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;
42 struct ceph_object_locator {
47 struct ceph_pg_mapping {
55 struct ceph_fsid fsid;
58 struct ceph_timespec created, modified;
60 u32 flags; /* CEPH_OSDMAP_* */
62 u32 max_osd; /* size of osd_state, _offload, _addr arrays */
63 u8 *osd_state; /* CEPH_OSD_* */
64 u32 *osd_weight; /* 0 = failed, 0x10000 = 100% normal */
65 struct ceph_entity_addr *osd_addr;
67 struct rb_root pg_temp;
68 struct rb_root pg_pools;
71 /* the CRUSH map specifies the mapping of placement groups to
72 * the list of osds that store+replicate them. */
73 struct crush_map *crush;
79 #define ceph_file_layout_su(l) ((__s32)le32_to_cpu((l).fl_stripe_unit))
80 #define ceph_file_layout_stripe_count(l) \
81 ((__s32)le32_to_cpu((l).fl_stripe_count))
82 #define ceph_file_layout_object_size(l) ((__s32)le32_to_cpu((l).fl_object_size))
83 #define ceph_file_layout_cas_hash(l) ((__s32)le32_to_cpu((l).fl_cas_hash))
84 #define ceph_file_layout_object_su(l) \
85 ((__s32)le32_to_cpu((l).fl_object_stripe_unit))
86 #define ceph_file_layout_pg_pool(l) \
87 ((__s32)le32_to_cpu((l).fl_pg_pool))
89 static inline unsigned ceph_file_layout_stripe_width(struct ceph_file_layout *l)
91 return le32_to_cpu(l->fl_stripe_unit) *
92 le32_to_cpu(l->fl_stripe_count);
95 /* "period" == bytes before i start on a new set of objects */
96 static inline unsigned ceph_file_layout_period(struct ceph_file_layout *l)
98 return le32_to_cpu(l->fl_object_size) *
99 le32_to_cpu(l->fl_stripe_count);
103 static inline int ceph_osd_is_up(struct ceph_osdmap *map, int osd)
105 return (osd < map->max_osd) && (map->osd_state[osd] & CEPH_OSD_UP);
108 static inline bool ceph_osdmap_flag(struct ceph_osdmap *map, int flag)
110 return map && (map->flags & flag);
113 extern char *ceph_osdmap_state_str(char *str, int len, int state);
115 static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map,
118 if (osd >= map->max_osd)
120 return &map->osd_addr[osd];
123 static inline int ceph_decode_pgid(void **p, void *end, struct ceph_pg *pgid)
127 if (!ceph_has_room(p, end, 1 + 8 + 4 + 4)) {
128 pr_warning("incomplete pg encoding");
132 version = ceph_decode_8(p);
134 pr_warning("do not understand pg encoding %d > 1",
139 pgid->pool = ceph_decode_64(p);
140 pgid->seed = ceph_decode_32(p);
141 *p += 4; /* skip deprecated preferred value */
146 extern struct ceph_osdmap *osdmap_decode(void **p, void *end);
147 extern struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
148 struct ceph_osdmap *map,
149 struct ceph_messenger *msgr);
150 extern void ceph_osdmap_destroy(struct ceph_osdmap *map);
152 /* calculate mapping of a file extent to an object */
153 extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout,
155 u64 *bno, u64 *oxoff, u64 *oxlen);
157 /* calculate mapping of object to a placement group */
158 extern int ceph_calc_ceph_pg(struct ceph_pg *pg, const char *oid,
159 struct ceph_osdmap *osdmap, uint64_t pool);
160 extern int ceph_calc_pg_acting(struct ceph_osdmap *osdmap,
163 extern int ceph_calc_pg_primary(struct ceph_osdmap *osdmap,
164 struct ceph_pg pgid);
166 extern const char *ceph_pg_pool_name_by_id(struct ceph_osdmap *map, u64 id);
167 extern int ceph_pg_poolid_by_name(struct ceph_osdmap *map, const char *name);