Rebase master to b121

module/zfs/include/sys/dmu.h

@@ -117,6 +117,7 @@ typedef enum dmu_object_type {
 	DMU_OT_SCRUB_QUEUE,		/* ZAP */
 	DMU_OT_USERGROUP_USED,		/* ZAP */
 	DMU_OT_USERGROUP_QUOTA,		/* ZAP */
+	DMU_OT_USERREFS,		/* ZAP */
 	DMU_OT_NUMTYPES
 } dmu_object_type_t;
 
@@ -174,8 +175,8 @@ int dmu_objset_evict_dbufs(objset_t *os);
 int dmu_objset_create(const char *name, dmu_objset_type_t type,
     objset_t *clone_parent, uint64_t flags,
     void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg);
-int dmu_objset_destroy(const char *name);
-int dmu_snapshots_destroy(char *fsname, char *snapname);
+int dmu_objset_destroy(const char *name, boolean_t defer);
+int dmu_snapshots_destroy(char *fsname, char *snapname, boolean_t defer);
 int dmu_objset_rollback(objset_t *os);
 int dmu_objset_snapshot(char *fsname, char *snapname, struct nvlist *props,
     boolean_t recursive);
@@ -646,10 +647,9 @@ typedef struct dmu_recv_cookie {
 } dmu_recv_cookie_t;
 
 int dmu_recv_begin(char *tofs, char *tosnap, struct drr_begin *,
-    boolean_t force, objset_t *origin, boolean_t online, dmu_recv_cookie_t *);
+    boolean_t force, objset_t *origin, dmu_recv_cookie_t *);
 int dmu_recv_stream(dmu_recv_cookie_t *drc, struct vnode *vp, offset_t *voffp);
 int dmu_recv_end(dmu_recv_cookie_t *drc);
-void dmu_recv_abort_cleanup(dmu_recv_cookie_t *drc);
 
 /* CRC64 table */
 #define	ZFS_CRC64_POLY	0xC96C5795D7870F42ULL	/* ECMA-182, reflected form */

module/zfs/include/sys/dmu_impl.h

@@ -19,7 +19,7 @@
  * CDDL HEADER END
  */
 /*
- * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
+ * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
  */
 
@@ -211,10 +211,11 @@ extern "C" {
  * ds_lock
  *    protects:
  *    	ds_user_ptr
- *    	ds_user_evice_func
+ *    	ds_user_evict_func
  *    	ds_open_refcount
  *    	ds_snapname
  *    	ds_phys accounting
+ *	ds_phys userrefs zapobj
  *	ds_reserved
  *    held from:
  *    	dsl_dataset_*

module/zfs/include/sys/dmu_objset.h

@@ -117,7 +117,7 @@ void dmu_objset_close(objset_t *os);
 int dmu_objset_create(const char *name, dmu_objset_type_t type,
     objset_t *clone_parent, uint64_t flags,
     void (*func)(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx), void *arg);
-int dmu_objset_destroy(const char *name);
+int dmu_objset_destroy(const char *name, boolean_t defer);
 int dmu_objset_rollback(objset_t *os);
 int dmu_objset_snapshot(char *fsname, char *snapname, nvlist_t *props,
     boolean_t recursive);

module/zfs/include/sys/dsl_dataset.h

@@ -62,6 +62,14 @@ typedef void dsl_dataset_evict_func_t(struct dsl_dataset *, void *);
  */
 #define	DS_FLAG_UNIQUE_ACCURATE	(1ULL<<2)
 
