[ovs-dev] [PATCH V2] datapath: Prevent panic

Fri Apr 20 16:36:37 UTC 2018

On 4/20/2018 9:03 AM, Gregory Rose wrote:
> On 4/20/2018 5:39 AM, Eric Garver wrote:
>> On Thu, Apr 19, 2018 at 08:07:33PM -0700, Gregory Rose wrote:
>> Fantastic, I'll test this and whip up a patch.
>>> Thanks!
>>>
>>> - Greg
>> I'll be on the lookout for it. Thanks.
>>
>> [..]
>
> Eric,
>
> with the above patch I'm getting this on a stock RHEL 7.4 kernel:
>
> [  599.659110] NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! 
> [ovstest:23936]
> [  599.659464] Modules linked in: nf_nat_tftp nf_conntrack_tftp 
> nf_nat_ftp nf_conntrack_ftp nf_conntrack_netlink bonding 8021q garp 
> mrp veth netconsole ip6t_rpfilter ipt_REJECT nf_reject_ipv4 
> ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set nfnetlink ebtable_nat 
> ebtable_broute bridge stp llc ip6table_nat nf_conntrack_ipv6 
> nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security 
> ip6table_raw iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 
> nf_nat nf_conntrack iptable_mangle iptable_security iptable_raw 
> ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter 
> snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hda_core 
> snd_hwdep snd_seq snd_seq_device snd_pcm iosf_mbi snd_timer ppdev snd 
> crc32_pclmul ghash_clmulni_intel aesni_intel sg lrw gf128mul 
> glue_helper ablk_helper pcspkr i2c_piix4 joydev cryptd soundcore 
> virtio_balloon parport_pc parport ip_tables xfs libcrc32c sr_mod cdrom 
> ata_generic pata_acpi qxl drm_kms_helper syscopyarea sysfillrect 
> sysimgblt fb_sys_fops ttm drm virtio_console virtio_blk virtio_net 
> ata_piix libata crct10dif_pclmul crct10dif_common crc32c_intel 
> serio_raw floppy i2c_core virtio_pci virtio_ring virtio dm_mirror 
> dm_region_hash dm_log dm_mod [last unloaded: ip6_udp_tunnel]
> [  599.662118] CPU: 3 PID: 23936 Comm: ovstest Tainted: G OE 
> ------------   3.10.0-693.el7.x86_64 #1
> [  599.662351] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
> [  599.662573] task: ffff880235871fa0 ti: ffff880218668000 task.ti: 
> ffff880218668000
> [  599.662823] RIP: 0010:[<ffffffffc0437210>] [<ffffffffc0437210>] 
> nf_ct_iterate_cleanup+0xb0/0x210 [nf_conntrack]
> [  599.663082] RSP: 0018:ffff88021866b870  EFLAGS: 00000246
> [  599.663325] RAX: 0000000000000001 RBX: ffff88017cebec00 RCX: 
> 0000000000000000
> [  599.663583] RDX: 0000000000006b35 RSI: 0000000000000000 RDI: 
> ffff88022de5e140
> [  599.663828] RBP: ffff88021866b8b0 R08: 0000000000000000 R09: 
> 0000000000000000
> [  599.664070] R10: ffff88017cebec00 R11: 0000000000000400 R12: 
> 0000000000000000
> [  599.664310] R13: ffff88021866b8b0 R14: 0000000000000000 R15: 
> 0000000000000000
> [  599.664548] FS:  00007f3c91f91a00(0000) GS:ffff88023fd80000(0000) 
> knlGS:0000000000000000
> [  599.664790] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [  599.665030] CR2: 00000000004be7a0 CR3: 000000021d745000 CR4: 
> 00000000001406e0
> [  599.665274] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 
> 0000000000000000
> [  599.665514] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 
> 0000000000000400
> [  599.665753] Stack:
> [  599.665984]  00005d8000000000 ffffffffc044c594 ffffffff81ad9d40 
> ffff88021866b948
> [  599.666251]  ffffffff81ad9d40 ffff8802347ed400 0000000000000014 
> ffffffff81ad9d40
> [  599.666535]  ffff88021866b8c0 ffffffffc043738c ffff88021866b938 
> ffffffffc053688b
> [  599.666809] Call Trace:
> [  599.667056]  [<ffffffffc043738c>] 
> nf_conntrack_flush_report+0x1c/0x20 [nf_conntrack]
> [  599.667310]  [<ffffffffc053688b>] 
> ctnetlink_del_conntrack+0x19b/0x1d0 [nf_conntrack_netlink]
> [  599.667587]  [<ffffffff8134bfa2>] ? nla_parse+0x32/0x120
> [  599.667842]  [<ffffffffc04e050f>] nfnetlink_rcv_msg+0x24f/0x260 
> [nfnetlink]
> [  599.668105]  [<ffffffffc04e02c0>] ? nfnetlink_bind+0x60/0x60 
> [nfnetlink]
> [  599.668356]  [<ffffffff815bd929>] netlink_rcv_skb+0xa9/0xc0
> [  599.668622]  [<ffffffffc04e088f>] nfnetlink_rcv+0x27f/0x50a 
> [nfnetlink]
> [  599.668921]  [<ffffffff815bd012>] netlink_unicast+0xf2/0x1b0
> [  599.669179]  [<ffffffff815bd3ef>] netlink_sendmsg+0x31f/0x6a0
> [  599.669434]  [<ffffffff812b4d65>] ? sock_has_perm+0x75/0x90
> [  599.669690]  [<ffffffff8156a580>] sock_sendmsg+0xb0/0xf0
> [  599.669939]  [<ffffffff81328da2>] ? radix_tree_lookup_slot+0x22/0x50
> [  599.670199]  [<ffffffff8156ae29>] ___sys_sendmsg+0x3a9/0x3c0
> [  599.670438]  [<ffffffff815baac0>] ? netlink_insert+0x1a0/0x2f0
> [  599.670701]  [<ffffffff812b4d65>] ? sock_has_perm+0x75/0x90
> [  599.670950]  [<ffffffff815691b2>] ? move_addr_to_user+0xb2/0xd0
> [  599.671207]  [<ffffffff8156929c>] ? SYSC_getsockname+0xcc/0xe0
> [  599.671458]  [<ffffffff8156b5f1>] __sys_sendmsg+0x51/0x90
> [  599.671704]  [<ffffffff8156b642>] SyS_sendmsg+0x12/0x20
> [  599.671943]  [<ffffffff816b4fc9>] system_call_fastpath+0x16/0x1b
> [  599.672186] Code: 74 16 e9 3c 01 00 00 0f 1f 40 00 4d 8b 3f 41 f6 
> c7 01 0f 85 2b 01 00 00 41 80 7f 37 00 75 ec 4d 8d 6f f0 4c 89 e6 4c 
> 89 ef ff d3 <85> c0 74 dc f0 41 ff 45 00 48 8b 7d c8 ff 14 25 50 31 a1 
> 81 e8
>
> So perhaps there's something else going on?
>

