[PATCH] audit: fix memory leak in nf_tables_commit

37 views
Skip to first unread message

Dongliang Mu

unread,
Jul 13, 2021, 5:42:15 AM7/13/21
to Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, Dongliang Mu, syzbot, netfilt...@vger.kernel.org, core...@netfilter.org, net...@vger.kernel.org, linux-...@vger.kernel.org
In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
free the adp variable.

Fix this by freeing the linked list with head adl.

backtrace:
kmalloc include/linux/slab.h:591 [inline]
kzalloc include/linux/slab.h:721 [inline]
nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
sock_sendmsg_nosec net/socket.c:702 [inline]
sock_sendmsg+0x56/0x80 net/socket.c:722

Reported-by: syzbot <syzk...@googlegroups.com>
Fixes: c520292f29b8 ("audit: log nftables configuration change events once per table")
Signed-off-by: Dongliang Mu <mudongl...@gmail.com>
---
net/netfilter/nf_tables_api.c | 12 ++++++++++++
1 file changed, 12 insertions(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 390d4466567f..7f45b291be13 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -8444,6 +8444,16 @@ static int nf_tables_commit_audit_alloc(struct list_head *adl,
return 0;
}

+static void nf_tables_commit_free(struct list_head *adl)
+{
+ struct nft_audit_data *adp, *adn;
+
+ list_for_each_entry_safe(adp, adn, adl, list) {
+ list_del(&adp->list);
+ kfree(adp);
+ }
+}
+
static void nf_tables_commit_audit_collect(struct list_head *adl,
struct nft_table *table, u32 op)
{
@@ -8508,6 +8518,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_audit_alloc(&adl, trans->ctx.table);
if (ret) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_free(adl);
return ret;
}
if (trans->msg_type == NFT_MSG_NEWRULE ||
@@ -8517,6 +8528,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_chain_prepare(net, chain);
if (ret < 0) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_free(adl);
return ret;
}
}
--
2.25.1

Lukas Bulwahn

unread,
Jul 13, 2021, 7:47:26 AM7/13/21
to Dongliang Mu, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, syzbot, netfilt...@vger.kernel.org, core...@netfilter.org, Netdev, Linux Kernel Mailing List
On Tue, Jul 13, 2021 at 11:42 AM Dongliang Mu <mudongl...@gmail.com> wrote:
>
> In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
> free the adp variable.
>
> Fix this by freeing the linked list with head adl.
>
> backtrace:
> kmalloc include/linux/slab.h:591 [inline]
> kzalloc include/linux/slab.h:721 [inline]
> nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
> nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
> nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
> nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
> nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
> netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
> netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
> netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
> sock_sendmsg_nosec net/socket.c:702 [inline]
> sock_sendmsg+0x56/0x80 net/socket.c:722
>
> Reported-by: syzbot <syzk...@googlegroups.com>

As far as I see, the more default way is to reference to syzbot by:

Reported-by: syzbot+[[20-letter hex reference]]@syzkaller.appspotmail.com

as in for example:

Reported-by: syzbot+fee641...@syzkaller.appspotmail.com

A rough count says that format above is used 1300 times, whereas

Reported-by: syzbot <syzk...@googlegroups.com>

is only used about 330 times.


Lukas
> --
> You received this message because you are subscribed to the Google Groups "syzkaller" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to syzkaller+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/syzkaller/20210713094158.450434-1-mudongliangabcd%40gmail.com.

Dongliang Mu

unread,
Jul 13, 2021, 7:52:10 AM7/13/21
to Lukas Bulwahn, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, syzbot, netfilt...@vger.kernel.org, core...@netfilter.org, Netdev, Linux Kernel Mailing List
On Tue, Jul 13, 2021 at 7:47 PM Lukas Bulwahn <lukas....@gmail.com> wrote:
>
> On Tue, Jul 13, 2021 at 11:42 AM Dongliang Mu <mudongl...@gmail.com> wrote:
> >
> > In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
> > free the adp variable.
> >
> > Fix this by freeing the linked list with head adl.
> >
> > backtrace:
> > kmalloc include/linux/slab.h:591 [inline]
> > kzalloc include/linux/slab.h:721 [inline]
> > nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
> > nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
> > nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
> > nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
> > nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
> > netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
> > netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
> > netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
> > sock_sendmsg_nosec net/socket.c:702 [inline]
> > sock_sendmsg+0x56/0x80 net/socket.c:722
> >
> > Reported-by: syzbot <syzk...@googlegroups.com>
>
> As far as I see, the more default way is to reference to syzbot by:
>
> Reported-by: syzbot+[[20-letter hex reference]]@syzkaller.appspotmail.com
>

