[PATCH ipsec 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[PATCH ipsec 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Before those patches, packets sent to a vti[6]/xfrm interface via
bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
no dst was attached.

 net/ipv4/ip_vti.c         | 10 ++++++++--
 net/ipv6/ip6_vti.c        | 11 +++++++++--
 net/xfrm/xfrm_interface.c | 21 +++++++++++++++++----
 3 files changed, 34 insertions(+), 8 deletions(-)

Regards,
Nicolas

[PATCH ipsec 1/2] vti[6]: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a vti[6] interface,
the packets are dropped because no dst is attached.

This could also be reproduced with an AF_PACKET socket, with the following
python script (vti1 is an ip_vti interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('vti1', 0x800, 0, 0))

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/ipv4/ip_vti.c  | 11 +++++++++--
 net/ipv6/ip6_vti.c | 11 +++++++++--
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/net/ipv4/ip_vti.c b/net/ipv4/ip_vti.c
index cfb025606793..341e0689456d 100644
--- a/net/ipv4/ip_vti.c
+++ b/net/ipv4/ip_vti.c
@@ -187,8 +187,15 @@ static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
 	int mtu;
 
 	if (!dst) {
-		dev->stats.tx_carrier_errors++;
-		goto tx_error_icmp;
+		struct rtable *rt = __ip_route_output_key(dev_net(dev),
+							  &fl->u.ip4);
+
+		if (IS_ERR(rt)) {
+			dev->stats.tx_carrier_errors++;
+			goto tx_error_icmp;
+		}
+		dst = &rt->dst;
+		skb_dst_set(skb, dst);
 	}
 
 	dst_hold(dst);
diff --git a/net/ipv6/ip6_vti.c b/net/ipv6/ip6_vti.c
index 024db17386d2..089b9d5dc15e 100644
--- a/net/ipv6/ip6_vti.c
+++ b/net/ipv6/ip6_vti.c
@@ -449,8 +449,15 @@ vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 	int err = -1;
 	int mtu;
 
-	if (!dst)
-		goto tx_err_link_failure;
+	if (!dst) {
+		dst = ip6_route_output(dev_net(dev), NULL, &fl->u.ip6);
+		if (dst->error) {
+			dst_release(dst);
+			dst = NULL;
+			goto tx_err_link_failure;
+		}
+		skb_dst_set(skb, dst);
+	}
 
 	dst_hold(dst);
 	dst = xfrm_lookup(t->net, dst, fl, NULL, 0);
-- 
2.24.0

[PATCH ipsec 2/2] xfrm interface: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a xfrm interface,
packets are dropped because no dst is attached to skb.

This could also be reproduced with an AF_PACKET socket, with the following
python script (xfrm1 is a xfrm interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('xfrm1', 0x800, 0, 0))

It was also not possible to send an ip packet through an AF_PACKET socket
because a LL header was expected. Let's remove those LL header constraints.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/xfrm/xfrm_interface.c | 22 ++++++++++++++++++----
 1 file changed, 18 insertions(+), 4 deletions(-)

diff --git a/net/xfrm/xfrm_interface.c b/net/xfrm/xfrm_interface.c
index 7ac1542feaf8..55978a1501ec 100644
--- a/net/xfrm/xfrm_interface.c
+++ b/net/xfrm/xfrm_interface.c
@@ -343,6 +343,7 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct xfrm_if *xi = netdev_priv(dev);
 	struct net_device_stats *stats = &xi->dev->stats;
+	struct dst_entry *dst = skb_dst(skb);
 	struct flowi fl;
 	int ret;
 
