/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
* vim:expandtab:shiftwidth=8:tabstop=8:
*
- * This file is part of Lustre, http://www.lustre.org
+ * GPL HEADER START
+ *
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 only,
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License version 2 for more details (a copy is included
+ * in the LICENSE file that accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
+ *
+ * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
+ * CA 95054 USA or visit www.sun.com if you need additional information or
+ * have any questions.
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright 2008 Sun Microsystems, Inc. All rights reserved
+ * Use is subject to license terms.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * 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
+ * detects the message format by checking the 'magic' field of the message
+ * (see lustre_msg_swabbed() below).
* ALL structs passing over the wire should be declared here. Structs
* that are used in interfaces with userspace should go in lustre_user.h.
*
* in the code to ensure that new/old clients that see this larger struct
* do not fail, otherwise you need to implement protocol compatibility).
*
- * We assume all nodes are either little-endian or big-endian, and we
- * always send messages in the sender's native format. The receiver
- * detects the message format by checking the 'magic' field of the message
- * (see lustre_msg_swabbed() below).
- *
* Each wire type has corresponding 'lustre_swab_xxxtypexxx()' routines,
* implemented either here, inline (trivial implementations) or in
* ptlrpc/pack_generic.c. These 'swabbers' convert the type from "other"
* 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_
#define _LUSTRE_IDL_H_
-#if defined(__linux__)
-#include <linux/lustre_types.h>
-#elif defined(__APPLE__)
-#include <darwin/lustre_types.h>
-#elif defined(__WINNT__)
-#include <winnt/lustre_types.h>
-#else
-#error Unsupported operating system.
-#endif
+#include <libcfs/libcfs.h>
/* Defn's shared with user-space. */
#include <lustre/lustre_user.h>
-
-/*
- * this file contains all data structures used in Lustre interfaces:
- * - obdo and obd_request records
- * - mds_request records
- * - ldlm data
- * - ioctl's
- */
+#include <lustre/ll_fiemap.h>
/*
* GENERAL STUFF
//#define PTLBD_BULK_PORTAL 21
#define MDS_SETATTR_PORTAL 22
#define MDS_READPAGE_PORTAL 23
+#define MDS_MDS_PORTAL 24
#define MGC_REPLY_PORTAL 25
#define MGS_REQUEST_PORTAL 26
#define MGS_REPLY_PORTAL 27
#define OST_REQUEST_PORTAL 28
+#define FLD_REQUEST_PORTAL 29
+#define SEQ_METADATA_PORTAL 30
+#define SEQ_DATA_PORTAL 31
+#define SEQ_CONTROLLER_PORTAL 32
#define SVC_KILLED 1
#define SVC_EVENT 2
#define LUSTRE_LOG_VERSION 0x00050000
#define LUSTRE_MGS_VERSION 0x00060000
+typedef __u64 mdsno_t;
+typedef __u64 seqno_t;
+
+struct lu_range {
+ __u64 lr_start;
+ __u64 lr_end;
+};
+
+static inline __u64 range_space(struct lu_range *r)
+{
+ return r->lr_end - r->lr_start;
+}
+
+static inline void range_zero(struct lu_range *r)
+{
+ r->lr_start = r->lr_end = 0;
+}
+
+static inline int range_within(struct lu_range *r,
+ __u64 s)
+{
+ return s >= r->lr_start && s < r->lr_end;
+}
+
+static inline void range_alloc(struct lu_range *r,
+ struct lu_range *s,
+ __u64 w)
+{
+ r->lr_start = s->lr_start;
+ r->lr_end = s->lr_start + w;
+ s->lr_start += w;
+}
+
+static inline int range_is_sane(struct lu_range *r)
+{
+ return (r->lr_end >= r->lr_start);
+}
+
+static inline int range_is_zero(struct lu_range *r)
+{
+ return (r->lr_start == 0 && r->lr_end == 0);
+}
+
+static inline int range_is_exhausted(struct lu_range *r)
+{
+ return range_space(r) == 0;
+}
+
+#define DRANGE "[%#16.16"LPF64"x-%#16.16"LPF64"x]"
+
+#define PRANGE(range) \
+ (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 {
+ /**
+ * 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 {
+ /* initial fid id value */
+ LUSTRE_FID_INIT_OID = 1UL
+};
+
+/** returns fid object sequence */
+static inline __u64 fid_seq(const struct lu_fid *fid)
+{
+ return fid->f_seq;
+}
+
+/** returns fid object id */
+static inline __u32 fid_oid(const struct lu_fid *fid)
+{
+ return fid->f_oid;
+}
+
+/** returns fid object version */
+static inline __u32 fid_ver(const struct lu_fid *fid)
+{
+ return fid->f_ver;
+}
+
+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
+
+/* IDIF sequence starts from this value, i.e. 1<<32 */
+#define IDIF_SEQ_START 0x100000000ULL
+
+/**
+ * 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_is_igif(const struct lu_fid *fid)
+{
+ return fid_seq(fid) > 0 && fid_seq(fid) < IDIF_SEQ_START;
+}
+
+/**
+ * Check if a fid is idif or not.
+ * \param fid the fid to be tested.
+ * \return true if the fid is a idif; otherwise false.
+ */
+static inline int fid_is_idif(const struct lu_fid *fid)
+{
+ return fid_seq(fid) >= IDIF_SEQ_START && fid_seq(fid) < FID_SEQ_START;
+}
+
+/**
+ * Get inode number from a igif.
+ * \param fid a igif to get inode number from.
+ * \return inode number for the igif.
+ */
+static inline ino_t lu_igif_ino(const struct lu_fid *fid)
+{
+ return fid_seq(fid);
+}
+
+/**
+ * Get inode generation from a igif.
+ * \param fid a igif to get inode generation from.
+ * \return inode generation for the igif.
