Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Sat, 18 Mar 2023 11:38:38 +0000
From:   Bobby Eshleman <bobbyeshleman@gmail.com>
To:     Stefano Garzarella <sgarzare@redhat.com>
Cc:     Bobby Eshleman <bobby.eshleman@bytedance.com>,
        Stefan Hajnoczi <stefanha@redhat.com>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>, kvm@vger.kernel.org,
        virtualization@lists.linux-foundation.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, Cong Wang <xiyou.wangcong@gmail.com>
Subject: Re: [PATCH net] virtio/vsock: fix leak due to missing skb owner
Message-ID: <ZBWivg3d7G/ETqqo@bullseye>
References: <20230327-vsock-fix-leak-v1-1-3fede367105f@bytedance.com>
 <jinx5oduhddyyaxnreey2riem3s7ju5zuszddmoiie6dcnyiiy@fr4cg33vi7aq>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <jinx5oduhddyyaxnreey2riem3s7ju5zuszddmoiie6dcnyiiy@fr4cg33vi7aq>
Precedence: bulk

On Tue, Mar 28, 2023 at 09:58:00AM +0200, Stefano Garzarella wrote:
> On Mon, Mar 27, 2023 at 10:01:05PM +0000, Bobby Eshleman wrote:
> > This patch sets the owner for the skb when being sent from a socket and
> > so solves the leak caused when virtio_transport_purge_skbs() finds
> > skb->sk is always NULL and therefore never matches it with the current
> > socket. Setting the owner upon allocation fixes this.
> > 
> > Fixes: 71dc9ec9ac7d ("virtio/vsock: replace virtio_vsock_pkt with sk_buff")
> > Signed-off-by: Bobby Eshleman <bobby.eshleman@bytedance.com>
> > Reported-by: Cong Wang <xiyou.wangcong@gmail.com>
> > Link: https://lore.kernel.org/all/ZCCbATwov4U+GBUv@pop-os.localdomain/
> > ---
> > net/vmw_vsock/virtio_transport_common.c | 3 +++
> > 1 file changed, 3 insertions(+)
> > 
> > diff --git a/net/vmw_vsock/virtio_transport_common.c b/net/vmw_vsock/virtio_transport_common.c
> > index 957cdc01c8e8..2a2f0c1a9fbd 100644
> > --- a/net/vmw_vsock/virtio_transport_common.c
> > +++ b/net/vmw_vsock/virtio_transport_common.c
> > @@ -94,6 +94,9 @@ virtio_transport_alloc_skb(struct virtio_vsock_pkt_info *info,
> > 					 info->op,
> > 					 info->flags);
> > 
> > +	if (info->vsk)
> > +		skb_set_owner_w(skb, sk_vsock(info->vsk));
> > +
> 
> Should we do the same also in virtio_transport_recv_pkt()?
> 
> The skb in that cases is allocated in drivers/vhost/vsock.c and
> net/vmw_vsock/virtio_transport.c using directly
> virtio_vsock_alloc_skb(), because we don't know in advance which socket
> it belongs to.
> 
> Then in virtio_transport_recv_pkt() we look for the socket and queue it
> up. This should also solve the problem in vsock_loopback.c where we move
> skb from one socket to another.
> 

That's a great point, skb_set_owner_r() in recv_pkt() will do all of the
right accounting when called by vsock_loopback_work.

I'll add that in a v2.

Thanks,
Bobby