[mirror_ubuntu-artful-kernel.git] / include / linux / ceph / rados.h

#ifndef CEPH_RADOS_H
#define CEPH_RADOS_H

/*
 * Data types for the Ceph distributed object storage layer RADOS
 * (Reliable Autonomic Distributed Object Store).
 */

#include <linux/ceph/msgr.h>

/*
 * fs id
 */
struct ceph_fsid {
	unsigned char fsid[16];
};

static inline int ceph_fsid_compare(const struct ceph_fsid *a,
				    const struct ceph_fsid *b)
{
	return memcmp(a, b, sizeof(*a));
}

/*
 * ino, object, etc.
 */
typedef __le64 ceph_snapid_t;
#define CEPH_SNAPDIR ((__u64)(-1))  /* reserved for hidden .snap dir */
#define CEPH_NOSNAP  ((__u64)(-2))  /* "head", "live" revision */
#define CEPH_MAXSNAP ((__u64)(-3))  /* largest valid snapid */

struct ceph_timespec {
	__le32 tv_sec;
	__le32 tv_nsec;
} __attribute__ ((packed));


/*
 * object layout - how objects are mapped into PGs
 */
#define CEPH_OBJECT_LAYOUT_HASH     1
#define CEPH_OBJECT_LAYOUT_LINEAR   2
#define CEPH_OBJECT_LAYOUT_HASHINO  3

/*
 * pg layout -- how PGs are mapped onto (sets of) OSDs
 */
#define CEPH_PG_LAYOUT_CRUSH  0
#define CEPH_PG_LAYOUT_HASH   1
#define CEPH_PG_LAYOUT_LINEAR 2
#define CEPH_PG_LAYOUT_HYBRID 3

#define CEPH_PG_MAX_SIZE      32  /* max # osds in a single pg */

/*
 * placement group.
 * we encode this into one __le64.
 */
struct ceph_pg_v1 {
	__le16 preferred; /* preferred primary osd */
	__le16 ps;        /* placement seed */
	__le32 pool;      /* object pool */
} __attribute__ ((packed));

/*
 * pg_pool is a set of pgs storing a pool of objects
 *
 *  pg_num -- base number of pseudorandomly placed pgs
 *
 *  pgp_num -- effective number when calculating pg placement.  this
 * is used for pg_num increases.  new pgs result in data being "split"
 * into new pgs.  for this to proceed smoothly, new pgs are intiially
 * colocated with their parents; that is, pgp_num doesn't increase
 * until the new pgs have successfully split.  only _then_ are the new
 * pgs placed independently.
 *
 *  lpg_num -- localized pg count (per device).  replicas are randomly
 * selected.
 *
 *  lpgp_num -- as above.
 */
#define CEPH_NOPOOL  ((__u64) (-1))  /* pool id not defined */

#define CEPH_POOL_TYPE_REP     1
#define CEPH_POOL_TYPE_RAID4   2 /* never implemented */
#define CEPH_POOL_TYPE_EC      3

/*
 * stable_mod func is used to control number of placement groups.
 * similar to straight-up modulo, but produces a stable mapping as b
 * increases over time.  b is the number of bins, and bmask is the
 * containing power of 2 minus 1.
 *
 * b <= bmask and bmask=(2**n)-1
 * e.g., b=12 -> bmask=15, b=123 -> bmask=127
 */
static inline int ceph_stable_mod(int x, int b, int bmask)
{
	if ((x & bmask) < b)
		return x & bmask;
	else
		return x & (bmask >> 1);
}

/*
 * object layout - how a given object should be stored.
 */
struct ceph_object_layout {
	struct ceph_pg_v1 ol_pgid;   /* raw pg, with _full_ ps precision. */
	__le32 ol_stripe_unit;    /* for per-object parity, if any */
} __attribute__ ((packed));

/*
 * compound epoch+version, used by storage layer to serialize mutations
 */
struct ceph_eversion {
	__le64 version;
	__le32 epoch;
} __attribute__ ((packed));

/*
 * osd map bits
 */

/* status bits */
#define CEPH_OSD_EXISTS  (1<<0)
#define CEPH_OSD_UP      (1<<1)
#define CEPH_OSD_AUTOOUT (1<<2)  /* osd was automatically marked out */
#define CEPH_OSD_NEW     (1<<3)  /* osd is new, never marked in */

extern const char *ceph_osd_state_name(int s);

/* osd weights.  fixed point value: 0x10000 == 1.0 ("in"), 0 == "out" */
#define CEPH_OSD_IN  0x10000
#define CEPH_OSD_OUT 0

/* osd primary-affinity.  fixed point value: 0x10000 == baseline */
#define CEPH_OSD_MAX_PRIMARY_AFFINITY 0x10000
#define CEPH_OSD_DEFAULT_PRIMARY_AFFINITY 0x10000


/*
 * osd map flag bits
 */
#define CEPH_OSDMAP_NEARFULL (1<<0)  /* sync writes (near ENOSPC) */
#define CEPH_OSDMAP_FULL     (1<<1)  /* no data writes (ENOSPC) */
#define CEPH_OSDMAP_PAUSERD  (1<<2)  /* pause all reads */
#define CEPH_OSDMAP_PAUSEWR  (1<<3)  /* pause all writes */
#define CEPH_OSDMAP_PAUSEREC (1<<4)  /* pause recovery */
#define CEPH_OSDMAP_NOUP     (1<<5)  /* block osd boot */
#define CEPH_OSDMAP_NODOWN   (1<<6)  /* block osd mark-down/failure */
#define CEPH_OSDMAP_NOOUT    (1<<7)  /* block osd auto mark-out */
#define CEPH_OSDMAP_NOIN     (1<<8)  /* block osd auto mark-in */
#define CEPH_OSDMAP_NOBACKFILL (1<<9) /* block osd backfill */
#define CEPH_OSDMAP_NORECOVER (1<<10) /* block osd recovery and backfill */
#define CEPH_OSDMAP_NOSCRUB  (1<<11) /* block periodic scrub */
#define CEPH_OSDMAP_NODEEP_SCRUB (1<<12) /* block periodic deep-scrub */
#define CEPH_OSDMAP_NOTIERAGENT (1<<13) /* disable tiering agent */
#define CEPH_OSDMAP_NOREBALANCE (1<<14) /* block osd backfill unless pg is degraded */
#define CEPH_OSDMAP_SORTBITWISE (1<<15) /* use bitwise hobject_t sort */
#define CEPH_OSDMAP_REQUIRE_JEWEL    (1<<16) /* require jewel for booting osds */
#define CEPH_OSDMAP_REQUIRE_KRAKEN   (1<<17) /* require kraken for booting osds */
#define CEPH_OSDMAP_REQUIRE_LUMINOUS (1<<18) /* require l for booting osds */
#define CEPH_OSDMAP_RECOVERY_DELETES (1<<19) /* deletes performed during recovery instead of peering */

/*
 * The error code to return when an OSD can't handle a write
 * because it is too large.
 */
#define OSD_WRITETOOBIG EMSGSIZE