+ */
+static inline __u32 lu_igif_gen(const struct lu_fid *fid)
+{
+ return fid_oid(fid);
+}
+
+#define DFID "[0x%16.16"LPF64"x/0x%8.8x:0x%8.8x]"
+
+#define PFID(fid) \
+ fid_seq(fid), \
+ fid_oid(fid), \
+ fid_ver(fid)
+
+static inline void fid_cpu_to_le(struct lu_fid *dst, const struct lu_fid *src)
+{
+ /* check that all fields are converted */
+ CLASSERT(sizeof *src ==
+ sizeof fid_seq(src) +
+ sizeof fid_oid(src) + sizeof fid_ver(src));
+ LASSERTF(fid_is_igif(src) || fid_ver(src) == 0, DFID"\n", PFID(src));
+ dst->f_seq = cpu_to_le64(fid_seq(src));
+ dst->f_oid = cpu_to_le32(fid_oid(src));
+ dst->f_ver = cpu_to_le32(fid_ver(src));
+}
+
+static inline void fid_le_to_cpu(struct lu_fid *dst, const struct lu_fid *src)
+{
+ /* check that all fields are converted */
+ CLASSERT(sizeof *src ==
+ sizeof fid_seq(src) +
+ sizeof fid_oid(src) + sizeof fid_ver(src));
+ dst->f_seq = le64_to_cpu(fid_seq(src));
+ dst->f_oid = le32_to_cpu(fid_oid(src));
+ dst->f_ver = le32_to_cpu(fid_ver(src));
+ LASSERTF(fid_is_igif(dst) || fid_ver(dst) == 0, DFID"\n", PFID(dst));
+}
+
+static inline void fid_cpu_to_be(struct lu_fid *dst, const struct lu_fid *src)
+{
+ /* check that all fields are converted */
+ CLASSERT(sizeof *src ==
+ sizeof fid_seq(src) +
+ sizeof fid_oid(src) + sizeof fid_ver(src));
+ LASSERTF(fid_is_igif(src) || fid_ver(src) == 0, DFID"\n", PFID(src));
+ dst->f_seq = cpu_to_be64(fid_seq(src));
+ dst->f_oid = cpu_to_be32(fid_oid(src));
+ dst->f_ver = cpu_to_be32(fid_ver(src));
+}
+
+static inline void fid_be_to_cpu(struct lu_fid *dst, const struct lu_fid *src)
+{
+ /* check that all fields are converted */
+ CLASSERT(sizeof *src ==
+ sizeof fid_seq(src) +
+ sizeof fid_oid(src) + sizeof fid_ver(src));
+ dst->f_seq = be64_to_cpu(fid_seq(src));
+ dst->f_oid = be32_to_cpu(fid_oid(src));
+ dst->f_ver = be32_to_cpu(fid_ver(src));
+ LASSERTF(fid_is_igif(dst) || fid_ver(dst) == 0, DFID"\n", PFID(dst));
+}
+
+/**
+ * Storage representation for fids.
+ *
+ * Variable size, first byte contains the length of the whole record.
+ */
+
+struct lu_fid_pack {
+ char fp_len;
+ char fp_area[sizeof(struct lu_fid)];
+};
+
+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);
+
+static inline int fid_is_sane(const struct lu_fid *fid)
+{
+ return
+ fid != NULL &&
+ ((fid_seq(fid) >= FID_SEQ_START && fid_oid(fid) != 0
+ && fid_ver(fid) == 0) ||
+ fid_is_igif(fid));
+}
+
+static inline int fid_is_zero(const struct lu_fid *fid)
+{
+ return fid_seq(fid) == 0 && fid_oid(fid) == 0;
+}
+
+extern void lustre_swab_lu_fid(struct lu_fid *fid);
+extern void lustre_swab_lu_range(struct lu_range *range);
+
+static inline int lu_fid_eq(const struct lu_fid *f0,
+ const struct lu_fid *f1)
+{
+ /* Check that there is no alignment padding. */
+ CLASSERT(sizeof *f0 ==
+ sizeof f0->f_seq + sizeof f0->f_oid + sizeof f0->f_ver);
+ LASSERTF(fid_is_igif(f0) || fid_ver(f0) == 0, DFID, PFID(f0));
+ LASSERTF(fid_is_igif(f1) || fid_ver(f1) == 0, DFID, PFID(f1));
+ 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 {
+ struct lu_fid lde_fid;
+ __u64 lde_hash;
+ __u16 lde_reclen;
+ __u16 lde_namelen;
+ __u32 lde_pad0;
+ char lde_name[0];
+};
+
+struct lu_dirpage {
+ __u64 ldp_hash_start;
+ __u64 ldp_hash_end;
+ __u32 ldp_flags;
+ __u32 ldp_pad0;
+ struct lu_dirent ldp_entries[0];
+};
+
+enum lu_dirpage_flags {
+ LDF_EMPTY = 1 << 0
+};
+
+static inline struct lu_dirent *lu_dirent_start(struct lu_dirpage *dp)
+{
+ if (le32_to_cpu(dp->ldp_flags) & LDF_EMPTY)
+ return NULL;
+ else
+ return dp->ldp_entries;
+}
+
+static inline struct lu_dirent *lu_dirent_next(struct lu_dirent *ent)
+{
+ struct lu_dirent *next;
+
+ if (le16_to_cpu(ent->lde_reclen) != 0)
+ next = ((void *)ent) + le16_to_cpu(ent->lde_reclen);
+ else
+ next = NULL;
+
+ return next;
+}
+
+static inline int lu_dirent_size(struct lu_dirent *ent)
+{
+ if (le16_to_cpu(ent->lde_reclen) == 0) {
+ return (sizeof(*ent) +
+ le16_to_cpu(ent->lde_namelen) + 7) & ~7;
+ }
+ return le16_to_cpu(ent->lde_reclen);
+}
+
+#define DIR_END_OFF 0xfffffffffffffffeULL
+
+/** @} lu_dir */
+
struct lustre_handle {
__u64 cookie;
};
tgt->cookie = src->cookie;
}
-/* we depend on this structure to be 8-byte aligned */
-/* this type is only endian-adjusted in lustre_unpack_msg() */
-struct lustre_msg_v1 {
- struct lustre_handle lm_handle;
- __u32 lm_magic;
- __u32 lm_type;
- __u32 lm_version;
- __u32 lm_opc;
- __u64 lm_last_xid;
