X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Finclude%2Flustre_disk.h;h=b0dec84a11d7e1acfd0bf8d9fde3792ef7c7b45a;hp=7ccce885d3d27d7cf672d14e52ae35ed1af3ed20;hb=ecd28d9b6cb691bda8184a7e07f1acc1ccded391;hpb=6a6561972406043efe41ae43b64fd278f360a4b9 diff --git a/lustre/include/lustre_disk.h b/lustre/include/lustre_disk.h index 7ccce88..b0dec84 100644 --- a/lustre/include/lustre_disk.h +++ b/lustre/include/lustre_disk.h @@ -27,7 +27,7 @@ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2012, Whamcloud, Inc. + * Copyright (c) 2011, 2014, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -53,19 +53,25 @@ /****************** on-disk files *********************/ -#define MDT_LOGS_DIR "LOGS" /* COMPAT_146 */ -#define MOUNT_CONFIGS_DIR "CONFIGS" -#define CONFIGS_FILE "mountdata" +#define MDT_LOGS_DIR "LOGS" /* COMPAT_146 */ +#define MOUNT_CONFIGS_DIR "CONFIGS" +#define CONFIGS_FILE "mountdata" /** Persistent mount data are stored on the disk in this file. */ -#define MOUNT_DATA_FILE MOUNT_CONFIGS_DIR"/"CONFIGS_FILE -#define LAST_RCVD "last_rcvd" -#define LOV_OBJID "lov_objid" -#define HEALTH_CHECK "health_check" -#define CAPA_KEYS "capa_keys" -#define CHANGELOG_USERS "changelog_users" -#define MGS_NIDTBL_DIR "NIDTBL_VERSIONS" -#define QMT_DIR "quota_master" -#define QSD_DIR "quota_slave" +#define MOUNT_DATA_FILE MOUNT_CONFIGS_DIR"/"CONFIGS_FILE +#define LAST_RCVD "last_rcvd" +#define LOV_OBJID "lov_objid" +#define LOV_OBJSEQ "lov_objseq" +#define HEALTH_CHECK "health_check" +#define CAPA_KEYS "capa_keys" +#define CHANGELOG_USERS "changelog_users" +#define MGS_NIDTBL_DIR "NIDTBL_VERSIONS" +#define QMT_DIR "quota_master" +#define QSD_DIR "quota_slave" +#define HSM_ACTIONS "hsm_actions" +#define LFSCK_DIR "LFSCK" +#define LFSCK_BOOKMARK "lfsck_bookmark" +#define LFSCK_LAYOUT "lfsck_layout" +#define LFSCK_NAMESPACE "lfsck_namespace" /****************** persistent mount data *********************/ @@ -90,14 +96,14 @@ #define LDD_F_UPGRADE14 0x0200 /** process as lctl conf_param */ #define LDD_F_PARAM 0x0400 -/** backend fs make use of IAM directory format. */ -#define LDD_F_IAM_DIR 0x0800 /** all nodes are specified as service nodes */ #define LDD_F_NO_PRIMNODE 0x1000 /** IR enable flag */ #define LDD_F_IR_CAPABLE 0x2000 /** the MGS refused to register the target. */ #define LDD_F_ERROR 0x4000 +/** process at lctl conf_param */ +#define LDD_F_PARAM2 0x8000 /* opc for target register */ #define LDD_F_OPC_REG 0x10000000 @@ -105,7 +111,9 @@ #define LDD_F_OPC_READY 0x40000000 #define LDD_F_OPC_MASK 0xf0000000 -#define LDD_F_ONDISK_MASK (LDD_F_SV_TYPE_MASK | LDD_F_IAM_DIR) +#define LDD_F_ONDISK_MASK (LDD_F_SV_TYPE_MASK) + +#define LDD_F_MASK 0xFFFF enum ldd_mount_type { LDD_MT_EXT3 = 0, @@ -171,10 +179,13 @@ struct lustre_disk_data { /*8192*/char ldd_params[4096]; /* key=value pairs */ }; -#define IS_MDT(data) ((data)->ldd_flags & LDD_F_SV_TYPE_MDT) -#define