/*
 * osd ops
 *
 * WARNING: do not use these op codes directly.  Use the helpers
 * defined below instead.  In certain cases, op code behavior was
 * redefined, resulting in special-cases in the helpers.
 */
#define CEPH_OSD_OP_MODE       0xf000
#define CEPH_OSD_OP_MODE_RD    0x1000
#define CEPH_OSD_OP_MODE_WR    0x2000
#define CEPH_OSD_OP_MODE_RMW   0x3000
#define CEPH_OSD_OP_MODE_SUB   0x4000
#define CEPH_OSD_OP_MODE_CACHE 0x8000

#define CEPH_OSD_OP_TYPE       0x0f00
#define CEPH_OSD_OP_TYPE_LOCK  0x0100
#define CEPH_OSD_OP_TYPE_DATA  0x0200
#define CEPH_OSD_OP_TYPE_ATTR  0x0300
#define CEPH_OSD_OP_TYPE_EXEC  0x0400
#define CEPH_OSD_OP_TYPE_PG    0x0500
#define CEPH_OSD_OP_TYPE_MULTI 0x0600 /* multiobject */

#define __CEPH_OSD_OP1(mode, nr) \
	(CEPH_OSD_OP_MODE_##mode | (nr))

#define __CEPH_OSD_OP(mode, type, nr) \
	(CEPH_OSD_OP_MODE_##mode | CEPH_OSD_OP_TYPE_##type | (nr))

#define __CEPH_FORALL_OSD_OPS(f)					    \
	/** data **/							    \
	/* read */							    \
	f(READ,		__CEPH_OSD_OP(RD, DATA, 1),	"read")		    \
	f(STAT,		__CEPH_OSD_OP(RD, DATA, 2),	"stat")		    \
	f(MAPEXT,	__CEPH_OSD_OP(RD, DATA, 3),	"mapext")	    \
									    \
	/* fancy read */						    \
	f(MASKTRUNC,	__CEPH_OSD_OP(RD, DATA, 4),	"masktrunc")	    \
	f(SPARSE_READ,	__CEPH_OSD_OP(RD, DATA, 5),	"sparse-read")	    \
									    \
	f(NOTIFY,	__CEPH_OSD_OP(RD, DATA, 6),	"notify")	    \
	f(NOTIFY_ACK,	__CEPH_OSD_OP(RD, DATA, 7),	"notify-ack")	    \
									    \
	/* versioning */						    \
	f(ASSERT_VER,	__CEPH_OSD_OP(RD, DATA, 8),	"assert-version")   \
									    \
	f(LIST_WATCHERS, __CEPH_OSD_OP(RD, DATA, 9),	"list-watchers")    \
									    \
	f(LIST_SNAPS,	__CEPH_OSD_OP(RD, DATA, 10),	"list-snaps")	    \
									    \
	/* sync */							    \
	f(SYNC_READ,	__CEPH_OSD_OP(RD, DATA, 11),	"sync_read")	    \
									    \
	/* write */							    \
	f(WRITE,	__CEPH_OSD_OP(WR, DATA, 1),	"write")	    \
	f(WRITEFULL,	__CEPH_OSD_OP(WR, DATA, 2),	"writefull")	    \
	f(TRUNCATE,	__CEPH_OSD_OP(WR, DATA, 3),	"truncate")	    \
	f(ZERO,		__CEPH_OSD_OP(WR, DATA, 4),	"zero")		    \
	f(DELETE,	__CEPH_OSD_OP(WR, DATA, 5),	"delete")	    \
									    \
	/* fancy write */						    \
	f(APPEND,	__CEPH_OSD_OP(WR, DATA, 6),	"append")	    \
	f(STARTSYNC,	__CEPH_OSD_OP(WR, DATA, 7),	"startsync")	    \
	f(SETTRUNC,	__CEPH_OSD_OP(WR, DATA, 8),	"settrunc")	    \
	f(TRIMTRUNC,	__CEPH_OSD_OP(WR, DATA, 9),	"trimtrunc")	    \
									    \
	f(TMAPUP,	__CEPH_OSD_OP(RMW, DATA, 10),	"tmapup")	    \
	f(TMAPPUT,	__CEPH_OSD_OP(WR, DATA, 11),	"tmapput")	    \
	f(TMAPGET,	__CEPH_OSD_OP(RD, DATA, 12),	"tmapget")	    \
									    \
	f(CREATE,	__CEPH_OSD_OP(WR, DATA, 13),	"create")	    \
	f(ROLLBACK,	__CEPH_OSD_OP(WR, DATA, 14),	"rollback")	    \
									    \
	f(WATCH,	__CEPH_OSD_OP(WR, DATA, 15),	"watch")	    \
									    \
	/* omap */							    \
	f(OMAPGETKEYS,	__CEPH_OSD_OP(RD, DATA, 17),	"omap-get-keys")    \
	f(OMAPGETVALS,	__CEPH_OSD_OP(RD, DATA, 18),	"omap-get-vals")    \
	f(OMAPGETHEADER, __CEPH_OSD_OP(RD, DATA, 19),	"omap-get-header")  \
	f(OMAPGETVALSBYKEYS, __CEPH_OSD_OP(RD, DATA, 20), "omap-get-vals-by-keys") \
	f(OMAPSETVALS,	__CEPH_OSD_OP(WR, DATA, 21),	"omap-set-vals")    \
	f(OMAPSETHEADER, __CEPH_OSD_OP(WR, DATA, 22),	"omap-set-header")  \
	f(OMAPCLEAR,	__CEPH_OSD_OP(WR, DATA, 23),	"omap-clear")	    \
	f(OMAPRMKEYS,	__CEPH_OSD_OP(WR, DATA, 24),	"omap-rm-keys")	    \
	f(OMAP_CMP,	__CEPH_OSD_OP(RD, DATA, 25),	"omap-cmp")	    \
									    \
	/* tiering */							    \
	f(COPY_FROM,	__CEPH_OSD_OP(WR, DATA, 26),	"copy-from")	    \
	f(COPY_GET_CLASSIC, __CEPH_OSD_OP(RD, DATA, 27), "copy-get-classic") \
	f(UNDIRTY,	__CEPH_OSD_OP(WR, DATA, 28),	"undirty")	    \
	f(ISDIRTY,	__CEPH_OSD_OP(RD, DATA, 29),	"isdirty")	    \
	f(COPY_GET,	__CEPH_OSD_OP(RD, DATA, 30),	"copy-get")	    \
	f(CACHE_FLUSH,	__CEPH_OSD_OP(CACHE, DATA, 31),	"cache-flush")	    \
	f(CACHE_EVICT,	__CEPH_OSD_OP(CACHE, DATA, 32),	"cache-evict")	    \
	f(CACHE_TRY_FLUSH, __CEPH_OSD_OP(CACHE, DATA, 33), "cache-try-flush") \
									    \
	/* convert tmap to omap */					    \
	f(TMAP2OMAP,	__CEPH_OSD_OP(RMW, DATA, 34),	"tmap2omap")	    \
									    \
	/* hints */							    \
	f(SETALLOCHINT,	__CEPH_OSD_OP(WR, DATA, 35),	"set-alloc-hint")   \
									    \
	/** multi **/							    \
	f(CLONERANGE,	__CEPH_OSD_OP(WR, MULTI, 1),	"clonerange")	    \
	f(ASSERT_SRC_VERSION, __CEPH_OSD_OP(RD, MULTI, 2), "assert-src-version") \
	f(SRC_CMPXATTR,	__CEPH_OSD_OP(RD, MULTI, 3),	"src-cmpxattr")	    \
									    \
	/** attrs **/							    \
	/* read */							    \
	f(GETXATTR,	__CEPH_OSD_OP(RD, ATTR, 1),	"getxattr")	    \
	f(GETXATTRS,	__CEPH_OSD_OP(RD, ATTR, 2),	"getxattrs")	    \
	f(CMPXATTR,	__CEPH_OSD_OP(RD, ATTR, 3),	"cmpxattr")	    \
									    \
	/* write */							    \
	f(SETXATTR,	__CEPH_OSD_OP(WR, ATTR, 1),	"setxattr")	    \
	f(SETXATTRS,	__CEPH_OSD_OP(WR, ATTR, 2),	"setxattrs")	    \
	f(RESETXATTRS,	__CEPH_OSD_OP(WR, ATTR, 3),	"resetxattrs")	    \
	f(RMXATTR,	__CEPH_OSD_OP(WR, ATTR, 4),	"rmxattr")	    \
									    \
	/** subop **/							    \
	f(PULL,		__CEPH_OSD_OP1(SUB, 1),		"pull")		    \
	f(PUSH,		__CEPH_OSD_OP1(SUB, 2),		"push")		    \
	f(BALANCEREADS,	__CEPH_OSD_OP1(SUB, 3),		"balance-reads")    \
	f(UNBALANCEREADS, __CEPH_OSD_OP1(SUB, 4),	"unbalance-reads")  \
	f(SCRUB,	__CEPH_OSD_OP1(SUB, 5),		"scrub")	    \
	f(SCRUB_RESERVE, __CEPH_OSD_OP1(SUB, 6),	"scrub-reserve")    \
	f(SCRUB_UNRESERVE, __CEPH_OSD_OP1(SUB, 7),	"scrub-unreserve")  \
	f(SCRUB_STOP,	__CEPH_OSD_OP1(SUB, 8),		"scrub-stop")	    \
	f(SCRUB_MAP,	__CEPH_OSD_OP1(SUB, 9),		"scrub-map")	    \
									    \
	/** lock **/							    \
	f(WRLOCK,	__CEPH_OSD_OP(WR, LOCK, 1),	"wrlock")	    \
	f(WRUNLOCK,	__CEPH_OSD_OP(WR, LOCK, 2),	"wrunlock")	    \
	f(RDLOCK,	__CEPH_OSD_OP(WR, LOCK, 3),	"rdlock")	    \
	f(RDUNLOCK,	__CEPH_OSD_OP(WR, LOCK, 4),	"rdunlock")	    \
	f(UPLOCK,	__CEPH_OSD_OP(WR, LOCK, 5),	"uplock")	    \
	f(DNLOCK,	__CEPH_OSD_OP(WR, LOCK, 6),	"dnlock")	    \
									    \
	/** exec **/							    \
	/* note: the RD bit here is wrong; see special-case below in helper */ \
	f(CALL,		__CEPH_OSD_OP(RD, EXEC, 1),	"call")		    \
									    \
	/** pg **/							    \
	f(PGLS,		__CEPH_OSD_OP(RD, PG, 1),	"pgls")		    \
	f(PGLS_FILTER,	__CEPH_OSD_OP(RD, PG, 2),	"pgls-filter")	    \
	f(PG_HITSET_LS,	__CEPH_OSD_OP(RD, PG, 3),	"pg-hitset-ls")	    \
	f(PG_HITSET_GET, __CEPH_OSD_OP(RD, PG, 4),	"pg-hitset-get")

