[deliverable/linux.git] / include / linux / ceph / osdmap.h

#ifndef _FS_CEPH_OSDMAP_H
#define _FS_CEPH_OSDMAP_H

#include <linux/rbtree.h>
#include <linux/ceph/types.h>
#include <linux/ceph/decode.h>
#include <linux/ceph/ceph_fs.h>
#include <linux/crush/crush.h>

/*
 * The osd map describes the current membership of the osd cluster and
 * specifies the mapping of objects to placement groups and placement
 * groups to (sets of) osds.  That is, it completely specifies the
 * (desired) distribution of all data objects in the system at some
 * point in time.
 *
 * Each map version is identified by an epoch, which increases monotonically.
 *
 * The map can be updated either via an incremental map (diff) describing
 * the change between two successive epochs, or as a fully encoded map.
 */
struct ceph_pg {
	uint64_t pool;
	uint32_t seed;
};

int ceph_pg_compare(const struct ceph_pg *lhs, const struct ceph_pg *rhs);

#define CEPH_POOL_FLAG_HASHPSPOOL	(1ULL << 0) /* hash pg seed and pool id
						       together */
#define CEPH_POOL_FLAG_FULL		(1ULL << 1) /* pool is full */

struct ceph_pg_pool_info {
	struct rb_node node;
	s64 id;
	u8 type; /* CEPH_POOL_TYPE_* */
	u8 size;
	u8 min_size;
	u8 crush_ruleset;
	u8 object_hash;
	u32 last_force_request_resend;
	u32 pg_num, pgp_num;
	int pg_num_mask, pgp_num_mask;
	s64 read_tier;
	s64 write_tier; /* wins for read+write ops */
	u64 flags; /* CEPH_POOL_FLAG_* */
	char *name;

	bool was_full;  /* for handle_one_map() */
};

static inline bool ceph_can_shift_osds(struct ceph_pg_pool_info *pool)
{
	switch (pool->type) {
	case CEPH_POOL_TYPE_REP:
		return true;
	case CEPH_POOL_TYPE_EC:
		return false;
	default:
		BUG_ON(1);
	}
}

struct ceph_object_locator {
	s64 pool;
	struct ceph_string *pool_ns;
};

static inline void ceph_oloc_init(struct ceph_object_locator *oloc)
{
	oloc->pool = -1;
	oloc->pool_ns = NULL;
}

static inline bool ceph_oloc_empty(const struct ceph_object_locator *oloc)
{
	return oloc->pool == -1;
}

void ceph_oloc_copy(struct ceph_object_locator *dest,
		    const struct ceph_object_locator *src);
void ceph_oloc_destroy(struct ceph_object_locator *oloc);

/*
 * Maximum supported by kernel client object name length
 *
 * (probably outdated: must be >= RBD_MAX_MD_NAME_LEN -- currently 100)
 */
#define CEPH_MAX_OID_NAME_LEN 100

/*
 * 51-char inline_name is long enough for all cephfs and all but one
 * rbd requests: <imgname> in "<imgname>.rbd"/"rbd_id.<imgname>" can be
 * arbitrarily long (~PAGE_SIZE).  It's done once during rbd map; all
 * other rbd requests fit into inline_name.
 *
 * Makes ceph_object_id 64 bytes on 64-bit.
 */
#define CEPH_OID_INLINE_LEN 52

/*
 * Both inline and external buffers have space for a NUL-terminator,
 * which is carried around.  It's not required though - RADOS object
 * names don't have to be NUL-terminated and may contain NULs.
 */
struct ceph_object_id {
	char *name;
	char inline_name[CEPH_OID_INLINE_LEN];
	int name_len;
};

static inline void ceph_oid_init(struct ceph_object_id *oid)
{
	oid->name = oid->inline_name;
	oid->name_len = 0;
}

#define CEPH_OID_INIT_ONSTACK(oid)					\
    ({ ceph_oid_init(&oid); oid; })
#define CEPH_DEFINE_OID_ONSTACK(oid)					\
	struct ceph_object_id oid = CEPH_OID_INIT_ONSTACK(oid)

static inline bool ceph_oid_empty(const struct ceph_object_id *oid)
{
	return oid->name == oid->inline_name && !oid->name_len;
}

void ceph_oid_copy(struct ceph_object_id *dest,
		   const struct ceph_object_id *src);
__printf(2, 3)
void ceph_oid_printf(struct ceph_object_id *oid, const char *fmt, ...);
__printf(3, 4)
int ceph_oid_aprintf(struct ceph_object_id *oid, gfp_t gfp,
		     const char *fmt, ...);
void ceph_oid_destroy(struct ceph_object_id *oid);