@@ -352,10 +353,26 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 	case htons(ETH_P_IPV6):
 		xfrm_decode_session(skb, &fl, AF_INET6);
 		memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
+		if (!dst) {
+			dst = ip6_route_output(dev_net(dev), NULL, &fl.u.ip6);
+			if (dst->error) {
+				dst_release(dst);
+				goto tx_err;
+			}
+			skb_dst_set(skb, dst);
+		}
 		break;
 	case htons(ETH_P_IP):
 		xfrm_decode_session(skb, &fl, AF_INET);
 		memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
+		if (!dst) {
+			struct rtable *rt = __ip_route_output_key(dev_net(dev),
+								  &fl.u.ip4);
+
+			if (IS_ERR(rt))
+				goto tx_err;
+			skb_dst_set(skb, &rt->dst);
+		}
 		break;
 	default:
 		goto tx_err;
@@ -563,12 +580,9 @@ static void xfrmi_dev_setup(struct net_device *dev)
 {
 	dev->netdev_ops 	= &xfrmi_netdev_ops;
 	dev->type		= ARPHRD_NONE;
-	dev->hard_header_len 	= ETH_HLEN;
-	dev->min_header_len	= ETH_HLEN;
 	dev->mtu		= ETH_DATA_LEN;
 	dev->min_mtu		= ETH_MIN_MTU;
-	dev->max_mtu		= ETH_DATA_LEN;
-	dev->addr_len		= ETH_ALEN;
+	dev->max_mtu		= IP_MAX_MTU;
 	dev->flags 		= IFF_NOARP;
 	dev->needs_free_netdev	= true;
 	dev->priv_destructor	= xfrmi_dev_free;
-- 
2.24.0

Re: [PATCH ipsec 2/2] xfrm interface: fix packet tx through bpf_redirect()

[Steffen Klassert]

On Tue, Dec 31, 2019 at 05:56:54PM +0100, Nicolas Dichtel wrote:
> With an ebpf program that redirects packets through a xfrm interface,
> packets are dropped because no dst is attached to skb.
> 
> This could also be reproduced with an AF_PACKET socket, with the following
> python script (xfrm1 is a xfrm interface):
> 
>  import socket
>  send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
>  # scapy
>  # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
>  # raw(p)
>  req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
>  send_s.sendto(req, ('xfrm1', 0x800, 0, 0))
> 
> It was also not possible to send an ip packet through an AF_PACKET socket
> because a LL header was expected. Let's remove those LL header constraints.
> 
> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
> ---
>  net/xfrm/xfrm_interface.c | 22 ++++++++++++++++++----
>  1 file changed, 18 insertions(+), 4 deletions(-)
> 
> diff --git a/net/xfrm/xfrm_interface.c b/net/xfrm/xfrm_interface.c
> index 7ac1542feaf8..55978a1501ec 100644
> --- a/net/xfrm/xfrm_interface.c
> +++ b/net/xfrm/xfrm_interface.c
> @@ -343,6 +343,7 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
>  {
>  	struct xfrm_if *xi = netdev_priv(dev);
>  	struct net_device_stats *stats = &xi->dev->stats;
> +	struct dst_entry *dst = skb_dst(skb);
>  	struct flowi fl;
>  	int ret;
>  
> @@ -352,10 +353,26 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
>  	case htons(ETH_P_IPV6):
>  		xfrm_decode_session(skb, &fl, AF_INET6);
>  		memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
> +		if (!dst) {
> +			dst = ip6_route_output(dev_net(dev), NULL, &fl.u.ip6);
> +			if (dst->error) {
> +				dst_release(dst);
> +				goto tx_err;
> +			}
> +			skb_dst_set(skb, dst);
> +		}
>  		break;
>  	case htons(ETH_P_IP):
>  		xfrm_decode_session(skb, &fl, AF_INET);
>  		memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
> +		if (!dst) {
> +			struct rtable *rt = __ip_route_output_key(dev_net(dev),
> +								  &fl.u.ip4);
> +
> +			if (IS_ERR(rt))
> +				goto tx_err;

With this change, the 'if (!dst)' in xfrmi_xmit2() is meaningless
and we don't handle this error as a link failure anymore.

Please make sure that the error path is not changed.

Re: [PATCH ipsec 2/2] xfrm interface: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

Le 09/01/2020 à 09:40, Steffen Klassert a écrit :
[snip]
>> @@ -352,10 +353,26 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
>>  	case htons(ETH_P_IPV6):
>>  		xfrm_decode_session(skb, &fl, AF_INET6);
>>  		memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
>> +		if (!dst) {
>> +			dst = ip6_route_output(dev_net(dev), NULL, &fl.u.ip6);
>> +			if (dst->error) {
>> +				dst_release(dst);
>> +				goto tx_err;
>> +			}
>> +			skb_dst_set(skb, dst);
>> +		}
>>  		break;
>>  	case htons(ETH_P_IP):
>>  		xfrm_decode_session(skb, &fl, AF_INET);
>>  		memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
>> +		if (!dst) {
>> +			struct rtable *rt = __ip_route_output_key(dev_net(dev),
>> +								  &fl.u.ip4);
>> +
>> +			if (IS_ERR(rt))
>> +				goto tx_err;
> 
> With this change, the 'if (!dst)' in xfrmi_xmit2() is meaningless
Yep, I was hesitant to remove it :)

> and we don't handle this error as a link failure anymore.
Good point, will send a v2.


