From e814e467e52fa4f982cf3d29a44b6b8fd8b1281e Mon Sep 17 00:00:00 2001 From: mav Date: Wed, 3 Oct 2018 02:16:22 +0000 Subject: [PATCH] MFC r337007: MFV r336991, r337001: 9102 zfs should be able to initialize storage devices The first access to a disk block can incur a performance penalty on some platforms (e.g. AWS's EBS, VMware VMDKs). Therefore it is recommended that volumes be "thick provisioned", where supported by the platform (VMware). Thick provisioning is time consuming and often is ignored. If the thick provision step is omitted, customers will see suboptimal performance until we have written to all parts of the LUN. ZFS should be able to initialize any unused storage to remove any first-write penalty that exists. illumos/illumos-gate@094e47e980b0796b94b1b8f51f462a64d246e516 Reviewed by: John Wren Kennedy Reviewed by: Matthew Ahrens Reviewed by: Pavel Zakharov Reviewed by: Prakash Surya Approved by: Richard Lowe Author: George Wilson --- cddl/contrib/opensolaris/cmd/zpool/zpool.8 | 31 + .../opensolaris/cmd/zpool/zpool_main.c | 155 ++++ cddl/contrib/opensolaris/cmd/ztest/ztest.c | 96 ++- .../opensolaris/lib/libzfs/common/libzfs.h | 5 + .../lib/libzfs/common/libzfs_pool.c | 94 +++ .../lib/libzfs/common/libzfs_util.c | 7 + .../lib/libzfs_core/common/libzfs_core.c | 37 + .../lib/libzfs_core/common/libzfs_core.h | 4 + .../opensolaris/uts/common/Makefile.files | 1 + .../opensolaris/uts/common/fs/zfs/metaslab.c | 23 +- .../opensolaris/uts/common/fs/zfs/spa.c | 158 +++- .../opensolaris/uts/common/fs/zfs/spa_misc.c | 7 + .../uts/common/fs/zfs/sys/metaslab_impl.h | 10 +- .../opensolaris/uts/common/fs/zfs/sys/spa.h | 2 + .../uts/common/fs/zfs/sys/vdev_impl.h | 31 + .../uts/common/fs/zfs/sys/vdev_initialize.h | 46 + .../uts/common/fs/zfs/sys/zio_priority.h | 3 +- .../opensolaris/uts/common/fs/zfs/vdev.c | 45 +- .../opensolaris/uts/common/fs/zfs/vdev_disk.c | 1 + .../opensolaris/uts/common/fs/zfs/vdev_file.c | 4 +- .../opensolaris/uts/common/fs/zfs/vdev_geom.c | 1 + .../uts/common/fs/zfs/vdev_indirect.c | 1 + .../uts/common/fs/zfs/vdev_initialize.c | 792 ++++++++++++++++++ .../uts/common/fs/zfs/vdev_mirror.c | 3 + .../uts/common/fs/zfs/vdev_missing.c | 4 +- .../uts/common/fs/zfs/vdev_queue.c | 16 +- .../uts/common/fs/zfs/vdev_raidz.c | 75 ++ .../uts/common/fs/zfs/vdev_removal.c | 13 + .../opensolaris/uts/common/fs/zfs/vdev_root.c | 3 +- .../opensolaris/uts/common/fs/zfs/zfs_ioctl.c | 80 ++ .../opensolaris/uts/common/sys/fs/zfs.h | 36 + sys/conf/files | 1 + 32 files changed, 1757 insertions(+), 28 deletions(-) create mode 100644 sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_initialize.h create mode 100644 sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_initialize.c diff --git a/cddl/contrib/opensolaris/cmd/zpool/zpool.8 b/cddl/contrib/opensolaris/cmd/zpool/zpool.8 index b0c21436308..d5a8d97b143 100644 --- a/cddl/contrib/opensolaris/cmd/zpool/zpool.8 +++ b/cddl/contrib/opensolaris/cmd/zpool/zpool.8 @@ -121,6 +121,11 @@ .Ar pool | id .Op Ar newpool .Nm +.Cm initialize +.Op Fl cs +.Ar pool +.Op Ar device Ns ... +.Nm .Cm iostat .Op Fl T Cm d Ns | Ns Cm u .Op Fl v @@ -1437,6 +1442,32 @@ to fully rewind. .El .It Xo .Nm +.Cm initialize +.Op Fl cs +.Ar pool +.Op Ar device Ns ... +.Xc +Begins initializing by writing to all unallocated regions on the specified +devices, or all eligible devices in the pool if no individual devices are +specified. +Only leaf data or log devices may be initialized. +.Bl -tag -width Ds +.It Fl c, -cancel +Cancel initializing on the specified devices, or all eligible devices if none +are specified. +If one or more target devices are invalid or are not currently being +initialized, the command will fail and no cancellation will occur on any device. +.It Fl s -suspend +Suspend initializing on the specified devices, or all eligible devices if none +are specified. +If one or more target devices are invalid or are not currently being +initialized, the command will fail and no suspension will occur on any device. +Initializing can then be resumed by running +.Nm zpool Cm initialize +with no flags on the relevant target devices. +.El +.It Xo +.Nm .Cm iostat .Op Fl T Cm d Ns | Ns Cm u .Op Fl v diff --git a/cddl/contrib/opensolaris/cmd/zpool/zpool_main.c b/cddl/contrib/opensolaris/cmd/zpool/zpool_main.c index cb0c777bb76..1a81991c1d2 100644 --- a/cddl/contrib/opensolaris/cmd/zpool/zpool_main.c +++ b/cddl/contrib/opensolaris/cmd/zpool/zpool_main.c @@ -87,6 +87,7 @@ static int zpool_do_detach(int, char **); static int zpool_do_replace(int, char **); static int zpool_do_split(int, char **); +static int zpool_do_initialize(int, char **); static int zpool_do_scrub(int, char **); static int zpool_do_import(int, char **); @@ -136,6 +137,7 @@ typedef enum { HELP_ONLINE, HELP_REPLACE, HELP_REMOVE, + HELP_INITIALIZE, HELP_SCRUB, HELP_STATUS, HELP_UPGRADE, @@ -187,6 +189,7 @@ static zpool_command_t command_table[] = { { "replace", zpool_do_replace, HELP_REPLACE }, { "split", zpool_do_split, HELP_SPLIT }, { NULL }, + { "initialize", zpool_do_initialize, HELP_INITIALIZE }, { "scrub", zpool_do_scrub, HELP_SCRUB }, { NULL }, { "import", zpool_do_import, HELP_IMPORT }, @@ -261,6 +264,8 @@ get_usage(zpool_help_t idx) return (gettext("\tremove [-nps] ...\n")); case HELP_REOPEN: return (gettext("\treopen \n")); + case HELP_INITIALIZE: + return (gettext("\tinitialize [-cs] [ ...]\n")); case HELP_SCRUB: return (gettext("\tscrub [-s | -p] ...\n")); case HELP_STATUS: @@ -1650,6 +1655,43 @@ print_status_config(zpool_handle_t *zhp, const char *name, nvlist_t *nv, "resilvering" : "repairing"); } + if ((vs->vs_initialize_state == VDEV_INITIALIZE_ACTIVE || + vs->vs_initialize_state == VDEV_INITIALIZE_SUSPENDED || + vs->vs_initialize_state == VDEV_INITIALIZE_COMPLETE) && + !vs->vs_scan_removing) { + char zbuf[1024]; + char tbuf[256]; + struct tm zaction_ts; + + time_t t = vs->vs_initialize_action_time; + int initialize_pct = 100; + if (vs->vs_initialize_state != VDEV_INITIALIZE_COMPLETE) { + initialize_pct = (vs->vs_initialize_bytes_done * 100 / + (vs->vs_initialize_bytes_est + 1)); + } + + (void) localtime_r(&t, &zaction_ts); + (void) strftime(tbuf, sizeof (tbuf), "%c", &zaction_ts); + + switch (vs->vs_initialize_state) { + case VDEV_INITIALIZE_SUSPENDED: + (void) snprintf(zbuf, sizeof (zbuf), + ", suspended, started at %s", tbuf); + break; + case VDEV_INITIALIZE_ACTIVE: + (void) snprintf(zbuf, sizeof (zbuf), + ", started at %s", tbuf); + break; + case VDEV_INITIALIZE_COMPLETE: + (void) snprintf(zbuf, sizeof (zbuf), + ", completed at %s", tbuf); + break; + } + + (void) printf(gettext(" (%d%% initialized%s)"), + initialize_pct, zbuf); + } + (void) printf("\n"); for (c = 0; c < children; c++) { @@ -4238,6 +4280,119 @@ zpool_do_scrub(int argc, char **argv) return (for_each_pool(argc, argv, B_TRUE, NULL, scrub_callback, &cb)); } +static void +zpool_collect_leaves(zpool_handle_t *zhp, nvlist_t *nvroot, nvlist_t *res) +{ + uint_t children = 0; + nvlist_t **child; + uint_t i; + + (void) nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, + &child, &children); + + if (children == 0) { + char *path = zpool_vdev_name(g_zfs, zhp, nvroot, B_FALSE); + fnvlist_add_boolean(res, path); + free(path); + return; + } + + for (i = 0; i < children; i++) { + zpool_collect_leaves(zhp, child[i], res); + } +} + +/* + * zpool initialize [-cs] [ ...] + * Initialize all unused blocks in the specified vdevs, or all vdevs in the pool + * if none specified. + * + * -c Cancel. Ends active initializing. + * -s Suspend. Initializing can then be restarted with no flags. + */ +int +zpool_do_initialize(int argc, char **argv) +{ + int c; + char *poolname; + zpool_handle_t *zhp; + nvlist_t *vdevs; + int err = 0; + + struct option long_options[] = { + {"cancel", no_argument, NULL, 'c'}, + {"suspend", no_argument, NULL, 's'}, + {0, 0, 0, 0} + }; + + pool_initialize_func_t cmd_type = POOL_INITIALIZE_DO; + while ((c = getopt_long(argc, argv, "cs", long_options, NULL)) != -1) { + switch (c) { + case 'c': + if (cmd_type != POOL_INITIALIZE_DO) { + (void) fprintf(stderr, gettext("-c cannot be " + "combined with other options\n")); + usage(B_FALSE); + } + cmd_type = POOL_INITIALIZE_CANCEL; + break; + case 's': + if (cmd_type != POOL_INITIALIZE_DO) { + (void) fprintf(stderr, gettext("-s cannot be " + "combined with other options\n")); + usage(B_FALSE); + } + cmd_type = POOL_INITIALIZE_SUSPEND; + break; + case '?': + if (optopt != 0) { + (void) fprintf(stderr, + gettext("invalid option '%c'\n"), optopt); + } else { + (void) fprintf(stderr, + gettext("invalid option '%s'\n"), + argv[optind - 1]); + } + usage(B_FALSE); + } + } + + argc -= optind; + argv += optind; + + if (argc < 1) { + (void) fprintf(stderr, gettext("missing pool name argument\n")); + usage(B_FALSE); + return (-1); + } + + poolname = argv[0]; + zhp = zpool_open(g_zfs, poolname); + if (zhp == NULL) + return (-1); + + vdevs = fnvlist_alloc(); + if (argc == 1) { + /* no individual leaf vdevs specified, so add them all */ + nvlist_t *config = zpool_get_config(zhp, NULL); + nvlist_t *nvroot = fnvlist_lookup_nvlist(config, + ZPOOL_CONFIG_VDEV_TREE); + zpool_collect_leaves(zhp, nvroot, vdevs); + } else { + int i; + for (i = 1; i < argc; i++) { + fnvlist_add_boolean(vdevs, argv[i]); + } + } + + err = zpool_initialize(zhp, cmd_type, vdevs); + + fnvlist_free(vdevs); + zpool_close(zhp); + + return (err); +} + typedef struct status_cbdata { int cb_count; boolean_t cb_allpools; diff --git a/cddl/contrib/opensolaris/cmd/ztest/ztest.c b/cddl/contrib/opensolaris/cmd/ztest/ztest.c index e5fcea98b62..af8bd1867da 100644 --- a/cddl/contrib/opensolaris/cmd/ztest/ztest.c +++ b/cddl/contrib/opensolaris/cmd/ztest/ztest.c @@ -104,6 +104,7 @@ #include #include #include +#include #include #include #include @@ -348,6 +349,7 @@ ztest_func_t ztest_spa_upgrade; ztest_func_t ztest_device_removal; ztest_func_t ztest_remap_blocks; ztest_func_t ztest_spa_checkpoint_create_discard; +ztest_func_t ztest_initialize; uint64_t zopt_always = 0ULL * NANOSEC; /* all the time */ uint64_t zopt_incessant = 1ULL * NANOSEC / 10; /* every 1/10 second */ @@ -391,7 +393,8 @@ ztest_info_t ztest_info[] = { &ztest_opts.zo_vdevtime }, { ztest_device_removal, 1, &zopt_sometimes }, { ztest_remap_blocks, 1, &zopt_sometimes }, - { ztest_spa_checkpoint_create_discard, 1, &zopt_rarely } + { ztest_spa_checkpoint_create_discard, 1, &zopt_rarely }, + { ztest_initialize, 1, &zopt_sometimes } }; #define ZTEST_FUNCS (sizeof (ztest_info) / sizeof (ztest_info_t)) @@ -5471,6 +5474,97 @@ ztest_spa_rename(ztest_ds_t *zd, uint64_t id) rw_exit(&ztest_name_lock); } +static vdev_t * +ztest_random_concrete_vdev_leaf(vdev_t *vd) +{ + if (vd == NULL) + return (NULL); + + if (vd->vdev_children == 0) + return (vd); + + vdev_t *eligible[vd->vdev_children]; + int eligible_idx = 0, i; + for (i = 0; i < vd->vdev_children; i++) { + vdev_t *cvd = vd->vdev_child[i]; + if (cvd->vdev_top->vdev_removing) + continue; + if (cvd->vdev_children > 0 || + (vdev_is_concrete(cvd) && !cvd->vdev_detached)) { + eligible[eligible_idx++] = cvd; + } + } + VERIFY(eligible_idx > 0); + + uint64_t child_no = ztest_random(eligible_idx); + return (ztest_random_concrete_vdev_leaf(eligible[child_no])); +} + +/* ARGSUSED */ +void +ztest_initialize(ztest_ds_t *zd, uint64_t id) +{ + spa_t *spa = ztest_spa; + int error = 0; + + mutex_enter(&ztest_vdev_lock); + + spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER); + + /* Random leaf vdev */ + vdev_t *rand_vd = ztest_random_concrete_vdev_leaf(spa->spa_root_vdev); + if (rand_vd == NULL) { + spa_config_exit(spa, SCL_VDEV, FTAG); + mutex_exit(&ztest_vdev_lock); + return; + } + + /* + * The random vdev we've selected may change as soon as we + * drop the spa_config_lock. We create local copies of things + * we're interested in. + */ + uint64_t guid = rand_vd->vdev_guid; + char *path = strdup(rand_vd->vdev_path); + boolean_t active = rand_vd->vdev_initialize_thread != NULL; + + zfs_dbgmsg("vd %p, guid %llu", rand_vd, guid); + spa_config_exit(spa, SCL_VDEV, FTAG); + + uint64_t cmd = ztest_random(POOL_INITIALIZE_FUNCS); + error = spa_vdev_initialize(spa, guid, cmd); + switch (cmd) { + case POOL_INITIALIZE_CANCEL: + if (ztest_opts.zo_verbose >= 4) { + (void) printf("Cancel initialize %s", path); + if (!active) + (void) printf(" failed (no initialize active)"); + (void) printf("\n"); + } + break; + case POOL_INITIALIZE_DO: + if (ztest_opts.zo_verbose >= 4) { + (void) printf("Start initialize %s", path); + if (active && error == 0) + (void) printf(" failed (already active)"); + else if (error != 0) + (void) printf(" failed (error %d)", error); + (void) printf("\n"); + } + break; + case POOL_INITIALIZE_SUSPEND: + if (ztest_opts.zo_verbose >= 4) { + (void) printf("Suspend initialize %s", path); + if (!active) + (void) printf(" failed (no initialize active)"); + (void) printf("\n"); + } + break; + } + free(path); + mutex_exit(&ztest_vdev_lock); +} + /* * Verify pool integrity by running zdb. */ diff --git a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs.h b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs.h index d518abb8664..0c50eb89eac 100644 --- a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs.h +++ b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs.h @@ -137,6 +137,9 @@ typedef enum zfs_error { EZFS_NO_CHECKPOINT, /* pool has no checkpoint */ EZFS_DEVRM_IN_PROGRESS, /* a device is currently being removed */ EZFS_VDEV_TOO_BIG, /* a device is too big to be used */ + EZFS_TOOMANY, /* argument list too long */ + EZFS_INITIALIZING, /* currently initializing */ + EZFS_NO_INITIALIZE, /* no active initialize */ EZFS_UNKNOWN } zfs_error_t; @@ -262,6 +265,8 @@ typedef struct splitflags { * Functions to manipulate pool and vdev state */ extern int zpool_scan(zpool_handle_t *, pool_scan_func_t, pool_scrub_cmd_t); +extern int zpool_initialize(zpool_handle_t *, pool_initialize_func_t, + nvlist_t *); extern int zpool_clear(zpool_handle_t *, const char *, nvlist_t *); extern int zpool_reguid(zpool_handle_t *); extern int zpool_reopen(zpool_handle_t *); diff --git a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_pool.c b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_pool.c index c94963bf571..392ed3cfa86 100644 --- a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_pool.c +++ b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_pool.c @@ -1981,6 +1981,100 @@ zpool_scan(zpool_handle_t *zhp, pool_scan_func_t func, pool_scrub_cmd_t cmd) } } +static int +xlate_init_err(int err) +{ + switch (err) { + case ENODEV: + return (EZFS_NODEVICE); + case EINVAL: + case EROFS: + return (EZFS_BADDEV); + case EBUSY: + return (EZFS_INITIALIZING); + case ESRCH: + return (EZFS_NO_INITIALIZE); + } + return (err); +} + +/* + * Begin, suspend, or cancel the initialization (initializing of all free + * blocks) for the given vdevs in the given pool. + */ +int +zpool_initialize(zpool_handle_t *zhp, pool_initialize_func_t cmd_type, + nvlist_t *vds) +{ + char msg[1024]; + libzfs_handle_t *hdl = zhp->zpool_hdl; + + nvlist_t *errlist; + + /* translate vdev names to guids */ + nvlist_t *vdev_guids = fnvlist_alloc(); + nvlist_t *guids_to_paths = fnvlist_alloc(); + boolean_t spare, cache; + nvlist_t *tgt; + nvpair_t *elem; + + for (elem = nvlist_next_nvpair(vds, NULL); elem != NULL; + elem = nvlist_next_nvpair(vds, elem)) { + char *vd_path = nvpair_name(elem); + tgt = zpool_find_vdev(zhp, vd_path, &spare, &cache, NULL); + + if ((tgt == NULL) || cache || spare) { + (void) snprintf(msg, sizeof (msg), + dgettext(TEXT_DOMAIN, "cannot initialize '%s'"), + vd_path); + int err = (tgt == NULL) ? EZFS_NODEVICE : + (spare ? EZFS_ISSPARE : EZFS_ISL2CACHE); + fnvlist_free(vdev_guids); + fnvlist_free(guids_to_paths); + return (zfs_error(hdl, err, msg)); + } + + uint64_t guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID); + fnvlist_add_uint64(vdev_guids, vd_path, guid); + + (void) snprintf(msg, sizeof (msg), "%llu", guid); + fnvlist_add_string(guids_to_paths, msg, vd_path); + } + + int err = lzc_initialize(zhp->zpool_name, cmd_type, vdev_guids, + &errlist); + fnvlist_free(vdev_guids); + + if (err == 0) { + fnvlist_free(guids_to_paths); + return (0); + } + + nvlist_t *vd_errlist = NULL; + if (errlist != NULL) { + vd_errlist = fnvlist_lookup_nvlist(errlist, + ZPOOL_INITIALIZE_VDEVS); + } + + (void) snprintf(msg, sizeof (msg), + dgettext(TEXT_DOMAIN, "operation failed")); + + for (elem = nvlist_next_nvpair(vd_errlist, NULL); elem != NULL; + elem = nvlist_next_nvpair(vd_errlist, elem)) { + int64_t vd_error = xlate_init_err(fnvpair_value_int64(elem)); + char *path = fnvlist_lookup_string(guids_to_paths, + nvpair_name(elem)); + (void) zfs_error_fmt(hdl, vd_error, "cannot initialize '%s'", + path); + } + + fnvlist_free(guids_to_paths); + if (vd_errlist != NULL) + return (-1); + + return (zpool_standard_error(hdl, err, msg)); +} + #ifdef illumos /* * This provides a very minimal check whether a given string is likely a diff --git a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_util.c b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_util.c index 6b52f9ad86d..faf00486751 100644 --- a/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_util.c +++ b/cddl/contrib/opensolaris/lib/libzfs/common/libzfs_util.c @@ -254,6 +254,13 @@ libzfs_error_description(libzfs_handle_t *hdl) return (dgettext(TEXT_DOMAIN, "device removal in progress")); case EZFS_VDEV_TOO_BIG: return (dgettext(TEXT_DOMAIN, "device exceeds supported size")); + case EZFS_TOOMANY: + return (dgettext(TEXT_DOMAIN, "argument list too long")); + case EZFS_INITIALIZING: + return (dgettext(TEXT_DOMAIN, "currently initializing")); + case EZFS_NO_INITIALIZE: + return (dgettext(TEXT_DOMAIN, "there is no active " + "initialization")); case EZFS_UNKNOWN: return (dgettext(TEXT_DOMAIN, "unknown error")); default: diff --git a/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.c b/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.c index 3dda53395a1..61fb4af2419 100644 --- a/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.c +++ b/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.c @@ -1085,3 +1085,40 @@ lzc_channel_program_nosync(const char *pool, const char *program, return (lzc_channel_program_impl(pool, program, B_FALSE, timeout, memlimit, argnvl, outnvl)); } + +/* + * Changes initializing state. + * + * vdevs should be a list of (, guid) where guid is a uint64 vdev GUID. + * The key is ignored. + * + * If there are errors related to vdev arguments, per-vdev errors are returned + * in an nvlist with the key "vdevs". Each error is a (guid, errno) pair where + * guid is stringified with PRIu64, and errno is one of the following as + * an int64_t: + * - ENODEV if the device was not found + * - EINVAL if the devices is not a leaf or is not concrete (e.g. missing) + * - EROFS if the device is not writeable + * - EBUSY start requested but the device is already being initialized + * - ESRCH cancel/suspend requested but device is not being initialized + * + * If the errlist is empty, then return value will be: + * - EINVAL if one or more arguments was invalid + * - Other spa_open failures + * - 0 if the operation succeeded + */ +int +lzc_initialize(const char *poolname, pool_initialize_func_t cmd_type, + nvlist_t *vdevs, nvlist_t **errlist) +{ + int error; + nvlist_t *args = fnvlist_alloc(); + fnvlist_add_uint64(args, ZPOOL_INITIALIZE_COMMAND, (uint64_t)cmd_type); + fnvlist_add_nvlist(args, ZPOOL_INITIALIZE_VDEVS, vdevs); + + error = lzc_ioctl(ZFS_IOC_POOL_INITIALIZE, poolname, args, errlist); + + fnvlist_free(args); + + return (error); +} diff --git a/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.h b/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.h index 0eaa89cc1e9..24781cb0fd6 100644 --- a/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.h +++ b/cddl/contrib/opensolaris/lib/libzfs_core/common/libzfs_core.h @@ -31,6 +31,8 @@ #include #include #include +#include + #ifdef __cplusplus extern "C" { @@ -56,6 +58,8 @@ int lzc_destroy_snaps(nvlist_t *, boolean_t, nvlist_t **); int lzc_bookmark(nvlist_t *, nvlist_t **); int lzc_get_bookmarks(const char *, nvlist_t *, nvlist_t **); int lzc_destroy_bookmarks(nvlist_t *, nvlist_t **); +int lzc_initialize(const char *, pool_initialize_func_t, nvlist_t *, + nvlist_t **); int lzc_snaprange_space(const char *, const char *, uint64_t *); diff --git a/sys/cddl/contrib/opensolaris/uts/common/Makefile.files b/sys/cddl/contrib/opensolaris/uts/common/Makefile.files index 8b6720e619f..27c5fa6e06d 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/Makefile.files +++ b/sys/cddl/contrib/opensolaris/uts/common/Makefile.files @@ -124,6 +124,7 @@ ZFS_COMMON_OBJS += \ vdev_indirect.o \ vdev_indirect_births.o \ vdev_indirect_mapping.o \ + vdev_initialize.o \ vdev_label.o \ vdev_mirror.o \ vdev_missing.o \ diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c index e4722b3146f..c562675b1ae 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/metaslab.c @@ -725,6 +725,8 @@ metaslab_group_create(metaslab_class_t *mc, vdev_t *vd, int allocators) mg = kmem_zalloc(sizeof (metaslab_group_t), KM_SLEEP); mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL); + mutex_init(&mg->mg_ms_initialize_lock, NULL, MUTEX_DEFAULT, NULL); + cv_init(&mg->mg_ms_initialize_cv, NULL, CV_DEFAULT, NULL); mg->mg_primaries = kmem_zalloc(allocators * sizeof (metaslab_t *), KM_SLEEP); mg->mg_secondaries = kmem_zalloc(allocators * sizeof (metaslab_t *), @@ -771,6 +773,8 @@ metaslab_group_destroy(metaslab_group_t *mg) kmem_free(mg->mg_secondaries, mg->mg_allocators * sizeof (metaslab_t *)); mutex_destroy(&mg->mg_lock); + mutex_destroy(&mg->mg_ms_initialize_lock); + cv_destroy(&mg->mg_ms_initialize_cv); for (int i = 0; i < mg->mg_allocators; i++) { refcount_destroy(&mg->mg_alloc_queue_depth[i]); @@ -1554,6 +1558,7 @@ metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object, uint64_t txg, mutex_init(&ms->ms_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&ms->ms_sync_lock, NULL, MUTEX_DEFAULT, NULL); cv_init(&ms->ms_load_cv, NULL, CV_DEFAULT, NULL); + ms->ms_id = id; ms->ms_start = id << vd->vdev_ms_shift; ms->ms_size = 1ULL << vd->vdev_ms_shift; @@ -2731,6 +2736,7 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg) * from it in 'metaslab_unload_delay' txgs, then unload it. */ if (msp->ms_loaded && + msp->ms_initializing == 0 && msp->ms_selected_txg + metaslab_unload_delay < txg) { for (int t = 1; t < TXG_CONCURRENT_STATES; t++) { VERIFY0(range_tree_space( @@ -2980,6 +2986,7 @@ metaslab_block_alloc(metaslab_t *msp, uint64_t size, uint64_t txg) metaslab_class_t *mc = msp->ms_group->mg_class; VERIFY(!msp->ms_condensing); + VERIFY0(msp->ms_initializing); start = mc->mc_ops->msop_alloc(msp, size); if (start != -1ULL) { @@ -3040,9 +3047,10 @@ find_valid_metaslab(metaslab_group_t *mg, uint64_t activation_weight, } /* - * If the selected metaslab is condensing, skip it. + * If the selected metaslab is condensing or being + * initialized, skip it. */ - if (msp->ms_condensing) + if (msp->ms_condensing || msp->ms_initializing > 0) continue; *was_active = msp->ms_allocator != -1; @@ -3207,7 +3215,9 @@ metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal, /* * If this metaslab is currently condensing then pick again as * we can't manipulate this metaslab until it's committed - * to disk. + * to disk. If this metaslab is being initialized, we shouldn't + * allocate from it since the allocated region might be + * overwritten after allocation. */ if (msp->ms_condensing) { metaslab_trace_add(zal, mg, msp, asize, d, @@ -3216,6 +3226,13 @@ metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal, ~METASLAB_ACTIVE_MASK); mutex_exit(&msp->ms_lock); continue; + } else if (msp->ms_initializing > 0) { + metaslab_trace_add(zal, mg, msp, asize, d, + TRACE_INITIALIZING, allocator); + metaslab_passivate(msp, msp->ms_weight & + ~METASLAB_ACTIVE_MASK); + mutex_exit(&msp->ms_lock); + continue; } offset = metaslab_block_alloc(msp, asize, txg); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c index ec06a1d108a..2d95e1f49eb 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa.c @@ -55,6 +55,7 @@ #include #include #include +#include #include #include #include @@ -443,8 +444,9 @@ spa_prop_get(spa_t *spa, nvlist_t **nvp) dp = spa_get_dsl(spa); dsl_pool_config_enter(dp, FTAG); - if (err = dsl_dataset_hold_obj(dp, - za.za_first_integer, FTAG, &ds)) { + err = dsl_dataset_hold_obj(dp, + za.za_first_integer, FTAG, &ds); + if (err != 0) { dsl_pool_config_exit(dp, FTAG); break; } @@ -599,7 +601,8 @@ spa_prop_validate(spa_t *spa, nvlist_t *props) break; } - if (error = dmu_objset_hold(strval, FTAG, &os)) + error = dmu_objset_hold(strval, FTAG, &os); + if (error != 0) break; /* @@ -1215,8 +1218,10 @@ spa_activate(spa_t *spa, int mode) */ trim_thread_create(spa); - for (size_t i = 0; i < TXG_SIZE; i++) - spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL, 0); + for (size_t i = 0; i < TXG_SIZE; i++) { + spa->spa_txg_zio[i] = zio_root(spa, NULL, NULL, + ZIO_FLAG_CANFAIL); + } list_create(&spa->spa_config_dirty_list, sizeof (vdev_t), offsetof(vdev_t, vdev_config_dirty_node)); @@ -1388,6 +1393,11 @@ spa_unload(spa_t *spa) */ spa_async_suspend(spa); + if (spa->spa_root_vdev) { + vdev_initialize_stop_all(spa->spa_root_vdev, + VDEV_INITIALIZE_ACTIVE); + } + /* * Stop syncing. */ @@ -1403,10 +1413,10 @@ spa_unload(spa_t *spa) * calling taskq_wait(mg_taskq). */ if (spa->spa_root_vdev != NULL) { - spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); + spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); for (int c = 0; c < spa->spa_root_vdev->vdev_children; c++) vdev_metaslab_fini(spa->spa_root_vdev->vdev_child[c]); - spa_config_exit(spa, SCL_ALL, FTAG); + spa_config_exit(spa, SCL_ALL, spa); } /* @@ -1440,7 +1450,7 @@ spa_unload(spa_t *spa) bpobj_close(&spa->spa_deferred_bpobj); - spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER); + spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); /* * Close all vdevs. @@ -1502,7 +1512,7 @@ spa_unload(spa_t *spa) spa->spa_comment = NULL; } - spa_config_exit(spa, SCL_ALL, FTAG); + spa_config_exit(spa, SCL_ALL, spa); } /* @@ -3954,6 +3964,10 @@ spa_load_impl(spa_t *spa, spa_import_type_t type, char **ereport) spa_restart_removal(spa); spa_spawn_aux_threads(spa); + + spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); + vdev_initialize_restart(spa->spa_root_vdev); + spa_config_exit(spa, SCL_CONFIG, FTAG); } spa_load_note(spa, "LOADED"); @@ -5676,6 +5690,7 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig, * in which case we can modify its state. */ if (spa->spa_state != POOL_STATE_UNINITIALIZED && spa->spa_sync_on) { + /* * Objsets may be open only because they're dirty, so we * have to force it to sync before checking spa_refcnt. @@ -5709,6 +5724,18 @@ spa_export_common(char *pool, int new_state, nvlist_t **oldconfig, return (SET_ERROR(EXDEV)); } + /* + * We're about to export or destroy this pool. Make sure + * we stop all initializtion activity here before we + * set the spa_final_txg. This will ensure that all + * dirty data resulting from the initialization is + * committed to disk before we unload the pool. + */ + if (spa->spa_root_vdev != NULL) { + vdev_initialize_stop_all(spa->spa_root_vdev, + VDEV_INITIALIZE_ACTIVE); + } + /* * We want this to be reflected on every label, * so mark them all dirty. spa_unload() will do the @@ -6399,6 +6426,86 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done) return (error); } +int +spa_vdev_initialize(spa_t *spa, uint64_t guid, uint64_t cmd_type) +{ + /* + * We hold the namespace lock through the whole function + * to prevent any changes to the pool while we're starting or + * stopping initialization. The config and state locks are held so that + * we can properly assess the vdev state before we commit to + * the initializing operation. + */ + mutex_enter(&spa_namespace_lock); + spa_config_enter(spa, SCL_CONFIG | SCL_STATE, FTAG, RW_READER); + + /* Look up vdev and ensure it's a leaf. */ + vdev_t *vd = spa_lookup_by_guid(spa, guid, B_FALSE); + if (vd == NULL || vd->vdev_detached) { + spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(ENODEV)); + } else if (!vd->vdev_ops->vdev_op_leaf || !vdev_is_concrete(vd)) { + spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(EINVAL)); + } else if (!vdev_writeable(vd)) { + spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(EROFS)); + } + mutex_enter(&vd->vdev_initialize_lock); + spa_config_exit(spa, SCL_CONFIG | SCL_STATE, FTAG); + + /* + * When we activate an initialize action we check to see + * if the vdev_initialize_thread is NULL. We do this instead + * of using the vdev_initialize_state since there might be + * a previous initialization process which has completed but + * the thread is not exited. + */ + if (cmd_type == POOL_INITIALIZE_DO && + (vd->vdev_initialize_thread != NULL || + vd->vdev_top->vdev_removing)) { + mutex_exit(&vd->vdev_initialize_lock); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(EBUSY)); + } else if (cmd_type == POOL_INITIALIZE_CANCEL && + (vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE && + vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED)) { + mutex_exit(&vd->vdev_initialize_lock); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(ESRCH)); + } else if (cmd_type == POOL_INITIALIZE_SUSPEND && + vd->vdev_initialize_state != VDEV_INITIALIZE_ACTIVE) { + mutex_exit(&vd->vdev_initialize_lock); + mutex_exit(&spa_namespace_lock); + return (SET_ERROR(ESRCH)); + } + + switch (cmd_type) { + case POOL_INITIALIZE_DO: + vdev_initialize(vd); + break; + case POOL_INITIALIZE_CANCEL: + vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED); + break; + case POOL_INITIALIZE_SUSPEND: + vdev_initialize_stop(vd, VDEV_INITIALIZE_SUSPENDED); + break; + default: + panic("invalid cmd_type %llu", (unsigned long long)cmd_type); + } + mutex_exit(&vd->vdev_initialize_lock); + + /* Sync out the initializing state */ + txg_wait_synced(spa->spa_dsl_pool, 0); + mutex_exit(&spa_namespace_lock); + + return (0); +} + + /* * Split a set of devices from their mirrors, and create a new pool from them. */ @@ -6606,6 +6713,19 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, spa_activate(newspa, spa_mode_global); spa_async_suspend(newspa); + for (c = 0; c < children; c++) { + if (vml[c] != NULL) { + /* + * Temporarily stop the initializing activity. We set + * the state to ACTIVE so that we know to resume + * the initializing once the split has completed. + */ + mutex_enter(&vml[c]->vdev_initialize_lock); + vdev_initialize_stop(vml[c], VDEV_INITIALIZE_ACTIVE); + mutex_exit(&vml[c]->vdev_initialize_lock); + } + } + #ifndef illumos /* mark that we are creating new spa by splitting */ newspa->spa_splitting_newspa = B_TRUE; @@ -6700,6 +6820,10 @@ spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, if (vml[c] != NULL) vml[c]->vdev_offline = B_FALSE; } + + /* restart initializing disks as necessary */ + spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART); + vdev_reopen(spa->spa_root_vdev); nvlist_free(spa->spa_config_splitting); @@ -7064,6 +7188,14 @@ spa_async_thread(void *arg) if (tasks & SPA_ASYNC_RESILVER) dsl_resilver_restart(spa->spa_dsl_pool, 0); + if (tasks & SPA_ASYNC_INITIALIZE_RESTART) { + mutex_enter(&spa_namespace_lock); + spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); + vdev_initialize_restart(spa->spa_root_vdev); + spa_config_exit(spa, SCL_CONFIG, FTAG); + mutex_exit(&spa_namespace_lock); + } + /* * Let the world know that we're done. */ @@ -7763,8 +7895,9 @@ spa_sync(spa_t *spa, uint64_t txg) * Wait for i/os issued in open context that need to complete * before this txg syncs. */ - VERIFY0(zio_wait(spa->spa_txg_zio[txg & TXG_MASK])); - spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL, 0); + (void) zio_wait(spa->spa_txg_zio[txg & TXG_MASK]); + spa->spa_txg_zio[txg & TXG_MASK] = zio_root(spa, NULL, NULL, + ZIO_FLAG_CANFAIL); /* * Lock out configuration changes. @@ -8066,7 +8199,8 @@ spa_sync(spa_t *spa, uint64_t txg) /* * Update usable space statistics. */ - while (vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))) + while ((vd = txg_list_remove(&spa->spa_vdev_txg_list, TXG_CLEAN(txg))) + != NULL) vdev_sync_done(vd, txg); spa_update_dspace(spa); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c index be96aa1c34a..b5eb820b1ec 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c @@ -41,6 +41,7 @@ #include #include #include +#include #include #include #include @@ -1317,6 +1318,12 @@ spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error, char *tag) if (vd != NULL) { ASSERT(!vd->vdev_detached || vd->vdev_dtl_sm == NULL); + if (vd->vdev_ops->vdev_op_leaf) { + mutex_enter(&vd->vdev_initialize_lock); + vdev_initialize_stop(vd, VDEV_INITIALIZE_CANCELED); + mutex_exit(&vd->vdev_initialize_lock); + } + spa_config_enter(spa, SCL_ALL, spa, RW_WRITER); vdev_free(vd); spa_config_exit(spa, SCL_ALL, spa); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/metaslab_impl.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/metaslab_impl.h index 4f3807dd5cf..5eb59df37e5 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/metaslab_impl.h +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/metaslab_impl.h @@ -68,7 +68,8 @@ typedef enum trace_alloc_type { TRACE_GROUP_FAILURE = -5ULL, TRACE_ENOSPC = -6ULL, TRACE_CONDENSING = -7ULL, - TRACE_VDEV_ERROR = -8ULL + TRACE_VDEV_ERROR = -8ULL, + TRACE_INITIALIZING = -9ULL } trace_alloc_type_t; #define METASLAB_WEIGHT_PRIMARY (1ULL << 63) @@ -271,6 +272,11 @@ struct metaslab_group { uint64_t mg_failed_allocations; uint64_t mg_fragmentation; uint64_t mg_histogram[RANGE_TREE_HISTOGRAM_SIZE]; + + int mg_ms_initializing; + boolean_t mg_initialize_updating; + kmutex_t mg_ms_initialize_lock; + kcondvar_t mg_ms_initialize_cv; }; /* @@ -361,6 +367,8 @@ struct metaslab { boolean_t ms_condense_wanted; uint64_t ms_condense_checked_txg; + uint64_t ms_initializing; /* leaves initializing this ms */ + /* * We must hold both ms_lock and ms_group->mg_lock in order to * modify ms_loaded. diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h index 34e481cb27a..b74740aa5ab 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h @@ -670,6 +670,7 @@ extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps); #define SPA_ASYNC_AUTOEXPAND 0x20 #define SPA_ASYNC_REMOVE_DONE 0x40 #define SPA_ASYNC_REMOVE_STOP 0x80 +#define SPA_ASYNC_INITIALIZE_RESTART 0x100 /* * Controls the behavior of spa_vdev_remove(). @@ -685,6 +686,7 @@ 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_initialize(spa_t *spa, uint64_t guid, uint64_t cmd_type); 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, diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_impl.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_impl.h index ae755ffcf5d..4a3af854d46 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_impl.h +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_impl.h @@ -80,6 +80,12 @@ typedef void vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd, uint64_t offset, uint64_t size, void *arg); typedef void vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size, vdev_remap_cb_t callback, void *arg); +/* + * Given a target vdev, translates the logical range "in" to the physical + * range "res" + */ +typedef void vdev_xlation_func_t(vdev_t *cvd, const range_seg_t *in, + range_seg_t *res); typedef struct vdev_ops { vdev_open_func_t *vdev_op_open; @@ -92,6 +98,11 @@ typedef struct vdev_ops { vdev_hold_func_t *vdev_op_hold; vdev_rele_func_t *vdev_op_rele; vdev_remap_func_t *vdev_op_remap; + /* + * For translating ranges from non-leaf vdevs (e.g. raidz) to leaves. + * Used when initializing vdevs. Isn't used by leaf ops. + */ + vdev_xlation_func_t *vdev_op_xlate; char vdev_op_type[16]; boolean_t vdev_op_leaf; } vdev_ops_t; @@ -252,6 +263,24 @@ struct vdev { /* pool checkpoint related */ space_map_t *vdev_checkpoint_sm; /* contains reserved blocks */ + + boolean_t vdev_initialize_exit_wanted; + vdev_initializing_state_t vdev_initialize_state; + kthread_t *vdev_initialize_thread; + /* Protects vdev_initialize_thread and vdev_initialize_state. */ + kmutex_t vdev_initialize_lock; + kcondvar_t vdev_initialize_cv; + uint64_t vdev_initialize_offset[TXG_SIZE]; + uint64_t vdev_initialize_last_offset; + range_tree_t *vdev_initialize_tree; /* valid while initializing */ + uint64_t vdev_initialize_bytes_est; + uint64_t vdev_initialize_bytes_done; + time_t vdev_initialize_action_time; /* start and end time */ + + /* for limiting outstanding I/Os */ + kmutex_t vdev_initialize_io_lock; + kcondvar_t vdev_initialize_io_cv; + uint64_t vdev_initialize_inflight; /* * Values stored in the config for an indirect or removing vdev. @@ -471,6 +500,8 @@ extern vdev_ops_t vdev_indirect_ops; /* * Common size functions */ +extern void vdev_default_xlate(vdev_t *vd, const range_seg_t *in, + range_seg_t *out); extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize); extern uint64_t vdev_get_min_asize(vdev_t *vd); extern void vdev_set_min_asize(vdev_t *vd); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_initialize.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_initialize.h new file mode 100644 index 00000000000..db4b0572cd6 --- /dev/null +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_initialize.h @@ -0,0 +1,46 @@ +/* + * 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) 2016 by Delphix. All rights reserved. + */ + +#ifndef _SYS_VDEV_INITIALIZE_H +#define _SYS_VDEV_INITIALIZE_H + +#ifdef __cplusplus +extern "C" { +#endif + +extern void vdev_initialize(vdev_t *vd); +extern void vdev_initialize_stop(vdev_t *vd, + vdev_initializing_state_t tgt_state); +extern void vdev_initialize_stop_all(vdev_t *vd, + vdev_initializing_state_t tgt_state); +extern void vdev_initialize_restart(vdev_t *vd); +extern void vdev_xlate(vdev_t *vd, const range_seg_t *logical_rs, + range_seg_t *physical_rs); + +#ifdef __cplusplus +} +#endif + +#endif /* _SYS_VDEV_INITIALIZE_H */ diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio_priority.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio_priority.h index 4db05ac7759..ebe05a09dc4 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio_priority.h +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zio_priority.h @@ -13,7 +13,7 @@ * CDDL HEADER END */ /* - * Copyright (c) 2014 by Delphix. All rights reserved. + * Copyright (c) 2014, 2016 by Delphix. All rights reserved. */ #ifndef _ZIO_PRIORITY_H #define _ZIO_PRIORITY_H @@ -30,6 +30,7 @@ typedef enum zio_priority { ZIO_PRIORITY_SCRUB, /* asynchronous scrub/resilver reads */ ZIO_PRIORITY_TRIM, /* free requests used for TRIM */ ZIO_PRIORITY_REMOVAL, /* reads/writes for vdev removal */ + ZIO_PRIORITY_INITIALIZING, /* initializing I/O */ ZIO_PRIORITY_NUM_QUEUEABLE, ZIO_PRIORITY_NOW /* non-queued i/os (e.g. free) */ diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c index 80087a2fccc..a875bfff508 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev.c @@ -51,6 +51,7 @@ #include #include #include +#include SYSCTL_DECL(_vfs_zfs); SYSCTL_NODE(_vfs_zfs, OID_AUTO, vdev, CTLFLAG_RW, 0, "ZFS VDEV"); @@ -289,6 +290,14 @@ vdev_getops(const char *type) return (ops); } +/* ARGSUSED */ +void +vdev_default_xlate(vdev_t *vd, const range_seg_t *in, range_seg_t *res) +{ + res->rs_start = in->rs_start; + res->rs_end = in->rs_end; +} + /* * Default asize function: return the MAX of psize with the asize of * all children. This is what's used by anything other than RAID-Z. @@ -560,7 +569,11 @@ vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops) mutex_init(&vd->vdev_probe_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_queue_lock, NULL, MUTEX_DEFAULT, NULL); mutex_init(&vd->vdev_scan_io_queue_lock, NULL, MUTEX_DEFAULT, NULL); - + mutex_init(&vd->vdev_initialize_lock, NULL, MUTEX_DEFAULT, NULL); + mutex_init(&vd->vdev_initialize_io_lock, NULL, MUTEX_DEFAULT, NULL); + cv_init(&vd->vdev_initialize_cv, NULL, CV_DEFAULT, NULL); + cv_init(&vd->vdev_initialize_io_cv, NULL, CV_DEFAULT, NULL); + for (int t = 0; t < DTL_TYPES; t++) { vd->vdev_dtl[t] = range_tree_create(NULL, NULL); } @@ -833,6 +846,7 @@ void vdev_free(vdev_t *vd) { spa_t *spa = vd->vdev_spa; + ASSERT3P(vd->vdev_initialize_thread, ==, NULL); /* * Scan queues are normally destroyed at the end of a scan. If the @@ -863,6 +877,7 @@ vdev_free(vdev_t *vd) ASSERT(vd->vdev_child == NULL); ASSERT(vd->vdev_guid_sum == vd->vdev_guid); + ASSERT(vd->vdev_initialize_thread == NULL); /* * Discard allocation state. @@ -936,6 +951,10 @@ vdev_free(vdev_t *vd) mutex_destroy(&vd->vdev_stat_lock); mutex_destroy(&vd->vdev_probe_lock); mutex_destroy(&vd->vdev_scan_io_queue_lock); + mutex_destroy(&vd->vdev_initialize_lock); + mutex_destroy(&vd->vdev_initialize_io_lock); + cv_destroy(&vd->vdev_initialize_io_cv); + cv_destroy(&vd->vdev_initialize_cv); if (vd == spa->spa_root_vdev) spa->spa_root_vdev = NULL; @@ -3030,7 +3049,8 @@ vdev_sync_done(vdev_t *vd, uint64_t txg) ASSERT(vdev_is_concrete(vd)); - while (msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg))) + while ((msp = txg_list_remove(&vd->vdev_ms_list, TXG_CLEAN(txg))) + != NULL) metaslab_sync_done(msp, txg); if (reassess) @@ -3256,6 +3276,15 @@ vdev_online(spa_t *spa, uint64_t guid, uint64_t flags, vdev_state_t *newstate) spa_async_request(spa, SPA_ASYNC_CONFIG_UPDATE); } + /* Restart initializing if necessary */ + mutex_enter(&vd->vdev_initialize_lock); + if (vdev_writeable(vd) && + vd->vdev_initialize_thread == NULL && + vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE) { + (void) vdev_initialize(vd); + } + mutex_exit(&vd->vdev_initialize_lock); + if (wasoffline || (oldstate < VDEV_STATE_DEGRADED && vd->vdev_state >= VDEV_STATE_DEGRADED)) @@ -3558,8 +3587,18 @@ vdev_get_stats(vdev_t *vd, vdev_stat_t *vs) vs->vs_timestamp = gethrtime() - vs->vs_timestamp; vs->vs_state = vd->vdev_state; vs->vs_rsize = vdev_get_min_asize(vd); - if (vd->vdev_ops->vdev_op_leaf) + if (vd->vdev_ops->vdev_op_leaf) { vs->vs_rsize += VDEV_LABEL_START_SIZE + VDEV_LABEL_END_SIZE; + /* + * Report intializing progress. Since we don't have the + * initializing locks held, this is only an estimate (although a + * fairly accurate one). + */ + vs->vs_initialize_bytes_done = vd->vdev_initialize_bytes_done; + vs->vs_initialize_bytes_est = vd->vdev_initialize_bytes_est; + vs->vs_initialize_state = vd->vdev_initialize_state; + vs->vs_initialize_action_time = vd->vdev_initialize_action_time; + } /* * Report expandable space on top-level, non-auxillary devices only. * The expandable space is reported in terms of metaslab sized units diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_disk.