Merge commit 'refs/top-bases/linux-zfs-branch' into linux-zfs-branch

cmd/zdb/zdb.c

@@ -55,7 +55,6 @@
 #include <sys/arc.h>
 #include <sys/ddt.h>
 #undef ZFS_MAXNAMELEN
-#undef verify
 #include <libzfs.h>
 
 #define	ZDB_COMPRESS_NAME(idx) ((idx) < ZIO_COMPRESS_FUNCTIONS ? \

cmd/zfs/zfs_main.c

@@ -311,6 +311,7 @@ safe_malloc(size_t size)
 	return (data);
 }
 
+#ifdef HAVE_ZPL
 static char *
 safe_strdup(char *str)
 {
@@ -321,6 +322,7 @@ safe_strdup(char *str)
 
 	return (dupstr);
 }
+#endif /* HAVE_ZPL */
 
 /*
  * Callback routine that will print out information for each of
@@ -488,6 +490,7 @@ parse_depth(char *opt, int *flags)
 
 #define	PROGRESS_DELAY 2		/* seconds */
 
+#ifdef HAVE_ZPL
 static char *pt_reverse = "\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b";
 static time_t pt_begin;
 static char *pt_header = NULL;
@@ -539,6 +542,8 @@ finish_progress(char *done)
 	free(pt_header);
 	pt_header = NULL;
 }
+#endif /* HAVE_ZPL */
+
 /*
  * zfs clone [-p] [-o prop=value] ... <snap> <fs | vol>
  *

cmd/zpool/zpool_main.c

@@ -1598,7 +1598,7 @@ do_import(nvlist_t *config, const char *newname, const char *mntopts,
  *	 -c	Read pool information from a cachefile instead of searching
  *		devices.
  *
- *       -d	Scan in a specific directory, other than /dev/dsk.  More than
+ *       -d	Scan in a specific directory, other than /dev/.  More than
  *		one directory can be specified using multiple '-d' options.
  *
  *       -D     Scan for previously destroyed pools or import all or only
@@ -1757,12 +1757,6 @@ zpool_do_import(int argc, char **argv)
 	    nvlist_add_uint32(policy, ZPOOL_REWIND_REQUEST, rewind_policy) != 0)
 		goto error;
 
-	if (searchdirs == NULL) {
-		searchdirs = safe_malloc(sizeof (char *));
-		searchdirs[0] = "/dev/dsk";
-		nsearch = 1;
-	}
-
 	/* check argument count */
 	if (do_all) {
 		if (argc != 0) {

cmd/ztest/ztest.c

@@ -1042,6 +1042,7 @@ ztest_pattern_set(void *buf, uint64_t size, uint64_t value)
 		*ip++ = value;
 }
 
+#ifndef NDEBUG
 static boolean_t
 ztest_pattern_match(void *buf, uint64_t size, uint64_t value)
 {
@@ -1054,6 +1055,7 @@ ztest_pattern_match(void *buf, uint64_t size, uint64_t value)
 
 	return (diff == 0);
 }
+#endif
 
 static void
 ztest_bt_generate(ztest_block_tag_t *bt, objset_t *os, uint64_t object,
@@ -4770,7 +4772,6 @@ ztest_run_zdb(char *pool)
 	    bin,
 	    zopt_verbose >= 3 ? "s" : "",
 	    zopt_verbose >= 4 ? "v" : "",
-	    spa_config_path,
 	    pool);
 
 	if (zopt_verbose >= 5)

lib/libzfs/libzfs_import.c

@@ -53,7 +53,6 @@
 #include <sys/vtoc.h>
 #include <sys/dktp/fdisk.h>
 #include <sys/efi_partition.h>
-#include <thread_pool.h>
 
 #include <sys/vdev_impl.h>
 #ifdef HAVE_LIBBLKID
@@ -1004,13 +1003,10 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 	vdev_entry_t *ve, *venext;
 	config_entry_t *ce, *cenext;
 	name_entry_t *ne, *nenext;
-	avl_tree_t slice_cache;
-	rdsk_node_t *slice;
-	void *cookie;
 
-	verify(poolname == NULL || guid == 0);
+	verify(iarg->poolname == NULL || iarg->guid == 0);
 
-	if (argc == 0) {
+	if (dirs == 0) {
 #ifdef HAVE_LIBBLKID
 		/* Use libblkid to scan all device for their type */
 		if (zpool_find_import_blkid(hdl, &pools) == 0)
@@ -1020,8 +1016,8 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 		    dgettext(TEXT_DOMAIN, "blkid failure falling back "
 		    "to manual probing"));
 #endif /* HAVE_LIBBLKID */
-		argc = 1;
+		dirs = 1;
-		argv = &default_dir;
+		dir = &default_dir;
 	}
 
 	/*
@@ -1030,7 +1026,6 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 	 * and toplevel GUID.
 	 */
 	for (i = 0; i < dirs; i++) {
-		tpool_t *t;
 		char *rdsk;
 		int dfd;
 
@@ -1064,8 +1059,6 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 			goto error;
 		}
 
-		avl_create(&slice_cache, slice_cache_compare,
-		    sizeof (rdsk_node_t), offsetof(rdsk_node_t, rn_node));
 		/*
 		 * This is not MT-safe, but we have no MT consumers of libzfs
 		 */
@@ -1076,11 +1069,23 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 				continue;
 
 			/*
-			 * Do not open /dev/watchdog to stat it because
+			 * Skip checking devices with well known prefixes:
-			 * it requires a special close or the watchdog
+			 * watchdog - A special close is required to avoid
-			 * with be triggered and the system reset.
+			 *            triggering it and resetting the system.
+			 * fuse     - Fuse control device.
+			 * ppp      - Generic PPP driver.
+			 * tty*     - Generic serial interface.
+			 * vcs*     - Virtual console memory.
+			 * parport* - Parallel port interface.
+			 * lp*      - Printer interface.
 			 */
-			if (strcmp(name, "watchdog") == 0)
+			if ((strncmp(name, "watchdog", 8) == 0) ||
+			    (strncmp(name, "fuse", 4) == 0)     ||
+			    (strncmp(name, "ppp", 3) == 0)      ||
+			    (strncmp(name, "tty", 3) == 0)      ||
+			    (strncmp(name, "vcs", 3) == 0)      ||
+			    (strncmp(name, "parport", 7) == 0)  ||
+			    (strncmp(name, "lp", 2) == 0))
 				continue;
 
 			if ((fd = openat64(dfd, name, O_RDONLY)) < 0)
@@ -1129,14 +1134,11 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 					continue;
 				}
 				/* use the non-raw path for the config */
-				(void) strlcpy(end, slice->rn_name, pathleft);
+				(void) strlcpy(end, name, pathleft);
 				if (add_config(hdl, &pools, path, config) != 0)
 					goto error;
 			}
-			free(slice->rn_name);
-			free(slice);
 		}
-		avl_destroy(&slice_cache);
 
 		(void) closedir(dirp);
 		dirp = NULL;
@@ -1145,7 +1147,7 @@ zpool_find_import_impl(libzfs_handle_t *hdl, importargs_t *iarg)
 #ifdef HAVE_LIBBLKID
 skip_scanning:
 #endif
-	ret = get_configs(hdl, &pools, active_ok);
+	ret = get_configs(hdl, &pools, iarg->can_be_active);
 
 error:
 	for (pe = pools.pools; pe != NULL; pe = penext) {

lib/libzfs/libzfs_pool.c

@@ -3435,7 +3435,7 @@ zpool_get_history(zpool_handle_t *zhp, nvlist_t **nvhisp)
 int
 zpool_events_next(libzfs_handle_t *hdl, nvlist_t **nvp, int *dropped, int block)
 {
-	zfs_cmd_t zc = { "\0", "\0", "\0", 0 };
+	zfs_cmd_t zc = { "\0", "\0", "\0", "\0", 0 };
 	int error = 0;
 
 	*nvp = NULL;
@@ -3493,7 +3493,7 @@ out:
 int
 zpool_events_clear(libzfs_handle_t *hdl, int *count)
 {
-	zfs_cmd_t zc = { "\0", "\0", "\0", 0 };
+	zfs_cmd_t zc = { "\0", "\0", "\0", "\0", 0 };
 	char msg[1024];
 
 	(void) snprintf(msg, sizeof (msg), dgettext(TEXT_DOMAIN,

module/nvpair/nvpair.c

@@ -3348,9 +3348,13 @@ EXPORT_SYMBOL(nvlist_add_int64_array);
 EXPORT_SYMBOL(nvlist_add_uint64_array);
 EXPORT_SYMBOL(nvlist_add_string_array);
 EXPORT_SYMBOL(nvlist_add_nvlist_array);
+EXPORT_SYMBOL(nvlist_next_nvpair);
+EXPORT_SYMBOL(nvlist_prev_nvpair);
+EXPORT_SYMBOL(nvlist_empty);
 EXPORT_SYMBOL(nvlist_add_hrtime);
 
 EXPORT_SYMBOL(nvlist_remove);
+EXPORT_SYMBOL(nvlist_remove_nvpair);
 EXPORT_SYMBOL(nvlist_remove_all);
 
 EXPORT_SYMBOL(nvlist_lookup_boolean);
@@ -3385,7 +3389,6 @@ EXPORT_SYMBOL(nvlist_lookup_nvpair);
 EXPORT_SYMBOL(nvlist_exists);
 
 /* processing nvpair */
-EXPORT_SYMBOL(nvlist_next_nvpair);
 EXPORT_SYMBOL(nvpair_name);
 EXPORT_SYMBOL(nvpair_type);
 EXPORT_SYMBOL(nvpair_value_boolean_value);

module/zcommon/zfs_comutil.c

@@ -203,4 +203,8 @@ const char *zfs_history_event_names[LOG_END] = {
 
 #if defined(_KERNEL) && defined(HAVE_SPL)
 EXPORT_SYMBOL(zfs_allocatable_devs);
+EXPORT_SYMBOL(zpool_get_rewind_policy);
+EXPORT_SYMBOL(zfs_zpl_version_map);
+EXPORT_SYMBOL(zfs_spa_version_map);
+EXPORT_SYMBOL(zfs_history_event_names);
 #endif

module/zcommon/zprop_common.c

@@ -427,17 +427,18 @@ zprop_width(int prop, boolean_t *fixed, zfs_type_t type)
 
 #if defined(_KERNEL) && defined(HAVE_SPL)
 /* Common routines to initialize property tables */
-EXPORT_SYMBOL(register_impl);
+EXPORT_SYMBOL(zprop_register_impl);
-EXPORT_SYMBOL(register_string);
+EXPORT_SYMBOL(zprop_register_string);
-EXPORT_SYMBOL(register_number);
+EXPORT_SYMBOL(zprop_register_number);
-EXPORT_SYMBOL(register_index);
+EXPORT_SYMBOL(zprop_register_index);
-EXPORT_SYMBOL(register_hidden);
+EXPORT_SYMBOL(zprop_register_hidden);
 
 /* Common routines for zfs and zpool property management */
 EXPORT_SYMBOL(zprop_iter_common);
 EXPORT_SYMBOL(zprop_name_to_prop);
 EXPORT_SYMBOL(zprop_string_to_index);
 EXPORT_SYMBOL(zprop_index_to_string);
+EXPORT_SYMBOL(zprop_random_value);
 EXPORT_SYMBOL(zprop_values);
 EXPORT_SYMBOL(zprop_valid_for_type);
 #endif

module/zfs/ddt.c

@@ -499,6 +499,7 @@ ddt_get_dedup_stats(spa_t *spa, ddt_stat_t *dds_total)
 {
 	ddt_histogram_t *ddh_total;
 
+	/* XXX: Move to a slab */
 	ddh_total = kmem_zalloc(sizeof (ddt_histogram_t), KM_SLEEP);
 	ddt_get_dedup_histogram(spa, ddh_total);
 	ddt_histogram_stat(dds_total, ddh_total);
@@ -647,6 +648,7 @@ ddt_alloc(const ddt_key_t *ddk)
 {
 	ddt_entry_t *dde;
 
+	/* XXX: Move to a slab */
 	dde = kmem_zalloc(sizeof (ddt_entry_t), KM_SLEEP);
 	cv_init(&dde->dde_cv, NULL, CV_DEFAULT, NULL);
 
@@ -795,7 +797,8 @@ ddt_table_alloc(spa_t *spa, enum zio_checksum c)
 {
 	ddt_t *ddt;
 
-	ddt = kmem_zalloc(sizeof (*ddt), KM_SLEEP);
+	/* XXX: Move to a slab */
+	ddt = kmem_zalloc(sizeof (*ddt), KM_SLEEP | KM_NODEBUG);
 
 	mutex_init(&ddt->ddt_lock, NULL, MUTEX_DEFAULT, NULL);
 	avl_create(&ddt->ddt_tree, ddt_entry_compare,

module/zfs/dmu_objset.c

@@ -1686,35 +1686,34 @@ dmu_objset_get_user(objset_t *os)
 }
 
 #if defined(_KERNEL) && defined(HAVE_SPL)
-EXPORT_SYMBOL(dmu_objset_spa);
+EXPORT_SYMBOL(dmu_objset_hold);
-EXPORT_SYMBOL(dmu_objset_zil);
+EXPORT_SYMBOL(dmu_objset_own);
-EXPORT_SYMBOL(dmu_objset_pool);
+EXPORT_SYMBOL(dmu_objset_rele);
-EXPORT_SYMBOL(dmu_objset_ds);
+EXPORT_SYMBOL(dmu_objset_disown);
-EXPORT_SYMBOL(dmu_objset_name);
+EXPORT_SYMBOL(dmu_objset_from_ds);
-EXPORT_SYMBOL(dmu_objset_type);
-EXPORT_SYMBOL(dmu_objset_id);
-EXPORT_SYMBOL(dmu_snapshot_list_next);
-EXPORT_SYMBOL(dmu_dir_list_next);
-EXPORT_SYMBOL(dmu_objset_set_user);
-EXPORT_SYMBOL(dmu_objset_get_user);
-
-/* Public routines to create, destroy, open, and close objsets. */
-EXPORT_SYMBOL(dmu_objset_open);
-EXPORT_SYMBOL(dmu_objset_open_ds);
-EXPORT_SYMBOL(dmu_objset_close);
-EXPORT_SYMBOL(dmu_objset_evict_dbufs);
 EXPORT_SYMBOL(dmu_objset_create);
-EXPORT_SYMBOL(dmu_objset_create_impl);
+EXPORT_SYMBOL(dmu_objset_clone);
 EXPORT_SYMBOL(dmu_objset_destroy);
-EXPORT_SYMBOL(dmu_snapshots_destroy);
-EXPORT_SYMBOL(dmu_objset_rollback);
 EXPORT_SYMBOL(dmu_objset_snapshot);
-EXPORT_SYMBOL(dmu_objset_rename);
-EXPORT_SYMBOL(dmu_objset_find);
-EXPORT_SYMBOL(dmu_objset_byteswap);
-
-/* Get stats on a dataset. */
-EXPORT_SYMBOL(dmu_objset_fast_stat);
 EXPORT_SYMBOL(dmu_objset_stats);
+EXPORT_SYMBOL(dmu_objset_fast_stat);
 EXPORT_SYMBOL(dmu_objset_space);
+EXPORT_SYMBOL(dmu_objset_fsid_guid);
+EXPORT_SYMBOL(dmu_objset_find);
+EXPORT_SYMBOL(dmu_objset_find_spa);
+EXPORT_SYMBOL(dmu_objset_prefetch);
+EXPORT_SYMBOL(dmu_objset_byteswap);
+EXPORT_SYMBOL(dmu_objset_evict_dbufs);
+EXPORT_SYMBOL(dmu_objset_snap_cmtime);
+
+EXPORT_SYMBOL(dmu_objset_sync);
+EXPORT_SYMBOL(dmu_objset_is_dirty);
+EXPORT_SYMBOL(dmu_objset_create_impl);
+EXPORT_SYMBOL(dmu_objset_open_impl);
+EXPORT_SYMBOL(dmu_objset_evict);
+EXPORT_SYMBOL(dmu_objset_do_userquota_updates);
+EXPORT_SYMBOL(dmu_objset_userquota_get_ids);
+EXPORT_SYMBOL(dmu_objset_userused_enabled);
+EXPORT_SYMBOL(dmu_objset_userspace_upgrade);
+EXPORT_SYMBOL(dmu_objset_userspace_present);
 #endif

