X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Finclude%2Flustre%2Flustre_idl.h;h=a9f6ba5bb5510c7558d93acff5c6070ff6fb6511;hb=0517160dd68ac026513ad1b8e3e6f7abd4acfdef;hp=89fa719116e3b95e15f7122b9a23e5c60e618885;hpb=fbfd488a2f87ea43332ae16341887f68c0ffbde5;p=fs%2Flustre-release.git diff --git a/lustre/include/lustre/lustre_idl.h b/lustre/include/lustre/lustre_idl.h index 89fa719..a9f6ba5 100644 --- a/lustre/include/lustre/lustre_idl.h +++ b/lustre/include/lustre/lustre_idl.h @@ -26,10 +26,13 @@ * GPL HEADER END */ /* - * Copyright 2008 Sun Microsystems, Inc. All rights reserved + * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. */ /* + * Copyright (c) 2011 Whamcloud, Inc. + */ +/* * This file is part of Lustre, http://www.lustre.org/ * Lustre is a trademark of Sun Microsystems, Inc. * @@ -91,11 +94,12 @@ #ifndef _LUSTRE_IDL_H_ #define _LUSTRE_IDL_H_ -#include +#if !defined(LASSERT) && !defined(LPU64) +#include /* for LASSERT, LPUX64, etc */ +#endif /* Defn's shared with user-space. */ #include -#include /* * GENERAL STUFF @@ -139,12 +143,7 @@ #define SEQ_DATA_PORTAL 31 #define SEQ_CONTROLLER_PORTAL 32 -#define SVC_KILLED 1 -#define SVC_EVENT 2 -#define SVC_SIGNAL 4 -#define SVC_RUNNING 8 -#define SVC_STOPPING 16 -#define SVC_STOPPED 32 +/* Portal 63 is reserved for the Cray Inc DVS - nic@cray.com, roe@cray.com, n8851@cray.com */ /* packet types */ #define PTL_RPC_MSG_REQUEST 4711 @@ -171,19 +170,35 @@ typedef __u32 mdsno_t; typedef __u64 seqno_t; +typedef __u64 obd_id; +typedef __u64 obd_seq; +typedef __s64 obd_time; +typedef __u64 obd_size; +typedef __u64 obd_off; +typedef __u64 obd_blocks; +typedef __u64 obd_valid; +typedef __u32 obd_blksize; +typedef __u32 obd_mode; +typedef __u32 obd_uid; +typedef __u32 obd_gid; +typedef __u32 obd_flag; +typedef __u32 obd_count; /** * Describes a range of sequence, lsr_start is included but lsr_end is * not in the range. - * Same structure is used in fld module where lsr_mdt field holds mdt id + * Same structure is used in fld module where lsr_index field holds mdt id * of the home mdt. */ +#define LU_SEQ_RANGE_MDT 0x0 +#define LU_SEQ_RANGE_OST 0x1 + struct lu_seq_range { __u64 lsr_start; __u64 lsr_end; - __u32 lsr_mdt; - __u32 lsr_padding; + __u32 lsr_index; + __u32 lsr_flags; }; /** @@ -201,7 +216,7 @@ static inline __u64 range_space(const struct lu_seq_range *range) static inline void range_init(struct lu_seq_range *range) { - range->lsr_start = range->lsr_end = range->lsr_mdt = 0; + range->lsr_start = range->lsr_end = range->lsr_index = 0; } /** @@ -214,18 +229,6 @@ static inline int range_within(const struct lu_seq_range *range, return s >= range->lsr_start && s < range->lsr_end; } -/** - * allocate \a w units of sequence from range \a from. - */ -static inline void range_alloc(struct lu_seq_range *to, - struct lu_seq_range *from, - __u64 width) -{ - to->lsr_start = from->lsr_start; - to->lsr_end = from->lsr_start + width; - from->lsr_start += width; -} - static inline int range_is_sane(const struct lu_seq_range *range) { return (range->lsr_end >= range->lsr_start); @@ -242,12 +245,21 @@ static inline int range_is_exhausted(const struct lu_seq_range *range) return range_space(range) == 0; } -#define DRANGE "[%#16.16"LPF64"x-%#16.16"LPF64"x):%x" +/* return 0 if two range have the same location */ +static inline int range_compare_loc(const struct lu_seq_range *r1, + const struct lu_seq_range *r2) +{ + return r1->lsr_index != r2->lsr_index || + r1->lsr_flags != r2->lsr_flags; +} + +#define DRANGE "[%#16.16"LPF64"x-%#16.16"LPF64"x):%x:%x" #define PRANGE(range) \ (range)->lsr_start, \ (range)->lsr_end, \ - (range)->lsr_mdt + (range)->lsr_index, \ + (range)->lsr_flags /** \defgroup lu_fid lu_fid * @{ */ @@ -279,7 +291,7 @@ struct lustre_mdt_attrs { * lma_self_fid and lma_flags are always available. */ __u32 lma_compat; - /** + /** * Per-file incompat feature list. Lustre version should support all * flags set in this field. The supported feature mask is available in * LMA_INCOMPAT_SUPP. @@ -287,13 +299,70 @@ struct lustre_mdt_attrs { __u32 lma_incompat; /** FID of this inode */ struct lu_fid lma_self_fid; - /** SOM state, mdt/ost type, others */ + /** mdt/ost type, others */ __u64 lma_flags; - /** total sectors in objects */ - __u64 lma_som_sectors; + /* IO Epoch SOM attributes belongs to */ + __u64 lma_ioepoch; + /** total file size in objects */ + __u64 lma_som_size; + /** total fs blocks in objects */ + __u64 lma_som_blocks; + /** mds mount id the size is valid for */ + __u64 lma_som_mountid; }; /** + * Fill \a lma with its first content. + * Only fid is stored. + */ +static inline void lustre_lma_init(struct lustre_mdt_attrs *lma, + const struct lu_fid *fid) +{ + lma->lma_compat = 0; + lma->lma_incompat = 0; + memcpy(&lma->lma_self_fid, fid, sizeof(*fid)); + lma->lma_flags = 0; + lma->lma_ioepoch = 0; + lma->lma_som_size = 0; + lma->lma_som_blocks = 0; + lma->lma_som_mountid = 0; + + /* If a field is added in struct lustre_mdt_attrs, zero it explicitly + * and change the test below. */ + LASSERT(sizeof(*lma) == + (offsetof(struct lustre_mdt_attrs, lma_som_mountid) + + sizeof(lma->lma_som_mountid))); +}; + +extern void lustre_swab_lu_fid(struct lu_fid *fid); + +/** + * Swab, if needed, lustre_mdt_attr struct to on-disk format. + * Otherwise, do not touch it. + */ +static inline void lustre_lma_swab(struct lustre_mdt_attrs *lma) +{ + /* Use LUSTRE_MSG_MAGIC to detect local endianess. */ + if (LUSTRE_MSG_MAGIC != cpu_to_le32(LUSTRE_MSG_MAGIC)) { + __swab32s(&lma->lma_compat); + __swab32s(&lma->lma_incompat); + lustre_swab_lu_fid(&lma->lma_self_fid); + __swab64s(&lma->lma_flags); + __swab64s(&lma->lma_ioepoch); + __swab64s(&lma->lma_som_size); + __swab64s(&lma->lma_som_blocks); + __swab64s(&lma->lma_som_mountid); + } +}; + +/* This is the maximum number of MDTs allowed in CMD testing until such + * a time that FID-on-OST is implemented. This is due to the limitations + * of packing non-0-MDT numbers into the FID SEQ namespace. Once FID-on-OST + * is implemented this limit will be virtually unlimited. */ +#define MAX_MDT_COUNT 8 + + +/** * fid constants */ enum { @@ -324,20 +393,85 @@ static inline void fid_zero(struct lu_fid *fid) memset(fid, 0, sizeof(*fid)); } -/* Normal FID sequence starts from this value, i.e. 1<<33 */ -#define FID_SEQ_START 0x200000000ULL +static inline obd_id fid_ver_oid(const struct lu_fid *fid) +{ + return ((__u64)fid_ver(fid) << 32 | fid_oid(fid)); +} -/* IDIF sequence starts from this value, i.e. 1<<32 */ -#define IDIF_SEQ_START 0x100000000ULL +/** + * Note that reserved SEQ numbers below 12 will conflict with ldiskfs + * inodes in the IGIF namespace, so these reserved SEQ numbers can be + * used for other purposes and not risk collisions with existing inodes. + * + * Different FID Format + * http://arch.lustre.org/index.php?title=Interoperability_fids_zfs#NEW.0 + */ +enum fid_seq { + FID_SEQ_OST_MDT0 = 0, + FID_SEQ_LLOG = 1, + FID_SEQ_ECHO = 2, + FID_SEQ_OST_MDT1 = 3, + FID_SEQ_OST_MAX = 9, /* Max MDT count before OST_on_FID */ + FID_SEQ_RSVD = 11, + FID_SEQ_IGIF = 12, + FID_SEQ_IGIF_MAX = 0x0ffffffffULL, + FID_SEQ_IDIF = 0x100000000ULL, + FID_SEQ_IDIF_MAX = 0x1ffffffffULL, + /* Normal FID sequence starts from this value, i.e. 1<<33 */ + FID_SEQ_START = 0x200000000ULL, + FID_SEQ_LOCAL_FILE = 0x200000001ULL, + FID_SEQ_DOT_LUSTRE = 0x200000002ULL, + FID_SEQ_NORMAL = 0x200000400ULL, + FID_SEQ_LOV_DEFAULT= 0xffffffffffffffffULL +}; + +#define OBIF_OID_MAX_BITS 32 +#define OBIF_MAX_OID (1ULL << OBIF_OID_MAX_BITS) +#define OBIF_OID_MASK ((1ULL << OBIF_OID_MAX_BITS) - 1) +#define IDIF_OID_MAX_BITS 48 +#define IDIF_MAX_OID (1ULL << IDIF_OID_MAX_BITS) +#define IDIF_OID_MASK ((1ULL << IDIF_OID_MAX_BITS) - 1) + + +static inline int fid_seq_is_mdt0(obd_seq seq) +{ + return (seq == FID_SEQ_OST_MDT0); +} + +static inline int fid_seq_is_cmd(const __u64 seq) +{ + return (seq >= FID_SEQ_OST_MDT1 && seq <= FID_SEQ_OST_MAX); +}; + +static inline int fid_seq_is_mdt(const __u64 seq) +{ + return seq == FID_SEQ_OST_MDT0 || + (seq >= FID_SEQ_OST_MDT1 && seq <= FID_SEQ_OST_MAX); +}; + +static inline int fid_seq_is_rsvd(const __u64 seq) +{ + return seq <= FID_SEQ_RSVD; +}; + +static inline int fid_is_mdt0(const struct lu_fid *fid) +{ + return fid_seq_is_mdt0(fid_seq(fid)); +} /** * Check if a fid is igif or not. * \param fid the fid to be tested. * \return true if the fid is a igif; otherwise false. */ +static inline int fid_seq_is_igif(const __u64 seq) +{ + return seq >= FID_SEQ_IGIF && seq <= FID_SEQ_IGIF_MAX; +} + static inline int fid_is_igif(const struct lu_fid *fid) { - return fid_seq(fid) > 0 && fid_seq(fid) < IDIF_SEQ_START; + return fid_seq_is_igif(fid_seq(fid)); } /** @@ -345,9 +479,177 @@ static inline int fid_is_igif(const struct lu_fid *fid) * \param fid the fid to be tested. * \return true if the fid is a idif; otherwise false. */ +static inline int fid_seq_is_idif(const __u64 seq) +{ + return seq >= FID_SEQ_IDIF && seq <= FID_SEQ_IDIF_MAX; +} + static inline int fid_is_idif(const struct lu_fid *fid) { - return fid_seq(fid) >= IDIF_SEQ_START && fid_seq(fid) < FID_SEQ_START; + return fid_seq_is_idif(fid_seq(fid)); +} + +struct ost_id { + obd_id oi_id; + obd_seq oi_seq; +}; + +static inline int fid_seq_is_norm(const __u64 seq) +{ + return (seq >= FID_SEQ_NORMAL); +} + +static inline int fid_is_norm(const struct lu_fid *fid) +{ + return fid_seq_is_norm(fid_seq(fid)); +} + +/* convert an OST objid into an IDIF FID SEQ number */ +static inline obd_seq fid_idif_seq(obd_id id, __u32 ost_idx) +{ + return FID_SEQ_IDIF | (ost_idx << 16) | ((id >> 32) & 0xffff); +} + +/* convert a packed IDIF FID into an OST objid */ +static inline obd_id fid_idif_id(obd_seq seq, __u32 oid, __u32 ver) +{ + return ((__u64)ver << 48) | ((seq & 0xffff) << 32) | oid; +} + +/* unpack an ostid (id/seq) from a wire/disk structure into an IDIF FID */ +static inline void ostid_idif_unpack(struct ost_id *ostid, + struct lu_fid *fid, __u32 ost_idx) +{ + fid->f_seq = fid_idif_seq(ostid->oi_id, ost_idx); + fid->f_oid = ostid->oi_id; /* truncate to 32 bits by assignment */ + fid->f_ver = ostid->oi_id >> 48; /* in theory, not currently used */ +} + +/* unpack an ostid (id/seq) from a wire/disk structure into a non-IDIF FID */ +static inline void ostid_fid_unpack(struct ost_id *ostid, struct lu_fid *fid) +{ + fid->f_seq = ostid->oi_seq; + fid->f_oid = ostid->oi_id; /* truncate to 32 bits by assignment */ + fid->f_ver = ostid->oi_id >> 32; /* in theory, not currently used */ +} + +/* Unpack an OST object id/seq (group) into a FID. This is needed for + * converting all obdo, lmm, lsm, etc. 64-bit id/seq pairs into proper + * FIDs. Note that if an id/seq is already in FID/IDIF format it will + * be passed through unchanged. Only legacy OST objects in "group 0" + * will be mapped into the IDIF namespace so that they can fit into the + * struct lu_fid fields without loss. For reference see: + * http://arch.lustre.org/index.php?title=Interoperability_fids_zfs + */ +static inline int fid_ostid_unpack(struct lu_fid *fid, struct ost_id *ostid, + __u32 ost_idx) +{ + if (ost_idx > 0xffff) { + CERROR("bad ost_idx, seq:"LPU64" id:"LPU64" ost_idx:%u\n", + ostid->oi_seq, ostid->oi_id, ost_idx); + return -EBADF; + } + + if (fid_seq_is_mdt0(ostid->oi_seq)) { + /* This is a "legacy" (old 1.x/2.early) OST object in "group 0" + * that we map into the IDIF namespace. It allows up to 2^48 + * objects per OST, as this is the object namespace that has + * been in production for years. This can handle create rates + * of 1M objects/s/OST for 9 years, or combinations thereof. */ + if (ostid->oi_id >= IDIF_MAX_OID) { + CERROR("bad MDT0 id, seq:"LPU64" id:"LPU64" ost_idx:%u\n", + ostid->oi_seq, ostid->oi_id, ost_idx); + return -EBADF; + } + ostid_idif_unpack(ostid, fid, ost_idx); + + } else if (fid_seq_is_rsvd(ostid->oi_seq)) { + /* These are legacy OST objects for LLOG/ECHO and CMD testing. + * We only support 2^32 objects in these groups, and cannot + * uniquely identify them in the system (i.e. they are the + * duplicated on all OSTs), but this is not strictly required + * for the old object protocol, which has a separate ost_idx. */ + if (ostid->oi_id >= 0xffffffffULL) { + CERROR("bad RSVD id, seq:"LPU64" id:"LPU64" ost_idx:%u\n", + ostid->oi_seq, ostid->oi_id, ost_idx); + return -EBADF; + } + ostid_fid_unpack(ostid, fid); + + } else if (unlikely(fid_seq_is_igif(ostid->oi_seq))) { + /* This is an MDT inode number, which should never collide with + * proper OST object IDs, and is probably a broken filesystem */ + CERROR("bad IGIF, seq:"LPU64" id:"LPU64" ost_idx:%u\n", + ostid->oi_seq, ostid->oi_id, ost_idx); + return -EBADF; + + } else /* if (fid_seq_is_idif(seq) || fid_seq_is_norm(seq)) */ { + /* This is either an IDIF object, which identifies objects across + * all OSTs, or a regular FID. The IDIF namespace maps legacy + * OST objects into the FID namespace. In both cases, we just + * pass the FID through, no conversion needed. */ + ostid_fid_unpack(ostid, fid); + } + + return 0; +} + +/* pack an IDIF FID into an ostid (id/seq) for the wire/disk */ +static inline void ostid_idif_pack(struct lu_fid *fid, struct ost_id *ostid) +{ + ostid->oi_seq = FID_SEQ_OST_MDT0; + ostid->oi_id = fid_idif_id(fid->f_seq, fid->f_oid, fid->f_ver); +} + +/* pack a non-IDIF FID into an ostid (id/seq) for the wire/disk */ +static inline void ostid_fid_pack(struct lu_fid *fid, struct ost_id *ostid) +{ + ostid->oi_seq = fid_seq(fid); + ostid->oi_id = fid_ver_oid(fid); +} + +/* pack any OST FID into an ostid (id/seq) for the wire/disk */ +static inline int fid_ostid_pack(struct lu_fid *fid, struct ost_id *ostid) +{ + if (unlikely(fid_seq_is_igif(fid->f_seq))) { + CERROR("bad IGIF, "DFID"\n", PFID(fid)); + return -EBADF; + } + + if (fid_is_idif(fid)) + ostid_idif_pack(fid, ostid); + else + ostid_fid_pack(fid, ostid); + + return 0; +} + +/* extract OST sequence (group) from a wire ost_id (id/seq) pair */ +static inline obd_seq ostid_seq(struct ost_id *ostid) +{ + if (unlikely(fid_seq_is_igif(ostid->oi_seq))) + CWARN("bad IGIF, oi_seq: "LPU64" oi_id: "LPX64"\n", + ostid->oi_seq, ostid->oi_id); + + if (unlikely(fid_seq_is_idif(ostid->oi_seq))) + return FID_SEQ_OST_MDT0; + + return ostid->oi_seq; +} + +/* extract OST objid from a wire ost_id (id/seq) pair */ +static inline obd_id ostid_id(struct ost_id *ostid) +{ + if (ostid->oi_seq == FID_SEQ_OST_MDT0) + return ostid->oi_id & IDIF_OID_MASK; + + if (fid_seq_is_rsvd(ostid->oi_seq)) + return ostid->oi_id & OBIF_OID_MASK; + + if (fid_seq_is_idif(ostid->oi_seq)) + return fid_idif_id(ostid->oi_seq, ostid->oi_id, 0); + + return ostid->oi_id; } /** @@ -361,6 +663,21 @@ static inline ino_t lu_igif_ino(const struct lu_fid *fid) } /** + * Build igif from the