Re: [PATCH 1/1]: Revised CTR mode implementation

On Thu, Sep 27, 2007 at 03:54:51PM -0500, Joy Latten wrote:

   So, for example,

		ctr(aes,4,8)

   specifies the counter block will be composed of 4 bytes from a
   nonce and 8 bytes from the IV and 4 bytes for counter, which is set.

Could you please add a check to verify that for

	ctr(X,Y,Z)

we have

	block_size(X) - Y - Z == 4

Return -EINVAL if this fails.

2. it is assumed that plaintext is  multiple of blocksize.

Yes blkcipher will fail if there's any left-over.

Sorry, I think I misled you earlier when you asked about the
block size and left-overs.

The block size of ctr(aes,X,Y) should not be that of the block
size of AES.  It should instead be 1 as CTR is a stream cipher.

The API currently doesn't allow that but I'll patch it so
that it does :)

3. currently nonce is extracted from the last 4 bytes of key.
   Thus keys entered through setkey() have an additional 32 bits.
   This causes problems for 256-bit keys. For example,
   crypto_ablkcipher_setkey() checks the maximum keysize and
   complains about keysize.
   This issue will be taken cared of with the new
   infrastructure/template for combined mode that is planned,
   and appropriate changes will be made to crypto_ctr_setkey()
   and testcases.

You should instead increase min_keysize/max_keysize accordingly.

Thanks,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt