Kernel Oops since the upgrade to kernel 3.5

[vlad59]

Hi,

I've been running a rock solid Dockstar with Archlinux and a Kernel 3.1. My max uptime was about 6 months, So it really was rock solid.

My main use are Nginx + Php and small test site, Samba share and Nzbget.

Yesterday I installed Kernel 3.5 (with pacman -S linux-kirkwood linux-headers-kirkwood). Everything went fine and I successfully rebooted into a 3.5 kernel.

This morning Nzbget was no longer working (but the process was here). After some test I finally found I had a Kernel Oops :

$this->bbcode_second_pass_code('', '
[root@minus var]# 2012 Sep 21 18:47:21 minus [ 2839.321737] Internal error: Oops: 817 [#1] PREEMPT ARM
2012 Sep 21 18:47:21 minus [ 2839.393061] Process nzbget (pid: 2101, stack limit = 0xc4582270)
2012 Sep 21 18:47:21 minus [ 2839.399087] Stack: (0xc4583c34 to 0xc4584000)
2012 Sep 21 18:47:21 minus [ 2839.403458] 3c20: 0009cb70 c4582000 00000400
2012 Sep 21 18:47:21 minus [ 2839.411676] 3c40: 00000000 0009cb70 00000400 c0272574 c0d78270 c7a70934 00000400 c4583ee4
2012 Sep 21 18:47:21 minus [ 2839.419890] 3c60: 00000400 00000400 c4bab06c c4582000 0000058c c03a1620 00000400 00000000
2012 Sep 21 18:47:21 minus [ 2839.428105] 3c80: 0000058c c7ab80c0 00000400 c03a2140 c7ab80c0 00000000 c4583edc c6ef9ae0
2012 Sep 21 18:47:21 minus [ 2839.436319] 3ca0: c7ab80c0 00000000 00000400 c4bab058 c6ef9b20 c6ef9e0c c4582000 c03e6380
2012 Sep 21 18:47:21 minus [ 2839.444534] 3cc0: c4583cf4 c7ab80c0 c7ab80c0 00000014 c4bab058 c6ef9ae0 c7ab80c0 00000014
2012 Sep 21 18:47:21 minus [ 2839.452748] 3ce0: c4bab058 00000000 c6ef9b20 c6ef9e0c c4582000 c03e97cc c6ef9ae0 00ab80c0
2012 Sep 21 18:47:21 minus [ 2839.460962] 3d00: c4bab058 c6ef9ae0 c7ab80c0 00000000 c6ef9fac 00000000 c6ef9e0c bf04dd38
2012 Sep 21 18:47:21 minus [ 2839.469175] 3d20: 00000000 00000000 00000000 c6ef9ae0 c6ef9ae0 00000000 c4583d94 c6ef9ae0
2012 Sep 21 18:47:21 minus [ 2839.477390] 3d40: 00000000 c6ef9e28 00000400 00000000 c6ef9e0c c03ddf88 c6ef9ae0 00000000
2012 Sep 21 18:47:21 minus [ 2839.485604] 3d60: c7bed500 c03e17f4 505c9a19 00001000 c4582000 00000000 00000400 00000000
2012 Sep 21 18:47:21 minus [ 2839.493819] 3d80: c4583ee4 c7bed500 00000001 00000000 c2ea4960 00001770 00000000 c4583ee4
2012 Sep 21 18:47:21 minus [ 2839.502033] 3da0: bf05ea9c 00000000 c4583ee4 00000000 c4582000 c579c6e0 b43fe8dc c03fe218
2012 Sep 21 18:47:21 minus [ 2839.510247] 3dc0: 00000000 00000000 c4583dd4 c003f4ec 00000000 00000000 00000000 c4583df0
2012 Sep 21 18:47:21 minus [ 2839.518461] 3de0: 00000400 c0396a44 00000000 c7991c00 c4583df0 c4583df0 00000000 00000400
2012 Sep 21 18:47:21 minus [ 2839.526676] 3e00: c579c6e0 c4583e30 00000000 c4583ee4 00000000 c4583ec8 c2ea4960 c024d580
2012 Sep 21 18:47:21 minus [ 2839.534889] 3e20: 01e98000 c008d8f0 00001000 00000000 91827364 c4583e34 00000000 00000001
2012 Sep 21 18:47:21 minus [ 2839.543104] 3e40: ffffffff 00000000 00000000 00000000 00000000 00000000 c7bed500 c4658088
2012 Sep 21 18:47:21 minus [ 2839.551318] 3e60: 00000000 00000000 c4583ec8 c0154380 c4583df0 00000000 c04420cc 00000000
2012 Sep 21 18:47:21 minus [ 2839.559533] 3e80: c7a1fe0c ffdfb1e0 00000000 c7a1fe0c ffdfa720 fffffdee 00000000 c4583f88
2012 Sep 21 18:47:21 minus [ 2839.567746] 3ea0: c6fc08c0 c00cab0c fffffff7 c4583ed4 c4583ed8 00000400 c579c6e0 00000000
2012 Sep 21 18:47:21 minus [ 2839.575961] 3ec0: 00000000 0009cb70 00000000 c0398480 00000000 fffffff7 00000001 0009cb70
2012 Sep 21 18:47:21 minus [ 2839.584175] 3ee0: 00000400 c4583f00 00000080 c4583edc 00000001 00000000 00000000 c0100c30
2012 Sep 21 18:47:21 minus [ 2839.592389] 3f00: 01e99000 00000000 00000000 00000002 00001000 00000000 00000000 00000000
2012 Sep 21 18:47:21 minus [ 2839.600603] 3f20: 00000000 00000000 c4582000 00000000 00000000 00000000 c4582000 c2ea4968
2012 Sep 21 18:47:21 minus [ 2839.608817] 3f40: 00000002 c4658088 00000000 00001000 c4582000 00000000 c2ea4960 00000000
2012 Sep 21 18:47:21 minus [ 2839.617032] 3f60: c4583f90 b43fe918 00000000 b43fe918 00000000 0000004e c0008dc4 b43fe918
2012 Sep 21 18:47:21 minus [ 2839.625246] 3f80: 00000000 ffffffff 00000000 00000123 c0008dc4 c03984e4 00000000 00000000
2012 Sep 21 18:47:21 minus [ 2839.633460] 3fa0: 00000400 c0008c40 00000000 ffffffff 00000008 0009cb70 00000400 00000000
2012 Sep 21 18:47:21 minus [ 2839.641675] 3fc0: 00000000 ffffffff 00000000 00000123 b5fcfd3c 00000000 b43fefa0 b43fe8dc
2012 Sep 21 18:47:21 minus [ 2839.649890] 3fe0: 00000000 b43fe8a8 b6be5a1c b6be6604 80000010 00000008 741f8c13 e61604c6
2012 Sep 21 18:47:21 minus [ 2839.658106] Code: e1a00000 e4803004 e4804004 e4805004 (e4806004)
')

