Jiri Pirko
55b40dbf0e
net: fix ifindex collision during namespace removal
...
Commit aca51397d014 ("netns: Fix arbitrary net_device-s corruptions
on net_ns stop.") introduced a possibility to hit a BUG in case device
is returning back to init_net and two following conditions are met:
1) dev->ifindex value is used in a name of another "dev%d"
device in init_net.
2) dev->name is used by another device in init_net.
Under real life circumstances this is hard to get. Therefore this has
been present happily for over 10 years. To reproduce:
$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 86:89:3f:86:61:29 brd ff:ff:ff:ff:ff:ff
3: enp0s2: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff
$ ip netns add ns1
$ ip -n ns1 link add dummy1ns1 type dummy
$ ip -n ns1 link add dummy2ns1 type dummy
$ ip link set enp0s2 netns ns1
$ ip -n ns1 link set enp0s2 name dummy0
[ 100.858894] virtio_net virtio0 dummy0: renamed from enp0s2
$ ip link add dev4 type dummy
$ ip -n ns1 a
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: dummy1ns1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 16:63:4c:38:3e:ff brd ff:ff:ff:ff:ff:ff
3: dummy2ns1: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether aa:9e:86:dd:6b:5d brd ff:ff:ff:ff:ff:ff
4: dummy0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 52:54:00:12:34:56 brd ff:ff:ff:ff:ff:ff
$ ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 86:89:3f:86:61:29 brd ff:ff:ff:ff:ff:ff
4: dev4: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 5a:e1:4a:b6:ec:f8 brd ff:ff:ff:ff:ff:ff
$ ip netns del ns1
[ 158.717795] default_device_exit: failed to move dummy0 to init_net: -17
[ 158.719316] ------------[ cut here ]------------
[ 158.720591] kernel BUG at net/core/dev.c:9824!
[ 158.722260] invalid opcode: 0000 [#1 ] SMP KASAN PTI
[ 158.723728] CPU: 0 PID: 56 Comm: kworker/u2:1 Not tainted 5.3.0-rc1+ #18
[ 158.725422] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-2.fc30 04/01/2014
[ 158.727508] Workqueue: netns cleanup_net
[ 158.728915] RIP: 0010:default_device_exit.cold+0x1d/0x1f
[ 158.730683] Code: 84 e8 18 c9 3e fe 0f 0b e9 70 90 ff ff e8 36 e4 52 fe 89 d9 4c 89 e2 48 c7 c6 80 d6 25 84 48 c7 c7 20 c0 25 84 e8 f4 c8 3e
[ 158.736854] RSP: 0018:ffff8880347e7b90 EFLAGS: 00010282
[ 158.738752] RAX: 000000000000003b RBX: 00000000ffffffef RCX: 0000000000000000
[ 158.741369] RDX: 0000000000000000 RSI: ffffffff8128013d RDI: ffffed10068fcf64
[ 158.743418] RBP: ffff888033550170 R08: 000000000000003b R09: fffffbfff0b94b9c
[ 158.745626] R10: fffffbfff0b94b9b R11: ffffffff85ca5cdf R12: ffff888032f28000
[ 158.748405] R13: dffffc0000000000 R14: ffff8880335501b8 R15: 1ffff110068fcf72
[ 158.750638] FS: 0000000000000000(0000) GS:ffff888036000000(0000) knlGS:0000000000000000
[ 158.752944] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 158.755245] CR2: 00007fe8b45d21d0 CR3: 00000000340b4005 CR4: 0000000000360ef0
[ 158.757654] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 158.760012] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 158.762758] Call Trace:
[ 158.763882] ? dev_change_net_namespace+0xbb0/0xbb0
[ 158.766148] ? devlink_nl_cmd_set_doit+0x520/0x520
[ 158.768034] ? dev_change_net_namespace+0xbb0/0xbb0
[ 158.769870] ops_exit_list.isra.0+0xa8/0x150
[ 158.771544] cleanup_net+0x446/0x8f0
[ 158.772945] ? unregister_pernet_operations+0x4a0/0x4a0
[ 158.775294] process_one_work+0xa1a/0x1740
[ 158.776896] ? pwq_dec_nr_in_flight+0x310/0x310
[ 158.779143] ? do_raw_spin_lock+0x11b/0x280
[ 158.780848] worker_thread+0x9e/0x1060
[ 158.782500] ? process_one_work+0x1740/0x1740
[ 158.784454] kthread+0x31b/0x420
[ 158.786082] ? __kthread_create_on_node+0x3f0/0x3f0
[ 158.788286] ret_from_fork+0x3a/0x50
[ 158.789871] ---[ end trace defd6c657c71f936 ]---
[ 158.792273] RIP: 0010:default_device_exit.cold+0x1d/0x1f
[ 158.795478] Code: 84 e8 18 c9 3e fe 0f 0b e9 70 90 ff ff e8 36 e4 52 fe 89 d9 4c 89 e2 48 c7 c6 80 d6 25 84 48 c7 c7 20 c0 25 84 e8 f4 c8 3e
[ 158.804854] RSP: 0018:ffff8880347e7b90 EFLAGS: 00010282
[ 158.807865] RAX: 000000000000003b RBX: 00000000ffffffef RCX: 0000000000000000
[ 158.811794] RDX: 0000000000000000 RSI: ffffffff8128013d RDI: ffffed10068fcf64
[ 158.816652] RBP: ffff888033550170 R08: 000000000000003b R09: fffffbfff0b94b9c
[ 158.820930] R10: fffffbfff0b94b9b R11: ffffffff85ca5cdf R12: ffff888032f28000
[ 158.825113] R13: dffffc0000000000 R14: ffff8880335501b8 R15: 1ffff110068fcf72
[ 158.829899] FS: 0000000000000000(0000) GS:ffff888036000000(0000) knlGS:0000000000000000
[ 158.834923] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 158.838164] CR2: 00007fe8b45d21d0 CR3: 00000000340b4005 CR4: 0000000000360ef0
[ 158.841917] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 158.845149] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Fix this by checking if a device with the same name exists in init_net
and fallback to original code - dev%d to allocate name - in case it does.