Hi Lukas,

this bug is not listed on the syzbot dashboard. I found this bug by
setting up a local syzkaller instance, so I only list syzbot other
than concrete syzbot id.

best regards,
Dongliang Mu

Lukas Bulwahn

unread,
Jul 13, 2021, 8:05:10 AM7/13/21
to Dongliang Mu, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, syzbot, netfilt...@vger.kernel.org, core...@netfilter.org, Netdev, Linux Kernel Mailing List
On Tue, Jul 13, 2021 at 1:52 PM Dongliang Mu <mudongl...@gmail.com> wrote:
>
> On Tue, Jul 13, 2021 at 7:47 PM Lukas Bulwahn <lukas....@gmail.com> wrote:
> >
> > On Tue, Jul 13, 2021 at 11:42 AM Dongliang Mu <mudongl...@gmail.com> wrote:
> > >

> > >
> > > Reported-by: syzbot <syzk...@googlegroups.com>
> >
> > As far as I see, the more default way is to reference to syzbot by:
> >
> > Reported-by: syzbot+[[20-letter hex reference]]@syzkaller.appspotmail.com
> >
>
> Hi Lukas,
>
> this bug is not listed on the syzbot dashboard. I found this bug by
> setting up a local syzkaller instance, so I only list syzbot other
> than concrete syzbot id.
>

I see. Thanks for your explanation.

Lukas

kernel test robot

unread,
Jul 13, 2021, 8:32:10 AM7/13/21
to Dongliang Mu, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, kbuil...@lists.01.org, net...@vger.kernel.org, Dongliang Mu, syzbot
Hi Dongliang,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on nf/master]
[also build test ERROR on nf-next/master ipvs/master v5.14-rc1 next-20210713]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url: https://github.com/0day-ci/linux/commits/Dongliang-Mu/audit-fix-memory-leak-in-nf_tables_commit/20210713-174434
base: https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git master
config: alpha-randconfig-r002-20210713 (attached as .config)
compiler: alpha-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/0day-ci/linux/commit/2112ee88ee1fa56b43d8d4ba2554d8d94199bd37
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Dongliang-Mu/audit-fix-memory-leak-in-nf_tables_commit/20210713-174434
git checkout 2112ee88ee1fa56b43d8d4ba2554d8d94199bd37
# save the attached .config to linux build tree
mkdir build_dir
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross O=build_dir ARCH=alpha SHELL=/bin/bash net/netfilter/

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <l...@intel.com>

All errors (new ones prefixed by >>):

net/netfilter/nf_tables_api.c: In function 'nf_tables_commit':
>> net/netfilter/nf_tables_api.c:8522:26: error: incompatible type for argument 1 of 'nf_tables_commit_free'
8522 | nf_tables_commit_free(adl);
| ^~~
| |
| struct list_head
net/netfilter/nf_tables_api.c:8448:53: note: expected 'struct list_head *' but argument is of type 'struct list_head'
8448 | static void nf_tables_commit_free(struct list_head *adl)
| ~~~~~~~~~~~~~~~~~~^~~
net/netfilter/nf_tables_api.c:8532:27: error: incompatible type for argument 1 of 'nf_tables_commit_free'
8532 | nf_tables_commit_free(adl);
| ^~~
| |
| struct list_head
net/netfilter/nf_tables_api.c:8448:53: note: expected 'struct list_head *' but argument is of type 'struct list_head'
8448 | static void nf_tables_commit_free(struct list_head *adl)
| ~~~~~~~~~~~~~~~~~~^~~