struct ceph_pg_mapping {
	struct rb_node node;
	struct ceph_pg pgid;

	union {
		struct {
			int len;
			int osds[];
		} pg_temp;
		struct {
			int osd;
		} primary_temp;
	};
};

struct ceph_osdmap {
	struct ceph_fsid fsid;
	u32 epoch;
	struct ceph_timespec created, modified;

	u32 flags;         /* CEPH_OSDMAP_* */

	u32 max_osd;       /* size of osd_state, _offload, _addr arrays */
	u8 *osd_state;     /* CEPH_OSD_* */
	u32 *osd_weight;   /* 0 = failed, 0x10000 = 100% normal */
	struct ceph_entity_addr *osd_addr;

	struct rb_root pg_temp;
	struct rb_root primary_temp;

	u32 *osd_primary_affinity;

	struct rb_root pg_pools;
	u32 pool_max;

	/* the CRUSH map specifies the mapping of placement groups to
	 * the list of osds that store+replicate them. */
	struct crush_map *crush;

	struct mutex crush_scratch_mutex;
	int crush_scratch_ary[CEPH_PG_MAX_SIZE * 3];
};

static inline bool ceph_osd_exists(struct ceph_osdmap *map, int osd)
{
	return osd >= 0 && osd < map->max_osd &&
	       (map->osd_state[osd] & CEPH_OSD_EXISTS);
}

static inline bool ceph_osd_is_up(struct ceph_osdmap *map, int osd)
{
	return ceph_osd_exists(map, osd) &&
	       (map->osd_state[osd] & CEPH_OSD_UP);
}

static inline bool ceph_osd_is_down(struct ceph_osdmap *map, int osd)
{
	return !ceph_osd_is_up(map, osd);
}

extern char *ceph_osdmap_state_str(char *str, int len, int state);
extern u32 ceph_get_primary_affinity(struct ceph_osdmap *map, int osd);

static inline struct ceph_entity_addr *ceph_osd_addr(struct ceph_osdmap *map,
						     int osd)
{
	if (osd >= map->max_osd)
		return NULL;
	return &map->osd_addr[osd];
}

static inline int ceph_decode_pgid(void **p, void *end, struct ceph_pg *pgid)
{
	__u8 version;

	if (!ceph_has_room(p, end, 1 + 8 + 4 + 4)) {
		pr_warn("incomplete pg encoding\n");
		return -EINVAL;
	}
	version = ceph_decode_8(p);
	if (version > 1) {
		pr_warn("do not understand pg encoding %d > 1\n",
			(int)version);
		return -EINVAL;
	}

	pgid->pool = ceph_decode_64(p);
	pgid->seed = ceph_decode_32(p);
	*p += 4;	/* skip deprecated preferred value */

	return 0;
}

struct ceph_osdmap *ceph_osdmap_alloc(void);
extern struct ceph_osdmap *ceph_osdmap_decode(void **p, void *end);
struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
					     struct ceph_osdmap *map);
extern void ceph_osdmap_destroy(struct ceph_osdmap *map);

struct ceph_osds {
	int osds[CEPH_PG_MAX_SIZE];
	int size;
	int primary; /* id, NOT index */
};

static inline void ceph_osds_init(struct ceph_osds *set)
{
	set->size = 0;
	set->primary = -1;
}

void ceph_osds_copy(struct ceph_osds *dest, const struct ceph_osds *src);

bool ceph_is_new_interval(const struct ceph_osds *old_acting,
			  const struct ceph_osds *new_acting,
			  const struct ceph_osds *old_up,
			  const struct ceph_osds *new_up,
			  int old_size,
			  int new_size,
			  int old_min_size,
			  int new_min_size,
			  u32 old_pg_num,
			  u32 new_pg_num,
			  bool old_sort_bitwise,
			  bool new_sort_bitwise,
			  const struct ceph_pg *pgid);
bool ceph_osds_changed(const struct ceph_osds *old_acting,
		       const struct ceph_osds *new_acting,
		       bool any_change);

