On Wed, Dec 24, 2008 at 11:42 PM, Luke Zhang (luzhan) <luz...@cisco.com> wrote:
> Hi Jeff,
>
> Thanks a lot for your detailed explain. I have my dmesg output attached.
> I actually have e1000_7.2.7-NAPI compiled in the kernel ( I thought it
> is 7.0.36 when I saw the change log in e1000_main.c), e1000e_0.2.9.5 and
> "Intel Gigabit Ethernet Network Driver - version 1.2.24" (igb?) loaded
> as modules. Not sure if it impact the system but according to the output
> of "cat /proc/bus/pci/devlist", the e1000 is the currently used driver.
>
> The symptom I have is that if I set the MTU to some smaller value than
> 1500 -- 768 for example, and ping the box with a bigger icmp packet, the
> box will hit kdb. See below console messages.
>
> Does this looks like some known issues to you? If it is, Is there any
> other way that this issue can affect the system beside setting a smaller
> mtu?
>
> Thanks a lot for your detailed explain and quick response, and Merry
> Christmas to the list!
>
> -- Luke
>
> Kernel BUG at skbuff:91
> invalid operand: 0000 [1] PREEMPT SMP
>
> Entering kdb (current=0x00000100dffa0e10, pid 0) on processor 1 Oops:
> <NULL>
> due to oops @ 0xffffffff804b3df2
> r15 = 0x00000101107373a0 r14 = 0x000000000000003b
> r13 = 0xffffff000007c740 r12 = 0x00000000ffffffff
> rbp = 0x00000101100fa500 rbx = 0x0000000000000000
> r11 = 0x0000000000000000 r10 = 0x00000000ffffffff
> r9 = 0x0000000000000000 r8 = 0x0000000000000102
> rax = 0x0000000000000032 rcx = 0x0000000000000001
> rdx = 0x0000000000000001 rsi = 0x0000000000000212
> rdi = 0xffffffff806836c0 orig_rax = 0xffffffffffffffff
> rip = 0xffffffff804b3df2 cs = 0x0000000000000010
> eflags = 0x0000000000010212 rsp = 0x00000100dffbfd70
> ss = 0x00000100dffbe000 ®s = 0x00000100dffbfcd8
> kdb_do_dump entered
> Forcing dump (if configured)
> LKCD not yet configured, can't take dump now
> NMI Watchdog detected LOCKUP on CPU1CPU 1
> Modules linked in: acceptfilter ukse streaming sendsocket kqueue mv_sata
> Pid: 0, comm: swapper Tainted: P 2.6.10
> RIP: 0010:[<ffffffff8011a1ea>]
> <ffffffff8011a1ea>{__smp_call_function+122}
> RSP: 0018:00000100dffbf978 EFLAGS: 00000097
> RAX: 0000000000000000 RBX: 0000000000000001 RCX: 0000000000000002
> RDX: 0000ffff0000ffff RSI: 0000000000000000 RDI: ffffffff80819d08
> RBP: 0000000000000000 R08: 0000000000000203 R09: 0000000000000000
> R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff80115040
> R13: 0000000000000000 R14: ffffffff80115040 R15: 0000000000000005
> FS: 0000000000000000(0000) GS:ffffffff808a4780(0000)
> knlGS:0000000000000000
> CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b
> CR2: 000000005592cc10 CR3: 00000000dffaa000 CR4: 00000000000006e0
> Process swapper (pid: 0, threadinfo 00000100dffa6000, task
> 00000100dffa0e10)
> Stack: ffffffff80115040 0000000000000000 ffffffff00000000
> 0000000000000000
> 0000000000000015 0000000000000000 0000000000000002
> 0000000000000001
> 0000000000000000 ffffffff8011a25f
> Call Trace:<IRQ> <ffffffff80115040>{machine_restart+0}
> <ffffffff8011a25f>{smp_call_function+63}
> <ffffffff801150b3>{machine_restart+115}
> <ffffffff802c6c7b>{kdb_reboot+11}
> <ffffffff802cad06>{kdb_main_loop+1478}
> <ffffffff804ab5ea>{kdba_getregcontents+330}
> <ffffffff804aba99>{kdba_main_loop+105}
> <ffffffff802c73c7>{kdb+1719}
> <ffffffff804abe36>{kdba_entry+86}
> <ffffffff801436e0>{notifier_call_chain+32}
> <ffffffff8010fef2>{__die+146} <ffffffff8010ff86>{die+54}
> <ffffffff8011015e>{do_trap+270}
> <ffffffff80131090>{__wake_up_common+64}
> <ffffffff801103c1>{do_invalid_op+145}
> <ffffffff804b3df2>{skb_over_panic+50}
> <ffffffff80135b28>{release_console_sem+440}
> <ffffffff80135ce0>{vprintk+432}
> <ffffffff80135d8d>{printk+141} <ffffffff8010f011>{error_exit+0}
> <ffffffff804b3df2>{skb_over_panic+50}
> <ffffffff804b3df2>{skb_over_panic+50}
> <ffffffff80345179>{e1000_clean_rx_irq+761}
> <ffffffff804b453d>{__kfree_skb+205}
> <ffffffff80344d4a>{e1000_clean+842}
> <ffffffff804ba6a4>{net_rx_action+132}
