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;
Feedback-ID: i494840e7:Fastmail
Date:   Fri, 4 Aug 2023 16:24:08 +0300
From:   Ido Schimmel <idosch@idosch.org>
To:     Fedor Pchelkin <pchelkin@ispras.ru>
Cc:     Roopa Prabhu <roopa@nvidia.com>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Nikolay Aleksandrov <razor@blackwall.org>,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        Alexey Khoroshilov <khoroshilov@ispras.ru>,
        lvc-project@linuxtesting.org
Subject: Re: [PATCH] drivers: vxlan: vnifilter: free percpu vni stats on
 error path
Message-ID: <ZMz7+Mk+OYj4q8xe@shredder>
References: <20230803193834.23340-1-pchelkin@ispras.ru>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230803193834.23340-1-pchelkin@ispras.ru>
Precedence: bulk

Prefix should be "PATCH net". See:
https://www.kernel.org/doc/html/latest/process/maintainer-netdev.html

On Thu, Aug 03, 2023 at 10:38:32PM +0300, Fedor Pchelkin wrote:
> In case rhashtable_lookup_insert_fast() fails inside vxlan_vni_add(), the
> allocated percpu vni stats are not freed on the error path.
> 
> Free them on the rhashtable_lookup_insert_fast() error path in
> vxlan_vni_add().
> 
> Found by Linux Verification Center (linuxtesting.org).
> 
> Fixes: 4095e0e1328a ("drivers: vxlan: vnifilter: per vni stats")
> Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
> ---
>  drivers/net/vxlan/vxlan_vnifilter.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/vxlan/vxlan_vnifilter.c b/drivers/net/vxlan/vxlan_vnifilter.c
> index a3de081cda5e..321cd0b450cc 100644
> --- a/drivers/net/vxlan/vxlan_vnifilter.c
> +++ b/drivers/net/vxlan/vxlan_vnifilter.c
> @@ -740,6 +740,7 @@ static int vxlan_vni_add(struct vxlan_dev *vxlan,
>  					    &vninode->vnode,
>  					    vxlan_vni_rht_params);
>  	if (err) {
> +		free_percpu(vninode->stats);

This oversight (and future ones) wouldn't have happened if
vxlan_vni_alloc() had a corresponding vxlan_vni_free(). I suggest
something like [1].

BTW, I think the GFP_ATOMIC in vxlan_vni_alloc() should be GFP_KERNEL. I
will take care of it in net-next.

>  		kfree(vninode);
>  		return err;
>  	}
> -- 
> 2.41.0

[1]
diff --git a/drivers/net/vxlan/vxlan_vnifilter.c b/drivers/net/vxlan/vxlan_vnifilter.c
index a3de081cda5e..c3ff30ab782e 100644
--- a/drivers/net/vxlan/vxlan_vnifilter.c
+++ b/drivers/net/vxlan/vxlan_vnifilter.c
@@ -713,6 +713,12 @@ static struct vxlan_vni_node *vxlan_vni_alloc(struct vxlan_dev *vxlan,
        return vninode;
 }
 
+static void vxlan_vni_free(struct vxlan_vni_node *vninode)
+{
+       free_percpu(vninode->stats);
+       kfree(vninode);
+}
+
 static int vxlan_vni_add(struct vxlan_dev *vxlan,
                         struct vxlan_vni_group *vg,
                         u32 vni, union vxlan_addr *group,
@@ -740,7 +746,7 @@ static int vxlan_vni_add(struct vxlan_dev *vxlan,
                                            &vninode->vnode,
                                            vxlan_vni_rht_params);
        if (err) {
-               kfree(vninode);
+               vxlan_vni_free(vninode);
                return err;
        }
 
@@ -763,8 +769,7 @@ static void vxlan_vni_node_rcu_free(struct rcu_head *rcu)
        struct vxlan_vni_node *v;
 
        v = container_of(rcu, struct vxlan_vni_node, rcu);
-       free_percpu(v->stats);
-       kfree(v);
+       vxlan_vni_free(v);
 }
 
 static int vxlan_vni_del(struct vxlan_dev *vxlan,