Thank you,
Nicolas

[PATCH ipsec v2 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Before those patches, packets sent to a vti[6]/xfrm interface via
bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
no dst was attached.

v1 -> v2:
  - remove useless check against skb_dst() in xfrmi_xmit2()
  - keep incrementing tx_carrier_errors in case of route lookup failure

 net/ipv4/ip_vti.c         | 11 +++++++++--
 net/ipv6/ip6_vti.c        | 11 +++++++++--
 net/xfrm/xfrm_interface.c | 28 +++++++++++++++++++++-------
 3 files changed, 39 insertions(+), 11 deletions(-)

Regards,
Nicolas

[PATCH ipsec v2 1/2] vti[6]: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a vti[6] interface,
the packets are dropped because no dst is attached.

This could also be reproduced with an AF_PACKET socket, with the following
python script (vti1 is an ip_vti interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('vti1', 0x800, 0, 0))

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/ipv4/ip_vti.c  | 11 +++++++++--
 net/ipv6/ip6_vti.c | 11 +++++++++--
 2 files changed, 18 insertions(+), 4 deletions(-)

diff --git a/net/ipv4/ip_vti.c b/net/ipv4/ip_vti.c
index e90b600c7a25..085d7aea82e7 100644
--- a/net/ipv4/ip_vti.c
+++ b/net/ipv4/ip_vti.c
@@ -187,8 +187,15 @@ static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
 	int mtu;
 
 	if (!dst) {
-		dev->stats.tx_carrier_errors++;
-		goto tx_error_icmp;
+		struct rtable *rt = __ip_route_output_key(dev_net(dev),
+							  &fl->u.ip4);
+
+		if (IS_ERR(rt)) {
+			dev->stats.tx_carrier_errors++;
+			goto tx_error_icmp;
+		}
+		dst = &rt->dst;
+		skb_dst_set(skb, dst);
 	}
 
 	dst_hold(dst);
diff --git a/net/ipv6/ip6_vti.c b/net/ipv6/ip6_vti.c
index 6f08b760c2a7..a6609803e8b1 100644
--- a/net/ipv6/ip6_vti.c
+++ b/net/ipv6/ip6_vti.c
@@ -449,8 +449,15 @@ vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 	int err = -1;
 	int mtu;
 
-	if (!dst)
-		goto tx_err_link_failure;
+	if (!dst) {
+		dst = ip6_route_output(dev_net(dev), NULL, &fl->u.ip6);
+		if (dst->error) {
+			dst_release(dst);
+			dst = NULL;
+			goto tx_err_link_failure;
+		}
+		skb_dst_set(skb, dst);
+	}
 
 	dst_hold(dst);
 	dst = xfrm_lookup(t->net, dst, fl, NULL, 0);
-- 
2.24.0

[PATCH ipsec v2 2/2] xfrm interface: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a xfrm interface,
packets are dropped because no dst is attached to skb.

This could also be reproduced with an AF_PACKET socket, with the following
python script (xfrm1 is a xfrm interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('xfrm1', 0x800, 0, 0))

It was also not possible to send an ip packet through an AF_PACKET socket
because a LL header was expected. Let's remove those LL header constraints.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/xfrm/xfrm_interface.c | 28 +++++++++++++++++++++-------
 1 file changed, 21 insertions(+), 7 deletions(-)

diff --git a/net/xfrm/xfrm_interface.c b/net/xfrm/xfrm_interface.c
index 7ac1542feaf8..60481acb10d2 100644
--- a/net/xfrm/xfrm_interface.c
+++ b/net/xfrm/xfrm_interface.c
@@ -268,9 +268,6 @@ xfrmi_xmit2(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 	int err = -1;
 	int mtu;
 
-	if (!dst)
-		goto tx_err_link_failure;
-
 	dst_hold(dst);
 	dst = xfrm_lookup_with_ifid(xi->net, dst, fl, NULL, 0, xi->p.if_id);
 	if (IS_ERR(dst)) {
@@ -343,6 +340,7 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct xfrm_if *xi = netdev_priv(dev);
 	struct net_device_stats *stats = &xi->dev->stats;
+	struct dst_entry *dst = skb_dst(skb);
 	struct flowi fl;
 	int ret;
 
