I am posed with a problem which I have no idea how to solve, essentially I have these Files which one may enter a "write" or "read" password (these passwords are different), and I need to Encrypt and Decrypt these files using a common algorithm such as AES.
I know how to use crypto libraries (I will be using the ones provided in Java), however, I am unsure how to make this secure, or how to implement this.
I originally thought of storing salted hashes of these passwords for write/read, and then checking the entered password, however, I am unsure how the Crypto would work, if I have a keystore in say Java, won't it need a password so the keys don't appear in plaintext; I cannot have the keystore password in plaintext anywhere and I am not allowed to just make one up. Is there any well known solution to this problem I am positing which I can study?
Essentially: is there anyway to get access to encrypting or decrypting using a key derived from read/write passwords DEPENDING on the type of password entered, i.e. a user entering a READ password cannot encrypt, just decrypt.
Public-key cryptography has two separate keys, one to decrypt, one to encrypt. Usually, one of the two is public, though. In your case, you'd want to keep both private, and protect them with your two passwords. I suppose that would work for RSA (where the public key cannot be derived from the private key, which in other schemes it could).
So you'd be creating an RSA keypair, and store the two keys separately, protected with the two passphrases. That way you can control access to them individually.
Use RSA encryption to encrypt and decrypt an AES key. Use the AES key to encrypt and decrypt the content file.
The Java Keystore API doesn't support encryption of public key entries, but it does encourage the use of a password to protect the integrity of public keys. Your application could be written to use the "encryption" password to load a key store with the "public" encryption key, and fail if the wrong password is entered. Of course, a user can circumvent this outside the application, and get access to the public key using something like keytool or their own custom code. But you can't really stop someone from performing encryption with some key; why do you care if they use this key?
Private key entries would be encrypted with the "decryption" password. This common use case is well supported by the Keystore API.
If you are not allowed to store the key(s) elsewhere, I would store them in the encrypted file. I would encrypt the file with AES and then encrypt that key (the "file-key") twice by two AES keys (read-key and write-key and store:
[length-of-next-section] [file-key-encrypted-by-read] [length] [file-key-encrypted-by-write-key] [length] [file-data-as-encrypted-by-file-key]
FWIW, this is akin to how MAC OS X's "File Vault 2" whole disk encryption works -- the disk encrypted with a key, and 1 or more people's passwords are used to encrypt that key, thus any of these person's passwords can be used to "unlock" the disk at boot time.
Related
I am looking for a best-practice or standard method to password protecting a console application. I have researched various methods and would like some feedback on my approach.
I have decided to hash my password using Argon2 so that I only have to store the one-way hash. Everything is working as expected. The question I have is where do I store the hash? Should it be hard-coded? Should I store it in a separate file and read it in? What is the most secure way to approach this? At the end of the day I am writing this application to learn and would very much like to learn to do it the correct way. Links to any relevant reading material would also be appreciated. I continue to google...
EDIT: So what would the potential drawbacks be if I stored program password as hash in a file. The user would have to know the password to use the application. Then let the program password that is protected by the hash be the encryption key to secure the sensitive information? Even if the source code and/or hash file is manipulated, the sensitive data would not be readable since the correct password is used as the key...what am I missing?
First of all, Argon2 is a fine key derivation function for turning passwords into encryption keys.
However, if you are using the Argon2 hash as an encryption key, then don't store it on disk, obviously. If you store the encryption key next to the encrypted data you might as well not encrypt at all. One could even argue that it's worse, because it gives a false sense of security.
Properly encrypted data is useless without the key, so you don't have to protect the application itself. Just ask for the password if and when you need to encrypt or decrypt something. You can consider keeping the hash in memory for a while so you don't have to ask for it repeatedly, but don't persist it.
This is exactly how GPG works, for instance. It doesn't store any password hashes anywhere. Instead it stores private keys encrypted and just asks for the passphrase if it needs to decrypt a private key.
