Hello!
On 10/03/17 12:55, Alexander Sverdlin wrote:
> Hello crypto maintainers!
>
> We've found and example of the ipsec algorithm combination, which doesn't fit
> into CRYPTO_MAX_ALG_NAME long buffers:
>
> ip x s add src 1.1.1.1 dst 1.1.1.2 proto esp spi 0 mode tunnel enc des3_ede
> 0x0 auth sha256 0x0 flag esn replay-window 256
>
> produces "echainiv(authencesn(hmac(sha256-generic),cbc(des3_ede-generic)))"
> on the machines without optimized crypto drivers, which doesn't fit into
> current
> 64-bytes buffers.
>
> I see two possible options:
>
> a) split CRYPTO_MAX_ALG_NAME into CRYPTO_MAX_ALG_NAME + CRYPTO_MAX_DRV_NAME
> pair
> and make later, say, 96, because the former probably cannot be changed
> because of
> numerous user-space exports. And change half of the code to use new define.
>
> b) rename *-generic algorithms to *-gen, so that cra_driver_name will be
> shortened,
> while MODULE_ALIAS_CRYPTO() could still be maintained in old and new form.
>
> What are your thoughts?
Any?
This is a regression caused by 856e3f4092
("crypto: seqiv - Add support for new AEAD interface")
As I've said above, I can offer one of the two solutions, which patch should I
send?
Or do you see any better alternatives?
--
Best regards,
Alexander Sverdlin.
