Handle server recovery in a new thread, instead of tying up request
processing threads. Centralizes the recovery management, including the
abort recovery checking.

I've seen the "server went back in time" message while testing HEAD,
so this will fix that.  Patch already on 1.4.

- with CMD each MDS has own group on OST, so group != 0 always. this is
  why filter_should_precreate() did return 1 always, so precreation path
  didn't take a place. I've comment out check for group out for a while

b=3869,1742

These are the remaining fixes from 1742 that are needed to allow a
client to reconnect during recover.  The request that triggered 3869
on the last run was actually a RESENT request that was sent before
recovery had finished, which this patch will fix.

2)add snap cow for obdfilter
3)some fix in smfs_cow

1) fix some bugs in create_indirect inode.

  lock on part of dir only
- minor API changes to pass the lock pointer

- fixed error code leakage in filter_precreate() what led to endless loop
in the case of -ENOSPC due to real lack of space or inability to allocate new inode.

- added various error messages in error code pathes in lmv code.
- small changes in existent error messages.

filter_sync_llog() was double locking a spin lock

2)fix bugs in fsfilt_commit in smfs, sometime inode will be null sometime

- b1_4 is basically the parent of HEAD, because that's the direction
  that changes flow, as strange as that sounds.  so there's a
  HEAD_BASE tag which sits on b1_4

b_cmobd_merge should not be used for any further development or testing!

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.

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

b=3284, b=3285, b=3043, b=3236, revert liblustre group changes

b=3043, b=3258, b=3219, b=2410

b=2912/2922, b=3145 too

b=3116, b=3064, b=3107, b=3122, b=1519, 2.6 kernel fixes, group lock fixes

b=3057, 2947, 2925, 3050, 3041, 2849, 1662

Don't deref NULL inode if we make it into filter_lvbo_init for (in
this case) the wrong object group

b=2947, b=2543, b=1557, b=2009

 to get Lustre built with 2.6.3+mm4, LTIME_S(dentry->d_inode->i_size) is not correct

Don't dereference dentry after dputting it.
Remove use of obdo in filter_lvb, since we don't actually need it.
b=2922

From b1_2.
b=2919

 to make Lustre build with 2.6.3+mm4 and suse 2.6

b=2939, b=2325, b=2059 (partial), b=2925, b=2399, b=2517

b=2633

export from the device from the export list.  This still isn't possible
for --force disconnects, so we don't do filter_grant_sanity_check() at
that time.

Also, change the LASSERTs to CERRORs in case this happens again.  First of
all to see if these errors are just transient during export disconnection
which means they are just overly-aggressive assertions, or if there is a
permanent accounting error that indicates some other problem.  Secondly, it
isn't possible to stop the use of existing export references, so there will
always be some small race during client eviction as completing operations
might update the export before it is destroyed.
b=2663

b=2464

Land initial b_bgl fixes.  Includes compile and a couple of swabbing
fixes.

b=2818

b=2733, b=2773, b=2529, b=2730, b=2819, b=2814, b=2822, b=1450, b=2676, b=2681
b=2817, b=2706, b=2816, b=1987, b=2884, b=1191, b=2809, b=2765, b=2805, b=1972

land b_size on HEAD for 1.2