b=15556
i=nikita.danilov
i=yury.umanets

- do not produce zero inode number;
- hash VFS inode by FID;
- start FID sequence from 0x100000400.

b=14714
i=adilger
i=johann

Move to use d_rehash_cond & d_move_locked.

i=yong.fan
i=rahul.deshmukh
i=nikita.danilov

- use req_capsule interface for client.
- add some interoperability support on server side.

- redesigned remote acl support, by FanYong
  b=11842
  r=pravin.shelar, eric.mei
- new sptlrpc configure interface
  b=13642
  r=yong.fan, rahul.deshmukh
- split kerberos keytab for lustre_root
  b=13873
  r=yong.fan, h.huang
- fix setuid for gss with linux keyring
  b=13899
  r=yong.fan, h.huang

add sles10 support.

  b:13816
  i:shadow,yury

b=13568
i=adilger
i=green

Remove 2.4 kernel check from 1.8.0 for llite subsystem.

b=11829
i=huanghua
i=tappro

early lock cancel for cmd

Move #ifdef above now-redundant checks if __iget() is not exported, per
Oleg's inspection comments.
b=12391
i=green

b:12123
Check if a directory has children when invalidating dentries
associated with an inode during lock cancellation.  This fixes
an incorrect ENOENT sometimes seen for valid filehandles during
testing with dbench.
r:green
r:shadow

Description: data loss for recently-modified files
Details    : In some cases it is possible that recently written or created
	     files may not be written to disk in a timely manner (this should
	     normally be within 30s unless client IO load is very high).
	     The problem appears as zero-length files or files that are a
	     multiple of 1MB in size after a client crash or client eviction
	     that are missing data at the end of the file.

	     This problem is more likely to be hit on clients where files are
	     repeatedly created and unlinked in the same directory, clients
	     have a large amount of RAM, have many CPUs, the filesystem has
	     many OSTs, the clients are rebooted frequently, and/or the files
	     are not accessed by other nodes after being written.

	     The presence of the problem can be detected by looking at
	     /proc/sys/fs/inode-state.  If the first number (nr_inodes) is
	     smaller than the second (nr_unused) then dirty files will not
	     be flushed automatically to disk.  "sync; sleep 10" should be
	     run several times on the node before unmounting it to update
	     Lustre (this is also safe to run on nodes without this problem).

	     There is also a related kernel bug in the RHEL4 4 2.6.9 kernel
	     that can cause this same problem, so customers using that kernel
	     also need to update the kernel in addition to Lustre.  In order
	     to properly fix this bug, the RHEL3 2.4.21 kernel is also updated.

	     It is normal that files written just before a client crash (less
	     than 30s) may not yet have been flushed to disk, even for local
	     filesystems.
i=green(original patch), i=shadow
b=12181, b=12203

Wrap lines at 80 columns.

Remove "#if FOO" compiler warnings from -Wundef.
(Committed by adilger)

Define FMODE_READ and FMODE_WRITE for use in liblustre
(Committed by green)

Allow test 6 and 7 to be run in isolation.
Fix test file names.
Create test files in $DIR instead of $MOUNT.
(Committed by adilger)

openclose.c: Save errno before calling fprintf().
(Committed by adilger)

sanityN.sh: Kill multiop before exiting, if test fails.
Test is currently excluded, was retesting to see if it worked yet.
b=3192
(Committed by adilger)

r=adilger
r=nathan
r=green

revert back last changes after testing.

 turn on stat auditing on client for testing purpose

stop STAT auditing on client

- fix for client STAT logging

 - debug addon: track how long iget() and ll_writepages() take

 - debug addon

land crypto api

 - b_size_on_mds landed (prototype for CMD2 w/o recovery support on OST)

- added canceling unused locks before disconnecting master/cache exports in cobd_iocontrol()

- added lmv_cancel_unused() to cancel unused locks from llite correctly instead of using
direct ldlm_cli_cancel_unused() on lmv namespace.

- removed not used ll_mdc_cancel_unused()

  - making it work in some stat cases (stat target is behind mount object)
  - making it work in sys_mount() control path (is mount point is behind mount object)

  - to avoid deadlock is mount case is_subdir() is used for detecting is current dentry is
    parent of dentry being in progress of mounting.

  - do not mount on revalidate if there is not valid @nd applied.
  - using nd->flags if possible for both lookup and revalidate cases.
  - added debug info printing in mounting cases.
  - totaly removed -ERESTARTSYS approach stuff.

  - fixed bug with many waiters on ->ll_gns_mount_finished - only first of them will find GNS
    in IDLE state and will start mounting and the rest will assert on wrong GNS state (MOUNTING)

  - reinit ->ll_gns_mount_finished after all waiters are awakaned, this fixes the bug when
    after first using this completion it does not functions, because complete_all() has
    slighly diff. semanthics then complete() and namely does not just decrease ->done but
    sets it to MAX_INT/2.

  - fully rewritten sanity-gns.sh to make it more flexible and easy to add new sanity tests.

- fixed sys_umount() code path. It did not release intent what caused memory leaks in sanity-gns.sh, as it uses umount for cleanups while testing GNS.
- do not free intent in GNS case if it is not lookup_it, only drop lock.
- added test_3a which checks disabling GNS mount point using chmod u-s

- using IT_CHDIR in sys_chdir() vfs-intent_api-vanilla-2.6.10-fc3.patch. This fixes GNS with cd into mnt point.

- different fixes in GNS stuff.
- gns_enabled flag is moved to separate space instead of using LL_GNS_DISABLED
- removed error messages after inode_setattr(). Only comment is added.
- many changes in sanity-gns.sh, added checks for "cd" case. Deadlock tests are disabled for a while.

- added IT_GETATTR to check if mount should be performed (experimental)

  - sleep_on() is known to be racy replacing it with OBD_SLEEP_ON()
  - added OBD_SLEEP_ON() which is inspired by wait_event() macro.
  - inter_module_{get|put}() replaced with symbol_{get|put}()

- fixes in GNS stuff:
  - ll_revalidate_it() calls missed ll_intent_release() in the case
    GNS dentry is found (fix for #6176)

  - added ll_gns_send_signal() which is needed to cause syscall restarting in the
    case of -ERESTARTSYS is returned.

  - cleanups in ll_gns_mount_object()

  - ll_gns_mount_object() does not wait unconditionaly for userspace upcall,
    instead it give it some time to complete (10s) and returns error if it
    did not mount dentry. This fixes the case when upcall script runs endless
    loop and never returns. This also fixes possible deadlock if another thread
    with wait on not finished mount (this schema is disabled by and -ERESTARTSYS
    is used instead. Both, waiting aproach and -ERESTARTSYS one need more
    investigation).

  - handling result of inode_setattr() call. Printing error message on error.
    This fixes wanings that result of calling this function should be checked.

  - fixed debug message about GNS inabilty to mount dentry (missed arg to CDEBUG
    format string)

  - cleanups in obdfilter_init()
  - fixed using "mount" command in sanity-gns.sh

  none of thread needs to wait for mount completion in kernel space.

- small cleanups in ll_gns_mount_object(). Some checks for input invariants are
  moved after checking gns state. Fixes in comments.

- added the ability to disable GNS at all via /proc/.../gns_enabled. This is useful
  for some complex tests (like recursive suid marked .mntinfo dir) at least. May be
  useful for something else.

- added two new sanity tests to sanity-gns.sh:

  a) test 2g: mounting /mnt/lustre/gns_entry/.mntinfo/.mntinfo/.mntinfo/.mntinfo where
     only last entry is correct GNS mount object.

  b) test 2h: upcall script starts mounting in background and retuns with no waiting
     for mount command.

- in sanity tests disable GNS when preparing mount objects and mount points.