePassword = (unsigned short int)guess;
: spStatus = RNBOsproWrite(ApiPacket, writePassword, cellAddress,
: writeData, accessCode );
: if (spStatus == 0) goto FoundWP;
:
: }
: printf( " Brute force WP failed \n";
:
: FoundWP: ... display the result


Bruteforcing the overwrite password(s) works similar, but is
_very_ time consuming. With a proper amount of bad luck
You might run the dongle for little less than a year.
So You better catch an old PC and place it in the garage
(or in the lab, as I did ;-) and let it run, just looking
once a day, how the scanning goes on...

To extend the stuff for a fully programmed dongle would
be a major project, yet under the assumption that the
simple query mechanism is known it should be feasible:
One would direct queries to all the cells until finding
one, which returns such a return value. This should be
possible after directing a set of queries to that cell.
I'll have a look when being back from holidays.
Then according to the statements of tgodd both the content
of the cell and the write password can be recovered, and
the brute force attack on the OP can start.

This seems to be everything, which is possible today.
But maybe some unkown genius takes a deeper look on the device
and solves the riddle.

Bye,
- Vox.

tgodd
August 30th, 2002, 16:02
AnyBody here have a SPRO manufactured from 91-95??
Let me know...

tgodd
August 30th, 2002, 18:13
Keep in mind that the Overwrite cells can be queried to acquire not only the password, but the Overwrite passwords as well.

cah
August 31st, 2002, 04:50
How to find SPRO is manufactured from 91-95?
I have one old sentinel dongle.
Explain me, how to find.

Cah...

tgodd
August 31st, 2002, 05:06
Sentinel makes several different types of locks (dongles).

Here is a list:
- C
- Scribe
- Scout
- Pro
- SuperPro

Keep in mind that these are all based on completely different
technologies.

The SuperPro however, (this is not a well known fact) has 2 flavours and it is not possible really to tell them apart, as far
as I know, whithout attempting a query on cell 2.
The newer SuperPros will not allow a query on cell 2.
There was a hole in the original design which allowed one to query cell 2 (Overwrite passwords).

And with a working knowledge of the lock internals one can
determine the write and overwrite passwords from the older
SuperPros.

Do not confuse the PRO with the SuperPro (SPRO), they are
distinctly different.

tgodd
August 31st, 2002, 14:42
Just a note because I forgot to mention.
Rainbow also bought MicroPahr, a lock which originates in france,
and Activator which was manufactured by a company called
Software Securities Inc.

MicroPhar = only memory.
Activator = 3 Counters and memory.

Morlac
September 2nd, 2002, 09:36
Hi everyone,

Have anyone noticed that cell 5 is actually readable and not hidden like the rest of cells from 2 to 7?
The driver does not allow reading cells 2 to 7 with a simple check.
If you remove that check you can read them. Cells 2 to 7 apart from 5 will return invalid data. If I remember correctly, 0xffff are return for them while cell 5 returns 0x0000.

Could somebody check if a dongle has anything stored there?
Mine contains 0x0000 only. Its a SuperPro Sentinel from Globetrotter FlexLM application.

Morlac.

tgodd
September 2nd, 2002, 14:30
I am affraid you have yet over-looked something.

If you read the SPRO through I/O to the parrallel port
you will notice the following facts:

cell/attribute
00/1
01/1
02/3
03/3
04/3
05/1
06/1
07/1

And I have no problems reading any of the cells defined with
attribute 1.
One thing as true that cell 5 is allways 0.

As I've been saying all along, is that if you query cell 2 and get an invalid response (you can only tell by the returned data), then you have a newer SPRO. If you get what appears to be proper encrypted info then you have an older style SPRO.

mueller5321
September 3rd, 2002, 10:40
@tgodd
Do you have some information about how to change the costumer
code of the hardlock dongle?.
You are right with the comment about the reprogramming of the hardlock. it work without any trouble.

I have a CPC (programmer card of hardlock) in front of me.
The magnifier you mentioned seams to be the booster circuit
ICL7662.
But is also obvious, that the dongle chip at it self is a asic.
It have the old FAST label printed on it is produced by Philips.
So it can't be a Microchip PIC :-) so you there also right.

Additional on CPC a 93C46 is present, which seams to hold the costumer id.

Are there any more detailed information about the hardlock?

Some seams to know which algorithm are used inside some dongle, but there are no implementations of this knowledge.
(knuth, feal e.g.)

tgodd
September 3rd, 2002, 13:05
I can assure you that there are only about 4-6 people world wide that know how the fast algo works.

And when I say, know how it works, I mean can also calculate the 3 seed values. There are also some people who make claims to know how the fast algo works, but are unable to calculate the seeds, and this is because the only thing they know about the Fast is what they have seen in somebody elses code.

I have to appologize first that I can not give you any more info than what I am about to give you.
In order to program a fast you need a fast.
You will notice that there is a fast asic on the programmer board.
The fast on the board has to be re-programmed.

