Linux Archive

Linux Archive (http://www.linux-archive.org/)
-   Debian Kernel (http://www.linux-archive.org/debian-kernel/)
-   -   Bug#673107: kirkwood: TCP checksum errors when using MTU 9000 (http://www.linux-archive.org/debian-kernel/667325-bug-673107-kirkwood-tcp-checksum-errors-when-using-mtu-9000-a.html)

Martin Michlmayr 05-16-2012 07:53 AM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Package: linux-2.6
Version: 2.6.32-45

A user on the Kirkwood platform reported TCP checksum errors when
using MTU 9000:
http://forum.qnap.com/viewtopic.php?f=147&t=59281&p=266527#p266527

This has been previously discussed:
http://lists.debian.org/debian-arm/2009/11/msg00087.html

iirc a patch went into the kernel to set the limits for the various
platforms (orion, kirkwood)... or maybe it was just discussed and
never went in.

Arnaud, can you investigate whether this patch really went in or
what's going on here?

--
Martin Michlmayr
http://www.cyrius.com/



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120516075339.GA28693@jirafa.cyrius.com">http://lists.debian.org/20120516075339.GA28693@jirafa.cyrius.com

Arnaud Patard (Rtp) 05-16-2012 02:31 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Martin Michlmayr <tbm@cyrius.com> writes:

Hi,

> Package: linux-2.6
> Version: 2.6.32-45
>
> A user on the Kirkwood platform reported TCP checksum errors when
> using MTU 9000:
> http://forum.qnap.com/viewtopic.php?f=147&t=59281&p=266527#p266527
>
> This has been previously discussed:
> http://lists.debian.org/debian-arm/2009/11/msg00087.html
>
> iirc a patch went into the kernel to set the limits for the various
> platforms (orion, kirkwood)... or maybe it was just discussed and
> never went in.

>From a quick look, it seems that the support to handle the smaller FIFO
on kirkwood/dove has been merged on the ethernet driver side but has
never been merged in the platform code. Will make a patch and come back
once I'll have more informations.

Thanks,
Arnaud



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 87k40cdwop.fsf@lebrac.rtp-net.org">http://lists.debian.org/87k40cdwop.fsf@lebrac.rtp-net.org

Arnaud Patard (Rtp) 05-21-2012 09:40 AM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Hi,

I've uploaded a test kernel at :
http://www.rtp-net.org/kirkwood/linux-image-2.6.32-5-kirkwood_2.6.32-46~bug673107_armel.deb

Can you please check that the issue is gone with it ?


Thanks,
Arnaud



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 87vcjpc1oh.fsf@lebrac.rtp-net.org">http://lists.debian.org/87vcjpc1oh.fsf@lebrac.rtp-net.org

Martin Michlmayr 05-21-2012 12:11 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
* Arnaud Patard <arnaud.patard@rtp-net.org> [2012-05-21 11:40]:
> I've uploaded a test kernel at :
> http://www.rtp-net.org/kirkwood/linux-image-2.6.32-5-kirkwood_2.6.32-46~bug673107_armel.deb
>
> Can you please check that the issue is gone with it ?

I'll ask the person who reported the bug. Thanks!
--
Martin Michlmayr
http://www.cyrius.com/



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120521121135.GB24041@jirafa.cyrius.com">http://lists.debian.org/20120521121135.GB24041@jirafa.cyrius.com

Martin Michlmayr 05-21-2012 09:20 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
* Arnaud Patard <arnaud.patard@rtp-net.org> [2012-05-21 11:40]:
> I've uploaded a test kernel at :
> http://www.rtp-net.org/kirkwood/linux-image-2.6.32-5-kirkwood_2.6.32-46~bug673107_armel.deb
>
> Can you please check that the issue is gone with it ?

Can you make a kernel image for Orion available?

--
Martin Michlmayr
http://www.cyrius.com/



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120521212050.GA32512@jirafa.cyrius.com">http://lists.debian.org/20120521212050.GA32512@jirafa.cyrius.com

Arnaud Patard (Rtp) 05-23-2012 08:48 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Martin Michlmayr <tbm@cyrius.com> writes:

> * Arnaud Patard <arnaud.patard@rtp-net.org> [2012-05-21 11:40]:
>> I've uploaded a test kernel at :
>> http://www.rtp-net.org/kirkwood/linux-image-2.6.32-5-kirkwood_2.6.32-46~bug673107_armel.deb
>>
>> Can you please check that the issue is gone with it ?
>
> Can you make a kernel image for Orion available?