vim +/nf_tables_commit_free +8522 net/netfilter/nf_tables_api.c

8491
8492 static int nf_tables_commit(struct net *net, struct sk_buff *skb)
8493 {
8494 struct nftables_pernet *nft_net = nft_pernet(net);
8495 struct nft_trans *trans, *next;
8496 struct nft_trans_elem *te;
8497 struct nft_chain *chain;
8498 struct nft_table *table;
8499 LIST_HEAD(adl);
8500 int err;
8501
8502 if (list_empty(&nft_net->commit_list)) {
8503 mutex_unlock(&nft_net->commit_mutex);
8504 return 0;
8505 }
8506
8507 /* 0. Validate ruleset, otherwise roll back for error reporting. */
8508 if (nf_tables_validate(net) < 0)
8509 return -EAGAIN;
8510
8511 err = nft_flow_rule_offload_commit(net);
8512 if (err < 0)
8513 return err;
8514
8515 /* 1. Allocate space for next generation rules_gen_X[] */
8516 list_for_each_entry_safe(trans, next, &nft_net->commit_list, list) {
8517 int ret;
8518
8519 ret = nf_tables_commit_audit_alloc(&adl, trans->ctx.table);
8520 if (ret) {
8521 nf_tables_commit_chain_prepare_cancel(net);
> 8522 nf_tables_commit_free(adl);
8523 return ret;
8524 }
8525 if (trans->msg_type == NFT_MSG_NEWRULE ||
8526 trans->msg_type == NFT_MSG_DELRULE) {
8527 chain = trans->ctx.chain;
8528
8529 ret = nf_tables_commit_chain_prepare(net, chain);
8530 if (ret < 0) {
8531 nf_tables_commit_chain_prepare_cancel(net);
8532 nf_tables_commit_free(adl);
8533 return ret;
8534 }
8535 }
8536 }
8537
8538 /* step 2. Make rules_gen_X visible to packet path */
8539 list_for_each_entry(table, &nft_net->tables, list) {
8540 list_for_each_entry(chain, &table->chains, list)
8541 nf_tables_commit_chain(net, chain);
8542 }
8543
8544 /*
8545 * Bump generation counter, invalidate any dump in progress.
8546 * Cannot fail after this point.
8547 */
8548 while (++nft_net->base_seq == 0)
8549 ;
8550
8551 /* step 3. Start new generation, rules_gen_X now in use. */
8552 net->nft.gencursor = nft_gencursor_next(net);
8553
8554 list_for_each_entry_safe(trans, next, &nft_net->commit_list, list) {
8555 nf_tables_commit_audit_collect(&adl, trans->ctx.table,
8556 trans->msg_type);
8557 switch (trans->msg_type) {
8558 case NFT_MSG_NEWTABLE:
8559 if (nft_trans_table_update(trans)) {
8560 if (!(trans->ctx.table->flags & __NFT_TABLE_F_UPDATE)) {
8561 nft_trans_destroy(trans);
8562 break;
8563 }
8564 if (trans->ctx.table->flags & NFT_TABLE_F_DORMANT)
8565 nf_tables_table_disable(net, trans->ctx.table);
8566
8567 trans->ctx.table->flags &= ~__NFT_TABLE_F_UPDATE;
8568 } else {
8569 nft_clear(net, trans->ctx.table);
8570 }
8571 nf_tables_table_notify(&trans->ctx, NFT_MSG_NEWTABLE);
8572 nft_trans_destroy(trans);
8573 break;
8574 case NFT_MSG_DELTABLE:
8575 list_del_rcu(&trans->ctx.table->list);
8576 nf_tables_table_notify(&trans->ctx, NFT_MSG_DELTABLE);
8577 break;
8578 case NFT_MSG_NEWCHAIN:
8579 if (nft_trans_chain_update(trans)) {
8580 nft_chain_commit_update(trans);
8581 nf_tables_chain_notify(&trans->ctx, NFT_MSG_NEWCHAIN);
8582 /* trans destroyed after rcu grace period */
8583 } else {
8584 nft_chain_commit_drop_policy(trans);
8585 nft_clear(net, trans->ctx.chain);
8586 nf_tables_chain_notify(&trans->ctx, NFT_MSG_NEWCHAIN);
8587 nft_trans_destroy(trans);
8588 }
8589 break;
8590 case NFT_MSG_DELCHAIN:
8591 nft_chain_del(trans->ctx.chain);
8592 nf_tables_chain_notify(&trans->ctx, NFT_MSG_DELCHAIN);
8593 nf_tables_unregister_hook(trans->ctx.net,
8594 trans->ctx.table,
8595 trans->ctx.chain);
8596 break;
8597 case NFT_MSG_NEWRULE:
8598 nft_clear(trans->ctx.net, nft_trans_rule(trans));
8599 nf_tables_rule_notify(&trans->ctx,
8600 nft_trans_rule(trans),
8601 NFT_MSG_NEWRULE);
8602 nft_trans_destroy(trans);
8603 break;
8604 case NFT_MSG_DELRULE:
8605 list_del_rcu(&nft_trans_rule(trans)->list);