+/*
+ * DS_FLAG_DEFER_DESTROY is set after 'zfs destroy -d' has been called
+ * on a dataset. This allows the dataset to be destroyed using 'zfs release'.
+ */
+#define	DS_FLAG_DEFER_DESTROY	(1ULL<<3)
+#define	DS_IS_DEFER_DESTROY(ds)	\
+	((ds)->ds_phys->ds_flags & DS_FLAG_DEFER_DESTROY)
+
 /*
  * DS_FLAG_CI_DATASET is set if the dataset contains a file system whose
  * name lookups should be performed case-insensitively.
@@ -93,7 +101,8 @@ typedef struct dsl_dataset_phys {
 	blkptr_t ds_bp;
 	uint64_t ds_next_clones_obj;	/* DMU_OT_DSL_CLONES */
 	uint64_t ds_props_obj;		/* DMU_OT_DSL_PROPS for snaps */
-	uint64_t ds_pad[6]; /* pad out to 320 bytes for good measure */
+	uint64_t ds_userrefs_obj;	/* DMU_OT_USERREFS */
+	uint64_t ds_pad[5]; /* pad out to 320 bytes for good measure */
 } dsl_dataset_phys_t;
 
 typedef struct dsl_dataset {
@@ -111,6 +120,9 @@ typedef struct dsl_dataset {
 	/* has internal locking: */
 	bplist_t ds_deadlist;
 
+	/* to protect against multiple concurrent incremental recv */
+	kmutex_t ds_recvlock;
+
 	/* protected by lock on pool's dp_dirty_datasets list */
 	txg_node_t ds_dirty_link;
 	list_node_t ds_synced_link;
@@ -122,6 +134,7 @@ typedef struct dsl_dataset {
 	kmutex_t ds_lock;
 	void *ds_user_ptr;
 	dsl_dataset_evict_func_t *ds_user_evict_func;
+	uint64_t ds_userrefs;
 
 	/*
 	 * ds_owner is protected by the ds_rwlock and the ds_lock
@@ -143,6 +156,15 @@ typedef struct dsl_dataset {
 	char ds_snapname[MAXNAMELEN];
 } dsl_dataset_t;
 
+struct dsl_ds_destroyarg {
+	dsl_dataset_t *ds;		/* ds to destroy */
+	dsl_dataset_t *rm_origin;	/* also remove our origin? */
+	boolean_t is_origin_rm;		/* set if removing origin snap */
+	boolean_t defer;		/* destroy -d requested? */
+	boolean_t releasing;		/* destroying due to release? */
+	boolean_t need_prep;		/* do we need to retry due to EBUSY? */
+};
+
 #define	dsl_dataset_is_snapshot(ds)	\
 	((ds)->ds_phys->ds_num_children != 0)
 
@@ -167,8 +189,8 @@ uint64_t dsl_dataset_create_sync(dsl_dir_t *pds, const char *lastname,
     dsl_dataset_t *origin, uint64_t flags, cred_t *, dmu_tx_t *);
 uint64_t dsl_dataset_create_sync_dd(dsl_dir_t *dd, dsl_dataset_t *origin,
     uint64_t flags, dmu_tx_t *tx);
-int dsl_dataset_destroy(dsl_dataset_t *ds, void *tag);
-int dsl_snapshots_destroy(char *fsname, char *snapname);
+int dsl_dataset_destroy(dsl_dataset_t *ds, void *tag, boolean_t defer);
+int dsl_snapshots_destroy(char *fsname, char *snapname, boolean_t defer);
 dsl_checkfunc_t dsl_dataset_destroy_check;
 dsl_syncfunc_t dsl_dataset_destroy_sync;
 dsl_checkfunc_t dsl_dataset_snapshot_check;
@@ -178,6 +200,11 @@ int dsl_dataset_rename(char *name, const char *newname, boolean_t recursive);
 int dsl_dataset_promote(const char *name);
 int dsl_dataset_clone_swap(dsl_dataset_t *clone, dsl_dataset_t *origin_head,
     boolean_t force);
+int dsl_dataset_user_hold(char *dsname, char *snapname, char *htag,
+    boolean_t recursive);
+int dsl_dataset_user_release(char *dsname, char *snapname, char *htag,
+    boolean_t recursive);
+int dsl_dataset_get_holds(const char *dsname, nvlist_t **nvp);
 
