From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH] net: sk_prot_alloc() should not blindly overwrite memory Date: Thu, 09 Jul 2009 07:36:05 +0200 Message-ID: <4A5581C5.5070409@gmail.com> References: <4A537469.3040207@gmail.com> <4A53CD39.7080407@gmail.com> <20090707.191424.167842005.davem@davemloft.net> <4A5441A0.3050504@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: emil.s.tantilov@intel.com, emils.tantilov@gmail.com, netdev@vger.kernel.org, jesse.brandeburg@intel.com, jeffrey.t.kirsher@intel.com, jolsa@redhat.com, Patrick McHardy , "Paul E. McKenney" To: David Miller Return-path: Received: from gw1.cosmosbay.com ([212.99.114.194]:51190 "EHLO gw1.cosmosbay.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750967AbZGIFgh (ORCPT ); Thu, 9 Jul 2009 01:36:37 -0400 In-Reply-To: <4A5441A0.3050504@gmail.com> Sender: netdev-owner@vger.kernel.org List-ID: Eric Dumazet a =E9crit : > David Miller a =E9crit : >> From: Eric Dumazet >> Date: Wed, 08 Jul 2009 00:33:29 +0200 >> >>> [PATCH] net: sk_prot_alloc() should not blindly overwrite memory >>> >>> Some sockets use SLAB_DESTROY_BY_RCU, and our RCU code rely that so= me >>> fields should not be blindly overwritten, even with null. >>> >>> These fields are sk->sk_refcnt and sk->sk_nulls_node.next >>> >>> Current sk_prot_alloc() implementation doesnt respect this hypothes= is, >>> calling kmem_cache_alloc() with __GFP_ZERO and setting sk_refcnt to= 1 >>> instead of atomically increment it. >>> >>> Reported-by: Emil S Tantilov >>> Signed-off-by: Eric Dumazet >> I've applied this but will wait for some more testing before >> I push it out for real to kernel.org >=20 > Thanks David >=20 > I forgot to CC Paul and Patrick, so I'll ask them to look at this pat= ch. >=20 > Patrick, a similar fix is needed in conntrack as well, we currently > uses "ct =3D kmem_cache_zalloc(nf_conntrack_cachep, gfp);" and thus > overwrite struct hlist_nulls_node hnnode; contained > in "struct nf_conntrack_tuple_hash", while lockless readers still > potentialy need them. Setting hnnode.next to NULL is dangerous > since last bit is not set (not a nulls value), a reader could > try to dereference this NULL pointer and trap. >=20 >=20 > Here is the patch again so that Paul & Patrick can comment on it. >=20 > I am not sure about the refcnt thing (blindly setting it to 0 again > should be OK in fact, since no reader should/can to the=20 > atomic_inc_if_not_zero on it), but the nulls.next thing is problemati= c. Here is an updated and much simpler patch, taking care of sk_node.next = being not set to 0 This patch applies to >=3D 2.6.29 kernels [PATCH] net: sk_prot_alloc() should not blindly overwrite memory Some sockets use SLAB_DESTROY_BY_RCU, and our RCU code correctness depends on sk->sk_nulls_node.next being always valid. A NULL value is not allowed as it might fault a lockless reader. Current sk_prot_alloc() implementation doesnt respect this hypothesis, calling kmem_cache_alloc() with __GFP_ZERO. Just call memset() around the forbidden field. Signed-off-by: Eric Dumazet --- diff --git a/net/core/sock.c b/net/core/sock.c index b0ba569..7b87ec0 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -939,8 +939,23 @@ static struct sock *sk_prot_alloc(struct proto *pr= ot, gfp_t priority, struct kmem_cache *slab; =20 slab =3D prot->slab; - if (slab !=3D NULL) - sk =3D kmem_cache_alloc(slab, priority); + if (slab !=3D NULL) { + sk =3D kmem_cache_alloc(slab, priority & ~__GFP_ZERO); + if (!sk) + return sk; + if (priority & __GFP_ZERO) { + /* + * caches using SLAB_DESTROY_BY_RCU should let + * sk_node.next un-modified. Special care is taken + * when initializing object to zero. + */ + if (offsetof(struct sock, sk_node.next) !=3D 0) + memset(sk, 0, offsetof(struct sock, sk_node.next)); + memset(&sk->sk_node.pprev, 0, + prot->obj_size - offsetof(struct sock, + sk_node.pprev)); + } + } else sk =3D kmalloc(prot->obj_size, priority); =20