[PATCH net] sctp: change to hold/put transport for proto_unreach_timer

[PATCH net] sctp: change to hold/put transport for proto_unreach_timer

[Xin Long]

A call trace was found in Hangbin's Codenomicon testing with debug kernel:

  [ 2615.981988] ODEBUG: free active (active state 0) object type: timer_list hint: sctp_generate_proto_unreach_event+0x0/0x3a0 [sctp]
  [ 2615.995050] WARNING: CPU: 17 PID: 0 at lib/debugobjects.c:328 debug_print_object+0x199/0x2b0
  [ 2616.095934] RIP: 0010:debug_print_object+0x199/0x2b0
  [ 2616.191533] Call Trace:
  [ 2616.194265]  <IRQ>
  [ 2616.202068]  debug_check_no_obj_freed+0x25e/0x3f0
  [ 2616.207336]  slab_free_freelist_hook+0xeb/0x140
  [ 2616.220971]  kfree+0xd6/0x2c0
  [ 2616.224293]  rcu_do_batch+0x3bd/0xc70
  [ 2616.243096]  rcu_core+0x8b9/0xd00
  [ 2616.256065]  __do_softirq+0x23d/0xacd
  [ 2616.260166]  irq_exit+0x236/0x2a0
  [ 2616.263879]  smp_apic_timer_interrupt+0x18d/0x620
  [ 2616.269138]  apic_timer_interrupt+0xf/0x20
  [ 2616.273711]  </IRQ>

This is because it holds asoc when transport->proto_unreach_timer starts
and puts asoc when the timer stops, and without holding transport the
transport could be freed when the timer is still running.

So fix it by holding/putting transport instead for proto_unreach_timer
in transport, just like other timers in transport.

Fixes: 50b5d6ad6382 ("sctp: Fix a race between ICMP protocol unreachable and connect()")
Reported-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
---
 net/sctp/input.c         | 4 ++--
 net/sctp/sm_sideeffect.c | 2 +-
 net/sctp/transport.c     | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/sctp/input.c b/net/sctp/input.c
index 55d4fc6..d508f6f 100644
--- a/net/sctp/input.c
+++ b/net/sctp/input.c
@@ -449,7 +449,7 @@ void sctp_icmp_proto_unreachable(struct sock *sk,
 		else {
 			if (!mod_timer(&t->proto_unreach_timer,
 						jiffies + (HZ/20)))
-				sctp_association_hold(asoc);
+				sctp_transport_hold(t);
 		}
 	} else {
 		struct net *net = sock_net(sk);
@@ -458,7 +458,7 @@ void sctp_icmp_proto_unreachable(struct sock *sk,
 			 "encountered!\n", __func__);
 
 		if (del_timer(&t->proto_unreach_timer))
-			sctp_association_put(asoc);
+			sctp_transport_put(t);
 
 		sctp_do_sm(net, SCTP_EVENT_T_OTHER,
 			   SCTP_ST_OTHER(SCTP_EVENT_ICMP_PROTO_UNREACH),
diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
index 813d307..0a51150 100644
--- a/net/sctp/sm_sideeffect.c
+++ b/net/sctp/sm_sideeffect.c
@@ -419,7 +419,7 @@ void sctp_generate_proto_unreach_event(struct timer_list *t)
 		/* Try again later.  */
 		if (!mod_timer(&transport->proto_unreach_timer,
 				jiffies + (HZ/20)))
-			sctp_association_hold(asoc);
+			sctp_transport_hold(transport);
 		goto out_unlock;
 	}
 
diff --git a/net/sctp/transport.c b/net/sctp/transport.c
index 806af58..60fcf31 100644
--- a/net/sctp/transport.c
+++ b/net/sctp/transport.c
@@ -133,7 +133,7 @@ void sctp_transport_free(struct sctp_transport *transport)
 
 	/* Delete the ICMP proto unreachable timer if it's active. */
 	if (del_timer(&transport->proto_unreach_timer))
-		sctp_association_put(transport->asoc);
+		sctp_transport_put(transport);
 
 	sctp_transport_put(transport);
 }
-- 
2.1.0

Re: [PATCH net] sctp: change to hold/put transport for proto_unreach_timer

[Marcelo Ricardo Leitner]

