[Click] dev_queue_xmit vs. hard_start_xmit

[Eddie Kohler]

Hi Ricard, all,

A patch from Joonwoo has been applied that should use dev_queue_xmit on newer 
kernels, as you suggest.  Thanks!

Eddie


On 1/11/11 12:21 AM, Ricard Vilalta wrote:
>   Hi Roman,
>
> To me looks like the module myri10ge is crashing. What is strange is that running click in user mode everything runs smoothly.
> As I told you in my previous e-mail, using dev_queue_xmit kernel module runs smoothly, too (not only for myricom NIC but intel 1Gb ethernet NIC).
>
> Thanks for your time,
>
> Ricard
>
> Jan 7 11:18:29 strongest-2 kernel: [160300.209737] click: starting router thread pid 21038 (f3d20180)
> Jan 7 11:20:52 strongest-2 kernel: [160443.541141] BUG: unable to handle kernel NULL pointer dereference at (null)
> Jan 7 11:20:52 strongest-2 kernel: [160443.541229] IP: [<f846aba4>] myri10ge_xmit+0x455/0x9c3 [myri10ge]
> Jan 7 11:20:52 strongest-2 kernel: [160443.541284] *pde = 00000000
> Jan 7 11:20:52 strongest-2 kernel: [160443.541325] Oops: 0002 [#1] SMP
> Jan 7 11:20:52 strongest-2 kernel: [160443.541369] last sysfs file: /sys/module/vt/parameters/default_utf8
> Jan 7 11:20:52 strongest-2 kernel: [160443.541420] Modules linked in: click proclikefs 8021q garp stp loop snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep snd_pcm snd_timer snd i2c_i801 soundcore snd_page_alloc i2c_core ioatdma button myri10ge inet_lro evdev pcspkr processor ext3 jbd mbcache sg sr_mod sd_mod cdrom crc_t10dif ata_generic ata_piix libata scsi_mod uhci_hcd e1000e ehci_hcd usbcore nls_base igb dca thermal thermal_sys [last unloaded: scsi_wait_scan]
> Jan 7 11:20:52 strongest-2 kernel: [160443.541834] Pid: 21038, comm: kclick Not tainted (2.6.32-5-686 #1) System Product Name
> Jan 7 11:20:52 strongest-2 kernel: [160443.541917] EIP: 0060:[<f846aba4>] EFLAGS: 00010246 CPU: 0
> Jan 7 11:20:52 strongest-2 kernel: [160443.541967] EIP is at myri10ge_xmit+0x455/0x9c3 [myri10ge]
> Jan 7 11:20:52 strongest-2 kernel: [160443.542015] EAX: 00000000 EBX: 00000000 ECX: 0000040a EDX: f6338800
> Jan 7 11:20:52 strongest-2 kernel: [160443.542066] ESI: 00000000 EDI: f50b9080 EBP: f63383e0 ESP: f671de9c
> Jan 7 11:20:52 strongest-2 kernel: [160443.542117] DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
> Jan 7 11:20:52 strongest-2 kernel: [160443.542166] Process kclick (pid: 21038, ti=f671c000 task=f5c24840 task.ti=f671c000)
> Jan 7 11:20:52 strongest-2 kernel: [160443.542247] Stack:
> Jan 7 11:20:52 strongest-2 kernel: [160443.542282] 00000000 00000000 00000000 00000000 f5d6b800 00000000 f5d6bba0 00000000
> Jan 7 11:20:52 strongest-2 kernel: [160443.542349]<0>  f5d6c040 c1386ba0 0000040a 00000000 00000000 09cc77ff 00000000 00000000
> Jan 7 11:20:52 strongest-2 kernel: [160443.542447]<0>  00000000 00000000 0000000c 00008100 f5c20000 c1007569 00000286 0000000f
> Jan 7 11:20:52 strongest-2 kernel: [160443.542574] Call Trace:
> Jan 7 11:20:52 strongest-2 kernel: [160443.542615] [<c1007569>] ? sched_clock+0x5/0x7
> Jan 7 11:20:52 strongest-2 kernel: [160443.542664] [<c126c876>] ? schedule+0x78f/0x7dc
