[GIT PULL] Btrfs updates for 2.6.30-rc

[GIT PULL] Btrfs updates for 2.6.30-rc

[Chris Mason]

Hello everyone,

The master branch of:

git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable.git

Has a few btrfs bug fixes, performance fixes and a few cleanups.  The
highlights are some code from Josef Bacik to for metadata block group
preallocation, which will help cut down on the chances of metadata
enospc.

Linus found an acl caching bug that made for many many btree searches
for acls on inodes that didn't have acls, and that is fixed up as well.

Chris Mason (7) commits (+144/-237):
    Btrfs: look for acls during btrfs_read_locked_inode (+62/-0)
    Btrfs: fix deadlocks and stalls on dead root removal (+6/-0)
    Btrfs: fix fallocate deadlock on inode extent lock (+29/-17)
    Btrfs: remove unused btrfs_bit_radix slab (+0/-8)
    Btrfs: ratelimit IO error printks (+31/-17)
    Btrfs: remove #if 0 code (+1/-188)
    Btrfs: fix acl caching (+15/-7)

Christoph Hellwig (3) commits (+30/-66):
    Btrfs: kill btrfs_cache_create (+28/-43)
    Btrfs: don't export symbols (+0/-6)
    Btrfs: simplify makefile (+2/-17)

Joel Becker (2) commits (+41/-24):
    Btrfs: Fix a trivial warning using max() of u64 vs ULL. (+1/-1)
    Btrfs: Fix a bunch of printk() warnings. (+40/-23)

Josef Bacik (1) commits (+42/-2):
    Btrfs: try to keep a healthy ratio of metadata vs data block groups

Chris Ball (1) commits (+27/-11):
    Btrfs: When shrinking, only update disk size on success

Total: (14) commits

 fs/btrfs/Makefile           |   19 -----
 fs/btrfs/acl.c              |   18 +++--
 fs/btrfs/ctree.h            |    6 +
 fs/btrfs/disk-io.c          |   93 ++++++++------------------
 fs/btrfs/extent-tree.c      |   49 +++++++++++---
 fs/btrfs/extent_io.c        |   81 +----------------------
 fs/btrfs/extent_map.c       |   17 ----
 fs/btrfs/file.c             |   89 +------------------------
 fs/btrfs/free-space-cache.c |   15 ++--
 fs/btrfs/inode-map.c        |    2 
 fs/btrfs/inode.c            |  153 ++++++++++++++++++++++++++++++--------------
 fs/btrfs/ioctl.c            |    9 +-
 fs/btrfs/super.c            |   27 +++++--
 fs/btrfs/transaction.c      |    6 +
 fs/btrfs/tree-log.c         |    2 
 fs/btrfs/volumes.c          |   35 ++++++----
 fs/btrfs/volumes.h          |    3 
 17 files changed, 284 insertions(+), 340 deletions(-)

Re: [GIT PULL] Btrfs updates for 2.6.30-rc

[Jeff Chua]

On Tue, Apr 28, 2009 at 2:05 AM, Chris Mason <chris.mason@oracle.com> wrote:
> Hello everyone,
>
> The master branch of:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable.git
>
> Has a few btrfs bug fixes, performance fixes and a few cleanups.  The
> highlights are some code from Josef Bacik to for metadata block group
> preallocation, which will help cut down on the chances of metadata
> enospc.


Chris,

Any chance to create a btrfs partition less than 256MB? I'm looking at
using btrfs when it's stable enough for all partitions rather than
some btrfs and smaller ones non-btrfs. Currently, reiserfs/reiser4
supports smaller than 100MB which is good for testing.

in btrfs-progs-0.18/mkfs.c ...
                                if (block_count < 256*1024*1024) {
                                        fprintf(stderr, "File system size "
                                                "%llu bytes is too small, "
                                                "256M is required at least\n",
                                                block_count);
                                        exit(1);
                                }

Also, would it be possible when creating a 300MB partition for "df" to
display "50MB free" instead of "300MB free"?

Thanks,
Jeff.

Re: [GIT PULL] Btrfs updates for 2.6.30-rc

[Jeff Chua]

On Fri, May 8, 2009 at 5:22 PM, Jeff Chua <jeff.chua.linux@gmail.com> wrote:
> Also, would it be possible when creating a 300MB partition for "df" to
> display "50MB free" instead of "300MB free"?

Should display "46MB free" (256MB used by journal?).


Thanks,
Jeff.

Re: [GIT PULL] Btrfs updates for 2.6.30-rc

[Chris Mason]

On Thu, May 14, 2009 at 03:19:15PM -0400, Chris Mason wrote:
> Hello everyone,
> 
> Here's a small btrfs update for the next rc.  The IO failure fixes help
> btrfs carry on when metadata reads fail because all copies of a given
> block are bad.  There is still a bunch of work to do in that area, but
> this a small start that avoids hopeless looping.

