Still Kernel Bug message and client hang

[Hans-Gunther Borrmann]

Hello,

I still see the following Kernel Bug message in /var/log/messages. Thereafter
any process trying to access a file, local or not, hangs The machine has to
be reset.

Kernel BUG at "fs/inode.c":1094
invalid operand: 0000 [1] SMP
CPU 1
ide_cd cdrom
Pid: 5316, comm: php.exe Tainted: P      2.6.12
RIP: 0010:[<ffffffff8018f77e>] <ffffffff8018f77e>{iput+30}
RSP: 0018:ffff810062db3dc8  EFLAGS: 00010246
RAX: ffffffff880c6540 RBX: ffff8100f778d800 RCX: 0000000000000000
RDX: ffff81007951ae90 RSI: ffff810062db3d58 RDI: ffff8100f778d800
RBP: ffff8100f778d800 R08: 0000000000000000 R09: ffff810062db3d58
R10: 000000000000000b R11: ffffffff8809df50 R12: ffff8100f309ecd0
R13: ffffffff880c5a00 R14: 00007fffffc17e30 R15: 00007fffffc1b210
FS:  00000000008f4ae0(0063) GS:ffffffff804627c0(0000) knlGS:0000000008208760
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 000000000051fa10 CR3: 00000000dbe9e000 CR4: 00000000000006e0
Process php.exe (pid: 5316, threadinfo ffff810062db2000, task
ffff81007951ae90)
Stack: ffffffff880c5a00 ffffffff880660a1 ffff8100f7eb4400 ffff8100f7eb4400
ffffffff8809e0d5 ffff810062db3e68 ffff810062db3ef8
0000000000000000 0000000000000296
Call Trace:<ffffffff880660a1>{:libafs:afs_PutVCache+161}
<ffffffff8809e0d5>{:libafs:afs_linux_getattr+389}
<ffffffff8017ed1d>{vfs_lstat+61} <ffffffff8017f13f>{sys_newlstat+31}
<ffffffff8010da46>{system_call+126}
Code: 0f 0b 40 1e 32 80 ff ff ff ff 46 04 66 66 90 66 66 90 48 85
RIP <ffffffff8018f77e>{iput+30} RSP <ffff810062db3dc8>

Hardware: AMD Dual Opteron
Operating system: Gentoo Linux
Kernel: vanilla kernel 2.12.6
OpenAFS 1.3.84
cacheinfo: /afs:/usr/vice/cache:1651000
real cache size: 2064208K
afsd parameters: "-fakestat -stat 10000 -dcache 4000 -daemons 5 -volumes 256
-files 50000"

This problem is urgent for me because I have to migrate our central webservice
to this machine. All our  web pages are  in AFS. So any help is appreciated.

Thanks in advance
Gunther
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Derrick Brashear-4]

So dsid you mention openafs version and i just didn't see it? if the
version was less than 1.3.85, well, upgrade.

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Hans-Gunther Borrmann]

On Wednesday 13 July 2005 16:20, Derrick J Brashear wrote:
> So dsid you mention openafs version and i just didn't see it?

Yes. It is 1.3.84

> if the version was less than 1.3.85, well, upgrade.
>

I' ll do so as quick as possible.

Thanks!
Gunther
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[David Botsch]

We've also been seeing this behavior in 1.2.13 ... is the fix for this
something that can be easily backported into the 1.2 tree?

Thanks.

On Wed, Jul 13, 2005 at 04:32:12PM +0200, Hans-Gunther Borrmann wrote:

> On Wednesday 13 July 2005 16:20, Derrick J Brashear wrote:
> > So dsid you mention openafs version and i just didn't see it?
>
> Yes. It is 1.3.84
>
> > if the version was less than 1.3.85, well, upgrade.
> >
>
> I' ll do so as quick as possible.
>
> Thanks!
> Gunther
> --
> ________________________________________________________________
> Hans-Gunther Borrmann <hans-gunther.borrmann@...>
> Rechenzentrum der Universitaet Freiburg
> Hermann-Herder-Str. 10, D79104 FREIBURG
> Tel.: [phone]
> Fax:  [phone]
>
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@...
> https://lists.openafs.org/mailman/listinfo/openafs-info

--
********************************
David William Botsch
Consultant/Advisor II
CCMR Computing Facility
dwb7@...
********************************
_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Derrick Brashear-4]

On Wed, 13 Jul 2005, David Botsch wrote:

> We've also been seeing this behavior in 1.2.13 ... is the fix for this
> something that can be easily backported into the 1.2 tree?

