memory leak in nf_tables_parse_netdev_hooks

14 views
Skip to first unread message

syzbot

unread,
Jan 16, 2020, 12:45:10 AM1/16/20
to core...@netfilter.org, da...@davemloft.net, f...@strlen.de, kad...@netfilter.org, linux-...@vger.kernel.org, net...@vger.kernel.org, netfilt...@vger.kernel.org, pa...@netfilter.org, syzkall...@googlegroups.com
Hello,

syzbot found the following crash on:

HEAD commit: 51d69817 Merge tag 'platform-drivers-x86-v5.5-3' of git://..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=10e32659e00000
kernel config: https://syzkaller.appspot.com/x/.config?x=a0eee3ce463efd
dashboard link: https://syzkaller.appspot.com/bug?extid=f9d4095107fc8749c69c
compiler: gcc (GCC) 9.0.0 20181231 (experimental)
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=16fae421e00000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=10e41c76e00000

IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+f9d409...@syzkaller.appspotmail.com

executing program
BUG: memory leak
unreferenced object 0xffff8881292d4580 (size 96):
comm "syz-executor052", pid 7129, jiffies 4294942632 (age 13.530s)
hex dump (first 32 bytes):
40 d2 15 1c 81 88 ff ff 40 d2 15 1c 81 88 ff ff @.......@.......
60 53 c3 82 ff ff ff ff 00 a0 2c 2a 81 88 ff ff `S........,*....
backtrace:
[<0000000059912bf5>] kmemleak_alloc_recursive
include/linux/kmemleak.h:43 [inline]
[<0000000059912bf5>] slab_post_alloc_hook mm/slab.h:586 [inline]
[<0000000059912bf5>] slab_alloc mm/slab.c:3320 [inline]
[<0000000059912bf5>] kmem_cache_alloc_trace+0x145/0x2c0 mm/slab.c:3549
[<00000000ab9f7d85>] kmalloc include/linux/slab.h:556 [inline]
[<00000000ab9f7d85>] nft_netdev_hook_alloc+0x3f/0xd0
net/netfilter/nf_tables_api.c:1624
[<0000000074e6bb65>] nf_tables_parse_netdev_hooks+0xac/0x230
net/netfilter/nf_tables_api.c:1673
[<00000000cd387efd>] nf_tables_flowtable_parse_hook
net/netfilter/nf_tables_api.c:5936 [inline]
[<00000000cd387efd>] nf_tables_newflowtable+0x41e/0x930
net/netfilter/nf_tables_api.c:6137
[<00000000526e3994>] nfnetlink_rcv_batch+0x662/0x8c0
net/netfilter/nfnetlink.c:433
[<000000006c5402bf>] nfnetlink_rcv_skb_batch
net/netfilter/nfnetlink.c:543 [inline]
[<000000006c5402bf>] nfnetlink_rcv+0x189/0x1c0
net/netfilter/nfnetlink.c:561
[<00000000b752f9f7>] netlink_unicast_kernel
net/netlink/af_netlink.c:1302 [inline]
[<00000000b752f9f7>] netlink_unicast+0x223/0x310
net/netlink/af_netlink.c:1328
[<00000000a0e23793>] netlink_sendmsg+0x2c0/0x570
net/netlink/af_netlink.c:1917
[<000000006e0aea94>] sock_sendmsg_nosec net/socket.c:639 [inline]
[<000000006e0aea94>] sock_sendmsg+0x54/0x70 net/socket.c:659
[<00000000a60c26dd>] ____sys_sendmsg+0x2d0/0x300 net/socket.c:2330
[<00000000d412f616>] ___sys_sendmsg+0x8a/0xd0 net/socket.c:2384
[<000000001179a6b0>] __sys_sendmsg+0x80/0xf0 net/socket.c:2417
[<000000004a6699d1>] __do_sys_sendmsg net/socket.c:2426 [inline]
[<000000004a6699d1>] __se_sys_sendmsg net/socket.c:2424 [inline]
[<000000004a6699d1>] __x64_sys_sendmsg+0x23/0x30 net/socket.c:2424
[<00000000a05aa7d1>] do_syscall_64+0x73/0x220
arch/x86/entry/common.c:294
[<00000000bad17b52>] entry_SYSCALL_64_after_hwframe+0x44/0xa9



---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches

Dan Carpenter

unread,
Jan 16, 2020, 5:10:39 AM1/16/20
to Pablo Neira Ayuso, core...@netfilter.org, linux-...@vger.kernel.org, net...@vger.kernel.org, netfilt...@vger.kernel.org, syzkall...@googlegroups.com, Jozsef Kadlecsik, Florian Westphal, David S. Miller, syzbot
Syzbot detected a leak in nf_tables_parse_netdev_hooks(). If the hook
already exists, then the error handling doesn't free the newest "hook".

Reported-by: syzbot+f9d409...@syzkaller.appspotmail.com
Fixes: b75a3e8371bc ("netfilter: nf_tables: allow netdevice to be used only once per flowtable")
Signed-off-by: Dan Carpenter <dan.ca...@oracle.com>
---
net/netfilter/nf_tables_api.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 273f3838318b..7728e9fd5de4 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1676,6 +1676,7 @@ static int nf_tables_parse_netdev_hooks(struct net *net,
goto err_hook;
}
if (nft_hook_list_find(hook_list, hook)) {
+ kfree(hook);
err = -EEXIST;
goto err_hook;
}
--
2.11.0

Florian Westphal

unread,
Jan 16, 2020, 5:14:25 AM1/16/20
to Dan Carpenter, Pablo Neira Ayuso, core...@netfilter.org, linux-...@vger.kernel.org, net...@vger.kernel.org, netfilt...@vger.kernel.org, syzkall...@googlegroups.com, Jozsef Kadlecsik, Florian Westphal, David S. Miller, syzbot
Dan Carpenter <dan.ca...@oracle.com> wrote:
> Syzbot detected a leak in nf_tables_parse_netdev_hooks(). If the hook
> already exists, then the error handling doesn't free the newest "hook".

Thanks.

Reviewed-by: Florian Westphal <f...@strlen.de>

Pablo Neira Ayuso

unread,
Jan 16, 2020, 8:59:35 AM1/16/20
to Dan Carpenter, core...@netfilter.org, linux-...@vger.kernel.org, net...@vger.kernel.org, netfilt...@vger.kernel.org, syzkall...@googlegroups.com, Jozsef Kadlecsik, Florian Westphal, David S. Miller, syzbot
On Thu, Jan 16, 2020 at 01:09:31PM +0300, Dan Carpenter wrote:
> Syzbot detected a leak in nf_tables_parse_netdev_hooks(). If the hook
> already exists, then the error handling doesn't free the newest "hook".

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