Re: kernel BUG at fs/btrfs/inode.c:806

[Chris Mason <chris.mason@oracle.com>]

Excerpts from Johannes Hirte's message of 2010-12-02 12:02:16 -0500:
> On Thursday 02 December 2010 17:52:50 Johannes Hirte wrote:
> > On Thursday 02 December 2010 17:19:56 Chris Mason wrote:
> > > Excerpts from Johannes Hirte's message of 2010-12-01 08:11:01 -0500:
> > > > On one of my machines with btrfs I got this bug:
> > > > 
> > > > entry offset 29085974528, bytes 4096, bitmap no
> > > > entry offset 29162995712, bytes 20480, bitmap yes
> > > > entry offset 29171744768, bytes 4096, bitmap no
> > > > block group has cluster?: no
> > > > 0 blocks of free space at or bigger than bytes is
> > > > block group 29834084352 has 1073741824 bytes, 1072648192 used 0 pinned 0 reserved
> > > 
> > > Well, you've had an ENOSPC explosion.
> > > 
> > > > 
> > > > The "block group" messages where way more, too much for the dmesg log buffer.
> > > > Kernel is a 2.6.37-rc3+ without the latest btrfs-fixes. The bug occurred when
> > > > compiling openoffice.org. After the bug a 'df -h' showed:
> > > > 
> > > > df -h:
> > > > Filesystem            Size  Used Avail Use% Mounted on
> > > > rootfs                 21G   17G  770M  96% /
> > > > /dev/root              21G   17G  770M  96% /
> > > > rc-svcdir             1.0M  108K  916K  11% /lib/rc/init.d
> > > > udev                   10M  116K  9.9M   2% /dev
> > > > shm                  1013M     0 1013M   0% /dev/shm
> > > > /dev/sda2              66G   46G   20G  71% /home
> > > > /dev/sdb1              75G   56G   19G  75% /mnt/windows
> > > 
> > > Which of these filesystems were you compiling on?
> > 
> > On /. It's a gentoo system and the bug happened during an 'emerge openoffice'.
> > The compilation ist usually done under /var/tmp/portage.
> 
> Btw, I was able to reproduce this with a second try to emerge openoffice.

Ok, there is one related fix in the git tree right now that you don't
have.  I'm not 100% sure it'll fix this, but it can't hurt.

-chris