/* Copyright (c) 2008-2012 Red Hat, Inc. This file is part of GlusterFS. This file is licensed to you under your choice of the GNU Lesser General Public License, version 3 or any later version (LGPLv3 or later), or the GNU General Public License, version 2 (GPLv2), in all cases as published by the Free Software Foundation. */ #ifndef __AFR_H__ #define __AFR_H__ #ifndef _CONFIG_H #define _CONFIG_H #include "config.h" #endif #include "call-stub.h" #include "compat-errno.h" #include "afr-mem-types.h" #include "libxlator.h" #include "timer.h" #include "syncop.h" #include "afr-self-heald.h" #include "afr-messages.h" #define AFR_XATTR_PREFIX "trusted.afr" #define AFR_PATHINFO_HEADER "REPLICATE:" #define AFR_SH_READDIR_SIZE_KEY "self-heal-readdir-size" #define AFR_SH_DATA_DOMAIN_FMT "%s:self-heal" #define AFR_DIRTY_DEFAULT AFR_XATTR_PREFIX ".dirty" #define AFR_DIRTY (((afr_private_t *) (THIS->private))->afr_dirty) #define AFR_LOCKEE_COUNT_MAX 3 #define AFR_DOM_COUNT_MAX 3 #define AFR_NUM_CHANGE_LOGS 3 /*data + metadata + entry*/ typedef int (*afr_lock_cbk_t) (call_frame_t *frame, xlator_t *this); typedef int (*afr_read_txn_wind_t) (call_frame_t *frame, xlator_t *this, int subvol); typedef int (*afr_inode_refresh_cbk_t) (call_frame_t *frame, xlator_t *this, int err); typedef int (*afr_changelog_resume_t) (call_frame_t *frame, xlator_t *this); #define alloca0(size) ({void *__ptr; __ptr = alloca(size); memset(__ptr, 0, size); __ptr;}) #define AFR_COUNT(array,max) ({int __i; int __res = 0; for (__i = 0; __i < max; __i++) if (array[__i]) __res++; __res;}) #define AFR_INTERSECT(dst,src1,src2,max) ({int __i; for (__i = 0; __i < max; __i++) dst[__i] = src1[__i] && src2[__i];}) #define AFR_CMP(a1,a2,len) ({int __cmp = 0; int __i; for (__i = 0; __i < len; __i++) if (a1[__i] != a2[__i]) { __cmp = 1; break;} __cmp;}) typedef struct _afr_private { gf_lock_t lock; /* to guard access to child_count, etc */ unsigned int child_count; /* total number of children */ xlator_t **children; inode_t *root_inode; unsigned char *child_up; char **pending_key; char *data_self_heal; /* on/off/open */ char * data_self_heal_algorithm; /* name of algorithm */ unsigned int data_self_heal_window_size; /* max number of pipelined read/writes */ unsigned int background_self_heal_count; unsigned int background_self_heals_started; gf_boolean_t metadata_self_heal; /* on/off */ gf_boolean_t entry_self_heal; /* on/off */ gf_boolean_t data_change_log; /* on/off */ gf_boolean_t metadata_change_log; /* on/off */ gf_boolean_t entry_change_log; /* on/off */ gf_boolean_t metadata_splitbrain_forced_heal; /* on/off */ int read_child; /* read-subvolume */ unsigned int hash_mode; /* for when read_child is not set */ int favorite_child; /* subvolume to be preferred in resolving split-brain cases */ gf_boolean_t inodelk_trace; gf_boolean_t entrylk_trace; unsigned int wait_count; /* # of servers to wait for success */ uint64_t up_count; /* number of CHILD_UPs we have seen */ uint64_t down_count; /* number of CHILD_DOWNs we have seen */ gf_boolean_t optimistic_change_log; gf_boolean_t eager_lock; gf_boolean_t pre_op_compat; /* on/off */ uint32_t post_op_delay_secs; unsigned int quorum_count; char vol_uuid[UUID_SIZE + 1]; int32_t *last_event; /* @event_generation: Keeps count of number of events received which can potentially impact consistency decisions. The events are CHILD_UP and CHILD_DOWN, when we have to recalculate the freshness/staleness of copies to detect if changes had happened while the