we need this becase client's timeout should be longer then inter-mds recovery may take
- few ugly hacks that allow client to reconnect with old (before failure took place) UUID
- mds_preprw() and mds_preprw() should use l_dput() instead of f_dput()
- SOCKNAL_IO_TIMEOUT has been set to 20 seconds to make discovery of
  stale connection faster
- lconf generates persisten UUIDs for LMV clients
  it each new LMV uses fresh UUID, then each recovered MDS looks like
  new client (target MDS doesn't recognize it's old because of new UUID).
  if target MDS gets restarted, then it'll find more clients in LAST_RCVD
  then it actually had

landed on HEAD

  MDS node with gotten ino/generation couple
- warning for cross-node rename in lmv_rename(). it isn't implemented yet.
  hence, we'd like to know when we hit this path
- mdc_getattr_name() accepts empty name in order to support getattr by fid request
- mds_get_parent_child_locked() forces MDS_INODELOCK_LOOKUP to be used
  if found dentry points to another MDS node
- mea_name2idx() and raw_name2idx() checks for last not to be zero and warns if it's

NOTE: this should fix the cause of the test #5 failure

  JBD routines must be called under BKL on 2.4

  fix the problem some apps confuses on: they see the same inode number for
  different objects. thanks to Andreas for the idea.

  recovery and looks for given ino/generation first. if no ino/generation are
  given then it simple recreates new object and returns ino/generation back
  to MDS holding a name. leaked inode is subject to rollback mechanism that
  is to be implemented yet
- when mds_reint_create() is asked to replay cross-node mkdir() it signals
  this to MDS holding dir inode and sends ino/generation couple if possible

NOTE: now cross-node mkdir() survives failure on any MDS node: holding the
      name and holding the inode

  given ino/generation. we need this for cross-node mkdir() which creates
  inode on remote MDS and adds dir entry then. after that remote MDS may
  fail and primary MDS will have to recreate inode with already gotten
  ino and generation numbers
- export refcounting bug in lmv fixed
- mdc_obj_create() stores just received ino/generation back to request.
  if request replay will happen mdt_obj_create() will use stored numbers
  to recreate lost inode
- mds_filter_recovery_request() accepts OST_CREATE request. need to be
  discussed though
- mdt_obj_create() rewritten to support recovery phase
- needless mds_lmv_disconnect() in mds_fs_cleanup() removed
- possible bug in retrieve_generation_numbers() fixed: it should understand
  cross-node dir entries and put proper mdsnum's

NOTE: please, re-apply extN-wantedi patch

- mds_commitrw() should use l_dput() instead of f_dput()
- removed needless CERROR() from mds_create_local_dentry()

NOTE: vanilla-2.4.20 runs CMD now

  MDS's reply with -ERESTART which notifies directory got splitted
- lmv_intent_open() recognizes a dir gets splitted during request,
  retrieves mea and repeats the request using proper MDS
- lmv_create() recognizes a dir gets splitted during request,
  retrieves mea and repeats the request using proper MDS
- bug fixed in lmv_getattr_name(): it passed wrong namelen to raw_name2idx()
- lmv_obd_create() sets OBD_MD_FLID to flag MDSs to mark created object
  unsplittable. we have to differ two possible requests: 1) to create
  remote inode for cross-node mkdir(); 2) to create slave objects. last
  ones must not be splitted recursively
- mdt_obj_create() has been rewritten to comply just described rules.
  also, it takes a lock on newly created inode. this is needed for recovery
- bug fixed in scan_and_distribute(): it tried to open an inode using decimal
  number and this caused iopen_lookup() to find alias dentries
- mds_get_lmv_attr() should return right mea size for given conf. in any case
- minor cleanups

- ldlm_cli_enqueue() takes 2nd number from resource name into account
  to recognize that returned lock changed. we need this because MDS
  may return LOOKUP lock with the same ino, but another generation.
  the case is very simple mkdir a; then touch a/b; remount; lookup a/b;
  a and b my live on different MDSs and have the same ino
- lmv_handle_remote_inode() changes intent from IT_LOOKUP to IT_GETATTR:
  caller expect to find attributes in reply
- lmv_intent_lookup() chooses right MDS to revalidate the name
- lmv_create() chooses right MDS where to forward request to
- lmv_link() chooses right mds where to forward request to
- lmv_unlink() chooses right mds where to forward request to
- lmv_readpage() removes . and .. from all the pages, but ones that come
  from master MDS for given directory
- lmv_obd_create_single() requests creation of single directory on given MDS
- mdt_obj_create() creates a directory for new mkdir() semantic
- each time new llite connects to MDS, the latter try to connect to own LMV
- mds_pack_inode2body() returns nlink=1 for directories
- mds_lmv_postsetup() pass valid cookiesize to lmv and down to the stack
- mds_reint_create() distributes new dir's inodes over the cluster
- bug in mds_create_local_dentry() fixed: rmdir may cause FIDS/ removal
- mds_get_parents_childer_lock() recognizes cross-ref dentries

cMD_PRE_CMD_SMASH are the relevant tags.

The pre-smashing contents of HEAD will be committed by yury into b_cmd as a
merge of the two branches.  These will be stabilized on b_cmd and eventually
landed onto HEAD.

Added group lock tests. Those are not built by default.

b=2094, b=3138, b=3335, b=3274, b=3293, b=2862, b=3147, b=2350, b=3313
b=3265, b=3315, b=3301, b=3325, b=3329, b=3125, b=3098

r=Robert

Update the regex of conf-sanity.sh test 13 to parse mds uuid from xml file properly.