Re: [PATCH net] net: 802: enforce underlying device type for GARP and MRP

[Ido Schimmel <idosch@idosch.org>]

On Wed, Feb 12, 2025 at 12:32:18PM +0100, Oscar Maes wrote:
> When creating a VLAN device, we initialize GARP (garp_init_applicant)
> and MRP (mrp_init_applicant) for the underlying device.
> 
> As part of the initialization process, we add the multicast address of
> each applicant to the underlying device, by calling dev_mc_add.
> 
> __dev_mc_add uses dev->addr_len to determine the length of the new
> multicast address.
> 
> This causes an out-of-bounds read if dev->addr_len is greater than 6,
> since the multicast addresses provided by GARP and MRP are only 6 bytes
> long.
> 
> This behaviour can be reproduced using the following commands:
> 
> ip tunnel add gretest mode ip6gre local ::1 remote ::2 dev lo
> ip l set up dev gretest
> ip link add link gretest name vlantest type vlan id 100
> 
> Then, the following command will display the address of garp_pdu_rcv:
> 
> ip maddr show | grep 01:80:c2:00:00:21
> 
> Fix this by enforcing the type and address length of
> the underlying device during GARP and MRP initialization.
> 
> Fixes: 22bedad3ce11 ("net: convert multicast list to list_head")
> Reported-by: syzbot <syzbot+91161fe81857b396c8a0@syzkaller.appspotmail.com>
> Closes: https://lore.kernel.org/netdev/000000000000ca9a81061a01ec20@google.com/
> Signed-off-by: Oscar Maes <oscmaes92@gmail.com>
> ---
>  net/802/garp.c | 5 +++++
>  net/802/mrp.c  | 5 +++++
>  2 files changed, 10 insertions(+)
> 
> diff --git a/net/802/garp.c b/net/802/garp.c
> index 27f0ab146..2f383ee73 100644
> --- a/net/802/garp.c
> +++ b/net/802/garp.c
> @@ -9,6 +9,7 @@
>  #include <linux/skbuff.h>
>  #include <linux/netdevice.h>
>  #include <linux/etherdevice.h>
> +#include <linux/if_arp.h>
>  #include <linux/rtnetlink.h>
>  #include <linux/llc.h>
>  #include <linux/slab.h>
> @@ -574,6 +575,10 @@ int garp_init_applicant(struct net_device *dev, struct garp_application *appl)
>  
>  	ASSERT_RTNL();
>  
> +	err = -EINVAL;
> +	if (dev->type != ARPHRD_ETHER || dev->addr_len != ETH_ALEN)

Checking for 'ARPHRD_ETHER' is not enough? Other virtual devices such as
macsec, macvlan and ipvlan only look at 'dev->type' AFAICT.

Also, how about moving this to vlan_check_real_dev()? It's common to
both the IOCTL and netlink paths.

> +		goto err1;
> +
>  	if (!rtnl_dereference(dev->garp_port)) {
>  		err = garp_init_port(dev);
>  		if (err < 0)
> diff --git a/net/802/mrp.c b/net/802/mrp.c
> index e0c96d0da..1efee0b39 100644
> --- a/net/802/mrp.c
> +++ b/net/802/mrp.c
> @@ -12,6 +12,7 @@
>  #include <linux/skbuff.h>
>  #include <linux/netdevice.h>
>  #include <linux/etherdevice.h>
> +#include <linux/if_arp.h>
>  #include <linux/rtnetlink.h>
>  #include <linux/slab.h>
>  #include <linux/module.h>
> @@ -859,6 +860,10 @@ int mrp_init_applicant(struct net_device *dev, struct mrp_application *appl)
>  
>  	ASSERT_RTNL();
>  
> +	err = -EINVAL;
> +	if (dev->type != ARPHRD_ETHER || dev->addr_len != ETH_ALEN)
> +		goto err1;
> +
>  	if (!rtnl_dereference(dev->mrp_port)) {
>  		err = mrp_init_port(dev);
>  		if (err < 0)
> -- 
> 2.39.5
> 
>