other server was down. CHILD_DOWN and CHILD_UP can also be received on network disconnect/reconnects and not necessarily server going down/up. Recalculating freshness/staleness on network events is equally important as we might have had a network split brain. */ uint32_t event_generation; gf_boolean_t choose_local; gf_boolean_t did_discovery; uint64_t sh_readdir_size; gf_boolean_t ensure_durability; char *sh_domain; char *afr_dirty; afr_self_heald_t shd; /* pump dependencies */ void *pump_private; gf_boolean_t use_afr_in_pump; } afr_private_t; typedef enum { AFR_DATA_TRANSACTION, /* truncate, write, ... */ AFR_METADATA_TRANSACTION, /* chmod, chown, ... */ AFR_ENTRY_TRANSACTION, /* create, rmdir, ... */ AFR_ENTRY_RENAME_TRANSACTION, /* rename */ } afr_transaction_type; typedef enum { AFR_TRANSACTION_LK, AFR_SELFHEAL_LK, } transaction_lk_type_t; typedef enum { AFR_LOCK_OP, AFR_UNLOCK_OP, } afr_lock_op_type_t; typedef enum { AFR_DATA_SELF_HEAL_LK, AFR_METADATA_SELF_HEAL_LK, AFR_ENTRY_SELF_HEAL_LK, }selfheal_lk_type_t; typedef enum { AFR_INODELK_TRANSACTION, AFR_INODELK_NB_TRANSACTION, AFR_ENTRYLK_TRANSACTION, AFR_ENTRYLK_NB_TRANSACTION, AFR_INODELK_SELFHEAL, AFR_INODELK_NB_SELFHEAL, AFR_ENTRYLK_SELFHEAL, AFR_ENTRYLK_NB_SELFHEAL, } afr_lock_call_type_t; /* xattr format: trusted.afr.volume = [x y z] x - data pending y - metadata pending z - entry pending */ static inline int afr_index_for_transaction_type (afr_transaction_type type) { switch (type) { case AFR_DATA_TRANSACTION: return 0; case AFR_METADATA_TRANSACTION: return 1; case AFR_ENTRY_TRANSACTION: case AFR_ENTRY_RENAME_TRANSACTION: return 2; } return -1; /* make gcc happy */ } typedef struct { loc_t loc; char *basename; unsigned char *locked_nodes; int locked_count; } afr_entry_lockee_t; int afr_entry_lockee_cmp (const void *l1, const void *l2); typedef struct { char *domain; /* Domain on which inodelk is taken */ struct gf_flock flock; unsigned char *locked_nodes; int32_t lock_count; } afr_inodelk_t; typedef struct { loc_t *lk_loc; int lockee_count; afr_entry_lockee_t lockee[AFR_LOCKEE_COUNT_MAX]; afr_inodelk_t inodelk[AFR_DOM_COUNT_MAX]; const char *lk_basename; const char *lower_basename; const char *higher_basename; char lower_locked; char higher_locked; unsigned char *locked_nodes; unsigned char *lower_locked_nodes; selfheal_lk_type_t selfheal_lk_type; transaction_lk_type_t transaction_lk_type; int32_t lock_count; int32_t entrylk_lock_count; uint64_t lock_number; int32_t lk_call_count; int32_t lk_expected_count; int32_t lk_attempted_count; int32_t lock_op_ret; int32_t lock_op_errno; afr_lock_cbk_t lock_cbk; char *domain; /* Domain on which inode/entry lock/unlock in progress.*/ } afr_internal_lock_t; struct afr_reply { int valid; int32_t op_ret; int32_t op_errno; dict_t *xdata; struct iatt poststat; struct iatt postparent; struct iatt prestat; struct iatt preparent; struct iatt preparent2; struct iatt postparent2; uint8_t checksum[MD5_DIGEST_LENGTH]; }; typedef enum { AFR_FD_NOT_OPENED, AFR_FD_OPENED, AFR_FD_OPENING } afr_fd_open_status_t; typedef struct { unsigned int *pre_op_done[AFR_NUM_CHANGE_LOGS]; int inherited[AFR_NUM_CHANGE_LOGS]; int on_disk[AFR_NUM_CHANGE_LOGS]; afr_fd_open_status_t *opened_on; /* which subvolumes the fd is open on */ unsigned int *lock_piggyback; unsigned int *lock_acquired; int flags; /* used for delayed-post-op optimization */ pthread_mutex_t delay_lock; gf_timer_t *delay_timer; call_frame_t *delay_frame; /* set if any write on this fd was a non