pegh is a file encryption tool using passwords with modern, standardized, and authenticated encryption. It is simple, secure, and returns proper exit codes so you can tell whether encryption or decryption failed or not.
pegh is Klingon for secret
# encrypt file.txt to file.txt.pegh with password SUPER_SECRET_1942 pegh -e SUPER_SECRET_1942 <file.txt >file.txt.pegh # decrypt file.txt.pegh to file.txt with password SUPER_SECRET_1942 pegh -d SUPER_SECRET_1942 <file.txt.pegh >file.txt # encrypt file.txt to file.txt.pegh with password from password.txt pegh -f password.txt -i file.txt -o file.txt.pegh # make encrypted backup tar czv -C /path/to/dir/ . | pegh SUPER_SECRET_1942 -o foo.tar.gz.pegh # extract encrypted backup, the "I'm ok with truncated data" way pegh SUPER_SECRET_1942 -d -i foo.tar.gz.pegh | tar xzv # safely extract only complete encrypted backup, the "I have more space than time" way pegh SUPER_SECRET_1942 -d -i foo.tar.gz.pegh -o foo.tar.gz && tar xzvf foo.tar.gz; rm -f foo.tar.gz # safely extract only complete encrypted backup, the "I have more time than space" way pegh SUPER_SECRET_1942 -d -i foo.tar.gz.pegh >/dev/null && pegh SUPER_SECRET_1942 -d -i foo.tar.gz.pegh | tar xzv
The easiest way to scale cost/time it takes for bruteforcing is simply to continue doubling -s, on both encryption and decryption commands.
$ pegh -h usage: pegh [options...] password -e encrypt input to output, default mode -d decrypt input to output -i <filename> file to use for input, - means stdin, default stdin -o <filename> file to use for output, - means stdout, default stdout -a append to -o instead of truncate -v pegh file format version to output, either 0 (AES-256-GCM) or 1 (Chacha20-Poly1305), default: 0 if AES is hardware accelerated, 1 otherwise -c <mb> chunk size for encryption, while decrypting/encrypting twice this ram will be used, the same amount will be needed for decryption as encryption. This value is saved in the file format, so decryption will fail if this isn't set high enough, these are only allocated after scrypt is finished so max usage will be the highest of these only, not both combined, max: 65535 (AES-256-GCM) or 262143 (Chacha20-Poly1305), default: 32 -m <max_mb> maximum megabytes of ram to use when deriving key from password with scrypt, applies for encryption AND decryption, must almost linearly scale with -N, if too low operation will fail, default: 64 -f <filename> read password from file instead of argument, - means stdin -N <num> scrypt parameter N, only applies for encryption, default 32768 this is rounded up to the next highest power of 2 -r <num> scrypt parameter r, only applies for encryption, default 8 -p <num> scrypt parameter p, only applies for encryption, default 1 -s <num> multiplication factor to apply to both -N and -m for easy work scaling, rounded up to the next highest power of 2, BEWARE: -s 32 requires 2G ram, -s 64 requires 4G and so on, default: 1 -h print this usage text -q do not print error output to stderr -V show version number and format version support then quit For additional info on scrypt params refer to: https://blog.filippo.io/the-scrypt-parameters/ https://tools.ietf.org/html/rfc7914#section-2
Each chunk is fully decrypted and authenticated in memory before being written out as plaintext, so an attacker may be able to truncate a file, but NEVER flip any bytes or corrupt it. Order is enforced by the incrementing the IV, so re-ordered chunks would be decrypted with the wrong IV and would fail authentication. The last chunk is flagged with AAD so we can detect files truncated even on chunk boundaries.
So only fully correct chunks will be output, and only ever in the correct order, but if later chunks are corrupted or the file is truncated, pegh will exit with an error code, and then it's the responsibility of the user/consuming application to do the proper thing with the truncated file it recieved.
Of course standard password bruteforcing is possible, but can be mitigated with increased scrypt parameters and longer password lengths.
pegh implements a simple versioned file format so encryption parameters can change in the future. Numbers here are inclusive 0-based byte array indices, 0th byte is always version number, everything else depends on version number, currently versions 0 and 1 exist.
Version 0, scrypt key derivation, AES-256-GCM encryption, 43 byte header, 16 byte auth tag per chunk. The 12-byte IV for the first chunk is 0, and is incremented by 1 for each successive chunk, if it ever rolls back over to 0 encryption should be aborted (chunk size should be increased). The last chunk has Additional Authenticated Data (AAD) sent in, a single byte value 0, which is used to flag the last chunk and detect file truncation.
|0||8 bit unsigned byte||pegh file format version|
|1-4||32 bit unsigned integer in big endian order||scrypt N parameter|
|5||8 bit unsigned byte||scrypt r parameter|
|6||8 bit unsigned byte||scrypt p parameter|
|7-10||32 bit unsigned integer in big endian order||aes encrypted chunk size|
|11-42||32 randomly generated bytes||scrypt key derivation seed|
|43+end||any number of chunks, chunk_size + 16 long||chunks followed by AES-256-GCM auth tag|
Version 1 has the exact same structure as version 0, except Chacha20-Poly1305 encryption instead of AES-256-GCM, key, IV, tag lengths are all the same.
pegh.c: AGPLv3 for now, message me if you have a problem with this
documentation/file format: consider this your choice of MIT, Apache 2, or public domain