DMA: Out of SW-IOMMU space for xx bytes at device [AR8151 network]

[hyper_ch]

I just setup Kubuntu Raring on my notebook yesterday and things seem to work fine except for samba/cifs. Sill using SMBv3 I can't transfer large files anymore. At about 100MB transfer the kernel log gets filled with those error messages:

DMA: Out of SW-IOMMU space for xx bytes at device ....

and the network becomes completely unusable. Only solution I have found was to reboot the system and at shut down, you also keep getting tons of those errors. However when using SCP or rsync over SSH there aren't any problems.

My network card: 05:00.0 Ethernet controller: Atheros Communications Inc. AR8151 v2.0 Gigabit Ethernet (rev c0)

Meanwhile I reverted back to using the 3.5.0-27 kernel from Quantal - that one has no issues.
---
ApportVersion: 2.8-0ubuntu4
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: hyper 2624 F.... pulseaudio
 /dev/snd/controlC1: hyper 2624 F.... pulseaudio
 /dev/snd/pcmC1D0p: hyper 2624 F...m pulseaudio
CRDA: Error: [Errno 2] No such file or directory
DistroRelease: Ubuntu 13.04
MachineType: ASUSTeK Computer Inc. U36SD
MarkForUpload: True
Package: linux (not installed)
ProcFB:
 0 inteldrmfb
 1 nouveaufb
ProcKernelCmdLine: BOOT_IMAGE=/vmlinuz-3.8.0-7-generic root=UUID=982daec3-56b7-423a-8d4a-f684b0855923 ro splash quiet vt.handoff=7
ProcVersionSignature: Ubuntu 3.8.0-7.15-generic 3.8.0
RelatedPackageVersions:
 linux-restricted-modules-3.8.0-7-generic N/A
 linux-backports-modules-3.8.0-7-generic N/A
 linux-firmware 1.103
Tags: raring
Uname: Linux 3.8.0-7-generic x86_64
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip fuse lpadmin plugdev sambashare sudo vboxusers
dmi.bios.date: 07/12/2011
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: U36SD.205
dmi.board.asset.tag: ATN12345678901234567
dmi.board.name: U36SD
dmi.board.vendor: ASUSTeK Computer Inc.
dmi.board.version: 1.0
dmi.chassis.asset.tag: No Asset Tag
dmi.chassis.type: 10
dmi.chassis.vendor: ASUSTeK Computer Inc.
dmi.chassis.version: 1.0
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrU36SD.205:bd07/12/2011:svnASUSTeKComputerInc.:pnU36SD:pvr1.0:rvnASUSTeKComputerInc.:rnU36SD:rvr1.0:cvnASUSTeKComputerInc.:ct10:cvr1.0:
dmi.product.name: U36SD
dmi.product.version: 1.0
dmi.sys.vendor: ASUSTeK Computer Inc.
---
ApportVersion: 2.8-0ubuntu4
Architecture: amd64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: hyper 2624 F.... pulseaudio
 /dev/snd/controlC1: hyper 2624 F.... pulseaudio
CRDA: Error: [Errno 2] No such file or directory
DistroRelease: Ubuntu 13.04
MachineType: ASUSTeK Computer Inc. U36SD
MarkForUpload: True
Package: linux (not installed)
ProcFB:
 0 inteldrmfb
 1 nouveaufb
ProcKernelCmdLine: BOOT_IMAGE=/vmlinuz-3.8.0-7-generic root=UUID=982daec3-56b7-423a-8d4a-f684b0855923 ro splash quiet vt.handoff=7
ProcVersionSignature: Ubuntu 3.8.0-7.15-generic 3.8.0
RelatedPackageVersions:
 linux-restricted-modules-3.8.0-7-generic N/A
 linux-backports-modules-3.8.0-7-generic N/A
 linux-firmware 1.103
Tags: raring
Uname: Linux 3.8.0-7-generic x86_64
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip fuse lpadmin plugdev sambashare sudo vboxusers
dmi.bios.date: 07/12/2011
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: U36SD.205
dmi.board.asset.tag: ATN12345678901234567
dmi.board.name: U36SD
dmi.board.vendor: ASUSTeK Computer Inc.
dmi.board.version: 1.0
dmi.chassis.asset.tag: No Asset Tag
dmi.chassis.type: 10
dmi.chassis.vendor: ASUSTeK Computer Inc.
dmi.chassis.version: 1.0
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvrU36SD.205:bd07/12/2011:svnASUSTeKComputerInc.:pnU36SD:pvr1.0:rvnASUSTeKComputerInc.:rnU36SD:rvr1.0:cvnASUSTeKComputerInc.:ct10:cvr1.0:
dmi.product.name: U36SD
dmi.product.version: 1.0
dmi.sys.vendor: ASUSTeK Computer Inc.

