1 /* -*- mode: c; c-basic-offset: 8; -*-
2 * vim: noexpandtab sw=8 ts=8 sts=0:
6 * Register ourselves with the heartbaet service, keep our node maps
7 * up to date, and fire off recovery when needed.
9 * Copyright (C) 2002, 2004 Oracle. All rights reserved.
11 * This program is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU General Public
13 * License as published by the Free Software Foundation; either
14 * version 2 of the License, or (at your option) any later version.
16 * This program is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * General Public License for more details.
21 * You should have received a copy of the GNU General Public
22 * License along with this program; if not, write to the
23 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
24 * Boston, MA 021110-1307, USA.
28 #include <linux/types.h>
29 #include <linux/slab.h>
30 #include <linux/highmem.h>
31 #include <linux/kmod.h>
33 #include <dlm/dlmapi.h>
35 #define MLOG_MASK_PREFIX ML_SUPER
36 #include <cluster/masklog.h>
41 #include "heartbeat.h"
45 #include "buffer_head_io.h"
47 static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map,
49 static inline void __ocfs2_node_map_clear_bit(struct ocfs2_node_map *map,
51 static inline int __ocfs2_node_map_is_empty(struct ocfs2_node_map *map);
52 static void __ocfs2_node_map_dup(struct ocfs2_node_map *target,
53 struct ocfs2_node_map *from);
54 static void __ocfs2_node_map_set(struct ocfs2_node_map *target,
55 struct ocfs2_node_map *from);
57 void ocfs2_init_node_maps(struct ocfs2_super *osb)
59 spin_lock_init(&osb->node_map_lock);
60 ocfs2_node_map_init(&osb->recovery_map);
61 ocfs2_node_map_init(&osb->osb_recovering_orphan_dirs);
64 static void ocfs2_do_node_down(int node_num,
65 struct ocfs2_super *osb)
67 BUG_ON(osb->node_num == node_num);
69 mlog(0, "ocfs2: node down event for %d\n", node_num);
73 * No DLM means we're not even ready to participate yet.
74 * We check the slots after the DLM comes up, so we will
75 * notice the node death then. We can safely ignore it
81 ocfs2_recovery_thread(osb, node_num);
84 /* Called from the dlm when it's about to evict a node. We may also
85 * get a heartbeat callback later. */
86 static void ocfs2_dlm_eviction_cb(int node_num,
89 struct ocfs2_super *osb = (struct ocfs2_super *) data;
90 struct super_block *sb = osb->sb;
92 mlog(ML_NOTICE, "device (%u,%u): dlm has evicted node %d\n",
93 MAJOR(sb->s_dev), MINOR(sb->s_dev), node_num);
95 ocfs2_do_node_down(node_num, osb);
98 void ocfs2_setup_hb_callbacks(struct ocfs2_super *osb)
100 /* Not exactly a heartbeat callback, but leads to essentially
101 * the same path so we set it up here. */
102 dlm_setup_eviction_cb(&osb->osb_eviction_cb,
103 ocfs2_dlm_eviction_cb,
107 void ocfs2_stop_heartbeat(struct ocfs2_super *osb)
110 char *argv[5], *envp[3];
112 if (ocfs2_mount_local(osb))
115 if (!osb->uuid_str) {
116 /* This can happen if we don't get far enough in mount... */
117 mlog(0, "No UUID with which to stop heartbeat!\n\n");
121 argv[0] = (char *)o2nm_get_hb_ctl_path();
124 argv[3] = osb->uuid_str;
127 mlog(0, "Run: %s %s %s %s\n", argv[0], argv[1], argv[2], argv[3]);
129 /* minimal command environment taken from cpu_run_sbin_hotplug */
131 envp[1] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin";
134 ret = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
139 /* special case -1 for now
