1 #ifndef _FS_CEPH_OSDMAP_H
2 #define _FS_CEPH_OSDMAP_H
4 #include <linux/rbtree.h>
5 #include <linux/ceph/types.h>
6 #include <linux/ceph/decode.h>
7 #include <linux/ceph/ceph_fs.h>
8 #include <linux/crush/crush.h>
11 * The osd map describes the current membership of the osd cluster and
12 * specifies the mapping of objects to placement groups and placement
13 * groups to (sets of) osds. That is, it completely specifies the
14 * (desired) distribution of all data objects in the system at some
17 * Each map version is identified by an epoch, which increases monotonically.
19 * The map can be updated either via an incremental map (diff) describing
20 * the change between two successive epochs, or as a fully encoded map.
27 #define CEPH_POOL_FLAG_HASHPSPOOL 1
29 struct ceph_pg_pool_info {
37 int pg_num_mask, pgp_num_mask;
39 s64 write_tier; /* wins for read+write ops */
44 struct ceph_object_locator {
49 * Maximum supported by kernel client object name length
51 * (probably outdated: must be >= RBD_MAX_MD_NAME_LEN -- currently 100)
53 #define CEPH_MAX_OID_NAME_LEN 100
55 struct ceph_object_id {
56 char name[CEPH_MAX_OID_NAME_LEN];
60 struct ceph_pg_mapping {
68 struct ceph_fsid fsid;
71 struct ceph_timespec created, modified;
73 u32 flags; /* CEPH_OSDMAP_* */
75 u32 max_osd; /* size of osd_state, _offload, _addr arrays */
76 u8 *osd_state; /* CEPH_OSD_* */
77 u32 *osd_weight; /* 0 = failed, 0x10000 = 100% normal */
78 struct ceph_entity_addr *osd_addr;
80 struct rb_root pg_temp;
81 struct rb_root pg_pools;
84 /* the CRUSH map specifies the mapping of placement groups to
85 * the list of osds that store+replicate them. */
86 struct crush_map *crush;
89 static inline void ceph_oid_set_name(struct ceph_object_id *oid,
95 if (len > sizeof(oid->name)) {
96 WARN(1, "ceph_oid_set_name '%s' len %d vs %zu, truncating\n",
97 name, len, sizeof(oid->name));
98 len = sizeof(oid->name);
101 memcpy(oid->name, name, len);
105 static inline void ceph_oid_copy(struct ceph_object_id *dest,
106 struct ceph_object_id *src)
108 BUG_ON(src->name_len > sizeof(dest->name));
109 memcpy(dest->name, src->name, src->name_len);
110 dest->name_len = src->name_len;
113 static inline int ceph_osd_is_up(struct ceph_osdmap *map, int osd)
115 return (osd < map->max_osd) && (map->osd_state[osd] & CEPH_OSD_UP);
118 static inline bool ceph_osdmap_flag(struct ceph_osdmap *map, int flag)
120 return map && (map->flags & flag);
123 extern char *ceph_osdmap_state_str(char *str, int len, int state);
125 static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map,
128 if (osd >= map->max_osd)
130 return &map->osd_addr[osd];
133 static inline int ceph_decode_pgid(void **p, void *end, struct ceph_pg *pgid)
137 if (!ceph_has_room(p, end, 1 + 8 + 4 + 4)) {
138 pr_warning("incomplete pg encoding");
142 version = ceph_decode_8(p);
144 pr_warning("do not understand pg encoding %d > 1",
149 pgid->pool = ceph_decode_64(p);
150 pgid->seed = ceph_decode_32(p);
151 *p += 4; /* skip deprecated preferred value */
156 extern struct ceph_osdmap *osdmap_decode(void **p, void *end);
157 extern struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
158 struct ceph_osdmap *map,
159 struct ceph_messenger *msgr);
160 extern void ceph_osdmap_destroy(struct ceph_osdmap *map);
162 /* calculate mapping of a file extent to an object */
163 extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout,
165 u64 *bno, u64 *oxoff, u64 *oxlen);
167 /* calculate mapping of object to a placement group */
168 extern int ceph_oloc_oid_to_pg(struct ceph_osdmap *osdmap,
169 struct ceph_object_locator *oloc,
170 struct ceph_object_id *oid,
171 struct ceph_pg *pg_out);
173 extern int ceph_calc_pg_acting(struct ceph_osdmap *osdmap,
176 extern int ceph_calc_pg_primary(struct ceph_osdmap *osdmap,
177 struct ceph_pg pgid);
179 extern struct ceph_pg_pool_info *ceph_pg_pool_by_id(struct ceph_osdmap *map,
182 extern const char *ceph_pg_pool_name_by_id(struct ceph_osdmap *map, u64 id);
183 extern int ceph_pg_poolid_by_name(struct ceph_osdmap *map, const char *name);