PDA

View Full Version : Fedora Core 5 hw csum failure?



GlobalAccess
27th November 2007, 10:02 AM
Lately i am getting Kernel and eth0 hw csum failures, so that the services on my servers crash. Anyone got a solution?

jcliburn
27th November 2007, 04:36 PM
Care to provide any details?

GlobalAccess
27th November 2007, 05:03 PM
Here is the message log:



Nov 27 16:59:49 x kernel: <unknown>: hw csum failure.
Nov 27 16:59:49 x kernel: [<c05b91fa>] skb_copy_and_csum_datagram_iovec+0xb0/0xcd
Nov 27 16:59:49 x kernel: [<c05f6c08>] udp_recvmsg+0xdd/0x1cd
Nov 27 16:59:49 x kernel: [<c05b3a2d>] sock_common_recvmsg+0x3e/0x54
Nov 27 16:59:49 x kernel: [<c05b2282>] sock_recvmsg+0xec/0x107
Nov 27 16:59:49 x kernel: [<c043d499>] autoremove_wake_function+0x0/0x35
Nov 27 16:59:49 x kernel: [<c04f6ab8>] copy_from_user+0x32/0x5e
Nov 27 16:59:49 x kernel: [<c05b3182>] sys_recvfrom+0xd8/0x12d
Nov 27 16:59:49 x kernel: [<c044662a>] futex_wake+0x81/0xb0
Nov 27 16:59:49 x kernel: [<c04466d8>] do_futex+0x7f/0x959
Nov 27 16:59:49 x kernel: [<c0441816>] getnstimeofday+0x30/0xbe
Nov 27 16:59:49 x kernel: [<c05b3710>] sys_socketcall+0x1cd/0x261
Nov 27 16:59:49 x kernel: [<c040518a>] syscall_call+0x7/0xb


as well as



Nov 27 16:59:39 x kernel: eth0: hw csum failure.
Nov 27 16:59:39 x kernel: [<c05b8f5f>] __skb_checksum_complete_head+0x45/0x58
Nov 27 16:59:39 x kernel: [<c05b8f7a>] __skb_checksum_complete+0x8/0xc
Nov 27 16:59:39 x kernel: [<c05f1bd5>] tcp_v4_rcv+0x10e/0x8d1
Nov 27 16:59:39 x kernel: [<c042536b>] inc_nr_running+0x13/0x26
Nov 27 16:59:39 x kernel: [<c05d8a69>] ip_local_deliver+0x189/0x230
Nov 27 16:59:39 x kernel: [<c05d88a7>] ip_rcv+0x481/0x4ba
Nov 27 16:59:39 x kernel: [<c042536b>] inc_nr_running+0x13/0x26
Nov 27 16:59:39 x kernel: [<c05bbb30>] netif_receive_skb+0x2e1/0x346
Nov 27 16:59:39 x kernel: [<c05bdab7>] process_backlog+0x92/0xe9
Nov 27 16:59:39 x kernel: [<c05bdc74>] net_rx_action+0x9a/0x196
Nov 27 16:59:39 x kernel: [<c0431e06>] __do_softirq+0x66/0xd3
Nov 27 16:59:39 x kernel: [<c04073d5>] do_softirq+0x6c/0xce
Nov 27 16:59:39 x kernel: [<c04446f3>] tick_do_update_jiffies64+0x93/0xa8
Nov 27 16:59:39 x kernel: [<c045ba21>] handle_fasteoi_irq+0x0/0xa6
Nov 27 16:59:39 x kernel: [<c0431cc9>] irq_exit+0x38/0x6b
Nov 27 16:59:39 x kernel: [<c04074d6>] do_IRQ+0x9f/0xb9
Nov 27 16:59:39 x kernel: [<c043ff60>] hrtimer_start+0xe6/0xf0
Nov 27 16:59:39 x kernel: [<c0403ddf>] default_idle+0x0/0x55
Nov 27 16:59:39 x kernel: [<c0405b6f>] common_interrupt+0x23/0x28
Nov 27 16:59:39 x kernel: [<c0403ddf>] default_idle+0x0/0x55
Nov 27 16:59:39 x kernel: [<c0421f78>] native_safe_halt+0x2/0x3
Nov 27 16:59:39 x kernel: [<c0403e18>] default_idle+0x39/0x55
Nov 27 16:59:39 x kernel: [<c040340b>] cpu_idle+0xab/0xcc
Nov 27 16:59:39 x kernel: [<c073fa6c>] start_kernel+0x32c/0x334
Nov 27 16:59:39 x kernel: [<c073f177>] unknown_bootoption+0x0/0x195
Nov 27 16:59:39 x kernel: =======================


anyone knows what is meant as cause and how to fix this?

jcliburn
27th November 2007, 06:13 PM
No one will know how to fix it unless you provide sufficient detail.

kernel version? (uname -a)

nic type? (su -c "lspci -vvxxx")

driver version? (su -c "ethtool -i eth0")

Is the nic driver native to your kernel, or did you acquire it and build it out-of-tree?

Does the problem occur under normal operating conditions, or during a particular heavy network load (such as transferring very large files or streaming video)?

Did the error recently begin, or has it been with you all along? Have you made any system changes that you can correlate to the onset of the problem?

Have you tried a newer kernel or distribution?

GlobalAccess
28th November 2007, 07:51 AM
Kernel:
2.6.23.1-49.fc8 #1 SMP Thu Nov 8 21:41:26 EST 2007 i686 athlon i386 GNU/Linux

driver version? (su -c "ethtool -i eth0"):
Operation not supportet

Does the problem occur under normal operating conditions, or during a particular heavy network load (such as transferring very large files or streaming video)? Under normal operationg conditions

Did the error recently begin, or has it been with you all along? Have you made any system changes that you can correlate to the onset of the problem? It began recently in all Fedora Systems, especially FC5 and Fedora 8.

Have you tried a newer kernel or distribution? Other distributions do not have this problem. The VM is running the newest kernel.

GlobalAccess
10th December 2007, 10:18 AM
Anyone got a solution. This is really getting annoying. Its becoming more and more frequent.

keisashankara
3rd March 2011, 03:07 AM
https://bugzilla.redhat.com/show_bug.cgi?id=91236

It seems to be a known bug related to your network drivers.

I am hitting the same bug in 7.3.

No solution is known, however this may help:

an intel guy said that in l-k:

-- http://marc.theaimsgroup.com/?t=105337876900002&r=1&w=2 --
can you try turning off Rx checksum offloading in e100? Set the module
parameter XsumRX=0 to turn it off.
--end--

Hlingler
3rd March 2011, 05:21 AM
Are you serious!? FC5 !? Last post dated 10th December 2007!?

Thread closed. Please don't resurrect the dead without a very good reason.

FedoraForum.org Posting Rules (http://www.fedoraforum.org/?view=guide)

V