209  * 2) negotiate who will take over all the locks for the dead node.
212  *    and moves aside those mastered by the dead guy
223  *    everything and recovery for this dead node is done
224  *10) go back to 2) while there are still dead nodes
233 	mlog(ML_NOTICE, "%s(%d): recovery info, state=%s, dead=%u, master=%u\n",  in dlm_print_reco_node_status()
248 				st = "dead";  in dlm_print_reco_node_status()
316  * could be dead or just not joined */
319 	int dead;  in dlm_is_node_dead()  local
321 	dead = !test_bit(node, dlm->domain_map);  in dlm_is_node_dead()
323 	return dead;  in dlm_is_node_dead()
327  * could be dead or just not joined */
375  * recovering a dead node (as the new master or not) and clear
392 		     "state=%d, master=%u, dead=%u\n",  in dlm_wait_for_recovery()
422 	       "dead node %u in domain %s\n", dlm->reco.new_master,  in dlm_print_recovery_master()
501 	 * because all of the dead node's lock resources  in dlm_do_recovery()
530 	/* continue and look for another dead node */  in dlm_do_recovery()
583 					     "dead? %s\n", ndata->node_num,  in dlm_remaster_locks()
615 				     "node %u for dead node %u\n",  in dlm_remaster_locks()
620 				     "node %u for dead node %u\n",  in dlm_remaster_locks()
625 				     "from node %u for dead node %u\n",  in dlm_remaster_locks()
685 			 * a new recovery for another dead node start  in dlm_remaster_locks()
707 			     "dead=%u, this=%u, new=%u\n", dlm->name,  in dlm_remaster_locks()
789 	mlog(0, "dlm_request_all_locks: dead node is %u, sending request "  in dlm_request_all_locks()
803 		     "to recover dead node %u\n", dlm->name, ret,  in dlm_request_all_locks()
878 	mlog(0, "%s: recovery worker started, dead=%u, master=%u\n",  in dlm_request_all_locks_worker()
887 			     "recovery master %u died, thread=(dead=%u,mas=%u)"  in dlm_request_all_locks_worker()
888 			     " current=(dead=%u,mas=%u)\n", dlm->name,  in dlm_request_all_locks_worker()
892 			mlog(ML_NOTICE, "%s: reco state invalid: reco(dead=%u, "  in dlm_request_all_locks_worker()
893 			     "master=%u), request(dead=%u, master=%u)\n",  in dlm_request_all_locks_worker()
902  	 * to a temp list if the dead owner matches.  note that the  in dlm_request_all_locks_worker()
917 			     "recovery state for dead node %u, ret=%d\n", dlm->name,  in dlm_request_all_locks_worker()
933 			     "recovery all-done for dead node %u, ret=%d\n",  in dlm_request_all_locks_worker()
951 	     "my node=%u, dead node=%u\n", send_to, done_msg.node_idx,  in dlm_send_all_done_msg()
958 		     "to recover dead node %u\n", dlm->name, ret, send_to,  in dlm_send_all_done_msg()
1043 		/* always prune any $RECOVERY entries for dead nodes,  in dlm_move_reco_locks_to_list()
1051 					     "a $RECOVERY lock for dead "  in dlm_move_reco_locks_to_list()
1067 			mlog(0, "found lockres owned by dead node while "  in dlm_move_reco_locks_to_list()
2087 				 * (the master is dead). */  in dlm_move_lockres_to_recovery_list()
2222 		/* if this node owned the lockres, and if the dead node  in dlm_revalidate_lvb()
2246 		mlog(0, "clearing %.*s lvb, dead node %u had EX\n",  in dlm_revalidate_lvb()
2259 	 * 1) remove any stale locks for the dead node  in dlm_free_dead_locks()
2260 	 * 2) if the dead node had an EX when he died, blank out the lvb  in dlm_free_dead_locks()
2298 		mlog(0, "%s:%.*s: freed %u locks for dead node %u, "  in dlm_free_dead_locks()
2302 			mlog(ML_ERROR, "%s:%.*s: freed %u locks for dead node %u, "  in dlm_free_dead_locks()
2310 		mlog(0, "%s:%.*s: dead node %u had a ref, but had "  in dlm_free_dead_locks()
2335 	 * 1) if the dead node was the master, move the lockres  in dlm_do_local_recovery_cleanup()
2342 	 *    dead node.  once recovery finishes, the dlm thread  in dlm_do_local_recovery_cleanup()
2349  			/* always prune any $RECOVERY entries for dead nodes,  in dlm_do_local_recovery_cleanup()
2357 						     "a $RECOVERY lock for dead "  in dlm_do_local_recovery_cleanup()
2389 						"dead node %u, this node was "  in dlm_do_local_recovery_cleanup()
2407 					mlog(0, "%s:%.*s: dead node %u had a ref, but had "  in dlm_do_local_recovery_cleanup()
2431 			mlog(0, "%s: dead master %d had reached "  in __dlm_hb_node_down()
2444 	/* check to see if the node is already considered dead */  in __dlm_hb_node_down()
2446 		mlog(0, "for domain %s, node %d is already dead. "  in __dlm_hb_node_down()
2551 	mlog(0, "starting recovery of %s at %lu, dead=%u, this=%u\n",  in dlm_pick_recovery_master()
2585 					     "but no dead node!\n",  in dlm_pick_recovery_master()
2596 			mlog(0, "%s: dead=%u, this=%u, sending "  in dlm_pick_recovery_master()
2677 	mlog(0, "%s: dead node is %u\n", dlm->name, dead_node);  in dlm_send_begin_reco_message()
2692 			mlog(0, "not sending begin reco to dead node "  in dlm_send_begin_reco_message()
2790 			mlog(0, "%s: new_master %u NOT DEAD, changing "  in dlm_begin_reco_handler()
2793 			/* may not have seen the new master as dead yet */  in dlm_begin_reco_handler()
2804 		mlog(0, "recovery master %u sees %u as dead, but this "  in dlm_begin_reco_handler()
2805 		     "node has not yet.  marking %u as dead\n",  in dlm_begin_reco_handler()
2909 		     "%u is supposed to be the new master, dead=%u\n",  in dlm_finalize_reco_handler()
2914 		mlog(ML_ERROR, "node %u sent recovery finalize msg for dead "  in dlm_finalize_reco_handler()
2915 		     "node %u, but node %u is supposed to be dead\n",  in dlm_finalize_reco_handler()
2925 				     "new master %u for dead node %u, but "  in dlm_finalize_reco_handler()
2937 				     "new master %u for dead node %u, but "  in dlm_finalize_reco_handler()
2950 	mlog(0, "%s: recovery done, reco master was %u, dead now %u, master now %u\n",  in dlm_finalize_reco_handler()