> <ffffffff8013aa71>{__do_softirq+113}
> <ffffffff8013ab25>{do_softirq+53}
> <ffffffff801577c3>{irq_exit+67} <ffffffff8011145f>{do_IRQ+63}
> <ffffffff8010eaef>{ret_from_intr+0} <EOI>
> <ffffffff8010c8b6>{mwait_idle+86}
> <ffffffff8010c81c>{cpu_idle+44}
> <ffffffff808bdfe3>{start_secondary+115}
>
>
> debugshell# cat /proc/bus/pci/devlist
> 0000 0600000c 80863590 0
> 0001 ff00000c 80863591 0
> 0008 0880000c 80863594 5
> 0010 0604000c 80863595 0
> 0020 0604000c 80863597 0
> 0030 0604000c 80863599 0
> 00e8 0c030002 808624d2 10 uhci_hcd
> 00e9 0c030002 808624d4 13 uhci_hcd
> 00ea 0c030002 808624d7 12 uhci_hcd
> 00eb 0c030002 808624de 10 uhci_hcd
> 00ef 0c032002 808624dd 17 ehci_hcd
> 00f0 060400c2 8086244e 0
> 00f8 06010002 808624d0 0
> 00f9 01018a02 808624db 12 PIIX_IDE
> 00fb 0c050002 808624d3 a
> 0100 06040009 80860329 0
> 0101 08002009 80860326 0
> 0102 06040009 8086032a 0
> 0103 08002009 80860327 0
> 0218 01000009 11ab6081 1c mvSata
> 0308 01000009 11ab6081 30 mvSata
> 0310 02000003 80861079 36 e1000
> 0311 02000003 80861079 37 e1000
> 0500 06040009 80860329 0
> 0501 08002009 80860326 0
> 0502 06040009 8086032a 0
> 0503 08002009 80860327 0
> 0608 02000003 80861079 48 e1000
> 0609 02000003 80861079 49 e1000
> 0708 02000003 80861079 60 e1000
> 0709 02000003 80861079 61 e1000
> 0808 03000027 10024752 a
>
> debugshell# lsmod
> Module Size Used by
> acceptfilter 8840 0
> ukse 377344 0
> streaming 91752 0
> sendsocket 13336 0
> kqueue 21920 1 sendsocket
> mptspi 9868 0
> mptctl 73352 0
> mptsas 21272 0
> mptscsih 41220 2 mptspi,mptsas
> mptbase 63072 4 mptspi,mptctl,mptsas,mptscsih
> e1000e 115492 0
> igb 105880 0
> mv_sata 81804 18
>
>
>
> -----Original Message-----
> From: tar...@gmail.com [mailto:tar...@gmail.com
> <mailto:tar...@gmail.com> ] On Behalf Of Jeff Kirsher
> Sent: Wednesday, December 24, 2008 7:37 PM
> To: Luke Zhang (luzhan)
> Cc: e1000-devel@lists.sourceforge.net
> Subject: Re: [E1000-devel] E1000 change log
>
> On Tue, Dec 23, 2008 at 9:49 PM, Luke Zhang (luzhan) <luz...@cisco.com>
> wrote:
>> Hi,
>>
>> I ran into a kernel panic and would like to see if it is some known
>> issue . I checked the sourceforge pages but can't find the changelog.
>> Where can I find the change log and issues fixed for the e1000 driver?
>>
>> And what is the relationship between e1000, e1000e and igb?
>>
>> I am using e1000 7.0.36+2.6.10 btw.
>>
>> Thanks.
>>
>> -- Luke
>>
>>
>
> To view the change log for each release, you need to click on the clip
> board icon for each release. Here is the sourceforge link for where you
> can see all the released versions of e1000.
> http://sourceforge.net/project/showfiles.php?group_id=42302&package_id=5
> 4835
> <http://sourceforge.net/project/showfiles.php?group_id=42302&package_id=
> 54835>
>
> The e1000 driver was the only driver we had for a long time which
> supported all gigabit Intel NICs, but as time passed, we found that we
> needed to split the driver into three drivers to make the driver(s)
> cleaner. So e1000 driver now supports PCI and PCI-X devices. E1000E
> supports early PCI-E devices and ICH devices, and IGB supports the
> latest PCI-E devices. Devices supported by IGB have never been
> supported in e1000, it is a new driver altogether. Most devices in the
> E1000E driver was once supported in the E1000 7.x driver.
>
> I see that you are running quite an old e1000 driver, which I would
> suggest running the latest 8.0.6 version of the driver unless your NIC
> requires the e1000e driver in which case I would suggest moving to the
> e1000e driver (version 0.5.8.2).
>
> If you supply the output for dmesg and lspci -vvv, it will tell what
> silicon you have and will let us know what driver is best for you.
>
> --
> Cheers,
> Jeff
>
FYI, attachments are stripped on the mailing list, so you will need to
put the output from dmesg and lspci inline, please.
--
Cheers,
Jeff
------------------------------------------------------------------------------
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
