netdev - Re: [PATCH v2] net/ipv6: Fix soft lockups in fib6_select

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <5efc180e-b719-4945-bd11-2ca392bde4a8@gmail.com>
Date: Tue, 9 Jul 2024 20:10:09 -0600
From: David Ahern <dsahern@...il.com>
To: Omid Ehtemam-Haghighi <omid.ehtemamhaghighi@...losecurity.com>,
 netdev@...r.kernel.org
Cc: adrian.oliver@...losecurity.com, Ido Schimmel <idosch@...dia.com>
Subject: Re: [PATCH v2] net/ipv6: Fix soft lockups in fib6_select_path under
 high next hop churn

On 7/9/24 9:37 AM, Omid Ehtemam-Haghighi wrote:
> diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
> index 83e4f9855ae1..6202575b2c20 100644
> --- a/net/ipv6/ip6_fib.c
> +++ b/net/ipv6/ip6_fib.c
> @@ -540,10 +540,16 @@ static int fib6_dump_node(struct fib6_walker *w)
>  		 * last sibling of this route (no need to dump the
>  		 * sibling routes again)
>  		 */
> -		if (rt->fib6_nsiblings)
> -			rt = list_last_entry(&rt->fib6_siblings,
> -					     struct fib6_info,
> -					     fib6_siblings);
> +		rcu_read_lock();
> +		if (rt->fib6_nsiblings) {
> +			last_sibling = rt;
> +			list_for_each_entry_rcu(sibling, &rt->fib6_siblings,
> +						fib6_siblings)
> +				last_sibling = sibling;
> +
> +			rt = last_sibling;
> +		}
> +		rcu_read_unlock();

as Eric noted fib6_dump_node is already called under rcu_read_lock().

>  	}
>  	w->leaf = NULL;
>  	return 0;


> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index 8d72ca0b086d..4bca06dce176 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -415,7 +415,7 @@ void fib6_select_path(const struct net *net, struct fib6_result *res,
>  		      struct flowi6 *fl6, int oif, bool have_oif_match,
>  		      const struct sk_buff *skb, int strict)
>  {
> -	struct fib6_info *sibling, *next_sibling;
> +	struct fib6_info *sibling;
>  	struct fib6_info *match = res->f6i;
>  
>  	if (!match->nh && (!match->fib6_nsiblings || have_oif_match))
> @@ -442,8 +442,9 @@ void fib6_select_path(const struct net *net, struct fib6_result *res,
>  	if (fl6->mp_hash <= atomic_read(&match->fib6_nh->fib_nh_upper_bound))
>  		goto out;
>  
> -	list_for_each_entry_safe(sibling, next_sibling, &match->fib6_siblings,
> -				 fib6_siblings) {
> +	rcu_read_lock();

put a newline after rcu_read_lock() and before _unlock(). Personally, I
like locking calls to 'pop' in the sense they are separate with
whitespace and not run together with other logic. Comment applies here
and other places since you need to respin.

overall, nothing else caught my eye. Added Ido to put this patch on his
radar as well.

> diff --git a/tools/testing/selftests/net/ipv6_route_update_soft_lockup.sh b/tools/testing/selftests/net/ipv6_route_update_soft_lockup.sh

thanks for the test case.