[linux-lvm] Kernel 2.4.12 bug?

[linux-lvm] Kernel 2.4.12 bug?

[GCS]

Hi!

Is there any known issues with kernel 2.4.12? My Debian Woody is complaining
about lvm during startup, and drops an oops when swapp is turned off, and lvm
is going to be deactivated. It is a paging error, eip seems to be trashed to
0x5a5a5a5a.

GCS

Re: [linux-lvm] Kernel 2.4.12 bug?

[Patrick Caulfield]

On Sat, Oct 20, 2001 at 11:13:57AM +0200, GCS wrote:
> Hi!
> 
> Is there any known issues with kernel 2.4.12? My Debian Woody is complaining
> about lvm during startup, and drops an oops when swapp is turned off, and lvm
> is going to be deactivated. It is a paging error, eip seems to be trashed to
> 0x5a5a5a5a.
 
Not that I've seen. I'm running 2.4.12 ATM quite happily, LVM is a module rather
than in the kernel though.
 

patrick

Re: [linux-lvm] Kernel 2.4.12 bug?

[GCS]

Hi,

On Mon, Oct 22, 2001 at 09:48:33AM +0100, Patrick Caulfield <caulfield@sistina.com> wrote:
> Not that I've seen. I'm running 2.4.12 ATM quite happily, LVM is a module rather
> than in the kernel though.
 My kernel is still complaining during vgscan, see the syslog:
lvm -- Module successfully initialized
invalidate: busy buffer
last message repeated 8 times

Also at reboot/halt I get:
 printing eip:
c014c022
Oops: 0000
CPU:    0
EIP:    0010:[proc_match+18/56]    Tainted: P
EFLAGS: 00010206
eax: d9fc2609   ebx: 03   ecx: 03   edx: 5a5a5a5a
esi: d9fc2609   edi: d9fc260d   ebp: d65f7d18   esp: d65f7d10
ds: 0018   es: 0018   ss: 0018
Process vgchange (pid: 345, stackpage=d65f7000)
etc...

Eip points inside c014bf4c t meminfo_read_proc. I have even patched it to
2.4.13pre3, but I get the same result. What should I check out? With 2.4.9
everything works normaly.

Cheers, GCS

Re: [linux-lvm] Kernel 2.4.12 bug?

[Patrick Caulfield]

On Wed, Oct 24, 2001 at 08:21:46AM +0200, GCS wrote:
> Hi,
> 
> On Mon, Oct 22, 2001 at 09:48:33AM +0100, Patrick Caulfield <caulfield@sistina.com> wrote:
> > Not that I've seen. I'm running 2.4.12 ATM quite happily, LVM is a module rather
> > than in the kernel though.
>  My kernel is still complaining during vgscan, see the syslog:

Are you swapping onto the LV ? It seems that swapoff -a does not remove the LV
from the list of swap partitions, so vgchange -an would either fail or pull the
rug out from underneath the paging mechanism.

patrick

Re: [linux-lvm] Kernel 2.4.12 bug?

[GCS]

On Wed, Oct 24, 2001 at 09:00:09AM +0100, Patrick Caulfield <caulfield@sistina.com> wrote:
> Are you swapping onto the LV ? It seems that swapoff -a does not remove the LV
> from the list of swap partitions, so vgchange -an would either fail or pull the
> rug out from underneath the paging mechanism.
 Not at all. LVM is on a 61.5Gb IBM disk, and my swapp is on a standard
partition, on an other disk. Ok, it happens after I get 'swapp is turned off'
message on the console. Thus if it is related to swapp, then it is a
software bug somewhere, as my swapp is nothing to do with LVM. If you want,
I can try my config without swapp. I mean remove it from fstab, reboot and see.
Also, do you have an idea for the boot time problem: buffer is busy message?
Thanks,
GCS

Re: [linux-lvm] Kernel 2.4.12 bug?

