05-28-2017, 04:49 PM
yeah, OpenCL support might make sense.
But there are some questions still open:
- what about the lenght of the data, is 1 block enough ?
- you need way better means to determine if the decryption was successfull, not just a test for a single byte (0x35 == "5"), this will produce way too many false postitives, there must be some other important/hard-coded etc data within the decrypted data, or?
- is it really worth the effort? as far as I understood, you only have 1 single "hash" that you need to crack... maybe it is better to try to remember the password and maybe generate some very specific dictionaries (small ones, for instance with the help of hashcat-utils etc) that you use as input to your rust code?
- it's also very important that you carefully test that you got the algorithm correct (yeah, also your rust code etc) with some additional test vectors that you should generate with the tool that was used to generate the encrypted data (the JS library)
But there are some questions still open:
- what about the lenght of the data, is 1 block enough ?
- you need way better means to determine if the decryption was successfull, not just a test for a single byte (0x35 == "5"), this will produce way too many false postitives, there must be some other important/hard-coded etc data within the decrypted data, or?
- is it really worth the effort? as far as I understood, you only have 1 single "hash" that you need to crack... maybe it is better to try to remember the password and maybe generate some very specific dictionaries (small ones, for instance with the help of hashcat-utils etc) that you use as input to your rust code?
- it's also very important that you carefully test that you got the algorithm correct (yeah, also your rust code etc) with some additional test vectors that you should generate with the tool that was used to generate the encrypted data (the JS library)