[PATCH] netfilter: don't reset nf_trace in nf_reset()

[PATCH] netfilter: don't reset nf_trace in nf_reset()

[Patrick McHardy]

Commit 130549fe added code to reset nf_trace in nf_reset(). This is wrong
and unnecessary.

nf_reset() is used in the following cases:

- when passing packets up the the socket layer, at which point we want to
  release all netfilter references that might keep modules pinned while
  the packet is queued. nf_trace doesn't matter anymore at this point.

- when encapsulating or decapsulating IPsec packets. We want to continue
  tracing these packets after IPsec processing.

- when passing packets through virtual network devices. Only devices on
  that encapsulate in IPv4/v6 matter since otherwise nf_trace is not
  used anymore. Its not entirely clear whether those packets should
  be traced after that, however we've always done that.

- when passing packets through virtual network devices that make the
  packet cross network namespace boundaries. This is the only cases
  where we clearly want to reset nf_trace and is also what the
  original patch intended to fix.

Add a new function nf_reset_trace() and use it in dev_forward_skb() to
fix this properly.

Signed-off-by: Patrick McHardy <kaber@trash.net>
---
 include/linux/skbuff.h | 4 ++++
 net/core/dev.c         | 1 +
 2 files changed, 5 insertions(+)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 72b3967..b8292d8 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -2641,6 +2641,10 @@ static inline void nf_reset(struct sk_buff *skb)
 	nf_bridge_put(skb->nf_bridge);
 	skb->nf_bridge = NULL;
 #endif
+}
+
+static inline void nf_reset_trace(struct sk_buff *skb)
+{
 #if IS_ENABLED(CONFIG_NETFILTER_XT_TARGET_TRACE)
 	skb->nf_trace = 0;
 #endif
diff --git a/net/core/dev.c b/net/core/dev.c
index 13e6447..e7d68ed 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -1639,6 +1639,7 @@ int dev_forward_skb(struct net_device *dev, struct sk_buff *skb)
 	skb->mark = 0;
 	secpath_reset(skb);
 	nf_reset(skb);
+	nf_reset_trace(skb);
 	return netif_rx(skb);
 }
 EXPORT_SYMBOL_GPL(dev_forward_skb);
-- 
1.8.1.4

Re: [PATCH] netfilter: don't reset nf_trace in nf_reset()

[David Miller]

From: Patrick McHardy <kaber@trash.net>
Date: Fri,  5 Apr 2013 20:42:05 +0200

> Commit 130549fe added code to reset nf_trace in nf_reset(). This is wrong
> and unnecessary.
> 
> nf_reset() is used in the following cases:
> 
> - when passing packets up the the socket layer, at which point we want to
>   release all netfilter references that might keep modules pinned while
>   the packet is queued. nf_trace doesn't matter anymore at this point.
> 
> - when encapsulating or decapsulating IPsec packets. We want to continue
>   tracing these packets after IPsec processing.
> 
> - when passing packets through virtual network devices. Only devices on
>   that encapsulate in IPv4/v6 matter since otherwise nf_trace is not
>   used anymore. Its not entirely clear whether those packets should
>   be traced after that, however we've always done that.
> 
> - when passing packets through virtual network devices that make the
>   packet cross network namespace boundaries. This is the only cases
>   where we clearly want to reset nf_trace and is also what the
>   original patch intended to fix.
> 
> Add a new function nf_reset_trace() and use it in dev_forward_skb() to
> fix this properly.
> 
> Signed-off-by: Patrick McHardy <kaber@trash.net>

I'll apply this directly because I'm trying to get a pull request out
to Linus.

Re: [PATCH] netfilter: don't reset nf_trace in nf_reset()

[Sergei Shtylyov]

On 04/05/2013 10:42 PM, Patrick McHardy wrote:

> Commit 130549fe

    Please also specify the summary line of this commit in parens.
David M. also seems to require it to be enclosed in quotes inside parens...

> added code to reset nf_trace in nf_reset(). This is wrong
> and unnecessary.
>
> nf_reset() is used in the following cases:
>
> - when passing packets up the the socket layer, at which point we want to
>    release all netfilter references that might keep modules pinned while
>    the packet is queued. nf_trace doesn't matter anymore at this point.
>
> - when encapsulating or decapsulating IPsec packets. We want to continue
>    tracing these packets after IPsec processing.
>
> - when passing packets through virtual network devices. Only devices on
>    that encapsulate in IPv4/v6 matter since otherwise nf_trace is not
>    used anymore. Its not entirely clear whether those packets should
>    be traced after that, however we've always done that.
>
> - when passing packets through virtual network devices that make the
>    packet cross network namespace boundaries. This is the only cases
>    where we clearly want to reset nf_trace and is also what the
>    original patch intended to fix.
>
> Add a new function nf_reset_trace() and use it in dev_forward_skb() to
> fix this properly.
>
> Signed-off-by: Patrick McHardy <kaber@trash.net>
>

WBR, Sergei

Re: [PATCH] netfilter: don't reset nf_trace in nf_reset()

[Patrick McHardy]

Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> schrieb:

>On 04/05/2013 10:42 PM, Patrick McHardy wrote:
>
>> Commit 130549fe
>
>    Please also specify the summary line of this commit in parens.
>David M. also seems to require it to be enclosed in quotes inside
>parens...

I thought I did. Well, apparently I forgot.

Re: [PATCH] netfilter: don't reset nf_trace in nf_reset()

[Gao feng]

On 2013/04/06 02:42, Patrick McHardy wrote:
> Commit 130549fe added code to reset nf_trace in nf_reset(). This is wrong
> and unnecessary.
> 
> nf_reset() is used in the following cases:
> 
> - when passing packets up the the socket layer, at which point we want to
>   release all netfilter references that might keep modules pinned while
>   the packet is queued. nf_trace doesn't matter anymore at this point.
> 
> - when encapsulating or decapsulating IPsec packets. We want to continue
>   tracing these packets after IPsec processing.
> 
> - when passing packets through virtual network devices. Only devices on
>   that encapsulate in IPv4/v6 matter since otherwise nf_trace is not
>   used anymore. Its not entirely clear whether those packets should
>   be traced after that, however we've always done that.
> 
> - when passing packets through virtual network devices that make the
>   packet cross network namespace boundaries. This is the only cases
>   where we clearly want to reset nf_trace and is also what the
>   original patch intended to fix.
> 
> Add a new function nf_reset_trace() and use it in dev_forward_skb() to
> fix this properly.
> 
> Signed-off-by: Patrick McHardy <kaber@trash.net>

Ok, I was confused by the nf_*reset*...
This patch is more proper,Since my commit will change the behavior of TRACE in some case.

Thanks for your fix and explanation.