[Dave Gilbert (ubuntu-treblig)]

Note also in that image is the message repeated a lot
CIFS VFS: No writable handles for inode

[Brad Figg (brad-figg)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1132477

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

apport information

[hyper_ch (bugs-launchpad-net-roleplayer)]

Kernel Log for 3.8.0-7

[hyper_ch (bugs-launchpad-net-roleplayer)]

Kernel Log for 3.5.0-27

[Dave Gilbert (ubuntu-treblig)]

High->problem with inbuilt networking

[Joseph Salisbury (jsalisbury)]

I'd like to perform a bisect to figure out what commit caused this regression. It would be very helpful to know the earliest kernel where the issue started happening as well as the latest kernel that did not have this issue.

Can you test the following kernels and report back? We are looking for the first kernel version that exhibits this bug:

v3.6 final: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.6-quantal/
v3.7 final: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.7-raring/
v3.8-rc1: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.8-rc1-raring/

You don't have to test every kernel, just up until the kernel that first has this bug.

Thanks in advance!

[hyper_ch (bugs-launchpad-net-roleplayer)]

I can't use any of those kernels... for some reasons I don't get prompted for luks password so they just hang

[Joseph Salisbury (jsalisbury)]

Did you install both the linux-image and linux-image-extras .deb packages?

If you did install both packages, can you see if you get prompted for the luks password with the latest upstream stable 3.5 kernel:
http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.5.7.6-quantal/

[hyper_ch (bugs-launchpad-net-roleplayer)]

no, just image and headers. Will also do the extras.

[hyper_ch (bugs-launchpad-net-roleplayer)]

There was a kernel update today and I just tried it at work and I could transfer 4GB and 6GB just fine. I'll test it at home. Maybe today's kernel update already fixed it.

[hyper_ch (bugs-launchpad-net-roleplayer)]

Ok, tested it at home with current 3.8 and same problem again. Maybe I used the 3.5 in the office - I'll have to try tomorrow again.

What I can say is that with 3.7 it all worked fine at home.

[hyper_ch (bugs-launchpad-net-roleplayer)]

Ok, I tested at work again and current 3.8 kernel gives same problems.

[Joseph Salisbury (jsalisbury)]

Can you confirm rc1 of v3.8 also exhibits this bug. That kernel can be downloaded from:
v3.8-rc1: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.8-rc1-raring/

If rc1 has this bug, I can bisect between v3.7 final and v3.8-rc1.

Also, ensure you install both the linux-image and linux-image-extras .deb packages.

Thanks again!

[hyper_ch (bugs-launchpad-net-roleplayer)]

Will do so... latest 3.8.0-8 still has same issue.

[hyper_ch (bugs-launchpad-net-roleplayer)]

Ok, I just checked.... v3.8-rc1 works without flaws.

[Joseph Salisbury (jsalisbury)]

Thanks for testing. To perform a bisect, we need to identify the last good and first bad kernel versions.

Can you test some of the other release candidates:

v3.8-rc2: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.8-rc2-quantal/
v3.8-rc4: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.8-rc4-quantal/
v3.8-rc6: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.8-rc6-quantal/

You don't have to test every kernel, just up until the kernel that first does not have this bug.

Thanks in advance!

[hyper_ch (bugs-launchpad-net-roleplayer)]

I can do that tomorrow :)

3 tries should be sufficient:

 v3.8-rc1-raring/ 22-Dec-2012 02:58 -
 v3.8-rc2-raring/ 03-Jan-2013 03:59 -
 v3.8-rc3-raring/ 10-Jan-2013 03:59 -
 v3.8-rc4-raring/ 18-Jan-2013 05:01 -
 v3.8-rc5-raring/ 25-Jan-2013 21:12 -
 v3.8-rc6-raring/ 01-Feb-2013 03:02 -
 v3.8-rc7-raring/ 08-Feb-2013 22:13 -

starting with 4... if it works, go to 6, if not, go to 2 and check them... and then either one above or below :)

[Joseph Salisbury (jsalisbury)]

Thanks!

[hyper_ch (bugs-launchpad-net-roleplayer)]

Ok, rc7 still works fine... current kernel doesn't.

[hyper_ch (bugs-launchpad-net-roleplayer)]

Is there any reason to not use rc7 kernel compare to quantal 3.5? Because then I could remove the quantal repos now :)

[Joseph Salisbury (jsalisbury)]

Sure, you should be able to use rc7, until this bug is fixed.

