diff mbox

[RFC] dst: check if dst is freed in dst_check()

Message ID 4C457120.9070105@6wind.com
State RFC, archived
Delegated to: David Miller
Headers show

Commit Message

Nicolas Dichtel July 20, 2010, 9:49 a.m. UTC
Hi,

I probably missed something, but I cannot find where obsolete field is checked 
when dst_check() is called. If dst->obsolete is > 1, dst cannot be used!

Attached is a proposal to fix this issue.


Regards,

Comments

Eric Dumazet July 21, 2010, 2:28 a.m. UTC | #1
Le mardi 20 juillet 2010 à 11:49 +0200, Nicolas Dichtel a écrit :
> Hi,
> 
> I probably missed something, but I cannot find where obsolete field is checked 
> when dst_check() is called. If dst->obsolete is > 1, dst cannot be used!
> 
> Attached is a proposal to fix this issue.
> 
> 

> diff --git a/include/net/dst.h b/include/net/dst.h
> index 81d1413..7bf4f9a 100644
> --- a/include/net/dst.h
> +++ b/include/net/dst.h
> @@ -319,6 +319,8 @@ static inline int dst_input(struct sk_buff *skb)
>  
>  static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
>  {
> +	if (dst->obsolete > 1)
> +		return NULL;
>  	if (dst->obsolete)
>  		dst = dst->ops->check(dst, cookie);
>  	return dst;

I believe this is not needed and redundant.

In what case do you think this matters ?

To my knowledge dst_check() is only used by net/xfrm/xfrm_policy.c

And xfrm_dst_check() does the necessary checks.

static struct dst_entry *xfrm_dst_check(struct dst_entry *dst, u32 cookie)
{
        /* Code (such as __xfrm4_bundle_create()) sets dst->obsolete
         * to "-1" to force all XFRM destinations to get validated by
         * dst_ops->check on every use.  We do this because when a
         * normal route referenced by an XFRM dst is obsoleted we do
         * not go looking around for all parent referencing XFRM dsts
         * so that we can invalidate them.  It is just too much work.
         * Instead we make the checks here on every use.  For example:
         *
         *      XFRM dst A --> IPv4 dst X
         *
         * X is the "xdst->route" of A (X is also the "dst->path" of A
         * in this example).  If X is marked obsolete, "A" will not
         * notice.  That's what we are validating here via the
         * stale_bundle() check.
         *
         * When a policy's bundle is pruned, we dst_free() the XFRM
         * dst which causes it's ->obsolete field to be set to a
         * positive non-zero integer.  If an XFRM dst has been pruned
         * like this, we want to force a new route lookup.
         */
        if (dst->obsolete < 0 && !stale_bundle(dst))
                return dst;

        return NULL;
}


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller July 21, 2010, 6:41 a.m. UTC | #2
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 21 Jul 2010 04:28:08 +0200

> Le mardi 20 juillet 2010 à 11:49 +0200, Nicolas Dichtel a écrit :
>> diff --git a/include/net/dst.h b/include/net/dst.h
>> index 81d1413..7bf4f9a 100644
>> --- a/include/net/dst.h
>> +++ b/include/net/dst.h
>> @@ -319,6 +319,8 @@ static inline int dst_input(struct sk_buff *skb)
>>  
>>  static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
>>  {
>> +	if (dst->obsolete > 1)
>> +		return NULL;
>>  	if (dst->obsolete)
>>  		dst = dst->ops->check(dst, cookie);
>>  	return dst;
> 
> I believe this is not needed and redundant.
> 
> In what case do you think this matters ?
> 
> To my knowledge dst_check() is only used by net/xfrm/xfrm_policy.c
> 
> And xfrm_dst_check() does the necessary checks.

Right, last time I was snooping around in here I came to the
same conclusion.  In fact I think I'm the author of that
enormous comment in xfrm_dst_check(). :-)


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Nicolas Dichtel July 21, 2010, 7:03 a.m. UTC | #3
Hi Eric,

I was thinking to use this function in sctp, but I misread xfrm part.
Sorry for the noise.


Regards,
Nicolas

Le 21.07.2010 04:28, Eric Dumazet a écrit :
> Le mardi 20 juillet 2010 à 11:49 +0200, Nicolas Dichtel a écrit :
>> Hi,
>>
>> I probably missed something, but I cannot find where obsolete field is checked 
>> when dst_check() is called. If dst->obsolete is > 1, dst cannot be used!
>>
>> Attached is a proposal to fix this issue.
>>
>>
> 
>> diff --git a/include/net/dst.h b/include/net/dst.h
>> index 81d1413..7bf4f9a 100644
>> --- a/include/net/dst.h
>> +++ b/include/net/dst.h
>> @@ -319,6 +319,8 @@ static inline int dst_input(struct sk_buff *skb)
>>  
>>  static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
>>  {
>> +	if (dst->obsolete > 1)
>> +		return NULL;
>>  	if (dst->obsolete)
>>  		dst = dst->ops->check(dst, cookie);
>>  	return dst;
> 
> I believe this is not needed and redundant.
> 
> In what case do you think this matters ?
> 
> To my knowledge dst_check() is only used by net/xfrm/xfrm_policy.c
> 
> And xfrm_dst_check() does the necessary checks.
> 
> static struct dst_entry *xfrm_dst_check(struct dst_entry *dst, u32 cookie)
> {
>         /* Code (such as __xfrm4_bundle_create()) sets dst->obsolete
>          * to "-1" to force all XFRM destinations to get validated by
>          * dst_ops->check on every use.  We do this because when a
>          * normal route referenced by an XFRM dst is obsoleted we do
>          * not go looking around for all parent referencing XFRM dsts
>          * so that we can invalidate them.  It is just too much work.
>          * Instead we make the checks here on every use.  For example:
>          *
>          *      XFRM dst A --> IPv4 dst X
>          *
>          * X is the "xdst->route" of A (X is also the "dst->path" of A
>          * in this example).  If X is marked obsolete, "A" will not
>          * notice.  That's what we are validating here via the
>          * stale_bundle() check.
>          *
>          * When a policy's bundle is pruned, we dst_free() the XFRM
>          * dst which causes it's ->obsolete field to be set to a
>          * positive non-zero integer.  If an XFRM dst has been pruned
>          * like this, we want to force a new route lookup.
>          */
>         if (dst->obsolete < 0 && !stale_bundle(dst))
>                 return dst;
> 
>         return NULL;
> }
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

From 69990a516f4b5b48608b0ea283dfac6f1fa110b3 Mon Sep 17 00:00:00 2001
From: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Date: Tue, 20 Jul 2010 11:35:53 +0200
Subject: [PATCH] dst: check if dst is freed in dst_check()

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
 include/net/dst.h |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/include/net/dst.h b/include/net/dst.h
index 81d1413..7bf4f9a 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -319,6 +319,8 @@  static inline int dst_input(struct sk_buff *skb)
 
 static inline struct dst_entry *dst_check(struct dst_entry *dst, u32 cookie)
 {
+	if (dst->obsolete > 1)
+		return NULL;
 	if (dst->obsolete)
 		dst = dst->ops->check(dst, cookie);
 	return dst;
-- 
1.5.4.5