|
On Thu, 2019-10-03 at 22:28 +0300, Julian Anastasov wrote:
> Hello,
hello Julian,
thanks a lot for reviewing this code.
> On Wed, 2 Oct 2019, Davide Caratti wrote:
>
> > if the IPVS module is removed while the sync daemon is starting, there is
> > a small possibility for stop_sync_thread() to end successfully just after
> > start_sync_thread() released sync_mutex, but before increasing the module
> > refcount. When this race happens, the following warning is seen in dmesg:
> >
> > IPVS: stopping master sync thread 14134 ...
> > ------------[ cut here ]------------
> > WARNING: CPU: 0 PID: 14133 at kernel/module.c:1137
> > module_put.part.44+0x167/0x2a0
[...]
> May be this is not early enough. When control hooks
> are the first thing to disconnect with ip_vs_unregister_nl_ioctl()
> how start_sync_thread() is called concurrently?
Initially I believed that __ip_vs_dev_cleanup() was called when
start_sync_thread() was already running (but it already released the RTNL
lock). But then, after your comments:
> ip_vs_sockopts is
> protected with .owner field and ip_vs_genl_family with .module field
> or more correctly with cb_lock rwsem in net/netlink/genetlink.c
> which avoids doit() handlers after genl_unregister_family() is called.
> So, both interfaces do not allow start_sync_thread() during module
> removal and it is interesting how your fix works actually.
I think that probably my patch just changes the timing of events in my
test setup (see script below), making the problem more difficult to
observe.
> Which interface calls start_sync_thread (sockopt/genl)? Note that
> do_ip_vs_set_ctl() already has ip_vs_use_count_inc() before the
> start_sync_thread() call. Also, the WARN in module_put warns
> about refcnt underflow, so there is an extra ip_vs_use_count_dec() ?
it's genl in my case, see call traces below.
> Another possibility is the missing checks for
> ip_vs_use_count_inc() result to cause underflow on ip_vs_use_count_dec().
thanks for the advice, the unbalance is probably the cause of the splat
I'm seeing. On a clean/unpatched kernel, I modified ip_vs_use_count() to
do this:
WARN_ON(!try_module_get(THIS_MODULE))
using the reproducer below, I still see the splat in module_put(), but
it's preceded by a failure in ip_vs_use_count_inc(), which accidentally is
the same one I moved in my patch).
IPVS: Registered protocols (TCP, UDP, SCTP, AH, ESP)
IPVS: Connection hash table configured (size=4096, memory=64Kbytes)
IPVS: ipvs loaded.
IPVS: sync thread started: state = MASTER, mcast_ifn = veth_syncd, syncid = 0,
id = 0
IPVS: stopping master sync thread 4005 ...
IPVS: sync thread started: state = MASTER, mcast_ifn = veth_syncd, syncid = 0,
id = 0
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4014 at net/netfilter/ipvs/ip_vs_ctl.c:235
ip_vs_use_count_inc+0x19/0x20 [ip_vs]
Modules linked in: ip_vs(-) nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 veth
ip6table_filter ip6_tables iptable_filter binfmt_misc intel_rapl_msr
intel_rapl_common crct10dif_pclmul crc32_pclmul ext4 mbcache jbd2
