[PATCH 1/1] net: possible use after free in dst_release

[PATCH 1/1] net: possible use after free in dst_release

[Francesco Ruggeri]

dst_release should not access dst->flags after decrementing
__refcnt to 0. The dst_entry may be in dst_busy_list and
dst_gc_task may dst_destroy it before dst_release gets a chance
to access dst->flags.

Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
---
 net/core/dst.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/net/core/dst.c b/net/core/dst.c
index e6dc772..a1656e3 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -301,12 +301,13 @@ void dst_release(struct dst_entry *dst)
 {
 	if (dst) {
 		int newrefcnt;
+		unsigned short nocache = dst->flags & DST_NOCACHE;
 
 		newrefcnt = atomic_dec_return(&dst->__refcnt);
 		if (unlikely(newrefcnt < 0))
 			net_warn_ratelimited("%s: dst:%p refcnt:%d\n",
 					     __func__, dst, newrefcnt);
-		if (!newrefcnt && unlikely(dst->flags & DST_NOCACHE))
+		if (!newrefcnt && unlikely(nocache))
 			call_rcu(&dst->rcu_head, dst_destroy_rcu);
 	}
 }
-- 
1.8.1.4

Re: [PATCH 1/1] net: possible use after free in dst_release

[Eric Dumazet]

On Wed, 2016-01-06 at 00:18 -0800, Francesco Ruggeri wrote:
> dst_release should not access dst->flags after decrementing
> __refcnt to 0. The dst_entry may be in dst_busy_list and
> dst_gc_task may dst_destroy it before dst_release gets a chance
> to access dst->flags.
> 
> Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
> ---
>  net/core/dst.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/net/core/dst.c b/net/core/dst.c
> index e6dc772..a1656e3 100644
> --- a/net/core/dst.c
> +++ b/net/core/dst.c
> @@ -301,12 +301,13 @@ void dst_release(struct dst_entry *dst)
>  {
>  	if (dst) {
>  		int newrefcnt;
> +		unsigned short nocache = dst->flags & DST_NOCACHE;
>  
>  		newrefcnt = atomic_dec_return(&dst->__refcnt);
>  		if (unlikely(newrefcnt < 0))
>  			net_warn_ratelimited("%s: dst:%p refcnt:%d\n",
>  					     __func__, dst, newrefcnt);
> -		if (!newrefcnt && unlikely(dst->flags & DST_NOCACHE))
> +		if (!newrefcnt && unlikely(nocache))
>  			call_rcu(&dst->rcu_head, dst_destroy_rcu);
>  	}
>  }

You're right.

My previous fix (commit d69bbf88c8d0) was okay only for DST_NOCACHE dst.

Fixes: d69bbf88c8d0 ("net: fix a race in dst_release()")
Fixes: 27b75c95f10d ("net: avoid RCU for NOCACHE dst")
Acked-by: Eric Dumazet <edumazet@google.com>

Thanks !

Re: [PATCH 1/1] net: possible use after free in dst_release

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 06 Jan 2016 05:57:24 -0800

> On Wed, 2016-01-06 at 00:18 -0800, Francesco Ruggeri wrote:
>> dst_release should not access dst->flags after decrementing
>> __refcnt to 0. The dst_entry may be in dst_busy_list and
>> dst_gc_task may dst_destroy it before dst_release gets a chance
>> to access dst->flags.
>> 
>> Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
>> ---
>>  net/core/dst.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>> 
>> diff --git a/net/core/dst.c b/net/core/dst.c
>> index e6dc772..a1656e3 100644
>> --- a/net/core/dst.c
>> +++ b/net/core/dst.c
>> @@ -301,12 +301,13 @@ void dst_release(struct dst_entry *dst)
>>  {
>>  	if (dst) {
>>  		int newrefcnt;
>> +		unsigned short nocache = dst->flags & DST_NOCACHE;
>>  
>>  		newrefcnt = atomic_dec_return(&dst->__refcnt);
>>  		if (unlikely(newrefcnt < 0))
>>  			net_warn_ratelimited("%s: dst:%p refcnt:%d\n",
>>  					     __func__, dst, newrefcnt);
>> -		if (!newrefcnt && unlikely(dst->flags & DST_NOCACHE))
>> +		if (!newrefcnt && unlikely(nocache))
>>  			call_rcu(&dst->rcu_head, dst_destroy_rcu);
>>  	}
>>  }
> 
> You're right.
> 
> My previous fix (commit d69bbf88c8d0) was okay only for DST_NOCACHE dst.
> 
> Fixes: d69bbf88c8d0 ("net: fix a race in dst_release()")
> Fixes: 27b75c95f10d ("net: avoid RCU for NOCACHE dst")
> Acked-by: Eric Dumazet <edumazet@google.com>

Applied and queued up for -stable, thanks everyone!