140 * TODO: should *really* make sure the calling func never passes -1!! */
141 void ocfs2_node_map_init(struct ocfs2_node_map *map)
143 map->num_nodes = OCFS2_NODE_MAP_MAX_NODES;
144 memset(map->map, 0, BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES) *
145 sizeof(unsigned long));
148 static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map,
151 set_bit(bit, map->map);
154 void ocfs2_node_map_set_bit(struct ocfs2_super *osb,
155 struct ocfs2_node_map *map,
160 BUG_ON(bit >= map->num_nodes);
161 spin_lock(&osb->node_map_lock);
162 __ocfs2_node_map_set_bit(map, bit);
163 spin_unlock(&osb->node_map_lock);
166 static inline void __ocfs2_node_map_clear_bit(struct ocfs2_node_map *map,
169 clear_bit(bit, map->map);
172 void ocfs2_node_map_clear_bit(struct ocfs2_super *osb,
173 struct ocfs2_node_map *map,
178 BUG_ON(bit >= map->num_nodes);
179 spin_lock(&osb->node_map_lock);
180 __ocfs2_node_map_clear_bit(map, bit);
181 spin_unlock(&osb->node_map_lock);
184 int ocfs2_node_map_test_bit(struct ocfs2_super *osb,
185 struct ocfs2_node_map *map,
189 if (bit >= map->num_nodes) {
190 mlog(ML_ERROR, "bit=%d map->num_nodes=%d\n", bit, map->num_nodes);
193 spin_lock(&osb->node_map_lock);
194 ret = test_bit(bit, map->map);
195 spin_unlock(&osb->node_map_lock);
199 static inline int __ocfs2_node_map_is_empty(struct ocfs2_node_map *map)
202 bit = find_next_bit(map->map, map->num_nodes, 0);
203 if (bit < map->num_nodes)
208 int ocfs2_node_map_is_empty(struct ocfs2_super *osb,
209 struct ocfs2_node_map *map)
212 BUG_ON(map->num_nodes == 0);
213 spin_lock(&osb->node_map_lock);
214 ret = __ocfs2_node_map_is_empty(map);
215 spin_unlock(&osb->node_map_lock);
219 static void __ocfs2_node_map_dup(struct ocfs2_node_map *target,
220 struct ocfs2_node_map *from)
222 BUG_ON(from->num_nodes == 0);
223 ocfs2_node_map_init(target);
224 __ocfs2_node_map_set(target, from);
227 /* returns 1 if bit is the only bit set in target, 0 otherwise */
228 int ocfs2_node_map_is_only(struct ocfs2_super *osb,
229 struct ocfs2_node_map *target,
232 struct ocfs2_node_map temp;
235 spin_lock(&osb->node_map_lock);
236 __ocfs2_node_map_dup(&temp, target);
237 __ocfs2_node_map_clear_bit(&temp, bit);
238 ret = __ocfs2_node_map_is_empty(&temp);
239 spin_unlock(&osb->node_map_lock);
244 static void __ocfs2_node_map_set(struct ocfs2_node_map *target,
245 struct ocfs2_node_map *from)
249 BUG_ON(target->num_nodes != from->num_nodes);
250 BUG_ON(target->num_nodes == 0);
252 num_longs = BITS_TO_LONGS(target->num_nodes);
253 for (i = 0; i < num_longs; i++)
254 target->map[i] = from->map[i];
257 /* Returns whether the recovery bit was actually set - it may not be
258 * if a node is still marked as needing recovery */
259 int ocfs2_recovery_map_set(struct ocfs2_super *osb,
264 spin_lock(&osb->node_map_lock);
266 if (!test_bit(num, osb->recovery_map.map)) {
267 __ocfs2_node_map_set_bit(&osb->recovery_map, num);
271 spin_unlock(&osb->node_map_lock);
276 void ocfs2_recovery_map_clear(struct ocfs2_super *osb,
279 ocfs2_node_map_clear_bit(osb, &osb->recovery_map, num);
282 int ocfs2_node_map_iterate(struct ocfs2_super *osb,
283 struct ocfs2_node_map *map,
288 idx = O2NM_INVALID_NODE_NUM;
289 spin_lock(&osb->node_map_lock);
290 if ((i != O2NM_INVALID_NODE_NUM) &&
292 (i < map->num_nodes)) {
293 while(i < map->num_nodes) {
294 if (test_bit(i, map->map)) {
301 spin_unlock(&osb->node_map_lock);