module/zfs/dsl_dataset.c

@@ -3935,12 +3935,13 @@ EXPORT_SYMBOL(dsl_dataset_destroy_check);
 EXPORT_SYMBOL(dsl_dataset_destroy_sync);
 EXPORT_SYMBOL(dsl_dataset_snapshot_check);
 EXPORT_SYMBOL(dsl_dataset_snapshot_sync);
-EXPORT_SYMBOL(dsl_dataset_rollback);
 EXPORT_SYMBOL(dsl_dataset_rename);
 EXPORT_SYMBOL(dsl_dataset_promote);
 EXPORT_SYMBOL(dsl_dataset_clone_swap);
-EXPORT_SYMBOL(dsl_dataset_set_user_ptr);
+EXPORT_SYMBOL(dsl_dataset_user_hold);
-EXPORT_SYMBOL(dsl_dataset_get_user_ptr);
+EXPORT_SYMBOL(dsl_dataset_user_release);
+EXPORT_SYMBOL(dsl_dataset_user_release_tmp);
+EXPORT_SYMBOL(dsl_dataset_get_holds);
 EXPORT_SYMBOL(dsl_dataset_get_blkptr);
 EXPORT_SYMBOL(dsl_dataset_set_blkptr);
 EXPORT_SYMBOL(dsl_dataset_get_spa);
@@ -3960,9 +3961,5 @@ EXPORT_SYMBOL(dsl_dataset_check_quota);
 EXPORT_SYMBOL(dsl_dataset_set_quota);
 EXPORT_SYMBOL(dsl_dataset_set_quota_sync);
 EXPORT_SYMBOL(dsl_dataset_set_reservation);
-EXPORT_SYMBOL(dsl_dataset_user_hold);
-EXPORT_SYMBOL(dsl_dataset_user_release);
-EXPORT_SYMBOL(dsl_dataset_user_release_tmp);
-EXPORT_SYMBOL(dsl_dataset_get_holds);
 EXPORT_SYMBOL(dsl_destroy_inconsistent);
 #endif

module/zfs/dsl_scan.c

@@ -50,9 +50,7 @@
 
 typedef int (scan_cb_t)(dsl_pool_t *, const blkptr_t *, const zbookmark_t *);
 
-static scan_cb_t dsl_scan_defrag_cb;
 static scan_cb_t dsl_scan_scrub_cb;
-static scan_cb_t dsl_scan_remove_cb;
 static dsl_syncfunc_t dsl_scan_cancel_sync;
 static void dsl_scan_sync_state(dsl_scan_t *, dmu_tx_t *tx);
 
@@ -189,9 +187,9 @@ dsl_scan_setup_sync(void *arg1, void *arg2, dmu_tx_t *tx)
 
 		if (vdev_resilver_needed(spa->spa_root_vdev,
 		    &scn->scn_phys.scn_min_txg, &scn->scn_phys.scn_max_txg)) {
-			spa_event_notify(spa, NULL, ESC_ZFS_RESILVER_START);
+			spa_event_notify(spa, NULL, FM_EREPORT_ZFS_RESILVER_START);
 		} else {
-			spa_event_notify(spa, NULL, ESC_ZFS_SCRUB_START);
+			spa_event_notify(spa, NULL, FM_EREPORT_ZFS_SCRUB_START);
 		}
 
 		spa->spa_scrub_started = B_TRUE;
@@ -292,7 +290,8 @@ dsl_scan_done(dsl_scan_t *scn, boolean_t complete, dmu_tx_t *tx)
 		    complete ? scn->scn_phys.scn_max_txg : 0, B_TRUE);
 		if (complete) {
 			spa_event_notify(spa, NULL, scn->scn_phys.scn_min_txg ?
-			    ESC_ZFS_RESILVER_FINISH : ESC_ZFS_SCRUB_FINISH);
+			    FM_EREPORT_ZFS_RESILVER_FINISH :
+			    FM_EREPORT_ZFS_SCRUB_FINISH);
 		}
 		spa_errlog_rotate(spa);
 

module/zfs/fm.c

@@ -432,7 +432,7 @@ static void
 fm_event_free(zevent_t *ev)
 {
 	/* Run provided cleanup callback */
-	ev->ev_cb(ev->ev_nvl);
+	ev->ev_cb(ev->ev_nvl, ev->ev_detector);
 
 	list_destroy(&ev->ev_zpd_list);
 	kmem_free(ev, sizeof(zevent_t));

module/zfs/include/sys/fm/util.h

@@ -79,7 +79,7 @@ typedef struct erpt_dump {
 
 #define ZEVENT_SHUTDOWN	0x1
 
-typedef void zevent_cb_t(nvlist_t *);
+typedef void zevent_cb_t(nvlist_t *, nvlist_t *);
 
 typedef struct zevent_s {
 	nvlist_t	*ev_nvl;       /* protected by the zevent_lock */
@@ -100,7 +100,7 @@ extern void fm_fini(void);
 extern void fm_nvprint(nvlist_t *);
 extern void fm_zevent_init(zfs_private_data_t *);
 extern void fm_zevent_fini(zfs_private_data_t *);
-extern void fm_zevent_post(nvlist_t *, zevent_cb_t *);
+extern void fm_zevent_post(nvlist_t *, nvlist_t *, zevent_cb_t *);
 extern void fm_zevent_drain_all(int *);
 extern int fm_zevent_next(zfs_private_data_t *, zfs_cmd_t *);
 extern int fm_zevent_wait(zfs_private_data_t *);

module/zfs/include/sys/spa_impl.h

@@ -208,7 +208,7 @@ struct spa {
 	kmutex_t	spa_proc_lock;		/* protects spa_proc* */
 	kcondvar_t	spa_proc_cv;		/* spa_proc_state transitions */
 	spa_proc_state_t spa_proc_state;	/* see definition */
-	struct proc	*spa_proc;		/* "zpool-poolname" process */
+	proc_t		*spa_proc;		/* "zpool-poolname" process */
 	uint64_t	spa_did;		/* if procp != p0, did of t1 */
 	boolean_t	spa_autoreplace;	/* autoreplace set in open */
 	int		spa_vdev_locks;		/* locks grabbed */

module/zfs/include/sys/zfs_fuid.h

@@ -33,6 +33,7 @@
 #include <sys/zfs_vfsops.h>
 #endif
 #include <sys/avl.h>
+#include <sys/list.h>
 
 #ifdef	__cplusplus
 extern "C" {

module/zfs/sa.c

@@ -1353,6 +1353,7 @@ sa_lookup(sa_handle_t *hdl, sa_attr_type_t attr, void *buf, uint32_t buflen)
 int
 sa_lookup_uio(sa_handle_t *hdl, sa_attr_type_t attr, uio_t *uio)
 {
+#ifdef HAVE_ZPL
 	int error;
 	sa_bulk_attr_t bulk;
 
@@ -1371,7 +1372,9 @@ sa_lookup_uio(sa_handle_t *hdl, sa_attr_type_t attr, uio_t *uio)
 	}
 	mutex_exit(&hdl->sa_lock);
 	return (error);
-
+#else
+	return ENOSYS;
+#endif /* HAVE_ZPL */
 }
 #endif
 

module/zfs/spa.c

@@ -4041,7 +4041,7 @@ spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, int replace_done)
 	vd->vdev_detached = B_TRUE;
 	vdev_dirty(tvd, VDD_DTL, vd, txg);
 
-	spa_event_notify(spa, vd, ESC_ZFS_VDEV_REMOVE);
+	spa_event_notify(spa, vd, FM_EREPORT_ZFS_DEVICE_REMOVE);
 
 	error = spa_vdev_exit(spa, vd, txg, 0);
 
@@ -5647,9 +5647,12 @@ spa_event_notify(spa_t *spa, vdev_t *vd, const char *name)
 #if defined(_KERNEL) && defined(HAVE_SPL)
 /* state manipulation functions */
 EXPORT_SYMBOL(spa_open);
+EXPORT_SYMBOL(spa_open_rewind);
 EXPORT_SYMBOL(spa_get_stats);
 EXPORT_SYMBOL(spa_create);
+EXPORT_SYMBOL(spa_import_rootpool);
 EXPORT_SYMBOL(spa_import);
+EXPORT_SYMBOL(spa_import_verbatim);
 EXPORT_SYMBOL(spa_tryimport);
 EXPORT_SYMBOL(spa_destroy);
 EXPORT_SYMBOL(spa_export);
@@ -5659,6 +5662,8 @@ EXPORT_SYMBOL(spa_async_suspend);
 EXPORT_SYMBOL(spa_async_resume);
 EXPORT_SYMBOL(spa_inject_addref);
 EXPORT_SYMBOL(spa_inject_delref);
+EXPORT_SYMBOL(spa_scan_stat_init);
+EXPORT_SYMBOL(spa_scan_get_stats);
 
 /* device maniion */
 EXPORT_SYMBOL(spa_vdev_add);
@@ -5666,6 +5671,8 @@ EXPORT_SYMBOL(spa_vdev_attach);
 EXPORT_SYMBOL(spa_vdev_detach);
 EXPORT_SYMBOL(spa_vdev_remove);
 EXPORT_SYMBOL(spa_vdev_setpath);
+EXPORT_SYMBOL(spa_vdev_setfru);
+EXPORT_SYMBOL(spa_vdev_split_mirror);
 