Now that we know v3.8 final introduced the regression, I can start a kernel bisect between v3.8-rc7 and v3.8 final. I'll build a test kernel and post a link to it shortly.

[Joseph Salisbury (jsalisbury)]

I started a kernel bisect between v3.7 final and v3.8-rc1. The kernel bisect will require testing of about 7-10 test kernels.

I built the first test kernel, up to the following commit:
42976ad0b26b2465f33c9a9146eb15f3a644d269

The test kernel can be downloaded from:
http://people.canonical.com/~jsalisbury/lp1132477

Can you test that kernel and report back if it has the bug or not. I will build the next test kernel based on your test results.

Thanks in advance

[hyper_ch (bugs-launchpad-net-roleplayer)]

That one works.

[hyper_ch (bugs-launchpad-net-roleplayer)]

btw, did you make an error when you said to bisect between 3.7 final and v3.8-rc1? Since v3.8-rc7-raring/ 08-Feb-2013 22:13 worked...

[Joseph Salisbury (jsalisbury)]

Whoops, yes that is an error from a common template I use.

I'm bisecting between v3.8-rc7 and v3.8 final.

[hyper_ch (bugs-launchpad-net-roleplayer)]

just tested with 3.8.0-9 still same error.

[meh (meh-deactivatedaccount)]

Fixed upstream with:

commit 7cb08d7f3a5ea6131f4f243c2080530ac41cb293
atl1c: restore buffer state

Queued for stable, but doesn't look like it made it for 3.8.2.
Please include said patch

[meh (meh-deactivatedaccount)]

http://www.spinics.net/lists/netdev/msg227820.html

[Dave Gilbert (ubuntu-treblig)]

Upstream patch identified as fix; see comment 44/45

[hyper_ch (bugs-launchpad-net-roleplayer)]

Will this be ported to 3.8.0-xx ?

[Bruno Pagani (b-pagani)]

At least it has been requested to be included in upstream 3.8.3.

[Joseph Salisbury (jsalisbury)]

It looks like commit 7cb08d7f3a5ea6131f4f243c2080530ac41cb293 was applied to v3.9-rc1 and queued up for linux-stable by David Miller.

The fix will make it into the Raring kernel when the 3.8.3 stable updates are applied to the Raring kernel.

[hyper_ch (bugs-launchpad-net-roleplayer)]

Thanks for the help :)

[hyper_ch (bugs-launchpad-net-roleplayer)]

Had to revert back to Quantal because KDE just wouldn't let me login anymore with 3.5 kernel... xserver crash.... 3.8 no fun with the network card. So when will that be included as you marked it now as Fix Committed?

[dyna (ubuntu-dyna)]

Same problem with 3.8.0-13 which should be rebased on 3.8.3 according to http://changelogs.ubuntu.com/changelogs/pool/main/l/linux/linux_3.8.0-13.23/changelog

[Sebastian Ilea (s3b4)]

I can confirm, not fixed yet.

[Jens Kreiensiek (jenx)]

Seems to be fixed with 3.8.0-14, but transmitting large files over NFS is way slower than before...

[Sebastian Ilea (s3b4)]

I tried today with 3.8.0-14. I tried to copy 1,2 GB of jpeg image files (1,5MB average size of the file) from a Windows share (using Nautilus). The transfer froze at 719,1MB. With previous 3.8.0-xx kernels, it usually would froze much sooner (frequently under 100MB, never much above 200MB).

I did not notice any speed drop.

[Sebastian Ilea (s3b4)]

After searching for a solution, I think that I do not have this bug. I probably have Bug #1075923 "nautilus hangs copying large directories from a samba share".

[hyper_ch (bugs-launchpad-net-roleplayer)]

#54

Same here.... I notice now also that with current kernel on the server the transfer has become much slower. I still use 3.5 kernel on the client though.

[Julian Wiedmann (jwiedmann)]

hyper_ch,

please test the very latest kernel from raring-updates (3.8.0-31.46), which contains the following patch:

commit bc421a035e6cc8420866754af35b355c165c6ed2
Author: Eric Dumazet <email address hidden>
Date: Mon Jul 29 10:24:04 2013 -0700

    atl1c: use custom skb allocator

    BugLink: http://bugs.launchpad.net/bugs/1221794

    [ Upstream commit 7b70176421993866e616f1cbc4d0dd4054f1bf78 ]

    We had reports ( https://bugzilla.kernel.org/show_bug.cgi?id=54021 )
    that using high order pages for skb allocations is problematic for atl1c

    We do not know exactly what the problem is, but we suspect that crossing
    4K pages is not well supported by this hardware.

    Use a custom allocator, using page allocator and 2K fragments for
    optimal stack behavior. We might make this allocator generic
    in future kernels.