X-Git-Url: http://ftp.safe.ca/?a=blobdiff_plain;f=fs%2Fdlm%2Frecoverd.c;h=fd677c8c3d3b041396817bc930bde7e5e5b0be87;hb=366ec51ba9622c44191fcf4d77ed4dc84acbdeec;hp=70103533677d9197fdf3355bdb76317dc8d61976;hpb=901359256b2666f52a3a7d3f31927677e91b3a2a;p=safe%2Fjmp%2Flinux-2.6

diff --git a/fs/dlm/recoverd.c b/fs/dlm/recoverd.c
index 7010353..fd677c8 100644
--- a/fs/dlm/recoverd.c
+++ b/fs/dlm/recoverd.c
@@ -2,7 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
-**  Copyright (C) 2004-2005 Red Hat, Inc.  All rights reserved.
+**  Copyright (C) 2004-2007 Red Hat, Inc.  All rights reserved.
 **
 **  This copyrighted material is made available to anyone wishing to use,
 **  modify, copy, or redistribute it subject to the terms and conditions
@@ -24,19 +24,28 @@
 
 
 /* If the start for which we're re-enabling locking (seq) has been superseded
-   by a newer stop (ls_recover_seq), we need to leave locking disabled. */
+   by a newer stop (ls_recover_seq), we need to leave locking disabled.
+
+   We suspend dlm_recv threads here to avoid the race where dlm_recv a) sees
+   locking stopped and b) adds a message to the requestqueue, but dlm_recoverd
+   enables locking and clears the requestqueue between a and b. */
 
 static int enable_locking(struct dlm_ls *ls, uint64_t seq)
 {
 	int error = -EINTR;
 
+	down_write(&ls->ls_recv_active);
+
 	spin_lock(&ls->ls_recover_lock);
 	if (ls->ls_recover_seq == seq) {
 		set_bit(LSFL_RUNNING, &ls->ls_flags);
+		/* unblocks processes waiting to enter the dlm */
 		up_write(&ls->ls_in_recovery);
 		error = 0;
 	}
 	spin_unlock(&ls->ls_recover_lock);
+
+	up_write(&ls->ls_recv_active);
 	return error;
 }
 
@@ -45,7 +54,7 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 	unsigned long start;
 	int error, neg = 0;
 
-	log_debug(ls, "recover %llx", rv->seq);
+	log_debug(ls, "recover %llx", (unsigned long long)rv->seq);
 
 	mutex_lock(&ls->ls_recoverd_active);
 
@@ -58,17 +67,18 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 	dlm_astd_resume();
 
 	/*
-	 * This list of root rsb's will be the basis of most of the recovery
-	 * routines.
+	 * Free non-master tossed rsb's.  Master rsb's are kept on toss
+	 * list and put on root list to be included in resdir recovery.
 	 */
 
-	dlm_create_root_list(ls);
+	dlm_clear_toss_list(ls);
 
 	/*
-	 * Free all the tossed rsb's so we don't have to recover them.
+	 * This list of root rsb's will be the basis of most of the recovery
+	 * routines.
 	 */
 
-	dlm_clear_toss_list(ls);
+	dlm_create_root_list(ls);
 
 	/*
 	 * Add or remove nodes from the lockspace's ls_nodes list.
@@ -77,7 +87,7 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 	error = dlm_recover_members(ls, rv, &neg);
 	if (error) {
-		log_error(ls, "recover_members failed %d", error);
+		log_debug(ls, "recover_members failed %d", error);
 		goto fail;
 	}
 	start = jiffies;
@@ -89,25 +99,17 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 	error = dlm_recover_directory(ls);
 	if (error) {
-		log_error(ls, "recover_directory failed %d", error);
+		log_debug(ls, "recover_directory failed %d", error);
 		goto fail;
 	}
 
 	/*
-	 * Purge directory-related requests that are saved in requestqueue.
-	 * All dir requests from before recovery are invalid now due to the dir
-	 * rebuild and will be resent by the requesting nodes.
-	 */
-
-	dlm_purge_requestqueue(ls);
-
-	/*
 	 * Wait for all nodes to complete directory rebuild.
 	 */
 
 	error = dlm_recover_directory_wait(ls);
 	if (error) {
-		log_error(ls, "recover_directory_wait failed %d", error);
+		log_debug(ls, "recover_directory_wait failed %d", error);
 		goto fail;
 	}
 
@@ -137,7 +139,7 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 		error = dlm_recover_masters(ls);
 		if (error) {
-			log_error(ls, "recover_masters failed %d", error);
+			log_debug(ls, "recover_masters failed %d", error);
 			goto fail;
 		}
 