inode number/generation. + */ +#define LU_IGIF_BUILD(fid, ino, gen) \ +do { \ + fid->f_seq = ino; \ + fid->f_oid = gen; \ + fid->f_ver = 0; \ +} while(0) +static inline void lu_igif_build(struct lu_fid *fid, __u32 ino, __u32 gen) +{ + LU_IGIF_BUILD(fid, ino, gen); + LASSERT(fid_is_igif(fid)); +} + +/** * Get inode generation from a igif. * \param fid a igif to get inode generation from. * \return inode generation for the igif. @@ -479,8 +796,9 @@ static inline int lu_fid_cmp(const struct lu_fid *f0, * enumeration. */ enum lu_dirent_attrs { - LUDA_FID = 0x0001, - LUDA_TYPE = 0x0002, + LUDA_FID = 0x0001, + LUDA_TYPE = 0x0002, + LUDA_64BITHASH = 0x0004, }; /** @@ -543,7 +861,14 @@ struct lu_dirpage { }; enum lu_dirpage_flags { - LDF_EMPTY = 1 << 0 + /** + * dirpage contains no entry. + */ + LDF_EMPTY = 1 << 0, + /** + * last entry's lde_hash equals ldp_hash_end. + */ + LDF_COLLIDE = 1 << 1 }; static inline struct lu_dirent *lu_dirent_start(struct lu_dirpage *dp) @@ -589,7 +914,22 @@ static inline int lu_dirent_size(struct lu_dirent *ent) return le16_to_cpu(ent->lde_reclen); } -#define DIR_END_OFF 0xfffffffffffffffeULL +#define MDS_DIR_END_OFF 0xfffffffffffffffeULL + +/** + * MDS_READPAGE page size + * + * This is the directory page size packed in MDS_READPAGE RPC. + * It's different than CFS_PAGE_SIZE because the client needs to + * access the struct lu_dirpage header packed at the beginning of + * the "page" and without this there isn't any way to know find the + * lu_dirpage header is if client and server CFS_PAGE_SIZE differ. + */ +#define LU_PAGE_SHIFT 12 +#define LU_PAGE_SIZE (1UL << LU_PAGE_SHIFT) +#define LU_PAGE_MASK (~(LU_PAGE_SIZE - 1)) + +#define LU_PAGE_COUNT 1 << (CFS_PAGE_SHIFT - LU_PAGE_SHIFT) /** @} lu_dir */ @@ -603,8 +943,8 @@ static inline int lustre_handle_is_used(struct lustre_handle *lh) return lh->cookie != 0ull; } -static inline int lustre_handle_equal(struct lustre_handle *lh1, - struct lustre_handle *lh2) +static inline int lustre_handle_equal(const struct lustre_handle *lh1, + const struct lustre_handle *lh2) { return lh1->cookie == lh2->cookie; } @@ -617,6 +957,7 @@ static inline void lustre_handle_copy(struct lustre_handle *tgt, /* flags for lm_flags */ #define MSGHDR_AT_SUPPORT 0x1 +#define MSGHDR_CKSUM_INCOMPAT18 0x2 #define lustre_msg lustre_msg_v2 /* we depend on this structure to be 8-byte aligned */ @@ -680,6 +1021,9 @@ 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 @@ -726,7 +1070,11 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); #define OBD_CONNECT_TRUNCLOCK 0x400ULL /*locks on server for punch */ #define OBD_CONNECT_TRANSNO 0x800ULL /*replay sends init transno */ #define OBD_CONNECT_IBITS 0x1000ULL /*support for inodebits locks*/ -#define OBD_CONNECT_JOIN 0x2000ULL /*files can be concatenated */ +#define OBD_CONNECT_JOIN 0x2000ULL /*files can be concatenated. + *We do not support JOIN FILE + *anymore, reserve this flags + *just for preventing such bit + *to be reused.*/ #define OBD_CONNECT_ATTRFID 0x4000ULL /*Server can GetAttr By Fid*/ #define OBD_CONNECT_NODEVOH 0x8000ULL /*No open hndl on specl nodes*/ #define OBD_CONNECT_RMT_CLIENT 0x10000ULL /*Remote client */ @@ -748,6 +1096,12 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); #define OBD_CONNECT_LOV_V3 0x100000000ULL /*client supports LOV v3 EA */ #define OBD_CONNECT_GRANT_SHRINK 0x200000000ULL /* support grant shrink */ #define OBD_CONNECT_SKIP_ORPHAN 0x400000000ULL /* don't reuse orphan objids */ +#define OBD_CONNECT_MAX_EASIZE 0x800000000ULL /* preserved for large EA */ +#define OBD_CONNECT_FULL20 0x1000000000ULL /* it is 2.0 client */ +#define OBD_CONNECT_LAYOUTLOCK 0x2000000000ULL /* client supports layout lock */ +#define OBD_CONNECT_64BITHASH 0x4000000000ULL /* client supports 64-bits + * directory hash */ +#define OBD_CONNECT_MAXBYTES 0x8000000000ULL /* max stripe size */ /* also update obd_connect_names[] for lprocfs_rd_connect_flags() * and lustre/utils/wirecheck.c */ @@ -759,15 +1113,17 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); #define MDT_CONNECT_SUPPORTED (OBD_CONNECT_RDONLY | OBD_CONNECT_VERSION | \ OBD_CONNECT_ACL | OBD_CONNECT_XATTR | \ - OBD_CONNECT_IBITS | OBD_CONNECT_JOIN | \ - OBD_CONNECT_NODEVOH |/* OBD_CONNECT_ATTRFID |*/\ + OBD_CONNECT_IBITS | \ + OBD_CONNECT_NODEVOH | OBD_CONNECT_ATTRFID | \ OBD_CONNECT_CANCELSET | OBD_CONNECT_AT | \ OBD_CONNECT_RMT_CLIENT | \ OBD_CONNECT_RMT_CLIENT_FORCE | \ - OBD_CONNECT_MDS_CAPA | OBD_CONNECT_OSS_CAPA | \ - OBD_CONNECT_MDS_MDS | OBD_CONNECT_FID | \ - LRU_RESIZE_CONNECT_FLAG | OBD_CONNECT_VBR | \ - OBD_CONNECT_LOV_V3 | OBD_CONNECT_SOM) + OBD_CONNECT_BRW_SIZE | OBD_CONNECT_MDS_CAPA | \ + OBD_CONNECT_OSS_CAPA | OBD_CONNECT_MDS_MDS | \ + OBD_CONNECT_FID | LRU_RESIZE_CONNECT_FLAG | \ + OBD_CONNECT_VBR | OBD_CONNECT_LOV_V3 | \ + OBD_CONNECT_SOM | OBD_CONNECT_FULL20 | \ + OBD_CONNECT_64BITHASH) #define OST_CONNECT_SUPPORTED (OBD_CONNECT_SRVLOCK | OBD_CONNECT_GRANT | \ OBD_CONNECT_REQPORTAL | OBD_CONNECT_VERSION | \ OBD_CONNECT_TRUNCLOCK | OBD_CONNECT_INDEX | \ @@ -778,9 +1134,15 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); OBD_CONNECT_OSS_CAPA | OBD_CONNECT_RMT_CLIENT | \ OBD_CONNECT_RMT_CLIENT_FORCE | OBD_CONNECT_VBR | \ OBD_CONNECT_MDS | OBD_CONNECT_SKIP_ORPHAN | \ - OBD_CONNECT_GRANT_SHRINK) + OBD_CONNECT_GRANT_SHRINK | OBD_CONNECT_FULL20 | \ + OBD_CONNECT_64BITHASH | OBD_CONNECT_MAXBYTES) #define ECHO_CONNECT_SUPPORTED (0) -#define MGS_CONNECT_SUPPORTED (OBD_CONNECT_VERSION | OBD_CONNECT_AT) +#define MGS_CONNECT_SUPPORTED (OBD_CONNECT_VERSION | OBD_CONNECT_AT | \ + OBD_CONNECT_FULL20) + +/* Features required for this version of the client to work with server */ +#define CLIENT_CONNECT_MDT_REQD (OBD_CONNECT_IBITS | OBD_CONNECT_FID | \ + OBD_CONNECT_FULL20) #define OBD_OCD_VERSION(major,minor,patch,fix) (((major)<<24) + ((minor)<<16) +\ ((patch)<<8) + (fix)) @@ -793,6 +1155,23 @@ extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb); * * If we eventually have separate connect data for different types, which we * almost certainly will, then perhaps we stick a union in here. */ +struct obd_connect_data_v1 { + __u64 ocd_connect_flags; /* OBD_CONNECT_* per above */ + __u32 ocd_version; /* lustre release version number */ + __u32 ocd_grant; /* initial cache grant amount (bytes) */ + __u32 ocd_index; /* LOV index to connect to */ + __u32 ocd_brw_size; /* Maximum BRW size in bytes */ + __u64 ocd_ibits_known; /* inode bits this client understands */ + __u32 ocd_nllu; /* non-local-lustre-user */ + __u32 ocd_nllg; /* non-local-lustre-group */ + __u64 ocd_transno; /* first transno from client to be replayed */ + __u32 ocd_group; /* MDS group on OST */ + __u32 ocd_cksum_types; /* supported checksum algorithms */ + __u32 ocd_max_easize; /* How big LOV EA can be on MDS */ + __u32 padding; /* also fix lustre_swab_connect */ + __u64 ocd_maxbytes; /* Maximum stripe size in bytes */ +}; + struct obd_connect_data { __u64 ocd_connect_flags; /* OBD_CONNECT_* per above */ __u32 ocd_version; /* lustre release version number */ @@ -805,10 +1184,31 @@ struct obd_connect_data { __u64 ocd_transno; /* first transno from client to be replayed */ __u32 ocd_group; /* MDS group on OST */ __u32 ocd_cksum_types; /* supported checksum algorithms */ - __u64 padding1; /* also fix lustre_swab_connect */ - __u64 padding2; /* also fix lustre_swab_connect */ + __u32 ocd_max_easize; /* How