references.
b=16450

first instance of a given type is created and last one is destroyed
respectively. This is need by CLIO.
b=16450

some system threads (ptlrpcd) are shared by multiple modules, and so cannot be
stopped during module shutdown.
b=16450

sufficient to locate and load all object state (inode). On a client, on the
other hand, more data are necessary instantiate an object. Change
lu_object_find() and friends to take additional `lu_conf' argument describing
object.  Typically this includes layout information.
b=16450

b=16450

b=16450

takes more than one ->lgh_lock lock.
b=16450

ldlm_cancel_shrink_policy() to estimate lock "value", instead of hard-coded
`number of pages' logic.
b=16450

addref a lock that might be being canceled concurrently.
b=16450

        lu_ref fields ->l_reference and ->lr_reference are added to ldlm_lock
        and ldlm_resource. LDLM interface has to be changed, because code that
        releases a reference on a lock, has to "know" what reference this is.
        In the most frequent case

                lock = ldlm_handle2lock(handle);
                ...
                LDLM_LOCK_PUT(lock);

        no changes are required. When any other reference (received _not_ from
        ldlm_handle2lock()) is released, LDLM_LOCK_RELEASE() has to be called
        instead of LDLM_LOCK_PUT().

        Arguably, changes are pervasive, and interface requires some discipline
        for proper use. On the other hand, it was very instrumental in finding
        a few leaked lock references.
b=16450

b=16450

errno values. CLIO needs this to prevent LDLM specific constants from escaping
ldlm and osc.
b=16450

ldlm_completion_ast() cannot handle as it waits until lock is
granted. Introduce new ldlm_completion_ast_async() for this. Also comment
ldlm_completion_ast().
b=16450

memory---a requirement implemented through ->o_join_lru() obd method in
HEAD. Remove this method and its users.
b=16450

wrap them into struct ldlm_callback_suite.
b=16450

is locked simultaneously. Also, fix possible dead-lock in
ldlm_lock_change_resource() by enforcing particular lock ordering.
b=16450

state.
b=16450

switch. This macro is to be used for consistency and sanity checks that are
too expensive to be left in `production' mode.
b=16450

situations where Lustre has to lock more than one inode mutex at a time.
b=16450

implemented---another is checked in yet to be landed core CLIO code.
b=16450

lock ordering. Update mdd code to use new locking interface.
b=16450

b=16450

the code.
b=16450

b=16450

b=16450

Fix comment.

Use $RM macro for portability.
Make it more clear when sub-makes are finished.

i=Adilger
fix for remote [mds|ost] with nodsh

i=Adilger
fix for remote [mds|ost] with nodsh

i=Brian
create machinefile on $TMP, cleanup machinefile

 b=12521
 To avoid extent lock conflicts, if avail_cb_nodes < stripe_count*CO,
 avail_cb_nodes should divide (stripe_count*CO) exactly. So that each OST
 can be accessed by one or more constant clients.

b=17403

i=adilger
i=yury.umanets
create objects in correct directory on OST.

b=17199
i=johann, bobijam
Patch provide by LLNL.

Skip dumping log if panic_on_lbug is set.

b=17038
i=johann

regression test case for getxattr upon symlink file.

i=Adilger
replace lustre proc by lctl [set|get]_param

 b=12521
 - set Lustre hints (except striping hints) "anywhere"
 - perform collective I/O for interleaving, no matter how big the req size is
 - keep the semantic information of cb_nodes
 - perform collective I/O by the same client if the whole file access portion
   is no bigger than stripe size and cb_nodes is changed by the user
 - fix some bugs in the error handling
 - remove redundant codes

i=Nathan
fix obsolete run_one CLEANUP

i=Johann
skip some tests if there are several ost services on oss node