[Patrick Caulfield]

On Wed, Oct 24, 2001 at 12:11:40PM +0200, GCS wrote:
> On Wed, Oct 24, 2001 at 09:00:09AM +0100, Patrick Caulfield <caulfield@sistina.com> wrote:
> > Are you swapping onto the LV ? It seems that swapoff -a does not remove the LV
> > from the list of swap partitions, so vgchange -an would either fail or pull the
> > rug out from underneath the paging mechanism.
>  Not at all. LVM is on a 61.5Gb IBM disk, and my swapp is on a standard
> partition, on an other disk. Ok, it happens after I get 'swapp is turned off'
> message on the console. Thus if it is related to swapp, then it is a
> software bug somewhere, as my swapp is nothing to do with LVM. If you want,
> I can try my config without swapp. I mean remove it from fstab, reboot and see.

It was just a thought. I'd like to see more of the ksymoops output - I'm also
worried by the "Tainted" flag in the extract you posted last time.


> Also, do you have an idea for the boot time problem: buffer is busy message?

I know it's harmless....


patrick

Re: [linux-lvm] Kernel 2.4.12 bug?

[GCS]

On Wed, Oct 24, 2001 at 11:37:47AM +0100, Patrick Caulfield <caulfield@sistina.com> wrote:
> It was just a thought. I'd like to see more of the ksymoops output - I'm also
> worried by the "Tainted" flag in the extract you posted last time.
 Ok, I will get a new one, as 2.4.13 is out. I cross my fingers to have this
bug fixed (if it is kernel related even). The tainted flag is set because the
vfat module I think. Kernel says: no license, will taint the kernel (or
similar to this).

GCS

Re: [linux-lvm] Kernel 2.4.12 bug?

[Patrick Caulfield]

On Wed, Oct 24, 2001 at 03:15:03PM +0200, GCS wrote:
> The tainted flag is set because the
> vfat module I think. Kernel says: no license, will taint the kernel (or
> similar to this).

Aha, I'd heard there were a few of those :-(

patrick

RE: [linux-lvm] Kernel 2.4.12 bug?

[Jean-Eric Cuendet]

> Also at reboot/halt I get:
>  printing eip:
> c014c022
> Oops: 0000
> CPU:    0
> EIP:    0010:[proc_match+18/56]    Tainted: P
> EFLAGS: 00010206
> eax: d9fc2609   ebx: 03   ecx: 03   edx: 5a5a5a5a
> esi: d9fc2609   edi: d9fc260d   ebp: d65f7d18   esp: d65f7d10
> ds: 0018   es: 0018   ss: 0018
> Process vgchange (pid: 345, stackpage=d65f7000)
> etc...

I also have problems with 2.4.12 + LVM rc4. I have a kernel panic with a
"stack trace" in the kernel *before* it mounts ROOT.
Any idea? Everything works fine with 2.4.7 / 2.4.9

-jec

Re: [linux-lvm] Kernel 2.4.12 bug?

[GCS]

On Wed, Oct 24, 2001 at 09:20:26AM +0200, Jean-Eric Cuendet <Jean-Eric.Cuendet@linkvest.com> wrote:
> I also have problems with 2.4.12 + LVM rc4. I have a kernel panic with a
> "stack trace" in the kernel *before* it mounts ROOT.
> Any idea? Everything works fine with 2.4.7 / 2.4.9
 I am using rc2. Anyway, try with 2.4.13. Ooops are gone, only 'buffer is
busy' remains, but as Patrik (?) wrote, this is harmless.

GCS

RE: [linux-lvm] Kernel 2.4.12 bug?

[Jean-Eric Cuendet]

> > The tainted flag is set because the
> > vfat module I think. Kernel says: no license, will taint 
> the kernel (or
> > similar to this).
> 
> Aha, I'd heard there were a few of those :-(

Could you please explain in more details what are those license things
and taint things?
Thanks
-jec