2017-12-05 22:55:16

by Kevin Cernekee

[permalink] [raw]
Subject: [PATCH] netlink: Add netns check on taps

Currently, a nlmon link inside a child namespace can observe systemwide
netlink activity. Filter the traffic so that in a non-init netns,
nlmon can only sniff netlink messages from its own netns.

Test case:

vpnns -- bash -c "ip link add nlmon0 type nlmon; \
ip link set nlmon0 up; \
tcpdump -i nlmon0 -q -w /tmp/nlmon.pcap -U" &
sudo ip xfrm state add src 10.1.1.1 dst 10.1.1.2 proto esp \
spi 0x1 mode transport \
auth sha1 0x6162633132330000000000000000000000000000 \
enc aes 0x00000000000000000000000000000000
grep abc123 /tmp/nlmon.pcap

Signed-off-by: Kevin Cernekee <[email protected]>
---
net/netlink/af_netlink.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index b9e0ee4..88381a2 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -253,6 +253,11 @@ static int __netlink_deliver_tap_skb(struct sk_buff *skb,
struct sock *sk = skb->sk;
int ret = -ENOMEM;

+ if (!net_eq(dev_net(dev), sock_net(sk)) &&
+ !net_eq(dev_net(dev), &init_net)) {
+ return 0;
+ }
+
dev_hold(dev);

if (is_vmalloc_addr(skb->head))
--
2.7.4


2017-12-06 02:19:54

by David Ahern

[permalink] [raw]
Subject: Re: [PATCH] netlink: Add netns check on taps

On 12/5/17 3:46 PM, Kevin Cernekee wrote:
> Currently, a nlmon link inside a child namespace can observe systemwide
> netlink activity. Filter the traffic so that in a non-init netns,
> nlmon can only sniff netlink messages from its own netns.
>
> Test case:
>
> vpnns -- bash -c "ip link add nlmon0 type nlmon; \
> ip link set nlmon0 up; \
> tcpdump -i nlmon0 -q -w /tmp/nlmon.pcap -U" &
> sudo ip xfrm state add src 10.1.1.1 dst 10.1.1.2 proto esp \
> spi 0x1 mode transport \
> auth sha1 0x6162633132330000000000000000000000000000 \
> enc aes 0x00000000000000000000000000000000
> grep abc123 /tmp/nlmon.pcap
>
> Signed-off-by: Kevin Cernekee <[email protected]>
> ---
> net/netlink/af_netlink.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
> index b9e0ee4..88381a2 100644
> --- a/net/netlink/af_netlink.c
> +++ b/net/netlink/af_netlink.c
> @@ -253,6 +253,11 @@ static int __netlink_deliver_tap_skb(struct sk_buff *skb,
> struct sock *sk = skb->sk;
> int ret = -ENOMEM;
>
> + if (!net_eq(dev_net(dev), sock_net(sk)) &&
> + !net_eq(dev_net(dev), &init_net)) {

Why is init_net special? Seems like snooping should be limited to the
namespace you are in.

2017-12-06 03:15:25

by Kevin Cernekee

[permalink] [raw]
Subject: Re: [PATCH] netlink: Add netns check on taps

On Tue, Dec 5, 2017 at 6:19 PM, David Ahern <[email protected]> wrote:
>> + if (!net_eq(dev_net(dev), sock_net(sk)) &&
>> + !net_eq(dev_net(dev), &init_net)) {
>
> Why is init_net special? Seems like snooping should be limited to the
> namespace you are in.

Depends how important it is to preserve the current "typical use case"
behavior, where the root user in the init netns can see all netlink
traffic on the system.

2017-12-06 19:40:08

by David Miller

[permalink] [raw]
Subject: Re: [PATCH] netlink: Add netns check on taps

From: Kevin Cernekee <[email protected]>
Date: Tue, 5 Dec 2017 14:46:22 -0800

> Currently, a nlmon link inside a child namespace can observe systemwide
> netlink activity. Filter the traffic so that in a non-init netns,
> nlmon can only sniff netlink messages from its own netns.
>
> Test case:
>
> vpnns -- bash -c "ip link add nlmon0 type nlmon; \
> ip link set nlmon0 up; \
> tcpdump -i nlmon0 -q -w /tmp/nlmon.pcap -U" &
> sudo ip xfrm state add src 10.1.1.1 dst 10.1.1.2 proto esp \
> spi 0x1 mode transport \
> auth sha1 0x6162633132330000000000000000000000000000 \
> enc aes 0x00000000000000000000000000000000
> grep abc123 /tmp/nlmon.pcap
>
> Signed-off-by: Kevin Cernekee <[email protected]>

Daniel, what behavior did you intend this to have?

Taps can see their own namespace only, or init_net is special
and can see all netlink activity.

I think letting init_net see everything could be confusing,
because there is no way to distinguish netlink events by
namespace just by looking at the messages that arrive at
the tap right?

So maybe own-namespace-only is the way to go.

Thanks.

2017-12-06 20:51:56

by Daniel Borkmann

[permalink] [raw]
Subject: Re: [PATCH] netlink: Add netns check on taps

On 12/06/2017 08:40 PM, David Miller wrote:
> From: Kevin Cernekee <[email protected]>
> Date: Tue, 5 Dec 2017 14:46:22 -0800
>
>> Currently, a nlmon link inside a child namespace can observe systemwide
>> netlink activity. Filter the traffic so that in a non-init netns,
>> nlmon can only sniff netlink messages from its own netns.
>>
>> Test case:
>>
>> vpnns -- bash -c "ip link add nlmon0 type nlmon; \
>> ip link set nlmon0 up; \
>> tcpdump -i nlmon0 -q -w /tmp/nlmon.pcap -U" &
>> sudo ip xfrm state add src 10.1.1.1 dst 10.1.1.2 proto esp \
>> spi 0x1 mode transport \
>> auth sha1 0x6162633132330000000000000000000000000000 \
>> enc aes 0x00000000000000000000000000000000
>> grep abc123 /tmp/nlmon.pcap
>>
>> Signed-off-by: Kevin Cernekee <[email protected]>
>
> Daniel, what behavior did you intend this to have?
>
> Taps can see their own namespace only, or init_net is special
> and can see all netlink activity.
>
> I think letting init_net see everything could be confusing,
> because there is no way to distinguish netlink events by
> namespace just by looking at the messages that arrive at
> the tap right?

Yeah, only snooping from own netns makes sense, lets limit
it to this.