/* calculate mapping of a file extent to an object */
extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout,
					 u64 off, u64 len,
					 u64 *bno, u64 *oxoff, u64 *oxlen);

int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap,
			      struct ceph_object_id *oid,
			      struct ceph_object_locator *oloc,
			      struct ceph_pg *raw_pgid);

void ceph_pg_to_up_acting_osds(struct ceph_osdmap *osdmap,
			       const struct ceph_pg *raw_pgid,
			       struct ceph_osds *up,
			       struct ceph_osds *acting);
int ceph_pg_to_acting_primary(struct ceph_osdmap *osdmap,
			      const struct ceph_pg *raw_pgid);

extern struct ceph_pg_pool_info *ceph_pg_pool_by_id(struct ceph_osdmap *map,
						    u64 id);

extern const char *ceph_pg_pool_name_by_id(struct ceph_osdmap *map, u64 id);
extern int ceph_pg_poolid_by_name(struct ceph_osdmap *map, const char *name);

#endif
Commit	Line	Data
f24e9980 SW	1	#ifndef _FS_CEPH_OSDMAP_H
	2	#define _FS_CEPH_OSDMAP_H
	3
	4	#include <linux/rbtree.h>
a1ce3928	5	#include <linux/ceph/types.h>
ef4859d6	6	#include <linux/ceph/decode.h>
a1ce3928	7	#include <linux/ceph/ceph_fs.h>
3d14c5d2	8	#include <linux/crush/crush.h>
f24e9980 SW	9
	10	/*
	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
	15	* point in time.
	16	*
	17	* Each map version is identified by an epoch, which increases monotonically.
	18	*
	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.
	21	*/
5b191d99 SW	22	struct ceph_pg {
	23	uint64_t pool;
	24	uint32_t seed;
	25	};
	26
f984cb76 ID	27	int ceph_pg_compare(const struct ceph_pg lhs, const struct ceph_pg rhs);
f984cb76 ID	28
04812acf ID	29	#define CEPH_POOL_FLAG_HASHPSPOOL (1ULL << 0) /* hash pg seed and pool id
04812acf ID	30	together */
63244fa1	31	#define CEPH_POOL_FLAG_FULL (1ULL << 1) /* pool is full */
83ca14fd	32
f24e9980	33	struct ceph_pg_pool_info {
4fc51be8	34	struct rb_node node;
4f6a7e5e	35	s64 id;
04812acf	36	u8 type; /* CEPH_POOL_TYPE_* */
4f6a7e5e	37	u8 size;
04812acf	38	u8 min_size;
4f6a7e5e SW	39	u8 crush_ruleset;
4f6a7e5e SW	40	u8 object_hash;
04812acf	41	u32 last_force_request_resend;
4f6a7e5e SW	42	u32 pg_num, pgp_num;
4f6a7e5e SW	43	int pg_num_mask, pgp_num_mask;
17a13e40 ID	44	s64 read_tier;
17a13e40 ID	45	s64 write_tier; /* wins for read+write ops */
04812acf	46	u64 flags; /* CEPH_POOL_FLAG_* */
2844a76a	47	char *name;
42c1b124 ID	48
42c1b124 ID	49	bool was_full; /* for handle_one_map() */
f24e9980 SW	50	};
f24e9980 SW	51
2abebdbc ID	52	static inline bool ceph_can_shift_osds(struct ceph_pg_pool_info *pool)
	53	{
	54	switch (pool->type) {
	55	case CEPH_POOL_TYPE_REP:
	56	return true;
	57	case CEPH_POOL_TYPE_EC:
	58	return false;
	59	default:
	60	BUG_ON(1);
	61	}
	62	}
	63
4f6a7e5e	64	struct ceph_object_locator {
22116525	65	s64 pool;
30c156d9	66	struct ceph_string *pool_ns;
4f6a7e5e SW	67	};
4f6a7e5e SW	68
63244fa1 ID	69	static inline void ceph_oloc_init(struct ceph_object_locator *oloc)
	70	{
	71	oloc->pool = -1;
30c156d9	72	oloc->pool_ns = NULL;
63244fa1 ID	73	}
	74
	75	static inline bool ceph_oloc_empty(const struct ceph_object_locator *oloc)
	76	{
	77	return oloc->pool == -1;
	78	}
	79
30c156d9 YZ	80	void ceph_oloc_copy(struct ceph_object_locator *dest,
	81	const struct ceph_object_locator *src);
	82	void ceph_oloc_destroy(struct ceph_object_locator *oloc);
63244fa1	83
4295f221 ID	84	/*
	85	* Maximum supported by kernel client object name length
	86	*