- __u64 lm_last_committed;
- __u64 lm_transno;
- __u32 lm_status;
- __u32 lm_flags;
- __u32 lm_conn_cnt;
- __u32 lm_bufcount;
- __u32 lm_buflens[0];
-};
+/* flags for lm_flags */
+#define MSGHDR_AT_SUPPORT 0x1
#define lustre_msg lustre_msg_v2
/* we depend on this structure to be 8-byte aligned */
__u32 lm_secflvr;
__u32 lm_magic;
__u32 lm_repsize;
- __u32 lm_timeout;
- __u32 lm_padding_1;
+ __u32 lm_cksum;
+ __u32 lm_flags;
__u32 lm_padding_2;
__u32 lm_padding_3;
__u32 lm_buflens[0];
};
-/* without security, ptlrpc_body is put in the first buffer. */
+/* without gss, ptlrpc_body is put at the first buffer. */
+#define PTLRPC_NUM_VERSIONS 4
struct ptlrpc_body {
struct lustre_handle pb_handle;
__u32 pb_type;
__u32 pb_flags;
__u32 pb_op_flags;
__u32 pb_conn_cnt;
- __u32 pb_padding_1;
- __u32 pb_padding_2;
- __u32 pb_padding_3;
- __u32 pb_padding_4;
- __u32 pb_padding_5;
+ __u32 pb_timeout; /* for req, the deadline, for rep, the service est */
+ __u32 pb_service_time; /* for rep, actual service time */
+ __u32 pb_limit;
+ __u64 pb_slv;
+ /* VBR: pre-versions */
+ __u64 pb_pre_versions[PTLRPC_NUM_VERSIONS];
+ /* padding for future needs */
+ __u64 pb_padding[4];
};
extern void lustre_swab_ptlrpc_body(struct ptlrpc_body *pb);
#define MSG_OP_FLAG_SHIFT 16
/* Flags that apply to all requests are in the bottom 16 bits */
-#define MSG_GEN_FLAG_MASK 0x0000ffff
-#define MSG_LAST_REPLAY 1
-#define MSG_RESENT 2
-#define MSG_REPLAY 4
+#define MSG_GEN_FLAG_MASK 0x0000ffff
+#define MSG_LAST_REPLAY 0x0001
+#define MSG_RESENT 0x0002
+#define MSG_REPLAY 0x0004
+/* #define MSG_AT_SUPPORT 0x0008
+ * This was used in early prototypes of adaptive timeouts, and while there
+ * shouldn't be any users of that code there also isn't a need for using this
+ * bits. Defer usage until at least 1.10 to avoid potential conflict. */
+#define MSG_DELAY_REPLAY 0x0010
+#define MSG_VERSION_REPLAY 0x0020
+#define MSG_REQ_REPLAY_DONE 0x0040
+#define MSG_LOCK_REPLAY_DONE 0x0080
/*
* Flags for all connect opcodes (MDS_CONNECT, OST_CONNECT)
#define MSG_CONNECT_LIBCLIENT 0x10
#define MSG_CONNECT_INITIAL 0x20
#define MSG_CONNECT_ASYNC 0x40
-#define MSG_CONNECT_NEXT_VER 0x80 /* use next version of lustre_msg */
+#define MSG_CONNECT_NEXT_VER 0x80 /* use next version of lustre_msg */
+#define MSG_CONNECT_TRANSNO 0x100 /* report transno */
/* Connect flags */
-#define OBD_CONNECT_RDONLY 0x1ULL /* client allowed read-only access */
-#define OBD_CONNECT_INDEX 0x2ULL /* connect to specific LOV idx */
-#define OBD_CONNECT_GRANT 0x8ULL /* OSC acquires grant at connect */
-#define OBD_CONNECT_SRVLOCK 0x10ULL /* server takes locks for client */
-#define OBD_CONNECT_VERSION 0x20ULL /* Server supports versions in ocd */
-#define OBD_CONNECT_REQPORTAL 0x40ULL /* Separate portal for non-IO reqs */
-#define OBD_CONNECT_ACL 0x80ULL /* client uses access control lists */
-#define OBD_CONNECT_XATTR 0x100ULL /* client using extended attributes*/
-#define OBD_CONNECT_CROW 0x200ULL /* MDS+OST create objects on write */
-#define OBD_CONNECT_TRUNCLOCK 0x400ULL /* locks on server for punch b=9528 */
-#define OBD_CONNECT_TRANSNO 0x800ULL /* replay sends initial transno */
-#define OBD_CONNECT_IBITS 0x1000ULL /* support for inodebits locks */
-#define OBD_CONNECT_JOIN 0x2000ULL /* files can be concatenated */
-#define OBD_CONNECT_ATTRFID 0x4000ULL /* Server supports GetAttr By Fid */
-#define OBD_CONNECT_NODEVOH 0x8000ULL /* No open handle for special nodes */
-#define OBD_CONNECT_LCL_CLIENT 0x10000ULL /* local 1.8 client */
-#define OBD_CONNECT_RMT_CLIENT 0x20000ULL /* Remote 1.8 client */
-#define OBD_CONNECT_BRW_SIZE 0x40000ULL /* Max bytes per rpc */
-#define OBD_CONNECT_QUOTA64 0x80000ULL /* 64bit qunit_data.qd_count b=10707*/
-#define OBD_CONNECT_FID_CAPA 0x100000ULL /* fid capability */
-#define OBD_CONNECT_OSS_CAPA 0x200000ULL /* OSS capability */
+#define OBD_CONNECT_RDONLY 0x00000001ULL /* client allowed read-only access */
+#define OBD_CONNECT_INDEX 0x00000002ULL /* connect to specific LOV idx */
+#define OBD_CONNECT_GRANT 0x00000008ULL /* OSC acquires grant at connect */
+#define OBD_CONNECT_SRVLOCK 0x00000010ULL /* server takes locks for client */
+#define OBD_CONNECT_VERSION 0x00000020ULL /* Server supports versions in ocd */
+#define OBD_CONNECT_REQPORTAL 0x00000040ULL /* Separate portal for non-IO reqs */
+#define OBD_CONNECT_ACL 0x00000080ULL /* client uses access control lists */
