Re: [PATCH] net: qualcomm: rmnet: fix endpoint use-after-free in rmnet_dellink()

[Paolo Abeni <pabeni@redhat.com>]

On 5/11/26 2:00 PM, Weiming Shi wrote:
> From: Security Analysis <xmei5@asu.edu>
> 
> rmnet_dellink() removes the endpoint from the hash table with
> hlist_del_init_rcu() and then immediately frees it with kfree(). However,
> RCU readers on the receive path (rmnet_rx_handler ->
> __rmnet_map_ingress_handler) may still hold a reference to the endpoint and
> dereference ep->egress_dev after the memory has been freed. The endpoint is
> a kmalloc-32 object, and the stale read at offset 8 corresponds to the
> egress_dev pointer.
> 
>   BUG: unable to handle page fault for address: ffffffffde942eef
>   Oops: 0002 [#1] SMP NOPTI
>   CPU: 1 UID: 0 PID: 137 Comm: poc_write Not tainted 7.0.0+ #4 PREEMPTLAZY
>   RIP: 0010:rmnet_vnd_rx_fixup (rmnet_vnd.c:27)
>   Call Trace:
>    <TASK>
>    __rmnet_map_ingress_handler (rmnet_handlers.c:48 rmnet_handlers.c:101)
>    rmnet_rx_handler (rmnet_handlers.c:129 rmnet_handlers.c:235)
>    __netif_receive_skb_core.constprop.0 (net/core/dev.c:6096)
>    __netif_receive_skb_one_core (net/core/dev.c:6208)
>    netif_receive_skb (net/core/dev.c:6467)
>    tun_get_user (drivers/net/tun.c:1955)
>    tun_chr_write_iter (drivers/net/tun.c:2003)
>    vfs_write (fs/read_write.c:688)
>    ksys_write (fs/read_write.c:740)
>    </TASK>
> 
> Replace kfree() with kfree_rcu_mightsleep() so the endpoint memory remains
> valid through the RCU grace period. Also remove the rmnet_vnd_dellink() call
> and inline only the nr_rmnet_devs decrement, since rmnet_vnd_dellink() would
> set ep->egress_dev to NULL during the grace period, creating a data race with
> lockless readers.
> 
> Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial implementation")
> Assisted-by: Claude:claude-opus-4-7
> Reported-by: Xiang Mei <xmei5@asu.edu>
> Signed-off-by: Weiming Shi <bestswngs@gmail.com>

SoB tag must match the 'From' header, and must be a real name, likely
'From' should be fixed.

Also you must specify the target tree in the subj prefix ('net' in this
case).

Please have an accurate read of

Documentation/process/maintainer-netdev.rst

before submitting the next revision

> ---
>  drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> index 269c0449760c..2e17a43aec5a 100644
> --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> @@ -213,8 +213,8 @@ static void rmnet_dellink(struct net_device *dev, struct list_head *head)
>  	ep = rmnet_get_endpoint(real_port, mux_id);
>  	if (ep) {
>  		hlist_del_init_rcu(&ep->hlnode);
> -		rmnet_vnd_dellink(mux_id, real_port, ep);
> -		kfree(ep);
> +		real_port->nr_rmnet_devs--;
> +		kfree_rcu_mightsleep(ep);

This is under the rtnl lock and will wait for an rcu grace period, which
is bad for rtnl lock contention. Please add an rcu field to `struct
rmnet_endpoint` and use kfree_rcu() instead.

/P