 /* spare statech is global across all pools) */
 EXPORT_SYMBOL(spa_spare_add);
@@ -5679,10 +5686,10 @@ EXPORT_SYMBOL(spa_l2cache_remove);
 EXPORT_SYMBOL(spa_l2cache_exists);
 EXPORT_SYMBOL(spa_l2cache_activate);
 EXPORT_SYMBOL(spa_l2cache_drop);
-EXPORT_SYMBOL(spa_l2cache_space_update);
 
-/* scrubbing */
+/* scanning */
-EXPORT_SYMBOL(spa_scrub);
+EXPORT_SYMBOL(spa_scan);
+EXPORT_SYMBOL(spa_scan_stop);
 
 /* spa syncing */
 EXPORT_SYMBOL(spa_sync); /* only for DMU use */
@@ -5693,8 +5700,6 @@ EXPORT_SYMBOL(spa_prop_set);
 EXPORT_SYMBOL(spa_prop_get);
 EXPORT_SYMBOL(spa_prop_clear_bootfs);
 
-#if defined(HAVE_SYSEVENT)
 /* asynchronous event notification */
 EXPORT_SYMBOL(spa_event_notify);
 #endif
-#endif

module/zfs/spa_history.c

@@ -441,11 +441,7 @@ log_internal(history_internal_events_t event, spa_t *spa,
 		return;
 
 	ha = kmem_alloc(sizeof (history_arg_t), KM_SLEEP);
-	ha->ha_history_str = kmem_alloc(vsnprintf(NULL, 0, fmt, adx) + 1,
+	ha->ha_history_str = kmem_asprintf(fmt, adx);
-	    KM_SLEEP);
-
-	(void) vsprintf(ha->ha_history_str, fmt, adx);
-
 	ha->ha_log_type = LOG_INTERNAL;
 	ha->ha_event = event;
 	ha->ha_zone = NULL;
@@ -509,5 +505,6 @@ spa_history_log_version(spa_t *spa, history_internal_events_t event)
 EXPORT_SYMBOL(spa_history_create_obj);
 EXPORT_SYMBOL(spa_history_get);
 EXPORT_SYMBOL(spa_history_log);
-EXPORT_SYMBOL(spa_history_internal_log);
+EXPORT_SYMBOL(spa_history_log_internal);
+EXPORT_SYMBOL(spa_history_log_version);
 #endif

module/zfs/spa_misc.c

@@ -1713,16 +1713,24 @@ EXPORT_SYMBOL(spa_name);
 EXPORT_SYMBOL(spa_guid);
 EXPORT_SYMBOL(spa_last_synced_txg);
 EXPORT_SYMBOL(spa_first_txg);
+EXPORT_SYMBOL(spa_syncing_txg);
 EXPORT_SYMBOL(spa_version);
 EXPORT_SYMBOL(spa_state);
+EXPORT_SYMBOL(spa_load_state);
 EXPORT_SYMBOL(spa_freeze_txg);
-EXPORT_SYMBOL(spa_get_alloc);
-EXPORT_SYMBOL(spa_get_space);
-EXPORT_SYMBOL(spa_get_dspace);
 EXPORT_SYMBOL(spa_get_asize);
+EXPORT_SYMBOL(spa_get_dspace);
+EXPORT_SYMBOL(spa_update_dspace);
+EXPORT_SYMBOL(spa_deflate);
+EXPORT_SYMBOL(spa_normal_class);
+EXPORT_SYMBOL(spa_log_class);
 EXPORT_SYMBOL(spa_max_replication);
+EXPORT_SYMBOL(spa_prev_software_version);
 EXPORT_SYMBOL(spa_get_failmode);
 EXPORT_SYMBOL(spa_suspended);
+EXPORT_SYMBOL(spa_bootfs);
+EXPORT_SYMBOL(spa_delegation);
+EXPORT_SYMBOL(spa_meta_objset);
 
 /* Miscellaneous support routines */
 EXPORT_SYMBOL(spa_rename);
@@ -1730,15 +1738,21 @@ EXPORT_SYMBOL(spa_guid_exists);
 EXPORT_SYMBOL(spa_strdup);
 EXPORT_SYMBOL(spa_strfree);
 EXPORT_SYMBOL(spa_get_random);
+EXPORT_SYMBOL(spa_generate_guid);
 EXPORT_SYMBOL(sprintf_blkptr);
 EXPORT_SYMBOL(spa_freeze);
 EXPORT_SYMBOL(spa_upgrade);
 EXPORT_SYMBOL(spa_evict_all);
 EXPORT_SYMBOL(spa_lookup_by_guid);
 EXPORT_SYMBOL(spa_has_spare);
-EXPORT_SYMBOL(bp_get_dasize);
+EXPORT_SYMBOL(dva_get_dsize_sync);
+EXPORT_SYMBOL(bp_get_dsize_sync);
+EXPORT_SYMBOL(bp_get_dsize);
 EXPORT_SYMBOL(spa_has_slogs);
 EXPORT_SYMBOL(spa_is_root);
+EXPORT_SYMBOL(spa_writeable);
+EXPORT_SYMBOL(spa_rewind_data_to_nvlist);
+EXPORT_SYMBOL(spa_mode);
 
 EXPORT_SYMBOL(spa_namespace_lock);
 #endif

module/zfs/txg.c

@@ -732,8 +732,6 @@ EXPORT_SYMBOL(txg_hold_open);
 EXPORT_SYMBOL(txg_rele_to_quiesce);
 EXPORT_SYMBOL(txg_rele_to_sync);
 EXPORT_SYMBOL(txg_register_callbacks);
-EXPORT_SYMBOL(txg_suspend);
-EXPORT_SYMBOL(txg_resume);
 EXPORT_SYMBOL(txg_delay);
 EXPORT_SYMBOL(txg_wait_synced);
 EXPORT_SYMBOL(txg_wait_open);

module/zfs/vdev.c

@@ -1072,6 +1072,12 @@ vdev_open_child(void *arg)
 boolean_t
 vdev_uses_zvols(vdev_t *vd)
 {
+/*
+ * NOTE: Disabled because under Linux I've choosen not to put all the zvols
+ * in their own directory.  This could be changed or this code can be updated
+ * to perhap run an ioctl() on the vdev path to determine if it is a zvol.
+ */
+#if 0
 	int c;
 
 	if (vd->vdev_path && strncmp(vd->vdev_path, ZVOL_DIR,
@@ -1080,6 +1086,7 @@ vdev_uses_zvols(vdev_t *vd)
 	for (c = 0; c < vd->vdev_children; c++)
 		if (vdev_uses_zvols(vd->vdev_child[c]))
 			return (B_TRUE);
+#endif
 	return (B_FALSE);
 }
 

module/zfs/vdev_disk.c

@@ -550,6 +550,35 @@ vdev_disk_io_done(zio_t *zio)
 	}
 }
 
+static void
+vdev_disk_hold(vdev_t *vd)
+{
+	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));
+
+	/* We must have a pathname, and it must be absolute. */
+	if (vd->vdev_path == NULL || vd->vdev_path[0] != '/')
+		return;
+
+	/*
+	 * Only prefetch path and devid info if the device has
+	 * never been opened.
+	 */
+	if (vd->vdev_tsd != NULL)
+		return;
+
+	/* XXX: Implement me as a vnode lookup for the device */
+	vd->vdev_name_vp = NULL;
+	vd->vdev_devid_vp = NULL;
+}
+
+static void
+vdev_disk_rele(vdev_t *vd)
+{
+	ASSERT(spa_config_held(vd->vdev_spa, SCL_STATE, RW_WRITER));
+
+	/* XXX: Implement me as a vnode rele for the device */
+}
+
 vdev_ops_t vdev_disk_ops = {
 	vdev_disk_open,
 	vdev_disk_close,
@@ -557,6 +586,8 @@ vdev_ops_t vdev_disk_ops = {
 	vdev_disk_io_start,
 	vdev_disk_io_done,
 	NULL,
+	vdev_disk_hold,
+	vdev_disk_rele,
 	VDEV_TYPE_DISK,		/* name of this vdev type */
 	B_TRUE			/* leaf vdev */
 };

module/zfs/zfs_fm.c

@@ -235,7 +235,6 @@ zfs_ereport_start(nvlist_t **ereport_out, nvlist_t **detector_out,
 	    vd != NULL ? vd->vdev_guid : 0);
 
 	fm_ereport_set(ereport, FM_EREPORT_VERSION, class, ena, detector, NULL);
-	fm_nvlist_destroy(detector, FM_NVA_FREE);
 
