Branch b1_8_gate

b=15825
o=green
i=shadow, bobijam

Description: Kernel BUG tries to release flock
Details    : Lustre does not destroy flock lock before last reference goes
             away. So always drop flock locks when client is evicted and
             perform unlock regardless of successfulness of speaking to MDS.

diff --git a/lustre/ldlm/ldlm_flock.c b/lustre/ldlm/ldlm_flock.c
index 1ec8b74..2877a3e 100644 (file)
--- a/lustre/ldlm/ldlm_flock.c
+++ b/lustre/ldlm/ldlm_flock.c
@@ -105,7 +105,7 @@ ldlm_flock_destroy(struct ldlm_lock *lock, ldlm_mode_t mode, int flags)
                 /* client side - set a flag to prevent sending a CANCEL */
                 lock->l_flags |= LDLM_FL_LOCAL_ONLY | LDLM_FL_CBPENDING;
 
-                /* when reaching here, it is under lock_res_and_lock(). Thus, 
+                /* when reaching here, it is under lock_res_and_lock(). Thus,
                    need call the nolock version of ldlm_lock_decref_internal*/
                 ldlm_lock_decref_internal_nolock(lock, mode);
         }
@@ -423,7 +423,7 @@ reprocess:
         if (*flags != LDLM_FL_WAIT_NOREPROC) {
                 if (first_enq) {
                         /* If this is an unlock, reprocess the waitq and
-                         * send completions ASTs for locks that can now be 
+                         * send completions ASTs for locks that can now be
                          * granted. The only problem with doing this
                          * reprocessing here is that the completion ASTs for
                          * newly granted locks will be sent before the unlock
@@ -515,7 +515,7 @@ ldlm_flock_completion_ast(struct ldlm_lock *lock, int flags, void *data)
          * holding the lock even if app still believes it has it, since
          * server already dropped it anyway. Only for granted locks too. */
         lock_res_and_lock(lock);
-        if ((lock->l_flags & (LDLM_FL_FAILED|LDLM_FL_LOCAL_ONLY)) == 
+        if ((lock->l_flags & (LDLM_FL_FAILED|LDLM_FL_LOCAL_ONLY)) ==
             (LDLM_FL_FAILED|LDLM_FL_LOCAL_ONLY)) {
                 unlock_res_and_lock(lock);
                 if (lock->l_req_mode == lock->l_granted_mode &&
@@ -556,7 +556,7 @@ ldlm_flock_completion_ast(struct ldlm_lock *lock, int flags, void *data)
 
         LDLM_DEBUG(lock, "client-side enqueue waking up: rc = %d", rc);
         RETURN(rc);
- 
+
 granted:
         OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_CP_CB_WAIT, 10);
         LDLM_DEBUG(lock, "client-side enqueue granted");