8606 nf_tables_rule_notify(&trans->ctx,
8607 nft_trans_rule(trans),
8608 NFT_MSG_DELRULE);
8609 nft_rule_expr_deactivate(&trans->ctx,
8610 nft_trans_rule(trans),
8611 NFT_TRANS_COMMIT);
8612 break;
8613 case NFT_MSG_NEWSET:
8614 nft_clear(net, nft_trans_set(trans));
8615 /* This avoids hitting -EBUSY when deleting the table
8616 * from the transaction.
8617 */
8618 if (nft_set_is_anonymous(nft_trans_set(trans)) &&
8619 !list_empty(&nft_trans_set(trans)->bindings))
8620 trans->ctx.table->use--;
8621
8622 nf_tables_set_notify(&trans->ctx, nft_trans_set(trans),
8623 NFT_MSG_NEWSET, GFP_KERNEL);
8624 nft_trans_destroy(trans);
8625 break;
8626 case NFT_MSG_DELSET:
8627 list_del_rcu(&nft_trans_set(trans)->list);
8628 nf_tables_set_notify(&trans->ctx, nft_trans_set(trans),
8629 NFT_MSG_DELSET, GFP_KERNEL);
8630 break;
8631 case NFT_MSG_NEWSETELEM:
8632 te = (struct nft_trans_elem *)trans->data;
8633
8634 nft_setelem_activate(net, te->set, &te->elem);
8635 nf_tables_setelem_notify(&trans->ctx, te->set,
8636 &te->elem,
8637 NFT_MSG_NEWSETELEM, 0);
8638 nft_trans_destroy(trans);
8639 break;
8640 case NFT_MSG_DELSETELEM:
8641 te = (struct nft_trans_elem *)trans->data;
8642
8643 nf_tables_setelem_notify(&trans->ctx, te->set,
8644 &te->elem,
8645 NFT_MSG_DELSETELEM, 0);
8646 nft_setelem_remove(net, te->set, &te->elem);
8647 if (!nft_setelem_is_catchall(te->set, &te->elem)) {
8648 atomic_dec(&te->set->nelems);
8649 te->set->ndeact--;
8650 }
8651 break;
8652 case NFT_MSG_NEWOBJ:
8653 if (nft_trans_obj_update(trans)) {
8654 nft_obj_commit_update(trans);
8655 nf_tables_obj_notify(&trans->ctx,
8656 nft_trans_obj(trans),
8657 NFT_MSG_NEWOBJ);
8658 } else {
8659 nft_clear(net, nft_trans_obj(trans));
8660 nf_tables_obj_notify(&trans->ctx,
8661 nft_trans_obj(trans),
8662 NFT_MSG_NEWOBJ);
8663 nft_trans_destroy(trans);
8664 }
8665 break;
8666 case NFT_MSG_DELOBJ:
8667 nft_obj_del(nft_trans_obj(trans));
8668 nf_tables_obj_notify(&trans->ctx, nft_trans_obj(trans),
8669 NFT_MSG_DELOBJ);
8670 break;
8671 case NFT_MSG_NEWFLOWTABLE:
8672 if (nft_trans_flowtable_update(trans)) {
8673 nft_trans_flowtable(trans)->data.flags =
8674 nft_trans_flowtable_flags(trans);
8675 nf_tables_flowtable_notify(&trans->ctx,
8676 nft_trans_flowtable(trans),
8677 &nft_trans_flowtable_hooks(trans),
8678 NFT_MSG_NEWFLOWTABLE);
8679 list_splice(&nft_trans_flowtable_hooks(trans),
8680 &nft_trans_flowtable(trans)->hook_list);
8681 } else {
8682 nft_clear(net, nft_trans_flowtable(trans));
8683 nf_tables_flowtable_notify(&trans->ctx,
8684 nft_trans_flowtable(trans),
8685 &nft_trans_flowtable(trans)->hook_list,
8686 NFT_MSG_NEWFLOWTABLE);
8687 }
8688 nft_trans_destroy(trans);
8689 break;
8690 case NFT_MSG_DELFLOWTABLE:
8691 if (nft_trans_flowtable_update(trans)) {
8692 nft_flowtable_hooks_del(nft_trans_flowtable(trans),
8693 &nft_trans_flowtable_hooks(trans));
8694 nf_tables_flowtable_notify(&trans->ctx,
8695 nft_trans_flowtable(trans),
8696 &nft_trans_flowtable_hooks(trans),
8697 NFT_MSG_DELFLOWTABLE);
8698 nft_unregister_flowtable_net_hooks(net,
8699 &nft_trans_flowtable_hooks(trans));
8700 } else {
8701 list_del_rcu(&nft_trans_flowtable(trans)->list);
8702 nf_tables_flowtable_notify(&trans->ctx,
8703 nft_trans_flowtable(trans),
8704 &nft_trans_flowtable(trans)->hook_list,
8705 NFT_MSG_DELFLOWTABLE);
8706 nft_unregister_flowtable_net_hooks(net,
8707 &nft_trans_flowtable(trans)->hook_list);
8708 }
8709 break;
8710 }
8711 }
8712
8713 nft_commit_notify(net, NETLINK_CB(skb).portid);
8714 nf_tables_gen_notify(net, skb, NFT_MSG_NEWGEN);
8715 nf_tables_commit_audit_log(&adl, nft_net->base_seq);
8716 nf_tables_commit_release(net);
8717
8718 return 0;
8719 }
8720