enum {
#define GENERATE_ENUM_ENTRY(op, opcode, str)	CEPH_OSD_OP_##op = (opcode),
__CEPH_FORALL_OSD_OPS(GENERATE_ENUM_ENTRY)
#undef GENERATE_ENUM_ENTRY
};

static inline int ceph_osd_op_type_lock(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_LOCK;
}
static inline int ceph_osd_op_type_data(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_DATA;
}
static inline int ceph_osd_op_type_attr(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_ATTR;
}
static inline int ceph_osd_op_type_exec(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_EXEC;
}
static inline int ceph_osd_op_type_pg(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_PG;
}
static inline int ceph_osd_op_type_multi(int op)
{
	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_MULTI;
}

static inline int ceph_osd_op_mode_subop(int op)
{
	return (op & CEPH_OSD_OP_MODE) == CEPH_OSD_OP_MODE_SUB;
}
static inline int ceph_osd_op_mode_read(int op)
{
	return (op & CEPH_OSD_OP_MODE_RD) &&
		op != CEPH_OSD_OP_CALL;
}
static inline int ceph_osd_op_mode_modify(int op)
{
	return op & CEPH_OSD_OP_MODE_WR;
}

/*
 * note that the following tmap stuff is also defined in the ceph librados.h
 * any modification here needs to be updated there
 */
#define CEPH_OSD_TMAP_HDR 'h'
#define CEPH_OSD_TMAP_SET 's'
#define CEPH_OSD_TMAP_CREATE 'c' /* create key */
#define CEPH_OSD_TMAP_RM  'r'
#define CEPH_OSD_TMAP_RMSLOPPY 'R'

extern const char *ceph_osd_op_name(int op);