IS_OST(data) ((data)->ldd_flags & LDD_F_SV_TYPE_OST) -#define IS_MGS(data) ((data)->ldd_flags & LDD_F_SV_TYPE_MGS) -#define MT_STR(data) mt_str((data)->ldd_mount_type) + +#define IS_MDT(data) ((data)->lsi_flags & LDD_F_SV_TYPE_MDT) +#define IS_OST(data) ((data)->lsi_flags & LDD_F_SV_TYPE_OST) +#define IS_MGS(data) ((data)->lsi_flags & LDD_F_SV_TYPE_MGS) +#define IS_SERVER(data) ((data)->lsi_flags & (LDD_F_SV_TYPE_MGS | \ + LDD_F_SV_TYPE_MDT | LDD_F_SV_TYPE_OST)) +#define MT_STR(data) mt_str((data)->ldd_mount_type) /* Make the mdt/ost server obd name based on the filesystem name */ static inline int server_make_name(__u32 flags, __u16 index, char *fs, @@ -196,16 +207,13 @@ static inline int server_make_name(__u32 flags, __u16 index, char *fs, return 0; } -/* Get the index from the obd name */ -int server_name2index(char *svname, __u32 *idx, char **endptr); - - /****************** mount command *********************/ /* The lmd is only used internally by Lustre; mount simply passes everything as string options */ -#define LMD_MAGIC 0xbdacbd03 +#define LMD_MAGIC 0xbdacbd03 +#define LMD_PARAMS_MAXLEN 4096 /* gleaned from the mount command - no persistent info here */ struct lustre_mount_data { @@ -226,20 +234,22 @@ struct lustre_mount_data { char *lmd_osd_type; /* OSD type */ }; -#define LMD_FLG_SERVER 0x0001 /* Mounting a server */ -#define LMD_FLG_CLIENT 0x0002 /* Mounting a client */ -#define LMD_FLG_ABORT_RECOV 0x0008 /* Abort recovery */ -#define LMD_FLG_NOSVC 0x0010 /* Only start MGS/MGC for servers, - no other services */ -#define LMD_FLG_NOMGS 0x0020 /* Only start target for servers, reusing - existing MGS services */ -#define LMD_FLG_WRITECONF 0x0040 /* Rewrite config log */ -#define LMD_FLG_NOIR 0x0080 /* NO imperative recovery */ -#define LMD_FLG_NOSCRUB 0x0100 /* Do not trigger scrub automatically */ -#define LMD_FLG_MGS 0x0200 /* Also start MGS along with server */ -#define LMD_FLG_IAM 0x0400 /* IAM dir */ -#define LMD_FLG_NO_PRIMNODE 0x0800 /* all nodes are service nodes */ -#define LMD_FLG_VIRGIN 0x1000 /* the service registers first time */ +#define LMD_FLG_SERVER 0x0001 /* Mounting a server */ +#define LMD_FLG_CLIENT 0x0002 /* Mounting a client */ +#define LMD_FLG_ABORT_RECOV 0x0008 /* Abort recovery */ +#define LMD_FLG_NOSVC 0x0010 /* Only start MGS/MGC for servers, + no other services */ +#define LMD_FLG_NOMGS 0x0020 /* Only start target for servers, reusing + existing MGS services */ +#define LMD_FLG_WRITECONF 0x0040 /* Rewrite config log */ +#define LMD_FLG_NOIR 0x0080 /* NO imperative recovery */ +#define LMD_FLG_NOSCRUB 0x0100 /* Do not trigger scrub automatically */ +#define LMD_FLG_MGS 0x0200 /* Also start MGS along with server */ +#define LMD_FLG_IAM 0x0400 /* IAM dir */ +#define LMD_FLG_NO_PRIMNODE 0x0800 /* all nodes are service nodes */ +#define LMD_FLG_VIRGIN 0x1000 /* the service registers first time */ +#define LMD_FLG_UPDATE 