XFS problem 2.6.7 vanilla

XFS problem 2.6.7 vanilla

[Fabio Coatti]

We are getting some error trace from xfs. I suppose that this can be due to a 
faulty HD sector, but it sound strange to me that a HD error can trigger an 
internal FS failure. We have tried several times to fix this error with XFS 
repair without succes, so I suppose a hw error, but is the aspected behaviour 
to get an internal FS error?
(2.6.7 vanilla)

Filesystem "hde2": corrupt inode 25375199 ((a)extents = 2034).  Unmount and 
run xfs_repair.
0x0: 49 4e 41 ed 01 02 00 03 00 00 07 f2 00 00 07 f2
Filesystem "hde2": XFS internal error xfs_iformat_extents(1) at line 678 of 
file fs/xfs/xfs_inode.c.  Caller 0xc029b5be
 [<c029bbb7>] xfs_iformat_extents+0x287/0x310
 [<c029b5be>] xfs_iformat+0x4ee/0x610
 [<c029b5be>] xfs_iformat+0x4ee/0x610
 [<c029b5be>] xfs_iformat+0x4ee/0x610
 [<c029c040>] xfs_xlate_dinode_core+0x160/0x810
 [<c029c99d>] xfs_iread+0x21d/0x270
 [<c0299c96>] xfs_iget_core+0xb6/0x4c0
 [<c029a1b9>] xfs_iget+0x119/0x150
 [<c02b6824>] xfs_dir_lookup_int+0xb4/0x130
 [<c02bc0e0>] xfs_lookup+0x50/0x90
 [<c02c82c7>] linvfs_lookup+0x67/0xa0
 [<c0151d2b>] real_lookup+0xcb/0xf0
 [<c0151f96>] do_lookup+0x96/0xb0
 [<c01523cf>] link_path_walk+0x41f/0x800
 [<c02f502e>] copy_to_user+0x3e/0x50
 [<c01529cf>] path_lookup+0x6f/0x150
 [<c0152c79>] __user_walk+0x49/0x80
 [<c014e4dc>] vfs_lstat+0x1c/0x60
 [<c014ebeb>] sys_lstat64+0x1b/0x40
 [<c0145810>] sys_close+0x50/0x60
 [<c0103efb>] syscall_call+0x7/0xb


-- 
Fabio "Cova" Coatti    http://members.ferrara.linux.it/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: XFS problem 2.6.7 vanilla

[Nathan Scott]

On Mon, Jul 05, 2004 at 10:33:10AM +0200, Fabio Coatti wrote:
> We are getting some error trace from xfs. I suppose that this can be due to a 
> faulty HD sector, but it sound strange to me that a HD error can trigger an 
> internal FS failure. We have tried several times to fix this error with XFS 
> repair without succes, so I suppose a hw error, but is the aspected behaviour 
> to get an internal FS error?
> (2.6.7 vanilla)

Could you try a current -bk tree (or XFS CVS on oss.sgi.com,
for 2.6.7 + XFS updates), this should be resolved there now.

thanks.

-- 
Nathan

Re: XFS problem 2.6.7 vanilla

[Fabio Coatti]

Alle 12:14, lunedì 5 luglio 2004, Nathan Scott ha scritto:
> On Mon, Jul 05, 2004 at 10:33:10AM +0200, Fabio Coatti wrote:
> > We are getting some error trace from xfs. I suppose that this can be due
> > to a faulty HD sector, but it sound strange to me that a HD error can
> > trigger an internal FS failure. We have tried several times to fix this
> > error with XFS repair without succes, so I suppose a hw error, but is the
> > aspected behaviour to get an internal FS error?
> > (2.6.7 vanilla)
>
> Could you try a current -bk tree (or XFS CVS on oss.sgi.com,
> for 2.6.7 + XFS updates), this should be resolved there now.


we aretrying right now bk18, after a xfs repair we don't see any error, so far 
so good. I'll keep pounding on FS and report back any issue.
Thanks for your help.


-- 
Fabio "Cova" Coatti    http://members.ferrara.linux.it/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.