Re: [PATCH net-next] net: ipv4: guard ip_mr_output() with rcu

[Petr Machata <petrm@nvidia.com>]

Eric Dumazet <edumazet@google.com> writes:

> syzbot found at least one path leads to an ip_mr_output()
> without RCU being held.
>
> Add guard(rcu)() to fix this in a concise way.
>
> WARNING: CPU: 0 PID: 0 at net/ipv4/ipmr.c:2302 ip_mr_output+0xbb1/0xe70 net/ipv4/ipmr.c:2302
> Call Trace:
>  <IRQ>
>   igmp_send_report+0x89e/0xdb0 net/ipv4/igmp.c:799
>  igmp_timer_expire+0x204/0x510 net/ipv4/igmp.c:-1
>   call_timer_fn+0x17e/0x5f0 kernel/time/timer.c:1747
>   expire_timers kernel/time/timer.c:1798 [inline]
>   __run_timers kernel/time/timer.c:2372 [inline]
>   __run_timer_base+0x61a/0x860 kernel/time/timer.c:2384
>   run_timer_base kernel/time/timer.c:2393 [inline]
>   run_timer_softirq+0xb7/0x180 kernel/time/timer.c:2403
>   handle_softirqs+0x286/0x870 kernel/softirq.c:579
>   __do_softirq kernel/softirq.c:613 [inline]
>   invoke_softirq kernel/softirq.c:453 [inline]
>   __irq_exit_rcu+0xca/0x1f0 kernel/softirq.c:680
>   irq_exit_rcu+0x9/0x30 kernel/softirq.c:696
>   instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1050 [inline]
>   sysvec_apic_timer_interrupt+0xa6/0xc0 arch/x86/kernel/apic/apic.c:1050
>
> Fixes: 35bec72a24ac ("net: ipv4: Add ip_mr_output()")
> Reported-by: syzbot+f02fb9e43bd85c6c66ae@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/netdev/685e841a.a00a0220.129264.0002.GAE@google.com/T/#u
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Petr Machata <petrm@nvidia.com>
> Cc: Roopa Prabhu <roopa@nvidia.com>
> Cc: Nikolay Aleksandrov <razor@blackwall.org>
> Cc: Benjamin Poirier <bpoirier@nvidia.com>
> Cc: Ido Schimmel <idosch@nvidia.com>

Hmm, reading the cleanup.h doco it calls out mixing goto and guards, but
I think it should work in this case, the guard is function scope, and
the gotos are back to this function scope.

Thanks for the fix!

Reviewed-by: Petr Machata <petrm@nvidia.com>

> ---
>  net/ipv4/ipmr.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/net/ipv4/ipmr.c b/net/ipv4/ipmr.c
> index f78c4e53dc8c161e334781970bbff6069c084ebb..3a2044e6033d5683bda678489f6eaf72ea0b8890 100644
> --- a/net/ipv4/ipmr.c
> +++ b/net/ipv4/ipmr.c
> @@ -2299,7 +2299,8 @@ int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb)
>  	struct mr_table *mrt;
>  	int vif;
>  
> -	WARN_ON_ONCE(!rcu_read_lock_held());
> +	guard(rcu)();
> +
>  	dev = rt->dst.dev;
>  
>  	if (IPCB(skb)->flags & IPSKB_FORWARDED)
> @@ -2313,7 +2314,6 @@ int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb)
>  	if (IS_ERR(mrt))
>  		goto mc_output;
>  
> -	/* already under rcu_read_lock() */
>  	cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr);
>  	if (!cache) {
>  		vif = ipmr_find_vif(mrt, dev);