big LOV EA can be on MDS */ + __u32 padding; /* also fix lustre_swab_connect */ + __u64 ocd_maxbytes; /* Maximum stripe size in bytes */ + /* Fields after ocd_maxbytes are only accessible by the receiver + * if the corresponding flag in ocd_connect_flags is set. Accessing + * any field after ocd_maxbytes on the receiver without a valid flag + * may result in out-of-bound memory access and kernel oops. */ + __u64 padding1; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding2; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding3; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding4; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding5; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding6; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding7; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding8; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 padding9; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingA; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingB; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingC; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingD; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingE; /* added 2.2.0. also fix lustre_swab_connect */ + __u64 paddingF; /* added 2.2.0. also fix lustre_swab_connect */ }; + extern void lustre_swab_connect(struct obd_connect_data *ocd); /* @@ -820,6 +1220,7 @@ extern void lustre_swab_connect(struct obd_connect_data *ocd); typedef enum { OBD_CKSUM_CRC32 = 0x00000001, OBD_CKSUM_ADLER = 0x00000002, + OBD_CKSUM_CRC32C= 0x00000004, } cksum_type_t; /* @@ -851,20 +1252,6 @@ typedef enum { } ost_cmd_t; #define OST_FIRST_OPC OST_REPLY -typedef __u64 obd_id; -typedef __u64 obd_gr; -typedef __u64 obd_time; -typedef __u64 obd_size; -typedef __u64 obd_off; -typedef __u64 obd_blocks; -typedef __u64 obd_valid; -typedef __u32 obd_blksize; -typedef __u32 obd_mode; -typedef __u32 obd_uid; -typedef __u32 obd_gid; -typedef __u32 obd_flag; -typedef __u32 obd_count; - enum obdo_flags { OBD_FL_INLINEDATA = 0x00000001, OBD_FL_OBDMDEXISTS = 0x00000002, @@ -876,25 +1263,29 @@ enum obdo_flags { OBD_FL_NO_USRQUOTA = 0x00000100, /* the object's owner is over quota */ OBD_FL_NO_GRPQUOTA = 0x00000200, /* the object's group is over quota */ OBD_FL_CREATE_CROW = 0x00000400, /* object should be create on write */ - OBD_FL_TRUNCLOCK = 0x00000800, /* delegate DLM locking during punch*/ + OBD_FL_SRVLOCK = 0x00000800, /* delegate DLM locking to server */ OBD_FL_CKSUM_CRC32 = 0x00001000, /* CRC32 checksum type */ OBD_FL_CKSUM_ADLER = 0x00002000, /* ADLER checksum type */ - OBD_FL_CKSUM_RSVD1 = 0x00004000, /* for future cksum types */ + OBD_FL_CKSUM_CRC32C = 0x00004000, /* CRC32C checksum type */ OBD_FL_CKSUM_RSVD2 = 0x00008000, /* for future cksum types */ OBD_FL_CKSUM_RSVD3 = 0x00010000, /* for future cksum types */ OBD_FL_SHRINK_GRANT = 0x00020000, /* object shrink the grant */ + OBD_FL_MMAP = 0x00040000, /* object is mmapped on the client */ + OBD_FL_RECOV_RESEND = 0x00080000, /* recoverable resent */ + OBD_FL_NOSPC_BLK = 0x00100000, /* no more block space on OST */ - OBD_FL_CKSUM_ALL = OBD_FL_CKSUM_CRC32 | OBD_FL_CKSUM_ADLER, + /* Note that while these checksum values are currently separate bits, + * in 2.x we can actually allow all values from 1-31 if we wanted. */ + OBD_FL_CKSUM_ALL = OBD_FL_CKSUM_CRC32 | OBD_FL_CKSUM_ADLER | + OBD_FL_CKSUM_CRC32C, /* mask for local-only flag, which won't be sent over network */ OBD_FL_LOCAL_MASK = 0xF0000000, - /* temporary OBDO used by osc_brw_async (see bug 18364) */ - OBD_FL_TEMPORARY = 0x10000000, }; #define LOV_MAGIC_V1 0x0BD10BD0 #define LOV_MAGIC LOV_MAGIC_V1 -#define LOV_MAGIC_JOIN 0x0BD20BD0 +#define LOV_MAGIC_JOIN_V1 0x0BD20BD0 #define LOV_MAGIC_V3 0x0BD30BD0 #define LOV_PATTERN_RAID0 0x001 /* stripes are used round-robin */ @@ -902,13 +1293,10 @@ enum obdo_flags { #define LOV_PATTERN_FIRST 0x100 /* first stripe is not in round-robin */ #define LOV_PATTERN_CMOBD 0x200 -#define LOV_OBJECT_GROUP_DEFAULT ~0ULL -#define LOV_OBJECT_GROUP_CLEAR 0ULL - #define lov_ost_data lov_ost_data_v1 struct lov_ost_data_v1 { /* per-stripe data structure (little-endian)*/ __u64 l_object_id; /* OST object ID */ - __u64 l_object_gr; /* OST object group (creating MDS number) */ + __u64 l_object_seq; /* OST object seq number */ __u32 l_ost_gen; /* generation of this l_ost_idx */ __u32 l_ost_idx; /* OST index in LOV (lov_tgt_desc->tgts) */ }; @@ -918,7 +1306,7 @@ struct lov_mds_md_v1 { /* LOV EA mds/wire data (little-endian) */ __u32 lmm_magic; /* magic number = LOV_MAGIC_V1 */ __u32 lmm_pattern; /* LOV_PATTERN_RAID0, LOV_PATTERN_RAID1 */ __u64 lmm_object_id; /* LOV object ID */ - __u64 lmm_object_gr; /* LOV object group */ + __u64 lmm_object_seq; /* LOV object seq number */ __u32 lmm_stripe_size; /* size of stripe in bytes */ __u32 lmm_stripe_count; /* num stripes in use for this object */ struct lov_ost_data_v1 lmm_objects[0]; /* per-stripe data */ @@ -946,7 +1334,7 @@ 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 */ + __u64 lmm_object_seq; /* LOV object seq number */ __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 */ @@ -979,7 +1367,7 @@ struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */ #define OBD_MD_FLCOOKIE (0x00800000ULL) /* log cancellation cookie */ #define OBD_MD_FLGROUP (0x01000000ULL) /* group */ #define OBD_MD_FLFID (0x02000000ULL) /* ->ost write inline fid */ -#define OBD_MD_FLEPOCH (0x04000000ULL) /* ->ost write easize is epoch */ +#define OBD_MD_FLEPOCH (0x04000000ULL) /* ->ost write with ioepoch */ /* ->mds if epoch opens or closes */ #define OBD_MD_FLGRANT (0x08000000ULL) /* ost preallocation space grant */ #define OBD_MD_FLDIREA (0x10000000ULL) /* dir's extended attribute data */ @@ -990,6 +1378,7 @@ struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */ #define OBD_MD_MDS (0x0000000100000000ULL) /* where an inode lives on */ #define OBD_MD_REINT (0x0000000200000000ULL) /* reintegrate oa */ #define OBD_MD_MEA (0x0000000400000000ULL) /* CMD split EA */ +#define OBD_MD_MDTIDX (0x0000000800000000ULL) /* Get MDT index */ #define OBD_MD_FLXATTR (0x0000001000000000ULL) /* xattr */ #define OBD_MD_FLXATTRLS (0x0000002000000000ULL) /* xattr list */ @@ -1000,7 +1389,8 @@ struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */ #define OBD_MD_FLOSSCAPA (0x0000040000000000ULL) /* OSS capability */ #define OBD_MD_FLCKSPLIT (0x0000080000000000ULL) /* Check split on server */ #define OBD_MD_FLCROSSREF (0x0000100000000000ULL) /* Cross-ref case */ - +#define OBD_MD_FLGETATTRLOCK (0x0000200000000000ULL) /* Get IOEpoch attributes + * under lock */ #define OBD_FL_TRUNC (0x0000200000000000ULL) /* for filter_truncate */ #define OBD_MD_FLRMTLSETFACL (0x0001000000000000ULL) /* lfs lsetfacl case */ @@ -1017,36 +1407,6 @@ struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */ /* don't forget obdo_fid which is way down at the bottom so it can * come after the definition of llog_cookie */ -enum obd_statfs_state { - OS_STATE_DEGRADED = 0x00000001, /**< RAID degraded/rebuilding */ - OS_STATE_READONLY = 0x00000002, /**< filesystem is read-only */ - OS_STATE_RDONLY_1 = 0x00000004, /**< obsolete 1.6, was EROFS=30 */ - OS_STATE_RDONLY_2 = 0x00000008, /**< obsolete 1.6, was EROFS=30 */ - OS_STATE_RDONLY_3 = 0x00000010, /**< obsolete 1.6, was EROFS=30 */ -}; - -struct obd_statfs { - __u64 os_type; - __u64 