This was found using syzkaller.
Fixes: aca51397d014 ("netns: Fix arbitrary net_device-s corruptions on net_ns stop.")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2019-07-29 11:07:04 -07:00
..
2019-07-06 12:50:01 +02:00
2019-07-12 17:31:19 -07:00
2019-06-19 17:09:55 +02:00
2019-06-22 08:59:24 -04:00
2019-06-05 17:37:10 +02:00
2019-05-30 11:26:32 -07:00
2019-06-16 14:22:37 -07:00
2019-07-08 19:48:57 -07:00
2019-07-11 10:55:49 -07:00
2019-05-30 11:29:53 -07:00
2019-07-12 16:03:16 -07:00
2019-07-29 09:50:05 -07:00
2019-05-30 11:29:22 -07:00
2019-07-24 11:19:03 +02:00
2019-07-18 11:05:25 -07:00
2019-07-29 11:07:04 -07:00
2019-05-30 11:29:52 -07:00
2019-07-18 17:08:07 -07:00
2019-05-24 17:36:42 +02:00
2019-07-10 18:43:43 -07:00
2019-07-19 21:27:45 -07:00
2019-06-07 11:00:14 -07:00
2019-07-11 14:37:45 -07:00
2019-06-19 11:37:47 -04:00
2019-05-21 10:50:46 +02:00
2019-07-25 17:35:03 -07:00
2019-07-26 14:17:44 -07:00
2019-07-29 10:26:14 -07:00
2019-06-19 17:09:55 +02:00
2019-07-08 19:48:57 -07:00
2019-07-08 19:38:03 -07:00
2019-06-23 13:24:17 -07:00
2019-06-17 20:20:36 -07:00
2019-06-05 17:36:36 +02:00
2019-07-20 21:40:32 +02:00
2019-05-30 11:26:41 -07:00
2019-07-18 17:08:07 -07:00
2019-05-30 11:26:32 -07:00
2019-07-19 21:27:45 -07:00
2019-05-21 11:28:45 +02:00
2019-06-14 19:50:33 -07:00
2019-07-24 15:49:05 -07:00
2019-07-08 12:46:24 -07:00
2019-06-19 17:09:55 +02:00
2019-07-24 15:45:11 -07:00
2019-06-27 21:06:39 -07:00
2019-06-05 17:37:07 +02:00
2019-06-19 17:09:55 +02:00
2019-06-05 17:36:37 +02:00
2019-07-27 13:58:12 -07:00
2019-06-19 17:09:55 +02:00
2019-05-30 11:26:32 -07:00
2019-07-18 17:08:07 -07:00
2019-07-29 09:46:58 -07:00
2019-07-19 10:06:06 -07:00
2019-06-27 21:06:39 -07:00
2019-06-22 08:59:24 -04:00
2019-07-19 10:42:02 -07:00
2019-05-30 11:26:32 -07:00
2019-07-21 11:41:01 -07:00
2019-07-22 16:04:17 +02:00
2019-06-07 11:00:14 -07:00
2019-07-08 15:35:17 -07:00
2019-06-05 17:30:29 +02:00
2019-07-15 20:44:49 -07:00
2019-05-24 17:27:12 +02:00
2019-07-12 15:02:21 +02:00
2019-07-08 19:48:57 -07:00
2019-05-31 15:30:03 -06:00
2019-06-17 13:56:26 -07:00
2019-07-19 10:42:02 -07:00
2019-05-21 10:50:45 +02:00