0x2000 /* update parameters */ +#define LMD_FLG_HSM 0x4000 /* Start coordinator */ #define lmd_is_client(x) ((x)->lmd_flags & LMD_FLG_CLIENT) @@ -261,14 +271,14 @@ struct lustre_mount_data { /* * This limit is arbitrary (131072 clients on x86), but it is convenient to use - * 2^n * CFS_PAGE_SIZE * 8 for the number of bits that fit an order-n allocation. + * 2^n * PAGE_CACHE_SIZE * 8 for the number of bits that fit an order-n allocation. * If we need more than 131072 clients (order-2 allocation on x86) then this * should become an array of single-page pointers that are allocated on demand. */ -#if (128 * 1024UL) > (CFS_PAGE_SIZE * 8) +#if (128 * 1024UL) > (PAGE_CACHE_SIZE * 8) #define LR_MAX_CLIENTS (128 * 1024UL) #else -#define LR_MAX_CLIENTS (CFS_PAGE_SIZE * 8) +#define LR_MAX_CLIENTS (PAGE_CACHE_SIZE * 8) #endif /** COMPAT_146: this is an OST (temporary) */ @@ -279,7 +289,9 @@ struct lustre_mount_data { #define OBD_COMPAT_20 0x00000008 /** MDS handles LOV_OBJID file */ -#define OBD_ROCOMPAT_LOVOBJID 0x00000001 +#define OBD_ROCOMPAT_LOVOBJID 0x00000001 +/** store OST index in the IDIF */ +#define OBD_ROCOMPAT_IDX_IN_IDIF 0x00000002 /** OST handles group subdirs */ #define OBD_INCOMPAT_GROUPS 0x00000001 @@ -323,8 +335,8 @@ struct lr_server_data { __u64 lsd_catalog_oid; /* recovery catalog object id */ __u32 lsd_catalog_ogen; /* recovery catalog inode generation */ __u8 lsd_peeruuid[40]; /* UUID of MDS associated with this OST */ - __u32 lsd_ost_index; /* index number of OST in LOV */ - __u32 lsd_mdt_index; /* index number of MDT in LMV */ + __u32 lsd_osd_index; /* index number of OST in LOV */ + __u32 lsd_padding1; /* was lsd_mdt_index, unused in 2.4.0 */ __u32 lsd_start_epoch; /* VBR: start epoch from last boot */ /** transaction values since lsd_trans_table_time */ __u64 lsd_trans_table[LR_EXPIRE_INTERVALS]; @@ -358,7 +370,7 @@ struct lsd_client_data { static inline void check_lcd(char *obd_name, int index, struct lsd_client_data *lcd) { - int length = sizeof(lcd->lcd_uuid); + size_t length = sizeof(lcd->lcd_uuid); if (strnlen((char*)lcd->lcd_uuid, length) == length) { lcd->lcd_uuid[length - 1] = '\0'; @@ -372,55 +384,55 @@ static inline void check_lcd(char *obd_name, int index, static inline void lsd_le_to_cpu(struct lr_server_data *buf, struct lr_server_data *lsd) { - int i; - memcpy(lsd->lsd_uuid, buf->lsd_uuid, sizeof (lsd->lsd_uuid)); - lsd->lsd_last_transno = le64_to_cpu(buf->lsd_last_transno); - lsd->lsd_compat14 = le64_to_cpu(buf->lsd_compat14); - lsd->lsd_mount_count = le64_to_cpu(buf->lsd_mount_count); - lsd->lsd_feature_compat = le32_to_cpu(buf->lsd_feature_compat); - lsd->lsd_feature_rocompat = le32_to_cpu(buf->lsd_feature_rocompat); - lsd->lsd_feature_incompat = le32_to_cpu(buf->lsd_feature_incompat); - lsd->lsd_server_size = le32_to_cpu(buf->lsd_server_size); - lsd->lsd_client_start = le32_to_cpu(buf->lsd_client_start); - lsd->lsd_client_size = le16_to_cpu(buf->lsd_client_size); - lsd->lsd_subdir_count = le16_to_cpu(buf->lsd_subdir_count); - lsd->lsd_catalog_oid = le64_to_cpu(buf->lsd_catalog_oid); - lsd->lsd_catalog_ogen = le32_to_cpu(buf->lsd_catalog_ogen); - memcpy(lsd->lsd_peeruuid, buf->lsd_peeruuid, sizeof(lsd->lsd_peeruuid)); - lsd->lsd_ost_index = le32_to_cpu(buf->lsd_ost_index); - lsd->lsd_mdt_index = le32_to_cpu(buf->lsd_mdt_index); - lsd->lsd_start_epoch = le32_to_cpu(buf->lsd_start_epoch); - for (i = 0; i < LR_EXPIRE_INTERVALS; i++) - lsd->lsd_trans_table[i] = le64_to_cpu(buf->lsd_trans_table[i]); - lsd->lsd_trans_table_time = le32_to_cpu(buf->lsd_trans_table_time); - lsd->lsd_expire_intervals = le32_to_cpu(buf->lsd_expire_intervals); + int i; + memcpy(lsd->lsd_uuid, buf->lsd_uuid, sizeof(lsd->lsd_uuid)); + lsd->lsd_last_transno = le64_to_cpu(buf->lsd_last_transno); + lsd->lsd_compat14 = le64_to_cpu(buf->lsd_compat14); + lsd->lsd_mount_count = le64_to_cpu(buf->lsd_mount_count); + lsd->lsd_feature_compat = le32_to_cpu(buf->lsd_feature_compat); + lsd->lsd_feature_rocompat = le32_to_cpu(buf->lsd_feature_rocompat); + lsd->lsd_feature_incompat = le32_to_cpu(buf->lsd_feature_incompat); + lsd->lsd_server_size = le32_to_cpu(buf->lsd_server_size); + lsd->lsd_client_start = le32_to_cpu(buf->lsd_client_start); + lsd->lsd_client_size = le16_to_cpu(buf->lsd_client_size); + lsd->lsd_subdir_count = le16_to_cpu(buf->lsd_subdir_count); + lsd->lsd_catalog_oid = le64_to_cpu(buf->lsd_catalog_oid); + lsd->lsd_catalog_ogen = le32_to_cpu(buf->lsd_catalog_ogen); + memcpy(lsd->lsd_peeruuid, buf->lsd_peeruuid, sizeof(lsd->lsd_peeruuid)); + lsd->lsd_osd_index = le32_to_cpu(buf->lsd_osd_index); + lsd->lsd_padding1 = le32_to_cpu(buf->lsd_padding1); + lsd->lsd_start_epoch = le32_to_cpu(buf->lsd_start_epoch); + for (i = 0; i < LR_EXPIRE_INTERVALS; i++) + lsd->lsd_trans_table[i] = le64_to_cpu(buf->lsd_trans_table[i]); + lsd->lsd_trans_table_time = le32_to_cpu(buf->lsd_trans_table_time); + lsd->lsd_expire_intervals = le32_to_cpu(buf->lsd_expire_intervals); } static inline void lsd_cpu_to_le(struct lr_server_data *lsd, struct lr_server_data *buf) { - int i; - memcpy(buf->lsd_uuid, lsd->lsd_uuid, sizeof (buf->lsd_uuid)); - buf->lsd_last_transno = cpu_to_le64(lsd->lsd_last_transno); - buf->lsd_compat14 = cpu_to_le64(lsd->lsd_compat14); - buf->lsd_mount_count = cpu_to_le64(lsd->lsd_mount_count); - buf->lsd_feature_compat = cpu_to_le32(lsd->lsd_feature_compat); - buf->lsd_feature_rocompat = cpu_to_le32(lsd->lsd_feature_rocompat); - buf->lsd_feature_incompat = cpu_to_le32(lsd->lsd_feature_incompat); - buf->lsd_server_size = cpu_to_le32(lsd->lsd_server_size); - buf->lsd_client_start = cpu_to_le32(lsd->lsd_client_start); - buf->lsd_client_size = cpu_to_le16(lsd->lsd_client_size); - buf->lsd_subdir_count = cpu_to_le16(lsd->lsd_subdir_count); - buf->lsd_catalog_oid = cpu_to_le64(lsd->lsd_catalog_oid); - buf->lsd_catalog_ogen = cpu_to_le32(lsd->lsd_catalog_ogen); - memcpy(buf->lsd_peeruuid, lsd->lsd_peeruuid, sizeof(buf->lsd_peeruuid)); - buf->lsd_ost_index = cpu_to_le32(lsd->lsd_ost_index); - buf->lsd_mdt_index = cpu_to_le32(lsd->lsd_mdt_index); - buf->lsd_start_epoch = cpu_to_le32(lsd->lsd_start_epoch); - for (i = 0; i < LR_EXPIRE_INTERVALS; i++) - buf->lsd_trans_table[i] = cpu_to_le64(lsd->lsd_trans_table[i]); - buf->lsd_trans_table_time = cpu_to_le32(lsd->lsd_trans_table_time); - buf->lsd_expire_intervals = cpu_to_le32(lsd->lsd_expire_intervals); + int i; + memcpy(buf->lsd_uuid, lsd->lsd_uuid, sizeof(buf->lsd_uuid)); + buf->lsd_last_transno = cpu_to_le64(lsd->lsd_last_transno); + buf->lsd_compat14 = cpu_to_le64(lsd->lsd_compat14); + buf->lsd_mount_count = cpu_to_le64(lsd->lsd_mount_count); + buf->lsd_feature_compat = cpu_to_le32(lsd->lsd_feature_compat); + buf->lsd_feature_rocompat = cpu_to_le32(lsd->lsd_feature_rocompat); + buf->lsd_feature_incompat = cpu_to_le32(lsd->lsd_feature_incompat); + buf->lsd_server_size = cpu_to_le32(lsd->lsd_server_size); + buf->lsd_client_start = cpu_to_le32(lsd->lsd_client_start); + buf->lsd_client_size = cpu_to_le16(lsd->lsd_client_size); + buf->lsd_subdir_count = cpu_to_le16(lsd->lsd_subdir_count); + buf->lsd_catalog_oid = cpu_to_le64(lsd->lsd_catalog_oid); + buf->lsd_catalog_ogen = cpu_to_le32(lsd->lsd_catalog_ogen); + memcpy(buf->lsd_peeruuid, lsd->lsd_peeruuid, sizeof(buf->lsd_peeruuid)); + buf->lsd_osd_index = cpu_to_le32(lsd->lsd_osd_index); + buf->lsd_padding1 = cpu_to_le32(lsd->lsd_padding1); + buf->lsd_start_epoch = cpu_to_le32(lsd->lsd_start_epoch); + for (i = 0; i < LR_EXPIRE_INTERVALS; i++) + buf->lsd_trans_table[i] = cpu_to_le64(lsd->lsd_trans_table[i]); + buf->lsd_trans_table_time = cpu_to_le32(lsd->lsd_trans_table_time); + buf->lsd_expire_intervals = cpu_to_le32(lsd->lsd_expire_intervals); } static inline void lcd_le_to_cpu(struct lsd_client_data *buf, @@ -481,67 +493,76 @@ static inline __u64 lcd_last_xid(struct lsd_client_data *lcd) struct ll_sb_info; struct lustre_sb_info { - int lsi_flags; - struct obd_device *lsi_mgc; /* mgc obd */ - struct lustre_mount_data *lsi_lmd; /* mount command info */ - struct lustre_disk_data *lsi_ldd; /* mount info on-disk */ - struct ll_sb_info *lsi_llsbi; /* add'l client sbi info */ - struct vfsmount *lsi_srv_mnt; /* the one server mount */ - cfs_atomic_t lsi_mounts; /* references to the srv_mnt */ + int lsi_flags; + struct obd_device *lsi_mgc; /* mgc obd */ + struct lustre_mount_data *lsi_lmd; /* mount command info */ + struct ll_sb_info *lsi_llsbi; /* add'l client sbi info */ + struct dt_device *lsi_dt_dev; /* dt device to access disk fs*/ + atomic_t lsi_mounts; /* references to the srv_mnt */ char lsi_svname[MTI_NAME_MAXLEN]; + char lsi_osd_obdname[64]; + char lsi_osd_uuid[64]; + struct obd_export *lsi_osd_exp; char lsi_osd_type[16]; - struct backing_dev_info lsi_bdi; /* each