Index: vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_deadlist.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_deadlist.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_deadlist.c (revision 319948) @@ -1,553 +1,553 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012 by Delphix. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright (c) 2014 Integros [integros.com] */ #include #include #include #include #include #include /* * Deadlist concurrency: * * Deadlists can only be modified from the syncing thread. * * Except for dsl_deadlist_insert(), it can only be modified with the * dp_config_rwlock held with RW_WRITER. * * The accessors (dsl_deadlist_space() and dsl_deadlist_space_range()) can * be called concurrently, from open context, with the dl_config_rwlock held * with RW_READER. * * Therefore, we only need to provide locking between dsl_deadlist_insert() and * the accessors, protecting: * dl_phys->dl_used,comp,uncomp * and protecting the dl_tree from being loaded. * The locking is provided by dl_lock. Note that locking on the bpobj_t * provides its own locking, and dl_oldfmt is immutable. */ static int dsl_deadlist_compare(const void *arg1, const void *arg2) { const dsl_deadlist_entry_t *dle1 = arg1; const dsl_deadlist_entry_t *dle2 = arg2; if (dle1->dle_mintxg < dle2->dle_mintxg) return (-1); else if (dle1->dle_mintxg > dle2->dle_mintxg) return (+1); else return (0); } static void dsl_deadlist_load_tree(dsl_deadlist_t *dl) { zap_cursor_t zc; zap_attribute_t za; ASSERT(MUTEX_HELD(&dl->dl_lock)); ASSERT(!dl->dl_oldfmt); if (dl->dl_havetree) return; avl_create(&dl->dl_tree, dsl_deadlist_compare, sizeof (dsl_deadlist_entry_t), offsetof(dsl_deadlist_entry_t, dle_node)); for (zap_cursor_init(&zc, dl->dl_os, dl->dl_object); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { dsl_deadlist_entry_t *dle = kmem_alloc(sizeof (*dle), KM_SLEEP); - dle->dle_mintxg = strtonum(za.za_name, NULL); + dle->dle_mintxg = zfs_strtonum(za.za_name, NULL); VERIFY3U(0, ==, bpobj_open(&dle->dle_bpobj, dl->dl_os, za.za_first_integer)); avl_add(&dl->dl_tree, dle); } zap_cursor_fini(&zc); dl->dl_havetree = B_TRUE; } void dsl_deadlist_open(dsl_deadlist_t *dl, objset_t *os, uint64_t object) { dmu_object_info_t doi; mutex_init(&dl->dl_lock, NULL, MUTEX_DEFAULT, NULL); dl->dl_os = os; dl->dl_object = object; VERIFY3U(0, ==, dmu_bonus_hold(os, object, dl, &dl->dl_dbuf)); dmu_object_info_from_db(dl->dl_dbuf, &doi); if (doi.doi_type == DMU_OT_BPOBJ) { dmu_buf_rele(dl->dl_dbuf, dl); dl->dl_dbuf = NULL; dl->dl_oldfmt = B_TRUE; VERIFY3U(0, ==, bpobj_open(&dl->dl_bpobj, os, object)); return; } dl->dl_oldfmt = B_FALSE; dl->dl_phys = dl->dl_dbuf->db_data; dl->dl_havetree = B_FALSE; } void dsl_deadlist_close(dsl_deadlist_t *dl) { void *cookie = NULL; dsl_deadlist_entry_t *dle; dl->dl_os = NULL; if (dl->dl_oldfmt) { dl->dl_oldfmt = B_FALSE; bpobj_close(&dl->dl_bpobj); return; } if (dl->dl_havetree) { while ((dle = avl_destroy_nodes(&dl->dl_tree, &cookie)) != NULL) { bpobj_close(&dle->dle_bpobj); kmem_free(dle, sizeof (*dle)); } avl_destroy(&dl->dl_tree); } dmu_buf_rele(dl->dl_dbuf, dl); mutex_destroy(&dl->dl_lock); dl->dl_dbuf = NULL; dl->dl_phys = NULL; } uint64_t dsl_deadlist_alloc(objset_t *os, dmu_tx_t *tx) { if (spa_version(dmu_objset_spa(os)) < SPA_VERSION_DEADLISTS) return (bpobj_alloc(os, SPA_OLD_MAXBLOCKSIZE, tx)); return (zap_create(os, DMU_OT_DEADLIST, DMU_OT_DEADLIST_HDR, sizeof (dsl_deadlist_phys_t), tx)); } void dsl_deadlist_free(objset_t *os, uint64_t dlobj, dmu_tx_t *tx) { dmu_object_info_t doi; zap_cursor_t zc; zap_attribute_t za; VERIFY3U(0, ==, dmu_object_info(os, dlobj, &doi)); if (doi.doi_type == DMU_OT_BPOBJ) { bpobj_free(os, dlobj, tx); return; } for (zap_cursor_init(&zc, os, dlobj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { uint64_t obj = za.za_first_integer; if (obj == dmu_objset_pool(os)->dp_empty_bpobj) bpobj_decr_empty(os, tx); else bpobj_free(os, obj, tx); } zap_cursor_fini(&zc); VERIFY3U(0, ==, dmu_object_free(os, dlobj, tx)); } static void dle_enqueue(dsl_deadlist_t *dl, dsl_deadlist_entry_t *dle, const blkptr_t *bp, dmu_tx_t *tx) { ASSERT(MUTEX_HELD(&dl->dl_lock)); if (dle->dle_bpobj.bpo_object == dmu_objset_pool(dl->dl_os)->dp_empty_bpobj) { uint64_t obj = bpobj_alloc(dl->dl_os, SPA_OLD_MAXBLOCKSIZE, tx); bpobj_close(&dle->dle_bpobj); bpobj_decr_empty(dl->dl_os, tx); VERIFY3U(0, ==, bpobj_open(&dle->dle_bpobj, dl->dl_os, obj)); VERIFY3U(0, ==, zap_update_int_key(dl->dl_os, dl->dl_object, dle->dle_mintxg, obj, tx)); } bpobj_enqueue(&dle->dle_bpobj, bp, tx); } static void dle_enqueue_subobj(dsl_deadlist_t *dl, dsl_deadlist_entry_t *dle, uint64_t obj, dmu_tx_t *tx) { ASSERT(MUTEX_HELD(&dl->dl_lock)); if (dle->dle_bpobj.bpo_object != dmu_objset_pool(dl->dl_os)->dp_empty_bpobj) { bpobj_enqueue_subobj(&dle->dle_bpobj, obj, tx); } else { bpobj_close(&dle->dle_bpobj); bpobj_decr_empty(dl->dl_os, tx); VERIFY3U(0, ==, bpobj_open(&dle->dle_bpobj, dl->dl_os, obj)); VERIFY3U(0, ==, zap_update_int_key(dl->dl_os, dl->dl_object, dle->dle_mintxg, obj, tx)); } } void dsl_deadlist_insert(dsl_deadlist_t *dl, const blkptr_t *bp, dmu_tx_t *tx) { dsl_deadlist_entry_t dle_tofind; dsl_deadlist_entry_t *dle; avl_index_t where; if (dl->dl_oldfmt) { bpobj_enqueue(&dl->dl_bpobj, bp, tx); return; } mutex_enter(&dl->dl_lock); dsl_deadlist_load_tree(dl); dmu_buf_will_dirty(dl->dl_dbuf, tx); dl->dl_phys->dl_used += bp_get_dsize_sync(dmu_objset_spa(dl->dl_os), bp); dl->dl_phys->dl_comp += BP_GET_PSIZE(bp); dl->dl_phys->dl_uncomp += BP_GET_UCSIZE(bp); dle_tofind.dle_mintxg = bp->blk_birth; dle = avl_find(&dl->dl_tree, &dle_tofind, &where); if (dle == NULL) dle = avl_nearest(&dl->dl_tree, where, AVL_BEFORE); else dle = AVL_PREV(&dl->dl_tree, dle); dle_enqueue(dl, dle, bp, tx); mutex_exit(&dl->dl_lock); } /* * Insert new key in deadlist, which must be > all current entries. * mintxg is not inclusive. */ void dsl_deadlist_add_key(dsl_deadlist_t *dl, uint64_t mintxg, dmu_tx_t *tx) { uint64_t obj; dsl_deadlist_entry_t *dle; if (dl->dl_oldfmt) return; dle = kmem_alloc(sizeof (*dle), KM_SLEEP); dle->dle_mintxg = mintxg; mutex_enter(&dl->dl_lock); dsl_deadlist_load_tree(dl); obj = bpobj_alloc_empty(dl->dl_os, SPA_OLD_MAXBLOCKSIZE, tx); VERIFY3U(0, ==, bpobj_open(&dle->dle_bpobj, dl->dl_os, obj)); avl_add(&dl->dl_tree, dle); VERIFY3U(0, ==, zap_add_int_key(dl->dl_os, dl->dl_object, mintxg, obj, tx)); mutex_exit(&dl->dl_lock); } /* * Remove this key, merging its entries into the previous key. */ void dsl_deadlist_remove_key(dsl_deadlist_t *dl, uint64_t mintxg, dmu_tx_t *tx) { dsl_deadlist_entry_t dle_tofind; dsl_deadlist_entry_t *dle, *dle_prev; if (dl->dl_oldfmt) return; mutex_enter(&dl->dl_lock); dsl_deadlist_load_tree(dl); dle_tofind.dle_mintxg = mintxg; dle = avl_find(&dl->dl_tree, &dle_tofind, NULL); dle_prev = AVL_PREV(&dl->dl_tree, dle); dle_enqueue_subobj(dl, dle_prev, dle->dle_bpobj.bpo_object, tx); avl_remove(&dl->dl_tree, dle); bpobj_close(&dle->dle_bpobj); kmem_free(dle, sizeof (*dle)); VERIFY3U(0, ==, zap_remove_int(dl->dl_os, dl->dl_object, mintxg, tx)); mutex_exit(&dl->dl_lock); } /* * Walk ds's snapshots to regenerate generate ZAP & AVL. */ static void dsl_deadlist_regenerate(objset_t *os, uint64_t dlobj, uint64_t mrs_obj, dmu_tx_t *tx) { dsl_deadlist_t dl; dsl_pool_t *dp = dmu_objset_pool(os); dsl_deadlist_open(&dl, os, dlobj); if (dl.dl_oldfmt) { dsl_deadlist_close(&dl); return; } while (mrs_obj != 0) { dsl_dataset_t *ds; VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, mrs_obj, FTAG, &ds)); dsl_deadlist_add_key(&dl, dsl_dataset_phys(ds)->ds_prev_snap_txg, tx); mrs_obj = dsl_dataset_phys(ds)->ds_prev_snap_obj; dsl_dataset_rele(ds, FTAG); } dsl_deadlist_close(&dl); } uint64_t dsl_deadlist_clone(dsl_deadlist_t *dl, uint64_t maxtxg, uint64_t mrs_obj, dmu_tx_t *tx) { dsl_deadlist_entry_t *dle; uint64_t newobj; newobj = dsl_deadlist_alloc(dl->dl_os, tx); if (dl->dl_oldfmt) { dsl_deadlist_regenerate(dl->dl_os, newobj, mrs_obj, tx); return (newobj); } mutex_enter(&dl->dl_lock); dsl_deadlist_load_tree(dl); for (dle = avl_first(&dl->dl_tree); dle; dle = AVL_NEXT(&dl->dl_tree, dle)) { uint64_t obj; if (dle->dle_mintxg >= maxtxg) break; obj = bpobj_alloc_empty(dl->dl_os, SPA_OLD_MAXBLOCKSIZE, tx); VERIFY3U(0, ==, zap_add_int_key(dl->dl_os, newobj, dle->dle_mintxg, obj, tx)); } mutex_exit(&dl->dl_lock); return (newobj); } void dsl_deadlist_space(dsl_deadlist_t *dl, uint64_t *usedp, uint64_t *compp, uint64_t *uncompp) { if (dl->dl_oldfmt) { VERIFY3U(0, ==, bpobj_space(&dl->dl_bpobj, usedp, compp, uncompp)); return; } mutex_enter(&dl->dl_lock); *usedp = dl->dl_phys->dl_used; *compp = dl->dl_phys->dl_comp; *uncompp = dl->dl_phys->dl_uncomp; mutex_exit(&dl->dl_lock); } /* * return space used in the range (mintxg, maxtxg]. * Includes maxtxg, does not include mintxg. * mintxg and maxtxg must both be keys in the deadlist (unless maxtxg is * larger than any bp in the deadlist (eg. UINT64_MAX)). */ void dsl_deadlist_space_range(dsl_deadlist_t *dl, uint64_t mintxg, uint64_t maxtxg, uint64_t *usedp, uint64_t *compp, uint64_t *uncompp) { dsl_deadlist_entry_t *dle; dsl_deadlist_entry_t dle_tofind; avl_index_t where; if (dl->dl_oldfmt) { VERIFY3U(0, ==, bpobj_space_range(&dl->dl_bpobj, mintxg, maxtxg, usedp, compp, uncompp)); return; } *usedp = *compp = *uncompp = 0; mutex_enter(&dl->dl_lock); dsl_deadlist_load_tree(dl); dle_tofind.dle_mintxg = mintxg; dle = avl_find(&dl->dl_tree, &dle_tofind, &where); /* * If we don't find this mintxg, there shouldn't be anything * after it either. */ ASSERT(dle != NULL || avl_nearest(&dl->dl_tree, where, AVL_AFTER) == NULL); for (; dle && dle->dle_mintxg < maxtxg; dle = AVL_NEXT(&dl->dl_tree, dle)) { uint64_t used, comp, uncomp; VERIFY3U(0, ==, bpobj_space(&dle->dle_bpobj, &used, &comp, &uncomp)); *usedp += used; *compp += comp; *uncompp += uncomp; } mutex_exit(&dl->dl_lock); } static void dsl_deadlist_insert_bpobj(dsl_deadlist_t *dl, uint64_t obj, uint64_t birth, dmu_tx_t *tx) { dsl_deadlist_entry_t dle_tofind; dsl_deadlist_entry_t *dle; avl_index_t where; uint64_t used, comp, uncomp; bpobj_t bpo; ASSERT(MUTEX_HELD(&dl->dl_lock)); VERIFY3U(0, ==, bpobj_open(&bpo, dl->dl_os, obj)); VERIFY3U(0, ==, bpobj_space(&bpo, &used, &comp, &uncomp)); bpobj_close(&bpo); dsl_deadlist_load_tree(dl); dmu_buf_will_dirty(dl->dl_dbuf, tx); dl->dl_phys->dl_used += used; dl->dl_phys->dl_comp += comp; dl->dl_phys->dl_uncomp += uncomp; dle_tofind.dle_mintxg = birth; dle = avl_find(&dl->dl_tree, &dle_tofind, &where); if (dle == NULL) dle = avl_nearest(&dl->dl_tree, where, AVL_BEFORE); dle_enqueue_subobj(dl, dle, obj, tx); } static int dsl_deadlist_insert_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_deadlist_t *dl = arg; dsl_deadlist_insert(dl, bp, tx); return (0); } /* * Merge the deadlist pointed to by 'obj' into dl. obj will be left as * an empty deadlist. */ void dsl_deadlist_merge(dsl_deadlist_t *dl, uint64_t obj, dmu_tx_t *tx) { zap_cursor_t zc; zap_attribute_t za; dmu_buf_t *bonus; dsl_deadlist_phys_t *dlp; dmu_object_info_t doi; VERIFY3U(0, ==, dmu_object_info(dl->dl_os, obj, &doi)); if (doi.doi_type == DMU_OT_BPOBJ) { bpobj_t bpo; VERIFY3U(0, ==, bpobj_open(&bpo, dl->dl_os, obj)); VERIFY3U(0, ==, bpobj_iterate(&bpo, dsl_deadlist_insert_cb, dl, tx)); bpobj_close(&bpo); return; } mutex_enter(&dl->dl_lock); for (zap_cursor_init(&zc, dl->dl_os, obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { - uint64_t mintxg = strtonum(za.za_name, NULL); + uint64_t mintxg = zfs_strtonum(za.za_name, NULL); dsl_deadlist_insert_bpobj(dl, za.za_first_integer, mintxg, tx); VERIFY3U(0, ==, zap_remove_int(dl->dl_os, obj, mintxg, tx)); } zap_cursor_fini(&zc); VERIFY3U(0, ==, dmu_bonus_hold(dl->dl_os, obj, FTAG, &bonus)); dlp = bonus->db_data; dmu_buf_will_dirty(bonus, tx); bzero(dlp, sizeof (*dlp)); dmu_buf_rele(bonus, FTAG); mutex_exit(&dl->dl_lock); } /* * Remove entries on dl that are >= mintxg, and put them on the bpobj. */ void dsl_deadlist_move_bpobj(dsl_deadlist_t *dl, bpobj_t *bpo, uint64_t mintxg, dmu_tx_t *tx) { dsl_deadlist_entry_t dle_tofind; dsl_deadlist_entry_t *dle; avl_index_t where; ASSERT(!dl->dl_oldfmt); mutex_enter(&dl->dl_lock); dmu_buf_will_dirty(dl->dl_dbuf, tx); dsl_deadlist_load_tree(dl); dle_tofind.dle_mintxg = mintxg; dle = avl_find(&dl->dl_tree, &dle_tofind, &where); if (dle == NULL) dle = avl_nearest(&dl->dl_tree, where, AVL_AFTER); while (dle) { uint64_t used, comp, uncomp; dsl_deadlist_entry_t *dle_next; bpobj_enqueue_subobj(bpo, dle->dle_bpobj.bpo_object, tx); VERIFY3U(0, ==, bpobj_space(&dle->dle_bpobj, &used, &comp, &uncomp)); ASSERT3U(dl->dl_phys->dl_used, >=, used); ASSERT3U(dl->dl_phys->dl_comp, >=, comp); ASSERT3U(dl->dl_phys->dl_uncomp, >=, uncomp); dl->dl_phys->dl_used -= used; dl->dl_phys->dl_comp -= comp; dl->dl_phys->dl_uncomp -= uncomp; VERIFY3U(0, ==, zap_remove_int(dl->dl_os, dl->dl_object, dle->dle_mintxg, tx)); dle_next = AVL_NEXT(&dl->dl_tree, dle); avl_remove(&dl->dl_tree, dle); bpobj_close(&dle->dle_bpobj); kmem_free(dle, sizeof (*dle)); dle = dle_next; } mutex_exit(&dl->dl_lock); } Index: vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_scan.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_scan.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_scan.c (revision 319948) @@ -1,1898 +1,1898 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2016 Gary Mills * Copyright (c) 2011, 2016 by Delphix. All rights reserved. * Copyright 2017 Joyent, Inc. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef _KERNEL #include #endif typedef int (scan_cb_t)(dsl_pool_t *, const blkptr_t *, const zbookmark_phys_t *); static scan_cb_t dsl_scan_scrub_cb; static void dsl_scan_cancel_sync(void *, dmu_tx_t *); static void dsl_scan_sync_state(dsl_scan_t *, dmu_tx_t *); static boolean_t dsl_scan_restarting(dsl_scan_t *, dmu_tx_t *); int zfs_top_maxinflight = 32; /* maximum I/Os per top-level */ int zfs_resilver_delay = 2; /* number of ticks to delay resilver */ int zfs_scrub_delay = 4; /* number of ticks to delay scrub */ int zfs_scan_idle = 50; /* idle window in clock ticks */ int zfs_scan_min_time_ms = 1000; /* min millisecs to scrub per txg */ int zfs_free_min_time_ms = 1000; /* min millisecs to free per txg */ int zfs_resilver_min_time_ms = 3000; /* min millisecs to resilver per txg */ boolean_t zfs_no_scrub_io = B_FALSE; /* set to disable scrub i/o */ boolean_t zfs_no_scrub_prefetch = B_FALSE; /* set to disable scrub prefetch */ enum ddt_class zfs_scrub_ddt_class_max = DDT_CLASS_DUPLICATE; int dsl_scan_delay_completion = B_FALSE; /* set to delay scan completion */ /* max number of blocks to free in a single TXG */ uint64_t zfs_free_max_blocks = UINT64_MAX; #define DSL_SCAN_IS_SCRUB_RESILVER(scn) \ ((scn)->scn_phys.scn_func == POOL_SCAN_SCRUB || \ (scn)->scn_phys.scn_func == POOL_SCAN_RESILVER) extern int zfs_txg_timeout; /* * Enable/disable the processing of the free_bpobj object. */ boolean_t zfs_free_bpobj_enabled = B_TRUE; /* the order has to match pool_scan_type */ static scan_cb_t *scan_funcs[POOL_SCAN_FUNCS] = { NULL, dsl_scan_scrub_cb, /* POOL_SCAN_SCRUB */ dsl_scan_scrub_cb, /* POOL_SCAN_RESILVER */ }; int dsl_scan_init(dsl_pool_t *dp, uint64_t txg) { int err; dsl_scan_t *scn; spa_t *spa = dp->dp_spa; uint64_t f; scn = dp->dp_scan = kmem_zalloc(sizeof (dsl_scan_t), KM_SLEEP); scn->scn_dp = dp; /* * It's possible that we're resuming a scan after a reboot so * make sure that the scan_async_destroying flag is initialized * appropriately. */ ASSERT(!scn->scn_async_destroying); scn->scn_async_destroying = spa_feature_is_active(dp->dp_spa, SPA_FEATURE_ASYNC_DESTROY); err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_func", sizeof (uint64_t), 1, &f); if (err == 0) { /* * There was an old-style scrub in progress. Restart a * new-style scrub from the beginning. */ scn->scn_restart_txg = txg; zfs_dbgmsg("old-style scrub was in progress; " "restarting new-style scrub in txg %llu", scn->scn_restart_txg); /* * Load the queue obj from the old location so that it * can be freed by dsl_scan_done(). */ (void) zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, "scrub_queue", sizeof (uint64_t), 1, &scn->scn_phys.scn_queue_obj); } else { err = zap_lookup(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys); if (err == ENOENT) return (0); else if (err) return (err); if (scn->scn_phys.scn_state == DSS_SCANNING && spa_prev_software_version(dp->dp_spa) < SPA_VERSION_SCAN) { /* * A new-type scrub was in progress on an old * pool, and the pool was accessed by old * software. Restart from the beginning, since * the old software may have changed the pool in * the meantime. */ scn->scn_restart_txg = txg; zfs_dbgmsg("new-style scrub was modified " "by old software; restarting in txg %llu", scn->scn_restart_txg); } } spa_scan_stat_init(spa); return (0); } void dsl_scan_fini(dsl_pool_t *dp) { if (dp->dp_scan) { kmem_free(dp->dp_scan, sizeof (dsl_scan_t)); dp->dp_scan = NULL; } } /* ARGSUSED */ static int dsl_scan_setup_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state == DSS_SCANNING) return (SET_ERROR(EBUSY)); return (0); } static void dsl_scan_setup_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; pool_scan_func_t *funcp = arg; dmu_object_type_t ot = 0; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; ASSERT(scn->scn_phys.scn_state != DSS_SCANNING); ASSERT(*funcp > POOL_SCAN_NONE && *funcp < POOL_SCAN_FUNCS); bzero(&scn->scn_phys, sizeof (scn->scn_phys)); scn->scn_phys.scn_func = *funcp; scn->scn_phys.scn_state = DSS_SCANNING; scn->scn_phys.scn_min_txg = 0; scn->scn_phys.scn_max_txg = tx->tx_txg; scn->scn_phys.scn_ddt_class_max = DDT_CLASSES - 1; /* the entire DDT */ scn->scn_phys.scn_start_time = gethrestime_sec(); scn->scn_phys.scn_errors = 0; scn->scn_phys.scn_to_examine = spa->spa_root_vdev->vdev_stat.vs_alloc; scn->scn_restart_txg = 0; scn->scn_done_txg = 0; spa_scan_stat_init(spa); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { scn->scn_phys.scn_ddt_class_max = zfs_scrub_ddt_class_max; /* rewrite all disk labels */ vdev_config_dirty(spa->spa_root_vdev); if (vdev_resilver_needed(spa->spa_root_vdev, &scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) { spa_event_notify(spa, NULL, NULL, ESC_ZFS_RESILVER_START); } else { spa_event_notify(spa, NULL, NULL, ESC_ZFS_SCRUB_START); } spa->spa_scrub_started = B_TRUE; /* * If this is an incremental scrub, limit the DDT scrub phase * to just the auto-ditto class (for correctness); the rest * of the scrub should go faster using top-down pruning. */ if (scn->scn_phys.scn_min_txg > TXG_INITIAL) scn->scn_phys.scn_ddt_class_max = DDT_CLASS_DITTO; } /* back to the generic stuff */ if (dp->dp_blkstats == NULL) { dp->dp_blkstats = kmem_alloc(sizeof (zfs_all_blkstats_t), KM_SLEEP); } bzero(dp->dp_blkstats, sizeof (zfs_all_blkstats_t)); if (spa_version(spa) < SPA_VERSION_DSL_SCRUB) ot = DMU_OT_ZAP_OTHER; scn->scn_phys.scn_queue_obj = zap_create(dp->dp_meta_objset, ot ? ot : DMU_OT_SCAN_QUEUE, DMU_OT_NONE, 0, tx); dsl_scan_sync_state(scn, tx); spa_history_log_internal(spa, "scan setup", tx, "func=%u mintxg=%llu maxtxg=%llu", *funcp, scn->scn_phys.scn_min_txg, scn->scn_phys.scn_max_txg); } /* ARGSUSED */ static void dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx) { static const char *old_names[] = { "scrub_bookmark", "scrub_ddt_bookmark", "scrub_ddt_class_max", "scrub_queue", "scrub_min_txg", "scrub_max_txg", "scrub_func", "scrub_errors", NULL }; dsl_pool_t *dp = scn->scn_dp; spa_t *spa = dp->dp_spa; int i; /* Remove any remnants of an old-style scrub. */ for (i = 0; old_names[i]; i++) { (void) zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, old_names[i], tx); } if (scn->scn_phys.scn_queue_obj != 0) { VERIFY(0 == dmu_object_free(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, tx)); scn->scn_phys.scn_queue_obj = 0; } /* * If we were "restarted" from a stopped state, don't bother * with anything else. */ if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (complete) scn->scn_phys.scn_state = DSS_FINISHED; else scn->scn_phys.scn_state = DSS_CANCELED; if (dsl_scan_restarting(scn, tx)) spa_history_log_internal(spa, "scan aborted, restarting", tx, "errors=%llu", spa_get_errlog_size(spa)); else if (!complete) spa_history_log_internal(spa, "scan cancelled", tx, "errors=%llu", spa_get_errlog_size(spa)); else spa_history_log_internal(spa, "scan done", tx, "errors=%llu", spa_get_errlog_size(spa)); if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); spa->spa_scrub_started = B_FALSE; spa->spa_scrub_active = B_FALSE; /* * If the scrub/resilver completed, update all DTLs to * reflect this. Whether it succeeded or not, vacate * all temporary scrub DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, tx->tx_txg, complete ? scn->scn_phys.scn_max_txg : 0, B_TRUE); if (complete) { spa_event_notify(spa, NULL, NULL, scn->scn_phys.scn_min_txg ? ESC_ZFS_RESILVER_FINISH : ESC_ZFS_SCRUB_FINISH); } spa_errlog_rotate(spa); /* * We may have finished replacing a device. * Let the async thread assess this and handle the detach. */ spa_async_request(spa, SPA_ASYNC_RESILVER_DONE); } scn->scn_phys.scn_end_time = gethrestime_sec(); } /* ARGSUSED */ static int dsl_scan_cancel_check(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; if (scn->scn_phys.scn_state != DSS_SCANNING) return (SET_ERROR(ENOENT)); return (0); } /* ARGSUSED */ static void dsl_scan_cancel_sync(void *arg, dmu_tx_t *tx) { dsl_scan_t *scn = dmu_tx_pool(tx)->dp_scan; dsl_scan_done(scn, B_FALSE, tx); dsl_scan_sync_state(scn, tx); } int dsl_scan_cancel(dsl_pool_t *dp) { return (dsl_sync_task(spa_name(dp->dp_spa), dsl_scan_cancel_check, dsl_scan_cancel_sync, NULL, 3, ZFS_SPACE_CHECK_RESERVED)); } static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx); static void dsl_scan_visitdnode(dsl_scan_t *, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx); void dsl_free(dsl_pool_t *dp, uint64_t txg, const blkptr_t *bp) { zio_free(dp->dp_spa, txg, bp); } void dsl_free_sync(zio_t *pio, dsl_pool_t *dp, uint64_t txg, const blkptr_t *bpp) { ASSERT(dsl_pool_sync_context(dp)); zio_nowait(zio_free_sync(pio, dp->dp_spa, txg, bpp, pio->io_flags)); } static uint64_t dsl_scan_ds_maxtxg(dsl_dataset_t *ds) { uint64_t smt = ds->ds_dir->dd_pool->dp_scan->scn_phys.scn_max_txg; if (ds->ds_is_snapshot) return (MIN(smt, dsl_dataset_phys(ds)->ds_creation_txg)); return (smt); } static void dsl_scan_sync_state(dsl_scan_t *scn, dmu_tx_t *tx) { VERIFY0(zap_update(scn->scn_dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_SCAN, sizeof (uint64_t), SCAN_PHYS_NUMINTS, &scn->scn_phys, tx)); } extern int zfs_vdev_async_write_active_min_dirty_percent; static boolean_t dsl_scan_check_pause(dsl_scan_t *scn, const zbookmark_phys_t *zb) { /* we never skip user/group accounting objects */ if (zb && (int64_t)zb->zb_object < 0) return (B_FALSE); if (scn->scn_pausing) return (B_TRUE); /* we're already pausing */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark)) return (B_FALSE); /* we're resuming */ /* We only know how to resume from level-0 blocks. */ if (zb && zb->zb_level != 0) return (B_FALSE); /* * We pause if: * - we have scanned for the maximum time: an entire txg * timeout (default 5 sec) * or * - we have scanned for at least the minimum time (default 1 sec * for scrub, 3 sec for resilver), and either we have sufficient * dirty data that we are starting to write more quickly * (default 30%), or someone is explicitly waiting for this txg * to complete. * or * - the spa is shutting down because this pool is being exported * or the machine is rebooting. */ int mintime = (scn->scn_phys.scn_func == POOL_SCAN_RESILVER) ? zfs_resilver_min_time_ms : zfs_scan_min_time_ms; uint64_t elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; int dirty_pct = scn->scn_dp->dp_dirty_total * 100 / zfs_dirty_data_max; if (elapsed_nanosecs / NANOSEC >= zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > mintime && (txg_sync_waiting(scn->scn_dp) || dirty_pct >= zfs_vdev_async_write_active_min_dirty_percent)) || spa_shutting_down(scn->scn_dp->dp_spa)) { if (zb) { dprintf("pausing at bookmark %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); scn->scn_phys.scn_bookmark = *zb; } dprintf("pausing at DDT bookmark %llx/%llx/%llx/%llx\n", (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); scn->scn_pausing = B_TRUE; return (B_TRUE); } return (B_FALSE); } typedef struct zil_scan_arg { dsl_pool_t *zsa_dp; zil_header_t *zsa_zh; } zil_scan_arg_t; /* ARGSUSED */ static int dsl_scan_zil_block(zilog_t *zilog, blkptr_t *bp, void *arg, uint64_t claim_txg) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; zbookmark_phys_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * One block ("stubby") can be allocated a long time ago; we * want to visit that one because it has been allocated * (on-disk) even if it hasn't been claimed (even though for * scrub there's nothing to do to it). */ if (claim_txg == 0 && bp->blk_birth >= spa_first_txg(dp->dp_spa)) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], ZB_ZIL_OBJECT, ZB_ZIL_LEVEL, bp->blk_cksum.zc_word[ZIL_ZC_SEQ]); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); return (0); } /* ARGSUSED */ static int dsl_scan_zil_record(zilog_t *zilog, lr_t *lrc, void *arg, uint64_t claim_txg) { if (lrc->lrc_txtype == TX_WRITE) { zil_scan_arg_t *zsa = arg; dsl_pool_t *dp = zsa->zsa_dp; dsl_scan_t *scn = dp->dp_scan; zil_header_t *zh = zsa->zsa_zh; lr_write_t *lr = (lr_write_t *)lrc; blkptr_t *bp = &lr->lr_blkptr; zbookmark_phys_t zb; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return (0); /* * birth can be < claim_txg if this record's txg is * already txg sync'ed (but this log block contains * other records that are not synced) */ if (claim_txg == 0 || bp->blk_birth < claim_txg) return (0); SET_BOOKMARK(&zb, zh->zh_log.blk_cksum.zc_word[ZIL_ZC_OBJSET], lr->lr_foid, ZB_ZIL_LEVEL, lr->lr_offset / BP_GET_LSIZE(bp)); VERIFY(0 == scan_funcs[scn->scn_phys.scn_func](dp, bp, &zb)); } return (0); } static void dsl_scan_zil(dsl_pool_t *dp, zil_header_t *zh) { uint64_t claim_txg = zh->zh_claim_txg; zil_scan_arg_t zsa = { dp, zh }; zilog_t *zilog; /* * We only want to visit blocks that have been claimed but not yet * replayed (or, in read-only mode, blocks that *would* be claimed). */ if (claim_txg == 0 && spa_writeable(dp->dp_spa)) return; zilog = zil_alloc(dp->dp_meta_objset, zh); (void) zil_parse(zilog, dsl_scan_zil_block, dsl_scan_zil_record, &zsa, claim_txg); zil_free(zilog); } /* ARGSUSED */ static void dsl_scan_prefetch(dsl_scan_t *scn, arc_buf_t *buf, blkptr_t *bp, uint64_t objset, uint64_t object, uint64_t blkid) { zbookmark_phys_t czb; arc_flags_t flags = ARC_FLAG_NOWAIT | ARC_FLAG_PREFETCH; if (zfs_no_scrub_prefetch) return; if (BP_IS_HOLE(bp) || bp->blk_birth <= scn->scn_phys.scn_min_txg || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_DNODE)) return; SET_BOOKMARK(&czb, objset, object, BP_GET_LEVEL(bp), blkid); (void) arc_read(scn->scn_zio_root, scn->scn_dp->dp_spa, bp, NULL, NULL, ZIO_PRIORITY_ASYNC_READ, ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD, &flags, &czb); } static boolean_t dsl_scan_check_resume(dsl_scan_t *scn, const dnode_phys_t *dnp, const zbookmark_phys_t *zb) { /* * We never skip over user/group accounting objects (obj<0) */ if (!ZB_IS_ZERO(&scn->scn_phys.scn_bookmark) && (int64_t)zb->zb_object >= 0) { /* * If we already visited this bp & everything below (in * a prior txg sync), don't bother doing it again. */ if (zbookmark_subtree_completed(dnp, zb, &scn->scn_phys.scn_bookmark)) return (B_TRUE); /* * If we found the block we're trying to resume from, or * we went past it to a different object, zero it out to * indicate that it's OK to start checking for pausing * again. */ if (bcmp(zb, &scn->scn_phys.scn_bookmark, sizeof (*zb)) == 0 || zb->zb_object > scn->scn_phys.scn_bookmark.zb_object) { dprintf("resuming at %llx/%llx/%llx/%llx\n", (longlong_t)zb->zb_objset, (longlong_t)zb->zb_object, (longlong_t)zb->zb_level, (longlong_t)zb->zb_blkid); bzero(&scn->scn_phys.scn_bookmark, sizeof (*zb)); } } return (B_FALSE); } /* * Return nonzero on i/o error. * Return new buf to write out in *bufp. */ static int dsl_scan_recurse(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, const blkptr_t *bp, const zbookmark_phys_t *zb, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; int zio_flags = ZIO_FLAG_CANFAIL | ZIO_FLAG_SCAN_THREAD; int err; if (BP_GET_LEVEL(bp) > 0) { arc_flags_t flags = ARC_FLAG_WAIT; int i; blkptr_t *cbp; int epb = BP_GET_LSIZE(bp) >> SPA_BLKPTRSHIFT; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { dsl_scan_prefetch(scn, buf, cbp, zb->zb_objset, zb->zb_object, zb->zb_blkid * epb + i); } for (i = 0, cbp = buf->b_data; i < epb; i++, cbp++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, zb->zb_objset, zb->zb_object, zb->zb_level - 1, zb->zb_blkid * epb + i); dsl_scan_visitbp(cbp, &czb, dnp, ds, scn, ostype, tx); } arc_buf_destroy(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_DNODE) { arc_flags_t flags = ARC_FLAG_WAIT; dnode_phys_t *cdnp; int i, j; int epb = BP_GET_LSIZE(bp) >> DNODE_SHIFT; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } for (i = 0, cdnp = buf->b_data; i < epb; i++, cdnp++) { for (j = 0; j < cdnp->dn_nblkptr; j++) { blkptr_t *cbp = &cdnp->dn_blkptr[j]; dsl_scan_prefetch(scn, buf, cbp, zb->zb_objset, zb->zb_blkid * epb + i, j); } } for (i = 0, cdnp = buf->b_data; i < epb; i++, cdnp++) { dsl_scan_visitdnode(scn, ds, ostype, cdnp, zb->zb_blkid * epb + i, tx); } arc_buf_destroy(buf, &buf); } else if (BP_GET_TYPE(bp) == DMU_OT_OBJSET) { arc_flags_t flags = ARC_FLAG_WAIT; objset_phys_t *osp; arc_buf_t *buf; err = arc_read(NULL, dp->dp_spa, bp, arc_getbuf_func, &buf, ZIO_PRIORITY_ASYNC_READ, zio_flags, &flags, zb); if (err) { scn->scn_phys.scn_errors++; return (err); } osp = buf->b_data; dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_meta_dnode, DMU_META_DNODE_OBJECT, tx); if (OBJSET_BUF_HAS_USERUSED(buf)) { /* * We also always visit user/group accounting * objects, and never skip them, even if we are * pausing. This is necessary so that the space * deltas from this txg get integrated. */ dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_groupused_dnode, DMU_GROUPUSED_OBJECT, tx); dsl_scan_visitdnode(scn, ds, osp->os_type, &osp->os_userused_dnode, DMU_USERUSED_OBJECT, tx); } arc_buf_destroy(buf, &buf); } return (0); } static void dsl_scan_visitdnode(dsl_scan_t *scn, dsl_dataset_t *ds, dmu_objset_type_t ostype, dnode_phys_t *dnp, uint64_t object, dmu_tx_t *tx) { int j; for (j = 0; j < dnp->dn_nblkptr; j++) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, dnp->dn_nlevels - 1, j); dsl_scan_visitbp(&dnp->dn_blkptr[j], &czb, dnp, ds, scn, ostype, tx); } if (dnp->dn_flags & DNODE_FLAG_SPILL_BLKPTR) { zbookmark_phys_t czb; SET_BOOKMARK(&czb, ds ? ds->ds_object : 0, object, 0, DMU_SPILL_BLKID); dsl_scan_visitbp(&dnp->dn_spill, &czb, dnp, ds, scn, ostype, tx); } } /* * The arguments are in this order because mdb can only print the * first 5; we want them to be useful. */ static void dsl_scan_visitbp(blkptr_t *bp, const zbookmark_phys_t *zb, dnode_phys_t *dnp, dsl_dataset_t *ds, dsl_scan_t *scn, dmu_objset_type_t ostype, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; arc_buf_t *buf = NULL; blkptr_t bp_toread = *bp; /* ASSERT(pbuf == NULL || arc_released(pbuf)); */ if (dsl_scan_check_pause(scn, zb)) return; if (dsl_scan_check_resume(scn, dnp, zb)) return; if (BP_IS_HOLE(bp)) return; scn->scn_visited_this_txg++; dprintf_bp(bp, "visiting ds=%p/%llu zb=%llx/%llx/%llx/%llx bp=%p", ds, ds ? ds->ds_object : 0, zb->zb_objset, zb->zb_object, zb->zb_level, zb->zb_blkid, bp); if (bp->blk_birth <= scn->scn_phys.scn_cur_min_txg) return; if (dsl_scan_recurse(scn, ds, ostype, dnp, &bp_toread, zb, tx) != 0) return; /* * If dsl_scan_ddt() has already visited this block, it will have * already done any translations or scrubbing, so don't call the * callback again. */ if (ddt_class_contains(dp->dp_spa, scn->scn_phys.scn_ddt_class_max, bp)) { ASSERT(buf == NULL); return; } /* * If this block is from the future (after cur_max_txg), then we * are doing this on behalf of a deleted snapshot, and we will * revisit the future block on the next pass of this dataset. * Don't scan it now unless we need to because something * under it was modified. */ if (BP_PHYSICAL_BIRTH(bp) <= scn->scn_phys.scn_cur_max_txg) { scan_funcs[scn->scn_phys.scn_func](dp, bp, zb); } } static void dsl_scan_visit_rootbp(dsl_scan_t *scn, dsl_dataset_t *ds, blkptr_t *bp, dmu_tx_t *tx) { zbookmark_phys_t zb; SET_BOOKMARK(&zb, ds ? ds->ds_object : DMU_META_OBJSET, ZB_ROOT_OBJECT, ZB_ROOT_LEVEL, ZB_ROOT_BLKID); dsl_scan_visitbp(bp, &zb, NULL, ds, scn, DMU_OST_NONE, tx); dprintf_ds(ds, "finished scan%s", ""); } void dsl_scan_ds_destroyed(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { if (ds->ds_is_snapshot) { /* * Note: * - scn_cur_{min,max}_txg stays the same. * - Setting the flag is not really necessary if * scn_cur_max_txg == scn_max_txg, because there * is nothing after this snapshot that we care * about. However, we set it anyway and then * ignore it when we retraverse it in * dsl_scan_visitds(). */ scn->scn_phys.scn_bookmark.zb_objset = dsl_dataset_phys(ds)->ds_next_snap_obj; zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)-> ds_next_snap_obj); scn->scn_phys.scn_flags |= DSF_VISIT_DS_AGAIN; } else { SET_BOOKMARK(&scn->scn_phys.scn_bookmark, ZB_DESTROYED_OBJSET, 0, 0, 0); zfs_dbgmsg("destroying ds %llu; currently traversing; " "reset bookmark to -1,0,0,0", (u_longlong_t)ds->ds_object); } } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { ASSERT3U(dsl_dataset_phys(ds)->ds_num_children, <=, 1); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); if (ds->ds_is_snapshot) { /* * We keep the same mintxg; it could be > * ds_creation_txg if the previous snapshot was * deleted too. */ VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_next_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("destroying ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)-> ds_next_snap_obj); } else { zfs_dbgmsg("destroying ds %llu; in queue; removing", (u_longlong_t)ds->ds_object); } } /* * dsl_scan_sync() should be called after this, and should sync * out our changed state, but just to be safe, do it here. */ dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_snapshotted(dsl_dataset_t *ds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; ASSERT(dsl_dataset_phys(ds)->ds_prev_snap_obj != 0); if (scn->scn_phys.scn_bookmark.zb_objset == ds->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = dsl_dataset_phys(ds)->ds_prev_snap_obj; zfs_dbgmsg("snapshotting ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)->ds_prev_snap_obj); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, &mintxg) == 0) { VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, tx)); VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_prev_snap_obj, mintxg, tx) == 0); zfs_dbgmsg("snapshotting ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds->ds_object, (u_longlong_t)dsl_dataset_phys(ds)->ds_prev_snap_obj); } dsl_scan_sync_state(scn, tx); } void dsl_scan_ds_clone_swapped(dsl_dataset_t *ds1, dsl_dataset_t *ds2, dmu_tx_t *tx) { dsl_pool_t *dp = ds1->ds_dir->dd_pool; dsl_scan_t *scn = dp->dp_scan; uint64_t mintxg; if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_phys.scn_bookmark.zb_objset == ds1->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds2->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (scn->scn_phys.scn_bookmark.zb_objset == ds2->ds_object) { scn->scn_phys.scn_bookmark.zb_objset = ds1->ds_object; zfs_dbgmsg("clone_swap ds %llu; currently traversing; " "reset zb_objset to %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, &mintxg) == 0) { int err; ASSERT3U(mintxg, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, tx)); err = zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, mintxg, tx); VERIFY(err == 0 || err == EEXIST); if (err == EEXIST) { /* Both were there to begin with */ VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); } zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds1->ds_object, (u_longlong_t)ds2->ds_object); } else if (zap_lookup_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, &mintxg) == 0) { ASSERT3U(mintxg, ==, dsl_dataset_phys(ds1)->ds_prev_snap_txg); ASSERT3U(mintxg, ==, dsl_dataset_phys(ds2)->ds_prev_snap_txg); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds2->ds_object, tx)); VERIFY(0 == zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds1->ds_object, mintxg, tx)); zfs_dbgmsg("clone_swap ds %llu; in queue; " "replacing with %llu", (u_longlong_t)ds2->ds_object, (u_longlong_t)ds1->ds_object); } dsl_scan_sync_state(scn, tx); } struct enqueue_clones_arg { dmu_tx_t *tx; uint64_t originobj; }; /* ARGSUSED */ static int enqueue_clones_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { struct enqueue_clones_arg *eca = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; if (dsl_dir_phys(hds->ds_dir)->dd_origin_obj != eca->originobj) return (0); err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (dsl_dataset_phys(ds)->ds_prev_snap_obj != eca->originobj) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &prev); dsl_dataset_rele(ds, FTAG); if (err) return (err); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, dsl_dataset_phys(ds)->ds_prev_snap_txg, eca->tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } static void dsl_scan_visitds(dsl_scan_t *scn, uint64_t dsobj, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; dsl_dataset_t *ds; objset_t *os; VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (scn->scn_phys.scn_cur_min_txg >= scn->scn_phys.scn_max_txg) { /* * This can happen if this snapshot was created after the * scan started, and we already completed a previous snapshot * that was created after the scan started. This snapshot * only references blocks with: * * birth < our ds_creation_txg * cur_min_txg is no less than ds_creation_txg. * We have already visited these blocks. * or * birth > scn_max_txg * The scan requested not to visit these blocks. * * Subsequent snapshots (and clones) can reference our * blocks, or blocks with even higher birth times. * Therefore we do not need to visit them either, * so we do not add them to the work queue. * * Note that checking for cur_min_txg >= cur_max_txg * is not sufficient, because in that case we may need to * visit subsequent snapshots. This happens when min_txg > 0, * which raises cur_min_txg. In this case we will visit * this dataset but skip all of its blocks, because the * rootbp's birth time is < cur_min_txg. Then we will * add the next snapshots/clones to the work queue. */ char *dsname = kmem_alloc(MAXNAMELEN, KM_SLEEP); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanning dataset %llu (%s) is unnecessary because " "cur_min_txg (%llu) >= max_txg (%llu)", dsobj, dsname, scn->scn_phys.scn_cur_min_txg, scn->scn_phys.scn_max_txg); kmem_free(dsname, MAXNAMELEN); goto out; } if (dmu_objset_from_ds(ds, &os)) goto out; /* * Only the ZIL in the head (non-snapshot) is valid. Even though * snapshots can have ZIL block pointers (which may be the same * BP as in the head), they must be ignored. So we traverse the * ZIL here, rather than in scan_recurse(), because the regular * snapshot block-sharing rules don't apply to it. */ if (DSL_SCAN_IS_SCRUB_RESILVER(scn) && !ds->ds_is_snapshot) dsl_scan_zil(dp, &os->os_zil_header); /* * Iterate over the bps in this ds. */ dmu_buf_will_dirty(ds->ds_dbuf, tx); rrw_enter(&ds->ds_bp_rwlock, RW_READER, FTAG); dsl_scan_visit_rootbp(scn, ds, &dsl_dataset_phys(ds)->ds_bp, tx); rrw_exit(&ds->ds_bp_rwlock, FTAG); char *dsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); dsl_dataset_name(ds, dsname); zfs_dbgmsg("scanned dataset %llu (%s) with min=%llu max=%llu; " "pausing=%u", (longlong_t)dsobj, dsname, (longlong_t)scn->scn_phys.scn_cur_min_txg, (longlong_t)scn->scn_phys.scn_cur_max_txg, (int)scn->scn_pausing); kmem_free(dsname, ZFS_MAX_DATASET_NAME_LEN); if (scn->scn_pausing) goto out; /* * We've finished this pass over this dataset. */ /* * If we did not completely visit this dataset, do another pass. */ if (scn->scn_phys.scn_flags & DSF_VISIT_DS_AGAIN) { zfs_dbgmsg("incomplete pass; visiting again"); scn->scn_phys.scn_flags &= ~DSF_VISIT_DS_AGAIN; VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, scn->scn_phys.scn_cur_max_txg, tx) == 0); goto out; } /* * Add descendent datasets to work queue. */ if (dsl_dataset_phys(ds)->ds_next_snap_obj != 0) { VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_next_snap_obj, dsl_dataset_phys(ds)->ds_creation_txg, tx) == 0); } if (dsl_dataset_phys(ds)->ds_num_children > 1) { boolean_t usenext = B_FALSE; if (dsl_dataset_phys(ds)->ds_next_clones_obj != 0) { uint64_t count; /* * A bug in a previous version of the code could * cause upgrade_clones_cb() to not set * ds_next_snap_obj when it should, leading to a * missing entry. Therefore we can only use the * next_clones_obj when its count is correct. */ int err = zap_count(dp->dp_meta_objset, dsl_dataset_phys(ds)->ds_next_clones_obj, &count); if (err == 0 && count == dsl_dataset_phys(ds)->ds_num_children - 1) usenext = B_TRUE; } if (usenext) { VERIFY0(zap_join_key(dp->dp_meta_objset, dsl_dataset_phys(ds)->ds_next_clones_obj, scn->scn_phys.scn_queue_obj, dsl_dataset_phys(ds)->ds_creation_txg, tx)); } else { struct enqueue_clones_arg eca; eca.tx = tx; eca.originobj = ds->ds_object; VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_clones_cb, &eca, DS_FIND_CHILDREN)); } } out: dsl_dataset_rele(ds, FTAG); } /* ARGSUSED */ static int enqueue_cb(dsl_pool_t *dp, dsl_dataset_t *hds, void *arg) { dmu_tx_t *tx = arg; dsl_dataset_t *ds; int err; dsl_scan_t *scn = dp->dp_scan; err = dsl_dataset_hold_obj(dp, hds->ds_object, FTAG, &ds); if (err) return (err); while (dsl_dataset_phys(ds)->ds_prev_snap_obj != 0) { dsl_dataset_t *prev; err = dsl_dataset_hold_obj(dp, dsl_dataset_phys(ds)->ds_prev_snap_obj, FTAG, &prev); if (err) { dsl_dataset_rele(ds, FTAG); return (err); } /* * If this is a clone, we don't need to worry about it for now. */ if (dsl_dataset_phys(prev)->ds_next_snap_obj != ds->ds_object) { dsl_dataset_rele(ds, FTAG); dsl_dataset_rele(prev, FTAG); return (0); } dsl_dataset_rele(ds, FTAG); ds = prev; } VERIFY(zap_add_int_key(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, ds->ds_object, dsl_dataset_phys(ds)->ds_prev_snap_txg, tx) == 0); dsl_dataset_rele(ds, FTAG); return (0); } /* * Scrub/dedup interaction. * * If there are N references to a deduped block, we don't want to scrub it * N times -- ideally, we should scrub it exactly once. * * We leverage the fact that the dde's replication class (enum ddt_class) * is ordered from highest replication class (DDT_CLASS_DITTO) to lowest * (DDT_CLASS_UNIQUE) so that we may walk the DDT in that order. * * To prevent excess scrubbing, the scrub begins by walking the DDT * to find all blocks with refcnt > 1, and scrubs each of these once. * Since there are two replication classes which contain blocks with * refcnt > 1, we scrub the highest replication class (DDT_CLASS_DITTO) first. * Finally the top-down scrub begins, only visiting blocks with refcnt == 1. * * There would be nothing more to say if a block's refcnt couldn't change * during a scrub, but of course it can so we must account for changes * in a block's replication class. * * Here's an example of what can occur: * * If a block has refcnt > 1 during the DDT scrub phase, but has refcnt == 1 * when visited during the top-down scrub phase, it will be scrubbed twice. * This negates our scrub optimization, but is otherwise harmless. * * If a block has refcnt == 1 during the DDT scrub phase, but has refcnt > 1 * on each visit during the top-down scrub phase, it will never be scrubbed. * To catch this, ddt_sync_entry() notifies the scrub code whenever a block's * reference class transitions to a higher level (i.e DDT_CLASS_UNIQUE to * DDT_CLASS_DUPLICATE); if it transitions from refcnt == 1 to refcnt > 1 * while a scrub is in progress, it scrubs the block right then. */ static void dsl_scan_ddt(dsl_scan_t *scn, dmu_tx_t *tx) { ddt_bookmark_t *ddb = &scn->scn_phys.scn_ddt_bookmark; ddt_entry_t dde = { 0 }; int error; uint64_t n = 0; while ((error = ddt_walk(scn->scn_dp->dp_spa, ddb, &dde)) == 0) { ddt_t *ddt; if (ddb->ddb_class > scn->scn_phys.scn_ddt_class_max) break; dprintf("visiting ddb=%llu/%llu/%llu/%llx\n", (longlong_t)ddb->ddb_class, (longlong_t)ddb->ddb_type, (longlong_t)ddb->ddb_checksum, (longlong_t)ddb->ddb_cursor); /* There should be no pending changes to the dedup table */ ddt = scn->scn_dp->dp_spa->spa_ddt[ddb->ddb_checksum]; ASSERT(avl_first(&ddt->ddt_tree) == NULL); dsl_scan_ddt_entry(scn, ddb->ddb_checksum, &dde, tx); n++; if (dsl_scan_check_pause(scn, NULL)) break; } zfs_dbgmsg("scanned %llu ddt entries with class_max = %u; pausing=%u", (longlong_t)n, (int)scn->scn_phys.scn_ddt_class_max, (int)scn->scn_pausing); ASSERT(error == 0 || error == ENOENT); ASSERT(error != ENOENT || ddb->ddb_class > scn->scn_phys.scn_ddt_class_max); } /* ARGSUSED */ void dsl_scan_ddt_entry(dsl_scan_t *scn, enum zio_checksum checksum, ddt_entry_t *dde, dmu_tx_t *tx) { const ddt_key_t *ddk = &dde->dde_key; ddt_phys_t *ddp = dde->dde_phys; blkptr_t bp; zbookmark_phys_t zb = { 0 }; if (scn->scn_phys.scn_state != DSS_SCANNING) return; for (int p = 0; p < DDT_PHYS_TYPES; p++, ddp++) { if (ddp->ddp_phys_birth == 0 || ddp->ddp_phys_birth > scn->scn_phys.scn_max_txg) continue; ddt_bp_create(checksum, ddk, ddp, &bp); scn->scn_visited_this_txg++; scan_funcs[scn->scn_phys.scn_func](scn->scn_dp, &bp, &zb); } } static void dsl_scan_visit(dsl_scan_t *scn, dmu_tx_t *tx) { dsl_pool_t *dp = scn->scn_dp; zap_cursor_t zc; zap_attribute_t za; if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_ddt(scn, tx); if (scn->scn_pausing) return; } if (scn->scn_phys.scn_bookmark.zb_objset == DMU_META_OBJSET) { /* First do the MOS & ORIGIN */ scn->scn_phys.scn_cur_min_txg = scn->scn_phys.scn_min_txg; scn->scn_phys.scn_cur_max_txg = scn->scn_phys.scn_max_txg; dsl_scan_visit_rootbp(scn, NULL, &dp->dp_meta_rootbp, tx); spa_set_rootblkptr(dp->dp_spa, &dp->dp_meta_rootbp); if (scn->scn_pausing) return; if (spa_version(dp->dp_spa) < SPA_VERSION_DSL_SCRUB) { VERIFY0(dmu_objset_find_dp(dp, dp->dp_root_dir_obj, enqueue_cb, tx, DS_FIND_CHILDREN)); } else { dsl_scan_visitds(scn, dp->dp_origin_snap->ds_object, tx); } ASSERT(!scn->scn_pausing); } else if (scn->scn_phys.scn_bookmark.zb_objset != ZB_DESTROYED_OBJSET) { /* * If we were paused, continue from here. Note if the * ds we were paused on was deleted, the zb_objset may * be -1, so we will skip this and find a new objset * below. */ dsl_scan_visitds(scn, scn->scn_phys.scn_bookmark.zb_objset, tx); if (scn->scn_pausing) return; } /* * In case we were paused right at the end of the ds, zero the * bookmark so we don't think that we're still trying to resume. */ bzero(&scn->scn_phys.scn_bookmark, sizeof (zbookmark_phys_t)); /* keep pulling things out of the zap-object-as-queue */ while (zap_cursor_init(&zc, dp->dp_meta_objset, scn->scn_phys.scn_queue_obj), zap_cursor_retrieve(&zc, &za) == 0) { dsl_dataset_t *ds; uint64_t dsobj; - dsobj = strtonum(za.za_name, NULL); + dsobj = zfs_strtonum(za.za_name, NULL); VERIFY3U(0, ==, zap_remove_int(dp->dp_meta_objset, scn->scn_phys.scn_queue_obj, dsobj, tx)); /* Set up min/max txg */ VERIFY3U(0, ==, dsl_dataset_hold_obj(dp, dsobj, FTAG, &ds)); if (za.za_first_integer != 0) { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, za.za_first_integer); } else { scn->scn_phys.scn_cur_min_txg = MAX(scn->scn_phys.scn_min_txg, dsl_dataset_phys(ds)->ds_prev_snap_txg); } scn->scn_phys.scn_cur_max_txg = dsl_scan_ds_maxtxg(ds); dsl_dataset_rele(ds, FTAG); dsl_scan_visitds(scn, dsobj, tx); zap_cursor_fini(&zc); if (scn->scn_pausing) return; } zap_cursor_fini(&zc); } static boolean_t dsl_scan_free_should_pause(dsl_scan_t *scn) { uint64_t elapsed_nanosecs; if (zfs_recover) return (B_FALSE); if (scn->scn_visited_this_txg >= zfs_free_max_blocks) return (B_TRUE); elapsed_nanosecs = gethrtime() - scn->scn_sync_start_time; return (elapsed_nanosecs / NANOSEC > zfs_txg_timeout || (NSEC2MSEC(elapsed_nanosecs) > zfs_free_min_time_ms && txg_sync_waiting(scn->scn_dp)) || spa_shutting_down(scn->scn_dp->dp_spa)); } static int dsl_scan_free_block_cb(void *arg, const blkptr_t *bp, dmu_tx_t *tx) { dsl_scan_t *scn = arg; if (!scn->scn_is_bptree || (BP_GET_LEVEL(bp) == 0 && BP_GET_TYPE(bp) != DMU_OT_OBJSET)) { if (dsl_scan_free_should_pause(scn)) return (SET_ERROR(ERESTART)); } zio_nowait(zio_free_sync(scn->scn_zio_root, scn->scn_dp->dp_spa, dmu_tx_get_txg(tx), bp, 0)); dsl_dir_diduse_space(tx->tx_pool->dp_free_dir, DD_USED_HEAD, -bp_get_dsize_sync(scn->scn_dp->dp_spa, bp), -BP_GET_PSIZE(bp), -BP_GET_UCSIZE(bp), tx); scn->scn_visited_this_txg++; return (0); } boolean_t dsl_scan_active(dsl_scan_t *scn) { spa_t *spa = scn->scn_dp->dp_spa; uint64_t used = 0, comp, uncomp; if (spa->spa_load_state != SPA_LOAD_NONE) return (B_FALSE); if (spa_shutting_down(spa)) return (B_FALSE); if (scn->scn_phys.scn_state == DSS_SCANNING || (scn->scn_async_destroying && !scn->scn_async_stalled)) return (B_TRUE); if (spa_version(scn->scn_dp->dp_spa) >= SPA_VERSION_DEADLISTS) { (void) bpobj_space(&scn->scn_dp->dp_free_bpobj, &used, &comp, &uncomp); } return (used != 0); } /* Called whenever a txg syncs. */ void dsl_scan_sync(dsl_pool_t *dp, dmu_tx_t *tx) { dsl_scan_t *scn = dp->dp_scan; spa_t *spa = dp->dp_spa; int err = 0; /* * Check for scn_restart_txg before checking spa_load_state, so * that we can restart an old-style scan while the pool is being * imported (see dsl_scan_init). */ if (dsl_scan_restarting(scn, tx)) { pool_scan_func_t func = POOL_SCAN_SCRUB; dsl_scan_done(scn, B_FALSE, tx); if (vdev_resilver_needed(spa->spa_root_vdev, NULL, NULL)) func = POOL_SCAN_RESILVER; zfs_dbgmsg("restarting scan func=%u txg=%llu", func, tx->tx_txg); dsl_scan_setup_sync(&func, tx); } /* * Only process scans in sync pass 1. */ if (spa_sync_pass(dp->dp_spa) > 1) return; /* * If the spa is shutting down, then stop scanning. This will * ensure that the scan does not dirty any new data during the * shutdown phase. */ if (spa_shutting_down(spa)) return; /* * If the scan is inactive due to a stalled async destroy, try again. */ if (!scn->scn_async_stalled && !dsl_scan_active(scn)) return; scn->scn_visited_this_txg = 0; scn->scn_pausing = B_FALSE; scn->scn_sync_start_time = gethrtime(); spa->spa_scrub_active = B_TRUE; /* * First process the async destroys. If we pause, don't do * any scrubbing or resilvering. This ensures that there are no * async destroys while we are scanning, so the scan code doesn't * have to worry about traversing it. It is also faster to free the * blocks than to scrub them. */ if (zfs_free_bpobj_enabled && spa_version(dp->dp_spa) >= SPA_VERSION_DEADLISTS) { scn->scn_is_bptree = B_FALSE; scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bpobj_iterate(&dp->dp_free_bpobj, dsl_scan_free_block_cb, scn, tx); VERIFY3U(0, ==, zio_wait(scn->scn_zio_root)); if (err != 0 && err != ERESTART) zfs_panic_recover("error %u from bpobj_iterate()", err); } if (err == 0 && spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)) { ASSERT(scn->scn_async_destroying); scn->scn_is_bptree = B_TRUE; scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_MUSTSUCCEED); err = bptree_iterate(dp->dp_meta_objset, dp->dp_bptree_obj, B_TRUE, dsl_scan_free_block_cb, scn, tx); VERIFY0(zio_wait(scn->scn_zio_root)); if (err == EIO || err == ECKSUM) { err = 0; } else if (err != 0 && err != ERESTART) { zfs_panic_recover("error %u from " "traverse_dataset_destroyed()", err); } if (bptree_is_empty(dp->dp_meta_objset, dp->dp_bptree_obj)) { /* finished; deactivate async destroy feature */ spa_feature_decr(spa, SPA_FEATURE_ASYNC_DESTROY, tx); ASSERT(!spa_feature_is_active(spa, SPA_FEATURE_ASYNC_DESTROY)); VERIFY0(zap_remove(dp->dp_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_BPTREE_OBJ, tx)); VERIFY0(bptree_free(dp->dp_meta_objset, dp->dp_bptree_obj, tx)); dp->dp_bptree_obj = 0; scn->scn_async_destroying = B_FALSE; scn->scn_async_stalled = B_FALSE; } else { /* * If we didn't make progress, mark the async * destroy as stalled, so that we will not initiate * a spa_sync() on its behalf. Note that we only * check this if we are not finished, because if the * bptree had no blocks for us to visit, we can * finish without "making progress". */ scn->scn_async_stalled = (scn->scn_visited_this_txg == 0); } } if (scn->scn_visited_this_txg) { zfs_dbgmsg("freed %llu blocks in %llums from " "free_bpobj/bptree txg %llu; err=%u", (longlong_t)scn->scn_visited_this_txg, (longlong_t) NSEC2MSEC(gethrtime() - scn->scn_sync_start_time), (longlong_t)tx->tx_txg, err); scn->scn_visited_this_txg = 0; /* * Write out changes to the DDT that may be required as a * result of the blocks freed. This ensures that the DDT * is clean when a scrub/resilver runs. */ ddt_sync(spa, tx->tx_txg); } if (err != 0) return; if (dp->dp_free_dir != NULL && !scn->scn_async_destroying && zfs_free_leak_on_eio && (dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes != 0 || dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes != 0 || dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes != 0)) { /* * We have finished background destroying, but there is still * some space left in the dp_free_dir. Transfer this leaked * space to the dp_leak_dir. */ if (dp->dp_leak_dir == NULL) { rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); (void) dsl_dir_create_sync(dp, dp->dp_root_dir, LEAK_DIR_NAME, tx); VERIFY0(dsl_pool_open_special_dir(dp, LEAK_DIR_NAME, &dp->dp_leak_dir)); rrw_exit(&dp->dp_config_rwlock, FTAG); } dsl_dir_diduse_space(dp->dp_leak_dir, DD_USED_HEAD, dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes, dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes, dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes, tx); dsl_dir_diduse_space(dp->dp_free_dir, DD_USED_HEAD, -dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes, -dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes, -dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes, tx); } if (dp->dp_free_dir != NULL && !scn->scn_async_destroying) { /* finished; verify that space accounting went to zero */ ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_used_bytes); ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_compressed_bytes); ASSERT0(dsl_dir_phys(dp->dp_free_dir)->dd_uncompressed_bytes); } if (scn->scn_phys.scn_state != DSS_SCANNING) return; if (scn->scn_done_txg == tx->tx_txg) { ASSERT(!scn->scn_pausing); /* finished with scan. */ zfs_dbgmsg("txg %llu scan complete", tx->tx_txg); dsl_scan_done(scn, B_TRUE, tx); ASSERT3U(spa->spa_scrub_inflight, ==, 0); dsl_scan_sync_state(scn, tx); return; } if (scn->scn_phys.scn_ddt_bookmark.ddb_class <= scn->scn_phys.scn_ddt_class_max) { zfs_dbgmsg("doing scan sync txg %llu; " "ddt bm=%llu/%llu/%llu/%llx", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_class, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_type, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_checksum, (longlong_t)scn->scn_phys.scn_ddt_bookmark.ddb_cursor); ASSERT(scn->scn_phys.scn_bookmark.zb_objset == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_object == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_level == 0); ASSERT(scn->scn_phys.scn_bookmark.zb_blkid == 0); } else { zfs_dbgmsg("doing scan sync txg %llu; bm=%llu/%llu/%llu/%llu", (longlong_t)tx->tx_txg, (longlong_t)scn->scn_phys.scn_bookmark.zb_objset, (longlong_t)scn->scn_phys.scn_bookmark.zb_object, (longlong_t)scn->scn_phys.scn_bookmark.zb_level, (longlong_t)scn->scn_phys.scn_bookmark.zb_blkid); } scn->scn_zio_root = zio_root(dp->dp_spa, NULL, NULL, ZIO_FLAG_CANFAIL); dsl_pool_config_enter(dp, FTAG); dsl_scan_visit(scn, tx); dsl_pool_config_exit(dp, FTAG); (void) zio_wait(scn->scn_zio_root); scn->scn_zio_root = NULL; zfs_dbgmsg("visited %llu blocks in %llums", (longlong_t)scn->scn_visited_this_txg, (longlong_t)NSEC2MSEC(gethrtime() - scn->scn_sync_start_time)); if (!scn->scn_pausing) { scn->scn_done_txg = tx->tx_txg + 1; zfs_dbgmsg("txg %llu traversal complete, waiting till txg %llu", tx->tx_txg, scn->scn_done_txg); } if (DSL_SCAN_IS_SCRUB_RESILVER(scn)) { mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight > 0) { cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); } mutex_exit(&spa->spa_scrub_lock); } dsl_scan_sync_state(scn, tx); } /* * This will start a new scan, or restart an existing one. */ void dsl_resilver_restart(dsl_pool_t *dp, uint64_t txg) { if (txg == 0) { dmu_tx_t *tx; tx = dmu_tx_create_dd(dp->dp_mos_dir); VERIFY(0 == dmu_tx_assign(tx, TXG_WAIT)); txg = dmu_tx_get_txg(tx); dp->dp_scan->scn_restart_txg = txg; dmu_tx_commit(tx); } else { dp->dp_scan->scn_restart_txg = txg; } zfs_dbgmsg("restarting resilver txg=%llu", txg); } boolean_t dsl_scan_resilvering(dsl_pool_t *dp) { return (dp->dp_scan->scn_phys.scn_state == DSS_SCANNING && dp->dp_scan->scn_phys.scn_func == POOL_SCAN_RESILVER); } /* * scrub consumers */ static void count_block(zfs_all_blkstats_t *zab, const blkptr_t *bp) { int i; /* * If we resume after a reboot, zab will be NULL; don't record * incomplete stats in that case. */ if (zab == NULL) return; for (i = 0; i < 4; i++) { int l = (i < 2) ? BP_GET_LEVEL(bp) : DN_MAX_LEVELS; int t = (i & 1) ? BP_GET_TYPE(bp) : DMU_OT_TOTAL; if (t & DMU_OT_NEWTYPE) t = DMU_OT_OTHER; zfs_blkstat_t *zb = &zab->zab_type[l][t]; int equal; zb->zb_count++; zb->zb_asize += BP_GET_ASIZE(bp); zb->zb_lsize += BP_GET_LSIZE(bp); zb->zb_psize += BP_GET_PSIZE(bp); zb->zb_gangs += BP_COUNT_GANG(bp); switch (BP_GET_NDVAS(bp)) { case 2: if (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) zb->zb_ditto_2_of_2_samevdev++; break; case 3: equal = (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[1])) + (DVA_GET_VDEV(&bp->blk_dva[0]) == DVA_GET_VDEV(&bp->blk_dva[2])) + (DVA_GET_VDEV(&bp->blk_dva[1]) == DVA_GET_VDEV(&bp->blk_dva[2])); if (equal == 1) zb->zb_ditto_2_of_3_samevdev++; else if (equal == 3) zb->zb_ditto_3_of_3_samevdev++; break; } } } static void dsl_scan_scrub_done(zio_t *zio) { spa_t *spa = zio->io_spa; abd_free(zio->io_abd); mutex_enter(&spa->spa_scrub_lock); spa->spa_scrub_inflight--; cv_broadcast(&spa->spa_scrub_io_cv); if (zio->io_error && (zio->io_error != ECKSUM || !