| |
| | [No title] |
 | | One approach might be to define SHA1 in terms of bit operations where Ak_j is the j-th bit of value A on round k, and xk_j is the j-th bit of input xk and include all of the 32 bit carries, rotations, xor, or, and negations in that overall equation set. |
 | | The fact that SHA1 encodes the length of the input that is hashed in the last 64 bits (Merkle-Damgard stengthening) will prevent you from successfully using this trick to find collisions with the whole SHA1. |
 | | So, while the intermediate state of SHA1 after processing a or ax will be the same, their final hashes will have to be different, because the two input messages aren't the same size. |
| www.cypherspace.org /adam/sha1int/readme.txt (2751 words) |
|