New: Tunnel device ignores TCP/UDP traffic

Bugme-new Applied Andrew Morton 5 participants 9 messages Jun 26, 2009 – Jun 28, 2009 1 message not in the archive

Subsystems (5)

Networking Network drivers IPv4/IPv6 Sockets +1 more subsystems

lore

Share & tools

lore

A message that is not in the archive (sent off-list or not yet received): [email protected]/

Andrew Morton

author

2009-06-26 19:45 UTC #1 lore raw

(switched to email. Please respond via emailed reply-to-all, not via the bugzilla web interface).

Quoted 17 lines · On Fri, 26 Jun 2009 14:45:11 GMT [email protected] wrote:

http://bugzilla.kernel.org/show_bug.cgi?id=13627

           Summary: Tunnel device ignores TCP/UDP traffic
           Product: Networking
           Version: 2.5
    Kernel Version: 2.6.31-rc1
          Platform: All
        OS/Version: Linux
              Tree: Mainline
            Status: NEW
          Severity: normal
          Priority: P1
         Component: IPV4
        AssignedTo: [email protected]
        ReportedBy: [email protected]
        Regression: Yes

It's a post-2.6.30 regression which Paul has bisected down to

commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700

net: Move rx skb_orphan call to where needed

(thanks for doing the bisection!)

Quoted 77 lines

Using OpenVPN on 2.6.29.4 and 2.6.30 works but 2.6.31-rc1 doesn't.

I can ping (ICMP) the remote end, and see the packets going back and forth using tcpdump, but they don't appear to be reaching the upper layers.

traceroute with -U (UDP) and -T (TCP SYN) options (ie. raw packet socket) works.

There are no iptables filters in place. Nothing unusual appears in dmesg or syslog.

root@thinkpad:~# ping 172.17.2.1
PING 172.17.2.1 (172.17.2.1) 56(84) bytes of data.
64 bytes from 172.17.2.1: icmp_seq=1 ttl=64 time=97.9 ms
64 bytes from 172.17.2.1: icmp_seq=2 ttl=64 time=108 ms
64 bytes from 172.17.2.1: icmp_seq=3 ttl=64 time=184 ms
64 bytes from 172.17.2.1: icmp_seq=4 ttl=64 time=96.0 ms
^C
--- 172.17.2.1 ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3004ms
rtt min/avg/max/mdev = 96.022/121.776/184.465/36.510 ms

# tcptraceroute -n 192.168.1.1 1
traceroute to 192.168.1.1 (192.168.1.1), 30 hops max, 60 byte packets
 1  192.168.1.1  90.230 ms  105.535 ms  104.833 ms

~# telnet 172.17.2.1 1
Trying 172.17.2.1...
^C

# tcpdump -n -p -i tun0
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on tun0, link-type RAW (Raw IP), capture size 96 bytes
15:28:44.450554 IP 172.17.2.10.46287 > 172.17.2.1.1: Flags [S], seq 434762695,
win 5840, options [mss 1460,sackOK,TS val 169108 ecr 0,nop,wscale 6], length 0
15:28:44.542303 IP 172.17.2.1.1 > 172.17.2.10.46287: Flags [R.], seq 0, ack
434762696, win 0, length 0

# ip link
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
4: wmaster0: <UP,LOWER_UP> mtu 0 qdisc pfifo_fast state UNKNOWN qlen 1000
    link/ieee802.11 00:14:a4:04:df:09 brd 00:00:00:00:00:00
5: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP
qlen 1000
    link/ether 00:14:a4:04:df:09 brd ff:ff:ff:ff:ff:ff
8: tun0: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 100
    link/[65534]

# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
4: wmaster0: <UP,LOWER_UP> mtu 0 qdisc pfifo_fast state UNKNOWN qlen 1000
    link/ieee802.11 00:14:a4:04:df:09 brd 00:00:00:00:00:00
5: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP
qlen 1000
    link/ether 00:14:a4:04:df:09 brd ff:ff:ff:ff:ff:ff
    inet 192.168.10.118/24 brd 192.168.10.255 scope global wlan0
    inet6 fe80::214:a4ff:fe04:df09/64 scope link
       valid_lft forever preferred_lft forever
8: tun0: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 100
    link/[65534]
    inet 172.17.2.10 peer 172.17.2.9/32 scope global tun0

# ip route
172.17.2.9 dev tun0  proto kernel  scope link  src 172.17.2.10
192.168.1.0/24 via 172.17.2.9 dev tun0  metric 20
172.17.2.0/24 via 172.17.2.9 dev tun0
192.168.10.0/24 dev wlan0  proto kernel  scope link  src 192.168.10.118
default via 192.168.10.1 dev wlan0

David Miller

maintainer Tested-by Oliver Hartkopp

2009-06-26 20:02 UTC #2 lore raw

From: Andrew Morton <[email protected]>
Date: Fri, 26 Jun 2009 12:44:44 -0700

Quoted 11 lines

It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700
net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)

Then it's good that we partially reverted that change just the other day :-)

commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f

Author: Herbert Xu <[email protected]>
Date:   Mon Jun 22 02:25:25 2009 +0000

net: Move rx skb_orphan call to where needed

In order to get the tun driver to account packets, we need to be able to receive packets with destructors set. To be on the safe side, I added an skb_orphan call for all protocols by default since some of them (IP in particular) cannot handle receiving packets destructors properly.

Now it seems that at least one protocol (CAN) expects to be able to pass skb->sk through the rx path without getting clobbered.

So this patch attempts to fix this properly by moving the skb_orphan call to where it's actually needed. In particular, I've added it to skb_set_owner_[rw] which is what most users of skb->destructor call.

This is actually an improvement for tun too since it means that we only give back the amount charged to the socket when the skb is passed to another socket that will also be charged accordingly.

Signed-off-by Herbert Xu <[email protected]>

Tested-by Oliver Hartkopp <[email protected]>

Signed-off-by David S. Miller <[email protected]>

include/net/sctp/sctp.h +1 -0

index 9f80a76..d16a304 100644
@@ -448,6 +448,7 @@ static inline void sctp_skb_set_owner_r(struct sk_buff *skb, struct sock *sk)
 {
 	struct sctp_ulpevent *event = sctp_skb2event(skb);
 
+	skb_orphan(skb);
 	skb->sk = sk;
 	skb->destructor = sctp_sock_rfree;
 	atomic_add(event->rmem_len, &sk->sk_rmem_alloc);

include/net/sock.h +2 -0

index 570c7a1..7f5c41c 100644
@@ -1250,6 +1250,7 @@ static inline int sk_has_allocations(const struct sock *sk)
 
 static inline void skb_set_owner_w(struct sk_buff *skb, struct sock *sk)
 {
+	skb_orphan(skb);
 	skb->sk = sk;
 	skb->destructor = sock_wfree;
 	/*

@@ -1262,6 +1263,7 @@ static inline void skb_set_owner_w(struct sk_buff *skb, struct sock *sk)
 
 static inline void skb_set_owner_r(struct sk_buff *skb, struct sock *sk)
 {
+	skb_orphan(skb);
 	skb->sk = sk;
 	skb->destructor = sock_rfree;
 	atomic_add(skb->truesize, &sk->sk_rmem_alloc);

net/ax25/ax25_in.c +1 -2

index 5f1d210..de56d39 100644
@@ -437,8 +437,7 @@ free:
 int ax25_kiss_rcv(struct sk_buff *skb, struct net_device *dev,
 		  struct packet_type *ptype, struct net_device *orig_dev)
 {
-	skb->sk = NULL;		/* Initially we don't know who it's for */
-	skb->destructor = NULL;	/* Who initializes this, dammit?! */
+	skb_orphan(skb);
 
 	if (!net_eq(dev_net(dev), &init_net)) {
 		kfree_skb(skb);

net/core/dev.c +0 -2

index baf2dc1..60b5728 100644
@@ -2310,8 +2310,6 @@ ncls:
 	if (!skb)
 		goto out;
 