client mountpoint needs - own backing_dev_info */ + char lsi_fstype[16]; + struct backing_dev_info lsi_bdi; /* each client mountpoint needs + own backing_dev_info */ + struct list_head lsi_lwp_list; + spinlock_t lsi_lwp_lock; + unsigned long lsi_lwp_started:1; }; -#define LSI_SERVER 0x00000001 -#define LSI_UMOUNT_FORCE 0x00000010 -#define LSI_UMOUNT_FAILOVER 0x00000020 -#define LSI_BDI_INITIALIZED 0x00000040 -#define LSI_IR_CAPABLE 0x00000080 +#define LSI_UMOUNT_FAILOVER 0x00200000 +#define LSI_BDI_INITIALIZED 0x00400000 #define s2lsi(sb) ((struct lustre_sb_info *)((sb)->s_fs_info)) #define s2lsi_nocast(sb) ((sb)->s_fs_info) #define get_profile_name(sb) (s2lsi(sb)->lsi_lmd->lmd_profile) #define get_mount_flags(sb) (s2lsi(sb)->lsi_lmd->lmd_flags) +#define get_mntdev_name(sb) (s2lsi(sb)->lsi_lmd->lmd_dev) #endif /* __KERNEL__ */ /****************** mount lookup info *********************/ struct lustre_mount_info { - char *lmi_name; - struct super_block *lmi_sb; - struct vfsmount *lmi_mnt; - cfs_list_t lmi_list_chain; + char *lmi_name; + struct super_block *lmi_sb; + struct list_head lmi_list_chain; }; /****************** prototypes *********************/ #ifdef __KERNEL__ - /* obd_mount.c */ +int server_name2fsname(const char *svname, char *fsname, const char **endptr); +int server_name2svname(const char *label, char *svname, const char **endptr, + size_t svsize); +int server_name_is_ost(const char *svname); +int target_name2index(const char *svname, __u32 *idx, const char **endptr); + +int lustre_put_lsi(struct super_block *sb); +int lustre_start_simple(char *obdname, char *type, char *uuid, + char *s1, char *s2, char *s3, char *s4); +int lustre_start_mgc(struct super_block *sb); void lustre_register_client_fill_super(int (*cfs)(struct super_block *sb, - struct vfsmount *mnt)); + struct vfsmount *mnt)); void lustre_register_kill_super_cb(void (*cfs)(struct super_block *sb)); - - int lustre_common_put_super(struct super_block *sb); -struct lustre_mount_info *server_find_mount_locked(const char *name); + +# ifdef HAVE_SERVER_SUPPORT +/* obd_mount_server.c */ +int server_fill_super(struct super_block *sb); struct lustre_mount_info *server_get_mount(const char *name); -struct lustre_mount_info *server_get_mount_2(const char *name); -int server_put_mount(const char *name, struct vfsmount *mnt); -int server_put_mount_2(const char *name, struct vfsmount *mnt); -int server_register_target(struct super_block *sb); +int server_put_mount(const char *name, bool dereg_mnt); struct mgs_target_info; -int server_mti_print(char *title, struct mgs_target_info *mti); +int server_mti_print(const char *title, struct mgs_target_info *mti); void server_calc_timeout(struct lustre_sb_info *lsi, struct obd_device *obd); +# endif -/* mgc_request.c */ int mgc_fsname2resid(char *fsname, struct ldlm_res_id *res_id, int type); - -#endif +#endif /* __KERNEL__ */ /** @} disk */