Each time I restart nzbget I have the problem (~ 1/2 hour later).

I really don't know what I can do, For now I'll try to avoid running Nzbget and see if the problem appear anyway.

Thanks in advance for your help.

EDIT : I don't know if it's important my root partition is still ext2 and the disk attached is 2.5" through USB (the same for the last year). I tested the disk on my laptop and it has no error.

[WarheadsSE]

This oops is a memory alignment issue. It's a nasty bugger to hunt down.

[moonman]

If it is memory allignment issue, maybe this will help:
$this->bbcode_second_pass_code('', ' echo 2 >/proc/cpu/alignment ')

[antipasta]

Ack, I'm running the same kernel version and I'm having the same issue w/ nzbget. Will see if the alignment fix helps any

Edit: Above command didn't seem to help. Downgraded to 3.5.3 and that exhibited similar strangeness with nzbget. Finally downgraded to 3.1 and now it's back to normal.

[vlad59]

@antipasta

Thanks for your answer. It feels good not to be alone in this.

@all

Also tried the command without success.

I removed official nzbget package and instead compiled NZBGet 9.0-r455 from sources. For now it seems fine but I'll post again after a few days. I hope it'll be stable again or I'll have to go back to 3.1 again.

Thanks for your answers.

[moonman]

You may wanna try this latest version I just compiled
$this->bbcode_second_pass_code('', 'pacman -U http://music4dance.mine.nu/Downloads/ArchLinuxArm/nzbget-9.0r455-0-arm.pkg.tar.xz')

[antipasta]

I still had issues with both the above package and building from source and kernel 3.5.4. Sticking with 3.1 for now I guess.

[vlad59]

Hi Antipasta,

Did you compile nzbget 8 or 9.0 ?
Did you use nzbgetweb ?

For now it seems fine for me. I'll add a big nzb tonight to stress it a little.

[antipasta]

I built the .9 release, and didnt build nzbgetweb because now there's a built in web server in the .9 release.

With the 3.5 kernel and the release I built from source I was experiencing strange issues like the client suddenly dying while running 'nzbget -C', and after the client died I wouldnt be able to relaunch the client, it would just hang.

Moved back to 3.1 and no longer have this issue.

[vlad59]

Hi,

Just got another kernel oops. So I'll be going back to 3.1 for now