A byte-oriented AES-256 implementation

So you’ve checked it wrong, check again. The demo.c in zip archive is exactly the test vector for AES-256 from FIPS-197, Appendix C.3

Ugh, must admit that it was annoying enough :)
I’ve been busy recently, still need to find a time to put my hands on EnRUPT.

Ugh… Once again:
a) this is AES-256, not AES-128
b) demo.c is a bloody test vector from FIPS-197
I’ve no idea on how and where you have mistested it, but you’ve tested it wrong. So the nonsense result is all yours. Sorry.

hello, sorry by my ignorance because I’m new at this, but I just tested your code and the demo works perfect… my question is that I increment the buffer from 16 to 32 and it only encrypt the first 16 bytes of the buffer.
it’s an error from my end? I can encrypt only 16 bytes strings?
hope you can help me.

Thanks

Have you take this code as reference?
http://fp.gladman.plus.com/AES/index.htm

Specialy, http://fp.gladman.plus.com/AES/aes-byte-29-08-08.zip ?

Hello, I’m brazilian,
is a cool demo and lib,

I’ve one question, I understand that block size is 128 bits (16 bytes),
but I don’t understand if I’ve 17 bytes, have I extend string with null-bytes? byte 90 (nop) or 00 ?

sorry my broken english, ^^
(if you can put one example I’ll be happy)

thanks

If I encrypt using your mthod and decrypt with http://fp.gladman.plus.com/cryptography_technology/rijndael/ method won’t work ??

LLya, your lib is RFC3602, RFC2406 or RFC3852 ?

you know one cool dll or lib for pascal with AES ?

I’m with the some error =/

Ondra,

It does sounds like a buffer overflow for me, but I cannot say anything without seeing your code. You may try to check the output buffer size and usage in your program. Take a closer look where and how you store a decrypted data.

You might want to double-check that emphasized seconds result of yours. It is certainly not the fastest implementation and, as it said, speed was not important. But you’ve got ridiculous and obviously inadequate timing.

Hi,

I experienced that the decryption is much slower than the encryption. Is that reasonable? I don’t think so, since the encryption runs through the same algorithm like the decryption just with another key.
Do you have any speed test values? The decryption of 2000 bytes took about 3 to 5 seconds, which seems too much for me. What do you think?

best regards,
Alex

This is nice and it’s just what I was looking for, a compact, free implementation.

However I too found that decryption was several hundred times slower than encryption. The problem is the aes_subBytes_inv() function, which tries all possible sbox values to find the inverse. Since computing the sbox already iterates over all possible values to find the mulinv, this is a doubly nested loop and takes a long time.

Studying the AES spec I figured out how to write a fast sbox inversion function. I manually (pencil and paper!) inverted their matrix 5.2, and it comes out that the inverse transform is b_i = b_((i+2) mod 8) xor b_((i+5) mod 8) xor b_((i+7) mod 8). So you do this step and then the mulinv, and it is a relatively fast (faster anyway) sbox inversion:

/* ————————————————————————– */
uint8_t rj_sbox_inv(uint8_t x)
{
uint8_t y, sb;

y = x ^ 0×63;
sb = y = (y<>7);
y = (y<>6); sb ^= y; y = (y<>5); sb ^= y;

return gf_mulinv(sb);
} /* rj_sbox_inv */

Then you can do:

/* ————————————————————————– */
void aes_subBytes_inv(uint8_t *buf)
{
register uint8_t i = 16, j;
while (i–)
{
buf[i] = rj_sbox_inv(buf[i]);
}
} /* aes_subBytes_inv */

and get rid of the double nesting. Now decryption is as fast as encryption, fast enough for my purposes.

However, having done that, it turned out that just putting in the sbox and inverse sbox as tables, 256 bytes each, was no bigger and considerably faster. You get rid of gf_alog(), gf_log(), gf_mulinv(), rj_sbox() and rj_sbox_inv(), and do:

#define rj_sbox(x) sbox[(x)]
#define rj_sbox_inv(x) sboxinv[(x)]

With the code you eliminate, the tables don’t make things any bigger, and least on a 32 bit machine. But I imagine there’s not much difference on most architectures.

Your way is a nice exercise in interpreting the spec, but the table do simplify things. But thanks for this implementation, I will use it!

Feel free to use my changes if you like, under the same licensing terms.

Hal

[...] moving onto the HSM side. The HSM itself uses a AES library from literatecode.com which allows me to peform AES encryption and decryption on the Arduino board. Some (and I must [...]

Syn, I’ve just uploaded the updated source code, you may take it. Decryption is faster now and the code has an option to be compiled with static tables.

Hi,

Thanks for the AES implementation.
If I want to change your code to AES-128, which part that I need to change?

Thank you.

Absolutely great stuff, thanks a lot Ilya!

In your demo you call “Init” both before you decode and encode a buffer. Do you need to cal Init before each separate encode or decode? If I’m trying to decode a buffer of size - say 16384 bytes, I assume I have to loop 1024 times and call decode each time, but do I also have to call init each loop?

Hi, boss !! done a great job !! :-)

Can you tell me changes to be made to your code available for hard code it to 128-bit Message Block and 128-bit Based Encryption Key for AES ?

i have tried to make some changes to it…
like :

1. Changing :

void aes_expandEncKey(uint8_t *k, uint8_t *rc)
{
register uint8_t i;

k[0] ^= rj_sbox(k[13]) ^ (*rc);
k[1] ^= rj_sbox(k[14]);
k[2] ^= rj_sbox(k[15]);
k[3] ^= rj_sbox(k[12]);
*rc = F( *rc);

for(i = 4; i < 16; i += 4) k[i] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
} /* aes_expandEncKey */

then similarly for:

2. void aes_expandDecKey(uint8_t *k, uint8_t *rc) ;

3. typedef struct {
uint8_t key[16];
uint8_t enckey[16];
uint8_t deckey[16];
} aes256_context;

4. uint8_t key[32]; to uint8_t key[16 ]; in demo.c file…
5. And few More changes…
6. but the input plain text with decrypted buffer varies totally….

I am using your license in my project report and bibliography too !! Dont mind for modification of work !! ;-)

And help me out !!