/* * Create "user holds" on snapshots. If there is a hold on a snapshot, * the snapshot can not be destroyed. (However, it can be marked for deletion * by lzc_destroy_snaps(defer=B_TRUE).) * * The keys in the nvlist are snapshot names. * The snapshots must all be in the same pool. * The value is the name of the hold (string type). * * If cleanup_fd is not -1, it must be the result of open("/dev/zfs", O_EXCL). * In this case, when the cleanup_fd is closed (including on process * termination), the holds will be released. If the system is shut down * uncleanly, the holds will be released when the pool is next opened * or imported. * * Holds for snapshots which don't exist will be skipped and have an entry * added to errlist, but will not cause an overall failure. * * The return value will be 0 if all holds, for snapshots that existed, * were succesfully created. * * Otherwise the return value will be the errno of a (unspecified) hold that * failed and no holds will be created. * * In all cases the errlist will have an entry for each hold that failed * (name = snapshot), with its value being the error code (int32). */ int lzc_hold(nvlist_t *holds, int cleanup_fd, nvlist_t **errlist) { char pool[ZFS_MAX_DATASET_NAME_LEN]; nvlist_t *args; nvpair_t *elem; int error; /* determine the pool name */ elem = nvlist_next_nvpair(holds, NULL); if (elem == NULL) return (0); (void) strlcpy(pool, nvpair_name(elem), sizeof (pool)); pool[strcspn(pool, "/@")] = '\0'; args = fnvlist_alloc(); fnvlist_add_nvlist(args, "holds", holds); if (cleanup_fd != -1) fnvlist_add_int32(args, "cleanup_fd", cleanup_fd); error = lzc_ioctl(ZFS_IOC_HOLD, pool, args, errlist); nvlist_free(args); return (error); }
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); }
/* * "from" can be NULL, a snapshot, or a bookmark. * * If from is NULL, a full (non-incremental) stream will be estimated. This * is calculated very efficiently. * * If from is a snapshot, lzc_send_space uses the deadlists attached to * each snapshot to efficiently estimate the stream size. * * If from is a bookmark, the indirect blocks in the destination snapshot * are traversed, looking for blocks with a birth time since the creation TXG of * the snapshot this bookmark was created from. This will result in * significantly more I/O and be less efficient than a send space estimation on * an equivalent snapshot. */ int lzc_send_space(const char *snapname, const char *from, enum lzc_send_flags flags, uint64_t *spacep) { nvlist_t *args; nvlist_t *result; int err; args = fnvlist_alloc(); if (from != NULL) fnvlist_add_string(args, "from", from); if (flags & LZC_SEND_FLAG_LARGE_BLOCK) fnvlist_add_boolean(args, "largeblockok"); if (flags & LZC_SEND_FLAG_EMBED_DATA) fnvlist_add_boolean(args, "embedok"); if (flags & LZC_SEND_FLAG_COMPRESS) fnvlist_add_boolean(args, "compressok"); err = lzc_ioctl(ZFS_IOC_SEND_SPACE, snapname, args, &result); nvlist_free(args); if (err == 0) *spacep = fnvlist_lookup_uint64(result, "space"); nvlist_free(result); return (err); }
/* * 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[MAXNAMELEN]; 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); fnvlist_free(ddura.ddura_todelete); fnvlist_free(ddura.ddura_chkholds); return (error); }
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 (!dsl_dataset_is_snapshot(ds)) return (SET_ERROR(EINVAL)); if (nvlist_empty(holds)) return (0); numholds = 0; mos = ds->ds_dir->dd_pool->dp_meta_objset; zapobj = ds->ds_phys->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) && ds->ds_phys->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 void zfs_ioc_input_tests(const char *pool) { char filepath[] = "/tmp/ioc_test_file_XXXXXX"; char dataset[ZFS_MAX_DATASET_NAME_LEN]; char snapbase[ZFS_MAX_DATASET_NAME_LEN + 32]; char snapshot[ZFS_MAX_DATASET_NAME_LEN + 32]; char bookmark[ZFS_MAX_DATASET_NAME_LEN + 32]; char backup[ZFS_MAX_DATASET_NAME_LEN]; char clone[ZFS_MAX_DATASET_NAME_LEN]; int tmpfd, err; /* * Setup names and create a working dataset */ (void) snprintf(dataset, sizeof (dataset), "%s/test-fs", pool); (void) snprintf(snapbase, sizeof (snapbase), "%s@snapbase", dataset); (void) snprintf(snapshot, sizeof (snapshot), "%s@snapshot", dataset); (void) snprintf(bookmark, sizeof (bookmark), "%s#bookmark", dataset); (void) snprintf(clone, sizeof (clone), "%s/test-fs-clone", pool); (void) snprintf(backup, sizeof (backup), "%s/backup", pool); err = lzc_create(dataset, DMU_OST_ZFS, NULL, NULL, 0); if (err) { (void) fprintf(stderr, "could not create '%s': %s\n", dataset, strerror(errno)); exit(2); } tmpfd = mkstemp(filepath); if (tmpfd < 0) { (void) fprintf(stderr, "could not create '%s': %s\n", filepath, strerror(errno)); exit(2); } /* * run a test for each ioctl * Note that some test build on previous test operations */ test_pool_sync(pool); test_pool_reopen(pool); test_pool_checkpoint(pool); test_pool_discard_checkpoint(pool); test_log_history(pool); test_create(dataset); test_snapshot(pool, snapbase); test_snapshot(pool, snapshot); test_space_snaps(snapshot); test_send_space(snapbase, snapshot); test_send_new(snapshot, tmpfd); test_recv_new(backup, tmpfd); test_bookmark(pool, snapshot, bookmark); test_get_bookmarks(dataset); test_destroy_bookmarks(pool, bookmark); test_hold(pool, snapshot); test_get_holds(snapshot); test_release(pool, snapshot); test_clone(snapshot, clone); zfs_destroy(clone); test_rollback(dataset, snapshot); test_destroy_snaps(pool, snapshot); test_destroy_snaps(pool, snapbase); test_remap(dataset); test_channel_program(pool); test_load_key(dataset); test_change_key(dataset); test_unload_key(dataset); /* * cleanup */ zfs_cmd_t zc = {"\0"}; nvlist_t *snaps = fnvlist_alloc(); fnvlist_add_boolean(snaps, snapshot); (void) lzc_destroy_snaps(snaps, B_FALSE, NULL); nvlist_free(snaps); (void) zfs_destroy(dataset); (void) zfs_destroy(backup); (void) close(tmpfd); (void) unlink(filepath); /* * All the unused slots should yield ZFS_ERR_IOC_CMD_UNAVAIL */ for (int i = 0; i < ARRAY_SIZE(ioc_skip); i++) { if (ioc_tested[ioc_skip[i] - ZFS_IOC_FIRST]) (void) fprintf(stderr, "cmd %d tested, not skipped!\n", (int)(ioc_skip[i] - ZFS_IOC_FIRST)); ioc_tested[ioc_skip[i] - ZFS_IOC_FIRST] = B_TRUE; } (void) strlcpy(zc.zc_name, pool, sizeof (zc.zc_name)); zc.zc_name[sizeof (zc.zc_name) - 1] = '\0'; for (unsigned ioc = ZFS_IOC_FIRST; ioc < ZFS_IOC_LAST; ioc++) { unsigned cmd = ioc - ZFS_IOC_FIRST; if (ioc_tested[cmd]) continue; if (ioctl(zfs_fd, ioc, &zc) != 0 && errno != ZFS_ERR_IOC_CMD_UNAVAIL) { (void) fprintf(stderr, "cmd %d is missing a test case " "(%d)\n", cmd, errno); } } }
/* * Test each ioc for the folowing ioctl input errors: * ZFS_ERR_IOC_ARG_UNAVAIL an input argument is not supported by kernel * ZFS_ERR_IOC_ARG_REQUIRED a required input argument is missing * ZFS_ERR_IOC_ARG_BADTYPE an input argument has an invalid type */ static int lzc_ioctl_test(zfs_ioc_t ioc, const char *name, nvlist_t *required, nvlist_t *optional, int expected_error, boolean_t wildcard) { nvlist_t *input = fnvlist_alloc(); nvlist_t *future = fnvlist_alloc(); int error = 0; if (required != NULL) { for (nvpair_t *pair = nvlist_next_nvpair(required, NULL); pair != NULL; pair = nvlist_next_nvpair(required, pair)) { fnvlist_add_nvpair(input, pair); } } if (optional != NULL) { for (nvpair_t *pair = nvlist_next_nvpair(optional, NULL); pair != NULL; pair = nvlist_next_nvpair(optional, pair)) { fnvlist_add_nvpair(input, pair); } } /* * Generic input run with 'optional' nvlist pair */ if (!wildcard) fnvlist_add_nvlist(input, "optional", future); lzc_ioctl_run(ioc, name, input, expected_error); if (!wildcard) fnvlist_remove(input, "optional"); /* * Bogus input value */ if (!wildcard) { fnvlist_add_string(input, "bogus_input", "bogus"); lzc_ioctl_run(ioc, name, input, ZFS_ERR_IOC_ARG_UNAVAIL); fnvlist_remove(input, "bogus_input"); } /* * Missing required inputs */ if (required != NULL) { nvlist_t *empty = fnvlist_alloc(); lzc_ioctl_run(ioc, name, empty, ZFS_ERR_IOC_ARG_REQUIRED); nvlist_free(empty); } /* * Wrong nvpair type */ if (required != NULL || optional != NULL) { /* * switch the type of one of the input pairs */ for (nvpair_t *pair = nvlist_next_nvpair(input, NULL); pair != NULL; pair = nvlist_next_nvpair(input, pair)) { char pname[MAXNAMELEN]; data_type_t ptype; strlcpy(pname, nvpair_name(pair), sizeof (pname)); pname[sizeof (pname) - 1] = '\0'; ptype = nvpair_type(pair); fnvlist_remove_nvpair(input, pair); switch (ptype) { case DATA_TYPE_STRING: fnvlist_add_uint64(input, pname, 42); break; default: fnvlist_add_string(input, pname, "bogus"); break; } } lzc_ioctl_run(ioc, name, input, ZFS_ERR_IOC_ARG_BADTYPE); } nvlist_free(future); nvlist_free(input); return (error); }
/* * Generate the pool's configuration based on the current in-core state. * * We infer whether to generate a complete config or just one top-level config * based on whether vd is the root vdev. */ nvlist_t * spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, int getstats) { nvlist_t *config, *nvroot; vdev_t *rvd = spa->spa_root_vdev; unsigned long hostid = 0; boolean_t locked = B_FALSE; uint64_t split_guid; char *pool_name; int config_gen_flags = 0; if (vd == NULL) { vd = rvd; locked = B_TRUE; spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); } ASSERT(spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_READER) == (SCL_CONFIG | SCL_STATE)); /* * If txg is -1, report the current value of spa->spa_config_txg. */ if (txg == -1ULL) txg = spa->spa_config_txg; /* * Originally, users had to handle spa namespace collisions by either * exporting the already imported pool or by specifying a new name for * the pool with a conflicting name. In the case of root pools from * virtual guests, neither approach to collision resolution is * reasonable. This is addressed by extending the new name syntax with * an option to specify that the new name is temporary. When specified, * ZFS_IMPORT_TEMP_NAME will be set in spa->spa_import_flags to tell us * to use the previous name, which we do below. */ if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) { VERIFY0(nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &pool_name)); } else pool_name = spa_name(spa); config = fnvlist_alloc(); fnvlist_add_uint64(config, ZPOOL_CONFIG_VERSION, spa_version(spa)); fnvlist_add_string(config, ZPOOL_CONFIG_POOL_NAME, pool_name); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_STATE, spa_state(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_TXG, txg); fnvlist_add_uint64(config, ZPOOL_CONFIG_POOL_GUID, spa_guid(spa)); fnvlist_add_uint64(config, ZPOOL_CONFIG_ERRATA, spa->spa_errata); if (spa->spa_comment != NULL) fnvlist_add_string(config, ZPOOL_CONFIG_COMMENT, spa->spa_comment); #ifdef _KERNEL hostid = zone_get_hostid(NULL); #else /* _KERNEL */ /* * We're emulating the system's hostid in userland, so we can't use * zone_get_hostid(). */ (void) ddi_strtoul(hw_serial, NULL, 10, &hostid); #endif /* _KERNEL */ if (hostid != 0) fnvlist_add_uint64(config, ZPOOL_CONFIG_HOSTID, hostid); fnvlist_add_string(config, ZPOOL_CONFIG_HOSTNAME, utsname()->nodename); if (vd != rvd) { fnvlist_add_uint64(config, ZPOOL_CONFIG_TOP_GUID, vd->vdev_top->vdev_guid); fnvlist_add_uint64(config, ZPOOL_CONFIG_GUID, vd->vdev_guid); if (vd->vdev_isspare) fnvlist_add_uint64(config, ZPOOL_CONFIG_IS_SPARE, 1ULL); if (vd->vdev_islog) fnvlist_add_uint64(config, ZPOOL_CONFIG_IS_LOG, 1ULL); vd = vd->vdev_top; /* label contains top config */ } else { /* * Only add the (potentially large) split information * in the mos config, and not in the vdev labels */ if (spa->spa_config_splitting != NULL) fnvlist_add_nvlist(config, ZPOOL_CONFIG_SPLIT, spa->spa_config_splitting); fnvlist_add_boolean(config, ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS); config_gen_flags |= VDEV_CONFIG_MOS; } /* * Add the top-level config. We even add this on pools which * don't support holes in the namespace. */ vdev_top_config_generate(spa, config); /* * If we're splitting, record the original pool's guid. */ if (spa->spa_config_splitting != NULL && nvlist_lookup_uint64(spa->spa_config_splitting, ZPOOL_CONFIG_SPLIT_GUID, &split_guid) == 0) { fnvlist_add_uint64(config, ZPOOL_CONFIG_SPLIT_GUID, split_guid); } nvroot = vdev_config_generate(spa, vd, getstats, config_gen_flags); fnvlist_add_nvlist(config, ZPOOL_CONFIG_VDEV_TREE, nvroot); nvlist_free(nvroot); /* * Store what's necessary for reading the MOS in the label. */ fnvlist_add_nvlist(config, ZPOOL_CONFIG_FEATURES_FOR_READ, spa->spa_label_features); if (getstats && spa_load_state(spa) == SPA_LOAD_NONE) { ddt_histogram_t *ddh; ddt_stat_t *dds; ddt_object_t *ddo; ddh = kmem_zalloc(sizeof (ddt_histogram_t), KM_SLEEP); ddt_get_dedup_histogram(spa, ddh); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_HISTOGRAM, (uint64_t *)ddh, sizeof (*ddh) / sizeof (uint64_t)); kmem_free(ddh, sizeof (ddt_histogram_t)); ddo = kmem_zalloc(sizeof (ddt_object_t), KM_SLEEP); ddt_get_dedup_object_stats(spa, ddo); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_OBJ_STATS, (uint64_t *)ddo, sizeof (*ddo) / sizeof (uint64_t)); kmem_free(ddo, sizeof (ddt_object_t)); dds = kmem_zalloc(sizeof (ddt_stat_t), KM_SLEEP); ddt_get_dedup_stats(spa, dds); fnvlist_add_uint64_array(config, ZPOOL_CONFIG_DDT_STATS, (uint64_t *)dds, sizeof (*dds) / sizeof (uint64_t)); kmem_free(dds, sizeof (ddt_stat_t)); } if (locked) spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); return (config); }
/* * Synchronize pool configuration to disk. This must be called with the * namespace lock held. Synchronizing the pool cache is typically done after * the configuration has been synced to the MOS. This exposes a window where * the MOS config will have been updated but the cache file has not. If * the system were to crash at that instant then the cached config may not * contain the correct information to open the pool and an explicit import * would be required. */ void spa_config_sync(spa_t *target, boolean_t removing, boolean_t postsysevent) { spa_config_dirent_t *dp, *tdp; nvlist_t *nvl; char *pool_name; boolean_t ccw_failure; int error = 0; ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (rootdir == NULL || !(spa_mode_global & FWRITE)) return; /* * Iterate over all cachefiles for the pool, past or present. When the * cachefile is changed, the new one is pushed onto this list, allowing * us to update previous cachefiles that no longer contain this pool. */ ccw_failure = B_FALSE; for (dp = list_head(&target->spa_config_list); dp != NULL; dp = list_next(&target->spa_config_list, dp)) { spa_t *spa = NULL; if (dp->scd_path == NULL) continue; /* * Iterate over all pools, adding any matching pools to 'nvl'. */ nvl = NULL; while ((spa = spa_next(spa)) != NULL) { /* * Skip over our own pool if we're about to remove * ourselves from the spa namespace or any pool that * is readonly. Since we cannot guarantee that a * readonly pool would successfully import upon reboot, * we don't allow them to be written to the cache file. */ if ((spa == target && removing) || !spa_writeable(spa)) continue; mutex_enter(&spa->spa_props_lock); tdp = list_head(&spa->spa_config_list); if (spa->spa_config == NULL || tdp == NULL || tdp->scd_path == NULL || strcmp(tdp->scd_path, dp->scd_path) != 0) { mutex_exit(&spa->spa_props_lock); continue; } if (nvl == NULL) nvl = fnvlist_alloc(); if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) pool_name = fnvlist_lookup_string( spa->spa_config, ZPOOL_CONFIG_POOL_NAME); else pool_name = spa_name(spa); fnvlist_add_nvlist(nvl, pool_name, spa->spa_config); mutex_exit(&spa->spa_props_lock); } error = spa_config_write(dp, nvl); if (error != 0) ccw_failure = B_TRUE; nvlist_free(nvl); } if (ccw_failure) { /* * Keep trying so that configuration data is * written if/when any temporary filesystem * resource issues are resolved. */ if (target->spa_ccw_fail_time == 0) { zfs_ereport_post(FM_EREPORT_ZFS_CONFIG_CACHE_WRITE, target, NULL, NULL, 0, 0); } target->spa_ccw_fail_time = gethrtime(); spa_async_request(target, SPA_ASYNC_CONFIG_UPDATE); } else { /* * Do not rate limit future attempts to update * the config cache. */ target->spa_ccw_fail_time = 0; } /* * Remove any config entries older than the current one. */ dp = list_head(&target->spa_config_list); while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) { list_remove(&target->spa_config_list, tdp); if (tdp->scd_path != NULL) spa_strfree(tdp->scd_path); kmem_free(tdp, sizeof (spa_config_dirent_t)); } spa_config_generation++; if (postsysevent) spa_event_notify(target, NULL, ESC_ZFS_CONFIG_SYNC); }