(zio->io_flags & ZIO_FLAG_SPECULATIVE))) { spa->spa_dsl_pool->dp_scan->scn_phys.scn_errors++; } mutex_exit(&spa->spa_scrub_lock); } static int dsl_scan_scrub_cb(dsl_pool_t *dp, const blkptr_t *bp, const zbookmark_phys_t *zb) { dsl_scan_t *scn = dp->dp_scan; size_t size = BP_GET_PSIZE(bp); spa_t *spa = dp->dp_spa; uint64_t phys_birth = BP_PHYSICAL_BIRTH(bp); boolean_t needs_io; int zio_flags = ZIO_FLAG_SCAN_THREAD | ZIO_FLAG_RAW | ZIO_FLAG_CANFAIL; int scan_delay = 0; if (phys_birth <= scn->scn_phys.scn_min_txg || phys_birth >= scn->scn_phys.scn_max_txg) return (0); count_block(dp->dp_blkstats, bp); if (BP_IS_EMBEDDED(bp)) return (0); ASSERT(DSL_SCAN_IS_SCRUB_RESILVER(scn)); if (scn->scn_phys.scn_func == POOL_SCAN_SCRUB) { zio_flags |= ZIO_FLAG_SCRUB; needs_io = B_TRUE; scan_delay = zfs_scrub_delay; } else { ASSERT3U(scn->scn_phys.scn_func, ==, POOL_SCAN_RESILVER); zio_flags |= ZIO_FLAG_RESILVER; needs_io = B_FALSE; scan_delay = zfs_resilver_delay; } /* If it's an intent log block, failure is expected. */ if (zb->zb_level == ZB_ZIL_LEVEL) zio_flags |= ZIO_FLAG_SPECULATIVE; for (int d = 0; d < BP_GET_NDVAS(bp); d++) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(&bp->blk_dva[d])); /* * Keep track of how much data we've examined so that * zpool(1M) status can make useful progress reports. */ scn->scn_phys.scn_examined += DVA_GET_ASIZE(&bp->blk_dva[d]); spa->spa_scan_pass_exam += DVA_GET_ASIZE(&bp->blk_dva[d]); /* if it's a resilver, this may not be in the target range */ if (!needs_io) { if (DVA_GET_GANG(&bp->blk_dva[d])) { /* * Gang members may be spread across multiple * vdevs, so the best estimate we have is the * scrub range, which has already been checked. * XXX -- it would be better to change our * allocation policy to ensure that all * gang members reside on the same vdev. */ needs_io = B_TRUE; } else { needs_io = vdev_dtl_contains(vd, DTL_PARTIAL, phys_birth, 1); } } } if (needs_io && !zfs_no_scrub_io) { vdev_t *rvd = spa->spa_root_vdev; uint64_t maxinflight = rvd->vdev_children * zfs_top_maxinflight; mutex_enter(&spa->spa_scrub_lock); while (spa->spa_scrub_inflight >= maxinflight) cv_wait(&spa->spa_scrub_io_cv, &spa->spa_scrub_lock); spa->spa_scrub_inflight++; mutex_exit(&spa->spa_scrub_lock); /* * If we're seeing recent (zfs_scan_idle) "important" I/Os * then throttle our workload to limit the impact of a scan. */ if (ddi_get_lbolt64() - spa->spa_last_io <= zfs_scan_idle) delay(scan_delay); zio_nowait(zio_read(NULL, spa, bp, abd_alloc_for_io(size, B_FALSE), size, dsl_scan_scrub_done, NULL, ZIO_PRIORITY_SCRUB, zio_flags, zb)); } /* do not relocate this block */ return (0); } /* Called by the ZFS_IOC_POOL_SCAN ioctl to start a scrub or resilver */ int dsl_scan(dsl_pool_t *dp, pool_scan_func_t func) { spa_t *spa = dp->dp_spa; /* * Purge all vdev caches and probe all devices. We do this here * rather than in sync context because this requires a writer lock * on the spa_config lock, which we can't do from sync context. The * spa_scrub_reopen flag indicates that vdev_open() should not * attempt to start another scrub. */ spa_vdev_state_enter(spa, SCL_NONE); spa->spa_scrub_reopen = B_TRUE; vdev_reopen(spa->spa_root_vdev); spa->spa_scrub_reopen = B_FALSE; (void) spa_vdev_state_exit(spa, NULL, 0); return (dsl_sync_task(spa_name(spa), dsl_scan_setup_check, dsl_scan_setup_sync, &func, 0, ZFS_SPACE_CHECK_NONE)); } static boolean_t dsl_scan_restarting(dsl_scan_t *scn, dmu_tx_t *tx) { return (scn->scn_restart_txg != 0 && scn->scn_restart_txg <= tx->tx_txg); } Index: vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_userhold.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_userhold.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/dsl_userhold.c (revision 319948) @@ -1,666 +1,666 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2015 by Delphix. All rights reserved. * Copyright (c) 2013 Steven Hartland. All rights reserved. */ #include #include #include #include #include #include #include #include #include #include #include typedef struct dsl_dataset_user_hold_arg { nvlist_t *dduha_holds; nvlist_t *dduha_chkholds; nvlist_t *dduha_errlist; minor_t dduha_minor; } dsl_dataset_user_hold_arg_t; /* * If you add new checks here, you may need to add additional checks to the * "temporary" case in snapshot_check() in dmu_objset.c. */ int dsl_dataset_user_hold_check_one(dsl_dataset_t *ds, const char *htag, boolean_t temphold, dmu_tx_t *tx) { dsl_pool_t *dp = dmu_tx_pool(tx); objset_t *mos = dp->dp_meta_objset; int error = 0; ASSERT(dsl_pool_config_held(dp)); if (strlen(htag) > MAXNAMELEN) return (SET_ERROR(E2BIG)); /* Tempholds have a more restricted length */ if (temphold && strlen(htag) + MAX_TAG_PREFIX_LEN >= MAXNAMELEN) return (SET_ERROR(E2BIG)); /* tags must be unique (if ds already exists) */ if (ds != NULL && dsl_dataset_phys(ds)->ds_userrefs_obj != 0) { uint64_t value; error = zap_lookup(mos, dsl_dataset_phys(ds)->ds_userrefs_obj, htag, 8, 1, &value); if (error == 0) error = SET_ERROR(EEXIST); else if (error == ENOENT) error = 0; } return (error); } static int dsl_dataset_user_hold_check(void *arg, dmu_tx_t *tx) { dsl_dataset_user_hold_arg_t *dduha = arg; dsl_pool_t *dp = dmu_tx_pool(tx); if (spa_version(dp->dp_spa) < SPA_VERSION_USERREFS) return (SET_ERROR(ENOTSUP)); if (!dmu_tx_is_syncing(tx)) return (0); for (nvpair_t *pair = nvlist_next_nvpair(dduha->dduha_holds, NULL); pair != NULL; pair = nvlist_next_nvpair(dduha->dduha_holds, pair)) { dsl_dataset_t *ds; int error = 0; char *htag, *name; /* must be a snapshot */ name = nvpair_name(pair); if (strchr(name, '@') == NULL) error = SET_ERROR(EINVAL); if (error == 0) error = nvpair_value_string(pair, &htag); if (error == 0) error = dsl_dataset_hold(dp, name, FTAG, &ds); if (error == 0) { error = dsl_dataset_user_hold_check_one(ds, htag, dduha->dduha_minor != 0, tx); dsl_dataset_rele(ds, FTAG); } if (error == 0) { fnvlist_add_string(dduha->dduha_chkholds, name, htag); } else { /* * We register ENOENT errors so they can be correctly * reported if needed, such as when all holds fail. */ fnvlist_add_int32(dduha->dduha_errlist, name, error); if (error != ENOENT) return (error); } } return (0); } static void dsl_dataset_user_hold_sync_one_impl(nvlist_t *tmpholds, dsl_dataset_t *ds, const char *htag, minor_t minor, uint64_t now, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; objset_t *mos = dp->dp_meta_objset; uint64_t zapobj; ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock)); if (dsl_dataset_phys(ds)->ds_userrefs_obj == 0) { /* * This is the first user hold for this dataset. Create * the userrefs zap object. */ dmu_buf_will_dirty(ds->ds_dbuf, tx); zapobj = dsl_dataset_phys(ds)->ds_userrefs_obj = zap_create(mos, DMU_OT_USERREFS, DMU_OT_NONE, 0, tx); } else { zapobj = dsl_dataset_phys(ds)->ds_userrefs_obj; } ds->ds_userrefs++; VERIFY0(zap_add(mos, zapobj, htag, 8, 1, &now, tx)); if (minor != 0) { char name[MAXNAMELEN]; nvlist_t *tags; VERIFY0(dsl_pool_user_hold(dp, ds->ds_object, htag, now, tx)); (void) snprintf(name, sizeof (name), "%llx", (u_longlong_t)ds->ds_object); if (nvlist_lookup_nvlist(tmpholds, name, &tags) != 0) { tags = fnvlist_alloc(); fnvlist_add_boolean(tags, htag); fnvlist_add_nvlist(tmpholds, name, tags); fnvlist_free(tags); } else { fnvlist_add_boolean(tags, htag); } } spa_history_log_internal_ds(ds, "hold", tx, "tag=%s temp=%d refs=%llu", htag, minor != 0, ds->ds_userrefs); } typedef struct zfs_hold_cleanup_arg { char zhca_spaname[ZFS_MAX_DATASET_NAME_LEN]; uint64_t zhca_spa_load_guid; nvlist_t *zhca_holds; } zfs_hold_cleanup_arg_t; static void dsl_dataset_user_release_onexit(void *arg) { zfs_hold_cleanup_arg_t *ca = arg; spa_t *spa; int error; error = spa_open(ca->zhca_spaname, &spa, FTAG); if (error != 0) { zfs_dbgmsg("couldn't release holds on pool=%s " "because pool is no longer loaded", ca->zhca_spaname); return; } if (spa_load_guid(spa) != ca->zhca_spa_load_guid) { zfs_dbgmsg("couldn't release holds on pool=%s " "because pool is no longer loaded (guid doesn't match)", ca->zhca_spaname); spa_close(spa, FTAG); return; } (void) dsl_dataset_user_release_tmp(spa_get_dsl(spa), ca->zhca_holds); fnvlist_free(ca->zhca_holds); kmem_free(ca, sizeof (zfs_hold_cleanup_arg_t)); spa_close(spa, FTAG); } static void dsl_onexit_hold_cleanup(spa_t *spa, nvlist_t *holds, minor_t minor) { zfs_hold_cleanup_arg_t *ca; if (minor == 0 || nvlist_empty(holds)) { fnvlist_free(holds); return; } ASSERT(spa != NULL); ca = kmem_alloc(sizeof (*ca), KM_SLEEP); (void) strlcpy(ca->zhca_spaname, spa_name(spa), sizeof (ca->zhca_spaname)); ca->zhca_spa_load_guid = spa_load_guid(spa); ca->zhca_holds = holds; VERIFY0(zfs_onexit_add_cb(minor, dsl_dataset_user_release_onexit, ca, NULL)); } void dsl_dataset_user_hold_sync_one(dsl_dataset_t *ds, const char *htag, minor_t minor, uint64_t now, dmu_tx_t *tx) { nvlist_t *tmpholds; if (minor != 0) tmpholds = fnvlist_alloc(); else tmpholds = NULL; dsl_dataset_user_hold_sync_one_impl(tmpholds, ds, htag, minor, now, tx); dsl_onexit_hold_cleanup(dsl_dataset_get_spa(ds), tmpholds, minor); } static void dsl_dataset_user_hold_sync(void *arg, dmu_tx_t *tx) { dsl_dataset_user_hold_arg_t *dduha = arg; dsl_pool_t *dp = dmu_tx_pool(tx); nvlist_t *tmpholds; uint64_t now = gethrestime_sec(); if (dduha->dduha_minor != 0) tmpholds = fnvlist_alloc(); else tmpholds = NULL; for (nvpair_t *pair = nvlist_next_nvpair(dduha->dduha_chkholds, NULL); pair != NULL; pair = nvlist_next_nvpair(dduha->dduha_chkholds, pair)) { dsl_dataset_t *ds; VERIFY0(dsl_dataset_hold(dp, nvpair_name(pair), FTAG, &ds)); dsl_dataset_user_hold_sync_one_impl(tmpholds, ds, fnvpair_value_string(pair), dduha->dduha_minor, now, tx); dsl_dataset_rele(ds, FTAG); } dsl_onexit_hold_cleanup(dp->dp_spa, tmpholds, dduha->dduha_minor); } /* * The full semantics of this function are described in the comment above * lzc_hold(). * * To summarize: * holds is nvl of snapname -> holdname * errlist will be filled in with snapname -> error * * The snaphosts must all be in the same pool. * * Holds for snapshots that don't exist will be skipped. * * If none of the snapshots for requested holds exist then ENOENT will be * returned. * * If cleanup_minor is not 0, the holds will be temporary, which will be cleaned * up when the process exits. * * On success all the holds, for snapshots that existed, will be created and 0 * will be returned. * * On failure no holds will be created, the errlist will be filled in, * and an errno will returned. * * In all cases the errlist will contain entries for holds where the snapshot * didn't exist. */ int dsl_dataset_user_hold(nvlist_t *holds, minor_t cleanup_minor, nvlist_t *errlist) { dsl_dataset_user_hold_arg_t dduha; nvpair_t *pair; int ret; pair = nvlist_next_nvpair(holds, NULL); if (pair == NULL) return (0); dduha.dduha_holds = holds; dduha.dduha_chkholds = fnvlist_alloc(); dduha.dduha_errlist = errlist; dduha.dduha_minor = cleanup_minor; ret = dsl_sync_task(nvpair_name(pair), dsl_dataset_user_hold_check, dsl_dataset_user_hold_sync, &dduha, fnvlist_num_pairs(holds), ZFS_SPACE_CHECK_RESERVED); fnvlist_free(dduha.dduha_chkholds); return (ret); } typedef int (dsl_holdfunc_t)(dsl_pool_t *dp, const char *name, void *tag, dsl_dataset_t **dsp); typedef struct dsl_dataset_user_release_arg { dsl_holdfunc_t *ddura_holdfunc; nvlist_t *ddura_holds; nvlist_t *ddura_todelete; nvlist_t *ddura_errlist; nvlist_t *ddura_chkholds; } dsl_dataset_user_release_arg_t; /* Place a dataset hold on the snapshot identified by passed dsobj string */ static int dsl_dataset_hold_obj_string(dsl_pool_t *dp, const char *dsobj, void *tag, dsl_dataset_t **dsp) { - return (dsl_dataset_hold_obj(dp, strtonum(dsobj, NULL), tag, dsp)); + return (dsl_dataset_hold_obj(dp, zfs_strtonum(dsobj, NULL), tag, dsp)); } static int dsl_dataset_user_release_check_one(dsl_dataset_user_release_arg_t *ddura, dsl_dataset_t *ds, nvlist_t *holds, const char *snapname) { uint64_t zapobj; nvlist_t *holds_found; objset_t *mos; int numholds; if (!ds->ds_is_snapshot) return (SET_ERROR(EINVAL)); if (nvlist_empty(holds)) return (0); numholds = 0; mos = ds->ds_dir->dd_pool->dp_meta_objset; zapobj = dsl_dataset_phys(ds)->ds_userrefs_obj; holds_found = fnvlist_alloc(); for (nvpair_t *pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { uint64_t tmp; int error; const char *holdname = nvpair_name(pair); if (zapobj != 0) error = zap_lookup(mos, zapobj, holdname, 8, 1, &tmp); else error = SET_ERROR(ENOENT); /* * Non-existent holds are put on the errlist, but don't * cause an overall failure. */ if (error == ENOENT) { if (ddura->ddura_errlist != NULL) { char *errtag = kmem_asprintf("%s#%s", snapname, holdname); fnvlist_add_int32(ddura->ddura_errlist, errtag, ENOENT); strfree(errtag); } continue; } if (error != 0) { fnvlist_free(holds_found); return (error); } fnvlist_add_boolean(holds_found, holdname); numholds++; } if (DS_IS_DEFER_DESTROY(ds) && dsl_dataset_phys(ds)->ds_num_children == 1 && ds->ds_userrefs == numholds) { /* we need to destroy the snapshot as well */ if (dsl_dataset_long_held(ds)) { fnvlist_free(holds_found); return (SET_ERROR(EBUSY)); } fnvlist_add_boolean(ddura->ddura_todelete, snapname); } if (numholds != 0) { fnvlist_add_nvlist(ddura->ddura_chkholds, snapname, holds_found); } fnvlist_free(holds_found); return (0); } static int dsl_dataset_user_release_check(void *arg, dmu_tx_t *tx) { dsl_dataset_user_release_arg_t *ddura; dsl_holdfunc_t *holdfunc; dsl_pool_t *dp; if (!dmu_tx_is_syncing(tx)) return (0); dp = dmu_tx_pool(tx); ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock)); ddura = arg; holdfunc = ddura->ddura_holdfunc; for (nvpair_t *pair = nvlist_next_nvpair(ddura->ddura_holds, NULL); pair != NULL; pair = nvlist_next_nvpair(ddura->ddura_holds, pair)) { int error; dsl_dataset_t *ds; nvlist_t *holds; const char *snapname = nvpair_name(pair); error = nvpair_value_nvlist(pair, &holds); if (error != 0) error = (SET_ERROR(EINVAL)); else error = holdfunc(dp, snapname, FTAG, &ds); if (error == 0) { error = dsl_dataset_user_release_check_one(ddura, ds, holds, snapname); dsl_dataset_rele(ds, FTAG); } if (error != 0) { if (ddura->ddura_errlist != NULL) { fnvlist_add_int32(ddura->ddura_errlist, snapname, error); } /* * Non-existent snapshots are put on the errlist, * but don't cause an overall failure. */ if (error != ENOENT) return (error); } } return (0); } static void dsl_dataset_user_release_sync_one(dsl_dataset_t *ds, nvlist_t *holds, dmu_tx_t *tx) { dsl_pool_t *dp = ds->ds_dir->dd_pool; objset_t *mos = dp->dp_meta_objset; for (nvpair_t *pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { int error; const char *holdname = nvpair_name(pair); /* Remove temporary hold if one exists. */ error = dsl_pool_user_release(dp, ds->ds_object, holdname, tx); VERIFY(error == 0 || error == ENOENT); VERIFY0(zap_remove(mos, dsl_dataset_phys(ds)->ds_userrefs_obj, holdname, tx)); ds->ds_userrefs--; spa_history_log_internal_ds(ds, "release", tx, "tag=%s refs=%lld", holdname, (longlong_t)ds->ds_userrefs); } } static void dsl_dataset_user_release_sync(void *arg, dmu_tx_t *tx) { dsl_dataset_user_release_arg_t *ddura = arg; dsl_holdfunc_t *holdfunc = ddura->ddura_holdfunc; dsl_pool_t *dp = dmu_tx_pool(tx); ASSERT(RRW_WRITE_HELD(&dp->dp_config_rwlock)); for (nvpair_t *pair = nvlist_next_nvpair(ddura->ddura_chkholds, NULL); pair != NULL; pair = nvlist_next_nvpair(ddura->ddura_chkholds, pair)) { dsl_dataset_t *ds; const char *name = nvpair_name(pair); VERIFY0(holdfunc(dp, name, FTAG, &ds)); dsl_dataset_user_release_sync_one(ds, fnvpair_value_nvlist(pair), tx); if (nvlist_exists(ddura->ddura_todelete, name)) { ASSERT(ds->ds_userrefs == 0 && dsl_dataset_phys(ds)->ds_num_children == 1 && DS_IS_DEFER_DESTROY(ds)); dsl_destroy_snapshot_sync_impl(ds, B_FALSE, tx); } dsl_dataset_rele(ds, FTAG); } } /* * The full semantics of this function are described in the comment above * lzc_release(). * * To summarize: * Releases holds specified in the nvl holds. * * holds is nvl of snapname -> { holdname, ... } * errlist will be filled in with snapname -> error * * If tmpdp is not NULL the names for holds should be the dsobj's of snapshots, * otherwise they should be the names of shapshots. * * As a release may cause snapshots to be destroyed this trys to ensure they * aren't mounted. * * The release of non-existent holds are skipped. * * At least one hold must have been released for the this function to succeed * and return 0. */ static int dsl_dataset_user_release_impl(nvlist_t *holds, nvlist_t *errlist, dsl_pool_t *tmpdp) { dsl_dataset_user_release_arg_t ddura; nvpair_t *pair; char *pool; int error; pair = nvlist_next_nvpair(holds, NULL); if (pair == NULL) return (0); /* * The release may cause snapshots to be destroyed; make sure they * are not mounted. */ if (tmpdp != NULL) { /* Temporary holds are specified by dsobj string. */ ddura.ddura_holdfunc = dsl_dataset_hold_obj_string; pool = spa_name(tmpdp->dp_spa); #ifdef _KERNEL for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { dsl_dataset_t *ds; dsl_pool_config_enter(tmpdp, FTAG); error = dsl_dataset_hold_obj_string(tmpdp, nvpair_name(pair), FTAG, &ds); if (error == 0) { char name[ZFS_MAX_DATASET_NAME_LEN]; dsl_dataset_name(ds, name); dsl_pool_config_exit(tmpdp, FTAG); dsl_dataset_rele(ds, FTAG); (void) zfs_unmount_snap(name); } else { dsl_pool_config_exit(tmpdp, FTAG); } } #endif } else { /* Non-temporary holds are specified by name. */ ddura.ddura_holdfunc = dsl_dataset_hold; pool = nvpair_name(pair); #ifdef _KERNEL for (pair = nvlist_next_nvpair(holds, NULL); pair != NULL; pair = nvlist_next_nvpair(holds, pair)) { (void) zfs_unmount_snap(nvpair_name(pair)); } #endif } ddura.ddura_holds = holds; ddura.ddura_errlist = errlist; ddura.ddura_todelete = fnvlist_alloc(); ddura.ddura_chkholds = fnvlist_alloc(); error = dsl_sync_task(pool, dsl_dataset_user_release_check, dsl_dataset_user_release_sync, &ddura, 0, ZFS_SPACE_CHECK_NONE); fnvlist_free(ddura.ddura_todelete); fnvlist_free(ddura.ddura_chkholds); return (error); } /* * holds is nvl of snapname -> { holdname, ... } * errlist will be filled in with snapname -> error */ int dsl_dataset_user_release(nvlist_t *holds, nvlist_t *errlist) { return (dsl_dataset_user_release_impl(holds, errlist, NULL)); } /* * holds is nvl of snapdsobj -> { holdname, ... } */ void dsl_dataset_user_release_tmp(struct dsl_pool *dp, nvlist_t *holds) { ASSERT(dp != NULL); (void) dsl_dataset_user_release_impl(holds, NULL, dp); } int dsl_dataset_get_holds(const char *dsname, nvlist_t *nvl) { dsl_pool_t *dp; dsl_dataset_t *ds; int err; err = dsl_pool_hold(dsname, FTAG, &dp); if (err != 0) return (err); err = dsl_dataset_hold(dp, dsname, FTAG, &ds); if (err != 0) { dsl_pool_rele(dp, FTAG); return (err); } if (dsl_dataset_phys(ds)->ds_userrefs_obj != 0) { zap_attribute_t *za; zap_cursor_t zc; za = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP); for (zap_cursor_init(&zc, ds->ds_dir->dd_pool->dp_meta_objset, dsl_dataset_phys(ds)->ds_userrefs_obj); zap_cursor_retrieve(&zc, za) == 0; zap_cursor_advance(&zc)) { fnvlist_add_uint64(nvl, za->za_name, za->za_first_integer); } zap_cursor_fini(&zc); kmem_free(za, sizeof (zap_attribute_t)); } dsl_dataset_rele(ds, FTAG); dsl_pool_rele(dp, FTAG); return (0); } Index: vendor-sys/illumos/dist/uts/common/fs/zfs/spa_errlog.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/spa_errlog.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/spa_errlog.c (revision 319948) @@ -1,406 +1,406 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2013, 2014 by Delphix. All rights reserved. */ /* * Routines to manage the on-disk persistent error log. * * Each pool stores a log of all logical data errors seen during normal * operation. This is actually the union of two distinct logs: the last log, * and the current log. All errors seen are logged to the current log. When a * scrub completes, the current log becomes the last log, the last log is thrown * out, and the current log is reinitialized. This way, if an error is somehow * corrected, a new scrub will show that that it no longer exists, and will be * deleted from the log when the scrub completes. * * The log is stored using a ZAP object whose key is a string form of the * zbookmark_phys tuple (objset, object, level, blkid), and whose contents is an * optional 'objset:object' human-readable string describing the data. When an * error is first logged, this string will be empty, indicating that no name is * known. This prevents us from having to issue a potentially large amount of * I/O to discover the object name during an error path. Instead, we do the * calculation when the data is requested, storing the result so future queries * will be faster. * * This log is then shipped into an nvlist where the key is the dataset name and * the value is the object name. Userland is then responsible for uniquifying * this list and displaying it to the user. */ #include #include #include #include #include /* * Convert a bookmark to a string. */ static void bookmark_to_name(zbookmark_phys_t *zb, char *buf, size_t len) { (void) snprintf(buf, len, "%llx:%llx:%llx:%llx", (u_longlong_t)zb->zb_objset, (u_longlong_t)zb->zb_object, (u_longlong_t)zb->zb_level, (u_longlong_t)zb->zb_blkid); } /* * Convert a string to a bookmark */ #ifdef _KERNEL static void name_to_bookmark(char *buf, zbookmark_phys_t *zb) { - zb->zb_objset = strtonum(buf, &buf); + zb->zb_objset = zfs_strtonum(buf, &buf); ASSERT(*buf == ':'); - zb->zb_object = strtonum(buf + 1, &buf); + zb->zb_object = zfs_strtonum(buf + 1, &buf); ASSERT(*buf == ':'); - zb->zb_level = (int)strtonum(buf + 1, &buf); + zb->zb_level = (int)zfs_strtonum(buf + 1, &buf); ASSERT(*buf == ':'); - zb->zb_blkid = strtonum(buf + 1, &buf); + zb->zb_blkid = zfs_strtonum(buf + 1, &buf); ASSERT(*buf == '\0'); } #endif /* * Log an uncorrectable error to the persistent error log. We add it to the * spa's list of pending errors. The changes are actually synced out to disk * during spa_errlog_sync(). */ void spa_log_error(spa_t *spa, zio_t *zio) { zbookmark_phys_t *zb = &zio->io_logical->io_bookmark; spa_error_entry_t search; spa_error_entry_t *new; avl_tree_t *tree; avl_index_t where; /* * If we are trying to import a pool, ignore any errors, as we won't be * writing to the pool any time soon. */ if (spa_load_state(spa) == SPA_LOAD_TRYIMPORT) return; mutex_enter(&spa->spa_errlist_lock); /* * If we have had a request to rotate the log, log it to the next list * instead of the current one. */ if (spa->spa_scrub_active || spa->spa_scrub_finished) tree = &spa->spa_errlist_scrub; else tree = &spa->spa_errlist_last; search.se_bookmark = *zb; if (avl_find(tree, &search, &where) != NULL) { mutex_exit(&spa->spa_errlist_lock); return; } new = kmem_zalloc(sizeof (spa_error_entry_t), KM_SLEEP); new->se_bookmark = *zb; avl_insert(tree, new, where); mutex_exit(&spa->spa_errlist_lock); } /* * Return the number of errors currently in the error log. This is actually the * sum of both the last log and the current log, since we don't know the union * of these logs until we reach userland. */ uint64_t spa_get_errlog_size(spa_t *spa) { uint64_t total = 0, count; mutex_enter(&spa->spa_errlog_lock); if (spa->spa_errlog_scrub != 0 && zap_count(spa->spa_meta_objset, spa->spa_errlog_scrub, &count) == 0) total += count; if (spa->spa_errlog_last != 0 && !spa->spa_scrub_finished && zap_count(spa->spa_meta_objset, spa->spa_errlog_last, &count) == 0) total += count; mutex_exit(&spa->spa_errlog_lock); mutex_enter(&spa->spa_errlist_lock); total += avl_numnodes(&spa->spa_errlist_last); total += avl_numnodes(&spa->spa_errlist_scrub); mutex_exit(&spa->spa_errlist_lock); return (total); } #ifdef _KERNEL static int process_error_log(spa_t *spa, uint64_t obj, void *addr, size_t *count) { zap_cursor_t zc; zap_attribute_t za; zbookmark_phys_t zb; if (obj == 0) return (0); for (zap_cursor_init(&zc, spa->spa_meta_objset, obj); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { if (*count == 0) { zap_cursor_fini(&zc); return (SET_ERROR(ENOMEM)); } name_to_bookmark(za.za_name, &zb); if (copyout(&zb, (char *)addr + (*count - 1) * sizeof (zbookmark_phys_t), sizeof (zbookmark_phys_t)) != 