@@ -147,13 +149,13 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 		error = dlm_recover_locks(ls);
 		if (error) {
-			log_error(ls, "recover_locks failed %d", error);
+			log_debug(ls, "recover_locks failed %d", error);
 			goto fail;
 		}
 
 		error = dlm_recover_locks_wait(ls);
 		if (error) {
-			log_error(ls, "recover_locks_wait failed %d", error);
+			log_debug(ls, "recover_locks_wait failed %d", error);
 			goto fail;
 		}
 
@@ -164,34 +166,57 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 		 */
 
 		dlm_recover_rsbs(ls);
+	} else {
+		/*
+		 * Other lockspace members may be going through the "neg" steps
+		 * while also adding us to the lockspace, in which case they'll
+		 * be doing the recover_locks (RS_LOCKS) barrier.
+		 */
+		dlm_set_recover_status(ls, DLM_RS_LOCKS);
+
+		error = dlm_recover_locks_wait(ls);
+		if (error) {
+			log_debug(ls, "recover_locks_wait failed %d", error);
+			goto fail;
+		}
 	}
 
 	dlm_release_root_list(ls);
 
+	/*
+	 * Purge directory-related requests that are saved in requestqueue.
+	 * All dir requests from before recovery are invalid now due to the dir
+	 * rebuild and will be resent by the requesting nodes.
+	 */
+
+	dlm_purge_requestqueue(ls);
+
 	dlm_set_recover_status(ls, DLM_RS_DONE);
 	error = dlm_recover_done_wait(ls);
 	if (error) {
-		log_error(ls, "recover_done_wait failed %d", error);
+		log_debug(ls, "recover_done_wait failed %d", error);
 		goto fail;
 	}
 
 	dlm_clear_members_gone(ls);
 
+	dlm_adjust_timeouts(ls);
+
 	error = enable_locking(ls, rv->seq);
 	if (error) {
-		log_error(ls, "enable_locking failed %d", error);
+		log_debug(ls, "enable_locking failed %d", error);
 		goto fail;
 	}
 
 	error = dlm_process_requestqueue(ls);
 	if (error) {
-		log_error(ls, "process_requestqueue failed %d", error);
+		log_debug(ls, "process_requestqueue failed %d", error);
 		goto fail;
 	}
 
 	error = dlm_recover_waiters_post(ls);
 	if (error) {
-		log_error(ls, "recover_waiters_post failed %d", error);
+		log_debug(ls, "recover_waiters_post failed %d", error);
 		goto fail;
 	}
 
@@ -199,7 +224,8 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 	dlm_astd_wake();
 
-	log_debug(ls, "recover %llx done: %u ms", rv->seq,
+	log_debug(ls, "recover %llx done: %u ms",
+		  (unsigned long long)rv->seq,
 		  jiffies_to_msecs(jiffies - start));
 	mutex_unlock(&ls->ls_recoverd_active);
 
@@ -207,11 +233,16 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
  fail:
 	dlm_release_root_list(ls);
-	log_debug(ls, "recover %llx error %d", rv->seq, error);
+	log_debug(ls, "recover %llx error %d",
+		  (unsigned long long)rv->seq, error);
 	mutex_unlock(&ls->ls_recoverd_active);
 	return error;
 }
 
+/* The dlm_ls_start() that created the rv we take here may already have been
+   stopped via dlm_ls_stop(); in that case we need to leave the RECOVERY_STOP
+   flag set. */
+
 static void do_ls_recovery(struct dlm_ls *ls)
 {
 	struct dlm_recover *rv = NULL;
@@ -219,12 +250,14 @@ static void do_ls_recovery(struct dlm_ls *ls)
 	spin_lock(&ls->ls_recover_lock);
 	rv = ls->ls_recover_args;
 	ls->ls_recover_args = NULL;
-	clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags);
+	if (rv && ls->ls_recover_seq == rv->seq)
+		clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags);
 	spin_unlock(&ls->ls_recover_lock);
 
 	if (rv) {
 		ls_recover(ls, rv);
 		kfree(rv->nodeids);
+		kfree(rv->new);
 		kfree(rv);
 	}
 }
@@ -234,6 +267,10 @@ static int dlm_recoverd(void *arg)
 	struct dlm_ls *ls;
 
 	ls = dlm_find_lockspace_local(arg);
+	if (!ls) {
+		log_print("dlm_recoverd: no lockspace %p", arg);
+		return -1;
+	}
 
 	while (!kthread_should_stop()) {
 		set_current_state(TASK_INTERRUPTIBLE);
@@ -275,6 +312,7 @@ void dlm_recoverd_stop(struct dlm_ls *ls)
 
 void dlm_recoverd_suspend(struct dlm_ls *ls)
 {
+	wake_up(&ls->ls_wait_general);
 	mutex_lock(&ls->ls_recoverd_active);
 }