stable write (i.e, without O_SYNC or O_DSYNC) */ gf_boolean_t witnessed_unstable_write; /* @open_fd_count: Number of open FDs queried from the server, as queried through xdata in FOPs. Currently, used to decide if eager-locking must be temporarily disabled. */ uint32_t open_fd_count; /* list of frames currently in progress */ struct list_head eager_locked; } afr_fd_ctx_t; typedef struct _afr_local { glusterfs_fop_t op; unsigned int call_count; /* @event_generation: copy of priv->event_generation taken at the time of starting the transaction. The copy is made so that we have a stable value through the various phases of the transaction. */ unsigned int event_generation; uint32_t open_fd_count; gf_boolean_t update_open_fd_count; gf_lkowner_t saved_lk_owner; int32_t op_ret; int32_t op_errno; int32_t **pending; int dirty[AFR_NUM_CHANGE_LOGS]; loc_t loc; loc_t newloc; fd_t *fd; afr_fd_ctx_t *fd_ctx; /* @child_up: copy of priv->child_up taken at the time of transaction start. The copy is taken so that we have a stable child_up array through the phases of the transaction as priv->child_up[i] can keep changing through time. */ unsigned char *child_up; /* @read_attempted: array of flags representing subvolumes where read operations of the read transaction have already been attempted. The array is first pre-filled with down subvolumes, and as reads are performed on other subvolumes, those are set as well. This way if the read operation fails we do not retry on that subvolume again. */ unsigned char *read_attempted; /* @readfn: pointer to function which will perform the read operation on a given subvolume. Used in read transactions. */ afr_read_txn_wind_t readfn; /* @refreshed: the inode was "refreshed" (i.e, pending xattrs from all subvols freshly inspected and inode ctx updated accordingly) as part of this transaction already. */ gf_boolean_t refreshed; /* @inode: the inode on which the read txn is performed on. ref'ed and copied from either fd->inode or loc.inode */ inode_t *inode; /* @parent[2]: parent inode[s] on which directory transactions are performed. */ inode_t *parent; inode_t *parent2; /* @readable: array of flags representing servers from which a read can be performed. This is the output of afr_inode_refresh() */ unsigned char *readable; afr_inode_refresh_cbk_t refreshfn; /* @refreshinode: Inode currently getting refreshed. */ inode_t *refreshinode; /* @pre_op_compat: compatibility mode of pre-op. send a separate pre-op and op operations as part of transaction, rather than combining */ gf_boolean_t pre_op_compat; dict_t *xattr_req; afr_internal_lock_t internal_lock; dict_t *dict; int optimistic_change_log; gf_boolean_t delayed_post_op; /* Is the current writev() going to perform a stable write? i.e, is fd->flags or @flags writev param have O_SYNC or O_DSYNC? */ gf_boolean_t stable_write; /* This write appended to the file. Nnot necessarily O_APPEND, just means the offset of write was at the end of file. */ gf_boolean_t append_write; /* This struct contains the arguments for the "continuation" (scheme-like) of fops */ struct { struct { gf_boolean_t needs_fresh_lookup; uuid_t gfid_req; } lookup; struct { unsigned char buf_set; struct statvfs buf; } statfs; struct { int32_t flags; } open; struct { int32_t cmd; struct gf_flock user_flock; struct gf_flock ret_flock; unsigned char *locked_nodes; } lk; /* inode read */ struct { int32_t mask; int last_index; /* index of the child we tried previously */ } access; struct { int last_index; } stat; struct { int last_index; } fstat; struct { size_t size; int last_index; } readlink; struct { char *name; int last_index; long xattr_len; } getxattr; struct { size_t size; off_t offset; int last_index; uint32_t flags; } readv; /* dir read */ struct { int success_count; int32_t op_ret; int32_t op_errno; uint32_t *checksum; } opendir; struct { int32_t op_ret; int32_t op_errno; size_t size; off_t offset; dict_t *dict; gf_boolean_t failed; int last_index; } readdir; /* inode write */ struct { struct iatt prebuf; struct iatt postbuf; } inode_wfop; //common structure for all inode-write-fops struct { int32_t op_ret; struct iovec *vector; struct iobref *iobref; int32_t count; off_t offset; uint32_t flags; } writev; struct { off_t offset; } truncate; struct { off_t offset; } ftruncate; struct { struct iatt in_buf; int32_t valid; } setattr; struct { struct iatt in_buf; int32_t valid; } fsetattr; struct { dict_t *dict; int32_t flags; } setxattr; struct { dict_t *dict; int32_t flags; } fsetxattr; struct { char *name; } removexattr; struct { dict_t *xattr; } xattrop; struct { dict_t *xattr; } fxattrop; /* dir write */ struct { inode_t *inode; struct iatt buf; struct iatt preparent; struct iatt postparent; struct iatt prenewparent; struct iatt postnewparent; } dir_fop; //common structure for all dir fops struct { fd_t *fd; dict_t *params; int32_t flags; mode_t mode; } create; struct { dev_t dev; mode_t mode; dict_t *params; } mknod; struct { int32_t mode; dict_t *params; } mkdir; struct { int flags; } rmdir; struct { dict_t *params; char *linkpath; } symlink; struct { int32_t mode; off_t offset; size_t len; } fallocate; struct { off_t offset; size_t len; } discard; struct { off_t offset; off_t len; struct iatt prebuf; struct iatt postbuf; } zerofill; struct { const char *volume; int32_t cmd; struct gf_flock flock; } inodelk; } cont; struct { off_t start, len; gf_boolean_t eager_lock_on; int *eager_lock; char *basename; char *new_basename; loc_t parent_loc; loc_t new_parent_loc; afr_transaction_type type; /* stub to resume on destruction of the transaction frame */ call_stub_t *resume_stub; struct list_head eager_locked; unsigned char *pre_op; /* @fop_subvols: subvolumes on which FOP will be attempted */ unsigned char *fop_subvols; /* @failed_subvols: subvolumes on which FOP failed. Always a subset of @fop_subvols */ unsigned char *failed_subvols; /* @dirtied: flag which indicates whether we set dirty flag in the OP. Typically true when we are performing operation on more than one subvol and optimistic changelog is disabled A 'true' value set in @dirtied flag means an 'undirtying' has to be done in POST-OP phase. */ gf_boolean_t dirtied; /* @inherited: flag which indicates that the dirty flags of the previous transaction were inherited */ gf_boolean_t inherited; /* @no_uninherit: flag which indicates that a pre_op_uninherit() must _not_ be attempted (and returned as failure) always. This flag is set when a hard pre-op is performed, but not accounted for it in fd_ctx->on_disk[]. Such transactions are "isolated" from the pre-op piggybacking entirely and therefore uninherit must not be attempted. */ gf_boolean_t no_uninherit; /* @uninherit_done: @uninherit_value: The above pair variables make pre_op_uninherit() idempotent. Both are FALSE initially. The first call to pre_op_uninherit sets @uninherit_done to TRUE and the return value to @uninherit_value. Further calls will check for @uninherit_done to be TRUE and if so will simply return @uninherit_value. */ gf_boolean_t uninherit_done; gf_boolean_t uninherit_value; /* @changelog_resume: function to be called after changlogging (either pre-op or post-op) is done */ afr_changelog_resume_t changelog_resume; call_frame_t *main_frame; int (*wind) (call_frame_t *frame, xlator_t *this, int subvol); int (*fop) (call_frame_t *frame, xlator_t *this); int (*done) (call_frame_t *frame, xlator_t *this); int (*resume) (call_frame_t *frame, xlator_t *this); int (*unwind) (call_frame_t *frame, xlator_t *this); /* post-op hook */ } transaction; syncbarrier_t barrier; struct marker_str marker; /* extra data for fops */ dict_t *xdata_req; dict_t *xdata_rsp; mode_t umask; int xflag; gf_boolean_t do_discovery; struct afr_reply *replies; } afr_local_t; /* did a call fail due to a child failing? */ #define child_went_down(op_ret, op_errno) (((op_ret) < 0) && \ ((op_errno == ENOTCONN) || \ (op_errno == EBADFD))) int afr_inode_read_subvol_get (inode_t *inode, xlator_t *this, unsigned char *data_subvols, unsigned char *metadata_subvols, int *event_generation); int __afr_inode_read_subvol_get (inode_t *inode, xlator_t *this, unsigned char *data_subvols, unsigned char *metadata_subvols, int *event_generation); int __afr_inode_read_subvol_set (inode_t *inode, xlator_t *this, unsigned char *data_subvols, unsigned char *metadata_subvol, int event_generation); int afr_inode_read_subvol_set (inode_t *inode, xlator_t *this, unsigned char *data_subvols, unsigned char *metadata_subvols, int event_generation); int afr_inode_read_subvol_reset (inode_t *inode, xlator_t *this); int afr_read_subvol_select_by_policy (inode_t *inode, xlator_t *this, unsigned char *readable); int afr_inode_read_subvol_type_get (inode_t *inode, xlator_t *this, unsigned char *readable, int *event_p, int type); int afr_read_subvol_get (inode_t *inode, xlator_t *this, int *subvol_p, int *event_p, afr_transaction_type type); #define afr_data_subvol_get(i, t, s, e) \ afr_read_subvol_get(i, t, s, e, AFR_DATA_TRANSACTION) #define afr_metadata_subvol_get(i, t, s, e) \ afr_read_subvol_get(i, t, s, e, AFR_METADATA_TRANSACTION) int afr_inode_refresh (call_frame_t *frame, xlator_t *this, inode_t *inode, afr_inode_refresh_cbk_t cbk); int32_t afr_notify (xlator_t *this, int32_t event, void *data, void *data2); int afr_init_entry_lockee (afr_entry_lockee_t *lockee, afr_local_t *local, loc_t *loc, char *basename, int child_count); void afr_entry_lockee_cleanup (afr_internal_lock_t *int_lock); int afr_attempt_lock_recovery (xlator_t *this, int32_t child_index); int afr_mark_locked_nodes (xlator_t *this, fd_t *fd, unsigned char *locked_nodes); void afr_set_lk_owner (call_frame_t *frame, xlator_t *this, void *lk_owner); int afr_set_lock_number (call_frame_t *frame, xlator_t *this); int32_t afr_unlock (call_frame_t *frame, xlator_t *this); int afr_nonblocking_entrylk (call_frame_t *frame, xlator_t *this); int afr_nonblocking_inodelk (call_frame_t *frame, xlator_t *this); int afr_blocking_lock (call_frame_t *frame, xlator_t *this); int afr_internal_lock_finish (call_frame_t *frame, xlator_t *this); int afr_lk_transfer_datalock (call_frame_t *dst, call_frame_t *src, char *dom, unsigned int child_count); int __afr_fd_ctx_set (xlator_t *this, fd_t *fd); int afr_fd_ctx_set (xlator_t *this, fd_t *fd); afr_fd_ctx_t * afr_fd_ctx_get (fd_t *fd, xlator_t *this); int afr_build_parent_loc (loc_t *parent, loc_t *child, int32_t *op_errno); int afr_locked_nodes_count (unsigned char *locked_nodes, int child_count); int