 	/*
 	 * Construct the per-ereport payload, depending on which parameters are
@@ -440,12 +439,13 @@ shrink_ranges(zfs_ecksum_info_t *eip)
 		uint32_t end = r[idx].zr_end;
 
 		while (idx < max - 1) {
+			uint32_t nstart, nend, gap;
+
 			idx++;
+			nstart = r[idx].zr_start;
+			nend = r[idx].zr_end;
 
-			uint32_t nstart = r[idx].zr_start;
+			gap = nstart - end;
-			uint32_t nend = r[idx].zr_end;
-
-			uint32_t gap = nstart - end;
 			if (gap < new_allowed_gap) {
 				end = nend;
 				continue;

module/zfs/zfs_ioctl.c

@@ -104,8 +104,8 @@ static const char *userquota_perms[] = {
 static int zfs_ioc_userspace_upgrade(struct file *filp, zfs_cmd_t *zc);
 static int zfs_check_settable(struct file *filp, const char *name,
     nvpair_t *property, cred_t *cr);
-static int zfs_check_clearable(char *dataset, nvlist_t *props,
+static int zfs_check_clearable(struct file *filp, char *dataset,
-    nvlist_t **errors);
+    nvlist_t *props, nvlist_t **errors);
 static int zfs_fill_zplprops_root(uint64_t, nvlist_t *, nvlist_t *,
     boolean_t *);
 int zfs_set_prop_nvlist(struct file *filp, const char *, zprop_source_t,
@@ -339,6 +339,7 @@ zfs_secpolicy_write_perms(const char *name, const char *perm, cred_t *cr)
 static int
 zfs_set_slabel_policy(const char *name, char *strval, cred_t *cr)
 {
+#if 0
 	char		ds_hexsl[MAXNAMELEN];
 	bslabel_t	ds_sl, new_sl;
 	boolean_t	new_default = FALSE;
@@ -426,6 +427,9 @@ out_check:
 	if (needed_priv != -1)
 		return (PRIV_POLICY(cr, needed_priv, B_FALSE, EPERM, NULL));
 	return (0);
+#else
+	return EPERM;
+#endif
 }
 
 static int
@@ -958,7 +962,7 @@ fit_error_list(zfs_cmd_t *zc, nvlist_t **errors)
 	return (0);
 }
 
-static int
+int
 put_nvlist(zfs_cmd_t *zc, nvlist_t *nvl)
 {
 	char *packed = NULL;
@@ -1009,6 +1013,7 @@ getzfsvfs(const char *dsname, zfsvfs_t **zfvp)
 	dmu_objset_rele(os, FTAG);
 	return (error);
 }
+#endif
 
 /*
  * Find a zfsvfs_t for a mounted filesystem, or create our own, in which
@@ -1017,6 +1022,7 @@ getzfsvfs(const char *dsname, zfsvfs_t **zfvp)
 static int
 zfsvfs_hold(const char *name, void *tag, zfsvfs_t **zfvp)
 {
+#ifdef HAVE_ZPL
 	int error = 0;
 
 	if (getzfsvfs(name, zfvp) != 0)
@@ -1034,11 +1040,15 @@ zfsvfs_hold(const char *name, void *tag, zfsvfs_t **zfvp)
 		}
 	}
 	return (error);
+#else
+	return ENOTSUP;
+#endif
 }
 
 static void
 zfsvfs_rele(zfsvfs_t *zfsvfs, void *tag)
 {
+#ifdef HAVE_ZPL
 	rrw_exit(&zfsvfs->z_teardown_lock, tag);
 
 	if (zfsvfs->z_vfs) {
@@ -1047,8 +1057,8 @@ zfsvfs_rele(zfsvfs_t *zfsvfs, void *tag)
 		dmu_objset_disown(zfsvfs->z_os, zfsvfs);
 		zfsvfs_free(zfsvfs);
 	}
+#endif
 }
-#endif /* HAVE_ZPL */
 
 static int
 zfs_ioc_pool_create(struct file *filp, zfs_cmd_t *zc)
@@ -1559,8 +1569,7 @@ zfs_ioc_vdev_split(struct file *filp, zfs_cmd_t *zc)
 }
 
 static int
-zfs_ioc_vdev_setpath(zfs_cmd_t *zc)
+zfs_ioc_vdev_setpath(struct file *filp, zfs_cmd_t *zc)
->>>>>>> refs/top-bases/linux-kernel-device
 {
 	spa_t *spa;
 	char *path = zc->zc_value;
@@ -1888,6 +1897,7 @@ top:
 static int
 zfs_prop_set_userquota(struct file *filp, const char *dsname, nvpair_t *pair)
 {
+#ifdef HAVE_ZPL
 	const char *propname = nvpair_name(pair);
 	uint64_t *valary;
 	unsigned int vallen;
@@ -1928,6 +1938,9 @@ zfs_prop_set_userquota(struct file *filp, const char *dsname, nvpair_t *pair)
 	}
 
 	return (err);
+#else
+	return ENOTSUP;
+#endif
 }
 
 /*
@@ -1949,7 +1962,7 @@ zfs_prop_set_special(struct file *filp, const char *dsname,
 
 	if (prop == ZPROP_INVAL) {
 		if (zfs_prop_userquota(propname))
-			return (zfs_prop_set_userquota(dsname, pair));
+			return (zfs_prop_set_userquota(filp, dsname, pair));
 		return (-1);
 	}
 
@@ -1979,8 +1992,7 @@ zfs_prop_set_special(struct file *filp, const char *dsname,
 		err = dsl_dataset_set_reservation(dsname, source, intval);
 		break;
 	case ZFS_PROP_VOLSIZE:
-		err = zvol_set_volsize(dsname, ddi_driver_major(zfs_dip),
+		err = zvol_set_volsize(dsname, intval);
-		    intval);
 		break;
 	case ZFS_PROP_VERSION:
 	{
@@ -1989,7 +2001,9 @@ zfs_prop_set_special(struct file *filp, const char *dsname,
 		if ((err = zfsvfs_hold(dsname, FTAG, &zfsvfs)) != 0)
 			break;
 
+#ifdef HAVE_ZPL
 		err = zfs_set_version(zfsvfs, intval);
+#endif
 		zfsvfs_rele(zfsvfs, FTAG);
 
 		if (err == 0 && intval >= ZPL_VERSION_USERSPACE) {
@@ -1997,7 +2011,7 @@ zfs_prop_set_special(struct file *filp, const char *dsname,
 
 			zc = kmem_zalloc(sizeof (zfs_cmd_t), KM_SLEEP);
 			(void) strcpy(zc->zc_name, dsname);
-			(void) zfs_ioc_userspace_upgrade(zc);
+			(void) zfs_ioc_userspace_upgrade(filp, zc);
 			kmem_free(zc, sizeof (zfs_cmd_t));
 		}
 		break;
@@ -2097,10 +2111,10 @@ retry:
 
 		/* Validate permissions */
 		if (err == 0)
-			err = zfs_check_settable(dsname, pair, CRED());
+			err = zfs_check_settable(filp, dsname, pair, CRED());
 
 		if (err == 0) {
-			err = zfs_prop_set_special(dsname, source, pair);
+			err = zfs_prop_set_special(filp, dsname, source, pair);
 			if (err == -1) {
 				/*
 				 * For better performance we build up a list of
@@ -2229,8 +2243,8 @@ props_skip(nvlist_t *props, nvlist_t *skipped, nvlist_t **newprops)
 }
 
 static int
-clear_received_props(objset_t *os, const char *fs, nvlist_t *props,
+clear_received_props(struct file *filp, objset_t *os,
-    nvlist_t *skipped)
+    const char *fs, nvlist_t *props, nvlist_t *skipped)
 {
 	int err = 0;
 	nvlist_t *cleared_props = NULL;
@@ -2242,7 +2256,7 @@ clear_received_props(objset_t *os, const char *fs, nvlist_t *props,
 		 */
 		zprop_source_t flags = (ZPROP_SRC_NONE |
 		    (dsl_prop_get_hasrecvd(os) ? ZPROP_SRC_RECEIVED : 0));
-		err = zfs_set_prop_nvlist(fs, flags, cleared_props, NULL);
+		err = zfs_set_prop_nvlist(filp, fs, flags, cleared_props, NULL);
 	}
 	nvlist_free(cleared_props);
 	return (err);
@@ -2355,7 +2369,7 @@ zfs_ioc_inherit_prop(struct file *filp, zfs_cmd_t *zc)
 		}
 
 		pair = nvlist_next_nvpair(dummy, NULL);
-		err = zfs_prop_set_special(zc->zc_name, source, pair);
+		err = zfs_prop_set_special(filp, zc->zc_name, source, pair);
 		nvlist_free(dummy);
 		if (err != -1)
 			return (err); /* special property already handled */
@@ -2521,6 +2535,7 @@ zfs_ioc_get_fsacl(struct file *filp, zfs_cmd_t *zc)
 	return (error);
 }
 