> Jan 7 11:20:52 strongest-2 kernel: [160443.542786] [<fa141a48>] ? _ZN8ToDevice12queue_packetEP6PacketP12netdev_queue+0xf8/0x1c0 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.542948] [<fa155bc5>] ? _ZN13FullNoteQueue4pushEiP6Packet+0x1a5/0x220 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543102] [<fa141bfd>] ? _ZN8ToDevice8run_taskEP4Task+0xed/0x290 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543225] [<fa13ebbf>] ? _ZN10FromDevice8run_taskEP4Task+0x7f/0x110 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543347] [<fa0f79d6>] ? _ZN12RouterThread6driverEv+0x1a6/0x4b0 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543433] [<fa0dbdcd>] ? click_lalloc+0x2d/0x60 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543511] [<fa0d3f50>] ? _ZN6VectorIPvE7reserveEi+0x90/0xb0 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543623] [<fa180f7a>] ? _ZL11click_schedPv+0xda/0x1a0 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543734] [<fa180ea0>] ? _ZL11click_schedPv+0x0/0x1a0 [click]
> Jan 7 11:20:52 strongest-2 kernel: [160443.543786] [<c1003d47>] ? kernel_thread_helper+0x7/0x10
> Jan 7 11:20:52 strongest-2 kernel: [160443.543833] Code: 00 00 00 c7 44 24 60 00 00 00 00 8b 5f 54 8b 4f 50 29 d9 89 4c 24 28 8b 55 20 8b 45 18 21 d0 89 44 24 2c 89 c6 8b 45 14 c1 e6 04<89>  3c 30 8b 5c 24 18 31 c0 8b 97 ac 00 00 00 8b 8b a4 00 00 00
> Jan 7 11:20:52 strongest-2 kernel: [160443.544100] EIP: [<f846aba4>] myri10ge_xmit+0x455/0x9c3 [myri10ge] SS:ESP 0068:f671de9c
> Jan 7 11:20:52 strongest-2 kernel: [160443.544186] CR2: 0000000000000000
> Jan 7 11:20:52 strongest-2 kernel: [160443.544714] ---[ end trace b519e64f26b52814 ]---
>
>
> ----------------original message-----------------
> From: "Roman Chertov" rchertov at cs.ucsb.edu
> To: click at amsterdam.lcs.mit.edu, "Ricard Vilalta" ricard.vilalta at cttc.es
> Date: Mon, 10 Jan 2011 09:33:37 -0800
> -------------------------------------------------
>
>
>> Ricard,
>>
>> Do you have the stack trace of the kernel panic?
>>
>> Roman
>>
>> On Mon, 10 Jan 2011 12:58:00 +0100 Ricard Vilalta ricard.vilalta at cttc.es
>> wrote
>>
>>> Hi all,
>>> I am not a linux kernel guru and I would like your support. I have
>>> lately been working with Myricom NIC's (using linux kernel module
>>> myri10ge) and it was impossible to forward paquets from another ethernet
>>> interface to the myricom interface. This leaded to kernel panic. I have
>>> been trying several solutions and the only one which seems to be working
>>> is using dev_queue_xmit instead of hard_start_xmit on todevice. I am
>>> using patchless click with linux kernel 2.6.32-35.
>>>
>>> I have fully tested the change and seems to work for me. What do you
>>> think of this change? May it lead to another type of errors or
>>> unexpected behaviors?
>>>
>>> Thanks in advance.
>>>
>>> Best regards,
>>> Ricard
>>>
>>> --
>>> ______________________________________________________________
>>>
>>> Ricard Vilalta
>>> Research Engineer
>>> Optical Networking Area (ONA) http://wikiona.cttc.es/
>>> CTTC - Centre Tecnològic de Telecomunicacions de Catalunya
>>> Parc Mediterrani de la Tecnologia (PMT)
>>> Av. Carl Friedrich Gauss 7,
>>> 08860 Castelldefels (Barcelona), Spain
>>> http://www.cttc.es/
>>> Phone: +34 93 396 71 70 (ext. 2232). Fax: +34 93 645 29 01
>>> E-mail: ricard.vilalta at cttc.es
>>>
>>>
>>> _______________________________________________
>>> click mailing list
>>> click at amsterdam.lcs.mit.edu
>>> https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>>
>>
>