]> git.baikalelectronics.ru Git - kernel.git/commit
ipv6: fix calling in6_ifa_hold incorrectly for dad work
authorXin Long <lucien.xin@gmail.com>
Thu, 15 Jun 2017 08:33:58 +0000 (16:33 +0800)
committerDavid S. Miller <davem@davemloft.net>
Thu, 15 Jun 2017 15:26:16 +0000 (11:26 -0400)
commitfa6cc18fc2f36085757d83fc3c058b329c763bb8
tree12a20e086cb52fed4037eaec3c40a0718d59a4ac
parent42e6d130669aa8bbe7179bf9a2973506a2891526
ipv6: fix calling in6_ifa_hold incorrectly for dad work

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>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv6/addrconf.c