X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Finclude%2Flustre%2Flustre_idl.h;h=c676d4dd4304924e074d917249eb872564470567;hp=13dbe4e4bf54cff16050159f75bb3f4c82fd8ea3;hb=5e9b1828ee5a9cb92b85005c0a34962bee56d137;hpb=6869932b552ac705f411de3362f01bd50c1f6f7d diff --git a/lustre/include/lustre/lustre_idl.h b/lustre/include/lustre/lustre_idl.h index 13dbe4e..c676d4d 100644 --- a/lustre/include/lustre/lustre_idl.h +++ b/lustre/include/lustre/lustre_idl.h @@ -36,7 +36,11 @@ * lustre/include/lustre/lustre_idl.h * * Lustre wire protocol definitions. + */ + +/** \defgroup lustreidl lustreidl * + * Lustre wire protocol definitions. * * We assume all nodes are either little-endian or big-endian, and we * always send messages in the sender's native format. The receiver @@ -79,6 +83,8 @@ * For variable length types, a second 'lustre_swab_v_xxxtypexxx()' routine * may be defined that swabs just the variable part, after the caller has * verified that the message buffer is large enough. + * + * @{ */ #ifndef _LUSTRE_IDL_H_ @@ -88,6 +94,7 @@ /* Defn's shared with user-space. */ #include +#include /* * GENERAL STUFF @@ -215,15 +222,38 @@ static inline int range_is_exhausted(struct lu_range *r) (range)->lr_start, \ (range)->lr_end +/** \defgroup lu_fid lu_fid + * @{ */ + +/** + * File identifier. + * + * Fid is a cluster-wide unique identifier of a file or an object + * (stripe). Fids are never reused. Fids are transmitted across network (in + * the sender byte-ordering), and stored on disk in a packed form (struct + * lu_fid_pack) in a big-endian order. + */ struct lu_fid { - __u64 f_seq; /* holds fid sequence. Lustre should support 2 ^ 64 - * objects, thus even if one sequence has one object we - * will never reach this value. */ - __u32 f_oid; /* fid number within its sequence. */ - __u32 f_ver; /* holds fid version. */ + /** + * fid sequence. Sequence is a unit of migration: all files (objects) + * with fids from a given sequence are stored on the same + * server. + * + * Lustre should support 2 ^ 64 objects, thus even if one + * sequence has one object we will never reach this value. + */ + __u64 f_seq; + /** fid number within sequence. */ + __u32 f_oid; + /** + * fid version, used to distinguish different versions (in the sense + * of snapshots, etc.) of the same file system object. Not currently + * used. + */ + __u32 f_ver; }; -/* +/** * fid constants */ enum { @@ -231,19 +261,19 @@ enum { LUSTRE_FID_INIT_OID = 1UL }; -/* get object sequence */ +/** returns fid object sequence */ static inline __u64 fid_seq(const struct lu_fid *fid) { return fid->f_seq; } -/* get object id */ +/** returns fid object id */ static inline __u32 fid_oid(const struct lu_fid *fid) { return fid->f_oid; } -/* get object version */ +/** returns fid object version */ static inline __u32 fid_ver(const struct lu_fid *fid) { return fid->f_ver; @@ -355,8 +385,7 @@ static inline void fid_be_to_cpu(struct lu_fid *dst, const struct lu_fid *src) LASSERTF(fid_is_igif(dst) || fid_ver(dst) == 0, DFID"\n", PFID(dst)); } -#ifdef __KERNEL__ -/* +/** * Storage representation for fids. * * Variable size, first byte contains the length of the whole record. @@ -371,9 +400,6 @@ void