0) { zap_cursor_fini(&zc); return (SET_ERROR(EFAULT)); } *count -= 1; } zap_cursor_fini(&zc); return (0); } static int process_error_list(avl_tree_t *list, void *addr, size_t *count) { spa_error_entry_t *se; for (se = avl_first(list); se != NULL; se = AVL_NEXT(list, se)) { if (*count == 0) return (SET_ERROR(ENOMEM)); if (copyout(&se->se_bookmark, (char *)addr + (*count - 1) * sizeof (zbookmark_phys_t), sizeof (zbookmark_phys_t)) != 0) return (SET_ERROR(EFAULT)); *count -= 1; } return (0); } #endif /* * Copy all known errors to userland as an array of bookmarks. This is * actually a union of the on-disk last log and current log, as well as any * pending error requests. * * Because the act of reading the on-disk log could cause errors to be * generated, we have two separate locks: one for the error log and one for the * in-core error lists. We only need the error list lock to log and error, so * we grab the error log lock while we read the on-disk logs, and only pick up * the error list lock when we are finished. */ int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count) { int ret = 0; #ifdef _KERNEL mutex_enter(&spa->spa_errlog_lock); ret = process_error_log(spa, spa->spa_errlog_scrub, uaddr, count); if (!ret && !spa->spa_scrub_finished) ret = process_error_log(spa, spa->spa_errlog_last, uaddr, count); mutex_enter(&spa->spa_errlist_lock); if (!ret) ret = process_error_list(&spa->spa_errlist_scrub, uaddr, count); if (!ret) ret = process_error_list(&spa->spa_errlist_last, uaddr, count); mutex_exit(&spa->spa_errlist_lock); mutex_exit(&spa->spa_errlog_lock); #endif return (ret); } /* * Called when a scrub completes. This simply set a bit which tells which AVL * tree to add new errors. spa_errlog_sync() is responsible for actually * syncing the changes to the underlying objects. */ void spa_errlog_rotate(spa_t *spa) { mutex_enter(&spa->spa_errlist_lock); spa->spa_scrub_finished = B_TRUE; mutex_exit(&spa->spa_errlist_lock); } /* * Discard any pending errors from the spa_t. Called when unloading a faulted * pool, as the errors encountered during the open cannot be synced to disk. */ void spa_errlog_drain(spa_t *spa) { spa_error_entry_t *se; void *cookie; mutex_enter(&spa->spa_errlist_lock); cookie = NULL; while ((se = avl_destroy_nodes(&spa->spa_errlist_last, &cookie)) != NULL) kmem_free(se, sizeof (spa_error_entry_t)); cookie = NULL; while ((se = avl_destroy_nodes(&spa->spa_errlist_scrub, &cookie)) != NULL) kmem_free(se, sizeof (spa_error_entry_t)); mutex_exit(&spa->spa_errlist_lock); } /* * Process a list of errors into the current on-disk log. */ static void sync_error_list(spa_t *spa, avl_tree_t *t, uint64_t *obj, dmu_tx_t *tx) { spa_error_entry_t *se; char buf[64]; void *cookie; if (avl_numnodes(t) != 0) { /* create log if necessary */ if (*obj == 0) *obj = zap_create(spa->spa_meta_objset, DMU_OT_ERROR_LOG, DMU_OT_NONE, 0, tx); /* add errors to the current log */ for (se = avl_first(t); se != NULL; se = AVL_NEXT(t, se)) { char *name = se->se_name ? se->se_name : ""; bookmark_to_name(&se->se_bookmark, buf, sizeof (buf)); (void) zap_update(spa->spa_meta_objset, *obj, buf, 1, strlen(name) + 1, name, tx); } /* purge the error list */ cookie = NULL; while ((se = avl_destroy_nodes(t, &cookie)) != NULL) kmem_free(se, sizeof (spa_error_entry_t)); } } /* * Sync the error log out to disk. This is a little tricky because the act of * writing the error log requires the spa_errlist_lock. So, we need to lock the * error lists, take a copy of the lists, and then reinitialize them. Then, we * drop the error list lock and take the error log lock, at which point we * do the errlog processing. Then, if we encounter an I/O error during this * process, we can successfully add the error to the list. Note that this will * result in the perpetual recycling of errors, but it is an unlikely situation * and not a performance critical operation. */ void spa_errlog_sync(spa_t *spa, uint64_t txg) { dmu_tx_t *tx; avl_tree_t scrub, last; int scrub_finished; mutex_enter(&spa->spa_errlist_lock); /* * Bail out early under normal circumstances. */ if (avl_numnodes(&spa->spa_errlist_scrub) == 0 && avl_numnodes(&spa->spa_errlist_last) == 0 && !spa->spa_scrub_finished) { mutex_exit(&spa->spa_errlist_lock); return; } spa_get_errlists(spa, &last, &scrub); scrub_finished = spa->spa_scrub_finished; spa->spa_scrub_finished = B_FALSE; mutex_exit(&spa->spa_errlist_lock); mutex_enter(&spa->spa_errlog_lock); tx = dmu_tx_create_assigned(spa->spa_dsl_pool, txg); /* * Sync out the current list of errors. */ sync_error_list(spa, &last, &spa->spa_errlog_last, tx); /* * Rotate the log if necessary. */ if (scrub_finished) { if (spa->spa_errlog_last != 0) VERIFY(dmu_object_free(spa->spa_meta_objset, spa->spa_errlog_last, tx) == 0); spa->spa_errlog_last = spa->spa_errlog_scrub; spa->spa_errlog_scrub = 0; sync_error_list(spa, &scrub, &spa->spa_errlog_last, tx); } /* * Sync out any pending scrub errors. */ sync_error_list(spa, &scrub, &spa->spa_errlog_scrub, tx); /* * Update the MOS to reflect the new values. */ (void) zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ERRLOG_LAST, sizeof (uint64_t), 1, &spa->spa_errlog_last, tx); (void) zap_update(spa->spa_meta_objset, DMU_POOL_DIRECTORY_OBJECT, DMU_POOL_ERRLOG_SCRUB, sizeof (uint64_t), 1, &spa->spa_errlog_scrub, tx); dmu_tx_commit(tx); mutex_exit(&spa->spa_errlog_lock); } Index: vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c (revision 319948) @@ -1,2058 +1,2058 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2017 by Delphix. All rights reserved. * Copyright 2015 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Integros [integros.com] */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_prop.h" #include /* * SPA locking * * There are four basic locks for managing spa_t structures: * * spa_namespace_lock (global mutex) * * This lock must be acquired to do any of the following: * * - Lookup a spa_t by name * - Add or remove a spa_t from the namespace * - Increase spa_refcount from non-zero * - Check if spa_refcount is zero * - Rename a spa_t * - add/remove/attach/detach devices * - Held for the duration of create/destroy/import/export * * It does not need to handle recursion. A create or destroy may * reference objects (files or zvols) in other pools, but by * definition they must have an existing reference, and will never need * to lookup a spa_t by name. * * spa_refcount (per-spa refcount_t protected by mutex) * * This reference count keep track of any active users of the spa_t. The * spa_t cannot be destroyed or freed while this is non-zero. Internally, * the refcount is never really 'zero' - opening a pool implicitly keeps * some references in the DMU. Internally we check against spa_minref, but * present the image of a zero/non-zero value to consumers. * * spa_config_lock[] (per-spa array of rwlocks) * * This protects the spa_t from config changes, and must be held in * the following circumstances: * * - RW_READER to perform I/O to the spa * - RW_WRITER to change the vdev config * * The locking order is fairly straightforward: * * spa_namespace_lock -> spa_refcount * * The namespace lock must be acquired to increase the refcount from 0 * or to check if it is zero. * * spa_refcount -> spa_config_lock[] * * There must be at least one valid reference on the spa_t to acquire * the config lock. * * spa_namespace_lock -> spa_config_lock[] * * The namespace lock must always be taken before the config lock. * * * The spa_namespace_lock can be acquired directly and is globally visible. * * The namespace is manipulated using the following functions, all of which * require the spa_namespace_lock to be held. * * spa_lookup() Lookup a spa_t by name. * * spa_add() Create a new spa_t in the namespace. * * spa_remove() Remove a spa_t from the namespace. This also * frees up any memory associated with the spa_t. * * spa_next() Returns the next spa_t in the system, or the * first if NULL is passed. * * spa_evict_all() Shutdown and remove all spa_t structures in * the system. * * spa_guid_exists() Determine whether a pool/device guid exists. * * The spa_refcount is manipulated using the following functions: * * spa_open_ref() Adds a reference to the given spa_t. Must be * called with spa_namespace_lock held if the * refcount is currently zero. * * spa_close() Remove a reference from the spa_t. This will * not free the spa_t or remove it from the * namespace. No locking is required. * * spa_refcount_zero() Returns true if the refcount is currently * zero. Must be called with spa_namespace_lock * held. * * The spa_config_lock[] is an array of rwlocks, ordered as follows: * SCL_CONFIG > SCL_STATE > SCL_ALLOC > SCL_ZIO > SCL_FREE > SCL_VDEV. * spa_config_lock[] is manipulated with spa_config_{enter,exit,held}(). * * To read the configuration, it suffices to hold one of these locks as reader. * To modify the configuration, you must hold all locks as writer. To modify * vdev state without altering the vdev tree's topology (e.g. online/offline), * you must hold SCL_STATE and SCL_ZIO as writer. * * We use these distinct config locks to avoid recursive lock entry. * For example, spa_sync() (which holds SCL_CONFIG as reader) induces * block allocations (SCL_ALLOC), which may require reading space maps * from disk (dmu_read() -> zio_read() -> SCL_ZIO). * * The spa config locks cannot be normal rwlocks because we need the * ability to hand off ownership. For example, SCL_ZIO is acquired * by the issuing thread and later released by an interrupt thread. * They do, however, obey the usual write-wanted semantics to prevent * writer (i.e. system administrator) starvation. * * The lock acquisition rules are as follows: * * SCL_CONFIG * Protects changes to the vdev tree topology, such as vdev * add/remove/attach/detach. Protects the dirty config list * (spa_config_dirty_list) and the set of spares and l2arc devices. * * SCL_STATE * Protects changes to pool state and vdev state, such as vdev * online/offline/fault/degrade/clear. Protects the dirty state list * (spa_state_dirty_list) and global pool state (spa_state). * * SCL_ALLOC * Protects changes to metaslab groups and classes. * Held as reader by metaslab_alloc() and metaslab_claim(). * * SCL_ZIO * Held by bp-level zios (those which have no io_vd upon entry) * to prevent changes to the vdev tree. The bp-level zio implicitly * protects all of its vdev child zios, which do not hold SCL_ZIO. * * SCL_FREE * Protects changes to metaslab groups and classes. * Held as reader by metaslab_free(). SCL_FREE is distinct from * SCL_ALLOC, and lower than SCL_ZIO, so that we can safely free * blocks in zio_done() while another i/o that holds either * SCL_ALLOC or SCL_ZIO is waiting for this i/o to complete. * * SCL_VDEV * Held as reader to prevent changes to the vdev tree during trivial * inquiries such as bp_get_dsize(). SCL_VDEV is distinct from the * other locks, and lower than all of them, to ensure that it's safe * to acquire regardless of caller context. * * In addition, the following rules apply: * * (a) spa_props_lock protects pool properties, spa_config and spa_config_list. * The lock ordering is SCL_CONFIG > spa_props_lock. * * (b) I/O operations on leaf vdevs. For any zio operation that takes * an explicit vdev_t argument -- such as zio_ioctl(), zio_read_phys(), * or zio_write_phys() -- the caller must ensure that the config cannot * cannot change in the interim, and that the vdev cannot be reopened. * SCL_STATE as reader suffices for both. * * The vdev configuration is protected by spa_vdev_enter() / spa_vdev_exit(). * * spa_vdev_enter() Acquire the namespace lock and the config lock * for writing. * * spa_vdev_exit() Release the config lock, wait for all I/O * to complete, sync the updated configs to the * cache, and release the namespace lock. * * vdev state is protected by spa_vdev_state_enter() / spa_vdev_state_exit(). * Like spa_vdev_enter/exit, these are convenience wrappers -- the actual * locking is, always, based on spa_namespace_lock and spa_config_lock[]. * * spa_rename() is also implemented within this file since it requires * manipulation of the namespace. */ static avl_tree_t spa_namespace_avl; kmutex_t spa_namespace_lock; static kcondvar_t spa_namespace_cv; static int spa_active_count; int spa_max_replication_override = SPA_DVAS_PER_BP; static kmutex_t spa_spare_lock; static avl_tree_t spa_spare_avl; static kmutex_t spa_l2cache_lock; static avl_tree_t spa_l2cache_avl; kmem_cache_t *spa_buffer_pool; int spa_mode_global; #ifdef ZFS_DEBUG /* Everything except dprintf and spa is on by default in debug builds */ int zfs_flags = ~(ZFS_DEBUG_DPRINTF | ZFS_DEBUG_SPA); #else int zfs_flags = 0; #endif /* * zfs_recover can be set to nonzero to attempt to recover from * otherwise-fatal errors, typically caused by on-disk corruption. When * set, calls to zfs_panic_recover() will turn into warning messages. * This should only be used as a last resort, as it typically results * in leaked space, or worse. */ boolean_t zfs_recover = B_FALSE; /* * If destroy encounters an EIO while reading metadata (e.g. indirect * blocks), space referenced by the missing metadata can not be freed. * Normally this causes the background destroy to become "stalled", as * it is unable to make forward progress. While in this stalled state, * all remaining space to free from the error-encountering filesystem is * "temporarily leaked". Set this flag to cause it to ignore the EIO, * permanently leak the space from indirect blocks that can not be read, * and continue to free everything else that it can. * * The default, "stalling" behavior is useful if the storage partially * fails (i.e. some but not all i/os fail), and then later recovers. In * this case, we will be able to continue pool operations while it is * partially failed, and when it recovers, we can continue to free the * space, with no leaks. However, note that this case is actually * fairly rare. * * Typically pools either (a) fail completely (but perhaps temporarily, * e.g. a top-level vdev going offline), or (b) have localized, * permanent errors (e.g. disk returns the wrong data due to bit flip or * firmware bug). In case (a), this setting does not matter because the * pool will be suspended and the sync thread will not be able to make * forward progress regardless. In case (b), because the error is * permanent, the best we can do is leak the minimum amount of space, * which is what setting this flag will do. Therefore, it is reasonable * for this flag to normally be set, but we chose the more conservative * approach of not setting it, so that there is no possibility of * leaking space in the "partial temporary" failure case. */ boolean_t zfs_free_leak_on_eio = B_FALSE; /* * Expiration time in milliseconds. This value has two meanings. First it is * used to determine when the spa_deadman() logic should fire. By default the * spa_deadman() will fire if spa_sync() has not completed in 1000 seconds. * Secondly, the value determines if an I/O is considered "hung". Any I/O that * has not completed in zfs_deadman_synctime_ms is considered "hung" resulting * in a system panic. */ uint64_t zfs_deadman_synctime_ms = 1000000ULL; /* * Check time in milliseconds. This defines the frequency at which we check * for hung I/O. */ uint64_t zfs_deadman_checktime_ms = 5000ULL; /* * Override the zfs deadman behavior via /etc/system. By default the * deadman is enabled except on VMware and sparc deployments. */ int zfs_deadman_enabled = -1; /* * The worst case is single-sector max-parity RAID-Z blocks, in which * case the space requirement is exactly (VDEV_RAIDZ_MAXPARITY + 1) * times the size; so just assume that. Add to this the fact that * we can have up to 3 DVAs per bp, and one more factor of 2 because * the block may be dittoed with up to 3 DVAs by ddt_sync(). All together, * the worst case is: * (VDEV_RAIDZ_MAXPARITY + 1) * SPA_DVAS_PER_BP * 2 == 24 */ int spa_asize_inflation = 24; /* * Normally, we don't allow the last 3.2% (1/(2^spa_slop_shift)) of space in * the pool to be consumed. This ensures that we don't run the pool * completely out of space, due to unaccounted changes (e.g. to the MOS). * It also limits the worst-case time to allocate space. If we have * less than this amount of free space, most ZPL operations (e.g. write, * create) will return ENOSPC. * * Certain operations (e.g. file removal, most administrative actions) can * use half the slop space. They will only return ENOSPC if less than half * the slop space is free. Typically, once the pool has less than the slop * space free, the user will use these operations to free up space in the pool. * These are the operations that call dsl_pool_adjustedsize() with the netfree * argument set to TRUE. * * A very restricted set of operations are always permitted, regardless of * the amount of free space. These are the operations that call * dsl_sync_task(ZFS_SPACE_CHECK_NONE), e.g. "zfs destroy". If these * operations result in a net increase in the amount of space used, * it is possible to run the pool completely out of space, causing it to * be permanently read-only. * * Note that on very small pools, the slop space will be larger than * 3.2%, in an effort to have it be at least spa_min_slop (128MB), * but we never allow it to be more than half the pool size. * * See also the comments in zfs_space_check_t. */ int spa_slop_shift = 5; uint64_t spa_min_slop = 128 * 1024 * 1024; /* * ========================================================================== * SPA config locking * ========================================================================== */ static void spa_config_lock_init(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_init(&scl->scl_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&scl->scl_cv, NULL, CV_DEFAULT, NULL); refcount_create_untracked(&scl->scl_count); scl->scl_writer = NULL; scl->scl_write_wanted = 0; } } static void spa_config_lock_destroy(spa_t *spa) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; mutex_destroy(&scl->scl_lock); cv_destroy(&scl->scl_cv); refcount_destroy(&scl->scl_count); ASSERT(scl->scl_writer == NULL); ASSERT(scl->scl_write_wanted == 0); } } int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw) { for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { if (scl->scl_writer || scl->scl_write_wanted) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } } else { ASSERT(scl->scl_writer != curthread); if (!refcount_is_zero(&scl->scl_count)) { mutex_exit(&scl->scl_lock); spa_config_exit(spa, locks & ((1 << i) - 1), tag); return (0); } scl->scl_writer = curthread; } (void) refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } return (1); } void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw) { int wlocks_held = 0; ASSERT3U(SCL_LOCKS, <, sizeof (wlocks_held) * NBBY); for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (scl->scl_writer == curthread) wlocks_held |= (1 << i); if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); if (rw == RW_READER) { while (scl->scl_writer || scl->scl_write_wanted) { cv_wait(&scl->scl_cv, &scl->scl_lock); } } else { ASSERT(scl->scl_writer != curthread); while (!refcount_is_zero(&scl->scl_count)) { scl->scl_write_wanted++; cv_wait(&scl->scl_cv, &scl->scl_lock); scl->scl_write_wanted--; } scl->scl_writer = curthread; } (void) refcount_add(&scl->scl_count, tag); mutex_exit(&scl->scl_lock); } ASSERT(wlocks_held <= locks); } void spa_config_exit(spa_t *spa, int locks, void *tag) { for (int i = SCL_LOCKS - 1; i >= 0; i--) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; mutex_enter(&scl->scl_lock); ASSERT(!refcount_is_zero(&scl->scl_count)); if (refcount_remove(&scl->scl_count, tag) == 0) { ASSERT(scl->scl_writer == NULL || scl->scl_writer == curthread); scl->scl_writer = NULL; /* OK in either case */ cv_broadcast(&scl->scl_cv); } mutex_exit(&scl->scl_lock); } } int spa_config_held(spa_t *spa, int locks, krw_t rw) { int locks_held = 0; for (int i = 0; i < SCL_LOCKS; i++) { spa_config_lock_t *scl = &spa->spa_config_lock[i]; if (!(locks & (1 << i))) continue; if ((rw == RW_READER && !refcount_is_zero(&scl->scl_count)) || (rw == RW_WRITER && scl->scl_writer == curthread)) locks_held |= 1 << i; } return (locks_held); } /* * ========================================================================== * SPA namespace functions * ========================================================================== */ /* * Lookup the named spa_t in the AVL tree. The spa_namespace_lock must be held. * Returns NULL if no matching spa_t is found. */ spa_t * spa_lookup(const char *name) { static spa_t search; /* spa_t is large; don't allocate on stack */ spa_t *spa; avl_index_t where; char *cp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); (void) strlcpy(search.spa_name, name, sizeof (search.spa_name)); /* * If it's a full dataset name, figure out the pool name and * just use that. */ cp = strpbrk(search.spa_name, "/@#"); if (cp != NULL) *cp = '\0'; spa = avl_find(&spa_namespace_avl, &search, &where); return (spa); } /* * Fires when spa_sync has not completed within zfs_deadman_synctime_ms. * If the zfs_deadman_enabled flag is set then it inspects all vdev queues * looking for potentially hung I/Os. */ void spa_deadman(void *arg) { spa_t *spa = arg; /* * Disable the deadman timer if the pool is suspended. */ if (spa_suspended(spa)) { VERIFY(cyclic_reprogram(spa->spa_deadman_cycid, CY_INFINITY)); return; } zfs_dbgmsg("slow spa_sync: started %llu seconds ago, calls %llu", (gethrtime() - spa->spa_sync_starttime) / NANOSEC, ++spa->spa_deadman_calls); if (zfs_deadman_enabled) vdev_deadman(spa->spa_root_vdev); } /* * Create an uninitialized spa_t with the given name. Requires * spa_namespace_lock. The caller must ensure that the spa_t doesn't already * exist by calling spa_lookup() first. */ spa_t * spa_add(const char *name, nvlist_t *config, const char *altroot) { spa_t *spa; spa_config_dirent_t *dp; cyc_handler_t hdlr; cyc_time_t when; ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa = kmem_zalloc(sizeof (spa_t), KM_SLEEP); mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_evicting_os_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_history_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_proc_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_cksum_tmpls_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_suspend_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_vdev_top_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_iokstat_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa->spa_alloc_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_evicting_os_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_proc_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL); cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL); for (int t = 0; t < TXG_SIZE; t++) bplist_create(&spa->spa_free_bplist[t]); (void) strlcpy(spa->spa_name, name, sizeof (spa->spa_name)); spa->spa_state = POOL_STATE_UNINITIALIZED; spa->spa_freeze_txg = UINT64_MAX; spa->spa_final_txg = UINT64_MAX; spa->spa_load_max_txg = UINT64_MAX; spa->spa_proc = &p0; spa->spa_proc_state = SPA_PROC_NONE; hdlr.cyh_func = spa_deadman; hdlr.cyh_arg = spa; hdlr.cyh_level = CY_LOW_LEVEL; spa->spa_deadman_synctime = MSEC2NSEC(zfs_deadman_synctime_ms); /* * This determines how often we need to check for hung I/Os after * the cyclic has already fired. Since checking for hung I/Os is * an expensive operation we don't want to check too frequently. * Instead wait for 5 seconds before checking again. */ when.cyt_interval = MSEC2NSEC(zfs_deadman_checktime_ms); when.cyt_when = CY_INFINITY; mutex_enter(&cpu_lock); spa->spa_deadman_cycid = cyclic_add(&hdlr, &when); mutex_exit(&cpu_lock); refcount_create(&spa->spa_refcount); spa_config_lock_init(spa); avl_add(&spa_namespace_avl, spa); /* * Set the alternate root, if there is one. */ if (altroot) { spa->spa_root = spa_strdup(altroot); spa_active_count++; } avl_create(&spa->spa_alloc_tree, zio_bookmark_compare, sizeof (zio_t), offsetof(zio_t, io_alloc_node)); /* * Every pool starts with the default cachefile */ list_create(&spa->spa_config_list, sizeof (spa_config_dirent_t), offsetof(spa_config_dirent_t, scd_link)); dp = kmem_zalloc(sizeof (spa_config_dirent_t), KM_SLEEP); dp->scd_path = altroot ? NULL : spa_strdup(spa_config_path); list_insert_head(&spa->spa_config_list, dp); VERIFY(nvlist_alloc(&spa->spa_load_info, NV_UNIQUE_NAME, KM_SLEEP) == 0); if (config != NULL) { nvlist_t *features; if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, &features) == 0) { VERIFY(nvlist_dup(features, &spa->spa_label_features, 0) == 0); } VERIFY(nvlist_dup(config, &spa->spa_config, 0) == 0); } if (spa->spa_label_features == NULL) { VERIFY(nvlist_alloc(&spa->spa_label_features, NV_UNIQUE_NAME, KM_SLEEP) == 0); } spa->spa_iokstat = kstat_create("zfs", 0, name, "disk", KSTAT_TYPE_IO, 1, 0); if (spa->spa_iokstat) { spa->spa_iokstat->ks_lock = &spa->spa_iokstat_lock; kstat_install(spa->spa_iokstat); } spa->spa_debug = ((zfs_flags & ZFS_DEBUG_SPA) != 0); spa->spa_min_ashift = INT_MAX; spa->spa_max_ashift = 0; /* * As a pool is being created, treat all features as disabled by * setting SPA_FEATURE_DISABLED for all entries in the feature * refcount cache. */ for (int i = 0; i < SPA_FEATURES; i++) { spa->spa_feat_refcount_cache[i] = SPA_FEATURE_DISABLED; } return (spa); } /* * Removes a spa_t from the namespace, freeing up any memory used. Requires * spa_namespace_lock. This is called only after the spa_t has been closed and * deactivated. */ void spa_remove(spa_t *spa) { spa_config_dirent_t *dp; ASSERT(MUTEX_HELD(&spa_namespace_lock)); ASSERT(spa->spa_state == POOL_STATE_UNINITIALIZED); ASSERT3U(refcount_count(&spa->spa_refcount), ==, 0); nvlist_free(spa->spa_config_splitting); avl_remove(&spa_namespace_avl, spa); cv_broadcast(&spa_namespace_cv); if (spa->spa_root) { spa_strfree(spa->spa_root); spa_active_count--; } while ((dp = list_head(&spa->spa_config_list)) != NULL) { list_remove(&spa->spa_config_list, dp); if (dp->scd_path != NULL) spa_strfree(dp->scd_path); kmem_free(dp, sizeof (spa_config_dirent_t)); } avl_destroy(&spa->spa_alloc_tree); list_destroy(&spa->spa_config_list); nvlist_free(spa->spa_label_features); nvlist_free(spa->spa_load_info); spa_config_set(spa, NULL); mutex_enter(&cpu_lock); if (spa->spa_deadman_cycid != CYCLIC_NONE) cyclic_remove(spa->spa_deadman_cycid); mutex_exit(&cpu_lock); spa->spa_deadman_cycid = CYCLIC_NONE; refcount_destroy(&spa->spa_refcount); spa_config_lock_destroy(spa); kstat_delete(spa->spa_iokstat); spa->spa_iokstat = NULL; for (int t = 0; t < TXG_SIZE; t++) bplist_destroy(&spa->spa_free_bplist[t]); zio_checksum_templates_free(spa); cv_destroy(&spa->spa_async_cv); cv_destroy(&spa->spa_evicting_os_cv); cv_destroy(&spa->spa_proc_cv); cv_destroy(&spa->spa_scrub_io_cv); cv_destroy(&spa->spa_suspend_cv); mutex_destroy(&spa->spa_alloc_lock); mutex_destroy(&spa->spa_async_lock); mutex_destroy(&spa->spa_errlist_lock); mutex_destroy(&spa->spa_errlog_lock); mutex_destroy(&spa->spa_evicting_os_lock); mutex_destroy(&spa->spa_history_lock); mutex_destroy(&spa->spa_proc_lock); mutex_destroy(&spa->spa_props_lock); mutex_destroy(&spa->spa_cksum_tmpls_lock); mutex_destroy(&spa->spa_scrub_lock); mutex_destroy(&spa->spa_suspend_lock); mutex_destroy(&spa->spa_vdev_top_lock); mutex_destroy(&spa->spa_iokstat_lock); kmem_free(spa, sizeof (spa_t)); } /* * Given a pool, return the next pool in the namespace, or NULL if there is * none. If 'prev' is NULL, return the first pool. */ spa_t * spa_next(spa_t *prev) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (prev) return (AVL_NEXT(&spa_namespace_avl, prev)); else return (avl_first(&spa_namespace_avl)); } /* * ========================================================================== * SPA refcount functions * ========================================================================== */ /* * Add a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_open_ref(spa_t *spa, void *tag) { ASSERT(refcount_count(&spa->spa_refcount) >= spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) refcount_add(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t. Must have at least one reference, or * have the namespace lock held. */ void spa_close(spa_t *spa, void *tag) { ASSERT(refcount_count(&spa->spa_refcount) > spa->spa_minref || MUTEX_HELD(&spa_namespace_lock)); (void) refcount_remove(&spa->spa_refcount, tag); } /* * Remove a reference to the given spa_t held by a dsl dir that is * being asynchronously released. Async releases occur from a taskq * performing eviction of dsl datasets and dirs. The namespace lock * isn't held and the hold by the object being evicted may contribute to * spa_minref (e.g. dataset or directory released during pool export), * so the asserts in spa_close() do not apply. */ void spa_async_close(spa_t *spa, void *tag) { (void) refcount_remove(&spa->spa_refcount, tag); } /* * Check to see if the spa refcount is zero. Must be called with * spa_namespace_lock held. We really compare against spa_minref, which is the * number of references acquired when opening a pool */ boolean_t spa_refcount_zero(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); return (refcount_count(&spa->spa_refcount) == spa->spa_minref); } /* * ========================================================================== * SPA spare and l2cache tracking * ========================================================================== */ /* * Hot spares and cache devices are tracked using the same code below, * for 'auxiliary' devices. */ typedef struct spa_aux { uint64_t aux_guid; uint64_t aux_pool; avl_node_t aux_avl; int aux_count; } spa_aux_t; static int spa_aux_compare(const void *a, const void *b) { const spa_aux_t *sa = a; const spa_aux_t *sb = b; if (sa->aux_guid < sb->aux_guid) return (-1); else if (sa->aux_guid > sb->aux_guid) return (1); else return (0); } void spa_aux_add(vdev_t *vd, avl_tree_t *avl) { avl_index_t where; spa_aux_t search; spa_aux_t *aux; search.aux_guid = vd->vdev_guid; if ((aux = avl_find(avl, &search, &where)) != NULL) { aux->aux_count++; } else { aux = kmem_zalloc(sizeof (spa_aux_t), KM_SLEEP); aux->aux_guid = vd->vdev_guid; aux->aux_count = 1; avl_insert(avl, aux, where); } } void spa_aux_remove(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search; spa_aux_t *aux; avl_index_t where; search.aux_guid = vd->vdev_guid; aux = avl_find(avl, &search, &where); ASSERT(aux != NULL); if (--aux->aux_count == 0) { avl_remove(avl, aux); kmem_free(aux, sizeof (spa_aux_t)); } else if (aux->aux_pool == spa_guid(vd->vdev_spa)) { aux->aux_pool = 0ULL; } } boolean_t spa_aux_exists(uint64_t guid, uint64_t *pool, int *refcnt, avl_tree_t *avl) { spa_aux_t search, *found; search.aux_guid = guid; found = avl_find(avl, &search, NULL); if (pool) { if (found) *pool = found->aux_pool; else *pool = 0ULL; } if (refcnt) { if (found) *refcnt = found->aux_count; else *refcnt = 0; } return (found != NULL); } void spa_aux_activate(vdev_t *vd, avl_tree_t *avl) { spa_aux_t search, *found; avl_index_t where; search.aux_guid = vd->vdev_guid; found = avl_find(avl, &search, &where); ASSERT(found != NULL); ASSERT(found->aux_pool == 0ULL); found->aux_pool = spa_guid(vd->vdev_spa); } /* * Spares are tracked globally due to the following constraints: * * - A spare may be part of multiple pools. * - A spare may be added to a pool even if it's actively in use within * another pool. * - A spare in use in any pool can only be the source of a replacement if * the target is a spare in the same pool. * * We keep track of all spares on the system through the use of a reference * counted AVL tree. When a vdev is added as a spare, or used as a replacement * spare, then we bump the reference count in the AVL tree. In addition, we set * the 'vdev_isspare' member to indicate that the device is a spare (active or * inactive). When a spare is made active (used to replace a device in the * pool), we also keep track of which pool its been made a part of. * * The 'spa_spare_lock' protects the AVL tree. These functions are normally * called under the spa_namespace lock as part of vdev reconfiguration. The * separate spare lock exists for the status query path, which does not need to * be completely consistent with respect to other vdev configuration changes. */ static int spa_spare_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_spare_add(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(!vd->vdev_isspare); spa_aux_add(vd, &spa_spare_avl); vd->vdev_isspare = B_TRUE; mutex_exit(&spa_spare_lock); } void spa_spare_remove(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_remove(vd, &spa_spare_avl); vd->vdev_isspare = B_FALSE; mutex_exit(&spa_spare_lock); } boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt) { boolean_t found; mutex_enter(&spa_spare_lock); found = spa_aux_exists(guid, pool, refcnt, &spa_spare_avl); mutex_exit(&spa_spare_lock); return (found); } void spa_spare_activate(vdev_t *vd) { mutex_enter(&spa_spare_lock); ASSERT(vd->vdev_isspare); spa_aux_activate(vd, &spa_spare_avl); mutex_exit(&spa_spare_lock); } /* * Level 2 ARC devices are tracked globally for the same reasons as spares. * Cache devices currently only support one pool per cache device, and so * for these devices the aux reference count is currently unused beyond 1. */ static int spa_l2cache_compare(const void *a, const void *b) { return (spa_aux_compare(a, b)); } void spa_l2cache_add(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(!vd->vdev_isl2cache); spa_aux_add(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_TRUE; mutex_exit(&spa_l2cache_lock); } void spa_l2cache_remove(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_remove(vd, &spa_l2cache_avl); vd->vdev_isl2cache = B_FALSE; mutex_exit(&spa_l2cache_lock); } boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool) { boolean_t found; mutex_enter(&spa_l2cache_lock); found = spa_aux_exists(guid, pool, NULL, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); return (found); } void spa_l2cache_activate(vdev_t *vd) { mutex_enter(&spa_l2cache_lock); ASSERT(vd->vdev_isl2cache); spa_aux_activate(vd, &spa_l2cache_avl); mutex_exit(&spa_l2cache_lock); } /* * ========================================================================== * SPA vdev locking * ========================================================================== */ /* * Lock the given spa_t for the purpose of adding or removing a vdev. * Grabs the global spa_namespace_lock plus the spa config lock for writing. * It returns the next transaction group for the spa_t. */ uint64_t spa_vdev_enter(spa_t *spa) { mutex_enter(&spa->spa_vdev_top_lock); mutex_enter(&spa_namespace_lock); return (spa_vdev_config_enter(spa)); } /* * Internal implementation for spa_vdev_enter(). Used when a vdev * operation requires multiple syncs (i.e. removing a device) while * keeping the spa_namespace_lock held. */ uint64_t spa_vdev_config_enter(spa_t *spa) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); return (spa_last_synced_txg(spa) + 1); } /* * Used in combination with spa_vdev_config_enter() to allow the syncing * of multiple transactions without releasing the spa_namespace_lock. */ void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag) { ASSERT(MUTEX_HELD(&spa_namespace_lock)); int config_changed = B_FALSE; ASSERT(txg > spa_last_synced_txg(spa)); spa->spa_pending_vdev = NULL; /* * Reassess the DTLs. */ vdev_dtl_reassess(spa->spa_root_vdev, 0, 0, B_FALSE); if (error == 0 && !list_is_empty(&spa->spa_config_dirty_list)) { config_changed = B_TRUE; spa->spa_config_generation++; } /* * Verify the metaslab classes. */ ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0); ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0); spa_config_exit(spa, SCL_ALL, spa); /* * Panic the system if the specified tag requires it. This * is useful for ensuring that configurations are updated * transactionally. */ if (zio_injection_enabled) zio_handle_panic_injection(spa, tag, 0); /* * Note: this txg_wait_synced() is important because it ensures * that there won't be more than one config change per txg. * This allows us to use the txg as the generation number. */ if (error == 0) txg_wait_synced(spa->spa_dsl_pool, txg); if (vd != NULL) { ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL); spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); vdev_free(vd); spa_config_exit(spa, SCL_ALL, spa); } /* * If the config changed, update the config cache. */ if (config_changed) spa_config_sync(spa, B_FALSE, B_TRUE); } /* * Unlock the spa_t after adding or removing a vdev. Besides undoing the * locking of spa_vdev_enter(), we also want make sure the transactions have * synced to disk, and then update the global configuration cache with the new * information. */ int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error) { spa_vdev_config_exit(spa, vd, txg, error, FTAG); mutex_exit(&spa_namespace_lock); mutex_exit(&spa->spa_vdev_top_lock); return (error); } /* * Lock the given spa_t for the purpose of changing vdev state. */ void spa_vdev_state_enter(spa_t *spa, int oplocks) { int locks = SCL_STATE_ALL | oplocks; /* * Root pools may need to read of the underlying devfs filesystem * when opening up a vdev. Unfortunately if we're holding the * SCL_ZIO lock it will result in a deadlock when we try to issue * the read from the root filesystem. Instead we "prefetch" * the associated vnodes that we need prior to opening the * underlying devices and cache them so that we can prevent * any I/O when we are doing the actual open. */ if (spa_is_root(spa)) { int low = locks & ~(SCL_ZIO - 1); int high = locks & ~low; spa_config_enter(spa, high, spa, RW_WRITER); vdev_hold(spa->spa_root_vdev); spa_config_enter(spa, low, spa, RW_WRITER); } else { spa_config_enter(spa, locks, spa, RW_WRITER); } spa->spa_vdev_locks = locks; } int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error) { boolean_t config_changed = B_FALSE; if (vd != NULL || error == 0) vdev_dtl_reassess(vd ? vd->vdev_top : spa->spa_root_vdev, 0, 0, B_FALSE); if (vd != NULL) { vdev_state_dirty(vd->vdev_top); config_changed = B_TRUE; spa->spa_config_generation++; } if (spa_is_root(spa)) vdev_rele(spa->spa_root_vdev); ASSERT3U(spa->spa_vdev_locks, >=, SCL_STATE_ALL); spa_config_exit(spa, spa->spa_vdev_locks, spa); /* * If anything changed, wait for it to sync. This ensures that, * from the system administrator's perspective, zpool(1M) commands * are synchronous. This is important for things like zpool offline: * when the command completes, you expect no further I/O from ZFS. */ if (vd != NULL) txg_wait_synced(spa->spa_dsl_pool, 0); /* * If the config changed, update the config cache. */ if (config_changed) { mutex_enter(&spa_namespace_lock); spa_config_sync(spa, B_FALSE, B_TRUE); mutex_exit(&spa_namespace_lock); } return (error); } /* * ========================================================================== * Miscellaneous functions * ========================================================================== */ void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx) { if (!nvlist_exists(spa->spa_label_features, feature)) { fnvlist_add_boolean(spa->spa_label_features, feature); /* * When we are creating the pool (tx_txg==TXG_INITIAL), we can't * dirty the vdev config because lock SCL_CONFIG is not held. * Thankfully, in this case we don't need to dirty the config * because it will be written out anyway when we finish * creating the pool. */ if (tx->tx_txg != TXG_INITIAL) vdev_config_dirty(spa->spa_root_vdev); } } void spa_deactivate_mos_feature(spa_t *spa, const char *feature) { if (nvlist_remove_all(spa->spa_label_features, feature) == 0) vdev_config_dirty(spa->spa_root_vdev); } /* * Rename a spa_t. */ int spa_rename(const char *name, const char *newname) { spa_t *spa; int err; /* * Lookup the spa_t and grab the config lock for writing. We need to * actually open the pool so that we can sync out the necessary labels. * It's OK to call spa_open() with the namespace lock held because we * allow recursive calls for other reasons. */ mutex_enter(&spa_namespace_lock); if ((err = spa_open(name, &spa, FTAG)) != 0) { mutex_exit(&spa_namespace_lock); return (err); } spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); avl_remove(&spa_namespace_avl, spa); (void) strlcpy(spa->spa_name, newname, sizeof (spa->spa_name)); avl_add(&spa_namespace_avl, spa); /* * Sync all labels to disk with the new names by marking the root vdev * dirty and waiting for it to sync. It will pick up the new pool name * during the sync. */ vdev_config_dirty(spa->spa_root_vdev); spa_config_exit(spa, SCL_ALL, FTAG); txg_wait_synced(spa->spa_dsl_pool, 0); /* * Sync the updated config cache. */ spa_config_sync(spa, B_FALSE, B_TRUE); spa_close(spa, FTAG); mutex_exit(&spa_namespace_lock); return (0); } /* * Return the spa_t associated with given pool_guid, if it exists. If * device_guid is non-zero, determine whether the pool exists *and* contains * a device with the specified device_guid. */ spa_t * spa_by_guid(uint64_t pool_guid, uint64_t device_guid) { spa_t *spa; avl_tree_t *t = &spa_namespace_avl; ASSERT(MUTEX_HELD(&spa_namespace_lock)); for (spa = avl_first(t); spa != NULL; spa = AVL_NEXT(t, spa)) { if (spa->spa_state == POOL_STATE_UNINITIALIZED) continue; if (spa->spa_root_vdev == NULL) continue; if (spa_guid(spa) == pool_guid) { if (device_guid == 0) break; if (vdev_lookup_by_guid(spa->spa_root_vdev, device_guid) != NULL) break; /* * Check any devices we may be in the process of adding. */ if (spa->spa_pending_vdev) { if (vdev_lookup_by_guid(spa->spa_pending_vdev, device_guid) != NULL) break; } } } return (spa); } /* * Determine whether a pool with the given pool_guid exists. */ boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid) { return (spa_by_guid(pool_guid, device_guid) != NULL); } char * spa_strdup(const char *s) { size_t len; char *new; len = strlen(s); new = kmem_alloc(len + 1, KM_SLEEP); bcopy(s, new, len); new[len] = '\0'; return (new); } void spa_strfree(char *s) { kmem_free(s, strlen(s) + 1); } uint64_t spa_get_random(uint64_t range) { uint64_t r; ASSERT(range != 0); (void) random_get_pseudo_bytes((void *)&r, sizeof (uint64_t)); return (r % range); } uint64_t spa_generate_guid(spa_t *spa) { uint64_t guid = spa_get_random(-1ULL); if (spa != NULL) { while (guid == 0 || spa_guid_exists(spa_guid(spa), guid)) guid = spa_get_random(-1ULL); } else { while (guid == 0 || spa_guid_exists(guid, 0)) guid = spa_get_random(-1ULL); } return (guid); } void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp) { char type[256]; char *checksum = NULL; char *compress = NULL; if (bp != NULL) { if (BP_GET_TYPE(bp) & DMU_OT_NEWTYPE) { dmu_object_byteswap_t bswap = DMU_OT_BYTESWAP(BP_GET_TYPE(bp)); (void) snprintf(type, sizeof (type), "bswap %s %s", DMU_OT_IS_METADATA(BP_GET_TYPE(bp)) ? "metadata" : "data", dmu_ot_byteswap[bswap].ob_name); } else { (void) strlcpy(type, dmu_ot[BP_GET_TYPE(bp)].ot_name, sizeof (type)); } if (!BP_IS_EMBEDDED(bp)) { checksum = zio_checksum_table[BP_GET_CHECKSUM(bp)].ci_name; } compress = zio_compress_table[BP_GET_COMPRESS(bp)].ci_name; } SNPRINTF_BLKPTR(snprintf, ' ', buf, buflen, bp, type, checksum, compress); } void spa_freeze(spa_t *spa) { uint64_t freeze_txg = 0; spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); if (spa->spa_freeze_txg == UINT64_MAX) { freeze_txg = spa_last_synced_txg(spa) + TXG_SIZE; spa->spa_freeze_txg = freeze_txg; } spa_config_exit(spa, SCL_ALL, FTAG); if (freeze_txg != 0) txg_wait_synced(spa_get_dsl(spa), freeze_txg); } void zfs_panic_recover(const char *fmt, ...) { va_list adx; va_start(adx, fmt); vcmn_err(zfs_recover ? CE_WARN : CE_PANIC, fmt, adx); va_end(adx); } /* * This is a stripped-down version of strtoull, suitable only for converting * lowercase hexadecimal numbers that don't overflow. */ uint64_t -strtonum(const char *str, char **nptr) +zfs_strtonum(const char *str, char **nptr) { uint64_t val = 0; char c; int digit; while ((c = *str) != '\0') { if (c >= '0' && c <= '9') digit = c - '0'; else if (c >= 'a' && c <= 'f') digit = 10 + c - 'a'; else break; val *= 16; val += digit; str++; } if (nptr) *nptr = (char *)str; return (val); } /* * ========================================================================== * Accessor functions * ========================================================================== */ boolean_t spa_shutting_down(spa_t *spa) { return (spa->spa_async_suspended); } dsl_pool_t * spa_get_dsl(spa_t *spa) { return (spa->spa_dsl_pool); } boolean_t spa_is_initializing(spa_t *spa) { return (spa->spa_is_initializing); } blkptr_t * spa_get_rootblkptr(spa_t *spa) { return (&spa->spa_ubsync.ub_rootbp); } void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp) { spa->spa_uberblock.ub_rootbp = *bp; } void spa_altroot(spa_t *spa, char *buf, size_t buflen) { if (spa->spa_root == NULL) buf[0] = '\0'; else (void) strncpy(buf, spa->spa_root, buflen); } int spa_sync_pass(spa_t *spa) { return (spa->spa_sync_pass); } char * spa_name(spa_t *spa) { return (spa->spa_name); } uint64_t spa_guid(spa_t *spa) { dsl_pool_t *dp = spa_get_dsl(spa); uint64_t guid; /* * If we fail to parse the config during spa_load(), we can go through * the error path (which posts an ereport) and end up here with no root * vdev. We stash the original pool guid in 'spa_config_guid' to handle * this case. */ if (spa->spa_root_vdev == NULL) return (spa->spa_config_guid); guid = spa->spa_last_synced_guid != 0 ? spa->spa_last_synced_guid : spa->spa_root_vdev->vdev_guid; /* * Return the most recently synced out guid unless we're * in syncing context. */ if (dp && dsl_pool_sync_context(dp)) return (spa->spa_root_vdev->vdev_guid); else return (guid); } uint64_t spa_load_guid(spa_t *spa) { /* * This is a GUID that exists solely as a reference for the * purposes of the arc. It is generated at load time, and * is never written to persistent storage. */ return (spa->spa_load_guid); } uint64_t spa_last_synced_txg(spa_t *spa) { return (spa->spa_ubsync.ub_txg); } uint64_t spa_first_txg(spa_t *spa) { return (spa->spa_first_txg); } uint64_t spa_syncing_txg(spa_t *spa) { return (spa->spa_syncing_txg); } /* * Return the last txg where data can be dirtied. The final txgs * will be used to just clear out any deferred frees that remain. */ uint64_t spa_final_dirty_txg(spa_t *spa) { return (spa->spa_final_txg - TXG_DEFER_SIZE); } pool_state_t spa_state(spa_t *spa) { return (spa->spa_state); } spa_load_state_t spa_load_state(spa_t *spa) { return (spa->spa_load_state); } uint64_t spa_freeze_txg(spa_t *spa) { return (spa->spa_freeze_txg); } /* ARGSUSED */ uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize) { return (lsize * spa_asize_inflation); } /* * Return the amount of slop space in bytes. It is 1/32 of the pool (3.2%), * or at least 128MB, unless that would cause it to be more than half the * pool size. * * See the comment above spa_slop_shift for details. */ uint64_t spa_get_slop_space(spa_t *spa) { uint64_t space = spa_get_dspace(spa); return (MAX(space >> spa_slop_shift, MIN(space >> 1, spa_min_slop))); } uint64_t spa_get_dspace(spa_t *spa) { return (spa->spa_dspace); } void spa_update_dspace(spa_t *spa) { spa->spa_dspace = metaslab_class_get_dspace(spa_normal_class(spa)) + ddt_get_dedup_dspace(spa); } /* * Return the failure mode that has been set to this pool. The default * behavior will be to block all I/Os when a complete failure occurs. */ uint8_t spa_get_failmode(spa_t *spa) { return (spa->spa_failmode); } boolean_t spa_suspended(spa_t *spa) { return (spa->spa_suspended); } uint64_t spa_version(spa_t *spa) { return (spa->spa_ubsync.ub_version); } boolean_t spa_deflate(spa_t *spa) { return (spa->spa_deflate); } metaslab_class_t * spa_normal_class(spa_t *spa) { return (spa->spa_normal_class); } metaslab_class_t * spa_log_class(spa_t *spa) { return (spa->spa_log_class); } void spa_evicting_os_register(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_insert_head(&spa->spa_evicting_os_list, os); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_deregister(spa_t *spa, objset_t *os) { mutex_enter(&spa->spa_evicting_os_lock); list_remove(&spa->spa_evicting_os_list, os); cv_broadcast(&spa->spa_evicting_os_cv); mutex_exit(&spa->spa_evicting_os_lock); } void spa_evicting_os_wait(spa_t *spa) { mutex_enter(&spa->spa_evicting_os_lock); while (!list_is_empty(&spa->spa_evicting_os_list)) cv_wait(&spa->spa_evicting_os_cv, &spa->spa_evicting_os_lock); mutex_exit(&spa->spa_evicting_os_lock); dmu_buf_user_evict_wait(); } int spa_max_replication(spa_t *spa) { /* * As of SPA_VERSION == SPA_VERSION_DITTO_BLOCKS, we are able to * handle BPs with more than one DVA allocated. Set our max * replication level accordingly. */ if (spa_version(spa) < SPA_VERSION_DITTO_BLOCKS) return (1); return (MIN(SPA_DVAS_PER_BP, spa_max_replication_override)); } int spa_prev_software_version(spa_t *spa) { return (spa->spa_prev_software_version); } uint64_t spa_deadman_synctime(spa_t *spa) { return (spa->spa_deadman_synctime); } uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva) { uint64_t asize = DVA_GET_ASIZE(dva); uint64_t dsize = asize; ASSERT(spa_config_held(spa, SCL_ALL, RW_READER) != 0); if (asize != 0 && spa->spa_deflate) { vdev_t *vd = vdev_lookup_top(spa, DVA_GET_VDEV(dva)); dsize = (asize >> SPA_MINBLOCKSHIFT) * vd->vdev_deflate_ratio; } return (dsize); } uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); return (dsize); } uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp) { uint64_t dsize = 0; spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); for (int d = 0; d < BP_GET_NDVAS(bp); d++) dsize += dva_get_dsize_sync(spa, &bp->blk_dva[d]); spa_config_exit(spa, SCL_VDEV, FTAG); return (dsize); } /* * ========================================================================== * Initialization and Termination * ========================================================================== */ static int spa_name_compare(const void *a1, const void *a2) { const spa_t *s1 = a1; const spa_t *s2 = a2; int s; s = strcmp(s1->spa_name, s2->spa_name); if (s > 0) return (1); if (s < 0) return (-1); return (0); } int spa_busy(void) { return (spa_active_count); } void spa_boot_init() { spa_config_load(); } void spa_init(int mode) { mutex_init(&spa_namespace_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_spare_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&spa_l2cache_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&spa_namespace_cv, NULL, CV_DEFAULT, NULL); avl_create(&spa_namespace_avl, spa_name_compare, sizeof (spa_t), offsetof(spa_t, spa_avl)); avl_create(&spa_spare_avl, spa_spare_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); avl_create(&spa_l2cache_avl, spa_l2cache_compare, sizeof (spa_aux_t), offsetof(spa_aux_t, aux_avl)); spa_mode_global = mode; #ifdef _KERNEL spa_arch_init(); #else if (spa_mode_global != FREAD && dprintf_find_string("watch")) { arc_procfd = open("/proc/self/ctl", O_WRONLY); if (arc_procfd == -1) { perror("could not enable watchpoints: " "opening /proc/self/ctl failed: "); } else { arc_watch = B_TRUE; } } #endif refcount_init(); unique_init(); range_tree_init(); metaslab_alloc_trace_init(); zio_init(); dmu_init(); zil_init(); vdev_cache_stat_init(); zfs_prop_init(); zpool_prop_init(); zpool_feature_init(); spa_config_load(); l2arc_start(); } void spa_fini(void) { l2arc_stop(); spa_evict_all(); vdev_cache_stat_fini(); zil_fini(); dmu_fini(); zio_fini(); metaslab_alloc_trace_fini(); range_tree_fini(); unique_fini(); refcount_fini(); avl_destroy(&spa_namespace_avl); avl_destroy(&spa_spare_avl); avl_destroy(&spa_l2cache_avl); cv_destroy(&spa_namespace_cv); mutex_destroy(&spa_namespace_lock); mutex_destroy(&spa_spare_lock); mutex_destroy(&spa_l2cache_lock); } /* * Return whether this pool has slogs. No locking needed. * It's not a problem if the wrong answer is returned as it's only for * performance and not correctness */ boolean_t spa_has_slogs(spa_t *spa) { return (spa->spa_log_class->mc_rotor != NULL); } spa_log_state_t spa_get_log_state(spa_t *spa) { return (spa->spa_log_state); } void spa_set_log_state(spa_t *spa, spa_log_state_t state) { spa->spa_log_state = state; } boolean_t spa_is_root(spa_t *spa) { return (spa->spa_is_root); } boolean_t spa_writeable(spa_t *spa) { return (!!(spa->spa_mode & FWRITE)); } /* * Returns true if there is a pending sync task in any of the current * syncing txg, the current quiescing txg, or the current open txg. */ boolean_t spa_has_pending_synctask(spa_t *spa) { return (!txg_all_lists_empty(&spa->spa_dsl_pool->dp_sync_tasks)); } int spa_mode(spa_t *spa) { return (spa->spa_mode); } uint64_t spa_bootfs(spa_t *spa) { return (spa->spa_bootfs); } uint64_t spa_delegation(spa_t *spa) { return (spa->spa_delegation); } objset_t * spa_meta_objset(spa_t *spa) { return (spa->spa_meta_objset); } enum zio_checksum spa_dedup_checksum(spa_t *spa) { return (spa->spa_dedup_checksum); } /* * Reset pool scan stat per scan pass (or reboot). */ void spa_scan_stat_init(spa_t *spa) { /* data not stored on disk */ spa->spa_scan_pass_start = gethrestime_sec(); spa->spa_scan_pass_exam = 0; vdev_scan_stat_init(spa->spa_root_vdev); } /* * Get scan stats for zpool status reports */ int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps) { dsl_scan_t *scn = spa->spa_dsl_pool ? spa->spa_dsl_pool->dp_scan : NULL; if (scn == NULL || scn->scn_phys.scn_func == POOL_SCAN_NONE) return (SET_ERROR(ENOENT)); bzero(ps, sizeof (pool_scan_stat_t)); /* data stored on disk */ ps->pss_func = scn->scn_phys.scn_func; ps->pss_start_time = scn->scn_phys.scn_start_time; ps->pss_end_time = scn->scn_phys.scn_end_time; ps->pss_to_examine = scn->scn_phys.scn_to_examine; ps->pss_examined = scn->scn_phys.scn_examined; ps->pss_to_process = scn->scn_phys.scn_to_process; ps->pss_processed = scn->scn_phys.scn_processed; ps->pss_errors = scn->scn_phys.scn_errors; ps->pss_state = scn->scn_phys.scn_state; /* data not stored on disk */ ps->pss_pass_start = spa->spa_scan_pass_start; ps->pss_pass_exam = spa->spa_scan_pass_exam; return (0); } boolean_t spa_debug_enabled(spa_t *spa) { return (spa->spa_debug); } int spa_maxblocksize(spa_t *spa) { if (spa_feature_is_enabled(spa, SPA_FEATURE_LARGE_BLOCKS)) return (SPA_MAXBLOCKSIZE); else return (SPA_OLD_MAXBLOCKSIZE); } Index: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h (revision 319948) @@ -1,909 +1,909 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2011, 2016 by Delphix. All rights reserved. * Copyright 2011 Nexenta Systems, Inc. All rights reserved. * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. * Copyright 2013 Saso Kiselkov. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2017 Joyent, Inc. */ #ifndef _SYS_SPA_H #define _SYS_SPA_H #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /* * Forward references that lots of things need. */ typedef struct spa spa_t; typedef struct vdev vdev_t; typedef struct metaslab metaslab_t; typedef struct metaslab_group metaslab_group_t; typedef struct metaslab_class metaslab_class_t; typedef struct zio zio_t; typedef struct zilog zilog_t; typedef struct spa_aux_vdev spa_aux_vdev_t; typedef struct ddt ddt_t; typedef struct ddt_entry ddt_entry_t; struct dsl_pool; struct dsl_dataset; /* * General-purpose 32-bit and 64-bit bitfield encodings. */ #define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len)) #define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len)) #define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low)) #define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low)) #define BF32_GET(x, low, len) BF32_DECODE(x, low, len) #define BF64_GET(x, low, len) BF64_DECODE(x, low, len) #define BF32_SET(x, low, len, val) do { \ ASSERT3U(val, <, 1U << (len)); \ ASSERT3U(low + len, <=, 32); \ (x) ^= BF32_ENCODE((x >> low) ^ (val), low, len); \ _NOTE(CONSTCOND) } while (0) #define BF64_SET(x, low, len, val) do { \ ASSERT3U(val, <, 1ULL << (len)); \ ASSERT3U(low + len, <=, 64); \ ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)); \ _NOTE(CONSTCOND) } while (0) #define BF32_GET_SB(x, low, len, shift, bias) \ ((BF32_GET(x, low, len) + (bias)) << (shift)) #define BF64_GET_SB(x, low, len, shift, bias) \ ((BF64_GET(x, low, len) + (bias)) << (shift)) #define BF32_SET_SB(x, low, len, shift, bias, val) do { \ ASSERT(IS_P2ALIGNED(val, 1U << shift)); \ ASSERT3S((val) >> (shift), >=, bias); \ BF32_SET(x, low, len, ((val) >> (shift)) - (bias)); \ _NOTE(CONSTCOND) } while (0) #define BF64_SET_SB(x, low, len, shift, bias, val) do { \ ASSERT(IS_P2ALIGNED(val, 1ULL << shift)); \ ASSERT3S((val) >> (shift), >=, bias); \ BF64_SET(x, low, len, ((val) >> (shift)) - (bias)); \ _NOTE(CONSTCOND) } while (0) /* * We currently support block sizes from 512 bytes to 16MB. * The benefits of larger blocks, and thus larger IO, need to be weighed * against the cost of COWing a giant block to modify one byte, and the * large latency of reading or writing a large block. * * Note that although blocks up to 16MB are supported, the recordsize * property can not be set larger than zfs_max_recordsize (default 1MB). * See the comment near zfs_max_recordsize in dsl_dataset.c for details. * * Note that although the LSIZE field of the blkptr_t can store sizes up * to 32MB, the dnode's dn_datablkszsec can only store sizes up to * 32MB - 512 bytes. Therefore, we limit SPA_MAXBLOCKSIZE to 16MB. */ #define SPA_MINBLOCKSHIFT 9 #define SPA_OLD_MAXBLOCKSHIFT 17 #define SPA_MAXBLOCKSHIFT 24 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) #define SPA_OLD_MAXBLOCKSIZE (1ULL << SPA_OLD_MAXBLOCKSHIFT) #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) /* * Size of block to hold the configuration data (a packed nvlist) */ #define SPA_CONFIG_BLOCKSIZE (1ULL << 14) /* * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB. * The ASIZE encoding should be at least 64 times larger (6 more bits) * to support up to 4-way RAID-Z mirror mode with worst-case gang block * overhead, three DVAs per bp, plus one more bit in case we do anything * else that expands the ASIZE. */ #define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */ #define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */ #define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */ #define SPA_COMPRESSBITS 7 /* * All SPA data is represented by 128-bit data virtual addresses (DVAs). * The members of the dva_t should be considered opaque outside the SPA. */ typedef struct dva { uint64_t dva_word[2]; } dva_t; /* * Each block has a 256-bit checksum -- strong enough for cryptographic hashes. */ typedef struct zio_cksum { uint64_t zc_word[4]; } zio_cksum_t; /* * Some checksums/hashes need a 256-bit initialization salt. This salt is kept * secret and is suitable for use in MAC algorithms as the key. */ typedef struct zio_cksum_salt { uint8_t zcs_bytes[32]; } zio_cksum_salt_t; /* * Each block is described by its DVAs, time of birth, checksum, etc. * The word-by-word, bit-by-bit layout of the blkptr is as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | vdev1 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 1 |G| offset1 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 2 | vdev2 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 3 |G| offset2 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 4 | vdev3 | GRID | ASIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 5 |G| offset3 | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | cksum |E| comp| PSIZE | LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 8 | padding | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 9 | physical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | fill count | * +-------+-------+-------+-------+-------+-------+-------+-------+ * c | checksum[0] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * d | checksum[1] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * e | checksum[2] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * f | checksum[3] | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * vdev virtual device ID * offset offset into virtual device * LSIZE logical size * PSIZE physical size (after compression) * ASIZE allocated size (including RAID-Z parity and gang block headers) * GRID RAID-Z layout information (reserved for future use) * cksum checksum function * comp compression function * G gang block indicator * B byteorder (endianness) * D dedup * X encryption (on version 30, which is not supported) * E blkptr_t contains embedded data (see below) * lvl level of indirection * type DMU object type * phys birth txg of block allocation; zero if same as logical birth txg * log. birth transaction group in which the block was logically born * fill count number of non-zero blocks under this bp * checksum[4] 256-bit checksum of the data this bp describes */ /* * "Embedded" blkptr_t's don't actually point to a block, instead they * have a data payload embedded in the blkptr_t itself. See the comment * in blkptr.c for more details. * * The blkptr_t is laid out as follows: * * 64 56 48 40 32 24 16 8 0 * +-------+-------+-------+-------+-------+-------+-------+-------+ * 0 | payload | * 1 | payload | * 2 | payload | * 3 | payload | * 4 | payload | * 5 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 6 |BDX|lvl| type | etype |E| comp| PSIZE| LSIZE | * +-------+-------+-------+-------+-------+-------+-------+-------+ * 7 | payload | * 8 | payload | * 9 | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * a | logical birth txg | * +-------+-------+-------+-------+-------+-------+-------+-------+ * b | payload | * c | payload | * d | payload | * e | payload | * f | payload | * +-------+-------+-------+-------+-------+-------+-------+-------+ * * Legend: * * payload contains the embedded data * B (byteorder) byteorder (endianness) * D (dedup) padding (set to zero) * X encryption (set to zero; see above) * E (embedded) set to one * lvl indirection level * type DMU object type * etype how to interpret embedded data (BP_EMBEDDED_TYPE_*) * comp compression function of payload * PSIZE size of payload after compression, in bytes * LSIZE logical size of payload, in bytes * note that 25 bits is enough to store the largest * "normal" BP's LSIZE (2^16 * 2^9) in bytes * log. birth transaction group in which the block was logically born * * Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded * bp's they are stored in units of SPA_MINBLOCKSHIFT. * Generally, the generic BP_GET_*() macros can be used on embedded BP's. * The B, D, X, lvl, type, and comp fields are stored the same as with normal * BP's so the BP_SET_* macros can be used with them. etype, PSIZE, LSIZE must * be set with the BPE_SET_* macros. BP_SET_EMBEDDED() should be called before * other macros, as they assert that they are only used on BP's of the correct * "embedded-ness". */ #define BPE_GET_ETYPE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET((bp)->blk_prop, 40, 8)) #define BPE_SET_ETYPE(bp, t) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, t); \ _NOTE(CONSTCOND) } while (0) #define BPE_GET_LSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1)) #define BPE_SET_LSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BPE_GET_PSIZE(bp) \ (ASSERT(BP_IS_EMBEDDED(bp)), \ BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1)) #define BPE_SET_PSIZE(bp, x) do { \ ASSERT(BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \ _NOTE(CONSTCOND) } while (0) typedef enum bp_embedded_type { BP_EMBEDDED_TYPE_DATA, BP_EMBEDDED_TYPE_RESERVED, /* Reserved for an unintegrated feature. */ NUM_BP_EMBEDDED_TYPES = BP_EMBEDDED_TYPE_RESERVED } bp_embedded_type_t; #define BPE_NUM_WORDS 14 #define BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t)) #define BPE_IS_PAYLOADWORD(bp, wp) \ ((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth) #define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */ #define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */ /* * A block is a hole when it has either 1) never been written to, or * 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads * without physically allocating disk space. Holes are represented in the * blkptr_t structure by zeroed blk_dva. Correct checking for holes is * done through the BP_IS_HOLE macro. For holes, the logical size, level, * DMU object type, and birth times are all also stored for holes that * were written to at some point (i.e. were punched after having been filled). */ typedef struct blkptr { dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */ uint64_t blk_prop; /* size, compression, type, etc */ uint64_t blk_pad[2]; /* Extra space for the future */ uint64_t blk_phys_birth; /* txg when block was allocated */ uint64_t blk_birth; /* transaction group at birth */ uint64_t blk_fill; /* fill count */ zio_cksum_t blk_cksum; /* 256-bit checksum */ } blkptr_t; /* * Macros to get and set fields in a bp or DVA. */ #define DVA_GET_ASIZE(dva) \ BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_ASIZE(dva, x) \ BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \ SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8) #define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x) #define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32) #define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x) #define DVA_GET_OFFSET(dva) \ BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0) #define DVA_SET_OFFSET(dva, x) \ BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x) #define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1) #define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x) #define BP_GET_LSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? \ (BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \ BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_LSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_PSIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1)) #define BP_SET_PSIZE(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET_SB((bp)->blk_prop, \ 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_COMPRESS(bp) \ BF64_GET((bp)->blk_prop, 32, SPA_COMPRESSBITS) #define BP_SET_COMPRESS(bp, x) \ BF64_SET((bp)->blk_prop, 32, SPA_COMPRESSBITS, x) #define BP_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1) #define BP_SET_EMBEDDED(bp, x) BF64_SET((bp)->blk_prop, 39, 1, x) #define BP_GET_CHECKSUM(bp) \ (BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \ BF64_GET((bp)->blk_prop, 40, 8)) #define BP_SET_CHECKSUM(bp, x) do { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ BF64_SET((bp)->blk_prop, 40, 8, x); \ _NOTE(CONSTCOND) } while (0) #define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8) #define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x) #define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5) #define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x) #define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1) #define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x) #define BP_GET_BYTEORDER(bp) BF64_GET((bp)->blk_prop, 63, 1) #define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x) #define BP_PHYSICAL_BIRTH(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth) #define BP_SET_BIRTH(bp, logical, physical) \ { \ ASSERT(!BP_IS_EMBEDDED(bp)); \ (bp)->blk_birth = (logical); \ (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \ } #define BP_GET_FILL(bp) (BP_IS_EMBEDDED(bp) ? 1 : (bp)->blk_fill) #define BP_IS_METADATA(bp) \ (BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) #define BP_GET_ASIZE(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ DVA_GET_ASIZE(&(bp)->blk_dva[2])) #define BP_GET_UCSIZE(bp) \ (BP_IS_METADATA(bp) ? BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)) #define BP_GET_NDVAS(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ !!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ !!DVA_GET_ASIZE(&(bp)->blk_dva[2])) #define BP_COUNT_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? 0 : \ (DVA_GET_GANG(&(bp)->blk_dva[0]) + \ DVA_GET_GANG(&(bp)->blk_dva[1]) + \ DVA_GET_GANG(&(bp)->blk_dva[2]))) #define DVA_EQUAL(dva1, dva2) \ ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \ (dva1)->dva_word[0] == (dva2)->dva_word[0]) #define BP_EQUAL(bp1, bp2) \ (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \ (bp1)->blk_birth == (bp2)->blk_birth && \ DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \ DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \ DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2])) #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \ (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \ ((zc1).zc_word[1] - (zc2).zc_word[1]) | \ ((zc1).zc_word[2] - (zc2).zc_word[2]) | \ ((zc1).zc_word[3] - (zc2).zc_word[3]))) #define ZIO_CHECKSUM_IS_ZERO(zc) \ (0 == ((zc)->zc_word[0] | (zc)->zc_word[1] | \ (zc)->zc_word[2] | (zc)->zc_word[3])) #define ZIO_CHECKSUM_BSWAP(zcp) \ { \ (zcp)->zc_word[0] = BSWAP_64((zcp)->zc_word[0]); \ (zcp)->zc_word[1] = BSWAP_64((zcp)->zc_word[1]); \ (zcp)->zc_word[2] = BSWAP_64((zcp)->zc_word[2]); \ (zcp)->zc_word[3] = BSWAP_64((zcp)->zc_word[3]); \ } #define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0) #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \ { \ (zcp)->zc_word[0] = w0; \ (zcp)->zc_word[1] = w1; \ (zcp)->zc_word[2] = w2; \ (zcp)->zc_word[3] = w3; \ } #define BP_IDENTITY(bp) (ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0]) #define BP_IS_GANG(bp) \ (BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp))) #define DVA_IS_EMPTY(dva) ((dva)->dva_word[0] == 0ULL && \ (dva)->dva_word[1] == 0ULL) #define BP_IS_HOLE(bp) \ (!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp))) /* BP_IS_RAIDZ(bp) assumes no block compression */ #define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \ BP_GET_PSIZE(bp)) #define BP_ZERO(bp) \ { \ (bp)->blk_dva[0].dva_word[0] = 0; \ (bp)->blk_dva[0].dva_word[1] = 0; \ (bp)->blk_dva[1].dva_word[0] = 0; \ (bp)->blk_dva[1].dva_word[1] = 0; \ (bp)->blk_dva[2].dva_word[0] = 0; \ (bp)->blk_dva[2].dva_word[1] = 0; \ (bp)->blk_prop = 0; \ (bp)->blk_pad[0] = 0; \ (bp)->blk_pad[1] = 0; \ (bp)->blk_phys_birth = 0; \ (bp)->blk_birth = 0; \ (bp)->blk_fill = 0; \ ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \ } #ifdef _BIG_ENDIAN #define ZFS_HOST_BYTEORDER (0ULL) #else #define ZFS_HOST_BYTEORDER (1ULL) #endif #define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER) #define BP_SPRINTF_LEN 320 /* * This macro allows code sharing between zfs, libzpool, and mdb. * 'func' is either snprintf() or mdb_snprintf(). * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line. */ #define SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \ { \ static const char *copyname[] = \ { "zero", "single", "double", "triple" }; \ int len = 0; \ int copies = 0; \ \ if (bp == NULL) { \ len += func(buf + len, size - len, ""); \ } else if (BP_IS_HOLE(bp)) { \ len += func(buf + len, size - len, \ "HOLE [L%llu %s] " \ "size=%llxL birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else if (BP_IS_EMBEDDED(bp)) { \ len = func(buf + len, size - len, \ "EMBEDDED [L%llu %s] et=%u %s " \ "size=%llxL/%llxP birth=%lluL", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ (int)BPE_GET_ETYPE(bp), \ compress, \ (u_longlong_t)BPE_GET_LSIZE(bp), \ (u_longlong_t)BPE_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth); \ } else { \ for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \ const dva_t *dva = &bp->blk_dva[d]; \ if (DVA_IS_VALID(dva)) \ copies++; \ len += func(buf + len, size - len, \ "DVA[%d]=<%llu:%llx:%llx>%c", d, \ (u_longlong_t)DVA_GET_VDEV(dva), \ (u_longlong_t)DVA_GET_OFFSET(dva), \ (u_longlong_t)DVA_GET_ASIZE(dva), \ ws); \ } \ if (BP_IS_GANG(bp) && \ DVA_GET_ASIZE(&bp->blk_dva[2]) <= \ DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \ copies--; \ len += func(buf + len, size - len, \ "[L%llu %s] %s %s %s %s %s %s%c" \ "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \ "cksum=%llx:%llx:%llx:%llx", \ (u_longlong_t)BP_GET_LEVEL(bp), \ type, \ checksum, \ compress, \ BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \ BP_IS_GANG(bp) ? "gang" : "contiguous", \ BP_GET_DEDUP(bp) ? "dedup" : "unique", \ copyname[copies], \ ws, \ (u_longlong_t)BP_GET_LSIZE(bp), \ (u_longlong_t)BP_GET_PSIZE(bp), \ (u_longlong_t)bp->blk_birth, \ (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \ (u_longlong_t)BP_GET_FILL(bp), \ ws, \ (u_longlong_t)bp->blk_cksum.zc_word[0], \ (u_longlong_t)bp->blk_cksum.zc_word[1], \ (u_longlong_t)bp->blk_cksum.zc_word[2], \ (u_longlong_t)bp->blk_cksum.zc_word[3]); \ } \ ASSERT(len < size); \ } #define BP_GET_BUFC_TYPE(bp) \ (BP_IS_METADATA(bp) ? ARC_BUFC_METADATA : ARC_BUFC_DATA) typedef enum spa_import_type { SPA_IMPORT_EXISTING, SPA_IMPORT_ASSEMBLE } spa_import_type_t; /* state manipulation functions */ extern int spa_open(const char *pool, spa_t **, void *tag); extern int spa_open_rewind(const char *pool, spa_t **, void *tag, nvlist_t *policy, nvlist_t **config); extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot, size_t buflen); extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props, nvlist_t *zplprops); extern int spa_import_rootpool(char *devpath, char *devid); extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props, uint64_t flags); extern nvlist_t *spa_tryimport(nvlist_t *tryconfig); extern int spa_destroy(char *pool); extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force, boolean_t hardforce); extern int spa_reset(char *pool); extern void spa_async_request(spa_t *spa, int flag); extern void spa_async_unrequest(spa_t *spa, int flag); extern void spa_async_suspend(spa_t *spa); extern void spa_async_resume(spa_t *spa); extern spa_t *spa_inject_addref(char *pool); extern void spa_inject_delref(spa_t *spa); extern void spa_scan_stat_init(spa_t *spa); extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps); #define SPA_ASYNC_CONFIG_UPDATE 0x01 #define SPA_ASYNC_REMOVE 0x02 #define SPA_ASYNC_PROBE 0x04 #define SPA_ASYNC_RESILVER_DONE 0x08 #define SPA_ASYNC_RESILVER 0x10 #define SPA_ASYNC_AUTOEXPAND 0x20 #define SPA_ASYNC_REMOVE_DONE 0x40 #define SPA_ASYNC_REMOVE_STOP 0x80 /* * Controls the behavior of spa_vdev_remove(). */ #define SPA_REMOVE_UNSPARE 0x01 #define SPA_REMOVE_DONE 0x02 /* device manipulation */ extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot); extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, int replacing); extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done); extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare); extern boolean_t spa_vdev_remove_active(spa_t *spa); extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath); extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru); extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, nvlist_t *props, boolean_t exp); /* spare state (which is global across all pools) */ extern void spa_spare_add(vdev_t *vd); extern void spa_spare_remove(vdev_t *vd); extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt); extern void spa_spare_activate(vdev_t *vd); /* L2ARC state (which is global across all pools) */ extern void spa_l2cache_add(vdev_t *vd); extern void spa_l2cache_remove(vdev_t *vd); extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool); extern void spa_l2cache_activate(vdev_t *vd); extern void spa_l2cache_drop(spa_t *spa); /* scanning */ extern int spa_scan(spa_t *spa, pool_scan_func_t func); extern int spa_scan_stop(spa_t *spa); /* spa syncing */ extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */ extern void spa_sync_allpools(void); /* spa namespace global mutex */ extern kmutex_t spa_namespace_lock; /* * SPA configuration functions in spa_config.c */ #define SPA_CONFIG_UPDATE_POOL 0 #define SPA_CONFIG_UPDATE_VDEVS 1 extern void spa_config_sync(spa_t *, boolean_t, boolean_t); extern void spa_config_load(void); extern nvlist_t *spa_all_configs(uint64_t *); extern void spa_config_set(spa_t *spa, nvlist_t *config); extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats); extern void spa_config_update(spa_t *spa, int what); /* * Miscellaneous SPA routines in spa_misc.c */ /* Namespace manipulation */ extern spa_t *spa_lookup(const char *name); extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot); extern void spa_remove(spa_t *spa); extern spa_t *spa_next(spa_t *prev); /* Refcount functions */ extern void spa_open_ref(spa_t *spa, void *tag); extern void spa_close(spa_t *spa, void *tag); extern void spa_async_close(spa_t *spa, void *tag); extern boolean_t spa_refcount_zero(spa_t *spa); #define SCL_NONE 0x00 #define SCL_CONFIG 0x01 #define SCL_STATE 0x02 #define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */ #define SCL_ALLOC 0x08 #define SCL_ZIO 0x10 #define SCL_FREE 0x20 #define SCL_VDEV 0x40 #define SCL_LOCKS 7 #define SCL_ALL ((1 << SCL_LOCKS) - 1) #define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO) /* Pool configuration locks */ extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw); extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw); extern void spa_config_exit(spa_t *spa, int locks, void *tag); extern int spa_config_held(spa_t *spa, int locks, krw_t rw); /* Pool vdev add/remove lock */ extern uint64_t spa_vdev_enter(spa_t *spa); extern uint64_t spa_vdev_config_enter(spa_t *spa); extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag); extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error); /* Pool vdev state change lock */ extern void spa_vdev_state_enter(spa_t *spa, int oplock); extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error); /* Log state */ typedef enum spa_log_state { SPA_LOG_UNKNOWN = 0, /* unknown log state */ SPA_LOG_MISSING, /* missing log(s) */ SPA_LOG_CLEAR, /* clear the log(s) */ SPA_LOG_GOOD, /* log(s) are good */ } spa_log_state_t; extern spa_log_state_t spa_get_log_state(spa_t *spa); extern void spa_set_log_state(spa_t *spa, spa_log_state_t state); extern int spa_offline_log(spa_t *spa); /* Log claim callback */ extern void spa_claim_notify(zio_t *zio); /* Accessor functions */ extern boolean_t spa_shutting_down(spa_t *spa); extern struct dsl_pool *spa_get_dsl(spa_t *spa); extern boolean_t spa_is_initializing(spa_t *spa); extern blkptr_t *spa_get_rootblkptr(spa_t *spa); extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp); extern void spa_altroot(spa_t *, char *, size_t); extern int spa_sync_pass(spa_t *spa); extern char *spa_name(spa_t *spa); extern uint64_t spa_guid(spa_t *spa); extern uint64_t spa_load_guid(spa_t *spa); extern uint64_t spa_last_synced_txg(spa_t *spa); extern uint64_t spa_first_txg(spa_t *spa); extern uint64_t spa_syncing_txg(spa_t *spa); extern uint64_t spa_final_dirty_txg(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern pool_state_t spa_state(spa_t *spa); extern spa_load_state_t spa_load_state(spa_t *spa); extern uint64_t spa_freeze_txg(spa_t *spa); extern uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize); extern uint64_t spa_get_dspace(spa_t *spa); extern uint64_t spa_get_slop_space(spa_t *spa); extern void spa_update_dspace(spa_t *spa); extern uint64_t spa_version(spa_t *spa); extern boolean_t spa_deflate(spa_t *spa); extern metaslab_class_t *spa_normal_class(spa_t *spa); extern metaslab_class_t *spa_log_class(spa_t *spa); extern void spa_evicting_os_register(spa_t *, objset_t *os); extern void spa_evicting_os_deregister(spa_t *, objset_t *os); extern void spa_evicting_os_wait(spa_t *spa); extern int spa_max_replication(spa_t *spa); extern int spa_prev_software_version(spa_t *spa); extern int spa_busy(void); extern uint8_t spa_get_failmode(spa_t *spa); extern boolean_t spa_suspended(spa_t *spa); extern uint64_t spa_bootfs(spa_t *spa); extern uint64_t spa_delegation(spa_t *spa); extern objset_t *spa_meta_objset(spa_t *spa); extern uint64_t spa_deadman_synctime(spa_t *spa); /* Miscellaneous support routines */ extern void spa_activate_mos_feature(spa_t *spa, const char *feature, dmu_tx_t *tx); extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature); extern int spa_rename(const char *oldname, const char *newname); extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid); extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid); extern char *spa_strdup(const char *); extern void spa_strfree(char *); extern uint64_t spa_get_random(uint64_t range); extern uint64_t spa_generate_guid(spa_t *spa); extern void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp); extern void spa_freeze(spa_t *spa); extern int spa_change_guid(spa_t *spa); extern void spa_upgrade(spa_t *spa, uint64_t version); extern void spa_evict_all(void); extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid, boolean_t l2cache); extern boolean_t spa_has_spare(spa_t *, uint64_t guid); extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva); extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp); extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp); extern boolean_t spa_has_slogs(spa_t *spa); extern boolean_t spa_is_root(spa_t *spa); extern boolean_t spa_writeable(spa_t *spa); extern boolean_t spa_has_pending_synctask(spa_t *spa); extern int spa_maxblocksize(spa_t *spa); extern void zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp); extern int spa_mode(spa_t *spa); -extern uint64_t strtonum(const char *str, char **nptr); +extern uint64_t zfs_strtonum(const char *str, char **nptr); extern char *spa_his_ievent_table[]; extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, char *his_buf); extern int spa_history_log(spa_t *spa, const char *his_buf); extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl); extern void spa_history_log_version(spa_t *spa, const char *operation); extern void spa_history_log_internal(spa_t *spa, const char *operation, dmu_tx_t *tx, const char *fmt, ...); extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op, dmu_tx_t *tx, const char *fmt, ...); extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, dmu_tx_t *tx, const char *fmt, ...); /* error handling */ struct zbookmark_phys; extern void spa_log_error(spa_t *spa, zio_t *zio); extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd, zio_t *zio, uint64_t stateoroffset, uint64_t length); extern void zfs_post_remove(spa_t *spa, vdev_t *vd); extern void zfs_post_state_change(spa_t *spa, vdev_t *vd); extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd); extern uint64_t spa_get_errlog_size(spa_t *spa); extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count); extern void spa_errlog_rotate(spa_t *spa); extern void spa_errlog_drain(spa_t *spa); extern void spa_errlog_sync(spa_t *spa, uint64_t txg); extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub); /* vdev cache */ extern void vdev_cache_stat_init(void); extern void vdev_cache_stat_fini(void); /* Initialization and termination */ extern void spa_init(int flags); extern void spa_fini(void); extern void spa_boot_init(); /* properties */ extern int spa_prop_set(spa_t *spa, nvlist_t *nvp); extern int spa_prop_get(spa_t *spa, nvlist_t **nvp); extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx); extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t); /* asynchronous event notification */ extern void spa_event_notify(spa_t *spa, vdev_t *vdev, nvlist_t *hist_nvl, const char *name); #ifdef ZFS_DEBUG #define dprintf_bp(bp, fmt, ...) do { \ if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \ snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \ dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \ kmem_free(__blkbuf, BP_SPRINTF_LEN); \ } \ _NOTE(CONSTCOND) } while (0) #else #define dprintf_bp(bp, fmt, ...) #endif extern boolean_t spa_debug_enabled(spa_t *spa); #define spa_dbgmsg(spa, ...) \ { \ if (spa_debug_enabled(spa)) \ zfs_dbgmsg(__VA_ARGS__); \ } extern int spa_mode_global; /* mode, e.g. FREAD | FWRITE */ #ifdef __cplusplus } #endif #endif /* _SYS_SPA_H */ Index: vendor-sys/illumos/dist/uts/common/fs/zfs/zfs_vfsops.c =================================================================== --- vendor-sys/illumos/dist/uts/common/fs/zfs/zfs_vfsops.c (revision 319947) +++ vendor-sys/illumos/dist/uts/common/fs/zfs/zfs_vfsops.c (revision 319948) @@ -1,2306 +1,2306 @@ /* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright (c) 2012, 2015 by Delphix. All rights reserved. * Copyright (c) 2014 Integros [integros.com] * Copyright 2016 Nexenta Systems, Inc. All rights reserved. */ /* Portions Copyright 2010 Robert Milkowski */ #include #include #include #include #include #include #include #include #include #include #include #include #include "fs/fs_subr.