[Ocfs2-tools-commits] branch, master, updated. ocfs2-tools-1.4.0-586-gec43bf6

svn-commits at oss.oracle.com svn-commits at oss.oracle.com
Fri Sep 10 01:10:24 PDT 2010


This is an automated email from the git hooks/post-receive script. It was
generated because a ref change was pushed to the repository containing
the project "Tools to manage the ocfs2 filesystem.".

The branch, master has been updated
       via  ec43bf6989ee91a68a9b39ae84ee3cc3d654f730 (commit)
       via  6fc7d15d8eb33c4eaa4ab5909a4e4ee9eec1ad65 (commit)
       via  9190ceae825a3e8e5701900ceed83e160e269aa1 (commit)
       via  d8857c20d9ce29161de202b598f966ba8ace303f (commit)
       via  285b04cea518137d0a4a5a353118b08fc5969568 (commit)
       via  17015fe03bbbbc3dacef13e166f183009fd26fb8 (commit)
       via  cdc69878698e9e6178b61de95afc6f51dfe8546d (commit)
       via  fd6abf1be1872ffdc060f80b395815c29b39e79d (commit)
       via  cdb51ac4a970fae86a7ecbf6a97f320ecc2cd0b9 (commit)
       via  963033ec4d145a6b5d5fe6c54681bf22be36832c (commit)
       via  aceadddfc82c63095cf4cf3f32e64661f77ad52d (commit)
       via  72216877ddac6e25acaa7a5e13ac1e12547da66f (commit)
       via  ddcafce505d9417c10d3712dae1990e764e3a769 (commit)
       via  12bd51271297b4c6c5d2f2b4cb1f56f6fc791384 (commit)
       via  7b759c354d4ddf4244405760a09aea2915df8f8c (commit)
       via  64a0abfcf248b2ccae9f23e8514d375489c281ab (commit)
       via  4bfd9b3308336eb791b63f26108994583fbed036 (commit)
       via  dde908fa3334c81c755f7065ff467cae94221c1b (commit)
       via  610c862f8724bbd78692e029253c03b7ea6c5070 (commit)
       via  be45dae440f5acef608fbc0279e33cb19def5b56 (commit)
       via  32d3d75656e757774872892fc4751b08ba31df99 (commit)
       via  54da71703065044458f1b865d120f04d9b48ce1b (commit)
       via  b74ea70d8eeef3ad6ab8d02d93530405f33844bb (commit)
       via  7878e2a891b88f6dee3eb0e59c7577afcf32e64b (commit)
       via  3caf13f53e13d591b9d2eeb7bfd35ee0cce4edf4 (commit)
      from  6e7008eec025ce12f6e3a8789fa26821e4512d83 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.

- Log -----------------------------------------------------------------
commit ec43bf6989ee91a68a9b39ae84ee3cc3d654f730
Author: Tao Ma <tao.ma at oracle.com>
Date:   Fri Sep 10 15:51:14 2010 +0800

    fsck.ocfs2: Avoid refcount a system file
    
    In pass1d, we try to adding refcount record for a multiple
    claimed clusters first. But if the cluster is found in a
    system file, we shouldn't try to add refcount for it and
    the old method should be carried out.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 6fc7d15d8eb33c4eaa4ab5909a4e4ee9eec1ad65
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:20:05 2010 +0800

    ocfs2-tools: Enable discontiguous block group support.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 9190ceae825a3e8e5701900ceed83e160e269aa1
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:19:55 2010 +0800

    o2image: Make it read group_desc properly.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit d8857c20d9ce29161de202b598f966ba8ace303f
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:19:40 2010 +0800

    fswreck: Add corrupt codes for extent list check.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 285b04cea518137d0a4a5a353118b08fc5969568
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:19:24 2010 +0800

    fsck.ocfs2: Add extent list check for discontig bg.
    
    So we will do some basic sanity check for discontig block group.
    The check includes:
    1. l_tree_depth must be set to 0.
    2. l_count must be set to ocfs2_extent_recs_per_gd.
    3. if l_next_free_rec is broken and the extent record list looks
       sane, update l_next_free_rec.
    4. if any extent record has an invalid e_cpos, dropt the group.
    5. if there is only one rec which has invalid e_leaf_clusters,
       update it according to other extent records.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>

commit 17015fe03bbbbc3dacef13e166f183009fd26fb8
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:18:37 2010 +0800

    ocfs2: Add ocfs2_extent_recs_per_gd.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit cdc69878698e9e6178b61de95afc6f51dfe8546d
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:17:06 2010 +0800

    fsck.ocfs2: Let pass1b handle discontig block group.
    
    In pass1b, in case we find duplicated clusters, we have
    to handle the discontig block group.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit fd6abf1be1872ffdc060f80b395815c29b39e79d
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:16:52 2010 +0800

    fsck.ocfs2: Make pass0 aware of discontig block group.
    