@@ -352,10 +350,29 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 	case htons(ETH_P_IPV6):
 		xfrm_decode_session(skb, &fl, AF_INET6);
 		memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
+		if (!dst) {
+			dst = ip6_route_output(dev_net(dev), NULL, &fl.u.ip6);
+			if (dst->error) {
+				dst_release(dst);
+				stats->tx_carrier_errors++;
+				goto tx_err;
+			}
+			skb_dst_set(skb, dst);
+		}
 		break;
 	case htons(ETH_P_IP):
 		xfrm_decode_session(skb, &fl, AF_INET);
 		memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
+		if (!dst) {
+			struct rtable *rt = __ip_route_output_key(dev_net(dev),
+								  &fl.u.ip4);
+
+			if (IS_ERR(rt)) {
+				stats->tx_carrier_errors++;
+				goto tx_err;
+			}
+			skb_dst_set(skb, &rt->dst);
+		}
 		break;
 	default:
 		goto tx_err;
@@ -563,12 +580,9 @@ static void xfrmi_dev_setup(struct net_device *dev)
 {
 	dev->netdev_ops 	= &xfrmi_netdev_ops;
 	dev->type		= ARPHRD_NONE;
-	dev->hard_header_len 	= ETH_HLEN;
-	dev->min_header_len	= ETH_HLEN;
 	dev->mtu		= ETH_DATA_LEN;
 	dev->min_mtu		= ETH_MIN_MTU;
-	dev->max_mtu		= ETH_DATA_LEN;
-	dev->addr_len		= ETH_ALEN;
+	dev->max_mtu		= IP_MAX_MTU;
 	dev->flags 		= IFF_NOARP;
 	dev->needs_free_netdev	= true;
 	dev->priv_destructor	= xfrmi_dev_free;
-- 
2.24.0

Re: [PATCH ipsec v2 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Le 09/01/2020 à 14:35, Nicolas Dichtel a écrit :
> Before those patches, packets sent to a vti[6]/xfrm interface via
> bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
> no dst was attached.
Please, drop this version, I will send an update.


Regards,
Nicolas

[PATCH ipsec v3 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Before those patches, packets sent to a vti[6]/xfrm interface via
bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
no dst was attached.

v2 -> v3:
  - fix flowi info for the route lookup

v1 -> v2:
  - remove useless check against skb_dst() in xfrmi_xmit2()
  - keep incrementing tx_carrier_errors in case of route lookup failure

 net/ipv4/ip_vti.c         | 13 +++++++++++--
 net/ipv6/ip6_vti.c        | 13 +++++++++++--
 net/xfrm/xfrm_interface.c | 32 +++++++++++++++++++++++++-------
 3 files changed, 47 insertions(+), 11 deletions(-)

Regards,
Nicolas

[PATCH ipsec v3 1/2] vti[6]: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a vti[6] interface,
the packets are dropped because no dst is attached.

This could also be reproduced with an AF_PACKET socket, with the following
python script (vti1 is an ip_vti interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('vti1', 0x800, 0, 0))

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/ipv4/ip_vti.c  | 13 +++++++++++--
 net/ipv6/ip6_vti.c | 13 +++++++++++--
 2 files changed, 22 insertions(+), 4 deletions(-)

diff --git a/net/ipv4/ip_vti.c b/net/ipv4/ip_vti.c
index e90b600c7a25..37cddd18f282 100644
--- a/net/ipv4/ip_vti.c
+++ b/net/ipv4/ip_vti.c
@@ -187,8 +187,17 @@ static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
 	int mtu;
 
 	if (!dst) {
-		dev->stats.tx_carrier_errors++;
-		goto tx_error_icmp;
+		struct rtable *rt;
+
+		fl->u.ip4.flowi4_oif = dev->ifindex;
+		fl->u.ip4.flowi4_flags |= FLOWI_FLAG_ANYSRC;
+		rt = __ip_route_output_key(dev_net(dev), &fl->u.ip4);
+		if (IS_ERR(rt)) {
+			dev->stats.tx_carrier_errors++;
+			goto tx_error_icmp;
+		}
+		dst = &rt->dst;
+		skb_dst_set(skb, dst);
 	}
 
 	dst_hold(dst);
diff --git a/net/ipv6/ip6_vti.c b/net/ipv6/ip6_vti.c
index 6f08b760c2a7..524006aa0d78 100644
--- a/net/ipv6/ip6_vti.c
+++ b/net/ipv6/ip6_vti.c
@@ -449,8 +449,17 @@ vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 	int err = -1;
 	int mtu;
 
