[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190704195504.GA20705@splinter>
Date: Thu, 4 Jul 2019 22:55:04 +0300
From: Ido Schimmel <idosch@...sch.org>
To: David Miller <davem@...emloft.net>
Cc: netdev@...r.kernel.org, dsahern@...il.com, jiri@...lanox.com,
shalomt@...lanox.com, mlxsw@...lanox.com, idosch@...lanox.com
Subject: Re: [PATCH net] ipv4: Fix NULL pointer dereference in
ipv4_neigh_lookup()
On Thu, Jul 04, 2019 at 12:24:49PM -0700, David Miller wrote:
> From: Ido Schimmel <idosch@...sch.org>
> Date: Thu, 4 Jul 2019 19:26:38 +0300
>
> > Both ip_neigh_gw4() and ip_neigh_gw6() can return either a valid pointer
> > or an error pointer, but the code currently checks that the pointer is
> > not NULL.
> ...
> > @@ -447,7 +447,7 @@ static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst,
> > n = ip_neigh_gw4(dev, pkey);
> > }
> >
> > - if (n && !refcount_inc_not_zero(&n->refcnt))
> > + if (!IS_ERR(n) && !refcount_inc_not_zero(&n->refcnt))
> > n = NULL;
> >
> > rcu_read_unlock_bh();
>
> Don't the callers expect only non-error pointers?
It is actually OK to return an error pointer here. In fact, before the
commit I cited the function returned the return value of neigh_create().
If you think it's clearer, we can do this instead:
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 8ea0735a6754..40697fcd2889 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -447,6 +447,9 @@ static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst,
n = ip_neigh_gw4(dev, pkey);
}
+ if (IS_ERR(n))
+ n = NULL;
+
if (n && !refcount_inc_not_zero(&n->refcnt))
n = NULL;
Powered by blists - more mailing lists