[Btrfs-devel] [PATCH 0/2] make root directory writeable

[Josef Bacik]

On Fri, Jun 29, 2007 at 08:05:37AM -0400, Chris Mason wrote:
> On Thu, Jun 28, 2007 at 11:36:59PM -0400, Josef Bacik wrote:
> > Hello,
> > 
> > Not really sure why but when you do a mkfs it creates a default directory and
> > the root directory is read-only.  The first patch makes the root directory
> > writeable, and the second patch fixes mkfs so it sets the root directoryy's
> > permission properly and doesn't create the default directory.  I tested this on
> > my box and ran iozone and it worked fine.  Thank you,
> 
> Thanks for sending these along, but the top level directory is special.
> It is meant to only have snapshots and subvolumes.  As we refine the way
> that subvolumes and snapshots are presented, the top level directory
> won't be needed anymore (at least won't need to be visible), but for
> right now, it does need to be read only.
> 
> The 'default' directory is a subvolume, where all the tree blocks and
> file extents inside are reference counted.  You can make more subvolumes
> appear in the root directory with the btrfsctl command:
> 
> btrfsctl -s new_subvol mount_point
> 
> The btree that makes up the root directory is written via copy-on-write,
> but it is not reference counted.  It is a special index of the
> subvolumes and snapshots on the FS, so we don't want to create files
> there.
>

Ahh ok, I couldn't see any explanation as to why you did it that way so I
figured it was just to make it easier to show off snapshots/subvolumes early on.
Thanks for the explanation.
 
> Here are a few ideas of things that won't conflict too much with my
> current enospc work.  But, if you have other ideas or bits you would
> like to improve, please feel free to suggest them, I'm happy to get all
> the help I can.
> 
> 1) On fsync, ext3 and reiserfs have code to let more writers try and
> come into the current transaction before a commit is forced (this is
> called let_transaction_grow in reiserfs).  btrfs needs something
> similar.  It's fairly simple, you look at the count of procs currently
> in the transaction, and if it is non-zero you sleep for a bit before
> forcing the commit.
> 
> It's important to sleep without any locks held and without being in the
> transaction.  Once coded, you can benchmark with Ric Wheeler's fs_mark:
> 
> http://developer.osdl.org/dev/doubt/fs_mark/index.html
> 
> If you're comparing btrfs to ext3 on a sata or ide drive, make sure to
> mount ext3 with -o barrier=1
> 
> 2) The in ram portion of the inode needs a field to indicate which
> transaction number last changed the inode.  This field can be checked by
> fsync to decide if a commit is actually required.
> 
> You would have to an index (use a radix tree) of the in memory
> transaction structs so that you can check to make sure a given
> transaction is actually done.
> 
> 3) struct btrfs_root_item has fields for block use accounting, but they
> currently don't get updated.  When blocks are allocated for a given
> root, we need to reflect that in struct btrfs_root_item and send it down
> to disk when the transaction commits.
> 
> The in memory struct btrfs_root has a duplicate of the root item, so the
> updates can be stored in the in memory version and then put into the
> tree once right before commit.
>

Awesome I will start working on these when I get home.  Thanks much,

Josef