I finally figured out how to perform a Diffie-Hellman key exchange in Java, the problem now is that I have to encrypt a txt file using the shared key. I do not know how to even start to do that. The txt file is just simply the integers 1 2 3 4 5 each on a new line. Is there a simple line of code I do not know about that can perform this function? The shared key that I have is 18.0.
encrypt a txt file after obtaining a public key shared between a client/server in Java
Let's assume you are working on a learning or toy project, otherwise there's no point of (re)inventing your own secure communication. If you want to take the security seriously, just use TLS/SSL (optionally with 2-way client auth ssl) for online content transmission.
I really hope for your own sake you are just learning or playing with this.
to encrypt a txt file using the shared key. I do not know how to even start to do that.
For encryption in rest (encrypting the file itself) - assuming from the question that the sender and receiver are having their own keypairs.
I suggest to create a random data encryption key to encrypt the content. Then you could simply use receiver's public key to encrypt the data encryption key and sign the content's hash using sender's private key. Never use plain (textbook) RSA, use propper padding.
I made a few notes about encryption for myself, you can take some inspiration.
The shared key that I have is 18.0.
Apprently you are using not long enough or not random enough public keys. Symmetric ciphers require keys of fixed length (e.g. AES requires 128, 192 or 256 bit keys). You could generate longer keys using any key derivation function or hash. If you want to use two-way communication, you need a separate key for each direction.
Regardless that - short key seed (small shared key, short public keys) would be feasible for brute-forcing. There are many ways how you could shoot yourself in your leg when dealing with crypto, so use some industrial standard when possible (e.g. TLS).
I am trying to encrypt a list of strings specified by the user. The AES key used to encrypt and decrypt this data must be saved on the device so it can be used to decrypt the data whenever the user needs to. I have read that using Keystore is a good option for storing keys, but most of the guides I have read about it relate it to signing your application, which is not exactly what I want. My application requires the user to input a password, which then decrypts the String data using the key which is unlocked with the password. How do I do this?
Thanks.
Its hard to answer your question without having you define "securely" in a precise manner. Are you talking about security against other apps? Security against other users? Security against theft of device? Or security against loss?
In addition, what level of security are you talking about? Is the data being encrypted under the given key sensitive enough to warrant serious considerations about how the key can be stored?
To my understanding, if the data is of highly sensitive nature, you should derive the key directly from the password using a password-based key-derivation function such as PBKDF #2, Bcrypt or Scrypt. In this way, the key material needs not be stored on your device directly.
If you intend to store the master encrypting key for whatever reason, encrypt under the password key and store as a Base64 string under SharedPreferences with MODE_PRIVATE to protect it against other apps. Even if the key was recovered by a malicious user/device theft, without the proper password key, the key material itself is useless.
If you wanted to protect the key against loss, you shouldn't store the key on the device but rather on your own server, again encrypted under the password key so that database compromise will not result in data compromise as well.
In all, I'd suggest you try not to store any sort of private key onto the device itself and instead used the password-derived key as your master key for all cryptographic operations. Leave the keystores for private-keys generated in public-key cryptography and known public key certificates.
I want to encrypt and decrypt a file containing usernames and passwords. Since this file will be accessed to check if the username and password entered is correct, I'm sure that generating a different key each time the program starts will not work, since it will decrypt the file incorrectly (because it was previously encrypted with a different key). Is there a better way of doing what I want to do? I want to just provide a static key here:
cipher.init(Cipher.ENCRYPT_MODE, KEY_HERE);
But I'm not sure what the data type for that parameter is, if ever that would work. I appreciate any help.
You don't want to use a reversible algorithm for passwords. Read this excellent article on the subject
Does anyone know good tutorials to change PBEWithMD5AndDES encryption algorithm to AES for a Java application? Specially , I want to know what precautions I should take while changing this algorithm to more secure one. Any important test cases to check before and after algorithm changes. Another question is since I have used PBEWithMD5AndDES , most of the user passwords are encrypted using that algorithm. So if I change my algorithm to AES , how do I make sure that decryption of passwords happen with old algorithm while I can still use new algorithm for any new encryption.
Normally you wouldn't encrypt a users password, you'd just hash it with a salt instead.