/*
 * osd op flags
 *
 * An op may be READ, WRITE, or READ|WRITE.
 */
enum {
	CEPH_OSD_FLAG_ACK =            0x0001,  /* want (or is) "ack" ack */
	CEPH_OSD_FLAG_ONNVRAM =        0x0002,  /* want (or is) "onnvram" ack */
	CEPH_OSD_FLAG_ONDISK =         0x0004,  /* want (or is) "ondisk" ack */
	CEPH_OSD_FLAG_RETRY =          0x0008,  /* resend attempt */
	CEPH_OSD_FLAG_READ =           0x0010,  /* op may read */
	CEPH_OSD_FLAG_WRITE =          0x0020,  /* op may write */
	CEPH_OSD_FLAG_ORDERSNAP =      0x0040,  /* EOLDSNAP if snapc is out of order */
	CEPH_OSD_FLAG_PEERSTAT_OLD =   0x0080,  /* DEPRECATED msg includes osd_peer_stat */
	CEPH_OSD_FLAG_BALANCE_READS =  0x0100,
	CEPH_OSD_FLAG_PARALLELEXEC =   0x0200,  /* execute op in parallel */
	CEPH_OSD_FLAG_PGOP =           0x0400,  /* pg op, no object */
	CEPH_OSD_FLAG_EXEC =           0x0800,  /* op may exec */
	CEPH_OSD_FLAG_EXEC_PUBLIC =    0x1000,  /* DEPRECATED op may exec (public) */
	CEPH_OSD_FLAG_LOCALIZE_READS = 0x2000,  /* read from nearby replica, if any */
	CEPH_OSD_FLAG_RWORDERED =      0x4000,  /* order wrt concurrent reads */
	CEPH_OSD_FLAG_IGNORE_CACHE =   0x8000,  /* ignore cache logic */
	CEPH_OSD_FLAG_SKIPRWLOCKS =   0x10000,  /* skip rw locks */
	CEPH_OSD_FLAG_IGNORE_OVERLAY = 0x20000, /* ignore pool overlay */
	CEPH_OSD_FLAG_FLUSH =         0x40000,  /* this is part of flush */
	CEPH_OSD_FLAG_MAP_SNAP_CLONE = 0x80000,  /* map snap direct to clone id */
	CEPH_OSD_FLAG_ENFORCE_SNAPC   = 0x100000,  /* use snapc provided even if
						      pool uses pool snaps */
	CEPH_OSD_FLAG_REDIRECTED   = 0x200000,  /* op has been redirected */
	CEPH_OSD_FLAG_KNOWN_REDIR = 0x400000,  /* redirect bit is authoritative */
	CEPH_OSD_FLAG_FULL_TRY =    0x800000,  /* try op despite full flag */
	CEPH_OSD_FLAG_FULL_FORCE = 0x1000000,  /* force op despite full flag */
};

enum {
	CEPH_OSD_OP_FLAG_EXCL = 1,      /* EXCL object create */
	CEPH_OSD_OP_FLAG_FAILOK = 2,    /* continue despite failure */
};

#define EOLDSNAPC    ERESTART  /* ORDERSNAP flag set; writer has old snapc*/
#define EBLACKLISTED ESHUTDOWN /* blacklisted */

/* xattr comparison */
enum {
	CEPH_OSD_CMPXATTR_OP_NOP = 0,
	CEPH_OSD_CMPXATTR_OP_EQ  = 1,
	CEPH_OSD_CMPXATTR_OP_NE  = 2,
	CEPH_OSD_CMPXATTR_OP_GT  = 3,
	CEPH_OSD_CMPXATTR_OP_GTE = 4,
	CEPH_OSD_CMPXATTR_OP_LT  = 5,
	CEPH_OSD_CMPXATTR_OP_LTE = 6
};

enum {
	CEPH_OSD_CMPXATTR_MODE_STRING = 1,
	CEPH_OSD_CMPXATTR_MODE_U64    = 2
};

enum {
	CEPH_OSD_WATCH_OP_UNWATCH = 0,
	CEPH_OSD_WATCH_OP_LEGACY_WATCH = 1,
	/* note: use only ODD ids to prevent pre-giant code from
	   interpreting the op as UNWATCH */
	CEPH_OSD_WATCH_OP_WATCH = 3,
	CEPH_OSD_WATCH_OP_RECONNECT = 5,
	CEPH_OSD_WATCH_OP_PING = 7,
};

const char *ceph_osd_watch_op_name(int o);

enum {
	CEPH_OSD_BACKOFF_OP_BLOCK = 1,
	CEPH_OSD_BACKOFF_OP_ACK_BLOCK = 2,
	CEPH_OSD_BACKOFF_OP_UNBLOCK = 3,
};

/*
 * an individual object operation.  each may be accompanied by some data
 * payload
 */
struct ceph_osd_op {
	__le16 op;           /* CEPH_OSD_OP_* */
	__le32 flags;        /* CEPH_OSD_OP_FLAG_* */
	union {
		struct {
			__le64 offset, length;
			__le64 truncate_size;
			__le32 truncate_seq;
		} __attribute__ ((packed)) extent;
		struct {
			__le32 name_len;
			__le32 value_len;
			__u8 cmp_op;       /* CEPH_OSD_CMPXATTR_OP_* */
			__u8 cmp_mode;     /* CEPH_OSD_CMPXATTR_MODE_* */
		} __attribute__ ((packed)) xattr;
		struct {
			__u8 class_len;
			__u8 method_len;
			__u8 argc;
			__le32 indata_len;
		} __attribute__ ((packed)) cls;
		struct {
			__le64 cookie, count;
		} __attribute__ ((packed)) pgls;
	        struct {
		        __le64 snapid;
	        } __attribute__ ((packed)) snap;
		struct {
			__le64 cookie;
			__le64 ver;     /* no longer used */
			__u8 op;	/* CEPH_OSD_WATCH_OP_* */
			__le32 gen;     /* registration generation */
		} __attribute__ ((packed)) watch;
		struct {
			__le64 cookie;
		} __attribute__ ((packed)) notify;
		struct {
			__le64 offset, length;
			__le64 src_offset;
		} __attribute__ ((packed)) clonerange;
		struct {
			__le64 expected_object_size;
			__le64 expected_write_size;
		} __attribute__ ((packed)) alloc_hint;
	};
	__le32 payload_len;
} __attribute__ ((packed));