    Make mark_group_used set the bitmap properly.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit cdb51ac4a970fae86a7ecbf6a97f320ecc2cd0b9
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:16:34 2010 +0800

    debugfs.ocfs2: Dump extent list if the group is discontiguous.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 963033ec4d145a6b5d5fe6c54681bf22be36832c
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:16:03 2010 +0800

    debugfs.ocfs2: Dump suballoc_loc if exists.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit aceadddfc82c63095cf4cf3f32e64661f77ad52d
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:15:10 2010 +0800

    tunefs.ocfs2: Add support for disabling discontig-bg.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 72216877ddac6e25acaa7a5e13ac1e12547da66f
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:14:11 2010 +0800

    tunefs.ocfs2: Add enable discontig-bg operations.
    
    This patch just adds the support for enable discontig-bg
    in tunefs.ocfs2 so that we can test it easily.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit ddcafce505d9417c10d3712dae1990e764e3a769
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:12:17 2010 +0800

    libocfs2: Set suballoc_bit properly for discontig block group.
    
    The old ocfs2_chain_alloc_with_io returns gb_blkno properly
    for us, that is cool for us. So we only need to pass back
    the proper suballoc_bit. And that's done for discontig
    block group.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 12bd51271297b4c6c5d2f2b4cb1f56f6fc791384
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:12:05 2010 +0800

    libocfs2: Let chainalloc handle discontig block group.
    
    chainalloc is used to allocate some bits from a suballocator.
    It will use bitmap_region to handle the real work of allocation.
    So for contiguous group, create one region for the whole group.
    For a discontiguous one, every leaf extent record will become an
    individual ocfs2_bitmap_region and be inserted.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 7b759c354d4ddf4244405760a09aea2915df8f8c
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:11:53 2010 +0800

    bitmap: Add 2 fields to facilitate discontiguous block group.
    
    For a discontiguous block group, we will have some bitmap
    region which doesn't start from the bit offset aligned to bytes.
    And I don't want to create a new bitmap region which start
    from that bit offset because in that case when we read/write
    from the original bg->bg_bitmap, we have to shift bits for
    every bytes in the region. That is too bad.
    
    So here the solution is: Add a new field named br_bitmap_start
    in ocfs2_bitmap_region which indicates the start bit in it.
    The bits ahead of it won't be touched by bitmap function now.
    
    With it, we can memcpy the original bitmap easily from the
    bg_bitmap and manipulate them.
    
    'br_total_bits' now includes both the useful and useless bits
    and another auxiliary paramter br_valid_bits is added to indicate
    the really bit length.
    
    Note:
    The only left problem is merge_region, but as we don't support
    it in chainalloc, we don't need to handle it by now.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 64a0abfcf248b2ccae9f23e8514d375489c281ab
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:11:33 2010 +0800

    libocfs2: Add functions to get the bit_set count.
    
    As for discontiguous block group, we need to get
    the total set_bits in a range of bitmap. So add
    the function ocfs2_get_bits_set.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 4bfd9b3308336eb791b63f26108994583fbed036
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:08:16 2010 +0800

    libocfs2: Make inode_scan notices discontig groups.
    
    inode_scan is used to iterate all the inodes in the
    ocfs2 volume. So this patch make inode_scan work
    with discontiguous block group.
    The change includes:
    1. Add a new function named get_next_read_blocks which
       will get the num_blocks to read and set cur_blkno
       properly if a discontig inode group is encountered.
    2. Change 'bpos' to b_offset. The old 'bpos' is used
       to indicate a block physical position in the block
       group, b_offset will be used as a virtual offset
       in the whole block group. It is almost the same.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit dde908fa3334c81c755f7065ff467cae94221c1b
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:07:31 2010 +0800

    libocfs2: Make ocfs2_init_group_desc aware of discontig group.
    
    Pass a parameter named suballoc to ocfs2_init_group_desc
    so that it can init bg_size properly.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 610c862f8724bbd78692e029253c03b7ea6c5070
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:07:18 2010 +0800

    libocfs2: Change swap_group_desc to read/write discontig groups.
    
    In read/write discontig groups, we will use
    offsetof(struct ocfs2_group_desc, bg_bitmap) + gd->bg_size ==
              offsetof(struct ocfs2_group_desc, bg_list)
    to check whether the group is discontiguous or not.
    And what' more, we change ocfs2_swap_group_desc to 2 separate
    functions *_from_cpu and *_to_cpu since we now have to swap
    the extent list if needed.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit be45dae440f5acef608fbc0279e33cb19def5b56
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:03:10 2010 +0800

    ocfs2: Add ocfs2_gd_is_discontig.
    
    Add ocfs2_gd_is_discontig so that we can test whether
    a group descriptor is discontiguous or not.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 32d3d75656e757774872892fc4751b08ba31df99
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:03:19 2010 +0800

    libocfs2: swap suballoc_loc properly.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 54da71703065044458f1b865d120f04d9b48ce1b
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:02:11 2010 +0800

    libocfs2: Add the 'discontig_bg' feature to the feature definitions.
    
