Decrypt AES with OpenSSL & Qt 5.5.1 Win32 VS2013
-
Do you mean the VB code ? Then you generate an IV in there, and use a different one when decrypting your string or am I missing something there ?
-
It is the same iv. There is no problem with the iv i believe, it does work to decrypt in openssl and VB. It is a common way to use the first 16 bytes of a string for iv. Like salting passwords, kind of, not really but i'm no expert and can't explain it better. Defending on the software and api you may have to remove the iv from the string before decrypting but i don't thing it is the case in qca - but i did try it anyways - no luck.
Maybe i don't use the API correctly. The example code:
cipher.setup( QCA::Decode, key, iv ); QCA::SecureArray cipherText = u.append(f); QCA::SecureArray plainText = cipher.update(cipherText); if (!cipher.ok()) { printf("Update failed\n"); } printf("Decryption using AES128 of [0x%s] is %s\n", qPrintable(QCA::arrayToHex(cipherText.toByteArray())), plainText.data()); plainText = cipher.final(); if (!cipher.ok()) { printf("Final failed\n"); } printf("Final decryption block using AES128 is %s\n", plainText.data()); printf("One step decryption using AES128: %s\n", QCA::SecureArray(cipher.process(cipherText)).data() );
So cipherText here is u.append(f); the update and the final results together.
Or maybe i found a bug. Who knows? But chances are i'm doing it wrong, somehow.
-
That's what I'm trying to clear up with you.
What you wrote is that you were using a part of your already encoded string as IV when setting up decoding and that's that part that is puzzling me. AFAIK, you should use the same IV that you generated, whatever the means, when you encoded that string. So basically (pseudo code):
Get secret key as sk Get password Generate IV as iv: 16 first chars of md5 of password Encode "Dr. Test" with sk + iv as encoded_str
and later:
Get secret key as sk Get password Generate IV as iv: 16 first char of md5 of password Decode encoded_str with sk + iv Get Dr. Test
What I understand of what you wrote for the decoding part is:
Get secret key as sk Get password Generate IV as iv: 16 first char of encoded_str Decode encoded_str with sk + iv Get garbage + Dr. Test
-
I did not wrote the VB code nor the encoder or decoder for this part. I just try to decrypt strings in XML files encoded using this VB code. For now i use a bash script on Linux but i fail to implement it in my Qt app so far.
Bash:
passmd5="$(echo -n "$1" | md5sum | cut -d '-' -f1 | tr -d '[[:space:]]')" theiv="$(echo $2 | xxd -l 16 -ps)" echo $2 | openssl enc -d -a -A -aes-128-cbc -iv $theiv -K $passmd5 | tail -c +17
Works perfectly. No luck in Qt.
-
Can you give a sample of input you pass to that script and what you should have as output ?
-
Of course:
#!/bin/bash password="test" # the password cstring="PpUr+LMHvaKmf0q6J7Oyzo4jbFO5kfWyXl0d8nD3hyM=" # the aes-128-cbc and base64 encoded string passmd5="$(echo -n "$password" | md5sum | cut -d '-' -f1 | tr -d '[[:space:]]')" # md5 hash the password theiv="$(echo $cstring | xxd -l 16 -ps)" # get the iv decoded="$(echo $cstring | openssl enc -d -a -A -aes-128-cbc -iv $theiv -K $passmd5 | tail -c +17)" # decode in openssl # expected: Dr. Test echo decoded: $decoded
-
Based on your script
QByteArray data = "PpUr+LMHvaKmf0q6J7Oyzo4jbFO5kfWyXl0d8nD3hyM="; QByteArray iv = data.left(16).toHex(); QCA::Hash hash( "md5" ); hash.update("test"); QCA::SecureArray key = hash.final(); QCA::Cipher cipher(QString("aes128"),QCA::Cipher::CBC, // use Default padding, which is equivalent to PKCS7 for CBC QCA::Cipher::DefaultPadding, // this object will encrypt QCA::Decode, key, iv); QCA::SecureArray decryptedData = cipher.process(QByteArray::fromBase64(data)); if (!cipher.ok()) { qDebug() << "Decryption failed ! "; } qDebug() << decryptedData.toByteArray().mid(16);
Note however that using tail like that looks suspicious. I'm still not convinced of your initial vector handling. It should be the same used when encoding the string. Here it clearly is not.
-
But again and I insist the IV handling is wrong. You should use the same IV for both encoding and decoding and following your code it's clearly not the case.
The IV should change each time you encode your data though to avoid repeat attack.
-
I am not involved in the encoding part. I did not write one single code for encoding. I get 3rd party XML files and i just want to decode them. To be very clear on this: I can't change the encoding, it is not my software, i do not develop it and i never had. I just got the source code for the de and encrypt part, so i may be able to import the XML files. I never wrote a single line VB code in my life. To be honest, i don't believe security is the issue, the developer just don't want anyone to be able to open it. That may not be an excuse for bad code but so far, i can't change it, i'm just happy now to be able to decode.