+#define OBD_CONNECT_XATTR 0x00000100ULL /* client using extended attributes*/
+#define OBD_CONNECT_TRUNCLOCK 0x00000400ULL /* locks on server for punch b=9528 */
+#define OBD_CONNECT_IBITS 0x00001000ULL /* support for inodebits locks */
+#define OBD_CONNECT_JOIN 0x00002000ULL /* files can be concatenated */
+#define OBD_CONNECT_ATTRFID 0x00004000ULL /* Server supports GetAttr By Fid */
+#define OBD_CONNECT_NODEVOH 0x00008000ULL /* No open handle for special nodes */
+#define OBD_CONNECT_LCL_CLIENT 0x00010000ULL /* local 1.8 client */
+#define OBD_CONNECT_RMT_CLIENT 0x00020000ULL /* Remote 1.8 client */
+#define OBD_CONNECT_BRW_SIZE 0x00040000ULL /* Max bytes per rpc */
+#define OBD_CONNECT_QUOTA64 0x00080000ULL /* 64bit qunit_data.qd_count b=10707*/
+#define OBD_CONNECT_MDS_CAPA 0x00100000ULL /* MDS capability */
+#define OBD_CONNECT_OSS_CAPA 0x00200000ULL /* OSS capability */
+#define OBD_CONNECT_CANCELSET 0x00400000ULL /* Early batched cancels. */
+#define OBD_CONNECT_SOM 0x00800000ULL /* SOM feature */
+#define OBD_CONNECT_AT 0x01000000ULL /* client uses adaptive timeouts */
+#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 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 */
-#define MDS_CONNECT_SUPPORTED (OBD_CONNECT_RDONLY | OBD_CONNECT_VERSION | \
+#ifdef HAVE_LRU_RESIZE_SUPPORT
+#define LRU_RESIZE_CONNECT_FLAG OBD_CONNECT_LRU_RESIZE
+#else
+#define LRU_RESIZE_CONNECT_FLAG 0
+#endif
+
+#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_NODEVOH |/* OBD_CONNECT_ATTRFID |*/\
+ OBD_CONNECT_LCL_CLIENT | \
+ OBD_CONNECT_RMT_CLIENT | \
+ 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 | \
+ 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 | \
- OBD_CONNECT_BRW_SIZE | OBD_CONNECT_QUOTA64)
+ OBD_CONNECT_BRW_SIZE | OBD_CONNECT_QUOTA64 | \
+ OBD_CONNECT_OSS_CAPA | OBD_CONNECT_CANCELSET | \
+ OBD_CONNECT_CKSUM | LRU_RESIZE_CONNECT_FLAG | \
+ OBD_CONNECT_AT)
#define ECHO_CONNECT_SUPPORTED (0)
-#define MGS_CONNECT_SUPPORTED (OBD_CONNECT_VERSION)
+#define MGS_CONNECT_SUPPORTED (OBD_CONNECT_VERSION | OBD_CONNECT_AT)
-#define MAX_QUOTA_COUNT32 ((0xffffffffULL >> QUOTABLOCK_BITS) << QUOTABLOCK_BITS)
+#define MAX_QUOTA_COUNT32 (0xffffffffULL)
#define OBD_OCD_VERSION(major,minor,patch,fix) (((major)<<24) + ((minor)<<16) +\
((patch)<<8) + (fix))
* 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 {
- __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 padding1; /* also fix lustre_swab_connect */
- __u64 padding2; /* also fix lustre_swab_connect */
- __u64 padding3; /* also fix lustre_swab_connect */
- __u64 padding4; /* also fix lustre_swab_connect */
+ __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 */
+ __u64 padding1; /* also fix lustre_swab_connect */
+ __u64 padding2; /* also fix lustre_swab_connect */
};
extern void lustre_swab_connect(struct obd_connect_data *ocd);
/*
+ * Supported checksum algorithms. Up to 32 checksum types are supported.
+ * (32-bit mask stored in obd_connect_data::ocd_cksum_types)
+ * Please update DECLARE_CKSUM_NAME/OBD_CKSUM_ALL in obd.h when adding a new
+ * algorithm and also the OBD_FL_CKSUM* flags.
+ */
+typedef enum {
+ OBD_CKSUM_CRC32 = 0x00000001,
+ OBD_CKSUM_ADLER = 0x00000002,
+} cksum_type_t;
+
+/*
* OST requests: OBDO & OBD request records
*/
OST_OPEN = 11,
OST_CLOSE = 12,
OST_STATFS = 13,
-/* OST_SAN_READ = 14, deprecated */
-/* OST_SAN_WRITE = 15, deprecated */
OST_SYNC = 16,
OST_SET_INFO = 17,
OST_QUOTACHECK = 18,
} ost_cmd_t;
#define OST_FIRST_OPC OST_REPLY
-typedef uint64_t obd_id;
-typedef uint64_t obd_gr;
-typedef uint64_t obd_time;
-typedef uint64_t obd_size;
-typedef uint64_t obd_off;
-typedef uint64_t obd_blocks;
-typedef uint32_t obd_blksize;
-typedef uint32_t obd_mode;
-typedef uint32_t obd_uid;
-typedef uint32_t obd_gid;
-typedef uint32_t obd_flag;
-typedef uint64_t obd_valid;
-typedef uint32_t obd_count;
+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 __u32 obd_blksize;
+typedef __u32 obd_mode;
+typedef __u32 obd_uid;
+typedef __u32 obd_gid;
+typedef __u32 obd_flag;
+typedef __u64 obd_valid;
+typedef __u32 obd_count;
#define OBD_FL_INLINEDATA (0x00000001)
#define OBD_FL_OBDMDEXISTS (0x00000002)
#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
+/**
+ * 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);
-
+/*
+ * Checksum types
+ */
+#define OBD_FL_CKSUM_CRC32 (0x00001000)