    Add discontig_bg to the selectable and printable feature definitions.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit b74ea70d8eeef3ad6ab8d02d93530405f33844bb
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:01:59 2010 +0800

    libocfs2: Change ocfs2_group_bitmap_size.
    
    Change ocfs2_group_bitmap_size to accept parameters suballocator
    and feature_incompat.currently all the callers set suballocator
    to 0. Some of them will be changed later.
    As for feature_incompat, some functions like ocfs2.h don't have
    a fs passed in, so we just use 0 and let it b a cluster group.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 7878e2a891b88f6dee3eb0e59c7577afcf32e64b
Author: Tao Ma <tao.ma at oracle.com>
Date:   Thu Sep 9 14:01:44 2010 +0800

    ocfs2-tools: Add disk layout for discontiguous block group.
    
    Add disk layout for discontiguous block group.
    
    Signed-off-by: Tao Ma <tao.ma at oracle.com>
    Signed-off-by: Joel Becker <joel.becker at oracle.com>

commit 3caf13f53e13d591b9d2eeb7bfd35ee0cce4edf4
Author: Goldwyn Rodrigues <rgoldwyn at suse.de>
Date:   Thu Sep 9 13:58:35 2010 +0800

    Fix - fsck.ocfs2 attempts to clear dx_dir blocks
    
    If a directory has dx_dir blocks, fsck attempts to clear them in global bitmap.
    
    Steps to reproduce:
    Make filesystem with indexed-dirs feature enabled. Make a directory and create
    a lot of files in it. Run fsck.ocfs2 -f on this filesystem.
    
    This fix accounts for the dx_leaf nodes and avoids clearing in global bitmap.
    The patch reuses the normal file extent functions, which includes code
    for refcounting which might not be relevant in this situation.
    
    Signed-off-by: Goldwyn Rodrigues <rgoldwyn at suse.de>
    Signed-off-by: Tao Ma <tao.ma at oracle.com>

-----------------------------------------------------------------------

Summary of changes:
 debugfs.ocfs2/dump.c                           |   43 +++-
 debugfs.ocfs2/find_block_inode.c               |    3 +-
 extras/compute_groups.c                        |    2 +-
 fsck.ocfs2/fsck.ocfs2.checks.8.in              |   70 +++++
 fsck.ocfs2/pass0.c                             |  232 ++++++++++++++++-
 fsck.ocfs2/pass1.c                             |    4 +
 fsck.ocfs2/pass1b.c                            |   71 +++++-
 fswreck/Makefile                               |    5 +-
 fswreck/chain.c                                |    3 +-
 fswreck/corrupt.c                              |    6 +
 fswreck/discontig_bg.c                         |  277 ++++++++++++++++++++
 fswreck/group.c                                |    3 +-
 fswreck/include/corrupt.h                      |    2 +
 fswreck/include/{refcount.h => discontig_bg.h} |   16 +-
 fswreck/include/fsck_type.h                    |   10 +
 fswreck/include/main.h                         |    1 +
 fswreck/main.c                                 |   20 ++
 include/ocfs2-kernel/ocfs2_fs.h                |  135 +++++++++-
 include/ocfs2/bitops.h                         |    1 +
 include/ocfs2/ocfs2.h                          |   21 ++-
 libocfs2/alloc.c                               |   65 +++--
 libocfs2/bitmap.c                              |   61 +++--
 libocfs2/bitmap.h                              |    3 +
 libocfs2/bitops.c                              |   16 ++
 libocfs2/chain.c                               |   60 ++++-
 libocfs2/chainalloc.c                          |  244 ++++++++++++++----
 libocfs2/extents.c                             |    1 +
 libocfs2/feature_string.c                      |   12 +-
 libocfs2/inode.c                               |    3 +-
 libocfs2/inode_scan.c                          |   74 +++++--
 libocfs2/refcount.c                            |    1 +
 libocfs2/xattr.c                               |    1 +
 mkfs.ocfs2/mkfs.c                              |   29 ++-
 o2image/o2image.c                              |   10 +-
 sizetest/sizetest.c                            |    6 +-
 tunefs.ocfs2/Makefile                          |    1 +
 tunefs.ocfs2/feature_discontig_bg.c            |  328 ++++++++++++++++++++++++
 tunefs.ocfs2/op_features.c                     |    2 +
 tunefs.ocfs2/op_resize_volume.c                |    2 +-
 39 files changed, 1647 insertions(+), 197 deletions(-)
 create mode 100644 fswreck/discontig_bg.c
 copy fswreck/include/{refcount.h => discontig_bg.h} (60%)
 create mode 100644 tunefs.ocfs2/feature_discontig_bg.c


hooks/post-receive
-- 
Tools to manage the ocfs2 filesystem.



More information about the Ocfs2-tools-commits mailing list