os_blocks; - __u64 os_bfree; - __u64 os_bavail; - __u64 os_files; - __u64 os_ffree; - __u8 os_fsid[40]; - __u32 os_bsize; - __u32 os_namelen; - __u64 os_maxbytes; - __u32 os_state; /**< obd_statfs_state OS_STATE_* flag */ - __u32 os_spare1; - __u32 os_spare2; - __u32 os_spare3; - __u32 os_spare4; - __u32 os_spare5; - __u32 os_spare6; - __u32 os_spare7; - __u32 os_spare8; - __u32 os_spare9; -}; extern void lustre_swab_obd_statfs (struct obd_statfs *os); #define OBD_STATFS_NODELAY 0x0001 /* requests should be send without delay @@ -1074,6 +1434,7 @@ extern void lustre_swab_obd_statfs (struct obd_statfs *os); #define OBD_BRW_NOCACHE 0x80 /* this page is a part of non-cached IO */ #define OBD_BRW_NOQUOTA 0x100 #define OBD_BRW_SRVLOCK 0x200 /* Client holds no lock over this page */ +#define OBD_BRW_ASYNC 0x400 /* Server may delay commit to disk */ #define OBD_BRW_MEMALLOC 0x800 /* Client runs in the "kswapd" context */ #define OBD_OBJECT_EOF 0xffffffffffffffffULL @@ -1083,7 +1444,7 @@ extern void lustre_swab_obd_statfs (struct obd_statfs *os); struct obd_ioobj { obd_id ioo_id; - obd_gr ioo_gr; + obd_seq ioo_seq; __u32 ioo_type; __u32 ioo_bufcnt; }; @@ -1112,11 +1473,11 @@ extern void lustre_swab_niobuf_remote (struct niobuf_remote *nbr); #define OST_LVB_GET_ERR(blocks) (int)(blocks - OST_LVB_ERR_INIT) struct ost_lvb { - __u64 lvb_size; - __u64 lvb_mtime; - __u64 lvb_atime; - __u64 lvb_ctime; - __u64 lvb_blocks; + __u64 lvb_size; + obd_time lvb_mtime; + obd_time lvb_atime; + obd_time lvb_ctime; + __u64 lvb_blocks; }; extern void lustre_swab_ost_lvb(struct ost_lvb *); @@ -1194,14 +1555,6 @@ extern void lustre_swab_generic_32s (__u32 *val); /* This FULL lock is useful to take on unlink sort of operations */ #define MDS_INODELOCK_FULL ((1<<(MDS_INODELOCK_MAXSHIFT+1))-1) -struct ll_fid { - __u64 id; /* holds object id */ - __u32 generation; /* holds object generation */ - - __u32 f_type; /* holds object type or stripe idx when passing it to - * OST for saving into EA. */ -}; - extern void lustre_swab_ll_fid (struct ll_fid *fid); #define MDS_STATUS_CONN 1 @@ -1225,13 +1578,15 @@ enum md_op_flags { MF_MDC_CANCEL_FID2 = (1 << 4), MF_MDC_CANCEL_FID3 = (1 << 5), MF_MDC_CANCEL_FID4 = (1 << 6), + /* There is a pending attribute update. */ + MF_SOM_AU = (1 << 7), + /* Cancel OST locks while getattr OST attributes. */ + MF_GETATTR_LOCK = (1 << 8), }; -#define MF_SOM_LOCAL_FLAGS (MF_MDC_CANCEL_FID1 | MF_MDC_CANCEL_FID2 | \ - MF_MDC_CANCEL_FID3 | MF_MDC_CANCEL_FID4) +#define MF_SOM_LOCAL_FLAGS (MF_SOM_CHANGE | MF_EPOCH_OPEN | MF_EPOCH_CLOSE) #define MDS_BFLAG_UNCOMMITTED_WRITES 0x1 -#define MDS_BFLAG_EXT_FLAGS 0x80000000 /* == EXT3_RESERVED_FL */ /* these should be identical to their EXT3_*_FL counterparts, and are * redefined here only to avoid dragging in ext3_fs.h */ @@ -1242,29 +1597,26 @@ enum md_op_flags { #define MDS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (dir only) */ #ifdef __KERNEL__ -/* If MDS_BFLAG_IOC_FLAGS is set it means we requested EXT3_*_FL inode flags - * and we need to decode these into local S_* flags in the inode. Otherwise - * we pass flags straight through (see bug 9486). */ +/* Convert wire MDS_*_FL to corresponding client local VFS S_* values + * for the client inode i_flags. The MDS_*_FL are the Lustre wire + * protocol equivalents of LDISKFS_*_FL values stored on disk, while + * the S_* flags are kernel-internal values that change between kernel + * versions. These flags are set/cleared via FSFILT_IOC_{GET,SET}_FLAGS. + * See b=16526 for a full history. */ static inline int ll_ext_to_inode_flags(int flags) { - return (flags & MDS_BFLAG_EXT_FLAGS) ? - (((flags & MDS_SYNC_FL) ? S_SYNC : 0) | + return (((flags & MDS_SYNC_FL) ? S_SYNC : 0) | ((flags & MDS_NOATIME_FL) ? S_NOATIME : 0) | ((flags & MDS_APPEND_FL) ? S_APPEND : 0) | #if defined(S_DIRSYNC) ((flags & MDS_DIRSYNC_FL) ? S_DIRSYNC : 0) | #endif - ((flags & MDS_IMMUTABLE_FL) ? S_IMMUTABLE : 0)) : - (flags & ~MDS_BFLAG_EXT_FLAGS); + ((flags & MDS_IMMUTABLE_FL) ? S_IMMUTABLE : 0)); } -/* If MDS_BFLAG_EXT_FLAGS is set it means we requested EXT3_*_FL inode flags - * and we pass these straight through. Otherwise we need to convert from - * S_* flags to their EXT3_*_FL equivalents (see bug 9486). */ -static inline int ll_inode_to_ext_flags(int oflags, int iflags) +static inline int ll_inode_to_ext_flags(int iflags) { - return (oflags & MDS_BFLAG_EXT_FLAGS) ? (oflags & ~MDS_BFLAG_EXT_FLAGS): - (((iflags & S_SYNC) ? MDS_SYNC_FL : 0) | + return (((iflags & S_SYNC) ? MDS_SYNC_FL : 0) | ((iflags & S_NOATIME) ? MDS_NOATIME_FL : 0) | ((iflags & S_APPEND) ? MDS_APPEND_FL : 0) | #if defined(S_DIRSYNC) @@ -1290,9 +1642,9 @@ struct mds_body { struct lustre_handle handle; __u64 valid; __u64 size; /* Offset, in the case of MDS_READPAGE */ - __u64 mtime; - __u64 atime; - __u64 ctime; + obd_time mtime; + obd_time atime; + obd_time ctime; __u64 blocks; /* XID, in the case of MDS_READPAGE */ __u64 io_epoch; __u64 ino; @@ -1322,9 +1674,9 @@ struct mdt_body { struct lustre_handle handle; __u64 valid; __u64 size; /* Offset, in the case of MDS_READPAGE */ - __u64 mtime; - __u64 atime; - __u64 ctime; + obd_time mtime; + obd_time atime; + obd_time ctime; __u64 blocks; /* XID, in the case of MDS_READPAGE */ __u64 ioepoch; __u64 ino; /* for 1.6 compatibility */ @@ -1343,8 +1695,9 @@ struct mdt_body { __u32 aclsize; __u32 max_mdsize; __u32 max_cookiesize; - __u32 padding_4; /* also fix lustre_swab_mdt_body */ - __u64 padding_5; + __u32 uid_h; /* high 32-bits of uid, for FUID */ + __u32 gid_h; /* high 32-bits of gid, for FUID */ + __u32 padding_5; /* also fix lustre_swab_mdt_body */ __u64 padding_6; __u64 padding_7; __u64 padding_8; @@ -1354,14 +1707,14 @@ struct mdt_body { extern void lustre_swab_mdt_body (struct mdt_body *b); -struct mdt_epoch { +struct mdt_ioepoch { struct lustre_handle handle; __u64 ioepoch; __u32 flags; __u32 padding; }; -extern void lustre_swab_mdt_epoch (struct mdt_epoch *b); +extern void lustre_swab_mdt_ioepoch (struct mdt_ioepoch *b); #define Q_QUOTACHECK 0x800100 #define Q_INITQUOTA 0x800101 /* init slave limits */ @@ -1414,6 +1767,8 @@ extern void lustre_swab_quota_adjust_qunit(struct quota_adjust_qunit *q); #define LQUOTA_FLAGS_ADJINO 8UL /* adjust the inode qunit size */ #define LQUOTA_FLAGS_CHG_QS 16UL /* indicate whether it has capability of * OBD_CONNECT_CHANGE_QS */ +#define LQUOTA_FLAGS_RECOVERY 32UL /* recovery is going on a uid/gid */ +#define LQUOTA_FLAGS_SETQUOTA 64UL /* being setquota on a uid/gid */ /* flags is specific for quota_adjust_qunit */ #define LQUOTA_QAQ_CREATE_LQS (1 << 31) /* when it is set, need create lqs */ @@ -1456,33 +1811,14 @@ struct mdt_remote_perm { __u32 rp_uid; __u32 rp_gid; __u32 rp_fsuid; + __u32 rp_fsuid_h; __u32 rp_fsgid; + __u32 rp_fsgid_h; __u32 rp_access_perm; /* MAY_READ/WRITE/EXEC */ }; extern void lustre_swab_mdt_remote_perm(struct mdt_remote_perm *p); -struct mds_rec_setattr { - __u32 sa_opcode; - __u32 sa_fsuid; - __u32 sa_fsgid; - __u32 sa_cap; - __u32 sa_suppgid; - __u32 sa_mode; - struct ll_fid sa_fid; - __u64 sa_valid; /* MDS_ATTR_* attributes */ - __u64 sa_size; - __u64 sa_mtime; - __u64 sa_atime; - __u64 sa_ctime; - __u32 sa_uid; - __u32 sa_gid; - __u32 sa_attr_flags; - __u32 sa_padding; /* also fix lustre_swab_mds_rec_setattr */ -}; - -extern void lustre_swab_mds_rec_setattr (struct mds_rec_setattr *sa); - struct