ghash_clmulni_intel snd_hda_codec_generic ledtrig_audio snd_hda_intel
snd_intel_nhlt snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device
snd_pcm aesni_intel crypto_simd cryptd glue_helper joydev pcspkr snd_timer
virtio_balloon snd soundcore i2c_piix4 nfsd auth_rpcgss nfs_acl lockd grace
sunrpc ip_tables xfs libcrc32c ata_generic pata_acpi virtio_net net_failover
virtio_blk failover virtio_console qxl drm_kms_helper syscopyarea sysfillrect
sysimgblt fb_sys_fops ata_piix ttm crc32c_intel serio_raw drm virtio_pci libata
virtio_ring virtio floppy dm_mirror dm_region_hash dm_log dm_mod [last
unloaded: nf_defrag_ipv6]
CPU: 0 PID: 4014 Comm: ipvsadm Not tainted 5.4.0-rc1.upstream+ #741
Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
RIP: 0010:ip_vs_use_count_inc+0x19/0x20 [ip_vs]
Code: e8 bc 5e dc f4 4c 8b 44 24 08 e9 db fd ff ff 66 90 0f 1f 44 00 00 48 c7
c7 00 02 42 c1 e8 5f 3c 9e f4 84 c0 74 04 0f b6 c0 c3 <0f> 0b 0f b6 c0 c3 90 0f
1f 44 00 00 55 48 89 e5 41 57 49 89 f7 41
RSP: 0018:ffff88804ba5f2c8 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffff88804f533ca8 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 1ffff1100974be3f RDI: ffffffffc1420200
RBP: ffff88804ba5f500 R08: fffffbfff71129d1 R09: fffffbfff71129d1
R10: 0000000000000001 R11: fffffbfff71129d0 R12: ffff88804f533300
R13: 0000000000000000 R14: 0000000000000001 R15: ffff888079442d00
FS: 00007fd42ff9c740(0000) GS:ffff888053800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000fc91d8 CR3: 000000004fe2c002 CR4: 00000000001606f0
Call Trace:
start_sync_thread+0x1c17/0x23a0 [ip_vs]
ip_vs_genl_new_daemon+0x3b0/0x550 [ip_vs]
ip_vs_genl_set_daemon+0x295/0x360 [ip_vs]
genl_rcv_msg+0x588/0x11e0
netlink_rcv_skb+0x121/0x350
genl_rcv+0x24/0x40
netlink_unicast+0x416/0x5d0
netlink_sendmsg+0x67f/0xb50
sock_sendmsg+0xe2/0x110
___sys_sendmsg+0x640/0x940
__sys_sendmsg+0xd2/0x170
do_syscall_64+0xa5/0x4e0
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7fd42f48f930
Code: c3 48 8b 05 42 75 2c 00 f7 db 64 89 18 48 83 cb ff eb dd 0f 1f 80 00 00
00 00 83 3d 6d d6 2c 00 00 75 10 b8 2e 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 31
c3 48 83 ec 08 e8 ae cc 00 00 48 89 04 24
RSP: 002b:00007ffcf69ecaa8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 0000000000fc2100 RCX: 00007fd42f48f930
RDX: 0000000000000000 RSI: 00007ffcf69ecb50 RDI: 0000000000000003
RBP: 0000000000fc2010 R08: 0000000000000001 R09: 0000000000fc2150
R10: 00007fd42f7577b8 R11: 0000000000000246 R12: 0000000000fc3160
R13: 00007ffcf69ecb50 R14: 0000000000000000 R15: 0000000000000000
irq event stamp: 3288
hardirqs last enabled at (3287): [<ffffffffb5bb276f>]
__local_bh_enable_ip+0xef/0x1b0
hardirqs last disabled at (3288): [<ffffffffb5a0556a>]
trace_hardirqs_off_thunk+0x1a/0x20
softirqs last enabled at (3286): [<ffffffffc13f3365>]
start_sync_thread+0x1c05/0x23a0 [ip_vs]
softirqs last disabled at (3284): [<ffffffffc13f32e1>]
start_sync_thread+0x1b81/0x23a0 [ip_vs]
---[ end trace 7b4e216440e510e6 ]---
IPVS: stopping master sync thread 4016 ...