+#ifdef HAVE_ZPL
 /*
  * Search the vfs list for a specified resource.  Returns a pointer to it
  * or NULL if no suitable entry is found. The caller of this routine
@@ -2847,9 +2862,18 @@ zfs_ioc_create(struct file *filp, zfs_cmd_t *zc)
 	if (error == 0) {
 		error = zfs_set_prop_nvlist(filp, zc->zc_name, ZPROP_SRC_LOCAL,
 		    nvprops, NULL);
+		if (error != 0) {
+			(void) dmu_objset_destroy(zc->zc_name, B_FALSE);
+			goto out;
+		}
+
+		if (type == DMU_OST_ZVOL) {
+			error = zvol_create_minor(zc->zc_name);
 			if (error != 0)
 				(void) dmu_objset_destroy(zc->zc_name, B_FALSE);
 		}
+	}
+out:
 	nvlist_free(nvprops);
 	return (error);
 }
@@ -3379,9 +3403,7 @@ static int
 zfs_ioc_recv(struct file *filp, zfs_cmd_t *zc)
 {
 	file_t *fp;
-#ifdef HAVE_ZPL
 	objset_t *os;
-#endif /* HAVE_ZPL */
 	dmu_recv_cookie_t drc;
 	boolean_t force = (boolean_t)zc->zc_guid;
 	int fd;
@@ -3417,7 +3439,6 @@ zfs_ioc_recv(struct file *filp, zfs_cmd_t *zc)
 		return (EBADF);
 	}
 
-#ifdef HAVE_ZPL
 	VERIFY(nvlist_alloc(&errors, NV_UNIQUE_NAME, KM_SLEEP) == 0);
 
 	if (props && dmu_objset_hold(tofs, FTAG, &os) == 0) {
@@ -3443,7 +3464,7 @@ zfs_ioc_recv(struct file *filp, zfs_cmd_t *zc)
 			 */
 			if (!first_recvd_props)
 				props_reduce(props, origprops);
-			if (zfs_check_clearable(tofs, origprops,
+			if (zfs_check_clearable(filp, tofs, origprops,
 			    &errlist) != 0)
 				(void) nvlist_merge(errors, errlist, 0);
 			nvlist_free(errlist);
@@ -3451,7 +3472,6 @@ zfs_ioc_recv(struct file *filp, zfs_cmd_t *zc)
 
 		dmu_objset_rele(os, FTAG);
 	}
-#endif /* HAVE_ZPL */
 
 	if (zc->zc_string[0]) {
 		error = dmu_objset_hold(zc->zc_string, FTAG, &origin);
@@ -3480,7 +3500,7 @@ zfs_ioc_recv(struct file *filp, zfs_cmd_t *zc)
 				    SPA_VERSION_RECVD_PROPS)
 					first_recvd_props = B_TRUE;
 			} else if (origprops != NULL) {
-				if (clear_received_props(os, tofs, origprops,
+				if (clear_received_props(filp,os,tofs,origprops,
 				    first_recvd_props ? NULL : props) != 0)
 					zc->zc_obj |= ZPROP_ERR_NOCLEAR;
 			} else {

module/zfs/zfs_sa.c

@@ -67,7 +67,7 @@ sa_attr_reg_t zfs_attr_table[ZPL_END+1] = {
 };
 
 #ifdef _KERNEL
-
+#ifdef HAVE_ZPL
 int
 zfs_sa_readlink(znode_t *zp, uio_t *uio)
 {
@@ -309,4 +309,5 @@ zfs_sa_upgrade_txholds(dmu_tx_t *tx, znode_t *zp)
 	}
 }
 
+#endif /* HAVE_ZPL */
 #endif

module/zfs/zvol.c

@@ -51,6 +51,7 @@ unsigned int zvol_threads = 0;
 static taskq_t *zvol_taskq;
 static kmutex_t zvol_state_lock;
 static list_t zvol_state_list;
+static char *zvol_tag = "zvol_tag";
 
 /*
  * The in-core state of each volume.
@@ -59,11 +60,12 @@ typedef struct zvol_state {
 	uint64_t		zv_volsize;	/* advertised space */
 	uint64_t		zv_volblocksize;/* volume block size */
 	objset_t		*zv_objset;	/* objset handle */
-	uint32_t		zv_mode;	/* DS_MODE_* at open time */
+	uint32_t		zv_flags;	/* ZVOL_* flags */
 	uint32_t		zv_open_count;	/* open counts */
 	uint32_t		zv_changed;	/* disk changed */
 	zilog_t			*zv_zilog;	/* ZIL handle */
 	znode_t			zv_znode;	/* for range locking */
+	dmu_buf_t		*zv_dbuf;	/* bonus handle */
 	dev_t			zv_dev;		/* device id */
 	struct gendisk		*zv_disk;	/* generic disk */
 	struct request_queue	*zv_queue;	/* request queue */
@@ -71,6 +73,8 @@ typedef struct zvol_state {
 	list_node_t		zv_next;	/* next zvol_state_t linkage */
 } zvol_state_t;
 
+#define	ZVOL_RDONLY	0x1
+
 /*
  * Find the next available range of ZVOL_MINORS minor numbers.  The
  * zvol_state_list is kept in ascending minor order so we simply need
@@ -197,15 +201,6 @@ zvol_get_stats(objset_t *os, nvlist_t *nv)
 	return (error);
 }
 
-/*
- * Notification handler for objset readonly property changes.
- */
-static void
-zvol_readonly_changed_cb(void *arg, uint64_t value)
-{
-	set_disk_ro(((zvol_state_t *)arg)->zv_disk, !!value);
-}
-
 /*
  * Sanity check volume size.
  */
@@ -278,40 +273,44 @@ int
 zvol_set_volsize(const char *name, uint64_t volsize)
 {
 	zvol_state_t *zv;
-	int error;
 	dmu_object_info_t doi;
-	uint64_t old_volsize = 0ULL;
+	objset_t *os = NULL;
 	zvol_state_t state = { 0 };
+	uint64_t old_volsize = 0ULL;
+	uint64_t readonly;
+	int error;
 
 	mutex_enter(&zvol_state_lock);
 
 	zv = zvol_find_by_name(name);
 	if (zv == NULL) {
-		/*
+		error = dmu_objset_hold(name, FTAG, &os);
-		 * If we are doing a "zfs clone -o volsize=", then the
+		if (error)
-		 * minor node won't exist yet.
-		 */
-		error = dmu_objset_open(name, DMU_OST_ZVOL, DS_MODE_OWNER,
-		    &state.zv_objset);
-		if (error != 0)
 			goto out;
 		zv = &state;
 	}
+
+	VERIFY(dsl_prop_get_integer(name, "readonly", &readonly, NULL) == 0);
+	if (readonly) {
+		error = EROFS;
+		goto out;
+	}
+
 	old_volsize = zv->zv_volsize;
 
-	if ((error = dmu_object_info(zv->zv_objset, ZVOL_OBJ, &doi)) != 0 ||
+	if ((error = dmu_object_info(os, ZVOL_OBJ, &doi)) != 0 ||
 	    (error = zvol_check_volsize(volsize,doi.doi_data_block_size)) != 0)
 		goto out;
 
-	if (get_disk_ro(zv->zv_disk) || (zv->zv_mode & DS_MODE_READONLY)) {
+	if (get_disk_ro(zv->zv_disk) || (zv->zv_flags & ZVOL_RDONLY)) {
 		error = EROFS;
 		goto out;
 	}
 
 	error = zvol_update_volsize(zv, volsize);
 out:
-	if (state.zv_objset)
+	if (os)
-		dmu_objset_close(state.zv_objset);
+		dmu_objset_rele(os, FTAG);
 
 	mutex_exit(&zvol_state_lock);
 
@@ -348,7 +347,7 @@ zvol_set_volblocksize(const char *name, uint64_t volblocksize)
 	if (zv == NULL)
 		return (ENXIO);
 
-	if (get_disk_ro(zv->zv_disk) || (zv->zv_mode & DS_MODE_READONLY))
+	if (get_disk_ro(zv->zv_disk) || (zv->zv_flags & ZVOL_RDONLY))
 		return (EROFS);
 
 	tx = dmu_tx_create(zv->zv_objset);
@@ -441,16 +440,9 @@ zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx,
 	zilog_t *zilog = zv->zv_zilog;
 	boolean_t slogging;
 
-	if (zil_disable)
+	if (zil_replaying(zilog, tx))
 		return;
 