static void run_tests(void) { const char *key = "key"; /* Note: maximum nvlist key length is 32KB */ int len = 1024 * 31; char *bigstring = malloc(len); for (int i = 0; i < len; i++) bigstring[i] = 'a' + i % 26; bigstring[len - 1] = '\0'; nvl = fnvlist_alloc(); fnvlist_add_boolean(nvl, key); test("boolean", B_TRUE, B_FALSE); fnvlist_add_boolean_value(nvl, key, B_TRUE); test("boolean_value", B_FALSE, B_FALSE); fnvlist_add_byte(nvl, key, 1); test("byte", B_FALSE, B_FALSE); fnvlist_add_int8(nvl, key, 1); test("int8", B_FALSE, B_FALSE); fnvlist_add_uint8(nvl, key, 1); test("uint8", B_FALSE, B_FALSE); fnvlist_add_int16(nvl, key, 1); test("int16", B_FALSE, B_FALSE); fnvlist_add_uint16(nvl, key, 1); test("uint16", B_FALSE, B_FALSE); fnvlist_add_int32(nvl, key, 1); test("int32", B_FALSE, B_FALSE); fnvlist_add_uint32(nvl, key, 1); test("uint32", B_FALSE, B_FALSE); fnvlist_add_int64(nvl, key, 1); test("int64", B_TRUE, B_TRUE); fnvlist_add_uint64(nvl, key, 1); test("uint64", B_FALSE, B_FALSE); fnvlist_add_string(nvl, key, "1"); test("string", B_TRUE, B_TRUE); { nvlist_t *val = fnvlist_alloc(); fnvlist_add_string(val, "subkey", "subvalue"); fnvlist_add_nvlist(nvl, key, val); fnvlist_free(val); test("nvlist", B_TRUE, B_TRUE); } { boolean_t val[2] = { B_FALSE, B_TRUE }; fnvlist_add_boolean_array(nvl, key, val, 2); test("boolean_array", B_FALSE, B_FALSE); } { uchar_t val[2] = { 0, 1 }; fnvlist_add_byte_array(nvl, key, val, 2); test("byte_array", B_FALSE, B_FALSE); } { int8_t val[2] = { 0, 1 }; fnvlist_add_int8_array(nvl, key, val, 2); test("int8_array", B_FALSE, B_FALSE); } { uint8_t val[2] = { 0, 1 }; fnvlist_add_uint8_array(nvl, key, val, 2); test("uint8_array", B_FALSE, B_FALSE); } { int16_t val[2] = { 0, 1 }; fnvlist_add_int16_array(nvl, key, val, 2); test("int16_array", B_FALSE, B_FALSE); } { uint16_t val[2] = { 0, 1 }; fnvlist_add_uint16_array(nvl, key, val, 2); test("uint16_array", B_FALSE, B_FALSE); } { int32_t val[2] = { 0, 1 }; fnvlist_add_int32_array(nvl, key, val, 2); test("int32_array", B_FALSE, B_FALSE); } { uint32_t val[2] = { 0, 1 }; fnvlist_add_uint32_array(nvl, key, val, 2); test("uint32_array", B_FALSE, B_FALSE); } { int64_t val[2] = { 0, 1 }; fnvlist_add_int64_array(nvl, key, val, 2); test("int64_array", B_TRUE, B_FALSE); } { uint64_t val[2] = { 0, 1 }; fnvlist_add_uint64_array(nvl, key, val, 2); test("uint64_array", B_FALSE, B_FALSE); } { char *const val[2] = { "0", "1" }; fnvlist_add_string_array(nvl, key, val, 2); test("string_array", B_TRUE, B_FALSE); } { nvlist_t *val[2]; val[0] = fnvlist_alloc(); fnvlist_add_string(val[0], "subkey", "subvalue"); val[1] = fnvlist_alloc(); fnvlist_add_string(val[1], "subkey2", "subvalue2"); fnvlist_add_nvlist_array(nvl, key, val, 2); fnvlist_free(val[0]); fnvlist_free(val[1]); test("nvlist_array", B_FALSE, B_FALSE); } { fnvlist_add_string(nvl, bigstring, "1"); test("large_key", B_TRUE, B_TRUE); } { fnvlist_add_string(nvl, key, bigstring); test("large_value", B_TRUE, B_TRUE); } { for (int i = 0; i < 1024; i++) { char buf[32]; (void) snprintf(buf, sizeof (buf), "key-%u", i); fnvlist_add_int64(nvl, buf, i); } test("many_keys", B_TRUE, B_TRUE); } #ifndef __sparc__ { for (int i = 0; i < 10; i++) { nvlist_t *newval = fnvlist_alloc(); fnvlist_add_nvlist(newval, "key", nvl); fnvlist_free(nvl); nvl = newval; } test("deeply_nested_pos", B_TRUE, B_TRUE); } { for (int i = 0; i < 90; i++) { nvlist_t *newval = fnvlist_alloc(); fnvlist_add_nvlist(newval, "key", nvl); fnvlist_free(nvl); nvl = newval; } test("deeply_nested_neg", B_FALSE, B_FALSE); } #endif free(bigstring); fnvlist_free(nvl); }
/* * Linux adds ZFS_IOC_RECV_NEW for resumable and raw streams and preserves the * legacy ZFS_IOC_RECV user/kernel interface. The new interface supports all * stream options but is currently only used for resumable streams. This way * updated user space utilities will interoperate with older kernel modules. * * Non-Linux OpenZFS platforms have opted to modify the legacy interface. */ static int recv_impl(const char *snapname, nvlist_t *recvdprops, nvlist_t *localprops, const char *origin, boolean_t force, boolean_t resumable, boolean_t raw, int input_fd, const dmu_replay_record_t *begin_record, int cleanup_fd, uint64_t *read_bytes, uint64_t *errflags, uint64_t *action_handle, nvlist_t **errors) { dmu_replay_record_t drr; char fsname[MAXPATHLEN]; char *atp; int error; ASSERT3S(g_refcount, >, 0); VERIFY3S(g_fd, !