#endif
Commit	Line	Data
5cd068c2 SW	1	#ifndef CEPH_RADOS_H
5cd068c2 SW	2	#define CEPH_RADOS_H
0dee3c28 SW	3
	4	/*
	5	* Data types for the Ceph distributed object storage layer RADOS
	6	* (Reliable Autonomic Distributed Object Store).
	7	*/
	8
a1ce3928	9	#include <linux/ceph/msgr.h>
0dee3c28 SW	10
	11	/*
	12	* fs id
	13	*/
	14	struct ceph_fsid {
	15	unsigned char fsid[16];
	16	};
	17
	18	static inline int ceph_fsid_compare(const struct ceph_fsid *a,
	19	const struct ceph_fsid *b)
	20	{
	21	return memcmp(a, b, sizeof(*a));
	22	}
	23
	24	/*
	25	* ino, object, etc.
	26	*/
	27	typedef __le64 ceph_snapid_t;
	28	#define CEPH_SNAPDIR ((__u64)(-1)) /* reserved for hidden .snap dir */
	29	#define CEPH_NOSNAP ((__u64)(-2)) /* "head", "live" revision */
	30	#define CEPH_MAXSNAP ((__u64)(-3)) /* largest valid snapid */
	31
	32	struct ceph_timespec {
	33	__le32 tv_sec;
	34	__le32 tv_nsec;
	35	} __attribute__ ((packed));
	36
	37
	38	/*
	39	* object layout - how objects are mapped into PGs
	40	*/
	41	#define CEPH_OBJECT_LAYOUT_HASH 1
	42	#define CEPH_OBJECT_LAYOUT_LINEAR 2
	43	#define CEPH_OBJECT_LAYOUT_HASHINO 3
	44
	45	/*
	46	* pg layout -- how PGs are mapped onto (sets of) OSDs
	47	*/
	48	#define CEPH_PG_LAYOUT_CRUSH 0
	49	#define CEPH_PG_LAYOUT_HASH 1
	50	#define CEPH_PG_LAYOUT_LINEAR 2
	51	#define CEPH_PG_LAYOUT_HYBRID 3
	52
083a51fb	53	#define CEPH_PG_MAX_SIZE 32 /* max # osds in a single pg */
0dee3c28 SW	54
	55	/*
	56	* placement group.
	57	* we encode this into one __le64.
	58	*/
12979354	59	struct ceph_pg_v1 {
51042122 SW	60	__le16 preferred; /* preferred primary osd */
	61	__le16 ps; /* placement seed */
	62	__le32 pool; /* object pool */
0dee3c28 SW	63	} __attribute__ ((packed));
	64
	65	/*
	66	* pg_pool is a set of pgs storing a pool of objects
	67	*
	68	* pg_num -- base number of pseudorandomly placed pgs
	69	*
	70	* pgp_num -- effective number when calculating pg placement. this
	71	* is used for pg_num increases. new pgs result in data being "split"
	72	* into new pgs. for this to proceed smoothly, new pgs are intiially
	73	* colocated with their parents; that is, pgp_num doesn't increase
	74	* until the new pgs have successfully split. only _then_ are the new
	75	* pgs placed independently.
	76	*
	77	* lpg_num -- localized pg count (per device). replicas are randomly
	78	* selected.
	79	*
	80	* lpgp_num -- as above.
	81	*/
86b00e0d AE	82	#define CEPH_NOPOOL ((__u64) (-1)) /* pool id not defined */
86b00e0d AE	83
2abebdbc ID	84	#define CEPH_POOL_TYPE_REP 1
	85	#define CEPH_POOL_TYPE_RAID4 2 /* never implemented */
	86	#define CEPH_POOL_TYPE_EC 3
0dee3c28 SW	87
	88	/*
	89	* stable_mod func is used to control number of placement groups.
	90	* similar to straight-up modulo, but produces a stable mapping as b
	91	* increases over time. b is the number of bins, and bmask is the
	92	* containing power of 2 minus 1.
	93	*
	94	* b <= bmask and bmask=(2**n)-1
	95	* e.g., b=12 -> bmask=15, b=123 -> bmask=127
	96	*/
	97	static inline int ceph_stable_mod(int x, int b, int bmask)
	98	{
	99	if ((x & bmask) < b)
	100	return x & bmask;
	101	else
	102	return x & (bmask >> 1);
	103	}
	104
	105	/*
	106	* object layout - how a given object should be stored.
	107	*/
	108	struct ceph_object_layout {
12979354	109	struct ceph_pg_v1 ol_pgid; /* raw pg, with _full_ ps precision. */
0dee3c28 SW	110	__le32 ol_stripe_unit; /* for per-object parity, if any */
	111	} __attribute__ ((packed));
	112
	113	/*
	114	* compound epoch+version, used by storage layer to serialize mutations
	115	*/
	116	struct ceph_eversion {
0dee3c28	117	__le64 version;
985c1673	118	__le32 epoch;
0dee3c28 SW	119	} __attribute__ ((packed));
	120
	121	/*
	122	* osd map bits
	123	*/
	124
	125	/* status bits */
0315a777 AE	126	#define CEPH_OSD_EXISTS (1<<0)
	127	#define CEPH_OSD_UP (1<<1)
	128	#define CEPH_OSD_AUTOOUT (1<<2) /* osd was automatically marked out */
	129	#define CEPH_OSD_NEW (1<<3) /* osd is new, never marked in */
0dee3c28	130
4b568b1a AE	131	extern const char *ceph_osd_state_name(int s);
4b568b1a AE	132
0dee3c28 SW	133	/* osd weights. fixed point value: 0x10000 == 1.0 ("in"), 0 == "out" */
	134	#define CEPH_OSD_IN 0x10000
	135	#define CEPH_OSD_OUT 0
	136
2cfa34f2 ID	137	/* osd primary-affinity. fixed point value: 0x10000 == baseline */
	138	#define CEPH_OSD_MAX_PRIMARY_AFFINITY 0x10000
	139	#define CEPH_OSD_DEFAULT_PRIMARY_AFFINITY 0x10000
	140
0dee3c28 SW	141
	142	/*
	143	* osd map flag bits
	144	*/
	145	#define CEPH_OSDMAP_NEARFULL (1<<0) /* sync writes (near ENOSPC) */
	146	#define CEPH_OSDMAP_FULL (1<<1) /* no data writes (ENOSPC) */
	147	#define CEPH_OSDMAP_PAUSERD (1<<2) /* pause all reads */
	148	#define CEPH_OSDMAP_PAUSEWR (1<<3) /* pause all writes */
	149	#define CEPH_OSDMAP_PAUSEREC (1<<4) /* pause recovery */