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_disk.c index 26828e069d7..be24cde54b9 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_disk.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_disk.c @@ -841,6 +841,7 @@ vdev_ops_t vdev_disk_ops = { vdev_disk_hold, vdev_disk_rele, NULL, + vdev_default_xlate, VDEV_TYPE_DISK, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c index b5caee2ec79..c198d77e21d 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c @@ -20,7 +20,7 @@ */ /* * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. - * Copyright (c) 2011, 2015 by Delphix. All rights reserved. + * Copyright (c) 2011, 2016 by Delphix. All rights reserved. */ #include @@ -271,6 +271,7 @@ vdev_ops_t vdev_file_ops = { vdev_file_hold, vdev_file_rele, NULL, + vdev_default_xlate, VDEV_TYPE_FILE, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; @@ -291,6 +292,7 @@ vdev_ops_t vdev_disk_ops = { vdev_file_hold, vdev_file_rele, NULL, + vdev_default_xlate, VDEV_TYPE_DISK, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c index d4a3aa4b43f..7794bd50552 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c @@ -1154,6 +1154,7 @@ vdev_ops_t vdev_geom_ops = { vdev_geom_hold, vdev_geom_rele, NULL, + vdev_default_xlate, VDEV_TYPE_DISK, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect.c index f6452791c11..c4e4835447f 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect.c @@ -1631,6 +1631,7 @@ vdev_ops_t vdev_indirect_ops = { NULL, NULL, vdev_indirect_remap, + NULL, VDEV_TYPE_INDIRECT, /* name of this vdev type */ B_FALSE /* leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_initialize.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_initialize.c new file mode 100644 index 00000000000..a2c39c2868e --- /dev/null +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_initialize.c @@ -0,0 +1,792 @@ +/* + * 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) 2016 by Delphix. All rights reserved. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* + * Maximum number of metaslabs per group that can be initialized + * simultaneously. + */ +int max_initialize_ms = 3; + +/* + * Value that is written to disk during initialization. + */ +uint64_t zfs_initialize_value = 0xdeadbeefdeadbeefULL; + +/* maximum number of I/Os outstanding per leaf vdev */ +int zfs_initialize_limit = 1; + +/* size of initializing writes; default 1MiB, see zfs_remove_max_segment */ +uint64_t zfs_initialize_chunk_size = 1024 * 1024; + +static boolean_t +vdev_initialize_should_stop(vdev_t *vd) +{ + return (vd->vdev_initialize_exit_wanted || !vdev_writeable(vd) || + vd->vdev_detached || vd->vdev_top->vdev_removing); +} + +static void +vdev_initialize_zap_update_sync(void *arg, dmu_tx_t *tx) +{ + /* + * We pass in the guid instead of the vdev_t since the vdev may + * have been freed prior to the sync task being processed. This + * happens when a vdev is detached as we call spa_config_vdev_exit(), + * stop the intializing thread, schedule the sync task, and free + * the vdev. Later when the scheduled sync task is invoked, it would + * find that the vdev has been freed. + */ + uint64_t guid = *(uint64_t *)arg; + uint64_t txg = dmu_tx_get_txg(tx); + kmem_free(arg, sizeof (uint64_t)); + + vdev_t *vd = spa_lookup_by_guid(tx->tx_pool->dp_spa, guid, B_FALSE); + if (vd == NULL || vd->vdev_top->vdev_removing || !vdev_is_concrete(vd)) + return; + + uint64_t last_offset = vd->vdev_initialize_offset[txg & TXG_MASK]; + vd->vdev_initialize_offset[txg & TXG_MASK] = 0; + + VERIFY(vd->vdev_leaf_zap != 0); + + objset_t *mos = vd->vdev_spa->spa_meta_objset; + + if (last_offset > 0) { + vd->vdev_initialize_last_offset = last_offset; + VERIFY0(zap_update(mos, vd->vdev_leaf_zap, + VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET, + sizeof (last_offset), 1, &last_offset, tx)); + } + if (vd->vdev_initialize_action_time > 0) { + uint64_t val = (uint64_t)vd->vdev_initialize_action_time; + VERIFY0(zap_update(mos, vd->vdev_leaf_zap, + VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME, sizeof (val), + 1, &val, tx)); + } + + uint64_t initialize_state = vd->vdev_initialize_state; + VERIFY0(zap_update(mos, vd->vdev_leaf_zap, + VDEV_LEAF_ZAP_INITIALIZE_STATE, sizeof (initialize_state), 1, + &initialize_state, tx)); +} + +static void +vdev_initialize_change_state(vdev_t *vd, vdev_initializing_state_t new_state) +{ + ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock)); + spa_t *spa = vd->vdev_spa; + + if (new_state == vd->vdev_initialize_state) + return; + + /* + * Copy the vd's guid, this will be freed by the sync task. + */ + uint64_t *guid = kmem_zalloc(sizeof (uint64_t), KM_SLEEP); + *guid = vd->vdev_guid; + + /* + * If we're suspending, then preserving the original start time. + */ + if (vd->vdev_initialize_state != VDEV_INITIALIZE_SUSPENDED) { + vd->vdev_initialize_action_time = gethrestime_sec(); + } + vd->vdev_initialize_state = new_state; + + dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); + VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); + dsl_sync_task_nowait(spa_get_dsl(spa), vdev_initialize_zap_update_sync, + guid, 2, ZFS_SPACE_CHECK_RESERVED, tx); + + switch (new_state) { + case VDEV_INITIALIZE_ACTIVE: + spa_history_log_internal(spa, "initialize", tx, + "vdev=%s activated", vd->vdev_path); + break; + case VDEV_INITIALIZE_SUSPENDED: + spa_history_log_internal(spa, "initialize", tx, + "vdev=%s suspended", vd->vdev_path); + break; + case VDEV_INITIALIZE_CANCELED: + spa_history_log_internal(spa, "initialize", tx, + "vdev=%s canceled", vd->vdev_path); + break; + case VDEV_INITIALIZE_COMPLETE: + spa_history_log_internal(spa, "initialize", tx, + "vdev=%s complete", vd->vdev_path); + break; + default: + panic("invalid state %llu", (unsigned long long)new_state); + } + + dmu_tx_commit(tx); +} + +static void +vdev_initialize_cb(zio_t *zio) +{ + vdev_t *vd = zio->io_vd; + mutex_enter(&vd->vdev_initialize_io_lock); + if (zio->io_error == ENXIO && !vdev_writeable(vd)) { + /* + * The I/O failed because the vdev was unavailable; roll the + * last offset back. (This works because spa_sync waits on + * spa_txg_zio before it runs sync tasks.) + */ + uint64_t *off = + &vd->vdev_initialize_offset[zio->io_txg & TXG_MASK]; + *off = MIN(*off, zio->io_offset); + } else { + /* + * Since initializing is best-effort, we ignore I/O errors and + * rely on vdev_probe to determine if the errors are more + * critical. + */ + if (zio->io_error != 0) + vd->vdev_stat.vs_initialize_errors++; + + vd->vdev_initialize_bytes_done += zio->io_orig_size; + } + ASSERT3U(vd->vdev_initialize_inflight, >, 0); + vd->vdev_initialize_inflight--; + cv_broadcast(&vd->vdev_initialize_io_cv); + mutex_exit(&vd->vdev_initialize_io_lock); + + spa_config_exit(vd->vdev_spa, SCL_STATE_ALL, vd); +} + +/* Takes care of physical writing and limiting # of concurrent ZIOs. */ +static int +vdev_initialize_write(vdev_t *vd, uint64_t start, uint64_t size, abd_t *data) +{ + spa_t *spa = vd->vdev_spa; + + /* Limit inflight initializing I/Os */ + mutex_enter(&vd->vdev_initialize_io_lock); + while (vd->vdev_initialize_inflight >= zfs_initialize_limit) { + cv_wait(&vd->vdev_initialize_io_cv, + &vd->vdev_initialize_io_lock); + } + vd->vdev_initialize_inflight++; + mutex_exit(&vd->vdev_initialize_io_lock); + + dmu_tx_t *tx = dmu_tx_create_dd(spa_get_dsl(spa)->dp_mos_dir); + VERIFY0(dmu_tx_assign(tx, TXG_WAIT)); + uint64_t txg = dmu_tx_get_txg(tx); + + spa_config_enter(spa, SCL_STATE_ALL, vd, RW_READER); + mutex_enter(&vd->vdev_initialize_lock); + + if (vd->vdev_initialize_offset[txg & TXG_MASK] == 0) { + uint64_t *guid = kmem_zalloc(sizeof (uint64_t), KM_SLEEP); + *guid = vd->vdev_guid; + + /* This is the first write of this txg. */ + dsl_sync_task_nowait(spa_get_dsl(spa), + vdev_initialize_zap_update_sync, guid, 2, + ZFS_SPACE_CHECK_RESERVED, tx); + } + + /* + * We know the vdev struct will still be around since all + * consumers of vdev_free must stop the initialization first. + */ + if (vdev_initialize_should_stop(vd)) { + mutex_enter(&vd->vdev_initialize_io_lock); + ASSERT3U(vd->vdev_initialize_inflight, >, 0); + vd->vdev_initialize_inflight--; + mutex_exit(&vd->vdev_initialize_io_lock); + spa_config_exit(vd->vdev_spa, SCL_STATE_ALL, vd); + mutex_exit(&vd->vdev_initialize_lock); + dmu_tx_commit(tx); + return (SET_ERROR(EINTR)); + } + mutex_exit(&vd->vdev_initialize_lock); + + vd->vdev_initialize_offset[txg & TXG_MASK] = start + size; + zio_nowait(zio_write_phys(spa->spa_txg_zio[txg & TXG_MASK], vd, start, + size, data, ZIO_CHECKSUM_OFF, vdev_initialize_cb, NULL, + ZIO_PRIORITY_INITIALIZING, ZIO_FLAG_CANFAIL, B_FALSE)); + /* vdev_initialize_cb releases SCL_STATE_ALL */ + + dmu_tx_commit(tx); + + return (0); +} + +/* + * Translate a logical range to the physical range for the specified vdev_t. + * This function is initially called with a leaf vdev and will walk each + * parent vdev until it reaches a top-level vdev. Once the top-level is + * reached the physical range is initialized and the recursive function + * begins to unwind. As it unwinds it calls the parent's vdev specific + * translation function to do the real conversion. + */ +void +vdev_xlate(vdev_t *vd, const range_seg_t *logical_rs, range_seg_t *physical_rs) +{ + /* + * Walk up the vdev tree + */ + if (vd != vd->vdev_top) { + vdev_xlate(vd->vdev_parent, logical_rs, physical_rs); + } else { + /* + * We've reached the top-level vdev, initialize the + * physical range to the logical range and start to + * unwind. + */ + physical_rs->rs_start = logical_rs->rs_start; + physical_rs->rs_end = logical_rs->rs_end; + return; + } + + vdev_t *pvd = vd->vdev_parent; + ASSERT3P(pvd, !