Done. Kernel available at :
http://www.rtp-net.org/kirkwood/linux-image-2.6.32-5-orion5x_2.6.32-46~bug673107_armel.deb

Thanks,
Arnaud



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 87txz6aak8.fsf@lebrac.rtp-net.org">http://lists.debian.org/87txz6aak8.fsf@lebrac.rtp-net.org

Damien Martins 05-27-2012 05:27 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Same behaviour :/

dmesg shows :

"[ 619.869032] [<c002f078>] (unwind_backtrace+0x0/0xdc) from
[<c00ab688>] (__alloc_pages_nodemask+0x4dc/0x57c)
[ 619.878846] [<c00ab688>] (__alloc_pages_nodemask+0x4dc/0x57c) from
[<c00ab73c>] (__get_free_pages+0x14/0x44)
[ 619.888727] [<c00ab73c>] (__get_free_pages+0x14/0x44) from
[<c00cfecc>] (__kmalloc_track_caller+0x40/0x19c)
[ 619.898526] [<c00cfecc>] (__kmalloc_track_caller+0x40/0x19c) from
[<c01fa504>] (__alloc_skb+0x50/0x10c)
[ 619.907976] [<c01fa504>] (__alloc_skb+0x50/0x10c) from [<c01fb5c0>]
(dev_alloc_skb+0x1c/0x44)
[ 619.916599] [<c01fb5c0>] (dev_alloc_skb+0x1c/0x44) from [<bf112264>]
(rxq_refill+0x7c/0x144 [mv643xx_eth])
[ 619.926363] [<bf112264>] (rxq_refill+0x7c/0x144 [mv643xx_eth]) from
[<bf113b0c>] (mv643xx_eth_poll+0x5e4/0x68c [mv643xx_eth])
[ 619.937757] [<bf113b0c>] (mv643xx_eth_poll+0x5e4/0x68c [mv643xx_eth])
from [<c02027b4>] (net_rx_action+0x90/0x208)
[ 619.948171] [<c02027b4>] (net_rx_action+0x90/0x208) from [<c00511d8>]
(__do_softirq+0xc0/0x1a8)
[ 619.956920] [<c00511d8>] (__do_softirq+0xc0/0x1a8) from [<c0051300>]
(irq_exit+0x40/0x94)
[ 619.965108] [<c0051300>] (irq_exit+0x40/0x94) from [<c0028070>]
(asm_do_IRQ+0x70/0x8c)
[ 619.973079] [<c0028070>] (asm_do_IRQ+0x70/0x8c) from [<c0028ad4>]
(__irq_svc+0x34/0x80)

[ 619.981114] Exception stack(0xd5393d50 to 0xd5393d98)
[ 619.986220] 3d40: 401a5d20
401a6000 08000075 00001000
[ 619.994399] 3d60: 00000000 d5393d9c 00000000 d358e268 00000001
d57dcd80 401a5000 d57dcd80

[ 620.002613] 3d80: d7894cdc d5393d98 c00309d4 c00327c4 00000013 ffffffff
[ 620.009295] [<c0028ad4>] (__irq_svc+0x34/0x80) from [<c00327c4>]
(feroceon_flush_user_cache_range+0x24/0x40)
[ 620.019190] [<c00327c4>] (feroceon_flush_user_cache_range+0x24/0x40)
from [<d74e9c70>] (0xd74e9c70)