0315a777 AE	150	#define CEPH_OSDMAP_NOUP (1<<5) /* block osd boot */
	151	#define CEPH_OSDMAP_NODOWN (1<<6) /* block osd mark-down/failure */
	152	#define CEPH_OSDMAP_NOOUT (1<<7) /* block osd auto mark-out */
	153	#define CEPH_OSDMAP_NOIN (1<<8) /* block osd auto mark-in */
	154	#define CEPH_OSDMAP_NOBACKFILL (1<<9) /* block osd backfill */
	155	#define CEPH_OSDMAP_NORECOVER (1<<10) /* block osd recovery and backfill */
63244fa1 ID	156	#define CEPH_OSDMAP_NOSCRUB (1<<11) /* block periodic scrub */
	157	#define CEPH_OSDMAP_NODEEP_SCRUB (1<<12) /* block periodic deep-scrub */
	158	#define CEPH_OSDMAP_NOTIERAGENT (1<<13) /* disable tiering agent */
	159	#define CEPH_OSDMAP_NOREBALANCE (1<<14) /* block osd backfill unless pg is degraded */
	160	#define CEPH_OSDMAP_SORTBITWISE (1<<15) /* use bitwise hobject_t sort */
ae78dd81 ID	161	#define CEPH_OSDMAP_REQUIRE_JEWEL (1<<16) /* require jewel for booting osds */
	162	#define CEPH_OSDMAP_REQUIRE_KRAKEN (1<<17) /* require kraken for booting osds */
	163	#define CEPH_OSDMAP_REQUIRE_LUMINOUS (1<<18) /* require l for booting osds */
	164	#define CEPH_OSDMAP_RECOVERY_DELETES (1<<19) /* deletes performed during recovery instead of peering */
0315a777 AE	165
	166	/*
	167	* The error code to return when an OSD can't handle a write
	168	* because it is too large.
	169	*/
	170	#define OSD_WRITETOOBIG EMSGSIZE
0dee3c28 SW	171
	172	/*
	173	* osd ops
0315a777 AE	174	*
	175	* WARNING: do not use these op codes directly. Use the helpers
	176	* defined below instead. In certain cases, op code behavior was
	177	* redefined, resulting in special-cases in the helpers.
0dee3c28 SW	178	*/
	179	#define CEPH_OSD_OP_MODE 0xf000
	180	#define CEPH_OSD_OP_MODE_RD 0x1000
	181	#define CEPH_OSD_OP_MODE_WR 0x2000
	182	#define CEPH_OSD_OP_MODE_RMW 0x3000
	183	#define CEPH_OSD_OP_MODE_SUB 0x4000
70b5bfa3	184	#define CEPH_OSD_OP_MODE_CACHE 0x8000
0dee3c28 SW	185
	186	#define CEPH_OSD_OP_TYPE 0x0f00
	187	#define CEPH_OSD_OP_TYPE_LOCK 0x0100
	188	#define CEPH_OSD_OP_TYPE_DATA 0x0200
	189	#define CEPH_OSD_OP_TYPE_ATTR 0x0300
	190	#define CEPH_OSD_OP_TYPE_EXEC 0x0400
	191	#define CEPH_OSD_OP_TYPE_PG 0x0500
0315a777	192	#define CEPH_OSD_OP_TYPE_MULTI 0x0600 /* multiobject */
0dee3c28	193
70b5bfa3 ID	194	#define __CEPH_OSD_OP1(mode, nr) \
	195	(CEPH_OSD_OP_MODE_##mode \| (nr))
	196
	197	#define __CEPH_OSD_OP(mode, type, nr) \
	198	(CEPH_OSD_OP_MODE_##mode \| CEPH_OSD_OP_TYPE_##type \| (nr))
	199
	200	#define __CEPH_FORALL_OSD_OPS(f) \
	201	/ data / \
	202	/* read */ \
	203	f(READ, __CEPH_OSD_OP(RD, DATA, 1), "read") \
	204	f(STAT, __CEPH_OSD_OP(RD, DATA, 2), "stat") \
	205	f(MAPEXT, __CEPH_OSD_OP(RD, DATA, 3), "mapext") \
	206	\
	207	/* fancy read */ \
	208	f(MASKTRUNC, __CEPH_OSD_OP(RD, DATA, 4), "masktrunc") \
	209	f(SPARSE_READ, __CEPH_OSD_OP(RD, DATA, 5), "sparse-read") \
	210	\
	211	f(NOTIFY, __CEPH_OSD_OP(RD, DATA, 6), "notify") \
	212	f(NOTIFY_ACK, __CEPH_OSD_OP(RD, DATA, 7), "notify-ack") \
	213	\
	214	/* versioning */ \
	215	f(ASSERT_VER, __CEPH_OSD_OP(RD, DATA, 8), "assert-version") \
	216	\
	217	f(LIST_WATCHERS, __CEPH_OSD_OP(RD, DATA, 9), "list-watchers") \
	218	\
	219	f(LIST_SNAPS, __CEPH_OSD_OP(RD, DATA, 10), "list-snaps") \
	220	\
	221	/* sync */ \
	222	f(SYNC_READ, __CEPH_OSD_OP(RD, DATA, 11), "sync_read") \
	223	\
	224	/* write */ \
	225	f(WRITE, __CEPH_OSD_OP(WR, DATA, 1), "write") \
	226	f(WRITEFULL, __CEPH_OSD_OP(WR, DATA, 2), "writefull") \
	227	f(TRUNCATE, __CEPH_OSD_OP(WR, DATA, 3), "truncate") \
	228	f(ZERO, __CEPH_OSD_OP(WR, DATA, 4), "zero") \
	229	f(DELETE, __CEPH_OSD_OP(WR, DATA, 5), "delete") \
	230	\
	231	/* fancy write */ \
	232	f(APPEND, __CEPH_OSD_OP(WR, DATA, 6), "append") \
	233	f(STARTSYNC, __CEPH_OSD_OP(WR, DATA, 7), "startsync") \
	234	f(SETTRUNC, __CEPH_OSD_OP(WR, DATA, 8), "settrunc") \
	235	f(TRIMTRUNC, __CEPH_OSD_OP(WR, DATA, 9), "trimtrunc") \
	236	\
	237	f(TMAPUP, __CEPH_OSD_OP(RMW, DATA, 10), "tmapup") \
	238	f(TMAPPUT, __CEPH_OSD_OP(WR, DATA, 11), "tmapput") \
	239	f(TMAPGET, __CEPH_OSD_OP(RD, DATA, 12), "tmapget") \
	240	\
	241	f(CREATE, __CEPH_OSD_OP(WR, DATA, 13), "create") \
	242	f(ROLLBACK, __CEPH_OSD_OP(WR, DATA, 14), "rollback") \
	243	\
	244	f(WATCH, __CEPH_OSD_OP(WR, DATA, 15), "watch") \
	245	\
	246	/* omap */ \
	247	f(OMAPGETKEYS, __CEPH_OSD_OP(RD, DATA, 17), "omap-get-keys") \
	248	f(OMAPGETVALS, __CEPH_OSD_OP(RD, DATA, 18), "omap-get-vals") \
	249	f(OMAPGETHEADER, __CEPH_OSD_OP(RD, DATA, 19), "omap-get-header") \
	250	f(OMAPGETVALSBYKEYS, __CEPH_OSD_OP(RD, DATA, 20), "omap-get-vals-by-keys") \
	251	f(OMAPSETVALS, __CEPH_OSD_OP(WR, DATA, 21), "omap-set-vals") \
	252	f(OMAPSETHEADER, __CEPH_OSD_OP(WR, DATA, 22), "omap-set-header") \
	253	f(OMAPCLEAR, __CEPH_OSD_OP(WR, DATA, 23), "omap-clear") \
	254	f(OMAPRMKEYS, __CEPH_OSD_OP(WR, DATA, 24), "omap-rm-keys") \
