Re: latest -git: BUG at fs/jfs/namei.c:512 assert(ip->i_nlink)

[Dave Kleikamp <shaggy@linux.vnet.ibm.com>]

On Fri, 2008-07-18 at 20:08 +0200, Vegard Nossum wrote:
> On Fri, Jul 18, 2008 at 7:29 PM, Dave Kleikamp
> <shaggy@linux.vnet.ibm.com> wrote:
> >
> > xtTruncate can call release_metapage() in several places.  If you still
> > have this kernel, could you please run addr2line against c02e4202?  If
> > you've rebuilt your kernel since then, I understand.
> 
> Sorry, it's gone. However, I just recompiled it and the size of the
> function is now 0x105f, and the offset 0xd42 matches with a call to
> release_metapage. So with this new address, it's this call:
> 
> $ addr2line -e vmlinux -i c02e428d
> fs/jfs/jfs_metapage.h:101
> fs/jfs/jfs_xtree.c:3723
> 
> ...which is this one:
> 
>     /* invalidate empty leaf page */
>     discard_metapage(mp);

Thanks.  This helps some, but I'm not sure why it should fail there.
Possibly a memory corruption problem caused by a bug in an earlier error
path.

> With your two patches applied, I now got this:
> 
> ERROR: (device loop0): remounting filesystem as read-only
> blkno = 200000687, nblocks = f
> ERROR: (device loop0): dbFree: block to be freed is outside the map
> BUG: unable to handle kernel paging request at 338f8038

I see I've got a few of these to track down.  I'm sure there are more
than a couple more bugs you will uncover.

> $ addr2line -e vmlinux -i c02ee352
> fs/jfs/jfs_dmap.c:2188
> 
> You can also have a copy of my scripts if you want to try to reproduce
> it locally. But I don't mind testing either :-)

Can you please send me a copy.  For some of these, I'd like to get a
repeatable test case, so I'm not debugging several problems at once.

Thanks,
Shaggy
-- 
David Kleikamp
IBM Linux Technology Center