Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 1 | /* -*- mode: c; c-basic-offset: 8; -*- |
| 2 | * vim: noexpandtab sw=8 ts=8 sts=0: |
| 3 | * |
| 4 | * heartbeat.c |
| 5 | * |
| 6 | * Register ourselves with the heartbaet service, keep our node maps |
| 7 | * up to date, and fire off recovery when needed. |
| 8 | * |
| 9 | * Copyright (C) 2002, 2004 Oracle. All rights reserved. |
| 10 | * |
| 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. |
| 15 | * |
| 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. |
| 20 | * |
| 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. |
| 25 | */ |
| 26 | |
| 27 | #include <linux/fs.h> |
| 28 | #include <linux/types.h> |
| 29 | #include <linux/slab.h> |
| 30 | #include <linux/highmem.h> |
| 31 | #include <linux/kmod.h> |
| 32 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 33 | #define MLOG_MASK_PREFIX ML_SUPER |
| 34 | #include <cluster/masklog.h> |
| 35 | |
| 36 | #include "ocfs2.h" |
| 37 | |
| 38 | #include "alloc.h" |
| 39 | #include "heartbeat.h" |
| 40 | #include "inode.h" |
| 41 | #include "journal.h" |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 42 | |
| 43 | #include "buffer_head_io.h" |
| 44 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 45 | static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map, |
| 46 | int bit); |
| 47 | static inline void __ocfs2_node_map_clear_bit(struct ocfs2_node_map *map, |
| 48 | int bit); |
Adrian Bunk | 0060005 | 2008-01-29 00:11:41 +0200 | [diff] [blame] | 49 | |
| 50 | /* special case -1 for now |
| 51 | * TODO: should *really* make sure the calling func never passes -1!! */ |
| 52 | static void ocfs2_node_map_init(struct ocfs2_node_map *map) |
| 53 | { |
| 54 | map->num_nodes = OCFS2_NODE_MAP_MAX_NODES; |
| 55 | memset(map->map, 0, BITS_TO_LONGS(OCFS2_NODE_MAP_MAX_NODES) * |
| 56 | sizeof(unsigned long)); |
| 57 | } |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 58 | |
| 59 | void ocfs2_init_node_maps(struct ocfs2_super *osb) |
| 60 | { |
| 61 | spin_lock_init(&osb->node_map_lock); |
Mark Fasheh | b4df6ed | 2006-02-22 17:35:08 -0800 | [diff] [blame] | 62 | ocfs2_node_map_init(&osb->osb_recovering_orphan_dirs); |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 63 | } |
| 64 | |
Joel Becker | 4670c46 | 2008-02-01 14:39:35 -0800 | [diff] [blame^] | 65 | void ocfs2_do_node_down(int node_num, void *data) |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 66 | { |
Joel Becker | 4670c46 | 2008-02-01 14:39:35 -0800 | [diff] [blame^] | 67 | struct ocfs2_super *osb = data; |
| 68 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 69 | BUG_ON(osb->node_num == node_num); |
| 70 | |
| 71 | mlog(0, "ocfs2: node down event for %d\n", node_num); |
| 72 | |
Joel Becker | 4670c46 | 2008-02-01 14:39:35 -0800 | [diff] [blame^] | 73 | if (!osb->cconn) { |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 74 | /* |
Joel Becker | 4670c46 | 2008-02-01 14:39:35 -0800 | [diff] [blame^] | 75 | * No cluster connection means we're not even ready to |
| 76 | * participate yet. We check the slots after the cluster |
| 77 | * comes up, so we will notice the node death then. We |
