[net] ipv6: fix calling in6_ifa_hold incorrectly for dad work

Submitted by Xin Long on June 15, 2017, 8:33 a.m.

Details

Message ID 2d710ef288455604efab24f72f4476ec5edb13bb.1497515638.git.lucien.xin@gmail.com
State Accepted
Delegated to: David Miller
Headers show

Commit Message

Xin Long June 15, 2017, 8:33 a.m.
Now when starting the dad work in addrconf_mod_dad_work, if the dad work
is idle and queued, it needs to hold ifa.

The problem is there's one gap in [1], during which if the pending dad work
is removed elsewhere. It will miss to hold ifa, but the dad word is still
idea and queue.

        if (!delayed_work_pending(&ifp->dad_work))
                in6_ifa_hold(ifp);
                    <--------------[1]
        mod_delayed_work(addrconf_wq, &ifp->dad_work, delay);

An use-after-free issue can be caused by this.

Chen Wei found this issue when WARN_ON(!hlist_unhashed(&ifp->addr_lst)) in
net6_ifa_finish_destroy was hit because of it.

As Hannes' suggestion, this patch is to fix it by holding ifa first in
addrconf_mod_dad_work, then calling mod_delayed_work and putting ifa if
the dad_work is already in queue.

Note that this patch did not choose to fix it with:

  if (!mod_delayed_work(delay))
          in6_ifa_hold(ifp);

As with it, when delay == 0, dad_work would be scheduled immediately, all
addrconf_mod_dad_work(0) callings had to be moved under ifp->lock.

Reported-by: Wei Chen <weichen@redhat.com>
Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
---
 net/ipv6/addrconf.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Comments

David Miller June 15, 2017, 2:58 p.m.
From: Xin Long <lucien.xin@gmail.com>
Date: Thu, 15 Jun 2017 16:33:58 +0800

> Now when starting the dad work in addrconf_mod_dad_work, if the dad work
> is idle and queued, it needs to hold ifa.
> 
> The problem is there's one gap in [1], during which if the pending dad work
> is removed elsewhere. It will miss to hold ifa, but the dad word is still
> idea and queue.
> 
>         if (!delayed_work_pending(&ifp->dad_work))
>                 in6_ifa_hold(ifp);
>                     <--------------[1]
>         mod_delayed_work(addrconf_wq, &ifp->dad_work, delay);
> 
> An use-after-free issue can be caused by this.
> 
> Chen Wei found this issue when WARN_ON(!hlist_unhashed(&ifp->addr_lst)) in
> net6_ifa_finish_destroy was hit because of it.
> 
> As Hannes' suggestion, this patch is to fix it by holding ifa first in
> addrconf_mod_dad_work, then calling mod_delayed_work and putting ifa if
> the dad_work is already in queue.
> 
> Note that this patch did not choose to fix it with:
> 
>   if (!mod_delayed_work(delay))
>           in6_ifa_hold(ifp);
> 
> As with it, when delay == 0, dad_work would be scheduled immediately, all
> addrconf_mod_dad_work(0) callings had to be moved under ifp->lock.
> 
> Reported-by: Wei Chen <weichen@redhat.com>
> Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
> Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
> Signed-off-by: Xin Long <lucien.xin@gmail.com>

Applied and queued up for -stable, thank you.

Patch hide | download patch | download mbox

diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 6a4fb1e..686c923 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -332,9 +332,9 @@  static void addrconf_mod_rs_timer(struct inet6_dev *idev,
 static void addrconf_mod_dad_work(struct inet6_ifaddr *ifp,
 				   unsigned long delay)
 {
-	if (!delayed_work_pending(&ifp->dad_work))
-		in6_ifa_hold(ifp);
-	mod_delayed_work(addrconf_wq, &ifp->dad_work, delay);
+	in6_ifa_hold(ifp);
+	if (mod_delayed_work(addrconf_wq, &ifp->dad_work, delay))
+		in6_ifa_put(ifp);
 }
 
 static int snmp6_alloc_dev(struct inet6_dev *idev)