Re: [PATCH net-next v1] icmp: fix ip_rt_bug race in icmp_route_lookup reverse path

[Paolo Abeni]

On 1/28/26 10:05 AM, Jiayuan Chen wrote:
> diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
> index 19c9c838967f..dc9dcc799824 100644
> --- a/net/ipv4/icmp.c
> +++ b/net/ipv4/icmp.c
> @@ -559,6 +559,23 @@ static struct rtable *icmp_route_lookup(struct net *net, struct flowi4 *fl4,
>  		/* steal dst entry from skb_in, don't drop refcnt */
>  		skb_dstref_steal(skb_in);
>  		skb_dstref_restore(skb_in, orefdst);
> +
> +		/*
> +		 * At this point, fl4_dec.daddr should NOT be local (we
> +		 * checked fl4_dec.saddr above). However, a race condition
> +		 * may occur if the address is added to the interface
> +		 * concurrently. In that case, ip_route_input() returns a
> +		 * LOCAL route with dst.output=ip_rt_bug, which must not
> +		 * be used for output.
> +		 */
> +		if (!err && rt2 && rt2->rt_type == RTN_LOCAL) {
> +			net_warn_ratelimited("%s: detected local route for %pI4 "
> +					     "during ICMP error handling (src %pI4), "
> +					     "possible address race\n",
> +					     __func__, &fl4_dec.daddr, &fl4_dec.saddr);

The fix looks correct to me, but this patch should target the 'net' tree
and the above warning message is a bit off: the text string should not
be broken to fit the 80 chars limit - it need to be greepable - it's
probably better to not include the function name.

/P