Re: md5sum (from libkcapi) fails as splice() returns -ENOKEY

[Stephan Mueller <smueller@chronox.de>]

Am Dienstag, 12. September 2017, 09:01:08 CEST schrieb Christophe LEROY:

Hi Christophe,

> Hi Stephan
> 
> Le 11/09/2017 à 21:17, Stephan Müller a écrit :
> > Am Montag, 11. September 2017, 19:07:31 CEST schrieb christophe leroy:
> > 
> > Hi christophe,
> > 
> >> Hello Stephan,
> >> 
> >> I'm trying to use md5sum from the latest libkcapi 0.14 and I getting a
> >> failure with return code -5.
> >> 
> >> What am I missing ? See strace below, splice() return -ENOKEY.
> > 
> > The ENOKEY error is due to an accept() at the wrong location. But I do not
> > see
> > that error:
> I did the test once more without the Talitos Crypto driver compiled in
> the kernel, and this time it works.
> I believe it must then be an issue with that driver.
> What could be the issue, what should I look for in the driver ?
> 

I think I see the error in talitos.c -- yet I do not have that hardware so I 
cannot create a patch and test.

In talitos_alg_alloc the function pointer setkey is set unconditional for 
CRYPTO_ALG_TYPE_AHASH. This is correct for HMAC/CMAC, but not correct for 
staight hashes. As both, md5 and hmac(md5) (and also for the SHA equivalents) 
are marked as CRYPTO_ALG_TYPE_AHASH in driver_algs, they all will get a setkey 
function.

This will trigger the following code in algif_hash:

static int hash_accept_parent(void *private, struct sock *sk)
{
        struct algif_hash_tfm *tfm = private;

        if (!tfm->has_key && crypto_ahash_has_setkey(tfm->hash))
                return -ENOKEY;

If the setkey would not be present for the straight hashes, ENOKEY would not 
be returned.

...

> 
> # autoreconf --version
> autoreconf (GNU Autoconf) 2.63

Thanks for the bug report. But I think the autoconf tools are not up to date. 
It should be 2.69.

Could you please check whether you can update?

I am thinking now to add AC_PREREQ([2.60]) to configure.ac.



Ciao
Stephan