-	skb_orphan(skb);
-
 	type = skb->protocol;
 	list_for_each_entry_rcu(ptype,
 			&ptype_base[ntohs(type) & PTYPE_HASH_MASK], list) {

net/irda/af_irda.c +0 -3

index 5922feb..cb762c8 100644
@@ -913,9 +913,6 @@ static int irda_accept(struct socket *sock, struct socket *newsock, int flags)
 	/* Clean up the original one to keep it in listen state */
 	irttp_listen(self->tsap);
 
-	/* Wow ! What is that ? Jean II */
-	skb->sk = NULL;
-	skb->destructor = NULL;
 	kfree_skb(skb);
 	sk->sk_ack_backlog--;

net/irda/ircomm/ircomm_lmp.c +1 -0

index 67c99d2..7ba9661 100644
@@ -196,6 +196,7 @@ static int ircomm_lmp_data_request(struct ircomm_cb *self,
 	/* Don't forget to refcount it - see ircomm_tty_do_softint() */
 	skb_get(skb);
 
+	skb_orphan(skb);
 	skb->destructor = ircomm_lmp_flow_control;
 
 	if ((self->pkt_count++ > 7) && (self->flow_status == FLOW_START)) {

David Miller

maintainer

2009-06-26 20:02 UTC #3 lore raw

From: David Miller <[email protected]>
Date: Fri, 26 Jun 2009 13:02:24 -0700 (PDT)

Quoted 11 lines

From: Andrew Morton <[email protected]>
Date: Fri, 26 Jun 2009 12:44:44 -0700
Quoted 8 lines
It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700

Quoted 6 lines

Quoted 3 lines
net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)
Then it's good that we partially reverted that change just the other day :-)

Ignore me, I'm an idiot...

Oliver Hartkopp

2009-06-26 20:52 UTC #4 lore raw

In reply to Andrew Morton

Quoted 32 lines · Andrew Morton

(switched to email. Please respond via emailed reply-to-all, not via the bugzilla web interface).
Quoted 17 lines · On Fri, 26 Jun 2009 14:45:11 GMT [email protected] wrote:
http://bugzilla.kernel.org/show_bug.cgi?id=13627
           Summary: Tunnel device ignores TCP/UDP traffic
           Product: Networking
           Version: 2.5
    Kernel Version: 2.6.31-rc1
          Platform: All
        OS/Version: Linux
              Tree: Mainline
            Status: NEW
          Severity: normal
          Priority: P1
         Component: IPV4
        AssignedTo: [email protected]
        ReportedBy: [email protected]
        Regression: Yes
It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700

Quoted 9 lines

net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)
Quoted 5 lines
Using OpenVPN on 2.6.29.4 and 2.6.30 works but 2.6.31-rc1 doesn't.
I can ping (ICMP) the remote end, and see the packets going back and forth using tcpdump, but they don't appear to be reaching the upper layers.

Sorry that i did not test tunnels also.

I just checked whether the patch from Herbert fixed the regression for PF_CAN sockets (which it does) - and of course i used 'normal' IP networking.

I hope the tunnel can be fixed without reverting the whole patch ...

Regards,
Oliver

Herbert Xu

2009-06-27 02:04 UTC #5 lore raw

In reply to Andrew Morton

Quoted 9 lines · Andrew Morton

It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700

Quoted 3 lines

net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)

Doh, I'd forgotten about transparent proxying.

inet: Call skb_orphan before tproxy activates

As transparent proxying looks up the socket early and assigns it to the skb for later processing, we must drop any existing socket ownership prior to that in order to distinguish between the case where tproxy is active and where it is not.

