Bug 13627

A git pull of Linus's tree from approximately a week ago doesn't show this problem.

Number 1 suspect is d55d87fdff8252d0e2f7c28c2d443aee17e9d70f. I'll try a recompile now with that commit reverted.

Confirmed: reverting d55d87fdff8252d0e2f7c28c2d443aee17e9d70f fixes the problem.

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

On Fri, 26 Jun 2009 14:45:11 GMT
bugzilla-daemon@bugzilla.kernel.org 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: shemminger@linux-foundation.org
>         ReportedBy: pm@debian.org
>         Regression: Yes
> 

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

  commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
  Author:     Herbert Xu <herbert@gondor.apana.org.au>
  AuthorDate: Mon Jun 22 02:25:25 2009 +0000
  Commit:     David S. Miller <davem@davemloft.net>
  CommitDate: Tue Jun 23 16:36:25 2009 -0700

      net: Move rx skb_orphan call to where needed

(thanks for doing the bisection!)

> 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

From: Andrew Morton <akpm@linux-foundation.org>
Date: Fri, 26 Jun 2009 12:44:44 -0700

> It's a post-2.6.30 regression which Paul has bisected down to
> 
>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>   Commit:     David S. Miller <davem@davemloft.net>
>   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 <herbert@gondor.apana.org.au>
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 <herbert@gondor.apana.org.au>
    Tested-by: Oliver Hartkopp <olver@hartkopp.net>
    Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/net/sctp/sctp.h b/include/net/sctp/sctp.h
index 9f80a76..d16a304 100644
--- a/include/net/sctp/sctp.h
+++ b/include/net/sctp/sctp.h
@@ -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);
diff --git a/include/net/sock.h b/include/net/sock.h
index 570c7a1..7f5c41c 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -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);
diff --git a/net/ax25/ax25_in.c b/net/ax25/ax25_in.c
index 5f1d210..de56d39 100644
--- a/net/ax25/ax25_in.c
+++ b/net/ax25/ax25_in.c
@@ -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);
diff --git a/net/core/dev.c b/net/core/dev.c
index baf2dc1..60b5728 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -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) {
diff --git a/net/irda/af_irda.c b/net/irda/af_irda.c
index 5922feb..cb762c8 100644
--- a/net/irda/af_irda.c
+++ b/net/irda/af_irda.c
@@ -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--;
 
diff --git a/net/irda/ircomm/ircomm_lmp.c b/net/irda/ircomm/ircomm_lmp.c
index 67c99d2..7ba9661 100644
--- a/net/irda/ircomm/ircomm_lmp.c
+++ b/net/irda/ircomm/ircomm_lmp.c
@@ -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)) {

From: David Miller <davem@davemloft.net>
Date: Fri, 26 Jun 2009 13:02:24 -0700 (PDT)

> From: Andrew Morton <akpm@linux-foundation.org>
> Date: Fri, 26 Jun 2009 12:44:44 -0700
> 
>> It's a post-2.6.30 regression which Paul has bisected down to
>> 
>>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>>   Commit:     David S. Miller <davem@davemloft.net>
>>   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 :-)

Ignore me, I'm an idiot...

Reply-To: oliver@hartkopp.net

Andrew Morton wrote:
> (switched to email.  Please respond via emailed reply-to-all, not via the
> bugzilla web interface).
> 
> On Fri, 26 Jun 2009 14:45:11 GMT
> bugzilla-daemon@bugzilla.kernel.org 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: shemminger@linux-foundation.org
>>         ReportedBy: pm@debian.org
>>         Regression: Yes
>>
> 
> It's a post-2.6.30 regression which Paul has bisected down to
> 
>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>   Commit:     David S. Miller <davem@davemloft.net>
>   CommitDate: Tue Jun 23 16:36:25 2009 -0700
> 
>       net: Move rx skb_orphan call to where needed
> 
> (thanks for doing the bisection!)
> 
>> 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

On Fri, Jun 26, 2009 at 12:44:44PM -0700, Andrew Morton wrote:
>
> It's a post-2.6.30 regression which Paul has bisected down to
> 
>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>   Commit:     David S. Miller <davem@davemloft.net>
>   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

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 <herbert@gondor.apana.org.au>

diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c
index 490ce20..db46b4b 100644
--- a/net/ipv4/ip_input.c
+++ b/net/ipv4/ip_input.c
@@ -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);
 
diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c
index c3a07d7..6d6a427 100644
--- a/net/ipv6/ip6_input.c
+++ b/net/ipv6/ip6_input.c
@@ -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,

From: Herbert Xu <herbert@gondor.apana.org.au>
Date: Sat, 27 Jun 2009 10:04:03 +0800

> On Fri, Jun 26, 2009 at 12:44:44PM -0700, Andrew Morton wrote:
>>
>> It's a post-2.6.30 regression which Paul has bisected down to
>> 
>>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>>   Commit:     David S. Miller <davem@davemloft.net>
>>   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
> 
> 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 <herbert@gondor.apana.org.au>

Applied, thanks Herbert.

Reply-To: oliver@hartkopp.net

David Miller wrote:
> From: Herbert Xu <herbert@gondor.apana.org.au>
> Date: Sat, 27 Jun 2009 10:04:03 +0800
> 
>> On Fri, Jun 26, 2009 at 12:44:44PM -0700, Andrew Morton wrote:
>>> It's a post-2.6.30 regression which Paul has bisected down to
>>>
>>>   commit d55d87fdff8252d0e2f7c28c2d443aee17e9d70f
>>>   Author:     Herbert Xu <herbert@gondor.apana.org.au>
>>>   AuthorDate: Mon Jun 22 02:25:25 2009 +0000
>>>   Commit:     David S. Miller <davem@davemloft.net>
>>>   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
>>
>> 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 <herbert@gondor.apana.org.au>
> 
> 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

From: Oliver Hartkopp <oliver@hartkopp.net>
Date: Sat, 27 Jun 2009 18:37:21 +0200

> David Miller wrote:
>> From: Herbert Xu <herbert@gondor.apana.org.au>
>> Date: Sat, 27 Jun 2009 10:04:03 +0800
>> 
>>> Doh, I'd forgotten about transparent proxying.
>>>
>>> inet: Call skb_orphan before tproxy activates
 ...
> 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.

On Fri, Jun 26, 2009 at 07:22:47PM -0700, David Miller wrote:
> From: Herbert Xu <herbert@gondor.apana.org.au>
> Date: Sat, 27 Jun 2009 10:04:03 +0800

> > inet: Call skb_orphan before tproxy activates

> > Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
> 
> Applied, thanks Herbert.

Checked. It fixes the problem for me.

*** Bug 13639 has been marked as a duplicate of this bug. ***

Handled-By : Herbert Xu <herbert@gondor.apana.org.au>
Patch      : http://patchwork.kernel.org/patch/32672/

*** Bug 13655 has been marked as a duplicate of this bug. ***

Fixed by commit 71f9dacd2e4d233029e9e956ca3f79531f411827 .