	87	* (probably outdated: must be >= RBD_MAX_MD_NAME_LEN -- currently 100)
	88	*/
	89	#define CEPH_MAX_OID_NAME_LEN 100
	90
d30291b9 ID	91	/*
	92	* 51-char inline_name is long enough for all cephfs and all but one
	93	* rbd requests: <imgname> in "<imgname>.rbd"/"rbd_id.<imgname>" can be
	94	* arbitrarily long (~PAGE_SIZE). It's done once during rbd map; all
	95	* other rbd requests fit into inline_name.
	96	*
	97	* Makes ceph_object_id 64 bytes on 64-bit.
	98	*/
	99	#define CEPH_OID_INLINE_LEN 52
	100
	101	/*
	102	* Both inline and external buffers have space for a NUL-terminator,
	103	* which is carried around. It's not required though - RADOS object
	104	* names don't have to be NUL-terminated and may contain NULs.
	105	*/
4295f221	106	struct ceph_object_id {
d30291b9 ID	107	char *name;
d30291b9 ID	108	char inline_name[CEPH_OID_INLINE_LEN];
4295f221 ID	109	int name_len;
	110	};
	111
d30291b9 ID	112	static inline void ceph_oid_init(struct ceph_object_id *oid)
	113	{
	114	oid->name = oid->inline_name;
	115	oid->name_len = 0;
	116	}
	117
281dbe5d ID	118	#define CEPH_OID_INIT_ONSTACK(oid) \
	119	({ ceph_oid_init(&oid); oid; })
	120	#define CEPH_DEFINE_OID_ONSTACK(oid) \
	121	struct ceph_object_id oid = CEPH_OID_INIT_ONSTACK(oid)
	122
d30291b9 ID	123	static inline bool ceph_oid_empty(const struct ceph_object_id *oid)
	124	{
	125	return oid->name == oid->inline_name && !oid->name_len;
	126	}
	127
	128	void ceph_oid_copy(struct ceph_object_id *dest,
	129	const struct ceph_object_id *src);
	130	__printf(2, 3)
	131	void ceph_oid_printf(struct ceph_object_id oid, const char fmt, ...);
	132	__printf(3, 4)
	133	int ceph_oid_aprintf(struct ceph_object_id *oid, gfp_t gfp,
	134	const char *fmt, ...);
	135	void ceph_oid_destroy(struct ceph_object_id *oid);
	136
f24e9980 SW	137	struct ceph_pg_mapping {
f24e9980 SW	138	struct rb_node node;
5b191d99	139	struct ceph_pg pgid;
35a935d7 ID	140
	141	union {
	142	struct {
	143	int len;
	144	int osds[];
	145	} pg_temp;
9686f94c ID	146	struct {
	147	int osd;
	148	} primary_temp;
35a935d7	149	};
f24e9980 SW	150	};
	151
	152	struct ceph_osdmap {
	153	struct ceph_fsid fsid;
	154	u32 epoch;
f24e9980 SW	155	struct ceph_timespec created, modified;
	156
	157	u32 flags; /* CEPH_OSDMAP_* */
	158
	159	u32 max_osd; /* size of osd_state, _offload, _addr arrays */
	160	u8 osd_state; / CEPH_OSD_* */
	161	u32 osd_weight; / 0 = failed, 0x10000 = 100% normal */
	162	struct ceph_entity_addr *osd_addr;
	163
	164	struct rb_root pg_temp;
9686f94c ID	165	struct rb_root primary_temp;
9686f94c ID	166
2cfa34f2 ID	167	u32 *osd_primary_affinity;
2cfa34f2 ID	168
4fc51be8 SW	169	struct rb_root pg_pools;
4fc51be8 SW	170	u32 pool_max;
f24e9980 SW	171
	172	/* the CRUSH map specifies the mapping of placement groups to
	173	* the list of osds that store+replicate them. */
	174	struct crush_map *crush;
9d521470 ID	175
	176	struct mutex crush_scratch_mutex;
	177	int crush_scratch_ary[CEPH_PG_MAX_SIZE * 3];
f24e9980 SW	178	};
f24e9980 SW	179
3b33f692	180	static inline bool ceph_osd_exists(struct ceph_osdmap *map, int osd)
246138fa ID	181	{
	182	return osd >= 0 && osd < map->max_osd &&
	183	(map->osd_state[osd] & CEPH_OSD_EXISTS);
	184	}
	185
3b33f692	186	static inline bool ceph_osd_is_up(struct ceph_osdmap *map, int osd)
f24e9980	187	{
246138fa ID	188	return ceph_osd_exists(map, osd) &&
	189	(map->osd_state[osd] & CEPH_OSD_UP);
	190	}
	191
3b33f692	192	static inline bool ceph_osd_is_down(struct ceph_osdmap *map, int osd)
