Nikos Mavrogiannopoulos <[email protected]> writes:
> It seems that ctr_crypt16() would not handle the whole input and that
> was complicating things.
I was afraid of that. Doing the extra block would be something like
done = ctr_crypt16(cipher, f, gcm_fill, ctx->ctr.b, length, dst, src);
if (done < length)
{
uint8_t block[16];
assert (done % 16 == 0);
assert (length - done < 16);
f(ctx, block, ctx->ctr.b, 16, 16);
memxor3(dst + done, src + done, block, length - done);
}
(if we skip updating the counter in this case; I don't think gcm
promises anything about the counter after a partial block).
But I agree it makes sense to let ctr_crypt16 do that.
More detailed comments later.
Regards,
/Niels
--
Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.
_______________________________________________
nettle-bugs mailing list
[email protected]
http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