I added back this part of the original patch:

diff --git a/tests/system-kmod-macros.at b/tests/system-kmod-macros.at
index f23a406..2b9b691 100644
--- a/tests/system-kmod-macros.at
+++ b/tests/system-kmod-macros.at
@@ -23,6 +23,7 @@ m4_define([OVS_TRAFFIC_VSWITCHD_START],
                 on_exit 'modprobe -q -r mod'
                ])
     on_exit 'ovs-dpctl del-dp ovs-system'
+   on_exit 'ovs-appctl dpctl/flush-conntrack'
     _OVS_VSWITCHD_START([])
     dnl Add bridges, ports, etc.
     AT_CHECK([ovs-vsctl -- _ADD_BR([br0]) -- $1 m4_if([$2], [], [], [| 
uuidfilt]
diff --git a/utilities/ovs-lib.in b/utilities/ovs-lib.in
index 4dc3151..4c3ad0f 100644
--- a/utilities/ovs-lib.in
+++ b/utilities/ovs-lib.in
@@ -616,6 +616,7 @@ force_reload_kmod () {
      for dp in `ovs-dpctl dump-dps`; do
          action "Removing datapath: $dp" ovs-dpctl del-dp "$dp"
      done
+    action "ovs-appctl dpctl/flush-conntrack"

      for vport in `awk '/^vport_/ { print $1 }' /proc/modules`; do
          action "Removing $vport module" rmmod $vport

And now the test is running successfully since then.  So I'll keep that 
part of the patch as well.

- Greg
