[qazwsx]

What we need is a CTC computer

Someone go and invent one please

What we need is a quantum computer, by the time this get's cracked on a normal processor, quantums might be the norm.
Some of you might say that quantum computing is years off, the truth is that they are becoming a reality, it might take a while, but I'd be surprised if they didn't appear some day in the not-too-distant future.
http://www.newscientist.com/article/dn3114-quantum-computing-making-tremendous-progress.html

[DarthWang]

Quantum computers are only better at solving certain types of problems, and not necessarily faster.

CTC would do it instantly

[qazwsx]

Ah, but you'd still have to leave a CTC running to come up with the answer, god knows how long it would take.
Quantum computers are good at breaking encryption because (in theory) they can do thousands of processes similtaneously, they can try many different passwords/keys at the same time, individually, these might be the same speed, or slower than trying a password on a normal computer, but because many attempts are being done at the same time, it becomes a hell of a lot easier.

[Stormstack]

The idea with a quantum computer is that it exists in multiple states at the same time. Like Schroedinger's cat being alive and dead.
The computer exists in a different state for each possible key, at the same time. So, it can try an infinite amount of keys a the same time.

[DarthWang]

No, it doesn't work that way. There are only 3 possible states: 1, 0, and superposition where it could be either 1 or 0. It's not necessarily faster than a normal computer, just better at solving certain types of problems. To solve this encryption, it would take the square root of all possible combinations.... still a hell of a lot of time.

Whereas with CTC, you would get the answer as soon as (or even before) you asked the question. Breaking the problem up into smaller steps makes it more efficient, as well.

However CTC is completely theoretical while there are actually quantum computer prototypes, so the point is rather academic.

[Stormstack]

There are ore than 2 possible states, because you are usually dealing with more than 1 bit of data at the same time. There is certainly more than 1 transistor in your processor, so the possibilities increase more and more. The computer may only work with 1's and 0's, but that doesn't mean that it can only be in 1 of 2 states at the same time.
Say you have a processor with (for simplicity) 3 transistors, you can have:
1 1 1 or
1 0 0 or
1 1 0 or
1 0 1 or
0 1 0 or
0 0 1 or
0 1 1 or
0 0 0

More than 2 possibilities, no?
So you get every transistor in a supposition of both states, giving you a number of states as near to infinite as makes no odds. A Core 2 Extreme QX6700 has 582 million transistors so there are 2^582000000 possibilities.

[Androgeos Exeunt]

I'MA FIRIN' MAH LAZER!

[tinfoil]

i thought we were trying to crack a password

[Droid803]

yeah, but you know, thread derailment happens. All the time.

[tinfoil]

i know. i happen to be quite good at it. however this is important and it would be good if we could come up with a solution rather than spam and **** that will just get this locked.

[FUBAR-BDHR]

Well I'm well over 10 Trillion passwords tried so far with one software and I don't even know how many with the other (found a faster program a week ago) still nothing.  So far I've pretty much ruled out all 5 character combinations.  Every combination of 6 characters between ASCII 32 and 127.  7 character run is up to thing beginning with ". 8 and 9 character word combinations  of the same sets are running but as you can imagine will take quite some time.  I also have 1 running 8 character combos of everything between 32 and 127 except caps.  All number combinations up to around 500000000000 are also out.  I'm not giving up though.  I'll let them run for years as long as the gods of power and hardware allow it.  Maybe I'll have it by the 20th anniversary of FS. 

Forgot to add I'm processing around 15 million passwords a second now.

[Bob-san]

I stopped--though I should probably continue. I was up a good deal of the way on ALL combos. I was thinking of something like bring it to a demonstration and have one of the password-hackers go at it for a few minutes--get us a definite answer!

[FUBAR-BDHR]

If you know of something that will run faster that about 4 million a second and is free let me know.  I have more processing power but the programs just aren't using it.  The fastest one uses an ini file so I can't run more than one copy on the same PC otherwise I could possibly be hitting 12 million as second on the one PC.

[haloboy100]

and that's bad?

[FUBAR-BDHR]

Well if I could quadruple it on 1 computer and double it on 3 others that is a significant decrease in the time it would take to crack it. 

[Tyrian]

But that's assuming that you could coordinate between the machines to avoid them retrying each other's combinations.

[FUBAR-BDHR]

That's the easy part.  You just run them with different first characters and let them run until they the first character has already been tried.  Say start one with a another with A another with m, M, 1, special characters, etc.  Might get a few hours of overlap but that's about it. 

[Bob-san]

I was running 12mil/second or so with Passware. Shoot me a PM and we'll see what we can come up with.

[Fabian]

I can now rule out WinZip as encryption tool, as the Zip file uses minimum version of 1.0 to extract stored files and minimum version of 2.0 to extract deflated files.

Also all WinZip versions of that time span (2002) (6.3,7.0,7.1,8.0,8.1) do not write an extended local header, which is also present.

This means that all attacks depending on the extremely weak PRNG of WinZip earlier 8.0 do not work.

The PRNG still is quite weak and the starting seed is (if Info-ZIP compatible was used) also guessable in between time(NULL) ^ PI as PI is used on win32 for xoring.

Also we know the plain text bytes of the 4 stored files that most probably start with the same header (PK...).

One file is unfortunately not enough as the time byte must be guessed correctly. (However this is only a manual brute force of 255 combinations).

However while this is enough to rule out the possible key combinations no published source code program exists, which uses plain text in several files at once.

So one you guys with the fast machines, you can try to get pkcrack running (on windows) and do a known some plain text attack on tables.zip with:

2d b0 50 4b 03 04 14 00  0a 00 08 00 ??

This assumes that the first file in tables.zip was deflated (08 00), is compatible using version 2 (14 00), and was using maximum compression settings and extended local headers (0a 00), using not another password (again 0 bit in 0a 00 is 0).

The other possibility would be:

2d b0 50 4b 03 04 14 00  08 00 08 00 ??

(Using standard compression settings, but still extended local headers)

Quite another possibility if another program was used for the other zip files is:

2d b0 50 4b 03 04 14 00  00 00 08 00 ??

(if no extended local headers were used)

The 2d b0 can be found in TM.zip as the time of day (in standard DOS format) (which is used instead of CRC-32 for files, which use extended local headers) and assumes that the zipper used for encryption was writing two checkbytes, however being compatible with version 1.0 this might hold true.

pkcrack needs to be started as follows then:

./pkcrack -C TM.zip -c tables.zip -p pt.zip -o -2

assuming that the pt.zip contains the bytes written above.

Another attack could assume that the zip file was packed together after the files were copied as we can see several files with 01 date have almost the same time stamp:

2d b0 50 4b 03 04 14 00  0a 00 08 00 ?? b0 36 2b

would then be our search string for tables.zip assuming that the first file in tables.zip has a last-modified date of approximately the same time as when it was packed.

(Risky but 4 times faster!)

That is all I can find for now,

Best Wishes,

Fabian

[FUBAR-BDHR]

Can you give those instructions in some no alien language?  Every time I try to run pkcrack with anything there I either get a message that the file isn't plaintext or the instructions for running pkcrack which seem to not even work when copied right from the readme for pkcrack.