Migrating from one encryption system to another is going to be a bit of a pain, as I see it you have two options:
During the upgrade process decrypt then re-encrypt all the passwords
Add a flag indicating the encryption method used. All existing passwords will obviously be set to the current standard. New users will be set to whatever method you choose and you can migrate other users when they change their password.
If you've already got data encrypted in format a, and you want to start using another encryption scheme, b, I can think of two ways to accomplish this:
Decrypt all of your data and re-encrypt it using `b`. This approach would be good when you can take your data store offline and "fix everything at once."
For each item you attempt to decrypt, try to decrypt it using `b` first. If that fails, decrypt it using `a`. The next time you try to encrypt something, make sure you use `b`. This approach could be used when you can't take your data store offline, but you want to encrypt all of your data using another algorithm. All of your data will eventually be encrypted using the other algorithm.
There's really no problem changing algorithms. What you need to do is decrypt the cipher text and then encrypt the resulting plain text with the new algorithm. That's straightforward. If you are going to perform this transition over time, I would suggest creating a new database table that keeps track of whether a particular entity (based on unique id) has been transfered to the new algorithm. If it has, then you simply use the new algorithm to decrypt it and you can forget about it, if not, then you use the old algorithm to decrypt it. Regardless though, all new encryption should be performed with the new algorithm.
Now, there's a second issue here. Why are you bothering to decrypt passwords? Just save the hash of the password and forget about it. If you are able to decrypt passwords, you introduce a potential vulnerability. If a malicious user can get a hold of your key you use to encrypt those passwords, then they could access the plain text of the password. Not only could the user then use that information to compromise your system, if your users use the same username/password combination for other sites, those accounts would be compromised as well. You should only store a hash of the password (SHA is a good one, don't use MD5) and then when the user attempts to log in, you hash the input and compare the two results. You have no need to know what the plain text password is.
you may look into ESAPI - java http://code.google.com/p/owasp-esapi-java/
ESAPI 1.4 was using PBEWithMD5AndDES, but in 2.0 they introduced AES
check their mail chain here
you may check the difference between the two implementations
PBEWithMD5AndDES is a method of taking a user's password and from it deriving an encryption scheme that can be used to protect further data. It is not a method of verifying a password, nor of encrypting one.
If you are only interested in password validation, then decrypt the passwords and replace them with a secure hash and in future match the hashes. You will also need your password reminder service to a password reset service.
The question is where is the password you are passing into the PBE algorithm coming from? If it is a fixed password for your application, then you just need to replace it and perform some kind of rolling upgrade. As an observation, if you are storing encrypted data as text, either hex or base-64 encoded, there are characters that cannot appear in the text output and which you can hence prepend to indicate a newer encryption scheme. For example the : character does not appear in base-64. That will allow you to identify what has been upgraded and what has not.
If the passwords are coming from the user, then each user has their own password derived cipher. In this case you can only re-encrypt whatever data has been encrypted with the user's cipher when the user provides their password.
The most direct replacement is going to be along the lines of PBEWithSHA256And256BitAES. Unfortunately, this is not supported by Java 6, so you will need a 3rd party JCE library such as Bouncy Castle. Bouncy Castle offers PBEWithSHA256And256BitAES-CBC-BC, which would be a suitable replacement.
The process of upgrading the cipher is a challenge. Whatever data has been encrypted with DES can only be decrypted with the user's password. I assume you do not have access to the passwords. This means you can only re-encrypt the data when the person who knows the password provides it. You are going to have a long period of time when your system contains a mixture of ciphers, so you need a way of identifying what is converted.
If we are talking about files, you could change the file suffix, or the folder they are stored in. If we are talking about BLOBs in a database, you could add an extra column to the database table to say what the encryption method is. If neither of those are possible you could add some form of header to the data to indicate that it has been encrypted in a new way. That's slightly risky as your existing data has no header and there is an outside chance it will match the new header by accident.
It may also be advisable to keep a list of which users have not yet had their data converted so you can prompt them to convert.