-	if (!dst)
-		goto tx_err_link_failure;
+	if (!dst) {
+		fl->u.ip6.flowi6_oif = dev->ifindex;
+		fl->u.ip6.flowi6_flags |= FLOWI_FLAG_ANYSRC;
+		dst = ip6_route_output(dev_net(dev), NULL, &fl->u.ip6);
+		if (dst->error) {
+			dst_release(dst);
+			dst = NULL;
+			goto tx_err_link_failure;
+		}
+		skb_dst_set(skb, dst);
+	}
 
 	dst_hold(dst);
 	dst = xfrm_lookup(t->net, dst, fl, NULL, 0);
-- 
2.24.0

[PATCH ipsec v3 2/2] xfrm interface: fix packet tx through bpf_redirect()

[Nicolas Dichtel]

With an ebpf program that redirects packets through a xfrm interface,
packets are dropped because no dst is attached to skb.

This could also be reproduced with an AF_PACKET socket, with the following
python script (xfrm1 is a xfrm interface):

 import socket
 send_s = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, 0)
 # scapy
 # p = IP(src='10.100.0.2', dst='10.200.0.1')/ICMP(type='echo-request')
 # raw(p)
 req = b'E\x00\x00\x1c\x00\x01\x00\x00@\x01e\xb2\nd\x00\x02\n\xc8\x00\x01\x08\x00\xf7\xff\x00\x00\x00\x00'
 send_s.sendto(req, ('xfrm1', 0x800, 0, 0))

It was also not possible to send an ip packet through an AF_PACKET socket
because a LL header was expected. Let's remove those LL header constraints.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 net/xfrm/xfrm_interface.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)

diff --git a/net/xfrm/xfrm_interface.c b/net/xfrm/xfrm_interface.c
index 7ac1542feaf8..00393179f185 100644
--- a/net/xfrm/xfrm_interface.c
+++ b/net/xfrm/xfrm_interface.c
@@ -268,9 +268,6 @@ xfrmi_xmit2(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 	int err = -1;
 	int mtu;
 
-	if (!dst)
-		goto tx_err_link_failure;
-
 	dst_hold(dst);
 	dst = xfrm_lookup_with_ifid(xi->net, dst, fl, NULL, 0, xi->p.if_id);
 	if (IS_ERR(dst)) {
@@ -343,6 +340,7 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct xfrm_if *xi = netdev_priv(dev);
 	struct net_device_stats *stats = &xi->dev->stats;
+	struct dst_entry *dst = skb_dst(skb);
 	struct flowi fl;
 	int ret;
 
@@ -352,10 +350,33 @@ static netdev_tx_t xfrmi_xmit(struct sk_buff *skb, struct net_device *dev)
 	case htons(ETH_P_IPV6):
 		xfrm_decode_session(skb, &fl, AF_INET6);
 		memset(IP6CB(skb), 0, sizeof(*IP6CB(skb)));
+		if (!dst) {
+			fl.u.ip6.flowi6_oif = dev->ifindex;
+			fl.u.ip6.flowi6_flags |= FLOWI_FLAG_ANYSRC;
+			dst = ip6_route_output(dev_net(dev), NULL, &fl.u.ip6);
+			if (dst->error) {
+				dst_release(dst);
+				stats->tx_carrier_errors++;
+				goto tx_err;
+			}
+			skb_dst_set(skb, dst);
+		}
 		break;
 	case htons(ETH_P_IP):
 		xfrm_decode_session(skb, &fl, AF_INET);
 		memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
+		if (!dst) {
+			struct rtable *rt;
+
+			fl.u.ip4.flowi4_oif = dev->ifindex;
+			fl.u.ip4.flowi4_flags |= FLOWI_FLAG_ANYSRC;
+			rt = __ip_route_output_key(dev_net(dev), &fl.u.ip4);
+			if (IS_ERR(rt)) {
+				stats->tx_carrier_errors++;
+				goto tx_err;
+			}
+			skb_dst_set(skb, &rt->dst);
+		}
 		break;
 	default:
 		goto tx_err;
@@ -563,12 +584,9 @@ static void xfrmi_dev_setup(struct net_device *dev)
 {
 	dev->netdev_ops 	= &xfrmi_netdev_ops;
 	dev->type		= ARPHRD_NONE;
-	dev->hard_header_len 	= ETH_HLEN;
-	dev->min_header_len	= ETH_HLEN;
 	dev->mtu		= ETH_DATA_LEN;
 	dev->min_mtu		= ETH_MIN_MTU;
-	dev->max_mtu		= ETH_DATA_LEN;
-	dev->addr_len		= ETH_ALEN;
+	dev->max_mtu		= IP_MAX_MTU;
 	dev->flags 		= IFF_NOARP;
 	dev->needs_free_netdev	= true;
 	dev->priv_destructor	= xfrmi_dev_free;
-- 
2.24.0