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuil...@lists.01.org
.config.gz

Dongliang Mu

unread,
Jul 13, 2021, 9:04:03 AM7/13/21
to Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, Dongliang Mu, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, net...@vger.kernel.org, linux-...@vger.kernel.org
In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
free the adp variable.

Fix this by freeing the linked list with head adl.

backtrace:
kmalloc include/linux/slab.h:591 [inline]
kzalloc include/linux/slab.h:721 [inline]
nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
sock_sendmsg_nosec net/socket.c:702 [inline]
sock_sendmsg+0x56/0x80 net/socket.c:722

Reported-by: syzbot <syzk...@googlegroups.com>
Reported-by: kernel test robot <l...@intel.com>
Fixes: c520292f29b8 ("audit: log nftables configuration change events once per table")
Signed-off-by: Dongliang Mu <mudongl...@gmail.com>
---
v1->v2: fix the compile issue
net/netfilter/nf_tables_api.c | 12 ++++++++++++
1 file changed, 12 insertions(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 390d4466567f..7f45b291be13 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -8444,6 +8444,16 @@ static int nf_tables_commit_audit_alloc(struct list_head *adl,
return 0;
}

+static void nf_tables_commit_free(struct list_head *adl)
+{
+ struct nft_audit_data *adp, *adn;
+
+ list_for_each_entry_safe(adp, adn, adl, list) {
+ list_del(&adp->list);
+ kfree(adp);
+ }
+}
+
static void nf_tables_commit_audit_collect(struct list_head *adl,
struct nft_table *table, u32 op)
{
@@ -8508,6 +8518,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_audit_alloc(&adl, trans->ctx.table);
if (ret) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_free(&adl);
return ret;
}
if (trans->msg_type == NFT_MSG_NEWRULE ||
@@ -8517,6 +8528,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_chain_prepare(net, chain);
if (ret < 0) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_free(&adl);
return ret;
}
}
--
2.25.1

Florian Westphal

unread,
Jul 13, 2021, 9:21:11 AM7/13/21
to Dongliang Mu, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, net...@vger.kernel.org, linux-...@vger.kernel.org
nf_tables_commit_audit_free?

