--- Comment #8 from Dave Chinner <david@xxxxxxxxxxxxx> 2010-04-04 19:29:10 CST
(In reply to comment #7)
> (In reply to comment #6)
> > Kernel version is 2.6.32-9, right? There should be fixes preventing this
> > corruption in the next 2.6.32.x kernel release. Alternatively, the fixes
> > are in
> > 2.6.33, so it might be worth downloading 2.6.33 and seeing if the problem
> > goes
> > away.
> I also use 2.6.28, but if you're right, this means that xfs_fsr changes
> metadata format to some unsupported by older kernels, this can't be true.
You're right, it isn't true. The problem is that the swap extents code that fsr
uses doesn't handle certain format inodes correctly, and when it swaps them it
leaves corrupted inodes behind. The fixes in 2.6.33 detect this situation and
abort the swap before it can happen.
So there is no new metadata format or any on disk change to prevent this
problem - just more careful kernel code. As such, there isn't anything wrong
with xfs_fsr and it as capable as it has always been of defragmenting large
Configure bugmail: http://oss.sgi.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.