246138fa ID	193	{
246138fa ID	194	return !ceph_osd_is_up(map, osd);
f24e9980 SW	195	}
f24e9980 SW	196
f24e9980	197	extern char ceph_osdmap_state_str(char str, int len, int state);
2cfa34f2	198	extern u32 ceph_get_primary_affinity(struct ceph_osdmap *map, int osd);
f24e9980 SW	199
	200	static inline struct ceph_entity_addr ceph_osd_addr(struct ceph_osdmap map,
	201	int osd)
	202	{
	203	if (osd >= map->max_osd)
	204	return NULL;
	205	return &map->osd_addr[osd];
	206	}
	207
ef4859d6 AE	208	static inline int ceph_decode_pgid(void *p, void end, struct ceph_pg *pgid)
	209	{
	210	__u8 version;
	211
	212	if (!ceph_has_room(p, end, 1 + 8 + 4 + 4)) {
3ef650d3	213	pr_warn("incomplete pg encoding\n");
ef4859d6 AE	214	return -EINVAL;
	215	}
	216	version = ceph_decode_8(p);
	217	if (version > 1) {
3ef650d3	218	pr_warn("do not understand pg encoding %d > 1\n",
ef4859d6 AE	219	(int)version);
	220	return -EINVAL;
	221	}
	222
	223	pgid->pool = ceph_decode_64(p);
	224	pgid->seed = ceph_decode_32(p);
	225	p += 4; / skip deprecated preferred value */
	226
	227	return 0;
	228	}
	229
e5253a7b	230	struct ceph_osdmap *ceph_osdmap_alloc(void);
a2505d63	231	extern struct ceph_osdmap ceph_osdmap_decode(void p, void end);
0c0a8de1 ID	232	struct ceph_osdmap osdmap_apply_incremental(void p, void end,
0c0a8de1 ID	233	struct ceph_osdmap *map);
f24e9980 SW	234	extern void ceph_osdmap_destroy(struct ceph_osdmap *map);
f24e9980 SW	235
6f3bfd45 ID	236	struct ceph_osds {
	237	int osds[CEPH_PG_MAX_SIZE];
	238	int size;
	239	int primary; /* id, NOT index */
	240	};
	241
	242	static inline void ceph_osds_init(struct ceph_osds *set)
	243	{
	244	set->size = 0;
	245	set->primary = -1;
	246	}
	247
	248	void ceph_osds_copy(struct ceph_osds dest, const struct ceph_osds src);
	249
63244fa1 ID	250	bool ceph_is_new_interval(const struct ceph_osds *old_acting,
	251	const struct ceph_osds *new_acting,
	252	const struct ceph_osds *old_up,
	253	const struct ceph_osds *new_up,
	254	int old_size,
	255	int new_size,
	256	int old_min_size,
	257	int new_min_size,
	258	u32 old_pg_num,
	259	u32 new_pg_num,
	260	bool old_sort_bitwise,
	261	bool new_sort_bitwise,
	262	const struct ceph_pg *pgid);
	263	bool ceph_osds_changed(const struct ceph_osds *old_acting,
	264	const struct ceph_osds *new_acting,
	265	bool any_change);
	266
f24e9980	267	/* calculate mapping of a file extent to an object */
d63b77f4	268	extern int ceph_calc_file_object_mapping(struct ceph_file_layout *layout,
e8afad65	269	u64 off, u64 len,
d63b77f4	270	u64 bno, u64 oxoff, u64 *oxlen);
f24e9980	271
d9591f5e ID	272	int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap,
	273	struct ceph_object_id *oid,
	274	struct ceph_object_locator *oloc,
	275	struct ceph_pg *raw_pgid);
7c13cb64	276
6f3bfd45 ID	277	void ceph_pg_to_up_acting_osds(struct ceph_osdmap *osdmap,
	278	const struct ceph_pg *raw_pgid,
	279	struct ceph_osds *up,
	280	struct ceph_osds *acting);
f81f1633 ID	281	int ceph_pg_to_acting_primary(struct ceph_osdmap *osdmap,
f81f1633 ID	282	const struct ceph_pg *raw_pgid);
f24e9980	283
ce7f6a27 ID	284	extern struct ceph_pg_pool_info ceph_pg_pool_by_id(struct ceph_osdmap map,
	285	u64 id);
	286
72afc71f	287	extern const char ceph_pg_pool_name_by_id(struct ceph_osdmap map, u64 id);
7669a2c9 YS	288	extern int ceph_pg_poolid_by_name(struct ceph_osdmap map, const char name);
7669a2c9 YS	289
f24e9980	290	#endif