Re: [PATCH] netlink: make sure nladdr has correct size in netlink_connect()

From: Eric Dumazet
Date: Wed Mar 14 2018 - 10:40:27 EST


On Wed, Mar 14, 2018 at 7:16 AM, Alexander Potapenko <glider@xxxxxxxxxx> wrote:
>
>
>
> On Wed, Mar 14, 2018 at 3:11 PM Eric Dumazet <edumazet@xxxxxxxxxx> wrote:
>>
>> On Wed, Mar 14, 2018 at 7:03 AM, Alexander Potapenko <glider@xxxxxxxxxx>
>> wrote:
>> > KMSAN reports use of uninitialized memory in the case when |alen| is
>> > smaller than sizeof(struct netlink_sock), and therefore |nladdr| isn't
>> > fully copied from the userspace.
>> >
>> > Signed-off-by: Alexander Potapenko <glider@xxxxxxxxxx>
>> > Fixes: 1da177e4c3f41524 ("Linux-2.6.12-rc2")
>> > ---
>> > net/netlink/af_netlink.c | 3 +++
>> > 1 file changed, 3 insertions(+)
>> >
>> > diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
>> > index 07e8478068f0..5d49b39e81c3 100644
>> > --- a/net/netlink/af_netlink.c
>> > +++ b/net/netlink/af_netlink.c
>> > @@ -1085,6 +1085,9 @@ static int netlink_connect(struct socket *sock,
>> > struct sockaddr *addr,
>> > if (addr->sa_family != AF_NETLINK)
>> > return -EINVAL;
>> >
>> > + if (alen < sizeof(struct netlink_sock))
>> > + return -EINVAL;
>> > +
>> >
>>
>> Hmmm. How was this patch tested exactly ?
>
> You're absolutely right, I should have been using sizeof(sockaddr_nl).
> The reproducer that I used to trigger the bug was passing alen=2, so the
> patch still worked despite being incorrect.
> Is there any generic set of networking tests that I can use for such bugs?

There are upstream tests in tools/testing/selftests/

In your case, making sure tools like iproute2 ss are still working
would have done the job.