Aside from that, there should be a followup patch (for nf-next),
adding empty inline functions in case of CONFIG_AUDITSYSCALL=n.

Right now it does pointless aggregation for the AUDIT=n case.

Dongliang Mu

unread,
Jul 13, 2021, 10:40:54 AM7/13/21
to Florian Westphal, Pablo Neira Ayuso, Jozsef Kadlecsik, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, open list:NETWORKING [GENERAL], linux-kernel
What do you mean? Modify the name of newly added function to
nf_tables_commit_audit_free?

>
> Aside from that, there should be a followup patch (for nf-next),
> adding empty inline functions in case of CONFIG_AUDITSYSCALL=n.

I see. I prefer to send them (two implementations of the newly added
function) in version v2.

Florian Westphal

unread,
Jul 13, 2021, 11:31:25 AM7/13/21
to Dongliang Mu, Florian Westphal, Pablo Neira Ayuso, Jozsef Kadlecsik, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, open list:NETWORKING [GENERAL], linux-kernel
Dongliang Mu <mudongl...@gmail.com> wrote:
> > > +static void nf_tables_commit_free(struct list_head *adl)
> >
> > nf_tables_commit_audit_free?
>
> What do you mean? Modify the name of newly added function to
> nf_tables_commit_audit_free?

Yes, this function is audit related, and it does the inverse
of existing '...audit_alloc'.

kernel test robot

unread,
Jul 13, 2021, 3:35:23 PM7/13/21
to Dongliang Mu, Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Paul Moore, Richard Guy Briggs, clang-bu...@googlegroups.com, kbuil...@lists.01.org, net...@vger.kernel.org, Dongliang Mu, syzbot
Hi Dongliang,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on nf/master]
[also build test ERROR on nf-next/master ipvs/master v5.14-rc1 next-20210713]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url: https://github.com/0day-ci/linux/commits/Dongliang-Mu/audit-fix-memory-leak-in-nf_tables_commit/20210713-174434
base: https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git master
config: x86_64-randconfig-a005-20210713 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project 8d69635ed9ecf36fd0ca85906bfde17949671cbe)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# install x86_64 cross compiling tool for clang build
# apt-get install binutils-x86-64-linux-gnu
# https://github.com/0day-ci/linux/commit/2112ee88ee1fa56b43d8d4ba2554d8d94199bd37
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Dongliang-Mu/audit-fix-memory-leak-in-nf_tables_commit/20210713-174434
git checkout 2112ee88ee1fa56b43d8d4ba2554d8d94199bd37
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <l...@intel.com>

All errors (new ones prefixed by >>):

>> net/netfilter/nf_tables_api.c:8522:26: error: passing 'struct list_head' to parameter of incompatible type 'struct list_head *'; take the address with &
nf_tables_commit_free(adl);
^~~
&
net/netfilter/nf_tables_api.c:8448:53: note: passing argument to parameter 'adl' here
static void nf_tables_commit_free(struct list_head *adl)
^
net/netfilter/nf_tables_api.c:8532:27: error: passing 'struct list_head' to parameter of incompatible type 'struct list_head *'; take the address with &
nf_tables_commit_free(adl);
^~~
&
net/netfilter/nf_tables_api.c:8448:53: note: passing argument to parameter 'adl' here
static void nf_tables_commit_free(struct list_head *adl)
^
2 errors generated.


vim +8522 net/netfilter/nf_tables_api.c
.config.gz

Dongliang Mu

unread,
Jul 13, 2021, 10:19:04 PM7/13/21
to Florian Westphal, Pablo Neira Ayuso, Jozsef Kadlecsik, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, open list:NETWORKING [GENERAL], linux-kernel
Hi Florian,

I double-check the patch, and it seems there is no need to send a
followup patch (nf-next).

