Last time I saw this, it was on a >1TB filesystem, and the inode numbers
had exceeded 32 bits (bad thing...) - mkfs.xfs has since changed to try
to avoid this by choosing bigger inode sizes on large filesystems, which
limits inode numbers (note: explanation of tricky XFS internals deleted)
. This should only matter, though, on ~>1TB filesystems.
If you can send the output of mkfs.xfs (or xfs_info on a mounted
filesystem), I can make sure the automatic inode sizing hasn't gone awry
on your 19G filesystem....
And I'll stress again that moving back to pre-2.4.10 would also be a big
help in figuring out where the problem is. My confidence in 2.4.11/12 is
not so great, there's just been no exposure yet.
Nigel Kukard wrote:
> xfs_create looping, dir ino 0x80, ino 0x6d09, ide0(3,3)
Eric Sandeen XFS for Linux http://oss.sgi.com/projects/xfs
sandeen@xxxxxxx SGI, Inc.