=, NULL); + ASSERT3P(pvd->vdev_ops->vdev_op_xlate, !=, NULL); + + /* + * As this recursive function unwinds, translate the logical + * range into its physical components by calling the + * vdev specific translate function. + */ + range_seg_t intermediate = { 0 }; + pvd->vdev_ops->vdev_op_xlate(vd, physical_rs, &intermediate); + + physical_rs->rs_start = intermediate.rs_start; + physical_rs->rs_end = intermediate.rs_end; +} + +/* + * Callback to fill each ABD chunk with zfs_initialize_value. len must be + * divisible by sizeof (uint64_t), and buf must be 8-byte aligned. The ABD + * allocation will guarantee these for us. + */ +/* ARGSUSED */ +static int +vdev_initialize_block_fill(void *buf, size_t len, void *unused) +{ + ASSERT0(len % sizeof (uint64_t)); + for (uint64_t i = 0; i < len; i += sizeof (uint64_t)) { + *(uint64_t *)((char *)(buf) + i) = zfs_initialize_value; + } + return (0); +} + +static abd_t * +vdev_initialize_block_alloc() +{ + /* Allocate ABD for filler data */ + abd_t *data = abd_alloc_for_io(zfs_initialize_chunk_size, B_FALSE); + + ASSERT0(zfs_initialize_chunk_size % sizeof (uint64_t)); + (void) abd_iterate_func(data, 0, zfs_initialize_chunk_size, + vdev_initialize_block_fill, NULL); + + return (data); +} + +static void +vdev_initialize_block_free(abd_t *data) +{ + abd_free(data); +} + +static int +vdev_initialize_ranges(vdev_t *vd, abd_t *data) +{ + avl_tree_t *rt = &vd->vdev_initialize_tree->rt_root; + + for (range_seg_t *rs = avl_first(rt); rs != NULL; + rs = AVL_NEXT(rt, rs)) { + uint64_t size = rs->rs_end - rs->rs_start; + + /* Split range into legally-sized physical chunks */ + uint64_t writes_required = + ((size - 1) / zfs_initialize_chunk_size) + 1; + + for (uint64_t w = 0; w < writes_required; w++) { + int error; + + error = vdev_initialize_write(vd, + VDEV_LABEL_START_SIZE + rs->rs_start + + (w * zfs_initialize_chunk_size), + MIN(size - (w * zfs_initialize_chunk_size), + zfs_initialize_chunk_size), data); + if (error != 0) + return (error); + } + } + return (0); +} + +static void +vdev_initialize_ms_load(metaslab_t *msp) +{ + ASSERT(MUTEX_HELD(&msp->ms_lock)); + + metaslab_load_wait(msp); + if (!msp->ms_loaded) + VERIFY0(metaslab_load(msp)); +} + +static void +vdev_initialize_mg_wait(metaslab_group_t *mg) +{ + ASSERT(MUTEX_HELD(&mg->mg_ms_initialize_lock)); + while (mg->mg_initialize_updating) { + cv_wait(&mg->mg_ms_initialize_cv, &mg->mg_ms_initialize_lock); + } +} + +static void +vdev_initialize_mg_mark(metaslab_group_t *mg) +{ + ASSERT(MUTEX_HELD(&mg->mg_ms_initialize_lock)); + ASSERT(mg->mg_initialize_updating); + + while (mg->mg_ms_initializing >= max_initialize_ms) { + cv_wait(&mg->mg_ms_initialize_cv, &mg->mg_ms_initialize_lock); + } + mg->mg_ms_initializing++; + ASSERT3U(mg->mg_ms_initializing, <=, max_initialize_ms); +} + +/* + * Mark the metaslab as being initialized to prevent any allocations + * on this metaslab. We must also track how many metaslabs are currently + * being initialized within a metaslab group and limit them to prevent + * allocation failures from occurring because all metaslabs are being + * initialized. + */ +static void +vdev_initialize_ms_mark(metaslab_t *msp) +{ + ASSERT(!MUTEX_HELD(&msp->ms_lock)); + metaslab_group_t *mg = msp->ms_group; + + mutex_enter(&mg->mg_ms_initialize_lock); + + /* + * To keep an accurate count of how many threads are initializing + * a specific metaslab group, we only allow one thread to mark + * the metaslab group at a time. This ensures that the value of + * ms_initializing will be accurate when we decide to mark a metaslab + * group as being initialized. To do this we force all other threads + * to wait till the metaslab's mg_initialize_updating flag is no + * longer set. + */ + vdev_initialize_mg_wait(mg); + mg->mg_initialize_updating = B_TRUE; + if (msp->ms_initializing == 0) { + vdev_initialize_mg_mark(mg); + } + mutex_enter(&msp->ms_lock); + msp->ms_initializing++; + mutex_exit(&msp->ms_lock); + + mg->mg_initialize_updating = B_FALSE; + cv_broadcast(&mg->mg_ms_initialize_cv); + mutex_exit(&mg->mg_ms_initialize_lock); +} + +static void +vdev_initialize_ms_unmark(metaslab_t *msp) +{ + ASSERT(!MUTEX_HELD(&msp->ms_lock)); + metaslab_group_t *mg = msp->ms_group; + mutex_enter(&mg->mg_ms_initialize_lock); + mutex_enter(&msp->ms_lock); + if (--msp->ms_initializing == 0) { + mg->mg_ms_initializing--; + cv_broadcast(&mg->mg_ms_initialize_cv); + } + mutex_exit(&msp->ms_lock); + mutex_exit(&mg->mg_ms_initialize_lock); +} + +static void +vdev_initialize_calculate_progress(vdev_t *vd) +{ + ASSERT(spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_READER) || + spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_WRITER)); + ASSERT(vd->vdev_leaf_zap != 0); + + vd->vdev_initialize_bytes_est = 0; + vd->vdev_initialize_bytes_done = 0; + + for (uint64_t i = 0; i < vd->vdev_top->vdev_ms_count; i++) { + metaslab_t *msp = vd->vdev_top->vdev_ms[i]; + mutex_enter(&msp->ms_lock); + + uint64_t ms_free = msp->ms_size - + space_map_allocated(msp->ms_sm); + + if (vd->vdev_top->vdev_ops == &vdev_raidz_ops) + ms_free /= vd->vdev_top->vdev_children; + + /* + * Convert the metaslab range to a physical range + * on our vdev. We use this to determine if we are + * in the middle of this metaslab range. + */ + range_seg_t logical_rs, physical_rs; + logical_rs.rs_start = msp->ms_start; + logical_rs.rs_end = msp->ms_start + msp->ms_size; + vdev_xlate(vd, &logical_rs, &physical_rs); + + if (vd->vdev_initialize_last_offset <= physical_rs.rs_start) { + vd->vdev_initialize_bytes_est += ms_free; + mutex_exit(&msp->ms_lock); + continue; + } else if (vd->vdev_initialize_last_offset > + physical_rs.rs_end) { + vd->vdev_initialize_bytes_done += ms_free; + vd->vdev_initialize_bytes_est += ms_free; + mutex_exit(&msp->ms_lock); + continue; + } + + /* + * If we get here, we're in the middle of initializing this + * metaslab. Load it and walk the free tree for more accurate + * progress estimation. + */ + vdev_initialize_ms_load(msp); + + for (range_seg_t *rs = avl_first(&msp->ms_allocatable->rt_root); rs; + rs = AVL_NEXT(&msp->ms_allocatable->rt_root, rs)) { + logical_rs.rs_start = rs->rs_start; + logical_rs.rs_end = rs->rs_end; + vdev_xlate(vd, &logical_rs, &physical_rs); + + uint64_t size = physical_rs.rs_end - + physical_rs.rs_start; + vd->vdev_initialize_bytes_est += size; + if (vd->vdev_initialize_last_offset > + physical_rs.rs_end) { + vd->vdev_initialize_bytes_done += size; + } else if (vd->vdev_initialize_last_offset > + physical_rs.rs_start && + vd->vdev_initialize_last_offset < + physical_rs.rs_end) { + vd->vdev_initialize_bytes_done += + vd->vdev_initialize_last_offset - + physical_rs.rs_start; + } + } + mutex_exit(&msp->ms_lock); + } +} + +static void +vdev_initialize_load(vdev_t *vd) +{ + ASSERT(spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_READER) || + spa_config_held(vd->vdev_spa, SCL_CONFIG, RW_WRITER)); + ASSERT(vd->vdev_leaf_zap != 0); + + if (vd->vdev_initialize_state == VDEV_INITIALIZE_ACTIVE || + vd->vdev_initialize_state == VDEV_INITIALIZE_SUSPENDED) { + int err = zap_lookup(vd->vdev_spa->spa_meta_objset, + vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET, + sizeof (vd->vdev_initialize_last_offset), 1, + &vd->vdev_initialize_last_offset); + ASSERT(err == 0 || err == ENOENT); + } + + vdev_initialize_calculate_progress(vd); +} + + +/* + * Convert the logical range into a physcial range and add it to our + * avl tree. + */ +void +vdev_initialize_range_add(void *arg, uint64_t start, uint64_t size) +{ + vdev_t *vd = arg; + range_seg_t logical_rs, physical_rs; + logical_rs.rs_start = start; + logical_rs.rs_end = start + size; + + ASSERT(vd->vdev_ops->vdev_op_leaf); + vdev_xlate(vd, &logical_rs, &physical_rs); + + IMPLY(vd->vdev_top == vd, + logical_rs.rs_start == physical_rs.rs_start); + IMPLY(vd->vdev_top == vd, + logical_rs.rs_end == physical_rs.rs_end); + + /* Only add segments that we have not visited yet */ + if (physical_rs.rs_end <= vd->vdev_initialize_last_offset) + return; + + /* Pick up where we left off mid-range. */ + if (vd->vdev_initialize_last_offset > physical_rs.rs_start) { + zfs_dbgmsg("range write: vd %s changed (%llu, %llu) to " + "(%llu, %llu)", vd->vdev_path, + (u_longlong_t)physical_rs.rs_start, + (u_longlong_t)physical_rs.rs_end, + (u_longlong_t)vd->vdev_initialize_last_offset, + (u_longlong_t)physical_rs.rs_end); + ASSERT3U(physical_rs.rs_end, >, + vd->vdev_initialize_last_offset); + physical_rs.rs_start = vd->vdev_initialize_last_offset; + } + ASSERT3U(physical_rs.rs_end, >=, physical_rs.rs_start); + + /* + * With raidz, it's possible that the logical range does not live on + * this leaf vdev. We only add the physical range to this vdev's if it + * has a length greater than 0. + */ + if (physical_rs.rs_end > physical_rs.rs_start) { + range_tree_add(vd->vdev_initialize_tree, physical_rs.rs_start, + physical_rs.rs_end - physical_rs.rs_start); + } else { + ASSERT3U(physical_rs.rs_end, ==, physical_rs.rs_start); + } +} + +static void +vdev_initialize_thread(void *arg) +{ + vdev_t *vd = arg; + spa_t *spa = vd->vdev_spa; + int error = 0; + uint64_t ms_count = 0; + + ASSERT(vdev_is_concrete(vd)); + spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); + + vd->vdev_initialize_last_offset = 0; + vdev_initialize_load(vd); + + abd_t *deadbeef = vdev_initialize_block_alloc(); + + vd->vdev_initialize_tree = range_tree_create(NULL, NULL); + + for (uint64_t i = 0; !vd->vdev_detached && + i < vd->vdev_top->vdev_ms_count; i++) { + metaslab_t *msp = vd->vdev_top->vdev_ms[i]; + + /* + * If we've expanded the top-level vdev or it's our + * first pass, calculate our progress. + */ + if (vd->vdev_top->vdev_ms_count != ms_count) { + vdev_initialize_calculate_progress(vd); + ms_count = vd->vdev_top->vdev_ms_count; + } + + vdev_initialize_ms_mark(msp); + mutex_enter(&msp->ms_lock); + vdev_initialize_ms_load(msp); + + range_tree_walk(msp->ms_allocatable, vdev_initialize_range_add, + vd); + mutex_exit(&msp->ms_lock); + + spa_config_exit(spa, SCL_CONFIG, FTAG); + error = vdev_initialize_ranges(vd, deadbeef); + vdev_initialize_ms_unmark(msp); + spa_config_enter(spa, SCL_CONFIG, FTAG, RW_READER); + + range_tree_vacate(vd->vdev_initialize_tree, NULL, NULL); + if (error != 0) + break; + } + + spa_config_exit(spa, SCL_CONFIG, FTAG); + mutex_enter(&vd->vdev_initialize_io_lock); + while (vd->vdev_initialize_inflight > 0) { + cv_wait(&vd->vdev_initialize_io_cv, + &vd->vdev_initialize_io_lock); + } + mutex_exit(&vd->vdev_initialize_io_lock); + + range_tree_destroy(vd->vdev_initialize_tree); + vdev_initialize_block_free(deadbeef); + vd->vdev_initialize_tree = NULL; + + mutex_enter(&vd->vdev_initialize_lock); + if (!vd->vdev_initialize_exit_wanted && vdev_writeable(vd)) { + vdev_initialize_change_state(vd, VDEV_INITIALIZE_COMPLETE); + } + ASSERT(vd->vdev_initialize_thread != NULL || + vd->vdev_initialize_inflight == 0); + + /* + * Drop the vdev_initialize_lock while we sync out the + * txg since it's possible that a device might be trying to + * come online and must check to see if it needs to restart an + * initialization. That thread will be holding the spa_config_lock + * which would prevent the txg_wait_synced from completing. + */ + mutex_exit(&vd->vdev_initialize_lock); + txg_wait_synced(spa_get_dsl(spa), 0); + mutex_enter(&vd->vdev_initialize_lock); + + vd->vdev_initialize_thread = NULL; + cv_broadcast(&vd->vdev_initialize_cv); + mutex_exit(&vd->vdev_initialize_lock); + thread_exit(); +} + +/* + * Initiates a device. Caller must hold vdev_initialize_lock. + * Device must be a leaf and not already be initializing. + */ +void +vdev_initialize(vdev_t *vd) +{ + ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock)); + ASSERT(vd->vdev_ops->vdev_op_leaf); + ASSERT(vdev_is_concrete(vd)); + ASSERT3P(vd->vdev_initialize_thread, ==, NULL); + ASSERT(!vd->vdev_detached); + ASSERT(!vd->vdev_initialize_exit_wanted); + ASSERT(!vd->vdev_top->vdev_removing); + + vdev_initialize_change_state(vd, VDEV_INITIALIZE_ACTIVE); + vd->vdev_initialize_thread = thread_create(NULL, 0, + vdev_initialize_thread, vd, 0, &p0, TS_RUN, maxclsyspri); +} + +/* + * Stop initializng a device, with the resultant initialing state being + * tgt_state. Blocks until the initializing thread has exited. + * Caller must hold vdev_initialize_lock and must not be writing to the spa + * config, as the initializing thread may try to enter the config as a reader + * before exiting. + */ +void +vdev_initialize_stop(vdev_t *vd, vdev_initializing_state_t tgt_state) +{ + spa_t *spa = vd->vdev_spa; + ASSERT(!spa_config_held(spa, SCL_CONFIG | SCL_STATE, RW_WRITER)); + + ASSERT(MUTEX_HELD(&vd->vdev_initialize_lock)); + ASSERT(vd->vdev_ops->vdev_op_leaf); + ASSERT(vdev_is_concrete(vd)); + + /* + * Allow cancel requests to proceed even if the initialize thread + * has stopped. + */ + if (vd->vdev_initialize_thread == NULL && + tgt_state != VDEV_INITIALIZE_CANCELED) { + return; + } + + vdev_initialize_change_state(vd, tgt_state); + vd->vdev_initialize_exit_wanted = B_TRUE; + while (vd->vdev_initialize_thread != NULL) + cv_wait(&vd->vdev_initialize_cv, &vd->vdev_initialize_lock); + + ASSERT3P(vd->vdev_initialize_thread, ==, NULL); + vd->vdev_initialize_exit_wanted = B_FALSE; +} + +static void +vdev_initialize_stop_all_impl(vdev_t *vd, vdev_initializing_state_t tgt_state) +{ + if (vd->vdev_ops->vdev_op_leaf && vdev_is_concrete(vd)) { + mutex_enter(&vd->vdev_initialize_lock); + vdev_initialize_stop(vd, tgt_state); + mutex_exit(&vd->vdev_initialize_lock); + return; + } + + for (uint64_t i = 0; i < vd->vdev_children; i++) { + vdev_initialize_stop_all_impl(vd->vdev_child[i], tgt_state); + } +} + +/* + * Convenience function to stop initializing of a vdev tree and set all + * initialize thread pointers to NULL. + */ +void +vdev_initialize_stop_all(vdev_t *vd, vdev_initializing_state_t tgt_state) +{ + vdev_initialize_stop_all_impl(vd, tgt_state); + + if (vd->vdev_spa->spa_sync_on) { + /* Make sure that our state has been synced to disk */ + txg_wait_synced(spa_get_dsl(vd->vdev_spa), 0); + } +} + +void +vdev_initialize_restart(vdev_t *vd) +{ + ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(!spa_config_held(vd->vdev_spa, SCL_ALL, RW_WRITER)); + + if (vd->vdev_leaf_zap != 0) { + mutex_enter(&vd->vdev_initialize_lock); + uint64_t initialize_state = VDEV_INITIALIZE_NONE; + int err = zap_lookup(vd->vdev_spa->spa_meta_objset, + vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_STATE, + sizeof (initialize_state), 1, &initialize_state); + ASSERT(err == 0 || err == ENOENT); + vd->vdev_initialize_state = initialize_state; + + uint64_t timestamp = 0; + err = zap_lookup(vd->vdev_spa->spa_meta_objset, + vd->vdev_leaf_zap, VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME, + sizeof (timestamp), 1, ×tamp); + ASSERT(err == 0 || err == ENOENT); + vd->vdev_initialize_action_time = (time_t)timestamp; + + if (vd->vdev_initialize_state == VDEV_INITIALIZE_SUSPENDED || + vd->vdev_offline) { + /* load progress for reporting, but don't resume */ + vdev_initialize_load(vd); + } else if (vd->vdev_initialize_state == + VDEV_INITIALIZE_ACTIVE && vdev_writeable(vd)) { + vdev_initialize(vd); + } + + mutex_exit(&vd->vdev_initialize_lock); + } + + for (uint64_t i = 0; i < vd->vdev_children; i++) { + vdev_initialize_restart(vd->vdev_child[i]); + } +} diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c index e23acf73fbc..26be35fc350 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c @@ -740,6 +740,7 @@ vdev_ops_t vdev_mirror_ops = { NULL, NULL, NULL, + vdev_default_xlate, VDEV_TYPE_MIRROR, /* name of this vdev type */ B_FALSE /* not a leaf vdev */ }; @@ -755,6 +756,7 @@ vdev_ops_t vdev_replacing_ops = { NULL, NULL, NULL, + vdev_default_xlate, VDEV_TYPE_REPLACING, /* name of this vdev type */ B_FALSE /* not a leaf vdev */ }; @@ -770,6 +772,7 @@ vdev_ops_t vdev_spare_ops = { NULL, NULL, NULL, + vdev_default_xlate, VDEV_TYPE_SPARE, /* name of this vdev type */ B_FALSE /* not a leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_missing.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_missing.c index 29194fc1106..6852de44504 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_missing.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_missing.c @@ -24,7 +24,7 @@ */ /* - * Copyright (c) 2012, 2014 by Delphix. All rights reserved. + * Copyright (c) 2012, 2016 by Delphix. All rights reserved. */ /* @@ -91,6 +91,7 @@ vdev_ops_t vdev_missing_ops = { NULL, NULL, NULL, + NULL, VDEV_TYPE_MISSING, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; @@ -106,6 +107,7 @@ vdev_ops_t vdev_hole_ops = { NULL, NULL, NULL, + NULL, VDEV_TYPE_HOLE, /* name of this vdev type */ B_TRUE /* leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c index 02f7232a2b7..0d60dfa1ac6 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_queue.c @@ -157,6 +157,8 @@ uint32_t zfs_vdev_trim_min_active = 1; uint32_t zfs_vdev_trim_max_active = 64; uint32_t zfs_vdev_removal_min_active = 1; uint32_t zfs_vdev_removal_max_active = 2; +uint32_t zfs_vdev_initializing_min_active = 1; +uint32_t zfs_vdev_initializing_max_active = 1; /* @@ -542,6 +544,8 @@ vdev_queue_class_min_active(zio_priority_t p) return (zfs_vdev_trim_min_active); case ZIO_PRIORITY_REMOVAL: return (zfs_vdev_removal_min_active); + case ZIO_PRIORITY_INITIALIZING: + return (zfs_vdev_initializing_min_active); default: panic("invalid priority %u", p); return (0); @@ -605,6 +609,8 @@ vdev_queue_class_max_active(spa_t *spa, zio_priority_t p) return (zfs_vdev_trim_max_active); case ZIO_PRIORITY_REMOVAL: return (zfs_vdev_removal_max_active); + case ZIO_PRIORITY_INITIALIZING: + return (zfs_vdev_initializing_max_active); default: panic("invalid priority %u", p); return (0); @@ -832,8 +838,8 @@ vdev_queue_io_to_issue(vdev_queue_t *vq) } /* - * For LBA-ordered queues (async / scrub), issue the i/o which follows - * the most recently issued i/o in LBA (offset) order. + * For LBA-ordered queues (async / scrub / initializing), issue the + * i/o which follows the most recently issued i/o in LBA (offset) order. * * For FIFO queues (sync), issue the i/o with the lowest timestamp. */ @@ -889,12 +895,14 @@ vdev_queue_io(zio_t *zio) if (zio->io_priority != ZIO_PRIORITY_SYNC_READ && zio->io_priority != ZIO_PRIORITY_ASYNC_READ && zio->io_priority != ZIO_PRIORITY_SCRUB && - zio->io_priority != ZIO_PRIORITY_REMOVAL) + zio->io_priority != ZIO_PRIORITY_REMOVAL && + zio->io_priority != ZIO_PRIORITY_INITIALIZING) zio->io_priority = ZIO_PRIORITY_ASYNC_READ; } else if (zio->io_type == ZIO_TYPE_WRITE) { if (zio->io_priority != ZIO_PRIORITY_SYNC_WRITE && zio->io_priority != ZIO_PRIORITY_ASYNC_WRITE && - zio->io_priority != ZIO_PRIORITY_REMOVAL) + zio->io_priority != ZIO_PRIORITY_REMOVAL && + zio->io_priority != ZIO_PRIORITY_INITIALIZING) zio->io_priority = ZIO_PRIORITY_ASYNC_WRITE; } else { ASSERT(zio->io_type == ZIO_TYPE_FREE); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_raidz.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_raidz.c index d6f4bbc4156..4df04c30aab 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_raidz.