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;
From:   Eelco Chaudron <echaudro@redhat.com>
To:     Aaron Conole <aconole@redhat.com>
Cc:     netdev@vger.kernel.org, Pravin B Shelar <pshelar@ovn.org>,
        Jakub Kicinski <kuba@kernel.org>,
        "David S. Miller" <davem@davemloft.net>,
        Paolo Abeni <pabeni@redhat.com>,
        Eric Dumazet <edumazet@google.com>,
        Thomas Graf <tgraf@suug.ch>, dev@openvswitch.org,
        Ilya Maximets <i.maximets@ovn.org>,
        wangchuanlei <wangchuanlei@inspur.com>,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH net] net: openvswitch: release vport resources on failure
Date:   Wed, 21 Dec 2022 10:48:37 +0100
Message-ID: <DDDE110D-D874-4DB0-9A8B-D796897976F7@redhat.com>
In-Reply-To: <20221220212717.526780-1-aconole@redhat.com>
References: <20221220212717.526780-1-aconole@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable
Precedence: bulk


On 20 Dec 2022, at 22:27, Aaron Conole wrote:

> A recent commit introducing upcall packet accounting failed to properly=

> release the vport object when the per-cpu stats struct couldn't be
> allocated.  This can cause dangling pointers to dp objects long after
> they've been released.
>
> Cc: Eelco Chaudron <echaudro@redhat.com>
> Cc: wangchuanlei <wangchuanlei@inspur.com>
> Fixes: 1933ea365aa7 ("net: openvswitch: Add support to count upcall pac=
kets")
> Reported-by: syzbot+8f4e2dcfcb3209ac35f9@syzkaller.appspotmail.com
> Signed-off-by: Aaron Conole <aconole@redhat.com>
> ---

Thanks for finding and fixing this! The changes look good to me.

Acked-by: Eelco Chaudron <echaudro@redhat.com>


>  net/openvswitch/datapath.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/net/openvswitch/datapath.c b/net/openvswitch/datapath.c
> index 932bcf766d63..6774baf9e212 100644
> --- a/net/openvswitch/datapath.c
> +++ b/net/openvswitch/datapath.c
> @@ -1854,7 +1854,7 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, st=
ruct genl_info *info)
>  	vport->upcall_stats =3D netdev_alloc_pcpu_stats(struct vport_upcall_s=
tats_percpu);
>  	if (!vport->upcall_stats) {
>  		err =3D -ENOMEM;
> -		goto err_destroy_portids;
> +		goto err_destroy_vport;
>  	}
>
>  	err =3D ovs_dp_cmd_fill_info(dp, reply, info->snd_portid,
> @@ -1869,6 +1869,8 @@ static int ovs_dp_cmd_new(struct sk_buff *skb, st=
ruct genl_info *info)
>  	ovs_notify(&dp_datapath_genl_family, reply, info);
>  	return 0;
>
> +err_destroy_vport:
> +	ovs_dp_detach_port(vport);
>  err_destroy_portids:
>  	kfree(rcu_dereference_raw(dp->upcall_portids));
>  err_unlock_and_destroy_meters:
> @@ -2316,7 +2318,7 @@ static int ovs_vport_cmd_new(struct sk_buff *skb,=
 struct genl_info *info)
>  	vport->upcall_stats =3D netdev_alloc_pcpu_stats(struct vport_upcall_s=
tats_percpu);
>  	if (!vport->upcall_stats) {
>  		err =3D -ENOMEM;
> -		goto exit_unlock_free;
> +		goto exit_unlock_free_vport;
>  	}
>
>  	err =3D ovs_vport_cmd_fill_info(vport, reply, genl_info_net(info),
> @@ -2336,6 +2338,8 @@ static int ovs_vport_cmd_new(struct sk_buff *skb,=
 struct genl_info *info)
>  	ovs_notify(&dp_vport_genl_family, reply, info);
>  	return 0;
>
> +exit_unlock_free_vport:
> +	ovs_dp_detach_port(vport);
>  exit_unlock_free:
>  	ovs_unlock();
>  	kfree_skb(reply);
> -- =

> 2.31.1