xfs: xfs_swap_extents needs to handle dynamic fork offsets

Saw this description from a patch set to lkml:
( http://lkml.indiana.edu/hypermail/linux/kernel/1004.2/02461.html )
========================================================
From: Greg KH
Date: Thu Apr 22 2010 - 16:43:40 EST
Subj: [022/197] xfs: xfs_swap_extents needs to handle dynamic fork offsets

2.6.32-stable review patch. If anyone has any objections, please let us know.
------------------
From: Dave Chinner <david@xxxxxxxxxxxxx>
commit e09f98606dcc156de1146c209d45a0d6d5f51c3f upstream
=========================================================
Description says...(brackets, [], added to highlight/break apart some sections):

...
Changing the way xfs_fsr sets up the attribute fork will prevent this situation from ever occurring, so in the kernel code we can get by with a preventative fix - check that the data fork in the defragmented inode is in a format valid for the inode it is being swapped into. [
]
This will lead to files that will silently and potentially repeatedly fail defragmentation, [
***] so _issue_ _a_ _warning_ to the log when this particular failure occurs to let us know that [
***]_xfs_fsr_ _needs_ _updating/fixing_.[***]

"To help identify how to improve xfs_fsr to avoid this issue, add trace points for the inodes being swapped so that we can determine why the swap was rejected and to confirm that the code is making the right decisions and modifications when swapping forks. "
....
=========================================================
lower in the code you added the message:
=========================================================
+ /* check inode formats now that data is flushed */
+ error = xfs_swap_extents_check_format(ip, tip);
+ if (error) {
+ xfs_fs_cmn_err(CE_NOTE, mp,
+ "%s: inode 0x%llx format is incompatible for exchanging.",
+ __FILE__, ip->i_ino);
goto out_unlock;
}
=========================================================

I'm seeing several of these in 'dmesg':

[873569.328221] Filesystem dm-6: fs/xfs/xfs_dfrag.c: inode 0x8227a310 format is incompatible for exchanging.
[873623.752724] Filesystem dm-6: fs/xfs/xfs_dfrag.c: inode 0x8227a310 format is incompatible for exchanging.
[873669.721127] Filesystem dm-6: fs/xfs/xfs_dfrag.c: inode 0x8227a310 format is incompatible for exchanging.

...

----
am running 2.6.38.3,

& version 3.1.4-2.1 of 'xfsprogs'....

Is there a version of xfs_fsr that fixes these problems, or is this this still something on the 'todo list' (I don't notice any other aberrant behaviors that might have anything to do with this)...Just thought I'd ask about the weird messages...

Thanks,
Linda

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

That's no surprise - there are still some formats of inodes that we cannot exchange because the code in xfs_fsr cannot precisely control the format of the inode forks from userspace. If you use the event tracing to capture the before and after swap extent trace points, I can tell you why they were incompatible for exchange, and whether xfs_fsr can be further modified to enable those files to be defragmented....

Cheers,

Dave.

Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs