lustre_idl.h

非常经典的一个分布式系统

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;

#define OBD_FL_INLINEDATA    (0x00000001)
#define OBD_FL_OBDMDEXISTS   (0x00000002)
#define OBD_FL_DELORPHAN     (0x00000004) /* if set in o_flags delete orphans */
#define OBD_FL_NORPC         (0x00000008) /* set in o_flags do in OSC not OST */
#define OBD_FL_IDONLY        (0x00000010) /* set in o_flags only adjust obj id*/
#define OBD_FL_RECREATE_OBJS (0x00000020) /* recreate missing obj */
#define OBD_FL_DEBUG_CHECK   (0x00000040) /* echo client/server debug check */
#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 */
#define OBD_FL_CREATE_CROW   (0x00000400) /* object should be create on write */

/*
 * 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.
 */
#define OBD_FL_TRUNCLOCK     (0x00000800)

/* this should be not smaller than sizeof(struct lustre_handle) + sizeof(struct
 * llog_cookie) + sizeof(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_PATTERN_RAID0 0x001   /* stripes are used round-robin */
#define LOV_PATTERN_RAID1 0x002   /* stripes are mirrors of each other */
#define LOV_PATTERN_FIRST 0x100   /* first stripe is not in round-robin */
#define LOV_PATTERN_CMOBD 0x200

#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) */
        __u32 l_ost_gen;          /* generation of this l_ost_idx */
        __u32 l_ost_idx;          /* OST index in LOV (lov_tgt_desc->tgts) */
};

#define lov_mds_md lov_mds_md_v1
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 */
        __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 */
};

#define OBD_MD_FLID        (0x00000001ULL) /* object ID */
#define OBD_MD_FLATIME     (0x00000002ULL) /* access time */
#define OBD_MD_FLMTIME     (0x00000004ULL) /* data modification time */
#define OBD_MD_FLCTIME     (0x00000008ULL) /* change time */
#define OBD_MD_FLSIZE      (0x00000010ULL) /* size */
#define OBD_MD_FLBLOCKS    (0x00000020ULL) /* allocated blocks count */
#define OBD_MD_FLBLKSZ     (0x00000040ULL) /* block size */
#define OBD_MD_FLMODE      (0x00000080ULL) /* access bits (mode & ~S_IFMT) */
#define OBD_MD_FLTYPE      (0x00000100ULL) /* object type (mode & S_IFMT) */
#define OBD_MD_FLUID       (0x00000200ULL) /* user ID */
#define OBD_MD_FLGID       (0x00000400ULL) /* group ID */
#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_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_FLCKSUM     (0x00100000ULL) /* bulk data checksum */
#define OBD_MD_FLQOS       (0x00200000ULL) /* quality of service stats */
#define OBD_MD_FLOSCOPQ    (0x00400000ULL) /* osc opaque data */
#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_FLGRANT     (0x08000000ULL) /* ost preallocation space grant */
#define OBD_MD_FLDIREA     (0x10000000ULL) /* dir's extended attribute data */
#define OBD_MD_FLUSRQUOTA  (0x20000000ULL) /* over quota flags sent from ost */
#define OBD_MD_FLGRPQUOTA  (0x40000000ULL) /* over quota flags sent from ost */
#define OBD_MD_FLMODEASIZE (0x80000000ULL) /* EA size will be changed */

#define OBD_MD_MDS         (0x0000000100000000ULL) /* where an inode lives on */
#define OBD_MD_REINT       (0x0000000200000000ULL) /* reintegrate oa */

#define OBD_MD_FLXATTR     (0x0000001000000000ULL) /* xattr */
#define OBD_MD_FLXATTRLS   (0x0000002000000000ULL) /* xattr list */
#define OBD_MD_FLXATTRRM   (0x0000004000000000ULL) /* xattr remove */
#define OBD_MD_FLACL       (0x0000008000000000ULL) /* ACL */

#define OBD_MD_FLGETATTR (OBD_MD_FLID    | OBD_MD_FLATIME | OBD_MD_FLMTIME | \
                          OBD_MD_FLCTIME | OBD_MD_FLSIZE  | OBD_MD_FLBLKSZ | \
                          OBD_MD_FLMODE  | OBD_MD_FLTYPE  | OBD_MD_FLUID   | \
                          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 */

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;       /* positive error code on server */
        __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);

/* ost_body.data values for OST_BRW */

#define OBD_BRW_READ            0x01
#define OBD_BRW_WRITE           0x02
#define OBD_BRW_RWMASK          (OBD_BRW_READ | OBD_BRW_WRITE)
#define OBD_BRW_SYNC            0x08
#define OBD_BRW_CHECK           0x10
#define OBD_BRW_FROM_GRANT      0x20 /* the osc manages this under llite */
#define OBD_BRW_GRANTED         0x40 /* the ost manages this */
#define OBD_BRW_DROP            0x80 /* drop the page after IO */
#define OBD_BRW_NOQUOTA        0x100
#define OBD_BRW_SRVLOCK        0x200 /* Client holds no lock over this page */

#define OBD_OBJECT_EOF 0xffffffffffffffffULL

#define OST_MIN_PRECREATE 32
#define OST_MAX_PRECREATE 20000

struct obd_ioobj {
        obd_id               ioo_id;
        obd_gr               ioo_gr;
        __u32                ioo_type;
        __u32                ioo_bufcnt;
};

extern void lustre_swab_obd_ioobj (struct obd_ioobj *ioo);

/* multiple of 8 bytes => can array */
struct niobuf_remote {
        __u64 offset;
        __u32 len;
        __u32 flags;
};

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 
 * negative, i.e. because ((MASK + rc) & MASK) != MASK. */
#define OST_LVB_ERR_INIT 0xffbadbad80000000ULL
#define OST_LVB_ERR_MASK 0xffbadbad00000000ULL
#define OST_LVB_IS_ERR(blocks)                                          \
        ((blocks & OST_LVB_ERR_MASK) == OST_LVB_ERR_MASK)
#define OST_LVB_SET_ERR(blocks, rc)                                     \
        do { blocks = OST_LVB_ERR_INIT + rc; } while (0)
#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;
};

extern void lustre_swab_ost_lvb(struct ost_lvb *);

/*
 *   MDS REQ RECORDS
 */

/* opcodes */
typedef enum {
        MDS_GETATTR      = 33,
        MDS_GETATTR_NAME = 34,
        MDS_CLOSE        = 35,
        MDS_REINT        = 36,
        MDS_READPAGE     = 37,
        MDS_CONNECT      = 38,
        MDS_DISCONNECT   = 39,
        MDS_GETSTATUS    = 40,
        MDS_STATFS       = 41,
        MDS_PIN          = 42,
        MDS_UNPIN        = 43,
        MDS_SYNC         = 44,
        MDS_DONE_WRITING = 45,
        MDS_SET_INFO     = 46,
        MDS_QUOTACHECK   = 47,
        MDS_QUOTACTL     = 48,
        MDS_GETXATTR     = 49,
        MDS_SETXATTR     = 50,
        MDS_LAST_OPC
} mds_cmd_t;

#define MDS_FIRST_OPC    MDS_GETATTR

/*
 * Do not exceed 63
 */

typedef enum {
        REINT_SETATTR  = 1,
        REINT_CREATE   = 2,
        REINT_LINK     = 3,
        REINT_UNLINK   = 4,
        REINT_RENAME   = 5,
        REINT_OPEN     = 6,
//      REINT_CLOSE    = 7,
//      REINT_WRITE    = 8,
        REINT_MAX
} mds_reint_t;

/* the disposition of the intent outlines what was executed */
#define DISP_IT_EXECD        0x00000001
#define DISP_LOOKUP_EXECD    0x00000002
#define DISP_LOOKUP_NEG      0x00000004
#define DISP_LOOKUP_POS      0x00000008
#define DISP_OPEN_CREATE     0x00000010
#define DISP_OPEN_OPEN       0x00000020
#define DISP_ENQ_COMPLETE    0x00400000
#define DISP_ENQ_OPEN_REF    0x00800000
#define DISP_ENQ_CREATE_REF  0x01000000
#define DISP_OPEN_LOCK       0x02000000

/* INODE LOCK PARTS */
#define MDS_INODELOCK_LOOKUP 0x000001       /* dentry, mode, owner, group */
#define MDS_INODELOCK_UPDATE 0x000002       /* size, links, timestamps */
#define MDS_INODELOCK_OPEN   0x000004       /* For opened files */

/* Do not forget to increase MDS_INODELOCK_MAXSHIFT when adding new bits */
#define MDS_INODELOCK_MAXSHIFT 2
/* 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
#define MDS_STATUS_LOV 2

struct mds_status_req {
        __u32  flags;
        __u32  repbuf;
};

extern void lustre_swab_mds_status_req (struct mds_status_req *r);

#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 */
#define MDS_SYNC_FL             0x00000008 /* Synchronous updates */
#define MDS_IMMUTABLE_FL        0x00000010 /* Immutable file */
#define MDS_APPEND_FL           0x00000020 /* writes to file may only append */
#define MDS_NOATIME_FL          0x00000080 /* do not update atime */
#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). */
static inline int ll_ext_to_inode_flags(int flags)
{
        return (flags & MDS_BFLAG_EXT_FLAGS) ?
               (((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);
}

/* If keep is set, we do not do anything with iflags, if it is not set, we
 * assume that iflags are inode flags and we need to conver those to
 * EXT3_*_FL flags (see bug 9486 and 12848) */