hw csum failure

Issue

• We noticed some networking errors in our OpenStack implementation.

Error message seen in /var/log/messages:

Mar 25 09:45:47  kernel: br-int: hw csum failure.
Mar 25 09:45:47  kernel: Pid: 0, comm: swapper Not tainted 2.6.32-431.8.1.el6.x86_64 #1
Mar 25 09:45:47  kernel: Call Trace:
Mar 25 09:45:47  kernel: <IRQ>  [<ffffffff8145bce2>] ? netdev_rx_csum_fault+0x42/0x50
Mar 25 09:45:47  kernel: [<ffffffff814542f0>] ? __skb_checksum_complete_head+0x60/0x70
Mar 25 09:45:47  kernel: [<ffffffff81454311>] ? __skb_checksum_complete+0x11/0x20
Mar 25 09:45:47  kernel: [<ffffffff814dbdcd>] ? nf_ip_checksum+0x5d/0x130
Mar 25 09:45:47  kernel: [<ffffffffa0371d31>] ? udp_error+0xb1/0x1e0 [nf_conntrack]
Mar 25 09:45:47  kernel: [<ffffffffa020efe2>] ? ovs_vport_send+0x22/0x90 [openvswitch]
Mar 25 09:45:47  kernel: [<ffffffff8144f05e>] ? __skb_clone+0x2e/0x120
Mar 25 09:45:47  kernel: [<ffffffffa036bc98>] ? nf_conntrack_in+0x138/0xa00 [nf_conntrack]
Mar 25 09:45:47  kernel: [<ffffffffa020db46>] ? find_bucket+0x66/0x70 [openvswitch]
Mar 25 09:45:47  kernel: [<ffffffffa020dd61>] ? ovs_flow_tbl_lookup+0x51/0xb0 [openvswitch]
Mar 25 09:45:47  kernel: [<ffffffffa0204721>] ? ipv4_conntrack_in+0x21/0x30 [nf_conntrack_ipv4]
Mar 25 09:45:47  kernel: [<ffffffff814896e9>] ? nf_iterate+0x69/0xb0
Mar 25 09:45:47  kernel: [<ffffffff81493be0>] ? ip_rcv_finish+0x0/0x440
Mar 25 09:45:47  kernel: [<ffffffff814898a6>] ? nf_hook_slow+0x76/0x120
Mar 25 09:45:47  kernel: [<ffffffff81493be0>] ? ip_rcv_finish+0x0/0x440
Mar 25 09:45:47  kernel: [<ffffffff81494284>] ? ip_rcv+0x264/0x350
Mar 25 09:45:47  kernel: [<ffffffffa0210513>] ? ovs_netdev_frame_hook+0xb3/0x110 [openvswitch]
Mar 25 09:45:47  kernel: [<ffffffff8145b83b>] ? __netif_receive_skb+0x4ab/0x750
Mar 25 09:45:47  kernel: [<ffffffff8145bb7a>] ? process_backlog+0x9a/0x100
Mar 25 09:45:47  kernel: [<ffffffff81460e33>] ? net_rx_action+0x103/0x2f0
Mar 25 09:45:47  kernel: [<ffffffff8107a8e1>] ? __do_softirq+0xc1/0x1e0
Mar 25 09:45:47  kernel: [<ffffffff810e6eb0>] ? handle_IRQ_event+0x60/0x170
Mar 25 09:45:47  kernel: [<ffffffff8100c30c>] ? call_softirq+0x1c/0x30
Mar 25 09:45:47  kernel: [<ffffffff8100fa75>] ? do_softirq+0x65/0xa0
Mar 25 09:45:47  kernel: [<ffffffff8107a795>] ? irq_exit+0x85/0x90
Mar 25 09:45:47  kernel: [<ffffffff815313e5>] ? do_IRQ+0x75/0xf0
Mar 25 09:45:47  kernel: [<ffffffff8100b9d3>] ? ret_from_intr+0x0/0x11
Mar 25 09:45:47  kernel: <EOI>  [<ffffffff812e0a6e>] ? intel_idle+0xde/0x170
Mar 25 09:45:47  kernel: [<ffffffff812e0a51>] ? intel_idle+0xc1/0x170

• Logs are filling up with "hw csum failure" messages.