Well, given the number of things fixed in 1.3.x recently, I'll go with no.
If 1.3.85 proves as stable as we believe and hope it will, it will form
the basis of the 1.4.0-rc1 in the next week anyway.

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[chas williams - CONTRACTOR]

is this 1.3.84?

In message <200507131615.33428.hans-gunther.borrmann@...>,Hans-G
unther Borrmann writes:
>Kernel BUG at "fs/inode.c":1094
>invalid operand: 0000 [1] SMP
>CPU 1
_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Hans-Gunther Borrmann]

On Wednesday 13 July 2005 16:52, chas williams - CONTRACTOR wrote:
> is this 1.3.84?
>

Yes. I'm just installing 1.3.85 as Derrick advised.
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Hans-Gunther Borrmann]

On Wednesday 13 July 2005 16:20, Derrick J Brashear wrote:
> So dsid you mention openafs version and i just didn't see it? if the
> version was less than 1.3.85, well, upgrade.
>

I upgraded to 1.3.85. Now /var/log/messages very quickly fills up with the
following message:

check_bad_parent(www): bad parent vcp->mvid->Fid.Volume != pvc->fid.Fid.Volume

Gunther
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[chas williams - CONTRACTOR]

that's a stupid debugging message i left in the code.  sorry.  it
can safely be ignored.  and removed from the source.

In message <200507131729.27831.hans-gunther.borrmann@...>,Hans-G
unther Borrmann writes:

>On Wednesday 13 July 2005 16:20, Derrick J Brashear wrote:
>> So dsid you mention openafs version and i just didn't see it? if the
>> version was less than 1.3.85, well, upgrade.
>>
>
>I upgraded to 1.3.85. Now /var/log/messages very quickly fills up with the
>following message:
>
>check_bad_parent(www): bad parent vcp->mvid->Fid.Volume != pvc->fid.Fid.Volume
>
>Gunther
>--
>________________________________________________________________
>Hans-Gunther Borrmann <hans-gunther.borrmann@...>
>Rechenzentrum der Universitaet Freiburg
>Hermann-Herder-Str. 10, D79104 FREIBURG
>Tel.: [phone]
>Fax:  [phone]
>
>_______________________________________________
>OpenAFS-info mailing list
>OpenAFS-info@...
>https://lists.openafs.org/mailman/listinfo/openafs-info
>

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Hans-Gunther Borrmann]

On Wednesday 13 July 2005 17:43, chas williams - CONTRACTOR wrote:
> that's a stupid debugging message i left in the code.  sorry.  it
> can safely be ignored.  and removed from the source.
>

Where is the message? I cannot let afs run because my /var fills up.

Gunther
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Derrick Brashear-4]

On Wed, 13 Jul 2005, Hans-Gunther Borrmann wrote:

> On Wednesday 13 July 2005 17:43, chas williams - CONTRACTOR wrote:
>> that's a stupid debugging message i left in the code.  sorry.  it
>> can safely be ignored.  and removed from the source.
>>
>
> Where is the message? I cannot let afs run because my /var fills up.

Dude, grep was probably invented before I was born.

It's in src/afs/LINUX/osi_vnodeops.c

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info

[Hans-Gunther Borrmann]

On Wednesday 13 July 2005 16:49, Derrick J Brashear wrote:
> If 1.3.85 proves as stable as we believe and hope it will, it will form
> the basis of the 1.4.0-rc1 in the next week anyway.

Just for feed back. I am runing now 1.3.85 for nearly 6 days on the university
webserver without any problems. Thanks to all who develope OpenAFS.
--
________________________________________________________________
Hans-Gunther Borrmann <hans-gunther.borrmann@...>
Rechenzentrum der Universitaet Freiburg
Hermann-Herder-Str. 10, D79104 FREIBURG
Tel.: [phone]
Fax:  [phone]

_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@...
https://lists.openafs.org/mailman/listinfo/openafs-info