Error whilst running "tune2fs -j"

Error whilst running "tune2fs -j"

[Martin J. Bligh]

Anyone see this before? This was 2.5.65-mjb2 running on my laptop,
at the start of a "tune2fs -j" ....

buffer layer error at fs/buffer.c:395
Pass this trace through ksymoops for reporting
Call Trace: [__buffer_error+51/56]  [__find_get_block_slow+126/176]  [unmap_underlying_metadata+18/72]  [__block_prepare_write+374/
1056]  [__block_commit_write+114/156]  [block_prepare_write+33/56]  [ext2_get_bl
ock+0/744]  [ext2_prepare_write+25/32]  [ext2_get_block+0/744]  [generic_file_ai
o_write_nolock+1527/2536]  [generic_commit_write+51/92]  [unlock_page+10/60]  [g
eneric_file_write_nolock+111/140]  [do_lookup+24/148]  [link_path_walk+1499/1884
]  [permission+43/56]  [may_open+99/376]  [open_namei+672/944]  [generic_file_wr
ite+72/96]  [vfs_write+165/268]  [sys_write+42/60]  [syscall_call+7/11] 

Re: Error whilst running "tune2fs -j"

[Andrew Morton]

"Martin J. Bligh" <mbligh@aracnet.com> wrote:
>
> Anyone see this before? This was 2.5.65-mjb2 running on my laptop,
> at the start of a "tune2fs -j" ....
> 
> buffer layer error at fs/buffer.c:395

grrr.  This means that the block_dev layer somehow managed to get the
filesystem softblocksize confused.  There is something lurking in there.  As
usual, if I could reproduce it I could fix it.

Was the fs mounted at the time?

What version of e2fsprogs?

Can you reproduce it?

Thanks.

Re: Error whilst running "tune2fs -j"

[Martin J. Bligh]

>> Anyone see this before? This was 2.5.65-mjb2 running on my laptop,
>> at the start of a "tune2fs -j" ....
>> 
>> buffer layer error at fs/buffer.c:395
> 
> grrr.  This means that the block_dev layer somehow managed to get the
> filesystem softblocksize confused.  There is something lurking in there.  As
> usual, if I could reproduce it I could fix it.
> 
> Was the fs mounted at the time?

Yes, was my root fs.
 
> What version of e2fsprogs?

Standard debian woody:

mbligh@flay:~/tmp$ apt-cache show e2fsprogs
Package: e2fsprogs
Essential: yes
Priority: required
Section: base
Installed-Size: 832
Maintainer: Yann Dirson <dirson@debian.org>
Architecture: i386
Version: 1.27-2
Provides: libcomerr2, libss2, libext2fs2, libe2p2, libuuid1
Pre-Depends: libc6 (>= 2.2.4-4)
Suggests: gpart, parted, e2fsck-static
Conflicts: e2fslibsg, dump (<< 0.4b4-4), quota (<< 1.55-8.1)
Filename: pool/main/e/e2fsprogs/e2fsprogs_1.27-2_i386.deb
Size: 335402

> Can you reproduce it?

Not easily, I fear ... since it's a pain in the butt to remove an ext3
journal on the root fs once mounted. I suppose I could boot from CD
or something ... will try to recreate on a less valuable box over
the weekend ;-)

I had been having some other trouble with the fs (power cycled a couple
of times for silly reasons), but it had just ext2 fscked. I suppose there
*might* have been some corruption, but seems unlikely.

M.

Re: Error whilst running "tune2fs -j"

[Andrew Morton]

"Martin J. Bligh" <mbligh@aracnet.com> wrote:
>
> > Can you reproduce it?
> 
> Not easily, I fear ... since it's a pain in the butt to remove an ext3
> journal on the root fs once mounted. I suppose I could boot from CD
> or something ... will try to recreate on a less valuable box over
> the weekend ;-)
> 
> I had been having some other trouble with the fs (power cycled a couple
> of times for silly reasons), but it had just ext2 fscked. I suppose there
> *might* have been some corruption, but seems unlikely.
> 

OK.  Please add the below to your patchset.  I've had this in -mm for *ages*,
precisely because this problem was reported a single time, maybe four months
ago.

I have not had a report of it triggering since then.  I'd like to know what
block number it was.

diff -puN fs/buffer.c~buffer-debug fs/buffer.c
--- 25/fs/buffer.c~buffer-debug	2003-04-02 22:24:27.000000000 -0800
+++ 25-akpm/fs/buffer.c	2003-04-02 22:24:27.000000000 -0800
@@ -397,6 +397,9 @@ __find_get_block_slow(struct block_devic
 		bh = bh->b_this_page;
 	} while (bh != head);
 	buffer_error();
+	printk("block=%llu, b_blocknr=%llu\n",
+		(unsigned long long)block, (unsigned long long)bh->b_blocknr);
+	printk("b_state=0x%08lx, b_size=%u\n", bh->b_state, bh->b_size);
 out_unlock:
 	spin_unlock(&bd_mapping->private_lock);
 	page_cache_release(page);

_