Re: [PATCH net] xfrm: fix one more kernel-infoleak in algo dumping

[Simon Horman <horms@kernel.org>]

On Wed, Oct 02, 2024 at 11:17:24AM +0500, Petr Vaganov wrote:
> During fuzz testing, the following issue was discovered:
> 
> BUG: KMSAN: kernel-infoleak in _copy_to_iter+0x598/0x2a30

...

> Fixes copying of xfrm algorithms where some random
> data of the structure fields can end up in userspace.
> Padding in structures may be filled with random (possibly sensitve)
> data and should never be given directly to user-space.
> 
> A similar issue was resolved in the commit
> 8222d5910dae ("xfrm: Zero padding when dumping algos and encap")
> 
> Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
> 
> Fixes: c7a5899eb26e ("xfrm: redact SA secret with lockdown confidentiality")
> Cc: stable@vger.kernel.org
> Co-developed-by: Boris Tonofa <b.tonofa@ideco.ru>
> Signed-off-by: Boris Tonofa <b.tonofa@ideco.ru>
> Signed-off-by: Petr Vaganov <p.vaganov@ideco.ru>
> ---
>  net/xfrm/xfrm_user.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
> index 55f039ec3d59..97faeb3574ea 100644
> --- a/net/xfrm/xfrm_user.c
> +++ b/net/xfrm/xfrm_user.c
> @@ -1098,7 +1098,9 @@ static int copy_to_user_auth(struct xfrm_algo_auth *auth, struct sk_buff *skb)
>  	if (!nla)
>  		return -EMSGSIZE;
>  	ap = nla_data(nla);
> -	memcpy(ap, auth, sizeof(struct xfrm_algo_auth));
> +	strscpy_pad(ap->alg_name, auth->alg_name, sizeof(sizeof(ap->alg_name)));

Hi Petr and Boris,

The nested sizeof doesn't look right to me.
I expect the length of the destination is simply sizeof(ap->alg_name).

And given that ap->alg_name is an array (which is why using sizeof is
correct here), I believe the two-argument variant of strscpy_pad() can be
used:

	strscpy_pad(ap->alg_name, auth->alg_name);

As an aside, and not for this patch, there is a usage of strncpy() just
above this hunk which looks like it could be converted to the two-argument
variant of strscpy() or strncpy_pad() if it ought to be zero-padded.


> +	ap->alg_key_len = auth->alg_key_len;
> +	ap->alg_trunc_len = auth->alg_trunc_len;
>  	if (redact_secret && auth->alg_key_len)
>  		memset(ap->alg_key, 0, (auth->alg_key_len + 7) / 8);
>  	else
> -- 
> 2.46.1
> 
>