	255	f(OMAP_CMP, __CEPH_OSD_OP(RD, DATA, 25), "omap-cmp") \
	256	\
	257	/* tiering */ \
258	f(COPY_FROM, __CEPH_OSD_OP(WR, DATA, 26), "copy-from") \
259	f(COPY_GET_CLASSIC, __CEPH_OSD_OP(RD, DATA, 27), "copy-get-classic") \
260	f(UNDIRTY, __CEPH_OSD_OP(WR, DATA, 28), "undirty") \
261	f(ISDIRTY, __CEPH_OSD_OP(RD, DATA, 29), "isdirty") \
262	f(COPY_GET, __CEPH_OSD_OP(RD, DATA, 30), "copy-get") \
263	f(CACHE_FLUSH, __CEPH_OSD_OP(CACHE, DATA, 31), "cache-flush") \
264	f(CACHE_EVICT, __CEPH_OSD_OP(CACHE, DATA, 32), "cache-evict") \
265	f(CACHE_TRY_FLUSH, __CEPH_OSD_OP(CACHE, DATA, 33), "cache-try-flush") \
266	\
267	/* convert tmap to omap */ \
268	f(TMAP2OMAP, __CEPH_OSD_OP(RMW, DATA, 34), "tmap2omap") \
269	\
270	/* hints */ \
271	f(SETALLOCHINT, __CEPH_OSD_OP(WR, DATA, 35), "set-alloc-hint") \
272	\
273	/ multi / \
274	f(CLONERANGE, __CEPH_OSD_OP(WR, MULTI, 1), "clonerange") \
275	f(ASSERT_SRC_VERSION, __CEPH_OSD_OP(RD, MULTI, 2), "assert-src-version") \
276	f(SRC_CMPXATTR, __CEPH_OSD_OP(RD, MULTI, 3), "src-cmpxattr") \
277	\
278	/ attrs / \
279	/* read */ \
280	f(GETXATTR, __CEPH_OSD_OP(RD, ATTR, 1), "getxattr") \
281	f(GETXATTRS, __CEPH_OSD_OP(RD, ATTR, 2), "getxattrs") \
282	f(CMPXATTR, __CEPH_OSD_OP(RD, ATTR, 3), "cmpxattr") \
283	\
284	/* write */ \
285	f(SETXATTR, __CEPH_OSD_OP(WR, ATTR, 1), "setxattr") \
286	f(SETXATTRS, __CEPH_OSD_OP(WR, ATTR, 2), "setxattrs") \
287	f(RESETXATTRS, __CEPH_OSD_OP(WR, ATTR, 3), "resetxattrs") \
288	f(RMXATTR, __CEPH_OSD_OP(WR, ATTR, 4), "rmxattr") \
289	\
290	/ subop / \
291	f(PULL, __CEPH_OSD_OP1(SUB, 1), "pull") \
292	f(PUSH, __CEPH_OSD_OP1(SUB, 2), "push") \
293	f(BALANCEREADS, __CEPH_OSD_OP1(SUB, 3), "balance-reads") \
294	f(UNBALANCEREADS, __CEPH_OSD_OP1(SUB, 4), "unbalance-reads") \
295	f(SCRUB, __CEPH_OSD_OP1(SUB, 5), "scrub") \
296	f(SCRUB_RESERVE, __CEPH_OSD_OP1(SUB, 6), "scrub-reserve") \
297	f(SCRUB_UNRESERVE, __CEPH_OSD_OP1(SUB, 7), "scrub-unreserve") \
298	f(SCRUB_STOP, __CEPH_OSD_OP1(SUB, 8), "scrub-stop") \
299	f(SCRUB_MAP, __CEPH_OSD_OP1(SUB, 9), "scrub-map") \
300	\
301	/ lock / \
302	f(WRLOCK, __CEPH_OSD_OP(WR, LOCK, 1), "wrlock") \
303	f(WRUNLOCK, __CEPH_OSD_OP(WR, LOCK, 2), "wrunlock") \
304	f(RDLOCK, __CEPH_OSD_OP(WR, LOCK, 3), "rdlock") \
305	f(RDUNLOCK, __CEPH_OSD_OP(WR, LOCK, 4), "rdunlock") \
306	f(UPLOCK, __CEPH_OSD_OP(WR, LOCK, 5), "uplock") \
307	f(DNLOCK, __CEPH_OSD_OP(WR, LOCK, 6), "dnlock") \
308	\
309	/ exec / \
310	/* note: the RD bit here is wrong; see special-case below in helper */ \
311	f(CALL, __CEPH_OSD_OP(RD, EXEC, 1), "call") \
312	\
313	/ pg / \
314	f(PGLS, __CEPH_OSD_OP(RD, PG, 1), "pgls") \
315	f(PGLS_FILTER, __CEPH_OSD_OP(RD, PG, 2), "pgls-filter") \
316	f(PG_HITSET_LS, __CEPH_OSD_OP(RD, PG, 3), "pg-hitset-ls") \
317	f(PG_HITSET_GET, __CEPH_OSD_OP(RD, PG, 4), "pg-hitset-get")
483fac71	318
70b5bfa3 ID	319	enum {
	320	#define GENERATE_ENUM_ENTRY(op, opcode, str) CEPH_OSD_OP_##op = (opcode),
	321	__CEPH_FORALL_OSD_OPS(GENERATE_ENUM_ENTRY)
	322	#undef GENERATE_ENUM_ENTRY
0dee3c28 SW	323	};
	324
	325	static inline int ceph_osd_op_type_lock(int op)
	326	{
	327	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_LOCK;
	328	}
	329	static inline int ceph_osd_op_type_data(int op)
	330	{
	331	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_DATA;
	332	}
	333	static inline int ceph_osd_op_type_attr(int op)
	334	{
	335	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_ATTR;
	336	}
	337	static inline int ceph_osd_op_type_exec(int op)
	338	{
	339	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_EXEC;
	340	}
	341	static inline int ceph_osd_op_type_pg(int op)
	342	{
	343	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_PG;
	344	}
0315a777 AE	345	static inline int ceph_osd_op_type_multi(int op)
	346	{
	347	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_MULTI;
	348	}
0dee3c28 SW	349
	350	static inline int ceph_osd_op_mode_subop(int op)
	351	{
	352	return (op & CEPH_OSD_OP_MODE) == CEPH_OSD_OP_MODE_SUB;
	353	}
	354	static inline int ceph_osd_op_mode_read(int op)
	355	{
0315a777 AE	356	return (op & CEPH_OSD_OP_MODE_RD) &&
0315a777 AE	357	op != CEPH_OSD_OP_CALL;
0dee3c28 SW	358	}
	359	static inline int ceph_osd_op_mode_modify(int op)
	360	{
0315a777	361	return op & CEPH_OSD_OP_MODE_WR;
0dee3c28 SW	362	}
0dee3c28 SW	363
f0b18d9f SW	364	/*
	365	* note that the following tmap stuff is also defined in the ceph librados.h
	366	* any modification here needs to be updated there
	367	*/
0dee3c28 SW	368	#define CEPH_OSD_TMAP_HDR 'h'
0dee3c28 SW	369	#define CEPH_OSD_TMAP_SET 's'
0315a777	370	#define CEPH_OSD_TMAP_CREATE 'c' /* create key */
0dee3c28	371	#define CEPH_OSD_TMAP_RM 'r'
0315a777	372	#define CEPH_OSD_TMAP_RMSLOPPY 'R'
0dee3c28 SW	373
	374	extern const char *ceph_osd_op_name(int op);
	375
0dee3c28 SW	376	/*
