Commit | Line | Data |
---|---|---|
f24e9980 SW |
1 | #ifndef _FS_CEPH_OSDMAP_H |
2 | #define _FS_CEPH_OSDMAP_H | |
3 | ||
4 | #include <linux/rbtree.h> | |
a1ce3928 DH |
5 | #include <linux/ceph/types.h> |
6 | #include <linux/ceph/ceph_fs.h> | |
3d14c5d2 | 7 | #include <linux/crush/crush.h> |
f24e9980 SW |
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 | */ | |
5b191d99 SW |
21 | struct ceph_pg { |
22 | uint64_t pool; | |
23 | uint32_t seed; | |
24 | }; | |
25 | ||
f24e9980 | 26 | struct ceph_pg_pool_info { |
4fc51be8 SW |
27 | struct rb_node node; |
28 | int id; | |
f24e9980 SW |
29 | struct ceph_pg_pool v; |
30 | int pg_num_mask, pgp_num_mask, lpg_num_mask, lpgp_num_mask; | |
2844a76a | 31 | char *name; |
f24e9980 SW |
32 | }; |
33 | ||
34 | struct ceph_pg_mapping { | |
35 | struct rb_node node; | |
5b191d99 | 36 | struct ceph_pg pgid; |
f24e9980 SW |
37 | int len; |
38 | int osds[]; | |
39 | }; | |
40 | ||
41 | struct ceph_osdmap { | |
42 | struct ceph_fsid fsid; | |
43 | u32 epoch; | |
44 | u32 mkfs_epoch; | |
45 | struct ceph_timespec created, modified; | |
46 | ||
47 | u32 flags; /* CEPH_OSDMAP_* */ | |
48 | ||
49 | u32 max_osd; /* size of osd_state, _offload, _addr arrays */ | |
50 | u8 *osd_state; /* CEPH_OSD_* */ | |
51 | u32 *osd_weight; /* 0 = failed, 0x10000 = 100% normal */ | |
52 | struct ceph_entity_addr *osd_addr; | |
53 | ||
54 | struct rb_root pg_temp; | |
4fc51be8 SW |
55 | struct rb_root pg_pools; |
56 | u32 pool_max; | |
f24e9980 SW |
57 | |
58 | /* the CRUSH map specifies the mapping of placement groups to | |
59 | * the list of osds that store+replicate them. */ | |
60 | struct crush_map *crush; | |
61 | }; | |
62 | ||
63 | /* | |
64 | * file layout helpers | |
65 | */ | |
66 | #define ceph_file_layout_su(l) ((__s32)le32_to_cpu((l).fl_stripe_unit)) | |
67 | #define ceph_file_layout_stripe_count(l) \ | |
68 | ((__s32)le32_to_cpu((l).fl_stripe_count)) | |
69 | #define ceph_file_layout_object_size(l) ((__s32)le32_to_cpu((l).fl_object_size)) | |
70 | #define ceph_file_layout_cas_hash(l) ((__s32)le32_to_cpu((l).fl_cas_hash)) | |
71 | #define ceph_file_layout_object_su(l) \ | |
72 | ((__s32)le32_to_cpu((l).fl_object_stripe_unit)) | |
f24e9980 SW |
73 | #define ceph_file_layout_pg_pool(l) \ |
74 | ((__s32)le32_to_cpu((l).fl_pg_pool)) | |
75 | ||
76 | static inline unsigned ceph_file_layout_stripe_width(struct ceph_file_layout *l) | |
77 | { | |
78 | return le32_to_cpu(l->fl_stripe_unit) * | |
79 | le32_to_cpu(l->fl_stripe_count); | |
80 | } | |
81 | ||
82 | /* "period" == bytes before i start on a new set of objects */ | |
83 | static inline unsigned ceph_file_layout_period(struct ceph_file_layout *l) | |
84 | { | |
85 | return le32_to_cpu(l->fl_object_size) * | |
86 | le32_to_cpu(l->fl_stripe_count); | |
87 | } | |
88 | ||
89 | ||
90 | static inline int ceph_osd_is_up(struct ceph_osdmap *map, int osd) | |
91 | { | |
92 | return (osd < map->max_osd) && (map->osd_state[osd] & CEPH_OSD_UP); | |
93 | } | |
94 | ||
95 | static inline bool ceph_osdmap_flag(struct ceph_osdmap *map, int flag) | |
96 | { | |
97 | return map && (map->flags & flag); | |
98 | } | |
99 | ||
100 | extern char *ceph_osdmap_state_str(char *str, int len, int state); | |
101 | ||
102 | static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map, | |
103 | int osd) | |
104 | { | |
105 | if (osd >= map->max_osd) | |
106 | return NULL; | |
107 | return &map->osd_addr[osd]; | |
108 | } | |
109 | ||
110 | extern struct ceph_osdmap *osdmap_decode(void **p, void *end); | |
111 | extern struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end, | |
112 | struct ceph_osdmap *map, | |
113 | struct ceph_messenger *msgr); | |
114 | extern void ceph_osdmap_destroy(struct ceph_osdmap *map); | |
115 | ||
116 | /* calculate mapping of a file extent to an object */ | |
d63b77f4 | 117 | extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout, |
e8afad65 | 118 | u64 off, u64 len, |
d63b77f4 | 119 | u64 *bno, u64 *oxoff, u64 *oxlen); |
f24e9980 SW |
120 | |
121 | /* calculate mapping of object to a placement group */ | |
122 | extern int ceph_calc_object_layout(struct ceph_object_layout *ol, | |
123 | const char *oid, | |
124 | struct ceph_file_layout *fl, | |
125 | struct ceph_osdmap *osdmap); | |
12979354 | 126 | extern int ceph_calc_pg_acting(struct ceph_osdmap *osdmap, |
5b191d99 | 127 | struct ceph_pg pgid, |
d85b7056 | 128 | int *acting); |
51042122 | 129 | extern int ceph_calc_pg_primary(struct ceph_osdmap *osdmap, |
5b191d99 | 130 | struct ceph_pg pgid); |
f24e9980 | 131 | |
72afc71f | 132 | extern const char *ceph_pg_pool_name_by_id(struct ceph_osdmap *map, u64 id); |
7669a2c9 YS |
133 | extern int ceph_pg_poolid_by_name(struct ceph_osdmap *map, const char *name); |
134 | ||
f24e9980 | 135 | #endif |