🚀 go-pugleaf

Hello,

I am doing a program using the PGP 6.5.1i source code to do a tool that
brute-forces some private-key algorithms. I've read the OpenPGP RFC and
found that I should check the 9th and 10th bytes of plainttext in order to
check for a (maybe) correct password.

I've done that for IDEA with success, but now I am stick trying to do it
also for CAST5 (or CAST128). I am using the following code:

  // MD5 implementation, nextPass is the tried password
  MD5Init (&md5Context);
  MD5Update (&md5Context, (unsigned char*) nextPass, strlen(nextPass));
  MD5Final (&md5Context);

  // now we decrypt the first 10 bytes of the file, the random ones, using
CAST5  implementation from PGP source code
  // md5Context.digest is originally declared as byte[8] and byte as
"unsigned char*", the same as PGPUInt8*, but I'm using different source code
for the MD5 implementation (a fast one)
  CAST5schedule(xkey, (PGPUInt8*) md5Context.digest);
  CAST5decryptCFBdbl(xkey,0,0,0,0,lFirstTen,firstTenDeciphered,1);

Well, I've readed the PGP source code and learned the OpenPGP by heart and,
as far as I know, this should work (as it does for IDEA). But it doesn't.
Trying to decrypt a file ciphered with the "llois" password, I get:

Cast SBoxes Valid  (this is a test)
CAST5 test OK      (this is a general encryption-decription test, it does it
Ok)

MD5(llois) = C5 9B BF DB C6 D3 4E D2 DF 35 1B 69 F9 8B 1A FD  // MD5 for
"llois", checked
Key       = C5 9B BF DB C6 D3 4E D2 DF 35 1B 69 F9 8B 1A FD  // key used for
CAST5
10b       = F2 DF 6D 94 64 C9 A9 44 23 B8  // 10 bytes of the ciphered file,
they are the correct ones (checked)
CAST5-10b = 78 D0 F7 D7 A8 94 9F C4 A9 B7  // supposed plaintext once
decrypted

As you can see, the bytes aren't the same althoght the password is correct.
I have done a lot of testing but I must say that Im a little bit stuck by
now. Any help would be terribly appreciated.

Thanks in advance,

C. J.