fid_pack(struct lu_fid_pack *pack, const struct lu_fid *fid, struct lu_fid *befider); int fid_unpack(const struct lu_fid_pack *pack, struct lu_fid *fid); -/* __KERNEL__ */ -#endif - static inline int fid_is_sane(const struct lu_fid *fid) { return @@ -402,7 +428,28 @@ static inline int lu_fid_eq(const struct lu_fid *f0, return memcmp(f0, f1, sizeof *f0) == 0; } -/* +#define __diff_normalize(val0, val1) \ +({ \ + typeof(val0) __val0 = (val0); \ + typeof(val1) __val1 = (val1); \ + \ + (__val0 == __val1 ? 0 : __val0 > __val1 ? +1 : -1); \ +}) + +static inline int lu_fid_cmp(const struct lu_fid *f0, + const struct lu_fid *f1) +{ + return + __diff_normalize(fid_seq(f0), fid_seq(f1)) ?: + __diff_normalize(fid_oid(f0), fid_oid(f1)) ?: + __diff_normalize(fid_ver(f0), fid_ver(f1)); +} + +/** @} lu_fid */ + +/** \defgroup lu_dir lu_dir + * @{ */ +/** * Layout of readdir pages, as transmitted on wire. */ struct lu_dirent { @@ -457,6 +504,8 @@ static inline int lu_dirent_size(struct lu_dirent *ent) #define DIR_END_OFF 0xfffffffffffffffeULL +/** @} lu_dir */ + struct lustre_handle { __u64 cookie; }; @@ -544,6 +593,9 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); #define DLM_LOCKREPLY_OFF 1 /* lockrep offset */ #define DLM_REPLY_REC_OFF 2 /* reply record offset */ +/* only use in req->rq_{req,rep}_swab_mask */ +#define MSG_PTLRPC_HEADER_OFF 31 + /* Flags that are operation-specific go in the top 16 bits. */ #define MSG_OP_FLAG_MASK 0xffff0000 #define MSG_OP_FLAG_SHIFT 16 @@ -602,10 +654,10 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); #define OBD_CONNECT_LRU_RESIZE 0x02000000ULL /* Lru resize feature. */ #define OBD_CONNECT_MDS_MDS 0x04000000ULL /* MDS-MDS connection*/ #define OBD_CONNECT_REAL 0x08000000ULL /* real connection */ -#define OBD_CONNECT_CHANGE_QS 0x10000000ULL /*shrink/enlarge qunit size - *b=10600 */ +#define OBD_CONNECT_CHANGE_QS 0x10000000ULL /* shrink/enlarge qunit b=10600 */ #define OBD_CONNECT_CKSUM 0x20000000ULL /* support several cksum algos */ #define OBD_CONNECT_FID 0x40000000ULL /* FID is supported by server */ +#define OBD_CONNECT_LOV_V3 0x100000000ULL /* client supports lov v3 ea */ /* also update obd_connect_names[] for lprocfs_rd_connect_flags() * and lustre/utils/wirecheck.c */ @@ -625,7 +677,8 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); OBD_CONNECT_MDS_CAPA | OBD_CONNECT_OSS_CAPA | \ OBD_CONNECT_MDS_MDS | OBD_CONNECT_CANCELSET | \ OBD_CONNECT_FID | \ - LRU_RESIZE_CONNECT_FLAG | OBD_CONNECT_AT) + LRU_RESIZE_CONNECT_FLAG | OBD_CONNECT_AT | \ + OBD_CONNECT_LOV_V3) #define OST_CONNECT_SUPPORTED (OBD_CONNECT_SRVLOCK | OBD_CONNECT_GRANT | \ OBD_CONNECT_REQPORTAL | OBD_CONNECT_VERSION | \ OBD_CONNECT_TRUNCLOCK | OBD_CONNECT_INDEX | \ @@ -730,7 +783,7 @@ typedef __u32 obd_count; #define OBD_FL_NO_USRQUOTA (0x00000100) /* the object's owner is over quota */ #define OBD_FL_NO_GRPQUOTA (0x00000200) /* the object's group is over quota */ -/* +/** * Set this to delegate DLM locking during obd_punch() to the OSTs. Only OSTs * that declared OBD_CONNECT_TRUNCLOCK in their connect flags support this * functionality. @@ -744,54 +797,10 @@ typedef __u32 obd_count; #define OBD_FL_CKSUM_ADLER (0x00002000) #define