Re: [PATCH 4.3 34/71] vrf: fix double free and memory corruption on register_netdevice failure

[Ben Hutchings]

On Sat, 2015-12-12 at 12:05 -0800, Greg Kroah-Hartman wrote:
> 4.3-stable review patch.ÂÂIf anyone has any objections, please let me
> know.
> 
> ------------------
> 
> From: Nikolay Aleksandrov <nikolay@xxxxxxxxxxxxxxxxxxx>
> 
> [ Upstream commit 7f109f7cc37108cba7243bc832988525b0d85909 ]
[...]
> --- a/drivers/net/vrf.c
> +++ b/drivers/net/vrf.c
> @@ -581,7 +581,6 @@ static int vrf_newlink(struct net *src_n
> Â{
> Â	struct net_vrf *vrf = netdev_priv(dev);
> Â	struct net_vrf_dev *vrf_ptr;
> -	int err;
> Â
> Â	if (!data || !data[IFLA_VRF_TABLE])
> Â		return -EINVAL;
> @@ -590,26 +589,16 @@ static int vrf_newlink(struct net *src_n
> Â
> Â	dev->priv_flags |= IFF_VRF_MASTER;
> Â
> -	err = -ENOMEM;
> Â	vrf_ptr = kmalloc(sizeof(*dev->vrf_ptr), GFP_KERNEL);
> Â	if (!vrf_ptr)
> -		goto out_fail;
> +		return -ENOMEM;
> Â
> Â	vrf_ptr->ifindex = dev->ifindex;
> Â	vrf_ptr->tb_id = vrf->tb_id;
> Â
> -	err = register_netdevice(dev);
> -	if (err < 0)
> -		goto out_fail;
> -
> Â	rcu_assign_pointer(dev->vrf_ptr, vrf_ptr);
> Â
> -	return 0;
> -
> -out_fail:
> -	kfree(vrf_ptr);
> -	free_netdev(dev);
> -	return err;
> +	return register_netdev(dev);
> Â}
> Â
> Âstatic size_t vrf_nl_getsize(const struct net_device *dev)

This leaks *dev->vrf_ptr if register_netdevice() fails. Â(This bug does
not exist in the mainline version, as net_device::vrf_ptr no longer
exists there.)

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

Attachment: signature.asc
Description: This is a digitally signed message part