 void *dsl_dataset_set_user_ptr(dsl_dataset_t *ds,
     void *p, dsl_dataset_evict_func_t func);

module/zfs/include/sys/dsl_deleg.h

@@ -53,6 +53,8 @@ extern "C" {
 #define	ZFS_DELEG_PERM_GROUPQUOTA	"groupquota"
 #define	ZFS_DELEG_PERM_USERUSED		"userused"
 #define	ZFS_DELEG_PERM_GROUPUSED	"groupused"
+#define	ZFS_DELEG_PERM_HOLD		"hold"
+#define	ZFS_DELEG_PERM_RELEASE		"release"
 
 /*
  * Note: the names of properties that are marked delegatable are also

module/zfs/include/sys/dsl_prop.h

@@ -69,7 +69,7 @@ dsl_syncfunc_t dsl_props_set_sync;
 int dsl_prop_set(const char *ddname, const char *propname,
     int intsz, int numints, const void *buf);
 int dsl_props_set(const char *dsname, nvlist_t *nvl);
-void dsl_prop_set_uint64_sync(dsl_dir_t *dd, const char *name, uint64_t val,
+void dsl_dir_prop_set_uint64_sync(dsl_dir_t *dd, const char *name, uint64_t val,
     cred_t *cr, dmu_tx_t *tx);
 
 void dsl_prop_nvlist_add_uint64(nvlist_t *nv, zfs_prop_t prop, uint64_t value);

module/zfs/include/sys/spa.h

@@ -500,8 +500,9 @@ extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
     char *his_buf);
 extern int spa_history_log(spa_t *spa, const char *his_buf,
     history_log_type_t what);
-void spa_history_internal_log(history_internal_events_t event, spa_t *spa,
-    dmu_tx_t *tx, cred_t *cr, const char *fmt, ...);
+extern void spa_history_internal_log(history_internal_events_t event,
+    spa_t *spa, dmu_tx_t *tx, cred_t *cr, const char *fmt, ...);
+extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt);
 
 /* error handling */
 struct zbookmark;

module/zfs/include/sys/spa_impl.h

@@ -105,6 +105,7 @@ struct spa {
 	int		spa_inject_ref;		/* injection references */
 	uint8_t		spa_sync_on;		/* sync threads are running */
 	spa_load_state_t spa_load_state;	/* current load operation */
+	boolean_t	spa_load_verbatim;	/* load the given config? */
 	taskq_t		*spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
 	dsl_pool_t	*spa_dsl_pool;
 	metaslab_class_t *spa_normal_class;	/* normal data class */

module/zfs/include/sys/vdev.h

@@ -47,6 +47,7 @@ typedef enum vdev_dtl_type {
 extern boolean_t zfs_nocacheflush;
 
 extern int vdev_open(vdev_t *);
+extern void vdev_open_children(vdev_t *vd);
 extern int vdev_validate(vdev_t *);
 extern void vdev_close(vdev_t *);
 extern int vdev_create(vdev_t *, uint64_t txg, boolean_t isreplace);

module/zfs/include/sys/vdev_impl.h

@@ -127,6 +127,8 @@ struct vdev {
 	space_map_t	vdev_dtl[DTL_TYPES]; /* in-core dirty time logs	*/
 	vdev_stat_t	vdev_stat;	/* virtual device statistics	*/
 	boolean_t	vdev_expanding;	/* expand the vdev?		*/
+	int		vdev_open_error; /* error on last open		*/
+	kthread_t	*vdev_open_thread; /* thread opening children	*/
 
 	/*
 	 * Top-level vdev state.

module/zfs/include/sys/zap.h

@@ -182,8 +182,7 @@ int zap_lookup_norm(objset_t *ds, uint64_t zapobj, const char *name,
     boolean_t *normalization_conflictp);
 
 int zap_count_write(objset_t *os, uint64_t zapobj, const char *name,
-    int add, uint64_t *towrite, uint64_t *tooverwrite,
-    uint64_t dn_datablkshift);
+    int add, uint64_t *towrite, uint64_t *tooverwrite);
 
