save calculated hashes or check against partial hash

[r00t]

hello guys, 
  
Is there any way to save hashes calculated with oclhashcat? I found a thread from 2013, which tells me it's not possible. Did something changed since then? If not, can I somehow find all strings corresponding to my partial hash?

my problem: 
I have the first 10 digits of a hash. The original string is double hashed to generate the corresponding hash, from which I only have the 10 first digits. I know there are about 4 billion possible unique variations from the original string. Now I need to try them all within a reasonable time and decided to use oclhashcat to do so. Is oclhashcat capable of solving my problem?  


Yes a lot of questions, sorry for this and I hope someone have a solution for me


//edit: it's using md5 for hashing

[epixoip]

Two part question, two part answer.

Is there any way to save hashes calculated with oclhashcat?

No.

I have the first 10 digits of a hash. The original string is double hashed to generate the corresponding hash, from which I only have the 10 first digits. it's using md5 for hashing

If it's stored in the database as a truncated hash you could easily write a cracker that could generate a collision within minutes.

If it's not stored in the database as a truncated hash and you just somehow "lost" the other 22 characters, then there's nothing you can do to get the original hash back. Even if you did generate a list of every possible double MD5 hash that started with those 10 characters, there will be thousands of collisions.

[r00t]

umm the other 22 digits are "lost" somehow, no way I can figure them out, but probably I should explain the structure of the original string a bit more. it looks like this:

Code:

<ip-address><proprietary string>

the <proprietary string> I'm able to figure out. The <ip-adress> is the missing part, but I know that the ip-adress is constant. 

so if I "create" a cropped hash from the original string through my proprietary application and use hashcat to crack it, I will get a set of this form as output from hashcat:

Code:

hash: <ip-address_1><proprietary string>
hash: <ip-address_2><proprietary string>
hash: <ip-address_3><proprietary string>
...

Since I have full controll over the <proprietary string> part (the only thing I'm able to influence), I can create another, different cropped hash by changing the <proprietary string> field and this time only use the ip-adresses from the hashcat output for cracking and repeat.

using this approach I should be able to determinate the ip-adress in a reasonable time, shouldn't I? The only thing I need is a fast way to create these sets of

Code:

hash: <ip-address><proprietary string>

Thank you so far

[epixoip]

Maybe. Hashcat won't help you with this, but you could just write a quick cracker and find out. You don't need to save the output of anything, just write a cracker that tests if substr(md5(md5(str)), 0, 10) === cropped-hash.

[r00t]

ok, thank you. It's too bad that I can't use hashcat for this, something with GPU / Multi-threading capability would be awesome in boosting my calculations, since I have to do this for about 30 IPs. If somebody has a hint how I can boost my calculation speed on windows/linux and with an amd card, that doesn't take to much effort (e.g. already written python+pyOpenCL code that need some small modifications), I would be glad to hear it.


greets

r00t

[undeath]

You can modify jtr.

[r00t]

ok thank you, I'll see how far I'll get with it, but there's a lot of code.