=, -1); /* Set 'fsname' to the name of containing filesystem */ (void) strlcpy(fsname, snapname, sizeof (fsname)); atp = strchr(fsname, '@'); if (atp == NULL) return (EINVAL); *atp = '\0'; /* If the fs does not exist, try its parent. */ if (!lzc_exists(fsname)) { char *slashp = strrchr(fsname, '/'); if (slashp == NULL) return (ENOENT); *slashp = '\0'; } /* * The begin_record is normally a non-byteswapped BEGIN record. * For resumable streams it may be set to any non-byteswapped * dmu_replay_record_t. */ if (begin_record == NULL) { error = recv_read(input_fd, &drr, sizeof (drr)); if (error != 0) return (error); } else { drr = *begin_record; } if (resumable || raw) { nvlist_t *outnvl = NULL; nvlist_t *innvl = fnvlist_alloc(); fnvlist_add_string(innvl, "snapname", snapname); if (recvdprops != NULL) fnvlist_add_nvlist(innvl, "props", recvdprops); if (localprops != NULL) fnvlist_add_nvlist(innvl, "localprops", localprops); if (origin != NULL && strlen(origin)) fnvlist_add_string(innvl, "origin", origin); fnvlist_add_byte_array(innvl, "begin_record", (uchar_t *)&drr, sizeof (drr)); fnvlist_add_int32(innvl, "input_fd", input_fd); if (force) fnvlist_add_boolean(innvl, "force"); if (resumable) fnvlist_add_boolean(innvl, "resumable"); if (cleanup_fd >= 0) fnvlist_add_int32(innvl, "cleanup_fd", cleanup_fd); if (action_handle != NULL) fnvlist_add_uint64(innvl, "action_handle", *action_handle); error = lzc_ioctl(ZFS_IOC_RECV_NEW, fsname, innvl, &outnvl); if (error == 0 && read_bytes != NULL) error = nvlist_lookup_uint64(outnvl, "read_bytes", read_bytes); if (error == 0 && errflags != NULL) error = nvlist_lookup_uint64(outnvl, "error_flags", errflags); if (error == 0 && action_handle != NULL) error = nvlist_lookup_uint64(outnvl, "action_handle", action_handle); if (error == 0 && errors != NULL) { nvlist_t *nvl; error = nvlist_lookup_nvlist(outnvl, "errors", &nvl); if (error == 0) *errors = fnvlist_dup(nvl); } fnvlist_free(innvl); fnvlist_free(outnvl); } else { zfs_cmd_t zc = {"\0"}; char *packed = NULL; size_t size; ASSERT3S(g_refcount, >, 0); (void) strlcpy(zc.zc_name, fsname, sizeof (zc.zc_value)); (void) strlcpy(zc.zc_value, snapname, sizeof (zc.zc_value)); if (recvdprops != NULL) { packed = fnvlist_pack(recvdprops, &size); zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed; zc.zc_nvlist_src_size = size; } if (localprops != NULL) { packed = fnvlist_pack(localprops, &size); zc.zc_nvlist_conf = (uint64_t)(uintptr_t)packed; zc.zc_nvlist_conf_size = size; } if (origin != NULL) (void) strlcpy(zc.zc_string, origin, sizeof (zc.zc_string)); ASSERT3S(drr.drr_type, ==, DRR_BEGIN); zc.zc_begin_record = drr.drr_u.drr_begin; zc.zc_guid = force; zc.zc_cookie = input_fd; zc.zc_cleanup_fd = -1; zc.zc_action_handle = 0; if (cleanup_fd >= 0) zc.zc_cleanup_fd = cleanup_fd; if (action_handle != NULL) zc.zc_action_handle = *action_handle; zc.zc_nvlist_dst_size = 128 * 1024; zc.zc_nvlist_dst = (uint64_t)(uintptr_t) malloc(zc.zc_nvlist_dst_size); error = ioctl(g_fd, ZFS_IOC_RECV, &zc); if (error != 0) { error = errno; } else { if (read_bytes != NULL) *read_bytes = zc.zc_cookie; if (errflags != NULL) *errflags = zc.zc_obj; if (action_handle != NULL) *action_handle = zc.zc_action_handle; if (errors != NULL) VERIFY0(nvlist_unpack( (void *)(uintptr_t)zc.zc_nvlist_dst, zc.zc_nvlist_dst_size, errors, KM_SLEEP)); } if (packed != NULL) fnvlist_pack_free(packed, size); free((void *)(uintptr_t)zc.zc_nvlist_dst); } return (error); }
static int lzc_ioctl(zfs_ioc_t ioc, const char *name, nvlist_t *source, nvlist_t **resultp) { zfs_cmd_t zc = { 0 }; int error = 0; char *packed; #ifdef __FreeBSD__ nvlist_t *oldsource; #endif size_t size; ASSERT3S(g_refcount, >, 0); (void) strlcpy(zc.zc_name, name, sizeof (zc.zc_name)); #ifdef __FreeBSD__ if (zfs_ioctl_version == ZFS_IOCVER_UNDEF) zfs_ioctl_version = get_zfs_ioctl_version(); if (zfs_ioctl_version < ZFS_IOCVER_LZC) { oldsource = source; error = lzc_compat_pre(&zc, &ioc, &source); if (error) return (error); } #endif packed = fnvlist_pack(source, &size); zc.zc_nvlist_src = (uint64_t)(uintptr_t)packed; zc.zc_nvlist_src_size = size; if (resultp != NULL) { *resultp = NULL; zc.zc_nvlist_dst_size = MAX(size * 2, 128 * 1024); zc.zc_nvlist_dst = (uint64_t)(uintptr_t) malloc(zc.zc_nvlist_dst_size); #ifdef illumos if (zc.zc_nvlist_dst == NULL) { #else if (zc.zc_nvlist_dst == 0) { #endif error = ENOMEM; goto out; } } while (ioctl(g_fd, ioc, &zc) != 0) { if (errno == ENOMEM && resultp != NULL) { free((void *)(uintptr_t)zc.zc_nvlist_dst); zc.zc_nvlist_dst_size *= 2; zc.zc_nvlist_dst = (uint64_t)(uintptr_t) malloc(zc.zc_nvlist_dst_size); #ifdef illumos if (zc.zc_nvlist_dst == NULL) { #else if (zc.zc_nvlist_dst == 0) { #endif error = ENOMEM; goto out; } } else { error = errno; break; } } #ifdef __FreeBSD__ if (zfs_ioctl_version < ZFS_IOCVER_LZC) lzc_compat_post(&zc, ioc); #endif if (zc.zc_nvlist_dst_filled) { *resultp = fnvlist_unpack((void *)(uintptr_t)zc.zc_nvlist_dst, zc.zc_nvlist_dst_size); } #ifdef __FreeBSD__ if (zfs_ioctl_version < ZFS_IOCVER_LZC) lzc_compat_outnvl(&zc, ioc, resultp); #endif out: #ifdef __FreeBSD__ if (zfs_ioctl_version < ZFS_IOCVER_LZC) { if (source != oldsource) nvlist_free(source); source = oldsource; } #endif fnvlist_pack_free(packed, size); free((void *)(uintptr_t)zc.zc_nvlist_dst); return (error); } int lzc_create(const char *fsname, enum lzc_dataset_type type, nvlist_t *props) { int error; nvlist_t *args = fnvlist_alloc(); fnvlist_add_int32(args, "type", (dmu_objset_type_t)type); if (props != NULL) fnvlist_add_nvlist(args, "props", props); error = lzc_ioctl(ZFS_IOC_CREATE, fsname, args, NULL); nvlist_free(args); return (error); }
/* * Find all 'allow' permissions from a given point and then continue * traversing up to the root. * * This function constructs an nvlist of nvlists. * each setpoint is an nvlist composed of an nvlist of an nvlist * of the individual * users/groups/everyone/create * permissions. * * The nvlist will look like this. * * { source fsname -> { whokeys { permissions,...}, ...}} * * The fsname nvpairs will be arranged in a bottom up order. For example, * if we have the following structure a/b/c then the nvpairs for the fsnames * will be ordered a/b/c, a/b, a. */ int dsl_deleg_get(const char *ddname, nvlist_t **nvp) { dsl_dir_t *dd, *startdd; dsl_pool_t *dp; int error; objset_t *mos; zap_cursor_t *basezc, *zc; zap_attribute_t *baseza, *za; char *source; error = dsl_pool_hold(ddname, FTAG, &dp); if (error != 0) return (error); error = dsl_dir_hold(dp, ddname, FTAG, &startdd, NULL); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } dp = startdd->dd_pool; mos = dp->dp_meta_objset; zc = kmem_alloc(sizeof (zap_cursor_t), KM_SLEEP); za = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP); basezc = kmem_alloc(sizeof (zap_cursor_t), KM_SLEEP); baseza = kmem_alloc(sizeof (zap_attribute_t), KM_SLEEP); source = kmem_alloc(MAXNAMELEN + strlen(MOS_DIR_NAME) + 1, KM_SLEEP); VERIFY(nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (dd = startdd; dd != NULL; dd = dd->dd_parent) { nvlist_t *sp_nvp; uint64_t n; if (dd->dd_phys->dd_deleg_zapobj == 0 || zap_count(mos, dd->dd_phys->dd_deleg_zapobj, &n) != 0 || n == 0) continue; sp_nvp = fnvlist_alloc(); for (zap_cursor_init(basezc, mos, dd->dd_phys->dd_deleg_zapobj); zap_cursor_retrieve(basezc, baseza) == 0; zap_cursor_advance(basezc)) { nvlist_t *perms_nvp; ASSERT(baseza->za_integer_length == 8); ASSERT(baseza->za_num_integers == 1); perms_nvp = fnvlist_alloc(); for (zap_cursor_init(zc, mos, baseza->za_first_integer); zap_cursor_retrieve(zc, za) == 0; zap_cursor_advance(zc)) { fnvlist_add_boolean(perms_nvp, za->za_name); } zap_cursor_fini(zc); fnvlist_add_nvlist(sp_nvp, baseza->za_name, perms_nvp); fnvlist_free(perms_nvp); } zap_cursor_fini(basezc); dsl_dir_name(dd, source); fnvlist_add_nvlist(*nvp, source, sp_nvp); nvlist_free(sp_nvp); } kmem_free(source, MAXNAMELEN + strlen(MOS_DIR_NAME) + 1); kmem_free(baseza, sizeof (zap_attribute_t)); kmem_free(basezc, sizeof (zap_cursor_t)); kmem_free(za, sizeof (zap_attribute_t)); kmem_free(zc, sizeof (zap_cursor_t)); dsl_dir_rele(startdd, FTAG); dsl_pool_rele(dp, FTAG); return (0); }