+#define OBD_FL_CKSUM_ADLER (0x00002000)
+#define OBD_FL_CKSUM_ALL (OBD_FL_CKSUM_CRC32 | OBD_FL_CKSUM_ADLER)
#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 */
#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 */
struct lov_ost_data_v1 lmm_objects[0]; /* per-stripe data */
};
+/* 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))
+
+#define XATTR_NAME_ACL_ACCESS "system.posix_acl_access"
+#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 */
#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 */
#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 */
+ /* ->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 */
#define OBD_MD_FLUSRQUOTA (0x20000000ULL) /* over quota flags sent from ost */
#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_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_FLRMTPERM (0x0000010000000000ULL) /* remote permission */
+#define OBD_MD_FLMDSCAPA (0x0000020000000000ULL) /* MDS capability */
+#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_FLRMTLSETFACL (0x0001000000000000ULL) /* lfs lsetfacl case */
+#define OBD_MD_FLRMTLGETFACL (0x0002000000000000ULL) /* lfs lgetfacl case */
+#define OBD_MD_FLRMTRSETFACL (0x0004000000000000ULL) /* lfs rsetfacl case */
+#define OBD_MD_FLRMTRGETFACL (0x0008000000000000ULL) /* lfs rgetfacl case */
#define OBD_MD_FLGETATTR (OBD_MD_FLID | OBD_MD_FLATIME | OBD_MD_FLMTIME | \
OBD_MD_FLCTIME | OBD_MD_FLSIZE | OBD_MD_FLBLKSZ | \
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 */
};
extern void lustre_swab_obd_statfs (struct obd_statfs *os);
+#define OBD_STATFS_NODELAY 0x0001 /* requests should be send without delay
+ * and resends for avoid deadlocks */
+
+#define OBD_STATFS_FROM_CACHE 0x0002 /* the statfs callback should not update
+ * obd_osfs_age */
/* ost_body.data values for OST_BRW */
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;
MDS_QUOTACHECK = 47,
MDS_QUOTACTL = 48,
MDS_GETXATTR = 49,
- MDS_SETXATTR = 50,
+ MDS_SETXATTR = 50, /* obsolete, now it's MDS_REINT op */
+ MDS_WRITEPAGE = 51,
+ MDS_IS_SUBDIR = 52,
MDS_LAST_OPC
} mds_cmd_t;
REINT_UNLINK = 4,
REINT_RENAME = 5,
REINT_OPEN = 6,
-// REINT_CLOSE = 7,
-// REINT_WRITE = 8,
+ REINT_SETXATTR = 7,
+// REINT_CLOSE = 8,
+// REINT_WRITE = 9,
REINT_MAX
-} mds_reint_t;
+} mds_reint_t, mdt_reint_t;
+
+extern void lustre_swab_generic_32s (__u32 *val);
/* the disposition of the intent outlines what was executed */
#define DISP_IT_EXECD 0x00000001
extern void lustre_swab_mds_status_req (struct mds_status_req *r);
+/* mdt_thread_info.mti_flags. */
+enum md_op_flags {
+ /* The flag indicates Size-on-MDS attributes are changed. */
+ MF_SOM_CHANGE = (1 << 0),
+ /* Flags indicates an epoch opens or closes. */
+ MF_EPOCH_OPEN = (1 << 1),
+ MF_EPOCH_CLOSE = (1 << 2),
+ MF_MDC_CANCEL_FID1 = (1 << 3),
+ MF_MDC_CANCEL_FID2 = (1 << 4),
+ MF_MDC_CANCEL_FID3 = (1 << 5),
+ MF_MDC_CANCEL_FID4 = (1 << 6),
+};
+
+#define MF_SOM_LOCAL_FLAGS (MF_MDC_CANCEL_FID1 | MF_MDC_CANCEL_FID2 | \
+ MF_MDC_CANCEL_FID3 | MF_MDC_CANCEL_FID4)
+
#define MDS_BFLAG_UNCOMMITTED_WRITES 0x1
#define MDS_BFLAG_EXT_FLAGS 0x80000000 /* == EXT3_RESERVED_FL */
}
#endif
+struct mdt_body {
+ struct lu_fid fid1;
+ struct lu_fid fid2;
+ struct lustre_handle handle;
+ __u64 valid;
+ __u64 size; /* Offset, in the case of MDS_READPAGE */
+ __u64 mtime;
+ __u64 atime;
+ __u64 ctime;
+ __u64 blocks; /* XID, in the case of MDS_READPAGE */
+ __u64 ioepoch;
+ __u64 ino; /* for 1.6 compatibility */
+ __u32 fsuid;
+ __u32 fsgid;
+ __u32 capability;
+ __u32 mode;
+ __u32 uid;
+ __u32 gid;
+ __u32 flags; /* from vfs for pin/unpin, MDS_BFLAG for close */
+ __u32 rdev;
+ __u32 nlink; /* #bytes to read in the case of MDS_READPAGE */
+ __u32 generation; /* for 1.6 compatibility */
+ __u32 suppgid;
+ __u32 eadatasize;
+ __u32 aclsize;
+ __u32 max_mdsize;
+ __u32 max_cookiesize;
+ __u32 padding_4; /* also fix lustre_swab_mdt_body */
+};
+
struct mds_body {
struct ll_fid fid1;
struct ll_fid fid2;
__u32 eadatasize;
__u32 aclsize;
__u32 max_mdsize;
- __u32 max_cookiesize; /* also fix lustre_swab_mds_body */
+ __u32 max_cookiesize;
__u32 padding_4; /* also fix lustre_swab_mds_body */
};
extern void lustre_swab_mds_body (struct mds_body *b);
+extern void lustre_swab_mdt_body (struct mdt_body *b);
+
+struct mdt_epoch {
+ struct lustre_handle handle;
+ __u64 ioepoch;
+ __u32 flags;
+ __u32 padding;
+};
+
+extern void lustre_swab_mdt_epoch (struct mdt_epoch *b);
#define Q_QUOTACHECK 0x800100
#define Q_INITQUOTA 0x800101 /* init slave limits */
extern void lustre_swab_obd_quotactl(struct obd_quotactl *q);
+/* inode access permission for remote user, the inode info are omitted,
+ * for client knows them. */
+struct mds_remote_perm {
+ __u32 rp_uid;
+ __u32 rp_gid;
+ __u32 rp_fsuid;
+ __u32 rp_fsgid;
+ __u32 rp_access_perm; /* MAY_READ/WRITE/EXEC */
+};
+
+/* permissions for md_perm.mp_perm */
+enum {
+ CFS_SETUID_PERM = 0x01,
+ CFS_SETGID_PERM = 0x02,
+ CFS_SETGRP_PERM = 0x04,
+ CFS_RMTACL_PERM = 0x08
+};
+
+extern void lustre_swab_mds_remote_perm(struct mds_remote_perm *p);
+
+struct mdt_remote_perm {
+ __u32 rp_uid;
+ __u32 rp_gid;
+ __u32 rp_fsuid;
+ __u32 rp_fsgid;
+ __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_suppgid;
__u32 sa_mode;
struct ll_fid sa_fid;
- __u64 sa_valid;
+ __u64 sa_valid; /* MDS_ATTR_* attributes */
__u64 sa_size;
__u64 sa_mtime;
__u64 sa_atime;
__u32 sa_padding; /* also fix lustre_swab_mds_rec_setattr */
};
-/* Remove this once we declare it in include/linux/fs.h (v21 kernel patch?) */
-#ifndef ATTR_CTIME_SET
-#define ATTR_CTIME_SET 0x2000
-#endif
-
extern void lustre_swab_mds_rec_setattr (struct mds_rec_setattr *sa);
+struct mdt_rec_setattr {
+ __u32 sa_opcode;
+ __u32 sa_fsuid;
+ __u32 sa_fsgid;
+ __u32 sa_cap;
+ __u32 sa_suppgid;
+ __u32 sa_padding_1;
+ struct lu_fid sa_fid;
+ __u64 sa_valid;
+ __u32 sa_uid;
+ __u32 sa_gid;
+ __u64 sa_size;
+ __u64 sa_blocks;
+ __u64 sa_mtime;
+ __u64 sa_atime;
+ __u64 sa_ctime;
+ __u32 sa_attr_flags;
+ __u32 sa_mode;
+ __u32 sa_padding_2;
+ __u32 sa_padding_3;
+ __u32 sa_padding_4;
+ __u32 sa_padding_5;
+};
+
+extern void lustre_swab_mdt_rec_setattr (struct mdt_rec_setattr *sa);
+
+/*
+ * Attribute flags used in mds_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.
+ */
+#define MDS_ATTR_MODE 0x1ULL /* = 1 */
+#define MDS_ATTR_UID 0x2ULL /* = 2 */
+#define MDS_ATTR_GID 0x4ULL /* = 4 */
+#define MDS_ATTR_SIZE 0x8ULL /* = 8 */
+#define MDS_ATTR_ATIME 0x10ULL /* = 16 */
+#define MDS_ATTR_MTIME 0x20ULL /* = 32 */
+#define MDS_ATTR_CTIME 0x40ULL /* = 64 */
+#define MDS_ATTR_ATIME_SET 0x80ULL /* = 128 */
+#define MDS_ATTR_MTIME_SET 0x100ULL /* = 256 */
+#define MDS_ATTR_FORCE 0x200ULL /* = 512, Not a change, but a change it */
+#define MDS_ATTR_ATTR_FLAG 0x400ULL /* = 1024 */
+#define MDS_ATTR_KILL_SUID 0x800ULL /* = 2048 */
+#define MDS_ATTR_KILL_SGID 0x1000ULL /* = 4096 */
+#define MDS_ATTR_CTIME_SET 0x2000ULL /* = 8192 */
+#define MDS_ATTR_FROM_OPEN 0x4000ULL /* = 16384, called from open path, ie O_TRUNC */
+#define MDS_ATTR_BLOCKS 0x8000ULL /* = 32768 */
+
#ifndef FMODE_READ
#define FMODE_READ 00000001
#define FMODE_WRITE 00000002
#endif
+
+#define FMODE_EPOCH 01000000
+#define FMODE_EPOCHLCK 02000000
+#define FMODE_SOM 04000000
+#define FMODE_CLOSED 0
+
+#define MDS_OPEN_CREATED 00000010
+#define MDS_OPEN_CROSS 00000020
+
#define MDS_FMODE_EXEC 00000004
#define MDS_OPEN_CREAT 00000100
#define MDS_OPEN_EXCL 00000200
#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_CREATE_RMT_ACL 01000000000 /* indicate create on remote server
+ * with default ACL */
+#define MDS_CREATE_SLAVE_OBJ 02000000000 /* indicate create slave object
+ * actually, this is for create, not
+ * conflict with other open flags */
#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 */
+/* permission for create non-directory file */
+#define MAY_CREATE (1 << 7)
+/* permission for create directory file */
+#define MAY_LINK (1 << 8)
+/* permission for delete from the directory */
+#define MAY_UNLINK (1 << 9)
+/* source's permission for rename */
+#define MAY_RENAME_SRC (1 << 10)
+/* target's permission for rename */
+#define MAY_RENAME_TAR (1 << 11)
+/* part (parent's) VTX permission check */
+#define MAY_VTX_PART (1 << 12)
+/* full VTX permission check */
+#define MAY_VTX_FULL (1 << 13)
+/* lfs rgetfacl permission check */
+#define MAY_RGETFACL (1 << 14)
+
+enum {
+ MDS_CHECK_SPLIT = 1 << 0,
+ MDS_CROSS_REF = 1 << 1,
+ MDS_VTX_BYPASS = 1 << 2,
+ MDS_PERM_BYPASS = 1 << 3,
+ MDS_SOM = 1 << 4
+};
+
+struct mds_rec_join {
+ struct ll_fid jr_fid;
+ __u64 jr_headsize;
+};
+
+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_create {
__u32 cr_opcode;
__u32 cr_fsuid;
extern void lustre_swab_mds_rec_create (struct mds_rec_create *cr);
-struct mds_rec_join {
- struct ll_fid jr_fid;