| 78 | * can safely ignore it here. |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 79 | */ |
| 80 | return; |
| 81 | } |
| 82 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 83 | ocfs2_recovery_thread(osb, node_num); |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 84 | } |
| 85 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 86 | void ocfs2_stop_heartbeat(struct ocfs2_super *osb) |
| 87 | { |
| 88 | int ret; |
| 89 | char *argv[5], *envp[3]; |
| 90 | |
Sunil Mushran | c271c5c | 2006-12-05 17:56:35 -0800 | [diff] [blame] | 91 | if (ocfs2_mount_local(osb)) |
| 92 | return; |
| 93 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 94 | if (!osb->uuid_str) { |
| 95 | /* This can happen if we don't get far enough in mount... */ |
| 96 | mlog(0, "No UUID with which to stop heartbeat!\n\n"); |
| 97 | return; |
| 98 | } |
| 99 | |
| 100 | argv[0] = (char *)o2nm_get_hb_ctl_path(); |
| 101 | argv[1] = "-K"; |
| 102 | argv[2] = "-u"; |
| 103 | argv[3] = osb->uuid_str; |
| 104 | argv[4] = NULL; |
| 105 | |
| 106 | mlog(0, "Run: %s %s %s %s\n", argv[0], argv[1], argv[2], argv[3]); |
| 107 | |
| 108 | /* minimal command environment taken from cpu_run_sbin_hotplug */ |
| 109 | envp[0] = "HOME=/"; |
| 110 | envp[1] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin"; |
| 111 | envp[2] = NULL; |
| 112 | |
Jeremy Fitzhardinge | 86313c4 | 2007-07-17 18:37:03 -0700 | [diff] [blame] | 113 | ret = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC); |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 114 | if (ret < 0) |
| 115 | mlog_errno(ret); |
| 116 | } |
| 117 | |
Mark Fasheh | ccd979b | 2005-12-15 14:31:24 -0800 | [diff] [blame] | 118 | static inline void __ocfs2_node_map_set_bit(struct ocfs2_node_map *map, |
| 119 | int bit) |
| 120 | { |
| 121 | set_bit(bit, map->map); |
| 122 | } |
| 123 | |
| 124 | void ocfs2_node_map_set_bit(struct ocfs2_super *osb, |
| 125 | struct ocfs2_node_map *map, |
| 126 | int bit) |
| 127 | { |
| 128 | if (bit==-1) |
| 129 | return; |
| 130 | BUG_ON(bit >= map->num_nodes); |
| 131 | spin_lock(&osb->node_map_lock); |
| 132 | __ocfs2_node_map_set_bit(map, bit); |
| 133 | spin_unlock(&osb->node_map_lock); |
| 134 | } |
| 135 | |
| 136 | static inline void __ocfs2_node_map_clear_bit(struct ocfs2_node_map *map, |
| 137 | int bit) |
| 138 | { |
| 139 | clear_bit(bit, map->map); |
| 140 | } |
| 141 | |
| 142 | void ocfs2_node_map_clear_bit(struct ocfs2_super *osb, |
| 143 | struct ocfs2_node_map *map, |
| 144 | int bit) |
| 145 | { |
| 146 | if (bit==-1) |
| 147 | return; |
| 148 | BUG_ON(bit >= map->num_nodes); |
| 149 | spin_lock(&osb->node_map_lock); |
| 150 | __ocfs2_node_map_clear_bit(map, bit); |
| 151 | spin_unlock(&osb->node_map_lock); |
| 152 | } |
| 153 | |
| 154 | int ocfs2_node_map_test_bit(struct ocfs2_super *osb, |
| 155 | struct ocfs2_node_map *map, |
| 156 | int bit) |
| 157 | { |
| 158 | int ret; |
| 159 | if (bit >= map->num_nodes) { |
| 160 | mlog(ML_ERROR, "bit=%d map->num_nodes=%d\n", bit, map->num_nodes); |
| 161 | BUG(); |
| 162 | } |
| 163 | spin_lock(&osb->node_map_lock); |
| 164 | ret = test_bit(bit, map->map); |
| 165 | spin_unlock(&osb->node_map_lock); |
| 166 | return ret; |
| 167 | } |
| 168 | |