Signed-off-by Herbert Xu <[email protected]>

net/ipv4/ip_input.c +3 -0

index 490ce20..db46b4b 100644
@@ -440,6 +440,9 @@ int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
 	/* Remove any debris in the socket control block */
 	memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));
 
+	/* Must drop socket now because of tproxy. */
+	skb_orphan(skb);
+
 	return NF_HOOK(PF_INET, NF_INET_PRE_ROUTING, skb, dev, NULL,
 		       ip_rcv_finish);

net/ipv6/ip6_input.c +3 -0

index c3a07d7..6d6a427 100644
@@ -139,6 +139,9 @@ int ipv6_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt
 
 	rcu_read_unlock();
 
+	/* Must drop socket now because of tproxy. */
+	skb_orphan(skb);
+
 	return NF_HOOK(PF_INET6, NF_INET_PRE_ROUTING, skb, dev, NULL,
 		       ip6_rcv_finish);
 err:

Cheers,

Signature · 4 lines

Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

David Miller

maintainer Applied

2009-06-27 02:22 UTC #6 lore raw

From: Herbert Xu <[email protected]>
Date: Sat, 27 Jun 2009 10:04:03 +0800

Quoted 16 lines · Andrew Morton

Quoted 9 lines
It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700
Quoted 3 lines
net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)
Doh, I'd forgotten about transparent proxying.

Quoted 8 lines

inet: Call skb_orphan before tproxy activates
As transparent proxying looks up the socket early and assigns it to the skb for later processing, we must drop any existing socket ownership prior to that in order to distinguish between the case where tproxy is active and where it is not.
Signed-off-by Herbert Xu <[email protected]>

Applied, thanks Herbert.

Oliver Hartkopp

2009-06-27 16:37 UTC #7 lore raw

Quoted 19 lines · David Miller

From: Herbert Xu <[email protected]>
Date: Sat, 27 Jun 2009 10:04:03 +0800
Quoted 16 lines · Andrew Morton
Quoted 11 lines
It's a post-2.6.30 regression which Paul has bisected down to
commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
Author: Herbert Xu <[email protected]>
AuthorDate: Mon Jun 22 02:25:25 2009 +0000
Commit: David S. Miller <[email protected]>
CommitDate: Tue Jun 23 16:36:25 2009 -0700
net: Move rx skb_orphan call to where needed
(thanks for doing the bisection!)
Doh, I'd forgotten about transparent proxying.
inet: Call skb_orphan before tproxy activates

Quoted 8 lines

Quoted 6 lines
As transparent proxying looks up the socket early and assigns it to the skb for later processing, we must drop any existing socket ownership prior to that in order to distinguish between the case where tproxy is active and where it is not.
Signed-off-by Herbert Xu <[email protected]>
Applied, thanks Herbert.

Hi Dave,

just a reminder, if you already queued up the original patch for 2.6.30-stable, that this patch has to follow then also.

Thanks,
Oliver

David Miller

maintainer

2009-06-28 04:55 UTC #8 lore raw

From: Oliver Hartkopp <[email protected]>
Date: Sat, 27 Jun 2009 18:37:21 +0200

Quoted 7 lines

Quoted 6 lines · David Miller
From: Herbert Xu <[email protected]>
Date: Sat, 27 Jun 2009 10:04:03 +0800
Quoted 3 lines
Doh, I'd forgotten about transparent proxying.
inet: Call skb_orphan before tproxy activates

...

Quoted 2 lines

just a reminder, if you already queued up the original patch for 2.6.30-stable, that this patch has to follow then also.

Indeed, I know.

Paul Martin

2009-06-28 11:07 UTC #9 lore raw

In reply to David Miller

Quoted 8 lines · David Miller

From: Herbert Xu <[email protected]>
Date: Sat, 27 Jun 2009 10:04:03 +0800
Quoted 3 lines
inet: Call skb_orphan before tproxy activates
Signed-off-by Herbert Xu <[email protected]>
Applied, thanks Herbert.

Checked. It fixes the problem for me.

Signature · 1 line

Paul Martin <[email protected]>