afr_replies_interpret (call_frame_t *frame, xlator_t *this, inode_t *inode); void afr_replies_wipe (afr_local_t *local, afr_private_t *priv); void afr_local_cleanup (afr_local_t *local, xlator_t *this); int afr_frame_return (call_frame_t *frame); int afr_open (call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags, fd_t *fd, dict_t *xdata); void afr_local_transaction_cleanup (afr_local_t *local, xlator_t *this); int afr_cleanup_fd_ctx (xlator_t *this, fd_t *fd); #define AFR_STACK_UNWIND(fop, frame, params ...) \ do { \ afr_local_t *__local = NULL; \ xlator_t *__this = NULL; \ if (frame) { \ __local = frame->local; \ __this = frame->this; \ frame->local = NULL; \ } \ STACK_UNWIND_STRICT (fop, frame, params); \ if (__local) { \ afr_local_cleanup (__local, __this); \ mem_put (__local); \ } \ } while (0) #define AFR_STACK_DESTROY(frame) \ do { \ afr_local_t *__local = NULL; \ xlator_t *__this = NULL; \ __local = frame->local; \ __this = frame->this; \ frame->local = NULL; \ STACK_DESTROY (frame->root); \ if (__local) { \ afr_local_cleanup (__local, __this); \ mem_put (__local); \ } \ } while (0); #define AFR_FRAME_INIT(frame, op_errno) \ ({frame->local = mem_get0 (THIS->local_pool); \ if (afr_local_init (frame->local, THIS->private, &op_errno)) { \ afr_local_cleanup (frame->local, THIS); \ mem_put (frame->local); \ frame->local = NULL; }; \ frame->local;}) #define AFR_STACK_RESET(frame) do { int opr; STACK_RESET (frame->root); AFR_FRAME_INIT(frame, opr);} while (0) /* allocate and return a string that is the basename of argument */ static inline char * AFR_BASENAME (const char *str) { char *__tmp_str = NULL; char *__basename_str = NULL; __tmp_str = gf_strdup (str); __basename_str = gf_strdup (basename (__tmp_str)); GF_FREE (__tmp_str); return __basename_str; } call_frame_t * afr_copy_frame (call_frame_t *base); int afr_transaction_local_init (afr_local_t *local, xlator_t *this); int32_t afr_marker_getxattr (call_frame_t *frame, xlator_t *this, loc_t *loc, const char *name,afr_local_t *local, afr_private_t *priv ); int afr_local_init (afr_local_t *local, afr_private_t *priv, int32_t *op_errno); int afr_internal_lock_init (afr_internal_lock_t *lk, size_t child_count, transaction_lk_type_t lk_type); int afr_higher_errno (int32_t old_errno, int32_t new_errno); int afr_final_errno (afr_local_t *local, afr_private_t *priv); int afr_xattr_req_prepare (xlator_t *this, dict_t *xattr_req); void afr_fix_open (fd_t *fd, xlator_t *this); afr_fd_ctx_t * afr_fd_ctx_get (fd_t *fd, xlator_t *this); void afr_set_low_priority (call_frame_t *frame); int afr_child_fd_ctx_set (xlator_t *this, fd_t *fd, int32_t child, int flags); void afr_matrix_cleanup (int32_t **pending, unsigned int m); int32_t** afr_matrix_create (unsigned int m, unsigned int n); void afr_filter_xattrs (dict_t *xattr); /* * Special value indicating we should use the "auto" quorum method instead of * a fixed value (including zero to turn off quorum enforcement). */ #define AFR_QUORUM_AUTO INT_MAX int afr_fd_report_unstable_write (xlator_t *this, fd_t *fd); gf_boolean_t afr_fd_has_witnessed_unstable_write (xlator_t *this, fd_t *fd); void afr_delayed_changelog_wake_resume (xlator_t *this, fd_t *fd, call_stub_t *stub); int afr_inodelk_init (afr_inodelk_t *lk, char *dom, size_t child_count); void afr_handle_open_fd_count (call_frame_t *frame, xlator_t *this); int afr_local_pathinfo (char *pathinfo, gf_boolean_t *is_local); void afr_remove_eager_lock_stub (afr_local_t *local); #endif /* __AFR_H__ */