 /*
  * Create an attribute with the given name and value.

module/zfs/include/sys/zfs_acl.h

@@ -203,6 +203,7 @@ void zfs_oldace_byteswap(ace_t *, int);
 void zfs_ace_byteswap(void *, size_t, boolean_t);
 extern boolean_t zfs_has_access(struct znode *zp, cred_t *cr);
 extern int zfs_zaccess(struct znode *, int, int, boolean_t, cred_t *);
+int zfs_fastaccesschk_execute(struct znode *, cred_t *);
 extern int zfs_zaccess_rwx(struct znode *, mode_t, int, cred_t *);
 extern int zfs_zaccess_unix(struct znode *, mode_t, cred_t *);
 extern int zfs_acl_access(struct znode *, int, cred_t *);

module/zfs/include/sys/zfs_ioctl.h

@@ -165,6 +165,7 @@ typedef struct zfs_cmd {
 	dmu_objset_stats_t zc_objset_stats;
 	struct drr_begin zc_begin_record;
 	zinject_record_t zc_inject_record;
+	boolean_t	zc_defer_destroy;
 } zfs_cmd_t;
 
 typedef struct zfs_useracct {

module/zfs/include/sys/zfs_vfsops.h

@@ -73,7 +73,6 @@ struct zfsvfs {
 	boolean_t	z_vscan;	/* virus scan on/off */
 	boolean_t	z_use_fuids;	/* version allows fuids */
 	boolean_t	z_replay;	/* set during ZIL replay */
-	kmutex_t	z_online_recv_lock; /* held while recv in progress */
 	uint64_t	z_version;	/* ZPL version */
 	uint64_t	z_shares_dir;	/* hidden shares dir */
 	kmutex_t	z_lock;

module/zfs/include/sys/zfs_znode.h

@@ -77,6 +77,7 @@ extern "C" {
 #define	ZFS_ACL_DEFAULTED	0x20		/* ACL should be defaulted */
 #define	ZFS_ACL_AUTO_INHERIT	0x40		/* ACL should be inherited */
 #define	ZFS_BONUS_SCANSTAMP	0x80		/* Scanstamp in bonus area */
+#define	ZFS_NO_EXECS_DENIED	0x100		/* exec was given to everyone */
 
 /*
  * Is ID ephemeral?
@@ -200,6 +201,7 @@ typedef struct znode {
 	uint64_t	z_gen;		/* generation (same as zp_gen) */
 	uint32_t	z_sync_cnt;	/* synchronous open count */
 	kmutex_t	z_acl_lock;	/* acl data lock */
+	zfs_acl_t	*z_acl_cached;	/* cached acl */
 	list_node_t	z_link_node;	/* all znodes in fs link */
 	/*
 	 * These are dmu managed fields.

module/zfs/include/sys/zio.h

@@ -143,6 +143,8 @@ enum zio_compress {
 #define	ZIO_FLAG_GODFATHER		0x080000
 
 #define	ZIO_FLAG_TRYHARD		0x100000
+#define	ZIO_FLAG_NODATA			0x200000
+#define	ZIO_FLAG_OPTIONAL		0x400000
 
 #define	ZIO_FLAG_GANG_INHERIT		\
 	(ZIO_FLAG_CANFAIL |		\
@@ -161,7 +163,9 @@ enum zio_compress {
 	ZIO_FLAG_IO_REPAIR |		\
 	ZIO_FLAG_IO_RETRY |		\
 	ZIO_FLAG_PROBE |		\
-	ZIO_FLAG_TRYHARD)
+	ZIO_FLAG_TRYHARD |		\
+	ZIO_FLAG_NODATA |		\
+	ZIO_FLAG_OPTIONAL)
 
 #define	ZIO_FLAG_AGG_INHERIT		\
 	(ZIO_FLAG_DONT_AGGREGATE |	\