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "zfs_comutil.h" int zfsfstype; vfsops_t *zfs_vfsops = NULL; static major_t zfs_major; static minor_t zfs_minor; static kmutex_t zfs_dev_mtx; extern int sys_shutdown; static int zfs_mount(vfs_t *vfsp, vnode_t *mvp, struct mounta *uap, cred_t *cr); static int zfs_umount(vfs_t *vfsp, int fflag, cred_t *cr); static int zfs_mountroot(vfs_t *vfsp, enum whymountroot); static int zfs_root(vfs_t *vfsp, vnode_t **vpp); static int zfs_statvfs(vfs_t *vfsp, struct statvfs64 *statp); static int zfs_vget(vfs_t *vfsp, vnode_t **vpp, fid_t *fidp); static void zfs_freevfs(vfs_t *vfsp); static const fs_operation_def_t zfs_vfsops_template[] = { VFSNAME_MOUNT, { .vfs_mount = zfs_mount }, VFSNAME_MOUNTROOT, { .vfs_mountroot = zfs_mountroot }, VFSNAME_UNMOUNT, { .vfs_unmount = zfs_umount }, VFSNAME_ROOT, { .vfs_root = zfs_root }, VFSNAME_STATVFS, { .vfs_statvfs = zfs_statvfs }, VFSNAME_SYNC, { .vfs_sync = zfs_sync }, VFSNAME_VGET, { .vfs_vget = zfs_vget }, VFSNAME_FREEVFS, { .vfs_freevfs = zfs_freevfs }, NULL, NULL }; /* * We need to keep a count of active fs's. * This is necessary to prevent our module * from being unloaded after a umount -f */ static uint32_t zfs_active_fs_count = 0; static char *noatime_cancel[] = { MNTOPT_ATIME, NULL }; static char *atime_cancel[] = { MNTOPT_NOATIME, NULL }; static char *noxattr_cancel[] = { MNTOPT_XATTR, NULL }; static char *xattr_cancel[] = { MNTOPT_NOXATTR, NULL }; /* * MO_DEFAULT is not used since the default value is determined * by the equivalent property. */ static mntopt_t mntopts[] = { { MNTOPT_NOXATTR, noxattr_cancel, NULL, 0, NULL }, { MNTOPT_XATTR, xattr_cancel, NULL, 0, NULL }, { MNTOPT_NOATIME, noatime_cancel, NULL, 0, NULL }, { MNTOPT_ATIME, atime_cancel, NULL, 0, NULL } }; static mntopts_t zfs_mntopts = { sizeof (mntopts) / sizeof (mntopt_t), mntopts }; /*ARGSUSED*/ int zfs_sync(vfs_t *vfsp, short flag, cred_t *cr) { /* * Data integrity is job one. We don't want a compromised kernel * writing to the storage pool, so we never sync during panic. */ if (panicstr) return (0); /* * SYNC_ATTR is used by fsflush() to force old filesystems like UFS * to sync metadata, which they would otherwise cache indefinitely. * Semantically, the only requirement is that the sync be initiated. * The DMU syncs out txgs frequently, so there's nothing to do. */ if (flag & SYNC_ATTR) return (0); if (vfsp != NULL) { /* * Sync a specific filesystem. */ zfsvfs_t *zfsvfs = vfsp->vfs_data; dsl_pool_t *dp; ZFS_ENTER(zfsvfs); dp = dmu_objset_pool(zfsvfs->z_os); /* * If the system is shutting down, then skip any * filesystems which may exist on a suspended pool. */ if (sys_shutdown && spa_suspended(dp->dp_spa)) { ZFS_EXIT(zfsvfs); return (0); } if (zfsvfs->z_log != NULL) zil_commit(zfsvfs->z_log, 0); ZFS_EXIT(zfsvfs); } else { /* * Sync all ZFS filesystems. This is what happens when you * run sync(1M). Unlike other filesystems, ZFS honors the * request by waiting for all pools to commit all dirty data. */ spa_sync_allpools(); } return (0); } static int zfs_create_unique_device(dev_t *dev) { major_t new_major; do { ASSERT3U(zfs_minor, <=, MAXMIN32); minor_t start = zfs_minor; do { mutex_enter(&zfs_dev_mtx); if (zfs_minor >= MAXMIN32) { /* * If we're still using the real major * keep out of /dev/zfs and /dev/zvol minor * number space. If we're using a getudev()'ed * major number, we can use all of its minors. */ if (zfs_major == ddi_name_to_major(ZFS_DRIVER)) zfs_minor = ZFS_MIN_MINOR; else zfs_minor = 0; } else { zfs_minor++; } *dev = makedevice(zfs_major, zfs_minor); mutex_exit(&zfs_dev_mtx); } while (vfs_devismounted(*dev) && zfs_minor != start); if (zfs_minor == start) { /* * We are using all ~262,000 minor numbers for the * current major number. Create a new major number. */ if ((new_major = getudev()) == (major_t)-1) { cmn_err(CE_WARN, "zfs_mount: Can't get unique major " "device number."); return (-1); } mutex_enter(&zfs_dev_mtx); zfs_major = new_major; zfs_minor = 0; mutex_exit(&zfs_dev_mtx); } else { break; } /* CONSTANTCONDITION */ } while (1); return (0); } static void atime_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == TRUE) { zfsvfs->z_atime = TRUE; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOATIME); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_ATIME, NULL, 0); } else { zfsvfs->z_atime = FALSE; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_ATIME); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOATIME, NULL, 0); } } static void xattr_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == TRUE) { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag |= VFS_XATTR; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOXATTR); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_XATTR, NULL, 0); } else { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag &= ~VFS_XATTR; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_XATTR); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOXATTR, NULL, 0); } } static void blksz_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; ASSERT3U(newval, <=, spa_maxblocksize(dmu_objset_spa(zfsvfs->z_os))); ASSERT3U(newval, >=, SPA_MINBLOCKSIZE); ASSERT(ISP2(newval)); zfsvfs->z_max_blksz = newval; zfsvfs->z_vfs->vfs_bsize = newval; } static void readonly_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval) { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag |= VFS_RDONLY; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_RW); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_RO, NULL, 0); } else { /* XXX locking on vfs_flag? */ zfsvfs->z_vfs->vfs_flag &= ~VFS_RDONLY; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_RO); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_RW, NULL, 0); } } static void devices_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { zfsvfs->z_vfs->vfs_flag |= VFS_NODEVICES; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_DEVICES); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NODEVICES, NULL, 0); } else { zfsvfs->z_vfs->vfs_flag &= ~VFS_NODEVICES; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NODEVICES); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_DEVICES, NULL, 0); } } static void setuid_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { zfsvfs->z_vfs->vfs_flag |= VFS_NOSETUID; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_SETUID); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOSETUID, NULL, 0); } else { zfsvfs->z_vfs->vfs_flag &= ~VFS_NOSETUID; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOSETUID); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_SETUID, NULL, 0); } } static void exec_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { zfsvfs->z_vfs->vfs_flag |= VFS_NOEXEC; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_EXEC); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NOEXEC, NULL, 0); } else { zfsvfs->z_vfs->vfs_flag &= ~VFS_NOEXEC; vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NOEXEC); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_EXEC, NULL, 0); } } /* * The nbmand mount option can be changed at mount time. * We can't allow it to be toggled on live file systems or incorrect * behavior may be seen from cifs clients * * This property isn't registered via dsl_prop_register(), but this callback * will be called when a file system is first mounted */ static void nbmand_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; if (newval == FALSE) { vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NBMAND); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NONBMAND, NULL, 0); } else { vfs_clearmntopt(zfsvfs->z_vfs, MNTOPT_NONBMAND); vfs_setmntopt(zfsvfs->z_vfs, MNTOPT_NBMAND, NULL, 0); } } static void snapdir_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_show_ctldir = newval; } static void vscan_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_vscan = newval; } static void acl_mode_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_mode = newval; } static void acl_inherit_changed_cb(void *arg, uint64_t newval) { zfsvfs_t *zfsvfs = arg; zfsvfs->z_acl_inherit = newval; } static int zfs_register_callbacks(vfs_t *vfsp) { struct dsl_dataset *ds = NULL; objset_t *os = NULL; zfsvfs_t *zfsvfs = NULL; uint64_t nbmand; boolean_t readonly = B_FALSE; boolean_t do_readonly = B_FALSE; boolean_t setuid = B_FALSE; boolean_t do_setuid = B_FALSE; boolean_t exec = B_FALSE; boolean_t do_exec = B_FALSE; boolean_t devices = B_FALSE; boolean_t do_devices = B_FALSE; boolean_t xattr = B_FALSE; boolean_t do_xattr = B_FALSE; boolean_t atime = B_FALSE; boolean_t do_atime = B_FALSE; int error = 0; ASSERT(vfsp); zfsvfs = vfsp->vfs_data; ASSERT(zfsvfs); os = zfsvfs->z_os; /* * The act of registering our callbacks will destroy any mount * options we may have. In order to enable temporary overrides * of mount options, we stash away the current values and * restore them after we register the callbacks. */ if (vfs_optionisset(vfsp, MNTOPT_RO, NULL) || !spa_writeable(dmu_objset_spa(os))) { readonly = B_TRUE; do_readonly = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_RW, NULL)) { readonly = B_FALSE; do_readonly = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOSUID, NULL)) { devices = B_FALSE; setuid = B_FALSE; do_devices = B_TRUE; do_setuid = B_TRUE; } else { if (vfs_optionisset(vfsp, MNTOPT_NODEVICES, NULL)) { devices = B_FALSE; do_devices = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_DEVICES, NULL)) { devices = B_TRUE; do_devices = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOSETUID, NULL)) { setuid = B_FALSE; do_setuid = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_SETUID, NULL)) { setuid = B_TRUE; do_setuid = B_TRUE; } } if (vfs_optionisset(vfsp, MNTOPT_NOEXEC, NULL)) { exec = B_FALSE; do_exec = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_EXEC, NULL)) { exec = B_TRUE; do_exec = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOXATTR, NULL)) { xattr = B_FALSE; do_xattr = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_XATTR, NULL)) { xattr = B_TRUE; do_xattr = B_TRUE; } if (vfs_optionisset(vfsp, MNTOPT_NOATIME, NULL)) { atime = B_FALSE; do_atime = B_TRUE; } else if (vfs_optionisset(vfsp, MNTOPT_ATIME, NULL)) { atime = B_TRUE; do_atime = B_TRUE; } /* * nbmand is a special property. It can only be changed at * mount time. * * This is weird, but it is documented to only be changeable * at mount time. */ if (vfs_optionisset(vfsp, MNTOPT_NONBMAND, NULL)) { nbmand = B_FALSE; } else if (vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL)) { nbmand = B_TRUE; } else { char osname[ZFS_MAX_DATASET_NAME_LEN]; dmu_objset_name(os, osname); if (error = dsl_prop_get_integer(osname, "nbmand", &nbmand, NULL)) { return (error); } } /* * Register property callbacks. * * It would probably be fine to just check for i/o error from * the first prop_register(), but I guess I like to go * overboard... */ ds = dmu_objset_ds(os); dsl_pool_config_enter(dmu_objset_pool(os), FTAG); error = dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_XATTR), xattr_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_RECORDSIZE), blksz_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_READONLY), readonly_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_DEVICES), devices_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SETUID), setuid_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_EXEC), exec_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_SNAPDIR), snapdir_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLMODE), acl_mode_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_ACLINHERIT), acl_inherit_changed_cb, zfsvfs); error = error ? error : dsl_prop_register(ds, zfs_prop_to_name(ZFS_PROP_VSCAN), vscan_changed_cb, zfsvfs); dsl_pool_config_exit(dmu_objset_pool(os), FTAG); if (error) goto unregister; /* * Invoke our callbacks to restore temporary mount options. */ if (do_readonly) readonly_changed_cb(zfsvfs, readonly); if (do_setuid) setuid_changed_cb(zfsvfs, setuid); if (do_exec) exec_changed_cb(zfsvfs, exec); if (do_devices) devices_changed_cb(zfsvfs, devices); if (do_xattr) xattr_changed_cb(zfsvfs, xattr); if (do_atime) atime_changed_cb(zfsvfs, atime); nbmand_changed_cb(zfsvfs, nbmand); return (0); unregister: dsl_prop_unregister_all(ds, zfsvfs); return (error); } static int zfs_space_delta_cb(dmu_object_type_t bonustype, void *data, uint64_t *userp, uint64_t *groupp) { /* * Is it a valid type of object to track? */ if (bonustype != DMU_OT_ZNODE && bonustype != DMU_OT_SA) return (SET_ERROR(ENOENT)); /* * If we have a NULL data pointer * then assume the id's aren't changing and * return EEXIST to the dmu to let it know to * use the same ids */ if (data == NULL) return (SET_ERROR(EEXIST)); if (bonustype == DMU_OT_ZNODE) { znode_phys_t *znp = data; *userp = znp->zp_uid; *groupp = znp->zp_gid; } else { int hdrsize; sa_hdr_phys_t *sap = data; sa_hdr_phys_t sa = *sap; boolean_t swap = B_FALSE; ASSERT(bonustype == DMU_OT_SA); if (sa.sa_magic == 0) { /* * This should only happen for newly created * files that haven't had the znode data filled * in yet. */ *userp = 0; *groupp = 0; return (0); } if (sa.sa_magic == BSWAP_32(SA_MAGIC)) { sa.sa_magic = SA_MAGIC; sa.sa_layout_info = BSWAP_16(sa.sa_layout_info); swap = B_TRUE; } else { VERIFY3U(sa.sa_magic, ==, SA_MAGIC); } hdrsize = sa_hdrsize(&sa); VERIFY3U(hdrsize, >=, sizeof (sa_hdr_phys_t)); *userp = *((uint64_t *)((uintptr_t)data + hdrsize + SA_UID_OFFSET)); *groupp = *((uint64_t *)((uintptr_t)data + hdrsize + SA_GID_OFFSET)); if (swap) { *userp = BSWAP_64(*userp); *groupp = BSWAP_64(*groupp); } } return (0); } static void fuidstr_to_sid(zfsvfs_t *zfsvfs, const char *fuidstr, char *domainbuf, int buflen, uid_t *ridp) { uint64_t fuid; const char *domain; - fuid = strtonum(fuidstr, NULL); + fuid = zfs_strtonum(fuidstr, NULL); domain = zfs_fuid_find_by_idx(zfsvfs, FUID_INDEX(fuid)); if (domain) (void) strlcpy(domainbuf, domain, buflen); else domainbuf[0] = '\0'; *ridp = FUID_RID(fuid); } static uint64_t zfs_userquota_prop_to_obj(zfsvfs_t *zfsvfs, zfs_userquota_prop_t type) { switch (type) { case ZFS_PROP_USERUSED: return (DMU_USERUSED_OBJECT); case ZFS_PROP_GROUPUSED: return (DMU_GROUPUSED_OBJECT); case ZFS_PROP_USERQUOTA: return (zfsvfs->z_userquota_obj); case ZFS_PROP_GROUPQUOTA: return (zfsvfs->z_groupquota_obj); } return (0); } int zfs_userspace_many(zfsvfs_t *zfsvfs, zfs_userquota_prop_t type, uint64_t *cookiep, void *vbuf, uint64_t *bufsizep) { int error; zap_cursor_t zc; zap_attribute_t za; zfs_useracct_t *buf = vbuf; uint64_t obj; if (!dmu_objset_userspace_present(zfsvfs->z_os)) return (SET_ERROR(ENOTSUP)); obj = zfs_userquota_prop_to_obj(zfsvfs, type); if (obj == 0) { *bufsizep = 0; return (0); } for (zap_cursor_init_serialized(&zc, zfsvfs->z_os, obj, *cookiep); (error = zap_cursor_retrieve(&zc, &za)) == 0; zap_cursor_advance(&zc)) { if ((uintptr_t)buf - (uintptr_t)vbuf + sizeof (zfs_useracct_t) > *bufsizep) break; fuidstr_to_sid(zfsvfs, za.za_name, buf->zu_domain, sizeof (buf->zu_domain), &buf->zu_rid); buf->zu_space = za.za_first_integer; buf++; } if (error == ENOENT) error = 0; ASSERT3U((uintptr_t)buf - (uintptr_t)vbuf, <=, *bufsizep); *bufsizep = (uintptr_t)buf - (uintptr_t)vbuf; *cookiep = zap_cursor_serialize(&zc); zap_cursor_fini(&zc); return (error); } /* * buf must be big enough (eg, 32 bytes) */ static int id_to_fuidstr(zfsvfs_t *zfsvfs, const char *domain, uid_t rid, char *buf, boolean_t addok) { uint64_t fuid; int domainid = 0; if (domain && domain[0]) { domainid = zfs_fuid_find_by_domain(zfsvfs, domain, NULL, addok); if (domainid == -1) return (SET_ERROR(ENOENT)); } fuid = FUID_ENCODE(domainid, rid); (void) sprintf(buf, "%llx", (longlong_t)fuid); return (0); } int zfs_userspace_one(zfsvfs_t *zfsvfs, zfs_userquota_prop_t type, const char *domain, uint64_t rid, uint64_t *valp) { char buf[32]; int err; uint64_t obj; *valp = 0; if (!dmu_objset_userspace_present(zfsvfs->z_os)) return (SET_ERROR(ENOTSUP)); obj = zfs_userquota_prop_to_obj(zfsvfs, type); if (obj == 0) return (0); err = id_to_fuidstr(zfsvfs, domain, rid, buf, B_FALSE); if (err) return (err); err = zap_lookup(zfsvfs->z_os, obj, buf, 8, 1, valp); if (err == ENOENT) err = 0; return (err); } int zfs_set_userquota(zfsvfs_t *zfsvfs, zfs_userquota_prop_t type, const char *domain, uint64_t rid, uint64_t quota) { char buf[32]; int err; dmu_tx_t *tx; uint64_t *objp; boolean_t fuid_dirtied; if (type != ZFS_PROP_USERQUOTA && type != ZFS_PROP_GROUPQUOTA) return (SET_ERROR(EINVAL)); if (zfsvfs->z_version < ZPL_VERSION_USERSPACE) return (SET_ERROR(ENOTSUP)); objp = (type == ZFS_PROP_USERQUOTA) ? &zfsvfs->z_userquota_obj : &zfsvfs->z_groupquota_obj; err = id_to_fuidstr(zfsvfs, domain, rid, buf, B_TRUE); if (err) return (err); fuid_dirtied = zfsvfs->z_fuid_dirty; tx = dmu_tx_create(zfsvfs->z_os); dmu_tx_hold_zap(tx, *objp ? *objp : DMU_NEW_OBJECT, B_TRUE, NULL); if (*objp == 0) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, zfs_userquota_prop_prefixes[type]); } if (fuid_dirtied) zfs_fuid_txhold(zfsvfs, tx); err = dmu_tx_assign(tx, TXG_WAIT); if (err) { dmu_tx_abort(tx); return (err); } mutex_enter(&zfsvfs->z_lock); if (*objp == 0) { *objp = zap_create(zfsvfs->z_os, DMU_OT_USERGROUP_QUOTA, DMU_OT_NONE, 0, tx); VERIFY(0 == zap_add(zfsvfs->z_os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[type], 8, 1, objp, tx)); } mutex_exit(&zfsvfs->z_lock); if (quota == 0) { err = zap_remove(zfsvfs->z_os, *objp, buf, tx); if (err == ENOENT) err = 0; } else { err = zap_update(zfsvfs->z_os, *objp, buf, 8, 1, "a, tx); } ASSERT(err == 0); if (fuid_dirtied) zfs_fuid_sync(zfsvfs, tx); dmu_tx_commit(tx); return (err); } boolean_t zfs_fuid_overquota(zfsvfs_t *zfsvfs, boolean_t isgroup, uint64_t fuid) { char buf[32]; uint64_t used, quota, usedobj, quotaobj; int err; usedobj = isgroup ? DMU_GROUPUSED_OBJECT : DMU_USERUSED_OBJECT; quotaobj = isgroup ? zfsvfs->z_groupquota_obj : zfsvfs->z_userquota_obj; if (quotaobj == 0 || zfsvfs->z_replay) return (B_FALSE); (void) sprintf(buf, "%llx", (longlong_t)fuid); err = zap_lookup(zfsvfs->z_os, quotaobj, buf, 8, 1, "a); if (err != 0) return (B_FALSE); err = zap_lookup(zfsvfs->z_os, usedobj, buf, 8, 1, &used); if (err != 0) return (B_FALSE); return (used >= quota); } boolean_t zfs_owner_overquota(zfsvfs_t *zfsvfs, znode_t *zp, boolean_t isgroup) { uint64_t fuid; uint64_t quotaobj; quotaobj = isgroup ? zfsvfs->z_groupquota_obj : zfsvfs->z_userquota_obj; fuid = isgroup ? zp->z_gid : zp->z_uid; if (quotaobj == 0 || zfsvfs->z_replay) return (B_FALSE); return (zfs_fuid_overquota(zfsvfs, isgroup, fuid)); } /* * Associate this zfsvfs with the given objset, which must be owned. * This will cache a bunch of on-disk state from the objset in the * zfsvfs. */ static int zfsvfs_init(zfsvfs_t *zfsvfs, objset_t *os) { int error; uint64_t val; zfsvfs->z_max_blksz = SPA_OLD_MAXBLOCKSIZE; zfsvfs->z_show_ctldir = ZFS_SNAPDIR_VISIBLE; zfsvfs->z_os = os; error = zfs_get_zplprop(os, ZFS_PROP_VERSION, &zfsvfs->z_version); if (error != 0) return (error); if (zfsvfs->z_version > zfs_zpl_version_map(spa_version(dmu_objset_spa(os)))) { (void) printf("Can't mount a version %lld file system " "on a version %lld pool\n. Pool must be upgraded to mount " "this file system.", (u_longlong_t)zfsvfs->z_version, (u_longlong_t)spa_version(dmu_objset_spa(os))); return (SET_ERROR(ENOTSUP)); } error = zfs_get_zplprop(os, ZFS_PROP_NORMALIZE, &val); if (error != 0) return (error); zfsvfs->z_norm = (int)val; error = zfs_get_zplprop(os, ZFS_PROP_UTF8ONLY, &val); if (error != 0) return (error); zfsvfs->z_utf8 = (val != 0); error = zfs_get_zplprop(os, ZFS_PROP_CASE, &val); if (error != 0) return (error); zfsvfs->z_case = (uint_t)val; /* * Fold case on file systems that are always or sometimes case * insensitive. */ if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE || zfsvfs->z_case == ZFS_CASE_MIXED) zfsvfs->z_norm |= U8_TEXTPREP_TOUPPER; zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); uint64_t sa_obj = 0; if (zfsvfs->z_use_sa) { /* should either have both of these objects or none */ error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj); if (error != 0) return (error); } error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, &zfsvfs->z_attr_table); if (error != 0) return (error); if (zfsvfs->z_version >= ZPL_VERSION_SA) sa_register_update_callback(os, zfs_sa_upgrade); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_ROOT_OBJ, 8, 1, &zfsvfs->z_root); if (error != 0) return (error); ASSERT(zfsvfs->z_root != 0); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_UNLINKED_SET, 8, 1, &zfsvfs->z_unlinkedobj); if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_USERQUOTA], 8, 1, &zfsvfs->z_userquota_obj); if (error == ENOENT) zfsvfs->z_userquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, zfs_userquota_prop_prefixes[ZFS_PROP_GROUPQUOTA], 8, 1, &zfsvfs->z_groupquota_obj); if (error == ENOENT) zfsvfs->z_groupquota_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_FUID_TABLES, 8, 1, &zfsvfs->z_fuid_obj); if (error == ENOENT) zfsvfs->z_fuid_obj = 0; else if (error != 0) return (error); error = zap_lookup(os, MASTER_NODE_OBJ, ZFS_SHARES_DIR, 8, 1, &zfsvfs->z_shares_dir); if (error == ENOENT) zfsvfs->z_shares_dir = 0; else if (error != 0) return (error); return (0); } int zfsvfs_create(const char *osname, zfsvfs_t **zfvp) { objset_t *os; zfsvfs_t *zfsvfs; int error; zfsvfs = kmem_zalloc(sizeof (zfsvfs_t), KM_SLEEP); /* * We claim to always be readonly so we can open snapshots; * other ZPL code will prevent us from writing to snapshots. */ error = dmu_objset_own(osname, DMU_OST_ZFS, B_TRUE, zfsvfs, &os); if (error) { kmem_free(zfsvfs, sizeof (zfsvfs_t)); return (error); } zfsvfs->z_vfs = NULL; zfsvfs->z_parent = zfsvfs; mutex_init(&zfsvfs->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&zfsvfs->z_lock, NULL, MUTEX_DEFAULT, NULL); list_create(&zfsvfs->z_all_znodes, sizeof (znode_t), offsetof(znode_t, z_link_node)); rrm_init(&zfsvfs->z_teardown_lock, B_FALSE); rw_init(&zfsvfs->z_teardown_inactive_lock, NULL, RW_DEFAULT, NULL); rw_init(&zfsvfs->z_fuid_lock, NULL, RW_DEFAULT, NULL); for (int i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_init(&zfsvfs->z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL); error = zfsvfs_init(zfsvfs, os); if (error != 0) { dmu_objset_disown(os, zfsvfs); *zfvp = NULL; kmem_free(zfsvfs, sizeof (zfsvfs_t)); return (error); } *zfvp = zfsvfs; return (0); } static int zfsvfs_setup(zfsvfs_t *zfsvfs, boolean_t mounting) { int error; error = zfs_register_callbacks(zfsvfs->z_vfs); if (error) return (error); zfsvfs->z_log = zil_open(zfsvfs->z_os, zfs_get_data); /* * If we are not mounting (ie: online recv), then we don't * have to worry about replaying the log as we blocked all * operations out since we closed the ZIL. */ if (mounting) { boolean_t readonly; /* * During replay we remove the read only flag to * allow replays to succeed. */ readonly = zfsvfs->z_vfs->vfs_flag & VFS_RDONLY; if (readonly != 0) zfsvfs->z_vfs->vfs_flag &= ~VFS_RDONLY; else zfs_unlinked_drain(zfsvfs); /* * Parse and replay the intent log. * * Because of ziltest, this must be done after * zfs_unlinked_drain(). (Further note: ziltest * doesn't use readonly mounts, where * zfs_unlinked_drain() isn't called.) This is because * ziltest causes spa_sync() to think it's committed, * but actually it is not, so the intent log contains * many txg's worth of changes. * * In particular, if object N is in the unlinked set in * the last txg to actually sync, then it could be * actually freed in a later txg and then reallocated * in a yet later txg. This would write a "create * object N" record to the intent log. Normally, this * would be fine because the spa_sync() would have * written out the fact that object N is free, before * we could write the "create object N" intent log * record. * * But when we are in ziltest mode, we advance the "open * txg" without actually spa_sync()-ing the changes to * disk. So we would see that object N is still * allocated and in the unlinked set, and there is an * intent log record saying to allocate it. */ if (spa_writeable(dmu_objset_spa(zfsvfs->z_os))) { if (zil_replay_disable) { zil_destroy(zfsvfs->z_log, B_FALSE); } else { zfsvfs->z_replay = B_TRUE; zil_replay(zfsvfs->z_os, zfsvfs, zfs_replay_vector); zfsvfs->z_replay = B_FALSE; } } zfsvfs->z_vfs->vfs_flag |= readonly; /* restore readonly bit */ } /* * Set the objset user_ptr to track its zfsvfs. */ mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); return (0); } void zfsvfs_free(zfsvfs_t *zfsvfs) { int i; extern krwlock_t zfsvfs_lock; /* in zfs_znode.c */ /* * This is a barrier to prevent the filesystem from going away in * zfs_znode_move() until we can safely ensure that the filesystem is * not unmounted. We consider the filesystem valid before the barrier * and invalid after the barrier. */ rw_enter(&zfsvfs_lock, RW_READER); rw_exit(&zfsvfs_lock); zfs_fuid_destroy(zfsvfs); mutex_destroy(&zfsvfs->z_znodes_lock); mutex_destroy(&zfsvfs->z_lock); list_destroy(&zfsvfs->z_all_znodes); rrm_destroy(&zfsvfs->z_teardown_lock); rw_destroy(&zfsvfs->z_teardown_inactive_lock); rw_destroy(&zfsvfs->z_fuid_lock); for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) mutex_destroy(&zfsvfs->z_hold_mtx[i]); kmem_free(zfsvfs, sizeof (zfsvfs_t)); } static void zfs_set_fuid_feature(zfsvfs_t *zfsvfs) { zfsvfs->z_use_fuids = USE_FUIDS(zfsvfs->z_version, zfsvfs->z_os); if (zfsvfs->z_vfs) { if (zfsvfs->z_use_fuids) { vfs_set_feature(zfsvfs->z_vfs, VFSFT_XVATTR); vfs_set_feature(zfsvfs->z_vfs, VFSFT_SYSATTR_VIEWS); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACEMASKONACCESS); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACLONCREATE); vfs_set_feature(zfsvfs->z_vfs, VFSFT_ACCESS_FILTER); vfs_set_feature(zfsvfs->z_vfs, VFSFT_REPARSE); } else { vfs_clear_feature(zfsvfs->z_vfs, VFSFT_XVATTR); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_SYSATTR_VIEWS); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACEMASKONACCESS); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACLONCREATE); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_ACCESS_FILTER); vfs_clear_feature(zfsvfs->z_vfs, VFSFT_REPARSE); } } zfsvfs->z_use_sa = USE_SA(zfsvfs->z_version, zfsvfs->z_os); } static int zfs_domount(vfs_t *vfsp, char *osname) { dev_t mount_dev; uint64_t recordsize, fsid_guid; int error = 0; zfsvfs_t *zfsvfs; ASSERT(vfsp); ASSERT(osname); error = zfsvfs_create(osname, &zfsvfs); if (error) return (error); zfsvfs->z_vfs = vfsp; /* Initialize the generic filesystem structure. */ vfsp->vfs_bcount = 0; vfsp->vfs_data = NULL; if (zfs_create_unique_device(&mount_dev) == -1) { error = SET_ERROR(ENODEV); goto out; } ASSERT(vfs_devismounted(mount_dev) == 0); if (error = dsl_prop_get_integer(osname, "recordsize", &recordsize, NULL)) goto out; vfsp->vfs_dev = mount_dev; vfsp->vfs_fstype = zfsfstype; vfsp->vfs_bsize = recordsize; vfsp->vfs_flag |= VFS_NOTRUNC; vfsp->vfs_data = zfsvfs; /* * The fsid is 64 bits, composed of an 8-bit fs type, which * separates our fsid from any other filesystem types, and a * 56-bit objset unique ID. The objset unique ID is unique to * all objsets open on this system, provided by unique_create(). * The 8-bit fs type must be put in the low bits of fsid[1] * because that's where other Solaris filesystems put it. */ fsid_guid = dmu_objset_fsid_guid(zfsvfs->z_os); ASSERT((fsid_guid & ~((1ULL<<56)-1)) == 0); vfsp->vfs_fsid.val[0] = fsid_guid; vfsp->vfs_fsid.val[1] = ((fsid_guid>>32) << 8) | zfsfstype & 0xFF; /* * Set features for file system. */ zfs_set_fuid_feature(zfsvfs); if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE) { vfs_set_feature(vfsp, VFSFT_DIRENTFLAGS); vfs_set_feature(vfsp, VFSFT_CASEINSENSITIVE); vfs_set_feature(vfsp, VFSFT_NOCASESENSITIVE); } else if (zfsvfs->z_case == ZFS_CASE_MIXED) { vfs_set_feature(vfsp, VFSFT_DIRENTFLAGS); vfs_set_feature(vfsp, VFSFT_CASEINSENSITIVE); } vfs_set_feature(vfsp, VFSFT_ZEROCOPY_SUPPORTED); if (dmu_objset_is_snapshot(zfsvfs->z_os)) { uint64_t pval; atime_changed_cb(zfsvfs, B_FALSE); readonly_changed_cb(zfsvfs, B_TRUE); if (error = dsl_prop_get_integer(osname, "xattr", &pval, NULL)) goto out; xattr_changed_cb(zfsvfs, pval); zfsvfs->z_issnap = B_TRUE; zfsvfs->z_os->os_sync = ZFS_SYNC_DISABLED; mutex_enter(&zfsvfs->z_os->os_user_ptr_lock); dmu_objset_set_user(zfsvfs->z_os, zfsvfs); mutex_exit(&zfsvfs->z_os->os_user_ptr_lock); } else { error = zfsvfs_setup(zfsvfs, B_TRUE); } if (!zfsvfs->z_issnap) zfsctl_create(zfsvfs); out: if (error) { dmu_objset_disown(zfsvfs->z_os, zfsvfs); zfsvfs_free(zfsvfs); } else { atomic_inc_32(&zfs_active_fs_count); } return (error); } void zfs_unregister_callbacks(zfsvfs_t *zfsvfs) { objset_t *os = zfsvfs->z_os; if (!dmu_objset_is_snapshot(os)) dsl_prop_unregister_all(dmu_objset_ds(os), zfsvfs); } /* * Convert a decimal digit string to a uint64_t integer. */ static int str_to_uint64(char *str, uint64_t *objnum) { uint64_t num = 0; while (*str) { if (*str < '0' || *str > '9') return (SET_ERROR(EINVAL)); num = num*10 + *str++ - '0'; } *objnum = num; return (0); } /* * The boot path passed from the boot loader is in the form of * "rootpool-name/root-filesystem-object-number'. Convert this * string to a dataset name: "rootpool-name/root-filesystem-name". */ static int zfs_parse_bootfs(char *bpath, char *outpath) { char *slashp; uint64_t objnum; int error; if (*bpath == 0 || *bpath == '/') return (SET_ERROR(EINVAL)); (void) strcpy(outpath, bpath); slashp = strchr(bpath, '/'); /* if no '/', just return the pool name */ if (slashp == NULL) { return (0); } /* if not a number, just return the root dataset name */ if (str_to_uint64(slashp+1, &objnum)) { return (0); } *slashp = '\0'; error = dsl_dsobj_to_dsname(bpath, objnum, outpath); *slashp = '/'; return (error); } /* * Check that the hex label string is appropriate for the dataset being * mounted into the global_zone proper. * * Return an error if the hex label string is not default or * admin_low/admin_high. For admin_low labels, the corresponding * dataset must be readonly. */ int zfs_check_global_label(const char *dsname, const char *hexsl) { if (strcasecmp(hexsl, ZFS_MLSLABEL_DEFAULT) == 0) return (0); if (strcasecmp(hexsl, ADMIN_HIGH) == 0) return (0); if (strcasecmp(hexsl, ADMIN_LOW) == 0) { /* must be readonly */ uint64_t rdonly; if (dsl_prop_get_integer(dsname, zfs_prop_to_name(ZFS_PROP_READONLY), &rdonly, NULL)) return (SET_ERROR(EACCES)); return (rdonly ? 0 : EACCES); } return (SET_ERROR(EACCES)); } /* * Determine whether the mount is allowed according to MAC check. * by comparing (where appropriate) label of the dataset against * the label of the zone being mounted into. If the dataset has * no label, create one. * * Returns 0 if access allowed, error otherwise (e.g. EACCES) */ static int zfs_mount_label_policy(vfs_t *vfsp, char *osname) { int error, retv; zone_t *mntzone = NULL; ts_label_t *mnt_tsl; bslabel_t *mnt_sl; bslabel_t ds_sl; char ds_hexsl[MAXNAMELEN]; retv = EACCES; /* assume the worst */ /* * Start by getting the dataset label if it exists. */ error = dsl_prop_get(osname, zfs_prop_to_name(ZFS_PROP_MLSLABEL), 1, sizeof (ds_hexsl), &ds_hexsl, NULL); if (error) return (SET_ERROR(EACCES)); /* * If labeling is NOT enabled, then disallow the mount of datasets * which have a non-default label already. No other label checks * are needed. */ if (!is_system_labeled()) { if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) == 0) return (0); return (SET_ERROR(EACCES)); } /* * Get the label of the mountpoint. If mounting into the global * zone (i.e. mountpoint is not within an active zone and the * zoned property is off), the label must be default or * admin_low/admin_high only; no other checks are needed. */ mntzone = zone_find_by_any_path(refstr_value(vfsp->vfs_mntpt), B_FALSE); if (mntzone->zone_id == GLOBAL_ZONEID) { uint64_t zoned; zone_rele(mntzone); if (dsl_prop_get_integer(osname, zfs_prop_to_name(ZFS_PROP_ZONED), &zoned, NULL)) return (SET_ERROR(EACCES)); if (!zoned) return (zfs_check_global_label(osname, ds_hexsl)); else /* * This is the case of a zone dataset being mounted * initially, before the zone has been fully created; * allow this mount into global zone. */ return (0); } mnt_tsl = mntzone->zone_slabel; ASSERT(mnt_tsl != NULL); label_hold(mnt_tsl); mnt_sl = label2bslabel(mnt_tsl); if (strcasecmp(ds_hexsl, ZFS_MLSLABEL_DEFAULT) == 0) { /* * The dataset doesn't have a real label, so fabricate one. */ char *str = NULL; if (l_to_str_internal(mnt_sl, &str) == 0 && dsl_prop_set_string(osname, zfs_prop_to_name(ZFS_PROP_MLSLABEL), ZPROP_SRC_LOCAL, str) == 0) retv = 0; if (str != NULL) kmem_free(str, strlen(str) + 1); } else if (hexstr_to_label(ds_hexsl, &ds_sl) == 0) { /* * Now compare labels to complete the MAC check. If the * labels are equal then allow access. If the mountpoint * label dominates the dataset label, allow readonly access. * Otherwise, access is denied. */ if (blequal(mnt_sl, &ds_sl)) retv = 0; else if (bldominates(mnt_sl, &ds_sl)) { vfs_setmntopt(vfsp, MNTOPT_RO, NULL, 0); retv = 0; } } label_rele(mnt_tsl); zone_rele(mntzone); return (retv); } static int zfs_mountroot(vfs_t *vfsp, enum whymountroot why) { int error = 0; static int zfsrootdone = 0; zfsvfs_t *zfsvfs = NULL; znode_t *zp = NULL; vnode_t *vp = NULL; char *zfs_bootfs; char *zfs_devid; ASSERT(vfsp); /* * The filesystem that we mount as root is defined in the * boot property "zfs-bootfs" with a format of * "poolname/root-dataset-objnum". */ if (why == ROOT_INIT) { if (zfsrootdone++) return (SET_ERROR(EBUSY)); /* * the process of doing a spa_load will require the * clock to be set before we could (for example) do * something better by looking at the timestamp on * an uberblock, so just set it to -1. */ clkset(-1); if ((zfs_bootfs = spa_get_bootprop("zfs-bootfs")) == NULL) { cmn_err(CE_NOTE, "spa_get_bootfs: can not get " "bootfs name"); return (SET_ERROR(EINVAL)); } zfs_devid = spa_get_bootprop("diskdevid"); error = spa_import_rootpool(rootfs.bo_name, zfs_devid); if (zfs_devid) spa_free_bootprop(zfs_devid); if (error) { spa_free_bootprop(zfs_bootfs); cmn_err(CE_NOTE, "spa_import_rootpool: error %d", error); return (error); } if (error = zfs_parse_bootfs(zfs_bootfs, rootfs.bo_name)) { spa_free_bootprop(zfs_bootfs); cmn_err(CE_NOTE, "zfs_parse_bootfs: error %d", error); return (error); } spa_free_bootprop(zfs_bootfs); if (error = vfs_lock(vfsp)) return (error); if (error = zfs_domount(vfsp, rootfs.bo_name)) { cmn_err(CE_NOTE, "zfs_domount: error %d", error); goto out; } zfsvfs = (zfsvfs_t *)vfsp->vfs_data; ASSERT(zfsvfs); if (error = zfs_zget(zfsvfs, zfsvfs->z_root, &zp)) { cmn_err(CE_NOTE, "zfs_zget: error %d", error); goto out; } vp = ZTOV(zp); mutex_enter(&vp->v_lock); vp->v_flag |= VROOT; mutex_exit(&vp->v_lock); rootvp = vp; /* * Leave rootvp held. The root file system is never unmounted. */ vfs_add((struct vnode *)0, vfsp, (vfsp->vfs_flag & VFS_RDONLY) ? MS_RDONLY : 0); out: vfs_unlock(vfsp); return (error); } else if (why == ROOT_REMOUNT) { readonly_changed_cb(vfsp->vfs_data, B_FALSE); vfsp->vfs_flag |= VFS_REMOUNT; /* refresh mount options */ zfs_unregister_callbacks(vfsp->vfs_data); return (zfs_register_callbacks(vfsp)); } else if (why == ROOT_UNMOUNT) { zfs_unregister_callbacks((zfsvfs_t *)vfsp->vfs_data); (void) zfs_sync(vfsp, 0, 0); return (0); } /* * if "why" is equal to anything else other than ROOT_INIT, * ROOT_REMOUNT, or ROOT_UNMOUNT, we do not support it. */ return (SET_ERROR(ENOTSUP)); } /*ARGSUSED*/ static int zfs_mount(vfs_t *vfsp, vnode_t *mvp, struct mounta *uap, cred_t *cr) { char *osname; pathname_t spn; int error = 0; uio_seg_t fromspace = (uap->flags & MS_SYSSPACE) ? UIO_SYSSPACE : UIO_USERSPACE; int canwrite; if (mvp->v_type != VDIR) return (SET_ERROR(ENOTDIR)); mutex_enter(&mvp->v_lock); if ((uap->flags & MS_REMOUNT) == 0 && (uap->flags & MS_OVERLAY) == 0 && (mvp->v_count != 1 || (mvp->v_flag & VROOT))) { mutex_exit(&mvp->v_lock); return (SET_ERROR(EBUSY)); } mutex_exit(&mvp->v_lock); /* * ZFS does not support passing unparsed data in via MS_DATA. * Users should use the MS_OPTIONSTR interface; this means * that all option parsing is already done and the options struct * can be interrogated. */ if ((uap->flags & MS_DATA) && uap->datalen > 0) return (SET_ERROR(EINVAL)); /* * Get the objset name (the "special" mount argument). */ if (error = pn_get(uap->spec, fromspace, &spn)) return (error); osname = spn.pn_path; /* * Check for mount privilege? * * If we don't have privilege then see if * we have local permission to allow it */ error = secpolicy_fs_mount(cr, mvp, vfsp); if (error) { if (dsl_deleg_access(osname, ZFS_DELEG_PERM_MOUNT, cr) == 0) { vattr_t vattr; /* * Make sure user is the owner of the mount point * or has sufficient privileges. */ vattr.va_mask = AT_UID; if (VOP_GETATTR(mvp, &vattr, 0, cr, NULL)) { goto out; } if (secpolicy_vnode_owner(cr, vattr.va_uid) != 0 && VOP_ACCESS(mvp, VWRITE, 0, cr, NULL) != 0) { goto out; } secpolicy_fs_mount_clearopts(cr, vfsp); } else { goto out; } } /* * Refuse to mount a filesystem if we are in a local zone and the * dataset is not visible. */ if (!INGLOBALZONE(curproc) && (!zone_dataset_visible(osname, &canwrite) || !canwrite)) { error = SET_ERROR(EPERM); goto out; } error = zfs_mount_label_policy(vfsp, osname); if (error) goto out; /* * When doing a remount, we simply refresh our temporary properties * according to those options set in the current VFS options. */ if (uap->flags & MS_REMOUNT) { /* refresh mount options */ zfs_unregister_callbacks(vfsp->vfs_data); error = zfs_register_callbacks(vfsp); goto out; } error = zfs_domount(vfsp, osname); /* * Add an extra VFS_HOLD on our parent vfs so that it can't * disappear due to a forced unmount. */ if (error == 0 && ((zfsvfs_t *)vfsp->vfs_data)->z_issnap) VFS_HOLD(mvp->v_vfsp); out: pn_free(&spn); return (error); } static int zfs_statvfs(vfs_t *vfsp, struct statvfs64 *statp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; dev32_t d32; uint64_t refdbytes, availbytes, usedobjs, availobjs; ZFS_ENTER(zfsvfs); dmu_objset_space(zfsvfs->z_os, &refdbytes, &availbytes, &usedobjs, &availobjs); /* * The underlying storage pool actually uses multiple block sizes. * We report the fragsize as the smallest block size we support, * and we report our blocksize as the filesystem's maximum blocksize. */ statp->f_frsize = 1UL << SPA_MINBLOCKSHIFT; statp->f_bsize = zfsvfs->z_max_blksz; /* * The following report "total" blocks of various kinds in the * file system, but reported in terms of f_frsize - the * "fragment" size. */ statp->f_blocks = (refdbytes + availbytes) >> SPA_MINBLOCKSHIFT; statp->f_bfree = availbytes >> SPA_MINBLOCKSHIFT; statp->f_bavail = statp->f_bfree; /* no root reservation */ /* * statvfs() should really be called statufs(), because it assumes * static metadata. ZFS doesn't preallocate files, so the best * we can do is report the max that could possibly fit in f_files, * and that minus the number actually used in f_ffree. * For f_ffree, report the smaller of the number of object available * and the number of blocks (each object will take at least a block). */ statp->f_ffree = MIN(availobjs, statp->f_bfree); statp->f_favail = statp->f_ffree; /* no "root reservation" */ statp->f_files = statp->f_ffree + usedobjs; (void) cmpldev(&d32, vfsp->vfs_dev); statp->f_fsid = d32; /* * We're a zfs filesystem. */ (void) strcpy(statp->f_basetype, vfssw[vfsp->vfs_fstype].vsw_name); statp->f_flag = vf_to_stf(vfsp->vfs_flag); statp->f_namemax = MAXNAMELEN - 1; /* * We have all of 32 characters to stuff a string here. * Is there anything useful we could/should provide? */ bzero(statp->f_fstr, sizeof (statp->f_fstr)); ZFS_EXIT(zfsvfs); return (0); } static int zfs_root(vfs_t *vfsp, vnode_t **vpp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; znode_t *rootzp; int error; ZFS_ENTER(zfsvfs); error = zfs_zget(zfsvfs, zfsvfs->z_root, &rootzp); if (error == 0) *vpp = ZTOV(rootzp); ZFS_EXIT(zfsvfs); return (error); } /* * Teardown the zfsvfs::z_os. * * Note, if 'unmounting' is FALSE, we return with the 'z_teardown_lock' * and 'z_teardown_inactive_lock' held. */ static int zfsvfs_teardown(zfsvfs_t *zfsvfs, boolean_t unmounting) { znode_t *zp; rrm_enter(&zfsvfs->z_teardown_lock, RW_WRITER, FTAG); if (!unmounting) { /* * We purge the parent filesystem's vfsp as the parent * filesystem and all of its snapshots have their vnode's * v_vfsp set to the parent's filesystem's vfsp. Note, * 'z_parent' is self referential for non-snapshots. */ (void) dnlc_purge_vfsp(zfsvfs->z_parent->z_vfs, 0); } /* * Close the zil. NB: Can't close the zil while zfs_inactive * threads are blocked as zil_close can call zfs_inactive. */ if (zfsvfs->z_log) { zil_close(zfsvfs->z_log); zfsvfs->z_log = NULL; } rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_WRITER); /* * If we are not unmounting (ie: online recv) and someone already * unmounted this file system while we were doing the switcheroo, * or a reopen of z_os failed then just bail out now. */ if (!unmounting && (zfsvfs->z_unmounted || zfsvfs->z_os == NULL)) { rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); return (SET_ERROR(EIO)); } /* * At this point there are no vops active, and any new vops will * fail with EIO since we have z_teardown_lock for writer (only * relavent for forced unmount). * * Release all holds on dbufs. */ mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp != NULL; zp = list_next(&zfsvfs->z_all_znodes, zp)) if (zp->z_sa_hdl) { ASSERT(ZTOV(zp)->v_count > 0); zfs_znode_dmu_fini(zp); } mutex_exit(&zfsvfs->z_znodes_lock); /* * If we are unmounting, set the unmounted flag and let new vops * unblock. zfs_inactive will have the unmounted behavior, and all * other vops will fail with EIO. */ if (unmounting) { zfsvfs->z_unmounted = B_TRUE; rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); } /* * z_os will be NULL if there was an error in attempting to reopen * zfsvfs, so just return as the properties had already been * unregistered and cached data had been evicted before. */ if (zfsvfs->z_os == NULL) return (0); /* * Unregister properties. */ zfs_unregister_callbacks(zfsvfs); /* * Evict cached data */ if (dsl_dataset_is_dirty(dmu_objset_ds(zfsvfs->z_os)) && !(zfsvfs->z_vfs->vfs_flag & VFS_RDONLY)) txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0); dmu_objset_evict_dbufs(zfsvfs->z_os); return (0); } /*ARGSUSED*/ static int zfs_umount(vfs_t *vfsp, int fflag, cred_t *cr) { zfsvfs_t *zfsvfs = vfsp->vfs_data; objset_t *os; int ret; ret = secpolicy_fs_unmount(cr, vfsp); if (ret) { if (dsl_deleg_access((char *)refstr_value(vfsp->vfs_resource), ZFS_DELEG_PERM_MOUNT, cr)) return (ret); } /* * We purge the parent filesystem's vfsp as the parent filesystem * and all of its snapshots have their vnode's v_vfsp set to the * parent's filesystem's vfsp. Note, 'z_parent' is self * referential for non-snapshots. */ (void) dnlc_purge_vfsp(zfsvfs->z_parent->z_vfs, 0); /* * Unmount any snapshots mounted under .zfs before unmounting the * dataset itself. */ if (zfsvfs->z_ctldir != NULL && (ret = zfsctl_umount_snapshots(vfsp, fflag, cr)) != 0) { return (ret); } if (!(fflag & MS_FORCE)) { /* * Check the number of active vnodes in the file system. * Our count is maintained in the vfs structure, but the * number is off by 1 to indicate a hold on the vfs * structure itself. * * The '.zfs' directory maintains a reference of its * own, and any active references underneath are * reflected in the vnode count. */ if (zfsvfs->z_ctldir == NULL) { if (vfsp->vfs_count > 1) return (SET_ERROR(EBUSY)); } else { if (vfsp->vfs_count > 2 || zfsvfs->z_ctldir->v_count > 1) return (SET_ERROR(EBUSY)); } } vfsp->vfs_flag |= VFS_UNMOUNTED; VERIFY(zfsvfs_teardown(zfsvfs, B_TRUE) == 0); os = zfsvfs->z_os; /* * z_os will be NULL if there was an error in * attempting to reopen zfsvfs. */ if (os != NULL) { /* * Unset the objset user_ptr. */ mutex_enter(&os->os_user_ptr_lock); dmu_objset_set_user(os, NULL); mutex_exit(&os->os_user_ptr_lock); /* * Finally release the objset */ dmu_objset_disown(os, zfsvfs); } /* * We can now safely destroy the '.zfs' directory node. */ if (zfsvfs->z_ctldir != NULL) zfsctl_destroy(zfsvfs); return (0); } static int zfs_vget(vfs_t *vfsp, vnode_t **vpp, fid_t *fidp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; znode_t *zp; uint64_t object = 0; uint64_t fid_gen = 0; uint64_t gen_mask; uint64_t zp_gen; int i, err; *vpp = NULL; ZFS_ENTER(zfsvfs); if (fidp->fid_len == LONG_FID_LEN) { zfid_long_t *zlfid = (zfid_long_t *)fidp; uint64_t objsetid = 0; uint64_t setgen = 0; for (i = 0; i < sizeof (zlfid->zf_setid); i++) objsetid |= ((uint64_t)zlfid->zf_setid[i]) << (8 * i); for (i = 0; i < sizeof (zlfid->zf_setgen); i++) setgen |= ((uint64_t)zlfid->zf_setgen[i]) << (8 * i); ZFS_EXIT(zfsvfs); err = zfsctl_lookup_objset(vfsp, objsetid, &zfsvfs); if (err) return (SET_ERROR(EINVAL)); ZFS_ENTER(zfsvfs); } if (fidp->fid_len == SHORT_FID_LEN || fidp->fid_len == LONG_FID_LEN) { zfid_short_t *zfid = (zfid_short_t *)fidp; for (i = 0; i < sizeof (zfid->zf_object); i++) object |= ((uint64_t)zfid->zf_object[i]) << (8 * i); for (i = 0; i < sizeof (zfid->zf_gen); i++) fid_gen |= ((uint64_t)zfid->zf_gen[i]) << (8 * i); } else { ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } /* A zero fid_gen means we are in the .zfs control directories */ if (fid_gen == 0 && (object == ZFSCTL_INO_ROOT || object == ZFSCTL_INO_SNAPDIR)) { *vpp = zfsvfs->z_ctldir; ASSERT(*vpp != NULL); if (object == ZFSCTL_INO_SNAPDIR) { VERIFY(zfsctl_root_lookup(*vpp, "snapshot", vpp, NULL, 0, NULL, NULL, NULL, NULL, NULL) == 0); } else { VN_HOLD(*vpp); } ZFS_EXIT(zfsvfs); return (0); } gen_mask = -1ULL >> (64 - 8 * i); dprintf("getting %llu [%u mask %llx]\n", object, fid_gen, gen_mask); if (err = zfs_zget(zfsvfs, object, &zp)) { ZFS_EXIT(zfsvfs); return (err); } (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs), &zp_gen, sizeof (uint64_t)); zp_gen = zp_gen & gen_mask; if (zp_gen == 0) zp_gen = 1; if (zp->z_unlinked || zp_gen != fid_gen) { dprintf("znode gen (%u) != fid gen (%u)\n", zp_gen, fid_gen); VN_RELE(ZTOV(zp)); ZFS_EXIT(zfsvfs); return (SET_ERROR(EINVAL)); } *vpp = ZTOV(zp); ZFS_EXIT(zfsvfs); return (0); } /* * Block out VOPs and close zfsvfs_t::z_os * * Note, if successful, then we return with the 'z_teardown_lock' and * 'z_teardown_inactive_lock' write held. We leave ownership of the underlying * dataset and objset intact so that they can be atomically handed off during * a subsequent rollback or recv operation and the resume thereafter. */ int zfs_suspend_fs(zfsvfs_t *zfsvfs) { int error; if ((error = zfsvfs_teardown(zfsvfs, B_FALSE)) != 0) return (error); return (0); } /* * Rebuild SA and release VOPs. Note that ownership of the underlying dataset * is an invariant across any of the operations that can be performed while the * filesystem was suspended. Whether it succeeded or failed, the preconditions * are the same: the relevant objset and associated dataset are owned by * zfsvfs, held, and long held on entry. */ int zfs_resume_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds) { int err; znode_t *zp; ASSERT(RRM_WRITE_HELD(&zfsvfs->z_teardown_lock)); ASSERT(RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)); /* * We already own this, so just update the objset_t, as the one we * had before may have been evicted. */ objset_t *os; VERIFY3P(ds->ds_owner, ==, zfsvfs); VERIFY(dsl_dataset_long_held(ds)); VERIFY0(dmu_objset_from_ds(ds, &os)); err = zfsvfs_init(zfsvfs, os); if (err != 0) goto bail; VERIFY(zfsvfs_setup(zfsvfs, B_FALSE) == 0); zfs_set_fuid_feature(zfsvfs); /* * Attempt to re-establish all the active znodes with * their dbufs. If a zfs_rezget() fails, then we'll let * any potential callers discover that via ZFS_ENTER_VERIFY_VP * when they try to use their znode. */ mutex_enter(&zfsvfs->z_znodes_lock); for (zp = list_head(&zfsvfs->z_all_znodes); zp; zp = list_next(&zfsvfs->z_all_znodes, zp)) { (void) zfs_rezget(zp); } mutex_exit(&zfsvfs->z_znodes_lock); bail: /* release the VOPs */ rw_exit(&zfsvfs->z_teardown_inactive_lock); rrm_exit(&zfsvfs->z_teardown_lock, FTAG); if (err) { /* * Since we couldn't setup the sa framework, try to force * unmount this file system. */ if (vn_vfswlock(zfsvfs->z_vfs->vfs_vnodecovered) == 0) (void) dounmount(zfsvfs->z_vfs, MS_FORCE, CRED()); } return (err); } static void zfs_freevfs(vfs_t *vfsp) { zfsvfs_t *zfsvfs = vfsp->vfs_data; /* * If this is a snapshot, we have an extra VFS_HOLD on our parent * from zfs_mount(). Release it here. If we came through * zfs_mountroot() instead, we didn't grab an extra hold, so * skip the VFS_RELE for rootvfs. */ if (zfsvfs->z_issnap && (vfsp != rootvfs)) VFS_RELE(zfsvfs->z_parent->z_vfs); zfsvfs_free(zfsvfs); atomic_dec_32(&zfs_active_fs_count); } /* * VFS_INIT() initialization. Note that there is no VFS_FINI(), * so we can't safely do any non-idempotent initialization here. * Leave that to zfs_init() and zfs_fini(), which are called * from the module's _init() and _fini() entry points. */ /*ARGSUSED*/ static int zfs_vfsinit(int fstype, char *name) { int error; zfsfstype = fstype; /* * Setup vfsops and vnodeops tables. */ error = vfs_setfsops(fstype, zfs_vfsops_template, &zfs_vfsops); if (error != 0) { cmn_err(CE_WARN, "zfs: bad vfs ops template"); } error = zfs_create_op_tables(); if (error) { zfs_remove_op_tables(); cmn_err(CE_WARN, "zfs: bad vnode ops template"); (void) vfs_freevfsops_by_type(zfsfstype); return (error); } mutex_init(&zfs_dev_mtx, NULL, MUTEX_DEFAULT, NULL); /* * Unique major number for all zfs mounts. * If we run out of 32-bit minors, we'll getudev() another major. */ zfs_major = ddi_name_to_major(ZFS_DRIVER); zfs_minor = ZFS_MIN_MINOR; return (0); } void zfs_init(void) { /* * Initialize .zfs directory structures */ zfsctl_init(); /* * Initialize znode cache, vnode ops, etc... */ zfs_znode_init(); dmu_objset_register_type(DMU_OST_ZFS, zfs_space_delta_cb); } void zfs_fini(void) { zfsctl_fini(); zfs_znode_fini(); } int zfs_busy(void) { return (zfs_active_fs_count != 0); } int zfs_set_version(zfsvfs_t *zfsvfs, uint64_t newvers) { int error; objset_t *os = zfsvfs->z_os; dmu_tx_t *tx; if (newvers < ZPL_VERSION_INITIAL || newvers > ZPL_VERSION) return (SET_ERROR(EINVAL)); if (newvers < zfsvfs->z_version) return (SET_ERROR(EINVAL)); if (zfs_spa_version_map(newvers) > spa_version(dmu_objset_spa(zfsvfs->z_os))) return (SET_ERROR(ENOTSUP)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_FALSE, ZPL_VERSION_STR); if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { dmu_tx_hold_zap(tx, MASTER_NODE_OBJ, B_TRUE, ZFS_SA_ATTRS); dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL); } error = dmu_tx_assign(tx, TXG_WAIT); if (error) { dmu_tx_abort(tx); return (error); } error = zap_update(os, MASTER_NODE_OBJ, ZPL_VERSION_STR, 8, 1, &newvers, tx); if (error) { dmu_tx_commit(tx); return (error); } if (newvers >= ZPL_VERSION_SA && !zfsvfs->z_use_sa) { uint64_t sa_obj; ASSERT3U(spa_version(dmu_objset_spa(zfsvfs->z_os)), >=, SPA_VERSION_SA); sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE, DMU_OT_NONE, 0, tx); error = zap_add(os, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx); ASSERT0(error); VERIFY(0 == sa_set_sa_object(os, sa_obj)); sa_register_update_callback(os, zfs_sa_upgrade); } spa_history_log_internal_ds(dmu_objset_ds(os), "upgrade", tx, "from %llu to %llu", zfsvfs->z_version, newvers); dmu_tx_commit(tx); zfsvfs->z_version = newvers; zfs_set_fuid_feature(zfsvfs); return (0); } /* * Read a property stored within the master node. */ int zfs_get_zplprop(objset_t *os, zfs_prop_t prop, uint64_t *value) { const char *pname; int error = ENOENT; /* * Look up the file system's value for the property. For the * version property, we look up a slightly different string. */ if (prop == ZFS_PROP_VERSION) pname = ZPL_VERSION_STR; else pname = zfs_prop_to_name(prop); if (os != NULL) { ASSERT3U(os->os_phys->os_type, ==, DMU_OST_ZFS); error = zap_lookup(os, MASTER_NODE_OBJ, pname, 8, 1, value); } if (error == ENOENT) { /* No value set, use the default value */ switch (prop) { case ZFS_PROP_VERSION: *value = ZPL_VERSION; break; case ZFS_PROP_NORMALIZE: case ZFS_PROP_UTF8ONLY: *value = 0; break; case ZFS_PROP_CASE: *value = ZFS_CASE_SENSITIVE; break; default: return (error); } error = 0; } return (error); } /* * Return true if the coresponding vfs's unmounted flag is set. * Otherwise return false. * If this function returns true we know VFS unmount has been initiated. */ boolean_t zfs_get_vfs_flag_unmounted(objset_t *os) { zfsvfs_t *zfvp; boolean_t unmounted = B_FALSE; ASSERT(dmu_objset_type(os) == DMU_OST_ZFS); mutex_enter(&os->os_user_ptr_lock); zfvp = dmu_objset_get_user(os); if (zfvp != NULL && zfvp->z_vfs != NULL && (zfvp->z_vfs->vfs_flag & VFS_UNMOUNTED)) unmounted = B_TRUE; mutex_exit(&os->os_user_ptr_lock); return (unmounted); } static vfsdef_t vfw = { VFSDEF_VERSION, MNTTYPE_ZFS, zfs_vfsinit, VSW_HASPROTO|VSW_CANRWRO|VSW_CANREMOUNT|VSW_VOLATILEDEV|VSW_STATS| VSW_XID|VSW_ZMOUNT, &zfs_mntopts }; struct modlfs zfs_modlfs = { &mod_fsops, "ZFS filesystem version " SPA_VERSION_STRING, &vfw };