I am unable to tell you anything more than this.

tgodd
September 3rd, 2002, 14:46
CaH

Sorry I can not give you more.
I hope that gives you enough to play with.

neemous
September 4th, 2002, 02:58
I mean no disrespect and will probably look like an
idiot for asking....
Why? Why wont people say what they know about dongles?
When bajunny fixit meteo blew hasp3, the whole world
knew about it....or were those guys just disgruntled
aladdin workers? Or does aladdin/rainbow have hired
thugs that would find and harm loud mouths? When new
aspr, vbox or some super packer come out with a new
version everybody is all over it...and speak openly
about it...what is the difference?

tgodd
September 4th, 2002, 05:19
All of what they figured out about the Hasp was first figured out in 91 - 92, and not by them I might add.

Most of what they have documented would appear to me to have been reversed from one of these emulators from the past. After all, they did not have anything on the hasp until 94 - 95.

Now do not mistake me, I am not saying that they didn't figure it out on their own, but if the wheel exists why reinvent it.

If you look at their code it does in fact look alot like the code from the earlier emulators.

Reversing is what we all do, so no disrespect to them, no matter what they did. At least this is my view.

You also have to realize that some people have to worry about feeding family, so you can not disrespect the fact that time spent, accounts for quite an investment.

So you see why some people can not be totally forth-right with information.

Believe me, I wish I had the Hints when I first started.

JMI
September 4th, 2002, 05:20
neemous:

Perhaps you noticed that tgodd stated he was "unable" to provide the information, not that he was "unwilling." You might give some thought to the possibility that he is under some "legal" constraint from disclosing certain information. This is just a guess on my part, but that is what it sounds like.

Regards.

tgodd
September 4th, 2002, 05:21
Yes you are right JMI.

I do have some legal obligations not to be overly forth-right.

tgodd
September 4th, 2002, 05:26
Alot can be said for the FUN of the chase, so to speek.

It is alot more fun to figure it out on your own, rather than
have the answer presented to you on a proverbial platter.

cah
September 19th, 2002, 08:40
To,
All Dongle RE,s

How to find & analyse of Sentinel dump, their algo/queries & their responses and seed codes.

If any one, publish an article on sentinel emulation, It will be very useful for newbies.

Cah...

melinelmago
September 23rd, 2002, 18:47
For read the sentinel superpro dongle I used sprodump.exe from +spath, and I known that the program search the write password but I don't know how.

Anybody know how to search the Write Password?

Thanks,
Melin

paulofutre10
February 14th, 2004, 10:27
Can you give me the email , of VoxQuietis

I need to do a brute force attack to the overwrite password , of sentinel superpro

Thanks
Paulo Futre

paulofutre10
February 14th, 2004, 10:49
I went to www.safe-key.com and download sentread.exe ( for read the sentinel superpro) , its generate a file . I open that file with a hex-editor .

in the 03C0h is (exemple) 0001 E08B 0000 0000 3254 0000 0000 0000
1º - Is the number of dongle ( i´m not sure)
2º - ID is 8BE0
5º- The Wirte password is 5432

I think that the 3º and 4º is the overwrite password , but the reader can´t read .


CAn you help me find the overwrite password ?
I read in this forum , that is a way with brute force attacak , Can you teach me how to do ?

tgodd
February 21st, 2004, 14:46
I beleive that in an earlier post, I had mentioned that the SuperPro Algo on the new generation Superpros had not yet been cracked.

I lied.


tgodd

Cata
March 5th, 2004, 02:57
Hi everyone,

Is it possible to recover overwrite passwords without brutte force for
superpro new generation?
I read all post about it but I don't have a clear answer.

Thanks for replaying!!!!!

PS After digging in sentinel driver seems the final check is in dongle.
(exists a check in driver but is eassy to bypass and it's not enought).

tgodd
March 5th, 2004, 21:19
Quote:
[Originally Posted by Cata]Hi everyone,

Is it possible to recover overwrite passwords without brutte force for
superpro new generation?
I read all post about it but I don't have a clear answer.

Thanks for replaying!!!!!

PS After digging in sentinel driver seems the final check is in dongle.
(exists a check in driver but is eassy to bypass and it's not enought).


NO

Cata
March 10th, 2004, 02:42
How is it possible to disable an alghorithm? (opposite of Activate function)

Thanks!

nikita@work
March 26th, 2004, 18:34
Quote:
[Originally Posted by Cata]How is it possible to disable an alghorithm? (opposite of Activate function)

Thanks!


It can be done only if you know algo descriptor and know overwrite passwords... both can be bruteforced in several days if there is a working dongle!

korvak
April 13th, 2004, 09:19
nikita is there a source for the program that will brute force the "Over-Write" password for the Sentinel SuperPro. i found the one by spath+ that will brute the "write" password, but i have not found one that will brute the "Over-Write" password. any help in pointing me in the write direction here has my many thanks.

korvak

nikita@work
April 13th, 2004, 11:58
Quote:
[Originally Posted by korvak]nikita is there a source for the program that will brute force the "Over-Write" password for the Sentinel SuperPro. i found the one by spath+ that will brute the "write" password, but i have not found one that will brute the "Over-Write" password. any help in pointing me in the write direction here has my many thanks.
korvak


as far as i know there are no freeware tools, but you can reverse sspro driver and rip all you need... if strip all checks it will increase speed in several times...

sapu
May 13th, 2004, 13:13
Edited by Woodmann