I will send a v3 patch with the new function name.

Best regards
Dongliang Mu

Dongliang Mu

unread,
Jul 13, 2021, 11:27:30 PM7/13/21
to Pablo Neira Ayuso, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, Dongliang Mu, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, net...@vger.kernel.org, linux-...@vger.kernel.org
In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
free the adp variable.

Fix this by adding nf_tables_commit_audit_free which frees
the linked list with the head node adl.

backtrace:
kmalloc include/linux/slab.h:591 [inline]
kzalloc include/linux/slab.h:721 [inline]
nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
sock_sendmsg_nosec net/socket.c:702 [inline]
sock_sendmsg+0x56/0x80 net/socket.c:722

Reported-by: syzbot <syzk...@googlegroups.com>
Reported-by: kernel test robot <l...@intel.com>
Fixes: c520292f29b8 ("audit: log nftables configuration change events once per table")
Signed-off-by: Dongliang Mu <mudongl...@gmail.com>
---
v1->v2: fix the compile issue
v2->v3: modify the added function to a new name
net/netfilter/nf_tables_api.c | 12 ++++++++++++
1 file changed, 12 insertions(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 390d4466567f..7f45b291be13 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -8444,6 +8444,16 @@ static int nf_tables_commit_audit_alloc(struct list_head *adl,
return 0;
}

+static void nf_tables_commit_audit_free(struct list_head *adl)
+{
+ struct nft_audit_data *adp, *adn;
+
+ list_for_each_entry_safe(adp, adn, adl, list) {
+ list_del(&adp->list);
+ kfree(adp);
+ }
+}
+
static void nf_tables_commit_audit_collect(struct list_head *adl,
struct nft_table *table, u32 op)
{
@@ -8508,6 +8518,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_audit_alloc(&adl, trans->ctx.table);
if (ret) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_audit_free(&adl);
return ret;
}
if (trans->msg_type == NFT_MSG_NEWRULE ||
@@ -8517,6 +8528,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
ret = nf_tables_commit_chain_prepare(net, chain);
if (ret < 0) {
nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_audit_free(&adl);
return ret;
}
}
--
2.25.1

Pablo Neira Ayuso

unread,
Jul 16, 2021, 8:22:47 PM7/16/21
to Dongliang Mu, Jozsef Kadlecsik, Florian Westphal, David S. Miller, Jakub Kicinski, Richard Guy Briggs, Paul Moore, syzbot, kernel test robot, netfilt...@vger.kernel.org, core...@netfilter.org, net...@vger.kernel.org, linux-...@vger.kernel.org
On Wed, Jul 14, 2021 at 11:27:03AM +0800, Dongliang Mu wrote:
> In nf_tables_commit, if nf_tables_commit_audit_alloc fails, it does not
> free the adp variable.
>
> Fix this by adding nf_tables_commit_audit_free which frees
> the linked list with the head node adl.
>
> backtrace:
> kmalloc include/linux/slab.h:591 [inline]
> kzalloc include/linux/slab.h:721 [inline]
> nf_tables_commit_audit_alloc net/netfilter/nf_tables_api.c:8439 [inline]
> nf_tables_commit+0x16e/0x1760 net/netfilter/nf_tables_api.c:8508
> nfnetlink_rcv_batch+0x512/0xa80 net/netfilter/nfnetlink.c:562
> nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
> nfnetlink_rcv+0x1fa/0x220 net/netfilter/nfnetlink.c:652
> netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
> netlink_unicast+0x2c7/0x3e0 net/netlink/af_netlink.c:1340
> netlink_sendmsg+0x36b/0x6b0 net/netlink/af_netlink.c:1929
> sock_sendmsg_nosec net/socket.c:702 [inline]
> sock_sendmsg+0x56/0x80 net/socket.c:722

Applied, thanks.
Reply all
Reply to author
Forward
0 new messages