[ 620.028258] Mem-info:
[ 620.030531] Normal per-cpu:
[ 620.033325] CPU 0: hi: 90, btch: 15 usd: 80
[ 620.038175] active_anon:12271 inactive_anon:12281 isolated_anon:0
[ 620.038191] active_file:10174 inactive_file:9996 isolated_file:0
[ 620.038208] unevictable:0 dirty:26 writeback:0 unstable:0
[ 620.038223] free:13631 slab_reclaimable:1922 slab_unreclaimable:993
[ 620.038240] mapped:5751 shmem:164 pagetables:506 bounce:0
[ 620.067688] Normal free:54524kB min:2032kB low:2540kB high:3048kB
active_anon:49084kB inactive_anon:49124kB active_file:40696kB
inactive_file:39984kB unevictable:0kB isolated(anon):0kB
isolated(file):0kB present:259072kB mlocked:0kB dirty:104kB
writeback:0kB mapped:23004kB shmem:656kB slab_reclaimable:7688kB
slab_unreclaimable:3972kB kernel_stack:1592kB pagetables:2024kB
unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0
all_unreclaimable? no

[ 620.107406] lowmem_reserve[]: 0 0
[ 620.110742] Normal: 12503*4kB 548*8kB 4*16kB 2*32kB 0*64kB 0*128kB
0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 54524kB

[ 620.121725] 21208 total pagecache pages
[ 620.125551] 872 pages in swap cache
[ 620.129078] Swap cache stats: add 1865, delete 993, find 362/382
[ 620.135074] Free swap = 973040kB
[ 620.138422] Total swap = 979832kB
[ 620.153822] 65536 pages of RAM
[ 620.156937] 14150 free pages
[ 620.159818] 1618 reserved pages
[ 620.162955] 2110 slab pages
[ 620.165783] 34061 pages shared
[ 620.168840] 872 pages swap cached

--
Regards,
Damien Martins



--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC263FD.1080408@makelofine.org">http://lists.debian.org/4FC263FD.1080408@makelofine.org

Damien Martins 07-19-2012 08:43 AM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Same behaviour :/

dmesg shows :

"[ 619.869032] [<c002f078>] (unwind_backtrace+0x0/0xdc) from
[<c00ab688>] (__alloc_pages_nodemask+0x4dc/0x57c)
[ 619.878846] [<c00ab688>] (__alloc_pages_nodemask+0x4dc/0x57c) from
[<c00ab73c>] (__get_free_pages+0x14/0x44)
[ 619.888727] [<c00ab73c>] (__get_free_pages+0x14/0x44) from
[<c00cfecc>] (__kmalloc_track_caller+0x40/0x19c)
[ 619.898526] [<c00cfecc>] (__kmalloc_track_caller+0x40/0x19c) from
[<c01fa504>] (__alloc_skb+0x50/0x10c)
[ 619.907976] [<c01fa504>] (__alloc_skb+0x50/0x10c) from [<c01fb5c0>]
(dev_alloc_skb+0x1c/0x44)
[ 619.916599] [<c01fb5c0>] (dev_alloc_skb+0x1c/0x44) from [<bf112264>]
(rxq_refill+0x7c/0x144 [mv643xx_eth])
[ 619.926363] [<bf112264>] (rxq_refill+0x7c/0x144 [mv643xx_eth]) from
[<bf113b0c>] (mv643xx_eth_poll+0x5e4/0x68c [mv643xx_eth])
[ 619.937757] [<bf113b0c>] (mv643xx_eth_poll+0x5e4/0x68c [mv643xx_eth])
from [<c02027b4>] (net_rx_action+0x90/0x208)
[ 619.948171] [<c02027b4>] (net_rx_action+0x90/0x208) from [<c00511d8>]
(__do_softirq+0xc0/0x1a8)
[ 619.956920] [<c00511d8>] (__do_softirq+0xc0/0x1a8) from [<c0051300>]
(irq_exit+0x40/0x94)
[ 619.965108] [<c0051300>] (irq_exit+0x40/0x94) from [<c0028070>]
(asm_do_IRQ+0x70/0x8c)
[ 619.973079] [<c0028070>] (asm_do_IRQ+0x70/0x8c) from [<c0028ad4>]
(__irq_svc+0x34/0x80)
[ 619.981114] Exception stack(0xd5393d50 to 0xd5393d98)
[ 619.986220] 3d40: 401a5d20
401a6000 08000075 00001000
[ 619.994399] 3d60: 00000000 d5393d9c 00000000 d358e268 00000001
d57dcd80 401a5000 d57dcd80
[ 620.002613] 3d80: d7894cdc d5393d98 c00309d4 c00327c4 00000013 ffffffff
[ 620.009295] [<c0028ad4>] (__irq_svc+0x34/0x80) from [<c00327c4>]
(feroceon_flush_user_cache_range+0x24/0x40)
[ 620.019190] [<c00327c4>] (feroceon_flush_user_cache_range+0x24/0x40)
from [<d74e9c70>] (0xd74e9c70)
[ 620.028258] Mem-info:
[ 620.030531] Normal per-cpu:
[ 620.033325] CPU 0: hi: 90, btch: 15 usd: 80
[ 620.038175] active_anon:12271 inactive_anon:12281 isolated_anon:0
[ 620.038191] active_file:10174 inactive_file:9996 isolated_file:0
[ 620.038208] unevictable:0 dirty:26 writeback:0 unstable:0
[ 620.038223] free:13631 slab_reclaimable:1922 slab_unreclaimable:993
[ 620.038240] mapped:5751 shmem:164 pagetables:506 bounce:0
[ 620.067688] Normal free:54524kB min:2032kB low:2540kB high:3048kB
active_anon:49084kB inactive_anon:49124kB active_file:40696kB
inactive_file:39984kB unevictable:0kB isolated(anon):0kB
isolated(file):0kB present:259072kB mlocked:0kB dirty:104kB
writeback:0kB mapped:23004kB shmem:656kB slab_reclaimable:7688kB
slab_unreclaimable:3972kB kernel_stack:1592kB pagetables:2024kB
unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0
all_unreclaimable? no
[ 620.107406] lowmem_reserve[]: 0 0
[ 620.110742] Normal: 12503*4kB 548*8kB 4*16kB 2*32kB 0*64kB 0*128kB
0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 54524kB
[ 620.121725] 21208 total pagecache pages
[ 620.125551] 872 pages in swap cache
[ 620.129078] Swap cache stats: add 1865, delete 993, find 362/382
[ 620.135074] Free swap = 973040kB
[ 620.138422] Total swap = 979832kB
[ 620.153822] 65536 pages of RAM
[ 620.156937] 14150 free pages
[ 620.159818] 1618 reserved pages
[ 620.162955] 2110 slab pages
[ 620.165783] 34061 pages shared
[ 620.168840] 872 pages swap cached