- __u64 jr_headsize;
+struct mdt_rec_create {
+ __u32 cr_opcode;
+ __u32 cr_fsuid;
+ __u32 cr_fsgid;
+ __u32 cr_cap;
+ __u32 cr_suppgid1;
+ __u32 cr_suppgid2;
+ struct lu_fid cr_fid1;
+ struct lu_fid cr_fid2;
+ struct lustre_handle cr_old_handle; /* u64 handle in case of open replay */
+ __u64 cr_time;
+ __u64 cr_rdev;
+ __u64 cr_ioepoch;
+ __u64 cr_padding_1; /* pad for 64 bits*/
+ __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_mds_rec_join (struct mds_rec_join *jr);
+extern void lustre_swab_mdt_rec_create (struct mdt_rec_create *cr);
struct mds_rec_link {
__u32 lk_opcode;
extern void lustre_swab_mds_rec_link (struct mds_rec_link *lk);
+struct mdt_rec_link {
+ __u32 lk_opcode;
+ __u32 lk_fsuid;
+ __u32 lk_fsgid;
+ __u32 lk_cap;
+ __u32 lk_suppgid1;
+ __u32 lk_suppgid2;
+ 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;
+ __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;
extern void lustre_swab_mds_rec_unlink (struct mds_rec_unlink *ul);
+struct mdt_rec_unlink {
+ __u32 ul_opcode;
+ __u32 ul_fsuid;
+ __u32 ul_fsgid;
+ __u32 ul_cap;
+ __u32 ul_suppgid1;
+ __u32 ul_suppgid2;
+ 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;
+ __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;
extern void lustre_swab_mds_rec_rename (struct mds_rec_rename *rn);
+struct mdt_rec_rename {
+ __u32 rn_opcode;
+ __u32 rn_fsuid;
+ __u32 rn_fsgid;
+ __u32 rn_cap;
+ __u32 rn_suppgid1;
+ __u32 rn_suppgid2;
+ 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;
+};
+
+struct mdt_rec_setxattr {
+ __u32 sx_opcode;
+ __u32 sx_fsuid;
+ __u32 sx_fsgid;
+ __u32 sx_cap;
+ __u32 sx_suppgid1;
+ __u32 sx_suppgid2;
+ struct lu_fid sx_fid;
+ __u64 sx_padding_1; /* These three members are lu_fid size */
+ __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;
+ __u32 sx_size;
+ __u32 sx_flags;
+ __u32 sx_padding_8;
+ __u32 sx_padding_9;
+ __u32 sx_padding_10;
+ __u32 sx_padding_11;
+};
+
+struct mdt_rec_reint {
+ __u32 rr_opcode;
+ __u32 rr_fsuid;
+ __u32 rr_fsgid;
+ __u32 rr_cap;
+ __u32 rr_suppgid1;
+ __u32 rr_suppgid2;
+ struct lu_fid rr_fid1;
+ struct lu_fid rr_fid2;
+ __u64 rr_mtime;
+ __u64 rr_atime;
+ __u64 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_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 */
+};
+
+extern void lustre_swab_mdt_rec_reint(struct mdt_rec_reint *rr);
+
+struct lmv_desc {
+ __u32 ld_tgt_count; /* how many MDS's */
+ __u32 ld_active_tgt_count; /* how many active */
+ struct obd_uuid ld_uuid;
+};
+
+extern void lustre_swab_lmv_desc (struct lmv_desc *ld);
+
+struct md_fld {
+ seqno_t mf_seq;
+ mdsno_t mf_mds;
+};
+
+extern void lustre_swab_md_fld (struct md_fld *mf);
+
+enum fld_rpc_opc {
+ FLD_QUERY = 600,
+ FLD_LAST_OPC,
+ FLD_FIRST_OPC = FLD_QUERY
+};
+
+enum seq_rpc_opc {
+ SEQ_QUERY = 700,
+ SEQ_LAST_OPC,
+ SEQ_FIRST_OPC = SEQ_QUERY
+};
+
+enum seq_op {
+ SEQ_ALLOC_SUPER = 0,
+ SEQ_ALLOC_META = 1
+};
+
/*
* LOV data structures
*/
#define LOV_MIN_STRIPE_SIZE 65536 /* maximum PAGE_SIZE (ia64), power of 2 */
#define LOV_MAX_STRIPE_COUNT 160 /* until bug 4424 is fixed */
+#define LOV_V1_INSANE_STRIPE_COUNT 65532 /* maximum stripe count bz13933 */
#define LOV_MAX_UUID_BUFFER_SIZE 8192
/* The size of the buffer the lov/mdc reserves for the
/* lock types */
typedef enum {
LCK_MINMODE = 0,
- LCK_EX = 1,
- LCK_PW = 2,
- LCK_PR = 4,
- LCK_CW = 8,
- LCK_CR = 16,
- LCK_NL = 32,
- LCK_GROUP = 64,
+ LCK_EX = 1,
+ LCK_PW = 2,
+ LCK_PR = 4,
+ LCK_CW = 8,
+ LCK_CR = 16,
+ LCK_NL = 32,
+ LCK_GROUP = 64,
+ LCK_COS = 128,
LCK_MAXMODE
} ldlm_mode_t;
+#define LCK_MODE_NUM 8
+
typedef enum {
LDLM_PLAIN = 10,
LDLM_EXTENT = 11,
__u64 gid;
};
+static inline int ldlm_extent_overlap(struct ldlm_extent *ex1,
+ struct ldlm_extent *ex2)
+{
+ return (ex1->start <= ex2->end) && (ex2->start <= ex1->end);
+}
+
struct ldlm_inodebits {
__u64 bits;
};
extern void lustre_swab_ldlm_lock_desc (struct ldlm_lock_desc *l);
+#define LDLM_LOCKREQ_HANDLES 2
+#define LDLM_ENQUEUE_CANCEL_OFF 1
+
struct ldlm_request {
__u32 lock_flags;
- __u32 lock_padding; /* also fix lustre_swab_ldlm_request */
+ __u32 lock_count;
struct ldlm_lock_desc lock_desc;
- struct lustre_handle lock_handle1;
- struct lustre_handle lock_handle2;
+ struct lustre_handle lock_handle[LDLM_LOCKREQ_HANDLES];
};
extern void lustre_swab_ldlm_request (struct ldlm_request *rq);
+/* If LDLM_ENQUEUE, 1 slot is already occupied, 1 is available.