-	if (zilog->zl_replay) {
-		dsl_dataset_dirty(dmu_objset_ds(zilog->zl_os), tx);
-		zilog->zl_replayed_seq[dmu_tx_get_txg(tx) & TXG_MASK] =
-		    zilog->zl_replaying_seq;
-		return;
-	}
-
 	slogging = spa_has_slogs(zilog->zl_spa);
 
 	while (size) {
@@ -480,8 +472,7 @@ zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx,
 		lr = (lr_write_t *)&itx->itx_lr;
 		if (write_state == WR_COPIED && dmu_read(zv->zv_objset,
 		    ZVOL_OBJ, offset, len, lr+1, DMU_READ_NO_PREFETCH) != 0) {
-			kmem_free(itx, offsetof(itx_t, itx_lr) +
+			zil_itx_destroy(itx);
-			    itx->itx_lr.lrc_reclen);
 			itx = zil_itx_create(TX_WRITE, sizeof (*lr));
 			lr = (lr_write_t *)&itx->itx_lr;
 			write_state = WR_NEED_COPY;
@@ -493,8 +484,7 @@ zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx,
 		lr->lr_foid = ZVOL_OBJ;
 		lr->lr_offset = offset;
 		lr->lr_length = len;
-		lr->lr_blkoff = offset -
+		lr->lr_blkoff = 0;
-			        P2ALIGN_TYPED(offset, blocksize, uint64_t);
 		BP_ZERO(&lr->lr_blkptr);
 
 		itx->itx_private = zv;
@@ -520,13 +510,10 @@ zvol_write(void *arg)
 	zvol_state_t *zv = q->queuedata;
 	uint64_t offset = blk_rq_pos(req) << 9;
 	uint64_t size = blk_rq_bytes(req);
-	int sync = 0, error = 0;
+	int error = 0;
 	dmu_tx_t *tx;
 	rl_t *rl;
 
-	if (rq_is_sync(req) && !zil_disable)
-		sync = 1;
-
 	rl = zfs_range_lock(&zv->zv_znode, offset, size, RL_WRITER);
 
 	tx = dmu_tx_create(zv->zv_objset);
@@ -541,13 +528,14 @@ zvol_write(void *arg)
 		return;
 	}
 
-	dmu_write_req(zv->zv_objset, ZVOL_OBJ, req, tx);
+	error = dmu_write_req(zv->zv_objset, ZVOL_OBJ, req, tx);
-	zvol_log_write(zv, tx, offset, size, sync);
+	if (error == 0)
+		zvol_log_write(zv, tx, offset, size, rq_is_sync(req));
 
 	dmu_tx_commit(tx);
 	zfs_range_unlock(rl);
 
-	if (sync)
+	if (rq_is_sync(req))
 		zil_commit(zv->zv_zilog, UINT64_MAX, ZVOL_OBJ);
 
 	blk_end_request(req, -error, size);
@@ -643,7 +631,7 @@ zvol_request(struct request_queue *q)
 			break;
 		case WRITE:
 			if (unlikely(get_disk_ro(zv->zv_disk)) ||
-			    unlikely(zv->zv_mode & DS_MODE_READONLY)) {
+			    unlikely(zv->zv_flags & ZVOL_RDONLY)) {
 				__blk_end_request(req, -EROFS, size);
 				break;
 			}
@@ -659,6 +647,77 @@ zvol_request(struct request_queue *q)
 	}
 }
 
+static void
+zvol_get_done(zgd_t *zgd, int error)
+{
+	if (zgd->zgd_db)
+		dmu_buf_rele(zgd->zgd_db, zgd);
+
+	zfs_range_unlock(zgd->zgd_rl);
+
+	if (error == 0 && zgd->zgd_bp)
+		zil_add_block(zgd->zgd_zilog, zgd->zgd_bp);
+
+	kmem_free(zgd, sizeof (zgd_t));
+}
+
+/*
+ * Get data to generate a TX_WRITE intent log record.
+ */
+static int
+zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio)
+{
+	zvol_state_t *zv = arg;
+	objset_t *os = zv->zv_objset;
+	uint64_t offset = lr->lr_offset;
+	uint64_t size = lr->lr_length;
+	dmu_buf_t *db;
+	zgd_t *zgd;
+	int error;
+
+	ASSERT(zio != NULL);
+	ASSERT(size != 0);
+
+	zgd = (zgd_t *)kmem_zalloc(sizeof (zgd_t), KM_SLEEP);
+	zgd->zgd_zilog = zv->zv_zilog;
+	zgd->zgd_rl = zfs_range_lock(&zv->zv_znode, offset, size, RL_READER);
+
+	/*
+	 * Write records come in two flavors: immediate and indirect.
+	 * For small writes it's cheaper to store the data with the
+	 * log record (immediate); for large writes it's cheaper to
+	 * sync the data and get a pointer to it (indirect) so that
+	 * we don't have to write the data twice.
+	 */
+	if (buf != NULL) { /* immediate write */
+		error = dmu_read(os, ZVOL_OBJ, offset, size, buf,
+		    DMU_READ_NO_PREFETCH);
+	} else {
+		size = zv->zv_volblocksize;
+		offset = P2ALIGN_TYPED(offset, size, uint64_t);
+		error = dmu_buf_hold(os, ZVOL_OBJ, offset, zgd, &db,
+		    DMU_READ_NO_PREFETCH);
+		if (error == 0) {
+			zgd->zgd_db = db;
+			zgd->zgd_bp = &lr->lr_blkptr;
+
+			ASSERT(db != NULL);
+			ASSERT(db->db_offset == offset);
+			ASSERT(db->db_size == size);
+
+			error = dmu_sync(zio, lr->lr_common.lrc_txg,
+			    zvol_get_done, zgd);
+
+			if (error == 0)
+				return (0);
+		}
+	}
+
+	zvol_get_done(zgd, error);
+
+	return (error);
+}
+
 /*
  * The zvol_state_t's are inserted in increasing MINOR(dev_t) order.
  */
@@ -688,27 +747,94 @@ zvol_remove(zvol_state_t *zv_remove)
 	list_remove(&zvol_state_list, zv_remove);
 }
 
+static int
+zvol_first_open(zvol_state_t *zv)
+{
+	objset_t *os;
+	uint64_t volsize;
+	int error;
+	uint64_t readonly;
+
+	/* lie and say we're read-only */
+	error = dmu_objset_own(zv->zv_disk->disk_name,
+	    DMU_OST_ZVOL, B_TRUE, zvol_tag, &os);
+	if (error)
+		return (-error);
+
+	error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
+	if (error) {
+	        dmu_objset_disown(os, zvol_tag);
+	        return (-error);
+	}
+
+	zv->zv_objset = os;
+	error = dmu_bonus_hold(os, ZVOL_OBJ, zvol_tag, &zv->zv_dbuf);
+	if (error) {
+	        dmu_objset_disown(os, zvol_tag);
+	        return (-error);
+	}
+
+	set_capacity(zv->zv_disk, volsize >> 9);
+	zv->zv_volsize = volsize;
+	zv->zv_zilog = zil_open(os, zvol_get_data);
+
+	VERIFY(dsl_prop_get_integer(zv->zv_disk->disk_name,
+	    "readonly", &readonly, NULL) == 0);
+	if (readonly || dmu_objset_is_snapshot(os)) {
+                set_disk_ro(zv->zv_disk, 1);
+	        zv->zv_flags |= ZVOL_RDONLY;
+	} else {
+                set_disk_ro(zv->zv_disk, 0);
+	        zv->zv_flags &= ~ZVOL_RDONLY;
+	}
+
+	return (-error);
+}
+
+static void
+zvol_last_close(zvol_state_t *zv)
+{
+	zil_close(zv->zv_zilog);
+	zv->zv_zilog = NULL;
+	dmu_buf_rele(zv->zv_dbuf, zvol_tag);
+	zv->zv_dbuf = NULL;
+	dmu_objset_disown(zv->zv_objset, zvol_tag);
+	zv->zv_objset = NULL;
+}
+
 static int
 zvol_open(struct block_device *bdev, fmode_t flag)
 {
 	zvol_state_t *zv = bdev->bd_disk->private_data;
+	int error = 0;
 
 	mutex_enter(&zvol_state_lock);
 	ASSERT3P(zv, !=, NULL);
-	ASSERT3P(zv->zv_objset, !=, NULL);
+
+	if (zv->zv_open_count == 0) {
+		error = zvol_first_open(zv);
+		if (error)
+			goto out_mutex;
+	}
 
 	if ((flag & FMODE_WRITE) &&
-	    (get_disk_ro(zv->zv_disk) || (zv->zv_mode & DS_MODE_READONLY))) {
+	    (get_disk_ro(zv->zv_disk) || (zv->zv_flags & ZVOL_RDONLY))) {
-		mutex_exit(&zvol_state_lock);
+		error = -EROFS;
-		return (-EROFS);
+		goto out_open_count;
 	}
 
 	zv->zv_open_count++;
+
+out_open_count:
+	if (zv->zv_open_count == 0)
+		zvol_last_close(zv);
+
+out_mutex:
 	mutex_exit(&zvol_state_lock);
 
 	check_disk_change(bdev);
 
