Kernel BUG

Kernel BUG

[Rudy Zijlstra]

Hi,

I have a system where whenever i trigger a ongoing consistent NFS load i 
get the following kernel BUG:

----------
kernel BUG at mm/truncate.c:311!
invalid opcode: 0000 [#1]
SMP
Modules linked in: cx88_alsa cx88_dvb cx88_vp3054_i2c or51132 
video_buf_dvb isl6421 zl10353
cx24123 cx22702 cx88_blackbird cx8802 cx2341x cx8800 cx88xx ir_common 
i2c_algo_bit tveeprom
compat_ioctl32 btcx_risc i2c_piix4 b2c2_flexcop_pci b2c2_flexcop mt352 
mt312 bcm3510 stv0297
 nxt200x lgdt330x budget_av dvb_pll saa7146_vv video_buf videodev 
v4l1_compat v4l2_common tu
a6100 budget_core saa7146 ttpci_eeprom tda10021 tda1004x stv0299 
dvb_core pcnet32
CPU:    1
EIP:    0060:[<c01489f0>]    Not tainted VLI
EFLAGS: 00010002   (2.6.19.1 #1)
EIP is at invalidate_complete_page2+0x45/0x84
eax: 8000082d   ebx: c1171e00   ecx: 00000000   edx: 00000000
esi: d2299db4   edi: 000b6096   ebp: 00000000   esp: d3f53db4
ds: 007b   es: 007b   ss: 0068
Process mythbackend (pid: 3584, ti=d3f52000 task=d5400550 task.ti=d3f52000)
Stack: c1171e00 000000d0 c1171e00 00000000 c0148b97 d2299db4 c1171e00 
000b6095
       0000000e 00000001 ffffffff 00000000 00000000 00000000 0000000e 
00000000
       c1171e00 c1171e20 c11a1e40 c11a1e60 c1245c00 c1245c20 c12694c0 
c12694e0
Call Trace:
 [<c0148b97>] invalidate_inode_pages2_range+0x168/0x264
 [<c0148cb2>] invalidate_inode_pages2+0x1f/0x25
 [<c01fa9ab>] nfs_revalidate_mapping+0x98/0x143
 [<c01632c4>] pipe_write+0x44d/0x459
 [<c01f9206>] nfs_file_read+0x91/0xf0
 [<c015d386>] do_sync_read+0xdd/0x11a
 [<c012daca>] autoremove_wake_function+0x0/0x4b
 [<c0398b12>] sock_ioctl+0x1b1/0x1d1
 [<c015d463>] vfs_read+0xa0/0x16b
 [<c015d7fe>] sys_read+0x4b/0x71
 [<c0102bc3>] syscall_call+0x7/0xb
 =======================
Code: c7 44 24 04 d0 00 00 00 89 1c 24 e8 1b b7 ff ff 31 d2 85 c0 74 4d 
8d 46 10 e8 4e fe 2b
 00 8b 03 a8 10 75 34 8b 03 f6 c4 08 74 08 <0f> 0b 37 01 07 b0 43 c0 89 
1c 24 e8 b0 86 ff ff
 f0 81 46 10 00
EIP: [<c01489f0>] invalidate_complete_page2+0x45/0x84 SS:ESP 0068:d3f53db4

---------

Captured using serial console.

Anybody an idea?

Its on a dual processor Netfinity 5000. Two PIII 500 processors, 512MB 
memory.

Cheers,

Rudy

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Kernel BUG

[Peter Zijlstra]

On Fri, 2006-12-22 at 12:59 +0100, Rudy Zijlstra wrote:
> Hi,
> 
> I have a system where whenever i trigger a ongoing consistent NFS load i 
> get the following kernel BUG:
> 
> ----------
> kernel BUG at mm/truncate.c:311!

Lotsa changes there:
try these:
 http://client.linux-nfs.org/Linux-2.6.x/2.6.19/linux-2.6.19-NFS_ALL.dif

And:
  http://lkml.org/lkml/2006/12/19/279




-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Kernel BUG

[Rudy Zijlstra]

Peter Zijlstra wrote:

>On Fri, 2006-12-22 at 12:59 +0100, Rudy Zijlstra wrote:
>  
>
>>Hi,
>>
>>I have a system where whenever i trigger a ongoing consistent NFS load i 
>>get the following kernel BUG:
>>
>>----------
>>kernel BUG at mm/truncate.c:311!
>>    
>>
>
>Lotsa changes there:
>try these:
> http://client.linux-nfs.org/Linux-2.6.x/2.6.19/linux-2.6.19-NFS_ALL.dif
>
>And:
>  http://lkml.org/lkml/2006/12/19/279
>
>  
>
Thanks, i have now a 2.6.19.1 patched with the above NFS_ALL patchset 
running under load since about 17:00 and still going strong. So seems to 
have solved the problem!

Also, performance seems to have significantly increased. I am testing 
this with MythTv with the storage over NFS. With HD LiveTV (which  is 
what i use to test this), this translates into about 43Mbps total load, 
from almost 15Mbps video stream which is present 3x on the ethernet:
1/ from card to NFS disk
2/ read from NFS disk
3/ stream to viewer app.

Doing this with HD streams before the patch set resulted in regular 
picture hickups. No hickups visible anymore (that i have seen, i have 
not been present all this time).

Regards,

Rudy

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs