b=12606
i=shadow, nikita

Description: don't use GFP_* in generic Lustre code.
Details    : Use cfs_alloc_* functions and CFS_* flags for code portability.

land 10800(adding llog ctxt refcount) to b1_6.
b=10800
i=green
i=shadow

b=11710
i=green
i=wangdi

b=12747
i=wangdi
i=scjody

land Early Lock Cancel code

LCONSOLE_ERROR(fmt, arg) to LCONSOLE_ERROR_MSG(0x000, fmt, ## arg).

b=10595
i=adilger

It also includes code fixes for first five improved error messages.

b=10595
i=adilger
i=braam

Always send IO to OST_IO_PORTAL regardless of what the default portal is.
Was previously broken for echo_client on catamount, but would also break
on 1.8 once the default request portal is REQPORTAL.

Detect if OST_WRITE or OST_READ requests go to non-IO portals, as that
would LBUG the server due to lack of IO buffers.
b=11813
i=green
i=shadow

Accept the old ost_num_threads parameter but warn that it
is deprecated, and fix an off-by-one error that caused an LBUG.

Original patch by adilger.

b=12470
i=nathan
i=scjody

clear grant info when osc do resent and replay bulk write.
b=11662
i=adilger
i=green

Add LNET router traceability for debug purposes

If a checksum failure occurs with a router as part of the IO path, the NID of
the last router that forwarded the bulk data is printed so it can be
identified.

Original patch by eeb.

b=11548
i=eeb
i=adilger

r=adilger
Insure min dynamic threads < max threads
Also make absolutely sure we don't start more than MAX
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.
ilter.c lustre/ost/ost_handler.c
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.
    (8 * 1024)
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.
s/lustre-core/mds/handler.c,v
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.
threads = MDS_THREADS_AUTO_MAX;
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.
andler.c	21 Feb 2007 21:22:08 -0000
on the biggest machines, we limit the MDS auto-threads to start at 32 and go to 128, nowhere near the 512 max limit.  On e.g.

 - increase timeout for bulk PUT/GET a little bit

 - additional debug message to simplify data gathering from script

 - dump time ost_brw_write() spent waiting for bulk completion

add support for ost-authorization-revocation and fid verification.

 - additional debug info

- store node id in audit records on OSS
- move audit hook from filter_commitrw() to fsfilt_smfs_map_inode_pages()
- fix invalid nid in osc_get_info()
- rework stat/open audit hooks to store file lustre id, not parent one
- audit errors while stat/open

 - additional debug info

 - dedicated threads for serving OST_DESTROY and MDS_CLOSE

  5984 - oops in mdc_set_lock_data
         we must drop lookup lock in ll_clear_inode()
  7216 - abort due to missed clients
	 replay already committed (open/close rpcs) requests
	 even if some clients haven't connected in time and
	 got evicted
  7057 - sanity: test_66: /mnt/lustre/f66 blocks 0 < 8
         send close rpc only when all dirty pages are written
	 (otherwise i_blocks isn't valid)
  6817 - don't update attrs. on mds if the client hasn't modified file
         using openhandle being closed

  - change LUSTRE_MDC_NAME (and others) to form like OBD_MDC_DEVICENAME to be coherent with
    OBD_LMV_DEVICENAME and others. Because some OBDs used one form and others used another one.

  - removed different guesses of OBD types from name of some functions in cmobd. Say
    cmobd_find_master() now is called cmobd_find_obd() as all it does is finds obd
    by uuid and may be used not only for finding master. Also cmobd_lmv_obd() is renamed to
    cmobd_md_obd() etc. Improved function (a bit smaller of code) wich looks for master.

- in memory leaks detector do not remove memory track in case of freeing memory chunk different
  of size. marking it instead as such to show in final log of leaks with special hint.

- fixes in lconf. COBD and CMOBD should use LMV and LOV devices with diffrent names to avoid
  issues in the case when both they run on the same node (like in test 45) and try to setup
  and clanup the same device twise (from lconf).

create dir across MDS's, subdir need default ACL of its parent to determine
its own default ACL and access ACL. API obd_create() is added 2 params
to hold acl data.

 - to avoid possible lock collision during replay, we should replay all
   request before any locks

 - clean more files at clean time

 * remote uid/gid handling
 * various gss fixes
 * fix broken local ACL

remove unused files.

remove unnecessary error msg, cleanup