mdt_rec_setattr { __u32 sa_opcode; __u32 sa_cap; @@ -1500,9 +1836,9 @@ struct mdt_rec_setattr { __u32 sa_gid; __u64 sa_size; __u64 sa_blocks; - __u64 sa_mtime; - __u64 sa_atime; - __u64 sa_ctime; + obd_time sa_mtime; + obd_time sa_atime; + obd_time sa_ctime; __u32 sa_attr_flags; __u32 sa_mode; __u32 sa_padding_2; @@ -1514,7 +1850,7 @@ struct mdt_rec_setattr { extern void lustre_swab_mdt_rec_setattr (struct mdt_rec_setattr *sa); /* - * Attribute flags used in mds_rec_setattr::sa_valid. + * Attribute flags used in mdt_rec_setattr::sa_valid. * The kernel's #defines for ATTR_* should not be used over the network * since the client and MDS may run different kernels (see bug 13828) * Therefore, we should only use MDS_ATTR_* attributes for sa_valid. @@ -1541,8 +1877,11 @@ extern void lustre_swab_mdt_rec_setattr (struct mdt_rec_setattr *sa); #define FMODE_WRITE 00000002 #endif +/* IO Epoch is opened on a closed file. */ #define FMODE_EPOCH 01000000 -#define FMODE_EPOCHLCK 02000000 +/* IO Epoch is opened on a file truncate. */ +#define FMODE_TRUNC 02000000 +/* Size-on-MDS Attribute Update is pending. */ #define FMODE_SOM 04000000 #define FMODE_CLOSED 0 @@ -1559,7 +1898,11 @@ extern void lustre_swab_mdt_rec_setattr (struct mdt_rec_setattr *sa); #define MDS_OPEN_DELAY_CREATE 0100000000 /* delay initial object create */ #define MDS_OPEN_OWNEROVERRIDE 0200000000 /* NFSD rw-reopen ro file for owner */ -#define MDS_OPEN_JOIN_FILE 0400000000 /* open for join file*/ +#define MDS_OPEN_JOIN_FILE 0400000000 /* open for join file. + * We do not support JOIN FILE + * anymore, reserve this flags + * just for preventing such bit + * to be reused. */ #define MDS_CREATE_RMT_ACL 01000000000 /* indicate create on remote server * with default ACL */ #define MDS_CREATE_SLAVE_OBJ 02000000000 /* indicate create slave object @@ -1568,6 +1911,9 @@ extern void lustre_swab_mdt_rec_setattr (struct mdt_rec_setattr *sa); #define MDS_OPEN_LOCK 04000000000 /* This open requires open lock */ #define MDS_OPEN_HAS_EA 010000000000 /* specify object create pattern */ #define MDS_OPEN_HAS_OBJS 020000000000 /* Just set the EA the obj exist */ +#define MDS_OPEN_NORESTORE 0100000000000ULL /* Do not restore file at open */ +#define MDS_OPEN_NEWSTRIPE 0200000000000ULL /* New stripe needed (restripe or + * hsm restore) */ /* permission for create non-directory file */ #define MAY_CREATE (1 << 7) @@ -1593,30 +1939,11 @@ enum { MDS_PERM_BYPASS = 1 << 3, MDS_SOM = 1 << 4, MDS_QUOTA_IGNORE = 1 << 5, - MDS_CLOSE_CLEANUP = 1 << 6 + MDS_CLOSE_CLEANUP = 1 << 6, + MDS_KEEP_ORPHAN = 1 << 7 }; -struct mds_rec_create { - __u32 cr_opcode; - __u32 cr_fsuid; - __u32 cr_fsgid; - __u32 cr_cap; - __u32 cr_flags; /* for use with open */ - __u32 cr_mode; - struct ll_fid cr_fid; - struct ll_fid cr_replayfid; - __u64 cr_time; - __u64 cr_rdev; - __u32 cr_suppgid; - __u32 cr_padding_1; /* also fix lustre_swab_mds_rec_create */ - __u32 cr_padding_2; /* also fix lustre_swab_mds_rec_create */ - __u32 cr_padding_3; /* also fix lustre_swab_mds_rec_create */ - __u32 cr_padding_4; /* also fix lustre_swab_mds_rec_create */ - __u32 cr_padding_5; /* also fix lustre_swab_mds_rec_create */ -}; - -extern void lustre_swab_mds_rec_create (struct mds_rec_create *cr); - +/* instance of mdt_reint_rec */ struct mdt_rec_create { __u32 cr_opcode; __u32 cr_cap; @@ -1631,52 +1958,33 @@ struct mdt_rec_create { struct lu_fid cr_fid1; struct lu_fid cr_fid2; struct lustre_handle cr_old_handle; /* handle in case of open replay */ - __u64 cr_time; + obd_time cr_time; __u64 cr_rdev; __u64 cr_ioepoch; - __u64 cr_padding_1; /* pad for 64 bits*/ + __u64 cr_padding_1; /* rr_blocks */ __u32 cr_mode; __u32 cr_bias; - __u32 cr_flags; /* for use with open */ - __u32 cr_padding_2; - __u32 cr_padding_3; - __u32 cr_padding_4; -}; - -extern void lustre_swab_mdt_rec_create (struct mdt_rec_create *cr); - -struct mds_rec_join { - struct ll_fid jr_fid; - __u64 jr_headsize; + /* use of helpers set/get_mrc_cr_flags() is needed to access + * 64 bits cr_flags [cr_flags_l, cr_flags_h], this is done to + * extend cr_flags size without breaking 1.8 compat */ + __u32 cr_flags_l; /* for use with open, low 32 bits */ + __u32 cr_flags_h; /* for use with open, high 32 bits */ + __u32 cr_padding_3; /* rr_padding_3 */ + __u32 cr_padding_4; /* rr_padding_4 */ }; -extern void lustre_swab_mds_rec_join (struct mds_rec_join *jr); - -struct mdt_rec_join { - struct lu_fid jr_fid; - __u64 jr_headsize; -}; - -extern void lustre_swab_mdt_rec_join (struct mdt_rec_join *jr); - -struct mds_rec_link { - __u32 lk_opcode; - __u32 lk_fsuid; - __u32 lk_fsgid; - __u32 lk_cap; - __u32 lk_suppgid1; - __u32 lk_suppgid2; - struct ll_fid lk_fid1; - struct ll_fid lk_fid2; - __u64 lk_time; - __u32 lk_padding_1; /* also fix lustre_swab_mds_rec_link */ - __u32 lk_padding_2; /* also fix lustre_swab_mds_rec_link */ - __u32 lk_padding_3; /* also fix lustre_swab_mds_rec_link */ - __u32 lk_padding_4; /* also fix lustre_swab_mds_rec_link */ -}; +static inline void set_mrc_cr_flags(struct mdt_rec_create *mrc, __u64 flags) +{ + mrc->cr_flags_l = (__u32)(flags & 0xFFFFFFFFUll); + mrc->cr_flags_h = (__u32)(flags >> 32); +} -extern void lustre_swab_mds_rec_link (struct mds_rec_link *lk); +static inline __u64 get_mrc_cr_flags(struct mdt_rec_create *mrc) +{ + return ((__u64)(mrc->cr_flags_l) | ((__u64)mrc->cr_flags_h << 32)); +} +/* instance of mdt_reint_rec */ struct mdt_rec_link { __u32 lk_opcode; __u32 lk_cap; @@ -1690,37 +1998,20 @@ struct mdt_rec_link { __u32 lk_suppgid2_h; struct lu_fid lk_fid1; struct lu_fid lk_fid2; - __u64 lk_time; - __u64 lk_padding_1; - __u64 lk_padding_2; - __u64 lk_padding_3; - __u64 lk_padding_4; + obd_time lk_time; + __u64 lk_padding_1; /* rr_atime */ + __u64 lk_padding_2; /* rr_ctime */ + __u64 lk_padding_3; /* rr_size */ + __u64 lk_padding_4; /* rr_blocks */ __u32 lk_bias; - __u32 lk_padding_5; - __u32 lk_padding_6; - __u32 lk_padding_7; - __u32 lk_padding_8; - __u32 lk_padding_9; -}; - -struct mds_rec_unlink { - __u32 ul_opcode; - __u32 ul_fsuid; - __u32 ul_fsgid; - __u32 ul_cap; - __u32 ul_suppgid; - __u32 ul_mode; - struct ll_fid ul_fid1; - struct ll_fid ul_fid2; - __u64 ul_time; - __u32 ul_padding_1; /* also fix lustre_swab_mds_rec_unlink */ - __u32 ul_padding_2; /* also fix lustre_swab_mds_rec_unlink */ - __u32 ul_padding_3; /* also fix lustre_swab_mds_rec_unlink */ - __u32 ul_padding_4; /* also fix lustre_swab_mds_rec_unlink */ + __u32 lk_padding_5; /* rr_mode */ + __u32 lk_padding_6; /* rr_flags */ + __u32 lk_padding_7; /* rr_padding_2 */ + __u32 lk_padding_8; /* rr_padding_3 */ + __u32 lk_padding_9; /* rr_padding_4 */ }; -extern void lustre_swab_mds_rec_unlink (struct mds_rec_unlink *ul); - +/* instance of mdt_reint_rec */ struct mdt_rec_unlink { __u32 ul_opcode; __u32 ul_cap; @@ -1734,37 +2025,20 @@ struct mdt_rec_unlink { __u32 ul_suppgid2_h; struct lu_fid ul_fid1; struct lu_fid ul_fid2; - __u64 ul_time; - __u64 ul_padding_2; - __u64 ul_padding_3; - __u64 ul_padding_4; - __u64 ul_padding_5; + obd_time ul_time; + __u64 ul_padding_2; /* rr_atime */ + __u64 ul_padding_3; /* rr_ctime */ + __u64 ul_padding_4; /* rr_size */ + __u64 ul_padding_5; /* rr_blocks */ __u32 ul_bias; __u32 ul_mode; - __u32 ul_padding_6; - __u32 ul_padding_7; - __u32 ul_padding_8; - __u32 ul_padding_9; -}; - -struct mds_rec_rename { - __u32 rn_opcode; - __u32 rn_fsuid; - __u32 rn_fsgid; - __u32 rn_cap; - __u32 