+ * Otherwise, 2 are available. */
+#define ldlm_request_bufsize(count,type) \
+({ \
+ int _avail = LDLM_LOCKREQ_HANDLES; \
+ _avail -= (type == LDLM_ENQUEUE ? LDLM_ENQUEUE_CANCEL_OFF : 0); \
+ sizeof(struct ldlm_request) + \
+ (count > _avail ? count - _avail : 0) * \
+ sizeof(struct lustre_handle); \
+})
+
struct ldlm_reply {
__u32 lock_flags;
__u32 lock_padding; /* also fix lustre_swab_ldlm_reply */
extern void lustre_swab_ldlm_reply (struct ldlm_reply *r);
-
/*
* Opcodes for mountconf (mgs and mgc)
*/
MGS_EXCEPTION, /* node died, etc. */
MGS_TARGET_REG, /* whenever target starts up */
MGS_TARGET_DEL,
+ MGS_SET_INFO,
MGS_LAST_OPC
} mgs_cmd_t;
+#define MGS_FIRST_OPC MGS_CONNECT
+
+#define MGS_PARAM_MAXLEN 1024
+#define KEY_SET_INFO "set_info"
+
+struct mgs_send_param {
+ char mgs_param[MGS_PARAM_MAXLEN];
+};
/* We pass this info to the MGS so it can write config logs */
#define MTI_NAME_MAXLEN 64
__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.
*/
/* 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;
__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*/
__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.
*/
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).
*/
__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;
struct llog_size_change_rec {
struct llog_rec_hdr lsc_hdr;
struct ll_fid lsc_fid;
- __u32 lsc_io_epoch;
+ __u32 lsc_ioepoch;
__u32 padding;
struct llog_rec_tail lsc_tail;
} __attribute__((packed));
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;
__u32 lgc_padding;
} __attribute__((packed));
-/* llog protocol */
+/** llog protocol */
enum llogd_rpc_ops {
LLOG_ORIGIN_HANDLE_CREATE = 501,
LLOG_ORIGIN_HANDLE_NEXT_BLOCK = 502,
LLOG_CATINFO = 507, /* for lfs catinfo */
LLOG_ORIGIN_HANDLE_PREV_BLOCK = 508,
LLOG_ORIGIN_HANDLE_DESTROY = 509, /* for destroy llog object*/
+ LLOG_LAST_OPC,
+ LLOG_FIRST_OPC = LLOG_ORIGIN_HANDLE_CREATE
};
struct llogd_body {
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 */
} quota_cmd_t;
#define JOIN_FILE_ALIGN 4096
+
+/** security opcodes */
+typedef enum {
+ SEC_CTX_INIT = 801,
+ SEC_CTX_INIT_CONT = 802,
+ SEC_CTX_FINI = 803,
+ SEC_LAST_OPC,
+ SEC_FIRST_OPC = SEC_CTX_INIT
+} sec_cmd_t;
+
+/*
+ * capa related definitions
+ */
+#define CAPA_HMAC_MAX_LEN 64
+#define CAPA_HMAC_KEY_MAX_LEN 56
+
+/* NB take care when changing the sequence of elements this struct,
+ * because the offset info is used in find_capa() */
+struct lustre_capa {
+ struct lu_fid lc_fid; /* fid */
+ __u64 lc_opc; /* operations allowed */
+ __u32 lc_uid; /* uid, it is obsolete, but maybe used in
+ * future, reserve it for 64-bits aligned.*/
+ __u32 lc_flags; /* HMAC algorithm & flags */
+ __u32 lc_keyid; /* key used for the capability */
+ __u32 lc_timeout; /* capa timeout value (sec) */
+ __u64 lc_expiry; /* expiry time (sec) */
+ __u8 lc_hmac[CAPA_HMAC_MAX_LEN]; /* HMAC */
+} __attribute__((packed));
+
+extern void lustre_swab_lustre_capa(struct lustre_capa *c);
+
+/** 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 */
+
+};
+
+#define CAPA_OPC_OSS_RW (CAPA_OPC_OSS_READ | CAPA_OPC_OSS_WRITE)
+#define CAPA_OPC_MDS_ONLY \
+ (CAPA_OPC_BODY_WRITE | CAPA_OPC_BODY_READ | CAPA_OPC_INDEX_LOOKUP | \
+ CAPA_OPC_INDEX_INSERT | CAPA_OPC_INDEX_DELETE)
+#define CAPA_OPC_OSS_ONLY \
+ (CAPA_OPC_OSS_WRITE | CAPA_OPC_OSS_READ | CAPA_OPC_OSS_TRUNC)
+#define CAPA_OPC_MDS_DEFAULT ~CAPA_OPC_OSS_ONLY
+#define CAPA_OPC_OSS_DEFAULT ~(CAPA_OPC_MDS_ONLY | CAPA_OPC_OSS_ONLY)
+
+/* MDS capability covers object capability for operations of body r/w
+ * (dir readpage/sendpage), index lookup/insert/delete and meta data r/w,
+ * while OSS capability only covers object capability for operations of
+ * oss data(file content) r/w/truncate.
+ */
+static inline int capa_for_mds(struct lustre_capa *c)
+{
+ return (c->lc_opc & CAPA_OPC_INDEX_LOOKUP) != 0;
+}
+
+static inline int capa_for_oss(struct lustre_capa *c)
+{
+ return (c->lc_opc & CAPA_OPC_INDEX_LOOKUP) == 0;
+}
+
+/* lustre_capa::lc_hmac_alg */
+enum {
+ CAPA_HMAC_ALG_SHA1 = 1, /**< sha1 algorithm */
+ CAPA_HMAC_ALG_MAX,
+};
+
+#define CAPA_FL_MASK 0x00ffffff
+#define CAPA_HMAC_ALG_MASK 0xff000000
+
+struct lustre_capa_key {
+ __u64 lk_mdsid; /**< mds# */
+ __u32 lk_keyid; /**< key# */
+ __u32 lk_padding;
+ __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 */
#endif
+
+/** @} lustreidl */
git://git.whamcloud.com / fs / lustre-release.git / blobdiff