blob: 07127c6fb134c3980b5f32f7f242fe73b0240c16 [file] [log] [blame]
Sage Weilf24e9982009-10-06 11:31:10 -07001#ifndef _FS_CEPH_OSDMAP_H
2#define _FS_CEPH_OSDMAP_H
3
4#include <linux/rbtree.h>
5#include "types.h"
6#include "ceph_fs.h"
7#include "crush/crush.h"
8
9/*
10 * The osd map describes the current membership of the osd cluster and
11 * specifies the mapping of objects to placement groups and placement
12 * groups to (sets of) osds. That is, it completely specifies the
13 * (desired) distribution of all data objects in the system at some
14 * point in time.
15 *
16 * Each map version is identified by an epoch, which increases monotonically.
17 *
18 * The map can be updated either via an incremental map (diff) describing
19 * the change between two successive epochs, or as a fully encoded map.
20 */
21struct ceph_pg_pool_info {
22 struct ceph_pg_pool v;
23 int pg_num_mask, pgp_num_mask, lpg_num_mask, lpgp_num_mask;
24};
25
26struct ceph_pg_mapping {
27 struct rb_node node;
28 u64 pgid;
29 int len;
30 int osds[];
31};
32
33struct ceph_osdmap {
34 struct ceph_fsid fsid;
35 u32 epoch;
36 u32 mkfs_epoch;
37 struct ceph_timespec created, modified;
38
39 u32 flags; /* CEPH_OSDMAP_* */
40
41 u32 max_osd; /* size of osd_state, _offload, _addr arrays */
42 u8 *osd_state; /* CEPH_OSD_* */
43 u32 *osd_weight; /* 0 = failed, 0x10000 = 100% normal */
44 struct ceph_entity_addr *osd_addr;
45
46 struct rb_root pg_temp;
47
48 u32 num_pools;
49 struct ceph_pg_pool_info *pg_pool;
50
51 /* the CRUSH map specifies the mapping of placement groups to
52 * the list of osds that store+replicate them. */
53 struct crush_map *crush;
54};
55
56/*
57 * file layout helpers
58 */
59#define ceph_file_layout_su(l) ((__s32)le32_to_cpu((l).fl_stripe_unit))
60#define ceph_file_layout_stripe_count(l) \
61 ((__s32)le32_to_cpu((l).fl_stripe_count))
62#define ceph_file_layout_object_size(l) ((__s32)le32_to_cpu((l).fl_object_size))
63#define ceph_file_layout_cas_hash(l) ((__s32)le32_to_cpu((l).fl_cas_hash))
64#define ceph_file_layout_object_su(l) \
65 ((__s32)le32_to_cpu((l).fl_object_stripe_unit))
66#define ceph_file_layout_pg_preferred(l) \
67 ((__s32)le32_to_cpu((l).fl_pg_preferred))
68#define ceph_file_layout_pg_pool(l) \
69 ((__s32)le32_to_cpu((l).fl_pg_pool))
70
71static inline unsigned ceph_file_layout_stripe_width(struct ceph_file_layout *l)
72{
73 return le32_to_cpu(l->fl_stripe_unit) *
74 le32_to_cpu(l->fl_stripe_count);
75}
76
77/* "period" == bytes before i start on a new set of objects */
78static inline unsigned ceph_file_layout_period(struct ceph_file_layout *l)
79{
80 return le32_to_cpu(l->fl_object_size) *
81 le32_to_cpu(l->fl_stripe_count);
82}
83
84
85static inline int ceph_osd_is_up(struct ceph_osdmap *map, int osd)
86{
87 return (osd < map->max_osd) && (map->osd_state[osd] & CEPH_OSD_UP);
88}
89
90static inline bool ceph_osdmap_flag(struct ceph_osdmap *map, int flag)
91{
92 return map && (map->flags & flag);
93}
94
95extern char *ceph_osdmap_state_str(char *str, int len, int state);
96
97static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map,
98 int osd)
99{
100 if (osd >= map->max_osd)
101 return NULL;
102 return &map->osd_addr[osd];
103}
104
105extern struct ceph_osdmap *osdmap_decode(void **p, void *end);
106extern struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
107 struct ceph_osdmap *map,
108 struct ceph_messenger *msgr);
109extern void ceph_osdmap_destroy(struct ceph_osdmap *map);
110
111/* calculate mapping of a file extent to an object */
112extern void ceph_calc_file_object_mapping(struct ceph_file_layout *layout,
113 u64 off, u64 *plen,
114 u64 *bno, u64 *oxoff, u64 *oxlen);
115
116/* calculate mapping of object to a placement group */
117extern int ceph_calc_object_layout(struct ceph_object_layout *ol,
118 const char *oid,
119 struct ceph_file_layout *fl,
120 struct ceph_osdmap *osdmap);
121extern int ceph_calc_pg_primary(struct ceph_osdmap *osdmap, union ceph_pg pgid);
122
123#endif