Sorry, the URL for the pull is the master branch of:

git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable.git

[full quote of the pull details below ]

-chris

> 
> Chris Mason (3) commits (+37/-6):
>     Btrfs: remove some WARN_ONs in the IO failure path (+0/-3)
>     Btrfs: Don't loop forever on metadata IO failures (+36/-3)
>     Btrfs: init inode ordered_data_close flag properly (+1/-0)
> 
> Li Hong (1) commits (+0/-4):
>     Btrfs: remove outdated comment in btrfs_ioctl_resize()
> 
> Sankar P (1) commits (+1/-1):
>     Btrfs: Spelling fix in btrfs_lookup_first_block_group comments
> 
> Sage Weil (1) commits (+2/-2):
>     Btrfs: make show_options result match actual option names
> 
> Total: (6) commits
> 
>  fs/btrfs/ctree.c       |   39 ++++++++++++++++++++++++++++++++++++---
>  fs/btrfs/disk-io.c     |    2 --
>  fs/btrfs/extent-tree.c |    2 +-
>  fs/btrfs/inode.c       |    2 +-
>  fs/btrfs/ioctl.c       |    4 ----
>  fs/btrfs/super.c       |    4 ++--
>  6 files changed, 40 insertions(+), 13 deletions(-)

[GIT PULL] Btrfs updates for 2.6.30-rc

[Chris Mason]

Hello everyone,

Here's a small btrfs update for the next rc.  The IO failure fixes help
btrfs carry on when metadata reads fail because all copies of a given
block are bad.  There is still a bunch of work to do in that area, but
this a small start that avoids hopeless looping.

Chris Mason (3) commits (+37/-6):
    Btrfs: remove some WARN_ONs in the IO failure path (+0/-3)
    Btrfs: Don't loop forever on metadata IO failures (+36/-3)
    Btrfs: init inode ordered_data_close flag properly (+1/-0)

Li Hong (1) commits (+0/-4):
    Btrfs: remove outdated comment in btrfs_ioctl_resize()

Sankar P (1) commits (+1/-1):
    Btrfs: Spelling fix in btrfs_lookup_first_block_group comments

Sage Weil (1) commits (+2/-2):
    Btrfs: make show_options result match actual option names

Total: (6) commits

 fs/btrfs/ctree.c       |   39 ++++++++++++++++++++++++++++++++++++---
 fs/btrfs/disk-io.c     |    2 --
 fs/btrfs/extent-tree.c |    2 +-
 fs/btrfs/inode.c       |    2 +-
 fs/btrfs/ioctl.c       |    4 ----
 fs/btrfs/super.c       |    4 ++--
 6 files changed, 40 insertions(+), 13 deletions(-)

[GIT PULL] Btrfs updates for 2.6.30-rc

[Chris Mason]

Hello everyone,

Here is the first round of btrfs fixes for the next merge window.  The
majority of this code is to reduce btrfs stack usage by doing extent
tree updates at a different time.  It generally makes btrfs usable on 4k
stacks, although more stack reduction work will be done later on.

The second commit from the bottom adds extra ordering for renames and
truncates to bring back the ext3 flavor everyone has been missing.  I
tried pretty hard to keep performance impact low, most workloads will
never notice it (including those that fsync).

Linus, please pull the master branch from:

git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable.git

Chris Mason (18):
      Btrfs: don't preallocate metadata blocks during btrfs_search_slot
      Btrfs: do extent allocation and reference count updates in the background
      Btrfs: reduce stack usage in some crucial tree balancing functions
      Btrfs: try to cleanup delayed refs while freeing extents
      Btrfs: process the delayed reference queue in clusters
      Btrfs: reduce stalls during transaction commit
      Btrfs: reduce stack in cow_file_range
      Btrfs: Check for a blocking lock before taking the spin
      Btrfs: Only let very young transactions grow during commit
      Btrfs: leave btree locks spinning more often
      Btrfs: readahead checksums during btrfs_finish_ordered_io
      Btrfs: limit balancing work while flushing delayed refs
      Btrfs: Make sure i_nlink doesn't hit zero too soon during log replay
      Btrfs: tree logging unlink/rename fixes
      Btrfs: optimize fsyncs on old files
      Btrfs: make sure btrfs_update_delayed_ref doesn't increase ref_mod
      Btrfs: add extra flushing for renames and truncates
      Btrfs: try to free metadata pages when we free btree blocks

 fs/btrfs/Makefile       |    2
 fs/btrfs/btrfs_inode.h  |   31
 fs/btrfs/ctree.c        |  588 +++++++++-------
 fs/btrfs/ctree.h        |   69 +
 fs/btrfs/delayed-ref.c  |  669 +++++++++++++++++++
 fs/btrfs/delayed-ref.h  |  193 +++++
 fs/btrfs/dir-item.c     |    3
 fs/btrfs/disk-io.c      |   81 +-
 fs/btrfs/disk-io.h      |    1
 fs/btrfs/extent-tree.c  | 1674 +++++++++++++++---------------------------------
 fs/btrfs/extent_io.c    |   51 -
 fs/btrfs/extent_io.h    |    3
 fs/btrfs/file-item.c    |    7
 fs/btrfs/file.c         |   50 +
 fs/btrfs/inode-item.c   |    3
 fs/btrfs/inode.c        |  194 ++++-
 fs/btrfs/locking.c      |   21
 fs/btrfs/ordered-data.c |  118 +++
 fs/btrfs/ordered-data.h |    4
 fs/btrfs/transaction.c  |  151 +++-
 fs/btrfs/transaction.h  |    8
 fs/btrfs/tree-defrag.c  |    2
 fs/btrfs/tree-log.c     |  444 ++++++++++--
 fs/btrfs/tree-log.h     |   17
 24 files changed, 2762 insertions(+), 1622 deletions(-)