Hi,

On Fri, Nov 13, 2020 at 05:18:24PM +0800, Xin Long wrote:
...
> diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
> index 813d307..0a51150 100644
> --- a/net/sctp/sm_sideeffect.c
> +++ b/net/sctp/sm_sideeffect.c
> @@ -419,7 +419,7 @@ void sctp_generate_proto_unreach_event(struct timer_list *t)
>  		/* Try again later.  */
>  		if (!mod_timer(&transport->proto_unreach_timer,
>  				jiffies + (HZ/20)))
> -			sctp_association_hold(asoc);
> +			sctp_transport_hold(transport);
>  		goto out_unlock;
>  	}
>  

The chunk above covers the socket busy case, but for the normal cases
it also needs:

@@ -435,7 +435,7 @@ void sctp_generate_proto_unreach_event(struct timer_list *t)

 out_unlock:
        bh_unlock_sock(sk);
-       sctp_association_put(asoc);
+       sctp_transport_put(asoc);
 }

  /* Handle the timeout of the RE-CONFIG timer. */

> diff --git a/net/sctp/transport.c b/net/sctp/transport.c
> index 806af58..60fcf31 100644
> --- a/net/sctp/transport.c
> +++ b/net/sctp/transport.c
> @@ -133,7 +133,7 @@ void sctp_transport_free(struct sctp_transport *transport)
>  
>  	/* Delete the ICMP proto unreachable timer if it's active. */
>  	if (del_timer(&transport->proto_unreach_timer))
> -		sctp_association_put(transport->asoc);
> +		sctp_transport_put(transport);
>  
>  	sctp_transport_put(transport);

Btw, quite noticeable on the above list of timers that only this timer
was using a reference on the asoc. Seems we're good now, then. :-)

  Marcelo

Re: [PATCH net] sctp: change to hold/put transport for proto_unreach_timer

[Xin Long]

On Fri, Nov 13, 2020 at 8:35 PM Marcelo Ricardo Leitner
<marcelo.leitner@gmail.com> wrote:
>
> Hi,
>
> On Fri, Nov 13, 2020 at 05:18:24PM +0800, Xin Long wrote:
> ...
> > diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
> > index 813d307..0a51150 100644
> > --- a/net/sctp/sm_sideeffect.c
> > +++ b/net/sctp/sm_sideeffect.c
> > @@ -419,7 +419,7 @@ void sctp_generate_proto_unreach_event(struct timer_list *t)
> >               /* Try again later.  */
> >               if (!mod_timer(&transport->proto_unreach_timer,
> >                               jiffies + (HZ/20)))
> > -                     sctp_association_hold(asoc);
> > +                     sctp_transport_hold(transport);
> >               goto out_unlock;
> >       }
> >
>
> The chunk above covers the socket busy case, but for the normal cases
> it also needs:
>
> @@ -435,7 +435,7 @@ void sctp_generate_proto_unreach_event(struct timer_list *t)
>
>  out_unlock:
>         bh_unlock_sock(sk);
> -       sctp_association_put(asoc);
> +       sctp_transport_put(asoc);
>  }
yeah, right, posted v2.

Thanks.
>
>   /* Handle the timeout of the RE-CONFIG timer. */
>
> > diff --git a/net/sctp/transport.c b/net/sctp/transport.c
> > index 806af58..60fcf31 100644
> > --- a/net/sctp/transport.c
> > +++ b/net/sctp/transport.c
> > @@ -133,7 +133,7 @@ void sctp_transport_free(struct sctp_transport *transport)
> >
> >       /* Delete the ICMP proto unreachable timer if it's active. */
> >       if (del_timer(&transport->proto_unreach_timer))
> > -             sctp_association_put(transport->asoc);
> > +             sctp_transport_put(transport);
> >
> >       sctp_transport_put(transport);
>
> Btw, quite noticeable on the above list of timers that only this timer
> was using a reference on the asoc. Seems we're good now, then. :-)
>
correct! :D