-	return (0);
+	return (error);
 }
 
 static int
@@ -720,88 +846,14 @@ zvol_release(struct gendisk *disk, fmode_t mode)
 	ASSERT3P(zv, !=, NULL);
 	ASSERT3U(zv->zv_open_count, >, 0);
 	zv->zv_open_count--;
+	if (zv->zv_open_count == 0)
+		zvol_last_close(zv);
+
 	mutex_exit(&zvol_state_lock);
 
 	return (0);
 }
 
-static void
-zvol_get_done(dmu_buf_t *db, void *vzgd)
-{
-	zgd_t *zgd = (zgd_t *)vzgd;
-	rl_t *rl = zgd->zgd_rl;
-
-	dmu_buf_rele(db, vzgd);
-	zfs_range_unlock(rl);
-	zil_add_block(zgd->zgd_zilog, zgd->zgd_bp);
-	kmem_free(zgd, sizeof (zgd_t));
-}
-
-/*
- * Get data to generate a TX_WRITE intent log record.
- */
-static int
-zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio)
-{
-	zvol_state_t *zv = arg;
-	objset_t *os = zv->zv_objset;
-	dmu_buf_t *db;
-	rl_t *rl;
-	zgd_t *zgd;
-	uint64_t boff;			/* block starting offset */
-	int dlen = lr->lr_length;	/* length of user data */
-	int error;
-
-	ASSERT(zio);
-	ASSERT(dlen != 0);
-
-	/*
-	 * Write records come in two flavors: immediate and indirect.
-	 * For small writes it's cheaper to store the data with the
-	 * log record (immediate); for large writes it's cheaper to
-	 * sync the data and get a pointer to it (indirect) so that
-	 * we don't have to write the data twice.
-	 */
-	if (buf != NULL) /* immediate write */
-		return (dmu_read(os, ZVOL_OBJ, lr->lr_offset, dlen, buf,
-		                 DMU_READ_NO_PREFETCH));
-
-	zgd = (zgd_t *)kmem_alloc(sizeof (zgd_t), KM_SLEEP);
-	zgd->zgd_zilog = zv->zv_zilog;
-	zgd->zgd_bp = &lr->lr_blkptr;
-
-	/*
-	 * Lock the range of the block to ensure that when the data is
-	 * written out and its checksum is being calculated that no other
-	 * thread can change the block.
-	 */
-	boff = P2ALIGN_TYPED(lr->lr_offset, zv->zv_volblocksize, uint64_t);
-	rl = zfs_range_lock(&zv->zv_znode, boff, zv->zv_volblocksize,
-	    RL_READER);
-	zgd->zgd_rl = rl;
-
-	VERIFY3S(dmu_buf_hold(os, ZVOL_OBJ, lr->lr_offset, zgd, &db), ==, 0);
-	error = dmu_sync(zio, db, &lr->lr_blkptr,
-	                 lr->lr_common.lrc_txg, zvol_get_done, zgd);
-	if (error == 0)
-		zil_add_block(zv->zv_zilog, &lr->lr_blkptr);
-	/*
-	 * If we get EINPROGRESS, then we need to wait for a
-	 * write IO initiated by dmu_sync() to complete before
-	 * we can release this dbuf.  We will finish everything
-	 * up in the zvol_get_done() callback.
-	 */
-	if (error == EINPROGRESS)
-		return (0);
-
-	dmu_buf_rele(db, zgd);
-	zfs_range_unlock(rl);
-	kmem_free(zgd, sizeof (zgd_t));
-
-	return (error);
-}
-
-
 static int
 zvol_ioctl(struct block_device *bdev, fmode_t mode,
            unsigned int cmd, unsigned long arg)
@@ -1026,9 +1078,7 @@ zvol_create_minor(const char *name)
 	zvol_state_t *zv;
 	objset_t *os;
 	dmu_object_info_t doi;
-	uint64_t volsize;
 	unsigned minor = 0;
-	int ds_mode = DS_MODE_OWNER;
 	int error = 0;
 
 	mutex_enter(&zvol_state_lock);
@@ -1039,57 +1089,40 @@ zvol_create_minor(const char *name)
 		goto out;
 	}
 
-	/* Snapshot may only be read-only */
+	error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, zvol_tag, &os);
-	if (strchr(name, '@') != 0)
-		ds_mode |= DS_MODE_READONLY;
-
-	error = dmu_objset_open(name, DMU_OST_ZVOL, ds_mode, &os);
 	if (error)
 		goto out;
 
 	error = dmu_object_info(os, ZVOL_OBJ, &doi);
 	if (error)
-		goto out_dmu_objset_close;
+		goto out_dmu_objset_disown;
-
-	error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
-	if (error)
-		goto out_dmu_objset_close;
 
 	error = zvol_find_minor(&minor);
 	if (error)
-		goto out_dmu_objset_close;
+		goto out_dmu_objset_disown;
 
 	zv = zvol_alloc(MKDEV(zvol_major, minor), name);
 	if (zv == NULL) {
 		error = EAGAIN;
-		goto out_dmu_objset_close;
+		goto out_dmu_objset_disown;
 	}
 
-	set_disk_ro(zv->zv_disk, !!(ds_mode & DS_MODE_READONLY));
+	if (dmu_objset_is_snapshot(os))
-	set_capacity(zv->zv_disk, volsize >> 9);
+		zv->zv_flags |= ZVOL_RDONLY;
 
-	zv->zv_volsize = volsize;
 	zv->zv_volblocksize = doi.doi_data_block_size;
-	zv->zv_objset = os;
+
-	zv->zv_mode = ds_mode;
+	if (zil_replay_disable)
-	zv->zv_zilog = zil_open(os, zvol_get_data);
+		zil_destroy(dmu_objset_zil(os), B_FALSE);
+	else
 		zil_replay(os, zv, zvol_replay_vector);
 
-        error = dsl_prop_register(dmu_objset_ds(zv->zv_objset), "readonly",
-				  zvol_readonly_changed_cb, zv);
-	if (error)
-		goto out_zvol_alloc;
-
 	zvol_insert(zv);
-	mutex_exit(&zvol_state_lock);
 	add_disk(zv->zv_disk);
+	error = 0;
 
-	return 0;
+out_dmu_objset_disown:
-
+	dmu_objset_disown(os, zvol_tag);
-out_zvol_alloc:
-	zvol_free(zv);
-out_dmu_objset_close:
-	dmu_objset_close(os);
 out:
 	mutex_exit(&zvol_state_lock);
 
@@ -1137,14 +1170,6 @@ zvol_remove_minor(const char *name)
 		goto out;
 	}
 
-	error = dsl_prop_unregister(dmu_objset_ds(zv->zv_objset),
-	    "readonly", zvol_readonly_changed_cb, zv);
-	if (error)
-		goto out;
-
-	zil_close(zv->zv_zilog);
-	dmu_objset_close(zv->zv_objset);
-
 	zvol_remove(zv);
 	zvol_free(zv);
 out:
@@ -1158,20 +1183,14 @@ out:
  * zvol_fini() which means the module reference count must have
  * dropped to zero and none of the zvol devices may be open.
  */
-static void
+void
-zvol_remove_minors(void)
+zvol_remove_minors(const char *name)
 {
 	zvol_state_t *zv;
 
 	mutex_enter(&zvol_state_lock);
 	while ((zv = list_head(&zvol_state_list)) != NULL) {
 		ASSERT3U(zv->zv_open_count, ==, 0);
-
-		(void)dsl_prop_unregister(dmu_objset_ds(zv->zv_objset),
-		    "readonly", zvol_readonly_changed_cb, zv);
-		zil_close(zv->zv_zilog);
-		dmu_objset_close(zv->zv_objset);
-
 		zvol_remove(zv);
 		zvol_free(zv);
 	}
@@ -1215,7 +1234,6 @@ zvol_init(void)
 void
 zvol_fini(void)
 {
-	zvol_remove_minors();
 	blk_unregister_region(MKDEV(zvol_major, 0), 1UL << MINORBITS);
 	unregister_blkdev(zvol_major, ZVOL_DRIVER);
 	taskq_destroy(zvol_taskq);