Re: [RFC PATCH] accounting for socket backlog

[Eric Dumazet <eric.dumazet@gmail.com>]

Le jeudi 25 février 2010 à 11:13 +0800, Zhu Yi a écrit :
> Hi,
> 
> We got system OOM while running some UDP netperf testing on the loopback
> device. The case is multiple senders sent stream UDP packets to a single
> receiver via loopback on local host. Of course, the receiver is not able
> to handle all the packets in time. But we surprisingly found that these
> packets were not discarded due to the receiver's sk->sk_rcvbuf limit.
> Instead, they are kept queuing to sk->sk_backlog and finally ate up all
> the memory. We believe this is a secure hole that a none privileged user
> can crash the system.
> 
> The root cause for this problem is, when the receiver is doing
> __release_sock() (i.e. after userspace recv, kernel udp_recvmsg ->
> skb_free_datagram_locked -> release_sock), it moves skbs from backlog to
> sk_receive_queue with the softirq enabled. In the above case, multiple
> busy senders will almost make it an endless loop. The skbs in the
> backlog end up eat all the system memory.
> 
> The patch fixed this problem by adding accounting for the socket
> backlog. So that the backlog size can be restricted by protocol's choice
> (i.e. UDP).
> 
> Signed-off-by: Zhu Yi <yi.zhu@intel.com>
> ---
> diff --git a/include/net/sock.h b/include/net/sock.h
> index 3f1a480..2e003b9 100644
> --- a/include/net/sock.h
> +++ b/include/net/sock.h
> @@ -253,6 +253,7 @@ struct sock {
>  	struct {
>  		struct sk_buff *head;
>  		struct sk_buff *tail;
> +		atomic_t len;

This adds a hole on 32bit arches.

I am pretty sure we dont need an atomic here, since we must own a lock
before manipulating sk_backlog{head,tail,len}.

UDP/IPV6 should be addressed too in your patch.

Other questions raised by your discovery :
- What about other protocols that also use a backlog ?
- __release_sock() could run forever with no preemption, even with a
limit on backlog.