/* * Synchronize pool configuration to disk. This must be called with the * namespace lock held. Synchronizing the pool cache is typically done after * the configuration has been synced to the MOS. This exposes a window where * the MOS config will have been updated but the cache file has not. If * the system were to crash at that instant then the cached config may not * contain the correct information to open the pool and an explicit import * would be required. */ void spa_write_cachefile(spa_t *target, boolean_t removing, boolean_t postsysevent) { spa_config_dirent_t *dp, *tdp; nvlist_t *nvl; char *pool_name; ASSERT(MUTEX_HELD(&spa_namespace_lock)); if (/*rootdir == NULL ||*/ !(spa_mode_global & FWRITE)) return; /* * Iterate over all cachefiles for the pool, past or present. When the * cachefile is changed, the new one is pushed onto this list, allowing * us to update previous cachefiles that no longer contain this pool. */ for (dp = list_head(&target->spa_config_list); dp != NULL; dp = list_next(&target->spa_config_list, dp)) { spa_t *spa = NULL; if (dp->scd_path == NULL) continue; /* * Iterate over all pools, adding any matching pools to 'nvl'. */ nvl = NULL; while ((spa = spa_next(spa)) != NULL) { /* * Skip over our own pool if we're about to remove * ourselves from the spa namespace or any pool that * is readonly. Since we cannot guarantee that a * readonly pool would successfully import upon reboot, * we don't allow them to be written to the cache file. */ if ((spa == target && removing) || !spa_writeable(spa)) continue; mutex_enter(&spa->spa_props_lock); tdp = list_head(&spa->spa_config_list); if (spa->spa_config == NULL || tdp->scd_path == NULL || strcmp(tdp->scd_path, dp->scd_path) != 0) { mutex_exit(&spa->spa_props_lock); continue; } if (nvl == NULL) nvl = fnvlist_alloc(); if (spa->spa_import_flags & ZFS_IMPORT_TEMP_NAME) { VERIFY0(nvlist_lookup_string(spa->spa_config, ZPOOL_CONFIG_POOL_NAME, &pool_name)); } else pool_name = spa_name(spa); fnvlist_add_nvlist(nvl, spa->spa_name, spa->spa_config); mutex_exit(&spa->spa_props_lock); } spa_config_write(dp, nvl); nvlist_free(nvl); } /* * Remove any config entries older than the current one. */ dp = list_head(&target->spa_config_list); while ((tdp = list_next(&target->spa_config_list, dp)) != NULL) { list_remove(&target->spa_config_list, tdp); if (tdp->scd_path != NULL) spa_strfree(tdp->scd_path); kmem_free(tdp, sizeof (spa_config_dirent_t)); } spa_config_generation++; if (postsysevent) { /* We don't have spa in spa_config_write, so handle the events here */ if (nvl == NULL) { zfs_ereport_post(FM_EREPORT_ZFS_CONFIG_REMOVE, target, NULL, NULL, NULL, (uint64_t)dp->scd_path, 0); } else { zfs_ereport_post(FM_EREPORT_ZFS_CONFIG_RENAME, target, NULL, NULL, NULL, (uint64_t)dp->scd_path, 0); } } }
/* * Find all 'allow' permissions from a given point and then continue * traversing up to the root. * * This function constructs an nvlist of nvlists. * each setpoint is an nvlist composed of an nvlist of an nvlist * of the individual * users/groups/everyone/create * permissions. * * The nvlist will look like this. * * { source fsname -> { whokeys { permissions,...}, ...}} * * The fsname nvpairs will be arranged in a bottom up order. For example, * if we have the following structure a/b/c then the nvpairs for the fsnames * will be ordered a/b/c, a/b, a. */ int dsl_deleg_get(const char *ddname, nvlist_t **nvp) { dsl_dir_t *dd, *startdd; dsl_pool_t *dp; int error; objset_t *mos; error = dsl_pool_hold(ddname, FTAG, &dp); if (error != 0) return (error); error = dsl_dir_hold(dp, ddname, FTAG, &startdd, NULL); if (error != 0) { dsl_pool_rele(dp, FTAG); return (error); } dp = startdd->dd_pool; mos = dp->dp_meta_objset; VERIFY(nvlist_alloc(nvp, NV_UNIQUE_NAME, KM_SLEEP) == 0); for (dd = startdd; dd != NULL; dd = dd->dd_parent) { zap_cursor_t basezc; zap_attribute_t baseza; nvlist_t *sp_nvp; uint64_t n; char source[ZFS_MAX_DATASET_NAME_LEN]; if (dsl_dir_phys(dd)->dd_deleg_zapobj == 0 || zap_count(mos, dsl_dir_phys(dd)->dd_deleg_zapobj, &n) != 0 || n == 0) continue; sp_nvp = fnvlist_alloc(); for (zap_cursor_init(&basezc, mos, dsl_dir_phys(dd)->dd_deleg_zapobj); zap_cursor_retrieve(&basezc, &baseza) == 0; zap_cursor_advance(&basezc)) { zap_cursor_t zc; zap_attribute_t za; nvlist_t *perms_nvp; ASSERT(baseza.za_integer_length == 8); ASSERT(baseza.za_num_integers == 1); perms_nvp = fnvlist_alloc(); for (zap_cursor_init(&zc, mos, baseza.za_first_integer); zap_cursor_retrieve(&zc, &za) == 0; zap_cursor_advance(&zc)) { fnvlist_add_boolean(perms_nvp, za.za_name); } zap_cursor_fini(&zc); fnvlist_add_nvlist(sp_nvp, baseza.za_name, perms_nvp); fnvlist_free(perms_nvp); } zap_cursor_fini(&basezc); dsl_dir_name(dd, source); fnvlist_add_nvlist(*nvp, source, sp_nvp); nvlist_free(sp_nvp); } dsl_dir_rele(startdd, FTAG); dsl_pool_rele(dp, FTAG); return (0); }