	377	* osd op flags
	378	*
	379	* An op may be READ, WRITE, or READ\|WRITE.
	380	*/
	381	enum {
0315a777 AE	382	CEPH_OSD_FLAG_ACK = 0x0001, /* want (or is) "ack" ack */
	383	CEPH_OSD_FLAG_ONNVRAM = 0x0002, /* want (or is) "onnvram" ack */
	384	CEPH_OSD_FLAG_ONDISK = 0x0004, /* want (or is) "ondisk" ack */
	385	CEPH_OSD_FLAG_RETRY = 0x0008, /* resend attempt */
	386	CEPH_OSD_FLAG_READ = 0x0010, /* op may read */
	387	CEPH_OSD_FLAG_WRITE = 0x0020, /* op may write */
	388	CEPH_OSD_FLAG_ORDERSNAP = 0x0040, /* EOLDSNAP if snapc is out of order */
	389	CEPH_OSD_FLAG_PEERSTAT_OLD = 0x0080, /* DEPRECATED msg includes osd_peer_stat */
	390	CEPH_OSD_FLAG_BALANCE_READS = 0x0100,
	391	CEPH_OSD_FLAG_PARALLELEXEC = 0x0200, /* execute op in parallel */
	392	CEPH_OSD_FLAG_PGOP = 0x0400, /* pg op, no object */
	393	CEPH_OSD_FLAG_EXEC = 0x0800, /* op may exec */
	394	CEPH_OSD_FLAG_EXEC_PUBLIC = 0x1000, /* DEPRECATED op may exec (public) */
	395	CEPH_OSD_FLAG_LOCALIZE_READS = 0x2000, /* read from nearby replica, if any */
	396	CEPH_OSD_FLAG_RWORDERED = 0x4000, /* order wrt concurrent reads */
1b3f2ab5 ID	397	CEPH_OSD_FLAG_IGNORE_CACHE = 0x8000, /* ignore cache logic */
	398	CEPH_OSD_FLAG_SKIPRWLOCKS = 0x10000, /* skip rw locks */
	399	CEPH_OSD_FLAG_IGNORE_OVERLAY = 0x20000, /* ignore pool overlay */
	400	CEPH_OSD_FLAG_FLUSH = 0x40000, /* this is part of flush */
bb873b53 ID	401	CEPH_OSD_FLAG_MAP_SNAP_CLONE = 0x80000, /* map snap direct to clone id */
	402	CEPH_OSD_FLAG_ENFORCE_SNAPC = 0x100000, /* use snapc provided even if
	403	pool uses pool snaps */
	404	CEPH_OSD_FLAG_REDIRECTED = 0x200000, /* op has been redirected */
	405	CEPH_OSD_FLAG_KNOWN_REDIR = 0x400000, /* redirect bit is authoritative */
	406	CEPH_OSD_FLAG_FULL_TRY = 0x800000, /* try op despite full flag */
	407	CEPH_OSD_FLAG_FULL_FORCE = 0x1000000, /* force op despite full flag */
0dee3c28 SW	408	};
	409
	410	enum {
	411	CEPH_OSD_OP_FLAG_EXCL = 1, /* EXCL object create */
0315a777	412	CEPH_OSD_OP_FLAG_FAILOK = 2, /* continue despite failure */
0dee3c28 SW	413	};
	414
	415	#define EOLDSNAPC ERESTART /* ORDERSNAP flag set; writer has old snapc*/
	416	#define EBLACKLISTED ESHUTDOWN /* blacklisted */
	417
ca9d93a2 SW	418	/* xattr comparison */
	419	enum {
	420	CEPH_OSD_CMPXATTR_OP_NOP = 0,
	421	CEPH_OSD_CMPXATTR_OP_EQ = 1,
	422	CEPH_OSD_CMPXATTR_OP_NE = 2,
	423	CEPH_OSD_CMPXATTR_OP_GT = 3,
	424	CEPH_OSD_CMPXATTR_OP_GTE = 4,
	425	CEPH_OSD_CMPXATTR_OP_LT = 5,
	426	CEPH_OSD_CMPXATTR_OP_LTE = 6
	427	};
	428
	429	enum {
	430	CEPH_OSD_CMPXATTR_MODE_STRING = 1,
	431	CEPH_OSD_CMPXATTR_MODE_U64 = 2
	432	};
	433
922dab61 ID	434	enum {
	435	CEPH_OSD_WATCH_OP_UNWATCH = 0,
	436	CEPH_OSD_WATCH_OP_LEGACY_WATCH = 1,
	437	/* note: use only ODD ids to prevent pre-giant code from
	438	interpreting the op as UNWATCH */
	439	CEPH_OSD_WATCH_OP_WATCH = 3,
	440	CEPH_OSD_WATCH_OP_RECONNECT = 5,
	441	CEPH_OSD_WATCH_OP_PING = 7,
	442	};
	443
	444	const char *ceph_osd_watch_op_name(int o);
483fac71	445
a02a946d ID	446	enum {
	447	CEPH_OSD_BACKOFF_OP_BLOCK = 1,
	448	CEPH_OSD_BACKOFF_OP_ACK_BLOCK = 2,
	449	CEPH_OSD_BACKOFF_OP_UNBLOCK = 3,
	450	};
	451
0dee3c28 SW	452	/*
	453	* an individual object operation. each may be accompanied by some data
	454	* payload
	455	*/
	456	struct ceph_osd_op {
	457	__le16 op; /* CEPH_OSD_OP_* */
7b25bf5f	458	__le32 flags; /* CEPH_OSD_OP_FLAG_* */
0dee3c28 SW	459	union {
	460	struct {
	461	__le64 offset, length;
0c948992 YS	462	__le64 truncate_size;
0c948992 YS	463	__le32 truncate_seq;
0dee3c28 SW	464	} __attribute__ ((packed)) extent;
	465	struct {
	466	__le32 name_len;
	467	__le32 value_len;
ca9d93a2 SW	468	__u8 cmp_op; /* CEPH_OSD_CMPXATTR_OP_* */
ca9d93a2 SW	469	__u8 cmp_mode; /* CEPH_OSD_CMPXATTR_MODE_* */
0dee3c28	470	} __attribute__ ((packed)) xattr;
0dee3c28 SW	471	struct {
	472	__u8 class_len;
	473	__u8 method_len;
	474	__u8 argc;
	475	__le32 indata_len;
	476	} __attribute__ ((packed)) cls;
	477	struct {
	478	__le64 cookie, count;
	479	} __attribute__ ((packed)) pgls;
f0b18d9f SW	480	struct {
	481	__le64 snapid;
	482	} __attribute__ ((packed)) snap;
483fac71 YS	483	struct {
483fac71 YS	484	__le64 cookie;
922dab61 ID	485	__le64 ver; /* no longer used */
	486	__u8 op; /* CEPH_OSD_WATCH_OP_* */
	487	__le32 gen; /* registration generation */
483fac71	488	} __attribute__ ((packed)) watch;
19079203 ID	489	struct {
	490	__le64 cookie;
	491	} __attribute__ ((packed)) notify;
0315a777 AE	492	struct {
	493	__le64 offset, length;
	494	__le64 src_offset;
	495	} __attribute__ ((packed)) clonerange;
c647b8a8 ID	496	struct {
	497	__le64 expected_object_size;
	498	__le64 expected_write_size;
	499	} __attribute__ ((packed)) alloc_hint;
0315a777	500	};
0dee3c28 SW	501	__le32 payload_len;
	502	} __attribute__ ((packed));
	503
0dee3c28 SW	504
0dee3c28 SW	505	#endif