Re: [PATCH] net: memory leak in nr_rx_frame

[Willem de Bruijn <willemdebruijn.kernel@gmail.com>]

Bragatheswaran Manickavel wrote:
> The condition (make = nr_make_new(sk)) == NULL suggests
> that nr_make_new allocates memory and returns a pointer.
> If this allocation fails (returns NULL), it indicates a
> potential memory leak.
> 
> Added sock_put() for make which can potentially solve
> this issue
> 
> Reported-by: syzbot+0145ea560de205bc09f0@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=0145ea560de205bc09f0
> Signed-off-by: Bragatheswaran Manickavel <bragathemanick0908@gmail.com>
> ---
>  net/netrom/af_netrom.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/net/netrom/af_netrom.c b/net/netrom/af_netrom.c
> index 0eed00184adf..7d7cda4ae300 100644
> --- a/net/netrom/af_netrom.c
> +++ b/net/netrom/af_netrom.c
> @@ -970,6 +970,8 @@ int nr_rx_frame(struct sk_buff *skb, struct net_device *dev)
>  		nr_transmit_refusal(skb, 0);
>  		if (sk)
>  			sock_put(sk);
> +		if (make)
> +			sock_put(make);
>  		return 0;
>  	}

If the branch is entered because (make = nr_make_new(sk)) == NULL then
make is.. NULL. If the branch is entered for one if the earlier two
conditions failing, then make is undefined.

The syzbot report is that make is allocated here and eventually
nothing points to this still allocated memory. It does not report
where the last reference was lost. Not inside this branch.