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_raidz.c @@ -41,6 +41,10 @@ #include #include +#ifdef ZFS_DEBUG +#include /* vdev_xlate testing */ +#endif + /* * Virtual device vector for RAID-Z. * @@ -1896,6 +1900,39 @@ vdev_raidz_child_done(zio_t *zio) rc->rc_skipped = 0; } +static void +vdev_raidz_io_verify(zio_t *zio, raidz_map_t *rm, int col) +{ +#ifdef ZFS_DEBUG + vdev_t *vd = zio->io_vd; + vdev_t *tvd = vd->vdev_top; + + range_seg_t logical_rs, physical_rs; + logical_rs.rs_start = zio->io_offset; + logical_rs.rs_end = logical_rs.rs_start + + vdev_raidz_asize(zio->io_vd, zio->io_size); + + raidz_col_t *rc = &rm->rm_col[col]; + vdev_t *cvd = vd->vdev_child[rc->rc_devidx]; + + vdev_xlate(cvd, &logical_rs, &physical_rs); + ASSERT3U(rc->rc_offset, ==, physical_rs.rs_start); + ASSERT3U(rc->rc_offset, <, physical_rs.rs_end); + /* + * It would be nice to assert that rs_end is equal + * to rc_offset + rc_size but there might be an + * optional I/O at the end that is not accounted in + * rc_size. + */ + if (physical_rs.rs_end > rc->rc_offset + rc->rc_size) { + ASSERT3U(physical_rs.rs_end, ==, rc->rc_offset + + rc->rc_size + (1 << tvd->vdev_ashift)); + } else { + ASSERT3U(physical_rs.rs_end, ==, rc->rc_offset + rc->rc_size); + } +#endif +} + /* * Start an IO operation on a RAIDZ VDev * @@ -1953,6 +1990,12 @@ vdev_raidz_io_start(zio_t *zio) for (c = 0; c < rm->rm_cols; c++) { rc = &rm->rm_col[c]; cvd = vd->vdev_child[rc->rc_devidx]; + + /* + * Verify physical to logical translation. + */ + vdev_raidz_io_verify(zio, rm, c); + zio_nowait(zio_vdev_child_io(zio, NULL, cvd, rc->rc_offset, rc->rc_abd, rc->rc_size, zio->io_type, zio->io_priority, 0, @@ -2622,6 +2665,37 @@ vdev_raidz_need_resilver(vdev_t *vd, uint64_t offset, size_t psize) return (B_FALSE); } +static void +vdev_raidz_xlate(vdev_t *cvd, const range_seg_t *in, range_seg_t *res) +{ + vdev_t *raidvd = cvd->vdev_parent; + ASSERT(raidvd->vdev_ops == &vdev_raidz_ops); + + uint64_t width = raidvd->vdev_children; + uint64_t tgt_col = cvd->vdev_id; + uint64_t ashift = raidvd->vdev_top->vdev_ashift; + + /* make sure the offsets are block-aligned */ + ASSERT0(in->rs_start % (1 << ashift)); + ASSERT0(in->rs_end % (1 << ashift)); + uint64_t b_start = in->rs_start >> ashift; + uint64_t b_end = in->rs_end >> ashift; + + uint64_t start_row = 0; + if (b_start > tgt_col) /* avoid underflow */ + start_row = ((b_start - tgt_col - 1) / width) + 1; + + uint64_t end_row = 0; + if (b_end > tgt_col) + end_row = ((b_end - tgt_col - 1) / width) + 1; + + res->rs_start = start_row << ashift; + res->rs_end = end_row << ashift; + + ASSERT3U(res->rs_start, <=, in->rs_start); + ASSERT3U(res->rs_end - res->rs_start, <=, in->rs_end - in->rs_start); +} + vdev_ops_t vdev_raidz_ops = { vdev_raidz_open, vdev_raidz_close, @@ -2633,6 +2707,7 @@ vdev_ops_t vdev_raidz_ops = { NULL, NULL, NULL, + vdev_raidz_xlate, VDEV_TYPE_RAIDZ, /* name of this vdev type */ B_FALSE /* not a leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_removal.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_removal.c index 44391755b1e..d19285beedb 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_removal.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_removal.c @@ -44,6 +44,7 @@ #include #include #include +#include /* * This file contains the necessary logic to remove vdevs from a @@ -1022,6 +1023,7 @@ vdev_remove_complete(spa_t *spa) txg_wait_synced(spa->spa_dsl_pool, 0); txg = spa_vdev_enter(spa); vdev_t *vd = vdev_lookup_top(spa, spa->spa_vdev_removal->svr_vdev_id); + ASSERT3P(vd->vdev_initialize_thread, ==, NULL); sysevent_t *ev = spa_event_create(spa, vd, NULL, ESC_ZFS_VDEV_REMOVE_DEV); @@ -1661,6 +1663,9 @@ spa_vdev_remove_log(vdev_t *vd, uint64_t *txg) /* Make sure these changes are sync'ed */ spa_vdev_config_exit(spa, NULL, *txg, 0, FTAG); + /* Stop initializing */ + (void) vdev_initialize_stop_all(vd, VDEV_INITIALIZE_CANCELED); + *txg = spa_vdev_config_enter(spa); sysevent_t *ev = spa_event_create(spa, vd, NULL, @@ -1821,6 +1826,13 @@ spa_vdev_remove_top(vdev_t *vd, uint64_t *txg) */ error = spa_reset_logs(spa); + /* + * We stop any initializing that is currently in progress but leave + * the state as "active". This will allow the initializing to resume + * if the removal is canceled sometime later. + */ + vdev_initialize_stop_all(vd, VDEV_INITIALIZE_ACTIVE); + *txg = spa_vdev_config_enter(spa); /* @@ -1832,6 +1844,7 @@ spa_vdev_remove_top(vdev_t *vd, uint64_t *txg) if (error != 0) { metaslab_group_activate(mg); + spa_async_request(spa, SPA_ASYNC_INITIALIZE_RESTART); return (error); } diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_root.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_root.c index 92c670d28b2..a03d18704df 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_root.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_root.c @@ -24,7 +24,7 @@ */ /* - * Copyright (c) 2012, 2014 by Delphix. All rights reserved. + * Copyright (c) 2012, 2016 by Delphix. All rights reserved. */ #include @@ -151,6 +151,7 @@ vdev_ops_t vdev_root_ops = { NULL, NULL, NULL, + NULL, VDEV_TYPE_ROOT, /* name of this vdev type */ B_FALSE /* not a leaf vdev */ }; diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c index e7819e9c9c1..d65acdc42c6 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c @@ -196,6 +196,8 @@ #include #include #include +#include +#include #include "zfs_namecheck.h" #include "zfs_prop.h" @@ -3865,6 +3867,80 @@ zfs_ioc_destroy(zfs_cmd_t *zc) return (err); } +/* + * innvl: { + * vdevs: { + * guid 1, guid 2, ... + * }, + * func: POOL_INITIALIZE_{CANCEL|DO|SUSPEND} + * } + * + * outnvl: { + * [func: EINVAL (if provided command type didn't make sense)], + * [vdevs: { + * guid1: errno, (see function body for possible errnos) + * ... + * }] + * } + * + */ +static int +zfs_ioc_pool_initialize(const char *poolname, nvlist_t *innvl, nvlist_t *outnvl) +{ + spa_t *spa; + int error; + + error = spa_open(poolname, &spa, FTAG); + if (error != 0) + return (error); + + uint64_t cmd_type; + if (nvlist_lookup_uint64(innvl, ZPOOL_INITIALIZE_COMMAND, + &cmd_type) != 0) { + spa_close(spa, FTAG); + return (SET_ERROR(EINVAL)); + } + if (!(cmd_type == POOL_INITIALIZE_CANCEL || + cmd_type == POOL_INITIALIZE_DO || + cmd_type == POOL_INITIALIZE_SUSPEND)) { + spa_close(spa, FTAG); + return (SET_ERROR(EINVAL)); + } + + nvlist_t *vdev_guids; + if (nvlist_lookup_nvlist(innvl, ZPOOL_INITIALIZE_VDEVS, + &vdev_guids) != 0) { + spa_close(spa, FTAG); + return (SET_ERROR(EINVAL)); + } + + nvlist_t *vdev_errlist = fnvlist_alloc(); + int total_errors = 0; + + for (nvpair_t *pair = nvlist_next_nvpair(vdev_guids, NULL); + pair != NULL; pair = nvlist_next_nvpair(vdev_guids, pair)) { + uint64_t vdev_guid = fnvpair_value_uint64(pair); + + error = spa_vdev_initialize(spa, vdev_guid, cmd_type); + if (error != 0) { + char guid_as_str[MAXNAMELEN]; + + (void) snprintf(guid_as_str, sizeof (guid_as_str), + "%llu", (unsigned long long)vdev_guid); + fnvlist_add_int64(vdev_errlist, guid_as_str, error); + total_errors++; + } + } + if (fnvlist_size(vdev_errlist) > 0) { + fnvlist_add_nvlist(outnvl, ZPOOL_INITIALIZE_VDEVS, + vdev_errlist); + } + fnvlist_free(vdev_errlist); + + spa_close(spa, FTAG); + return (total_errors > 0 ? EINVAL : 0); +} + /* * fsname is name of dataset to rollback (to most recent snapshot) * @@ -6125,6 +6201,10 @@ zfs_ioctl_init(void) zfs_secpolicy_config, POOL_NAME, POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + zfs_ioctl_register("initialize", ZFS_IOC_POOL_INITIALIZE, + zfs_ioc_pool_initialize, zfs_secpolicy_config, POOL_NAME, + POOL_CHECK_SUSPENDED | POOL_CHECK_READONLY, B_TRUE, B_TRUE); + /* IOCTLS that use the legacy function signature */ zfs_ioctl_register_legacy(ZFS_IOC_POOL_FREEZE, zfs_ioc_pool_freeze, diff --git a/sys/cddl/contrib/opensolaris/uts/common/sys/fs/zfs.h b/sys/cddl/contrib/opensolaris/uts/common/sys/fs/zfs.h index c347b63a1a6..d1f62aaa042 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/sys/fs/zfs.h +++ b/sys/cddl/contrib/opensolaris/uts/common/sys/fs/zfs.h @@ -637,6 +637,13 @@ typedef struct zpool_load_policy { #define VDEV_TOP_ZAP_POOL_CHECKPOINT_SM \ "com.delphix:pool_checkpoint_sm" +#define VDEV_LEAF_ZAP_INITIALIZE_LAST_OFFSET \ + "com.delphix:next_offset_to_initialize" +#define VDEV_LEAF_ZAP_INITIALIZE_STATE \ + "com.delphix:vdev_initialize_state" +#define VDEV_LEAF_ZAP_INITIALIZE_ACTION_TIME \ + "com.delphix:vdev_initialize_action_time" + /* * This is needed in userland to report the minimum necessary device size. * @@ -735,6 +742,15 @@ typedef enum pool_scrub_cmd { POOL_SCRUB_FLAGS_END } pool_scrub_cmd_t; +/* + * Initialize functions. + */ +typedef enum pool_initialize_func { + POOL_INITIALIZE_DO, + POOL_INITIALIZE_CANCEL, + POOL_INITIALIZE_SUSPEND, + POOL_INITIALIZE_FUNCS +} pool_initialize_func_t; /* * ZIO types. Needed to interpret vdev statistics below. @@ -814,6 +830,14 @@ typedef struct pool_checkpoint_stat { uint64_t pcs_space; /* checkpointed space */ } pool_checkpoint_stat_t; +typedef enum { + VDEV_INITIALIZE_NONE, + VDEV_INITIALIZE_ACTIVE, + VDEV_INITIALIZE_CANCELED, + VDEV_INITIALIZE_SUSPENDED, + VDEV_INITIALIZE_COMPLETE +} vdev_initializing_state_t; + /* * Vdev statistics. Note: all fields should be 64-bit because this * is passed between kernel and userland as an nvlist uint64 array. @@ -840,6 +864,11 @@ typedef struct vdev_stat { uint64_t vs_physical_ashift; /* vdev_physical_ashift */ uint64_t vs_fragmentation; /* device fragmentation */ uint64_t vs_checkpoint_space; /* checkpoint-consumed space */ + uint64_t vs_initialize_errors; /* initializing errors */ + uint64_t vs_initialize_bytes_done; /* bytes initialized */ + uint64_t vs_initialize_bytes_est; /* total bytes to initialize */ + uint64_t vs_initialize_state; /* vdev_initialzing_state_t */ + uint64_t vs_initialize_action_time; /* time_t */ } vdev_stat_t; #define VDEV_STAT_VALID(field, uint64_t_field_count) \ ((uint64_t_field_count * sizeof(uint64_t)) >= \ @@ -974,6 +1003,7 @@ typedef enum zfs_ioc { ZFS_IOC_REMAP, ZFS_IOC_POOL_CHECKPOINT, ZFS_IOC_POOL_DISCARD_CHECKPOINT, + ZFS_IOC_POOL_INITIALIZE, ZFS_IOC_LAST } zfs_ioc_t; @@ -1036,6 +1066,12 @@ typedef enum { #define ZPOOL_HIST_DSID "dsid" #define ZPOOL_HIST_ERRNO "errno" +/* + * The following are names used when invoking ZFS_IOC_POOL_INITIALIZE. + */ +#define ZPOOL_INITIALIZE_COMMAND "initialize_command" +#define ZPOOL_INITIALIZE_VDEVS "initialize_vdevs" + /* * Flags for ZFS_IOC_VDEV_SET_STATE */ diff --git a/sys/conf/files b/sys/conf/files index 3bc45233829..0a1f854c878 100644 --- a/sys/conf/files +++ b/sys/conf/files @@ -222,6 +222,7 @@ cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c optional zfs compile-wi cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect_births.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_indirect_mapping.c optional zfs compile-with "${ZFS_C}" +cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_initialize.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_geom.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_label.c optional zfs compile-with "${ZFS_C}" cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_mirror.c optional zfs compile-with "${ZFS_C}" -- 2.45.0