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4013 at kernel/module.c:1146
module_put.part.44+0x15b/0x290
Modules linked in: ip_vs(-) nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 veth
ip6table_filter ip6_tables iptable_filter binfmt_misc intel_rapl_msr
intel_rapl_common crct10dif_pclmul crc32_pclmul ext4 mbcache jbd2
ghash_clmulni_intel snd_hda_codec_generic ledtrig_audio snd_hda_intel
snd_intel_nhlt snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device
snd_pcm aesni_intel crypto_simd cryptd glue_helper joydev pcspkr snd_timer
virtio_balloon snd soundcore i2c_piix4 nfsd auth_rpcgss nfs_acl lockd grace
sunrpc ip_tables xfs libcrc32c ata_generic pata_acpi virtio_net net_failover
virtio_blk failover virtio_console qxl drm_kms_helper syscopyarea sysfillrect
sysimgblt fb_sys_fops ata_piix ttm crc32c_intel serio_raw drm virtio_pci libata
virtio_ring virtio floppy dm_mirror dm_region_hash dm_log dm_mod [last
unloaded: nf_defrag_ipv6]
CPU: 0 PID: 4013 Comm: modprobe Tainted: G W
5.4.0-rc1.upstream+ #741
Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
RIP: 0010:module_put.part.44+0x15b/0x290
Code: 04 25 28 00 00 00 0f 85 18 01 00 00 48 83 c4 68 5b 5d 41 5c 41 5d 41 5e
41 5f c3 89 44 24 28 83 e8 01 89 c5 0f 89 57 ff ff ff <0f> 0b e9 78 ff ff ff 65
8b 1d 67 83 26 4a 89 db be 08 00 00 00 48
RSP: 0018:ffff888050607c78 EFLAGS: 00010297
RAX: 0000000000000003 RBX: ffffffffc1420590 RCX: ffffffffb5db0ef9
RDX: 0000000000000000 RSI: 0000000000000004 RDI: ffffffffc1420590
RBP: 00000000ffffffff R08: fffffbfff82840b3 R09: fffffbfff82840b3
R10: 0000000000000001 R11: fffffbfff82840b2 R12: 1ffff1100a0c0f90
R13: ffffffffc1420200 R14: ffff88804f533300 R15: ffff88804f533ca0
FS: 00007f8ea9720740(0000) GS:ffff888053800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f3245abe000 CR3: 000000004c28a006 CR4: 00000000001606f0
Call Trace:
stop_sync_thread+0x3a3/0x7c0 [ip_vs]
ip_vs_sync_net_cleanup+0x13/0x50 [ip_vs]
ops_exit_list.isra.5+0x94/0x140
unregister_pernet_operations+0x29d/0x460
unregister_pernet_device+0x26/0x60
ip_vs_cleanup+0x11/0x38 [ip_vs]
__x64_sys_delete_module+0x2d5/0x400
do_syscall_64+0xa5/0x4e0
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x7f8ea8bf0db7
Code: 73 01 c3 48 8b 0d b9 80 2c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f
84 00 00 00 00 00 0f 1f 44 00 00 b8 b0 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01
c3 48 8b 0d 89 80 2c 00 f7 d8 64 89 01 48
RSP: 002b:00007ffcd38d2fe8 EFLAGS: 00000206 ORIG_RAX: 00000000000000b0
RAX: ffffffffffffffda RBX: 0000000002436240 RCX: 00007f8ea8bf0db7
RDX: 0000000000000000 RSI: 0000000000000800 RDI: 00000000024362a8
RBP: 0000000000000000 R08: 00007f8ea8eba060 R09: 00007f8ea8c658a0
R10: 00007ffcd38d2a60 R11: 0000000000000206 R12: 0000000000000000
R13: 0000000000000001 R14: 00000000024362a8 R15: 0000000000000000
irq event stamp: 4538
hardirqs last enabled at (4537): [<ffffffffb6193dde>]
quarantine_put+0x9e/0x170
hardirqs last disabled at (4538): [<ffffffffb5a0556a>]
trace_hardirqs_off_thunk+0x1a/0x20
softirqs last enabled at (4522): [<ffffffffb6f8ebe9>]
sk_common_release+0x169/0x2d0
softirqs last disabled at (4520): [<ffffffffb6f8eb3e>]
sk_common_release+0xbe/0x2d0
---[ end trace 7b4e216440e510e7 ]---
IPVS: ipvs unloaded.
so, this is try_module_get() in ip_vs_start_sync_thread() failing to refcount
ip_vs.
When the module is unloaded and stop_sync_thread() is called, the refcounts are
unbalanced and then we see the splat when the last module_put() is called.
> Are there any steps to reproduce?
yes, a script that does the following:
while true; do
if modprobe -r ip_vs; then
sleep 0.5
fi
done >/dev/null 2>&1 &
while true; do
ipvsadm --start-daemon master --mcast-interface veth_syncd
done >/dev/null 2>&1 &
while true; do
ipvsadm --stop-daemon master
done >/dev/null 2>&1
I'm about to send a v2 patch where I test the return value of each call to
ip_vs_use_count_inc(), and jump to the error path of caller functions
(returning -EAGAIN) on failure of try_module_get(). For
ip_vs_start_sync_thread(), the resulting code is very similar to what I
did in the first post, with the only addition of an additional 'goto
out_unlock;' in case the attempt to refcount the module failed.
thank you in advance for any further feedback!
--
davide
|
