Fwd: btrfsck and ctree version

[polack christian <baobab874@gmail.com>]

i know that the proposed ctree.c file is from a kernel source but
btrfsck is user space only, since the btrfs-next is newer than
btrfs-prog i was hoping for a commit of this change for the user-space
version.

since this file-system have been created prior kernel 3.2 there is no
tree root backup

 i was hoping using  btrfsck to regenerate the  csum which are failing
during mount time (Input/output error)

/var/log/messages: btrfs csum failed ino 1048522 off 5124096 csum
1219517398 private 836806197

 i didn't find any way to deactivate csum check  with a mount option

or as chris mention is there a way to regenerate  the cache on the block device.

is there a solution ?

thanks for your responses

olivier


2013/1/29 Chris Mason <chris.mason@fusionio.com>
>
> On Mon, Jan 28, 2013 at 03:03:08PM -0700, David Sterba wrote:
> > On Mon, Jan 28, 2013 at 03:07:13PM +0100, polack christian wrote:
> > > i did use btrfsck to recover it
> > > i got the tool from
> > >
> > > git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-progs.git
> > >
> > > and i got this error message:
> > > ...
> > > Check tree block failed, want=294555648, have=0
> > > Check tree block failed, want=294559744, have=0
> > > Check tree block failed, want=294559744, have=0
> > > btrfsck: ctree.c:1690: leaf_space_used: Assertion `!(data_len < 0)' failed.
> > > Aborted (core dumped)
> > >
> > > looking at
> > >
> > > git://git.kernel.org/pub/scm/linux/kernel/git/josef/btrfs-next.git
> >
> > but this is a kernel source repository, not progs, I wonder
> >
> > > this error in ctree.c have been corrected by this commit
> > >
> > > http://git.kernel.org/?p=linux/kernel/git/josef/btrfs-next.git;a=commit;h=41be1f3b40b87de33cd2e7463dce88596dbdccc4
> >
> > how this could happen. I have looked at the whether it does not silently
> > fix a bug, nothing wrong I can see now.  How did you verify that the
> > patch fixes the fsck problem?
>
> It sounds much more like the reboot or remount cleared the cache on the
> block device.
>
> -chris
>