--
Regards,
Damien Martins


--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 5007C8AB.4080409@makelofine.org">http://lists.debian.org/5007C8AB.4080409@makelofine.org

Ben Hutchings 07-19-2012 11:24 AM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
On Thu, 2012-07-19 at 10:43 +0200, Damien Martins wrote:
> Same behaviour :/
>
> dmesg shows :
[...]

MTU of 9000 requires 4 contiguous pages of memory for each packet. On a
machine with only 256 MB of memory, that tends to be hard to find. This
is not a bug.

Are the checksum errors gone?

Ben.

--
Ben Hutchings
DNRC Motto: I can please only one person per day.
Today is not your day. Tomorrow isn't looking good either.

Damien Martins 07-29-2012 09:41 PM

Bug#673107: kirkwood: TCP checksum errors when using MTU 9000
 
Hi,

As I'm not a TCP/IP guru (neither a good network admin), please find
netowrk capture from Wireshark to analyze network traffic on TCP port
5001 (which is used for iperf testing) on this URL :

https://www.makelofine.org/perso/debug/orion_full.pcap

log : debug
pass : 5c400a761e688

192.168.1.1 is my workstation, 192.168.1.3 is the Orion device (a QNAP
TS-209 Pro 2)



Cordialement,
Damien Martins

Le 19/07/2012 13:24, Ben Hutchings a écrit :

On Thu, 2012-07-19 at 10:43 +0200, Damien Martins wrote:

Same behaviour :/

dmesg shows :

[...]

MTU of 9000 requires 4 contiguous pages of memory for each packet. On a
machine with only 256 MB of memory, that tends to be hard to find. This
is not a bug.

Are the checksum errors gone?

Ben.




--
To UNSUBSCRIBE, email to debian-kernel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 5015AE06.5000403@makelofine.org">http://lists.debian.org/5015AE06.5000403@makelofine.org


All times are GMT. The time now is 01:23 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.