[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Ouch, here's an odd one.



Hi,

On Tue, Feb 06, 2001 at 07:16:04PM -0800, Jay Weber wrote:
> 
> Well, simply doing the following (without a forced sync call) leads to
> either the lack of copied file after reboot, or a bogus copy of the file,
> when done on ext3.  Forcing a sync prior to reboot seems to keep the data
> intact.
> 
> cp /etc/fstab /
> reboot -nf

Ahh, my fault: one of the recent ext3 diffs got a logic test the wrong
way round.  Patch below.  Only ext3-0.0.5e is affected.

I'm currently testing 0.0.6a with the new dirty buffer handling and
the above fix.  I've also got a handle on the recovery hangs (orphaned
special inodes seem to be the problem) so as soon as I've got all of
that sorted I'll push out an official 0.0.6a release.

The dirty buffer changes will not only fix the occassional "buffer
already locked" error we see, it should also make ext3 robust against
odd block devices such as loop or lvm which can sleep in their IO
request code.

Cheers,
 Stephen





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]