Nilfs users (Log-Structured Snapshotting Filesystem)

Slicky Devil | 6 Mar 2012 14:24

Re: BUG: unable to handle kernel NULL pointer dereference at 00000048

Luckily, I seem to have been able to reproduce the bug (after
shrinking the partition down to some hundred MB). Your patch seems to
have fixed it!

Now I get the following, which, I think, is pretty much expected,
since I chopped a large part of the filesystem off:

15:52 root ~ # mount /dev/1/buggy-home test
[  461.137823] NILFS: error searching super root.
mount.nilfs2: Error while mounting /dev/mapper/1-buggy--home on
/root/test: Input/output error

Please, fix the bug in the official kernel sources, too. Your
filesystem proved very useful here for storing documents and other
changeable stuff -- we don't want it get broken unexpectedly. :)

Still, thanks a lot! The things you are doing here are really cool!

Cheers!

On Mon, Mar 5, 2012 at 5:30 PM, Ryusuke Konishi
<konishi.ryusuke@...> wrote:
>
> Thank you for reporting this issue.
>
> I found a bug in the nilfs_load_super_block function which has
> potential to cause this oops.
>
> Could you try the following patch if you still have the partition ?
>

(Continue reading)

headers

Ryusuke Konishi | 6 Mar 2012 15:54

Re: BUG: unable to handle kernel NULL pointer dereference at 00000048

Hi,
On Tue, 6 Mar 2012 16:24:40 +0300, Slicky Devil wrote:
> Luckily, I seem to have been able to reproduce the bug (after
> shrinking the partition down to some hundred MB). Your patch seems to
> have fixed it!
> 
> Now I get the following, which, I think, is pretty much expected,
> since I chopped a large part of the filesystem off:
> 
> 15:52 root ~ # mount /dev/1/buggy-home test
> [  461.137823] NILFS: error searching super root.
> mount.nilfs2: Error while mounting /dev/mapper/1-buggy--home on
> /root/test: Input/output error

Hmmm.  Did you shrink the filesystem with nilfs-resize tool
before shrinking the partition ?

This error (I/O error) also looks undesirable to me.

 
> Please, fix the bug in the official kernel sources, too. Your
> filesystem proved very useful here for storing documents and other
> changeable stuff -- we don't want it get broken unexpectedly. :)

Sure, I will send the fix upstream (and stable trees).

Thanks,
Ryusuke Konishi

> Still, thanks a lot! The things you are doing here are really cool!

(Continue reading)

headers

Slicky Devil | 6 Mar 2012 22:07

Re: BUG: unable to handle kernel NULL pointer dereference at 00000048

On Tue, Mar 6, 2012 at 5:54 PM, Ryusuke Konishi
<konishi.ryusuke@...> wrote:
> Hmmm.  Did you shrink the filesystem with nilfs-resize tool
> before shrinking the partition ?
>
> This error (I/O error) also looks undesirable to me.

I couldn't shrink the filesystem, because I couldn't mount it because
of the bug. The filesystem was large and I was in desparate need of
some spare diskspace. At first, I was going to completely get rid of
it, but then I thought you might want to have a look at the
superblock, or some othe block, or whatever else to analyze the bug.
Since the bug was some sort of initialization issue, I finally decided
to keep the first 500 MB of the filesystem and reclaimed the rest of
it (by shrinking the partition).

And all that had happened *before* I posted the bug report on this mailing list.

Luckily, when I tried to mount the "broken" filesystem today, the
nilfs driver exposed the very same oops again. I applied your
patch/recompiled the kernel and the oops disappeared.

So, no. I didn't use any nilfs-resize. And those IO error are pretty
much expected to me.
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to majordomo@...
More majordomo info at  http://vger.kernel.org/majordomo-info.html

(Continue reading)

headers

Ryusuke Konishi | 10 Mar 2012 04:08

Re: BUG: unable to handle kernel NULL pointer dereference at 00000048

On Wed, 7 Mar 2012 00:07:24 +0300, Slicky Devil wrote:
> On Tue, Mar 6, 2012 at 5:54 PM, Ryusuke Konishi
> <konishi.ryusuke@...> wrote:
> > Hmmm.  Did you shrink the filesystem with nilfs-resize tool
> > before shrinking the partition ?
> >
> > This error (I/O error) also looks undesirable to me.
> 
> I couldn't shrink the filesystem, because I couldn't mount it because
> of the bug. The filesystem was large and I was in desparate need of
> some spare diskspace. At first, I was going to completely get rid of
> it, but then I thought you might want to have a look at the
> superblock, or some othe block, or whatever else to analyze the bug.
> Since the bug was some sort of initialization issue, I finally decided
> to keep the first 500 MB of the filesystem and reclaimed the rest of
> it (by shrinking the partition).
> 
> And all that had happened *before* I posted the bug report on this mailing list.
> 
> Luckily, when I tried to mount the "broken" filesystem today, the
> nilfs driver exposed the very same oops again. I applied your
> patch/recompiled the kernel and the oops disappeared.
> 
> So, no. I didn't use any nilfs-resize. And those IO error are pretty
> much expected to me.

Understood, thank you for letting me know the situation.

Regards,
Ryusuke Konishi

(Continue reading)