rn_suppgid1; - __u32 rn_suppgid2; - struct ll_fid rn_fid1; - struct ll_fid rn_fid2; - __u64 rn_time; - __u32 rn_padding_1; /* also fix lustre_swab_mds_rec_rename */ - __u32 rn_padding_2; /* also fix lustre_swab_mds_rec_rename */ - __u32 rn_padding_3; /* also fix lustre_swab_mds_rec_rename */ - __u32 rn_padding_4; /* also fix lustre_swab_mds_rec_rename */ + __u32 ul_padding_6; /* rr_flags */ + __u32 ul_padding_7; /* rr_padding_2 */ + __u32 ul_padding_8; /* rr_padding_3 */ + __u32 ul_padding_9; /* rr_padding_4 */ }; -extern void lustre_swab_mds_rec_rename (struct mds_rec_rename *rn); - +/* instance of mdt_reint_rec */ struct mdt_rec_rename { __u32 rn_opcode; __u32 rn_cap; @@ -1778,19 +2052,20 @@ struct mdt_rec_rename { __u32 rn_suppgid2_h; struct lu_fid rn_fid1; struct lu_fid rn_fid2; - __u64 rn_time; - __u64 rn_padding_1; - __u64 rn_padding_2; - __u64 rn_padding_3; - __u64 rn_padding_4; - __u32 rn_bias; /* some operation flags */ - __u32 rn_mode; /* cross-ref rename has mode */ - __u32 rn_padding_5; - __u32 rn_padding_6; - __u32 rn_padding_7; - __u32 rn_padding_8; -}; - + obd_time rn_time; + __u64 rn_padding_1; /* rr_atime */ + __u64 rn_padding_2; /* rr_ctime */ + __u64 rn_padding_3; /* rr_size */ + __u64 rn_padding_4; /* rr_blocks */ + __u32 rn_bias; /* some operation flags */ + __u32 rn_mode; /* cross-ref rename has mode */ + __u32 rn_padding_5; /* rr_flags */ + __u32 rn_padding_6; /* rr_padding_2 */ + __u32 rn_padding_7; /* rr_padding_3 */ + __u32 rn_padding_8; /* rr_padding_4 */ +}; + +/* instance of mdt_reint_rec */ struct mdt_rec_setxattr { __u32 sx_opcode; __u32 sx_cap; @@ -1803,22 +2078,30 @@ struct mdt_rec_setxattr { __u32 sx_suppgid2; __u32 sx_suppgid2_h; struct lu_fid sx_fid; - __u64 sx_padding_1; /* These three members are lu_fid size */ + __u64 sx_padding_1; /* These three are rr_fid2 */ __u32 sx_padding_2; __u32 sx_padding_3; __u64 sx_valid; - __u64 sx_time; - __u64 sx_padding_5; - __u64 sx_padding_6; - __u64 sx_padding_7; + obd_time sx_time; + __u64 sx_padding_5; /* rr_ctime */ + __u64 sx_padding_6; /* rr_size */ + __u64 sx_padding_7; /* rr_blocks */ __u32 sx_size; __u32 sx_flags; - __u32 sx_padding_8; - __u32 sx_padding_9; - __u32 sx_padding_10; - __u32 sx_padding_11; + __u32 sx_padding_8; /* rr_flags */ + __u32 sx_padding_9; /* rr_padding_2 */ + __u32 sx_padding_10; /* rr_padding_3 */ + __u32 sx_padding_11; /* rr_padding_4 */ }; +/* + * mdt_rec_reint is the template for all mdt_reint_xxx structures. + * Do NOT change the size of various members, otherwise the value + * will be broken in lustre_swab_mdt_rec_reint(). + * + * If you add new members in other mdt_reint_xxx structres and need to use the + * rr_padding_x fields, then update lustre_swab_mdt_rec_reint() also. + */ struct mdt_rec_reint { __u32 rr_opcode; __u32 rr_cap; @@ -1832,14 +2115,14 @@ struct mdt_rec_reint { __u32 rr_suppgid2_h; struct lu_fid rr_fid1; struct lu_fid rr_fid2; - __u64 rr_mtime; - __u64 rr_atime; - __u64 rr_ctime; + obd_time rr_mtime; + obd_time rr_atime; + obd_time rr_ctime; __u64 rr_size; __u64 rr_blocks; __u32 rr_bias; __u32 rr_mode; - __u32 rr_padding_1; /* also fix lustre_swab_mdt_rec_reint */ + __u32 rr_flags; __u32 rr_padding_2; /* also fix lustre_swab_mdt_rec_reint */ __u32 rr_padding_3; /* also fix lustre_swab_mdt_rec_reint */ __u32 rr_padding_4; /* also fix lustre_swab_mdt_rec_reint */ @@ -1954,6 +2237,7 @@ typedef enum { LDLM_BL_CALLBACK = 104, LDLM_CP_CALLBACK = 105, LDLM_GL_CALLBACK = 106, + LDLM_SET_INFO = 107, LDLM_LAST_OPC } ldlm_cmd_t; #define LDLM_FIRST_OPC LDLM_ENQUEUE @@ -2007,12 +2291,12 @@ struct ldlm_inodebits { __u64 bits; }; -struct ldlm_flock { - __u64 start; - __u64 end; - __u64 blocking_export; /* not actually used over the wire */ - __u32 blocking_pid; /* not actually used over the wire */ - __u32 pid; +struct ldlm_flock_wire { + __u64 lfw_start; + __u64 lfw_end; + __u64 lfw_owner; + __u32 lfw_padding; + __u32 lfw_pid; }; /* it's important that the fields of the ldlm_extent structure match @@ -2023,11 +2307,11 @@ struct ldlm_flock { typedef union { struct ldlm_extent l_extent; - struct ldlm_flock l_flock; + struct ldlm_flock_wire l_flock; struct ldlm_inodebits l_inodebits; -} ldlm_policy_data_t; +} ldlm_wire_policy_data_t; -extern void lustre_swab_ldlm_policy_data (ldlm_policy_data_t *d); +extern void lustre_swab_ldlm_policy_data (ldlm_wire_policy_data_t *d); struct ldlm_intent { __u64 opc; @@ -2047,7 +2331,7 @@ struct ldlm_lock_desc { struct ldlm_resource_desc l_resource; ldlm_mode_t l_req_mode; ldlm_mode_t l_granted_mode; - ldlm_policy_data_t l_policy_data; + ldlm_wire_policy_data_t l_policy_data; }; extern void lustre_swab_ldlm_lock_desc (struct ldlm_lock_desc *l); @@ -2140,8 +2424,8 @@ struct cfg_marker { __u32 cm_flags; __u32 cm_vers; /* lustre release version number */ __u32 padding; /* 64 bit align */ - __u64 cm_createtime; /*when this record was first created */ - __u64 cm_canceltime; /*when this record is no longer valid*/ + obd_time cm_createtime; /*when this record was first created */ + obd_time cm_canceltime; /*when this record is no longer valid*/ char cm_tgtname[MTI_NAME_MAXLEN]; char cm_comment[MTI_NAME_MAXLEN]; }; @@ -2166,7 +2450,7 @@ typedef enum { /** Identifier for a single log object */ struct llog_logid { __u64 lgl_oid; - __u64 lgl_ogr; + __u64 lgl_oseq; __u32 lgl_ogen; } __attribute__((packed)); @@ -2179,14 +2463,6 @@ struct llog_catid { __u32 lci_padding3; } __attribute__((packed)); -/** join file lov mds md*/ -struct lov_mds_md_join { - struct lov_mds_md lmmj_md; - /*join private info*/ - struct llog_logid lmmj_array_id; /*array object id*/ - __u32 lmmj_extent_count; /*array extent count*/ -}; - /* Log data record types - there is no specific reason that these need to * be related to the RPC opcodes, but no reason not to (may be handy later?) */ @@ -2203,7 +2479,7 @@ typedef enum { OBD_CFG_REC = LLOG_OP_MAGIC | 0x20000, PTL_CFG_REC = LLOG_OP_MAGIC | 0x30000, /* obsolete */ LLOG_GEN_REC = LLOG_OP_MAGIC | 0x40000, - LLOG_JOIN_REC = LLOG_OP_MAGIC | 0x50000, + LLOG_JOIN_REC = LLOG_OP_MAGIC | 0x50000, /* obsolete */ CHANGELOG_REC = LLOG_OP_MAGIC | 0x60000, CHANGELOG_USER_REC = LLOG_OP_MAGIC | 0x70000, LLOG_HDR_MAGIC = LLOG_OP_MAGIC | 0x45539, @@ -2246,28 +2522,11 @@ struct llog_logid_rec { struct llog_rec_tail lid_tail; } __attribute__((packed)); -/** MDS extent description - * It is for joined file extent info, each extent info for joined file - * just like (start, end, lmm). - */ -struct mds_extent_desc { - __u64 med_start; /* extent start */ - __u64 med_len; /* extent length */ - struct lov_mds_md med_lmm; /* extent's lmm */ -}; - -/** Joined file array extent log record*/ -struct llog_array_rec { - struct llog_rec_hdr lmr_hdr; - struct mds_extent_desc lmr_med; - struct llog_rec_tail lmr_tail; -}; - struct llog_create_rec { struct llog_rec_hdr lcr_hdr; struct ll_fid lcr_fid; obd_id lcr_oid; - obd_count lcr_ogr; + obd_count lcr_oseq; __u32 padding; struct llog_rec_tail lcr_tail; } __attribute__((packed)); @@ -2283,7 +2542,7 @@ struct llog_orphan_rec { struct llog_unlink_rec { struct llog_rec_hdr lur_hdr; obd_id lur_oid; - obd_count lur_ogr; + obd_count lur_oseq; obd_count lur_count; struct llog_rec_tail lur_tail; } __attribute__((packed)); @@ -2291,7 +2550,7 @@ struct llog_unlink_rec { struct llog_setattr_rec { struct llog_rec_hdr lsr_hdr; obd_id lsr_oid; - obd_count lsr_ogr; + obd_count