Re: [PATCH ipsec v3 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Steffen Klassert]

On Mon, Jan 13, 2020 at 09:32:45AM +0100, Nicolas Dichtel wrote:
> Before those patches, packets sent to a vti[6]/xfrm interface via
> bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
> no dst was attached.
> 
> v2 -> v3:
>   - fix flowi info for the route lookup
> 
> v1 -> v2:
>   - remove useless check against skb_dst() in xfrmi_xmit2()
>   - keep incrementing tx_carrier_errors in case of route lookup failure
> 
>  net/ipv4/ip_vti.c         | 13 +++++++++++--
>  net/ipv6/ip6_vti.c        | 13 +++++++++++--
>  net/xfrm/xfrm_interface.c | 32 +++++++++++++++++++++++++-------
>  3 files changed, 47 insertions(+), 11 deletions(-)

Applied to the ipsec tree, thanks a lot!

Re: [PATCH ipsec v3 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Le 15/01/2020 à 10:56, Steffen Klassert a écrit :
> On Mon, Jan 13, 2020 at 09:32:45AM +0100, Nicolas Dichtel wrote:
>> Before those patches, packets sent to a vti[6]/xfrm interface via
>> bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
>> no dst was attached.
>>
>> v2 -> v3:
>>   - fix flowi info for the route lookup
>>
>> v1 -> v2:
>>   - remove useless check against skb_dst() in xfrmi_xmit2()
>>   - keep incrementing tx_carrier_errors in case of route lookup failure
>>
>>  net/ipv4/ip_vti.c         | 13 +++++++++++--
>>  net/ipv6/ip6_vti.c        | 13 +++++++++++--
>>  net/xfrm/xfrm_interface.c | 32 +++++++++++++++++++++++++-------
>>  3 files changed, 47 insertions(+), 11 deletions(-)
> 
> Applied to the ipsec tree, thanks a lot!
> 
Thanks. Those patches are now in Linus tree:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=95224166a903
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=f042365dbffe

Could you queue them for stable trees?


Thank you,
Nicolas

Re: [PATCH ipsec v3 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Steffen Klassert]

On Mon, Jan 27, 2020 at 10:28:10AM +0100, Nicolas Dichtel wrote:
> Le 15/01/2020 à 10:56, Steffen Klassert a écrit :
> > On Mon, Jan 13, 2020 at 09:32:45AM +0100, Nicolas Dichtel wrote:
> >> Before those patches, packets sent to a vti[6]/xfrm interface via
> >> bpf_redirect() or via an AF_PACKET socket were dropped, mostly because
> >> no dst was attached.
> >>
> >> v2 -> v3:
> >>   - fix flowi info for the route lookup
> >>
> >> v1 -> v2:
> >>   - remove useless check against skb_dst() in xfrmi_xmit2()
> >>   - keep incrementing tx_carrier_errors in case of route lookup failure
> >>
> >>  net/ipv4/ip_vti.c         | 13 +++++++++++--
> >>  net/ipv6/ip6_vti.c        | 13 +++++++++++--
> >>  net/xfrm/xfrm_interface.c | 32 +++++++++++++++++++++++++-------
> >>  3 files changed, 47 insertions(+), 11 deletions(-)
> > 
> > Applied to the ipsec tree, thanks a lot!
> > 
> Thanks. Those patches are now in Linus tree:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=95224166a903
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=f042365dbffe
> 
> Could you queue them for stable trees?

I'll do an ipsec stable submission the next days.

Re: [PATCH ipsec v3 0/2] ipsec interfaces: fix sending with bpf_redirect() / AF_PACKET sockets

[Nicolas Dichtel]

Le 27/01/2020 à 10:58, Steffen Klassert a écrit :
[snip]
> I'll do an ipsec stable submission the next days.
> 

Thanks!