OBD_FL_CKSUM_ALL (OBD_FL_CKSUM_CRC32 | OBD_FL_CKSUM_ADLER) -/* - * This should not be smaller than sizeof(struct lustre_handle) + sizeof(struct - * llog_cookie) + sizeof(struct ll_fid). Nevertheless struct ll_fid is not - * longer stored in o_inline, we keep this just for case. - */ -#define OBD_INLINESZ 80 - -/* Note: 64-bit types are 64-bit aligned in structure */ -struct obdo { - obd_valid o_valid; /* hot fields in this obdo */ - obd_id o_id; - obd_gr o_gr; - obd_id o_fid; - obd_size o_size; /* o_size-o_blocks == ost_lvb */ - obd_time o_mtime; - obd_time o_atime; - obd_time o_ctime; - obd_blocks o_blocks; /* brw: cli sent cached bytes */ - obd_size o_grant; - - /* 32-bit fields start here: keep an even number of them via padding */ - obd_blksize o_blksize; /* optimal IO blocksize */ - obd_mode o_mode; /* brw: cli sent cache remain */ - obd_uid o_uid; - obd_gid o_gid; - obd_flag o_flags; - obd_count o_nlink; /* brw: checksum */ - obd_count o_generation; - obd_count o_misc; /* brw: o_dropped */ - __u32 o_easize; /* epoch in ost writes */ - __u32 o_mds; - __u32 o_stripe_idx; /* holds stripe idx */ - __u32 o_padding_1; - char o_inline[OBD_INLINESZ]; - /* lustre_handle + llog_cookie */ -}; - -#define o_dirty o_blocks -#define o_undirty o_mode -#define o_dropped o_misc -#define o_cksum o_nlink - -extern void lustre_swab_obdo (struct obdo *o); - - #define LOV_MAGIC_V1 0x0BD10BD0 #define LOV_MAGIC LOV_MAGIC_V1 #define LOV_MAGIC_JOIN 0x0BD20BD0 +#define LOV_MAGIC_V3 0x0BD30BD0 #define LOV_PATTERN_RAID0 0x001 /* stripes are used round-robin */ #define LOV_PATTERN_RAID1 0x002 /* stripes are mirrors of each other */ @@ -820,7 +829,7 @@ struct lov_mds_md_v1 { /* LOV EA mds/wire data (little-endian) */ struct lov_ost_data_v1 lmm_objects[0]; /* per-stripe data */ }; -extern void lustre_swab_lov_mds_md(struct lov_mds_md *llm); +/* extern void lustre_swab_lov_mds_md(struct lov_mds_md *llm); */ #define MAX_MD_SIZE (sizeof(struct lov_mds_md) + 4 * sizeof(struct lov_ost_data)) #define MIN_MD_SIZE (sizeof(struct lov_mds_md) + 1 * sizeof(struct lov_ost_data)) @@ -829,6 +838,18 @@ extern void lustre_swab_lov_mds_md(struct lov_mds_md *llm); #define XATTR_NAME_ACL_DEFAULT "system.posix_acl_default" #define XATTR_NAME_LOV "trusted.lov" +struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */ + __u32 lmm_magic; /* magic number = LOV_MAGIC_V3 */ + __u32 lmm_pattern; /* LOV_PATTERN_RAID0, LOV_PATTERN_RAID1 */ + __u64 lmm_object_id; /* LOV object ID */ + __u64 lmm_object_gr; /* LOV object group */ + __u32 lmm_stripe_size; /* size of stripe in bytes */ + __u32 lmm_stripe_count; /* num stripes in use for this object */ + char lmm_pool_name[LOV_MAXPOOLNAME]; /* must be 32bit aligned */ + struct lov_ost_data_v1 lmm_objects[0]; /* per-stripe data */ +}; + + #define OBD_MD_FLID (0x00000001ULL) /* object ID */ #define OBD_MD_FLATIME (0x00000002ULL) /* access time */ #define OBD_MD_FLMTIME (0x00000004ULL) /* data modification time */ @@ -843,11 +864,11 @@ extern void lustre_swab_lov_mds_md(struct lov_mds_md *llm); #define OBD_MD_FLFLAGS (0x00000800ULL) /* flags word */ #define OBD_MD_FLNLINK (0x00002000ULL) /* link count */ #define OBD_MD_FLGENER (0x00004000ULL) /* generation number */ -#define OBD_MD_FLINLINE (0x00008000ULL) /* inline data */ +/*#define OBD_MD_FLINLINE (0x00008000ULL) inline data. used until 1.6.5 */ #define OBD_MD_FLRDEV (0x00010000ULL) /* device number */ #define OBD_MD_FLEASIZE (0x00020000ULL) /* extended attribute data */ #define OBD_MD_LINKNAME (0x00040000ULL) /* symbolic link target */ -#define OBD_MD_FLHANDLE (0x00080000ULL) /* file handle */ +#define OBD_MD_FLHANDLE (0x00080000ULL) /* file/lock handle */ #define OBD_MD_FLCKSUM (0x00100000ULL) /* bulk data checksum */ #define OBD_MD_FLQOS (0x00200000ULL) /* quality of service stats */ #define OBD_MD_FLOSCOPQ (0x00400000ULL) /* osc opaque data */ @@ -887,16 +908,6 @@ extern void lustre_swab_lov_mds_md(struct lov_mds_md *llm); OBD_MD_FLGID | OBD_MD_FLFLAGS | OBD_MD_FLNLINK | \ OBD_MD_FLGENER | OBD_MD_FLRDEV | OBD_MD_FLGROUP) -static inline struct lustre_handle *obdo_handle(struct obdo *oa) -{ - return (struct lustre_handle *)oa->o_inline; -} - -static inline struct llog_cookie *obdo_logcookie(struct obdo *oa) -{ - return (struct llog_cookie *)(oa->o_inline + - sizeof(struct lustre_handle)); -} /* don't forget obdo_fid which is way down at the bottom so it can * come after the definition of llog_cookie */ @@ -966,15 +977,6 @@ struct niobuf_remote { extern void lustre_swab_niobuf_remote (struct niobuf_remote *nbr); -/* request structure for OST's */ - -struct ost_body { - struct obdo oa; -}; - -extern void lustre_swab_ost_body (struct ost_body *b); -extern void lustre_swab_ost_last_id(obd_id *id); - /* lock value block communicated between the filter and llite */ /* OST_LVB_ERR_INIT is needed because the return code in rc is @@ -1726,10 +1728,11 @@ typedef enum { LCK_CR = 16, LCK_NL = 32, LCK_GROUP = 64, + LCK_COS = 128, LCK_MAXMODE } ldlm_mode_t; -#define LCK_MODE_NUM 7 +#define LCK_MODE_NUM 8 typedef enum { LDLM_PLAIN = 10, @@ -1890,12 +1893,15 @@ struct cfg_marker { __u32 cm_flags; __u32 cm_vers; /* lustre release version number */ __u32 padding; /* 64 bit align */ - time_t cm_createtime; /*when this record was first created */ - time_t cm_canceltime; /*when this record is no longer valid*/ + __u64 cm_createtime; /*when this record was first created */ + __u64 cm_canceltime; /*when this record is no longer valid*/ char cm_tgtname[MTI_NAME_MAXLEN]; char cm_comment[MTI_NAME_MAXLEN]; }; +extern void lustre_swab_cfg_marker(struct cfg_marker *marker, + int swab, int size); + /* * Opcodes for multiple servers. */ @@ -1910,14 +1916,14 @@ typedef enum { /* catalog of log objects */ -/* Identifier for a single log object */ +/** Identifier for a single log object */ struct llog_logid { __u64 lgl_oid; __u64 lgl_ogr; __u32 lgl_ogen; } __attribute__((packed)); -/* Records written to the CATALOGS list */ +/** Records written to the CATALOGS list */ #define CATLIST "CATALOGS" struct llog_catid { struct llog_logid lci_logid; @@ -1926,7 +1932,7 @@ struct llog_catid { __u32 lci_padding3; } __attribute__((packed)); -/*join file lov mds md*/ +/** join file lov mds md*/ struct lov_mds_md_join { struct lov_mds_md lmmj_md; /*join private info*/ @@ -1963,7 +1969,7 @@ typedef enum { __swab32(LLOG_OP_MAGIC) || \ (((r)->lrh_type == 