lsr_oseq; __u32 lsr_uid; __u32 lsr_gid; __u32 padding; @@ -2301,7 +2560,7 @@ struct llog_setattr_rec { struct llog_setattr64_rec { struct llog_rec_hdr lsr_hdr; obd_id lsr_oid; - obd_count lsr_ogr; + obd_count lsr_oseq; __u32 padding; __u32 lsr_uid; __u32 lsr_uid_h; @@ -2319,47 +2578,13 @@ struct llog_size_change_rec { } __attribute__((packed)); #define CHANGELOG_MAGIC 0xca103000 -/** Changelog record types - * When adding record types, update mdd_lproc.c's changelog_str - */ -enum changelog_rec_type { - CL_MARK = 0, - CL_CREATE = 1, /* namespace */ - CL_MKDIR = 2, /* namespace */ - CL_HARDLINK = 3, /* namespace */ - CL_SOFTLINK = 4, /* namespace */ - CL_MKNOD = 5, /* namespace */ - CL_UNLINK = 6, /* namespace */ - CL_RMDIR = 7, /* namespace */ - CL_RENAME = 8, /* namespace */ - CL_EXT = 9, /* namespace extended record (2nd half of rename) */ - CL_OPEN = 10, /* not currently used */ - CL_CLOSE = 11, /* may be written to log only with mtime change */ - CL_IOCTL = 12, - CL_TRUNC = 13, - CL_SETATTR = 14, - CL_XATTR = 15, - CL_LAST -}; - -/** Changelog entry type names. Must be defined in the same order as the - * \a changelog_rec_type enum. - */ -#define DECLARE_CHANGELOG_NAMES static const char *changelog_str[] = \ - {"MARK","CREAT","MKDIR","HLINK","SLINK","MKNOD","UNLNK","RMDIR", \ - "RNMFM","RNMTO","OPEN","CLOSE","IOCTL","TRUNC","SATTR","XATTR"} /** \a changelog_rec_type's that can't be masked */ #define CHANGELOG_MINMASK (1 << CL_MARK) /** bits covering all \a changelog_rec_type's */ -#define CHANGELOG_ALLMASK 0XFFFF +#define CHANGELOG_ALLMASK 0XFFFFFFFF /** default \a changelog_rec_type mask */ -#define CHANGELOG_DEFMASK CHANGELOG_ALLMASK - -/* per-record flags */ -#define CLF_VERSION 0x1000 -#define CLF_FLAGMASK 0x0FFF -#define CLF_HSM 0x0001 +#define CHANGELOG_DEFMASK CHANGELOG_ALLMASK & ~(1 << CL_ATIME) /* changelog llog name, needed by client replicators */ #define CHANGELOG_CATALOG "changelog_catalog" @@ -2371,22 +2596,9 @@ struct changelog_setinfo { /** changelog record */ struct llog_changelog_rec { - struct llog_rec_hdr cr_hdr; - __u16 cr_flags; /**< (flags&CLF_FLAGMASK)|CLF_VERSION */ - __u16 cr_namelen; - __u32 cr_type; /**< \a changelog_rec_type */ - __u64 cr_index; - __u64 cr_prev; /**< last index for this target fid */ - __u64 cr_time; - union { - struct lu_fid cr_tfid; /**< target fid */ - __u32 cr_markerflags; /**< CL_MARK flags */ - }; - struct lu_fid cr_pfid; /**< parent fid */ - union { - char cr_name[0]; /**< last element */ - struct llog_rec_tail cr_tail; /**< for_sizezof_only */ - }; + struct llog_rec_hdr cr_hdr; + struct changelog_rec cr; + struct llog_rec_tail cr_tail; /**< for_sizezof_only */ } __attribute__((packed)); #define CHANGELOG_USER_PREFIX "cl" @@ -2423,7 +2635,7 @@ struct llog_gen_rec { struct llog_log_hdr { struct llog_rec_hdr llh_hdr; - __u64 llh_timestamp; + obd_time llh_timestamp; __u32 llh_count; __u32 llh_bitmap_offset; __u32 llh_size; @@ -2479,35 +2691,11 @@ struct llogd_conn_body { __u32 lgdc_ctxt_idx; } __attribute__((packed)); -struct lov_user_ost_data_join { /* per-stripe data structure */ - __u64 l_extent_start; /* extent start*/ - __u64 l_extent_end; /* extent end*/ - __u64 l_object_id; /* OST object ID */ - __u64 l_object_gr; /* OST object group (creating MDS number) */ - __u32 l_ost_gen; /* generation of this OST index */ - __u32 l_ost_idx; /* OST index in LOV */ -} __attribute__((packed)); - -struct lov_user_md_join { /* LOV EA user data (host-endian) */ - __u32 lmm_magic; /* magic number = LOV_MAGIC_JOIN */ - __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 */ - __u32 lmm_extent_count; /* extent count of lmm*/ - __u64 lmm_tree_id; /* mds tree object id */ - __u64 lmm_tree_gen; /* mds tree object gen */ - struct llog_logid lmm_array_id; /* mds extent desc llog object id */ - struct lov_user_ost_data_join lmm_objects[0]; /* per-stripe data */ -} __attribute__((packed)); - /* 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; + struct ost_id o_oi; + obd_id o_parent_seq; obd_size o_size; /* o_size-o_blocks == ost_lvb */ obd_time o_mtime; obd_time o_atime; @@ -2522,22 +2710,24 @@ struct obdo { obd_gid o_gid; obd_flag o_flags; obd_count o_nlink; /* brw: checksum */ - obd_count o_generation; + obd_count o_parent_oid; obd_count o_misc; /* brw: o_dropped */ - __u32 o_easize; /* epoch in ost writes */ - __u32 o_mds; + __u64 o_ioepoch; /* epoch in ost writes */ __u32 o_stripe_idx; /* holds stripe idx */ - __u32 o_padding_1; + __u32 o_parent_ver; 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; + __u32 o_uid_h; + __u32 o_gid_h; __u64 o_padding_3; __u64 o_padding_4; __u64 o_padding_5; __u64 o_padding_6; }; +#define o_id o_oi.oi_id +#define o_seq o_oi.oi_seq #define o_dirty o_blocks #define o_undirty o_mode #define o_dropped o_misc @@ -2551,13 +2741,19 @@ static inline void lustre_set_wire_obdo(struct obdo *wobdo, struct obdo *lobdo) static inline void lustre_get_wire_obdo(struct obdo *lobdo, struct obdo *wobdo) { - obd_flag local_flags = lobdo->o_flags & OBD_FL_LOCAL_MASK; + obd_flag local_flags = 0; + + if (lobdo->o_valid & OBD_MD_FLFLAGS) + local_flags = lobdo->o_flags & OBD_FL_LOCAL_MASK; LASSERT(!(wobdo->o_flags & OBD_FL_LOCAL_MASK)); - + memcpy(lobdo, wobdo, sizeof(*lobdo)); - lobdo->o_flags &= ~OBD_FL_LOCAL_MASK; - lobdo->o_flags |= local_flags; + if (local_flags != 0) { + lobdo->o_valid |= OBD_MD_FLFLAGS; + lobdo->o_flags &= ~OBD_FL_LOCAL_MASK; + lobdo->o_flags |= local_flags; + } } extern void lustre_swab_obdo (struct obdo *o); @@ -2582,7 +2778,6 @@ 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); extern void lustre_swab_lov_mds_md(struct lov_mds_md *lmm); /* llog_swab.c */ @@ -2595,6 +2790,13 @@ extern void lustre_swab_llog_rec(struct llog_rec_hdr *rec, struct lustre_cfg; extern void lustre_swab_lustre_cfg(struct lustre_cfg *lcfg); +/* Functions for dumping PTLRPC fields */ +void dump_rniobuf(struct niobuf_remote *rnb); +void dump_ioo(struct obd_ioobj *nb); +void dump_obdo(struct obdo *oa); +void dump_ost_body(struct ost_body *ob); +void dump_rcs(__u32 *rc); + /* this will be used when OBD_CONNECT_CHANGE_QS is set */ struct qunit_data { /** @@ -2633,7 +2835,7 @@ struct qunit_data { #define QDATA_CLR_CHANGE_QS(qdata) ((qdata)->qd_flags &= ~LQUOTA_FLAGS_CHG_QS) extern void lustre_swab_qdata(struct qunit_data *d); -extern struct qunit_data *quota_get_qdata(void*req, int is_req, int is_exp); +extern struct qunit_data *quota_get_qdata(void *req, int is_req, int is_exp); extern int quota_copy_qdata(void *request, struct qunit_data *qdata, int is_req, int is_exp); @@ -2644,8 +2846,6 @@ typedef enum { } quota_cmd_t; #define QUOTA_FIRST_OPC QUOTA_DQACQ -#define JOIN_FILE_ALIGN 4096 - #define QUOTA_REQUEST 1 #define QUOTA_REPLY 0 #define QUOTA_EXPORT 1 @@ -2739,7 +2939,7 @@ enum { #define CAPA_HMAC_ALG_MASK 0xff000000 struct lustre_capa_key { - __u64 lk_mdsid; /**< mds# */ + __u64 lk_seq; /**< mds# */ __u32 lk_keyid; /**< key# */ __u32 lk_padding; __u8 lk_key[CAPA_HMAC_KEY_MAX_LEN]; /**< key */ @@ -2762,10 +2962,9 @@ struct link_ea_header { * Stored in this crazy struct for maximum packing and endian-neutrality */ struct link_ea_entry { - struct lu_fid lee_parent_fid; /** __u16 stored big-endian, unaligned */ - char lee_reclen[2]; - __u16 lee_padding; + unsigned char lee_reclen[2]; + unsigned char lee_parent_fid[sizeof(struct lu_fid)]; char lee_name[0]; }__attribute__((packed));