0) && ((r)->lrh_len > LLOG_CHUNK_SIZE))) -/* Log record header - stored in little endian order. +/** Log record header - stored in little endian order. * Each record must start with this struct, end with a llog_rec_tail, * and be a multiple of 256 bits in size. */ @@ -1990,7 +1996,7 @@ struct llog_logid_rec { struct llog_rec_tail lid_tail; } __attribute__((packed)); -/* MDS extent description +/** MDS extent description * It is for joined file extent info, each extent info for joined file * just like (start, end, lmm). */ @@ -1999,7 +2005,8 @@ struct mds_extent_desc { __u64 med_len; /* extent length */ struct lov_mds_md med_lmm; /* extent's lmm */ }; -/*Joined file array extent log record*/ + +/** Joined file array extent log record*/ struct llog_array_rec { struct llog_rec_hdr lmr_hdr; struct mds_extent_desc lmr_med; @@ -2090,7 +2097,7 @@ struct llog_log_hdr { llh->llh_bitmap_offset - \ sizeof(llh->llh_tail)) * 8) -/* log cookies are used to reference a specific log file and a record therein */ +/** log cookies are used to reference a specific log file and a record therein */ struct llog_cookie { struct llog_logid lgc_lgl; __u32 lgc_subsys; @@ -2098,7 +2105,7 @@ struct llog_cookie { __u32 lgc_padding; } __attribute__((packed)); -/* llog protocol */ +/** llog protocol */ enum llogd_rpc_ops { LLOG_ORIGIN_HANDLE_CREATE = 501, LLOG_ORIGIN_HANDLE_NEXT_BLOCK = 502, @@ -2152,8 +2159,70 @@ struct lov_user_md_join { /* LOV EA user data (host-endian) */ struct lov_user_ost_data_join lmm_objects[0]; /* per-stripe data */ } __attribute__((packed)); -extern void lustre_swab_lov_user_md(struct lov_user_md *lum); -extern void lustre_swab_lov_user_md_objects(struct lov_user_md *lum); +/* Note: 64-bit types are 64-bit aligned in structure */ +struct obdo { + obd_valid o_valid; /* hot fields in this obdo */ + obd_id o_id; + obd_gr o_gr; + obd_id o_fid; + obd_size o_size; /* o_size-o_blocks == ost_lvb */ + obd_time o_mtime; + obd_time o_atime; + obd_time o_ctime; + obd_blocks o_blocks; /* brw: cli sent cached bytes */ + obd_size o_grant; + + /* 32-bit fields start here: keep an even number of them via padding */ + obd_blksize o_blksize; /* optimal IO blocksize */ + obd_mode o_mode; /* brw: cli sent cache remain */ + obd_uid o_uid; + obd_gid o_gid; + obd_flag o_flags; + obd_count o_nlink; /* brw: checksum */ + obd_count o_generation; + obd_count o_misc; /* brw: o_dropped */ + __u32 o_easize; /* epoch in ost writes */ + __u32 o_mds; + __u32 o_stripe_idx; /* holds stripe idx */ + __u32 o_padding_1; + struct lustre_handle o_handle; /* brw: lock handle to prolong locks */ + struct llog_cookie o_lcookie; /* destroy: unlink cookie from MDS */ + + __u64 o_padding_2; + __u64 o_padding_3; + __u64 o_padding_4; + __u64 o_padding_5; + __u64 o_padding_6; +}; + +#define o_dirty o_blocks +#define o_undirty o_mode +#define o_dropped o_misc +#define o_cksum o_nlink + +extern void lustre_swab_obdo (struct obdo *o); + +/* request structure for OST's */ + +struct ost_body { + struct obdo oa; +}; + +/* Key for FIEMAP to be used in get_info calls */ +struct ll_fiemap_info_key { + char name[8]; + struct obdo oa; + struct ll_user_fiemap fiemap; +}; + +extern void lustre_swab_ost_body (struct ost_body *b); +extern void lustre_swab_ost_last_id(obd_id *id); +extern void lustre_swab_fiemap(struct ll_user_fiemap *fiemap); + +extern void lustre_swab_lov_user_md_v1(struct lov_user_md_v1 *lum); +extern void lustre_swab_lov_user_md_v3(struct lov_user_md_v3 *lum); +extern void lustre_swab_lov_user_md_objects(struct lov_user_ost_data *lod, + int stripe_count); extern void lustre_swab_lov_user_md_join(struct lov_user_md_join *lumj); /* llog_swab.c */ @@ -2196,7 +2265,7 @@ typedef enum { #define JOIN_FILE_ALIGN 4096 -/* security opcodes */ +/** security opcodes */ typedef enum { SEC_CTX_INIT = 801, SEC_CTX_INIT_CONT = 802, @@ -2227,18 +2296,18 @@ struct lustre_capa { extern void lustre_swab_lustre_capa(struct lustre_capa *c); -/* lustre_capa.lc_opc */ +/** lustre_capa::lc_opc */ enum { - CAPA_OPC_BODY_WRITE = 1<<0, /* write object data */ - CAPA_OPC_BODY_READ = 1<<1, /* read object data */ - CAPA_OPC_INDEX_LOOKUP = 1<<2, /* lookup object fid */ - CAPA_OPC_INDEX_INSERT = 1<<3, /* insert object fid */ - CAPA_OPC_INDEX_DELETE = 1<<4, /* delete object fid */ - CAPA_OPC_OSS_WRITE = 1<<5, /* write oss object data */ - CAPA_OPC_OSS_READ = 1<<6, /* read oss object data */ - CAPA_OPC_OSS_TRUNC = 1<<7, /* truncate oss object */ - CAPA_OPC_META_WRITE = 1<<8, /* write object meta data */ - CAPA_OPC_META_READ = 1<<9, /* read object meta data */ + CAPA_OPC_BODY_WRITE = 1<<0, /**< write object data */ + CAPA_OPC_BODY_READ = 1<<1, /**< read object data */ + CAPA_OPC_INDEX_LOOKUP = 1<<2, /**< lookup object fid */ + CAPA_OPC_INDEX_INSERT = 1<<3, /**< insert object fid */ + CAPA_OPC_INDEX_DELETE = 1<<4, /**< delete object fid */ + CAPA_OPC_OSS_WRITE = 1<<5, /**< write oss object data */ + CAPA_OPC_OSS_READ = 1<<6, /**< read oss object data */ + CAPA_OPC_OSS_TRUNC = 1<<7, /**< truncate oss object */ + CAPA_OPC_META_WRITE = 1<<8, /**< write object meta data */ + CAPA_OPC_META_READ = 1<<9, /**< read object meta data */ }; @@ -2266,9 +2335,9 @@ static inline int capa_for_oss(struct lustre_capa *c) return (c->lc_opc & CAPA_OPC_INDEX_LOOKUP) == 0; } -/* lustre_capa.lc_hmac_alg */ +/* lustre_capa::lc_hmac_alg */ enum { - CAPA_HMAC_ALG_SHA1 = 1, /* sha1 algorithm */ + CAPA_HMAC_ALG_SHA1 = 1, /**< sha1 algorithm */ CAPA_HMAC_ALG_MAX, }; @@ -2276,17 +2345,19 @@ enum { #define CAPA_HMAC_ALG_MASK 0xff000000 struct lustre_capa_key { - __u64 lk_mdsid; /* mds# */ - __u32 lk_keyid; /* key# */ + __u64 lk_mdsid; /**< mds# */ + __u32 lk_keyid; /**< key# */ __u32 lk_padding; - __u8 lk_key[CAPA_HMAC_KEY_MAX_LEN]; /* key */ + __u8 lk_key[CAPA_HMAC_KEY_MAX_LEN]; /**< key */ } __attribute__((packed)); extern void lustre_swab_lustre_capa_key(struct lustre_capa_key *k); /* quota check function */ -#define QUOTA_RET_OK 0 /* return successfully */ -#define QUOTA_RET_NOQUOTA 1 /* not support quota */ -#define QUOTA_RET_NOLIMIT 2 /* quota limit isn't set */ -#define QUOTA_RET_ACQUOTA 3 /* need to acquire extra quota */ +#define QUOTA_RET_OK 0 /**< return successfully */ +#define QUOTA_RET_NOQUOTA 1 /**< not support quota */ +#define QUOTA_RET_NOLIMIT 2 /**< quota limit isn't set */ +#define QUOTA_RET_ACQUOTA 3 /**< need to acquire extra quota */ #endif + +/** @} lustreidl */