Mercurial > illumos > illumos-gate

changeset 9234:bffdc4fc05c4

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
6792139 recovering from a suspended pool needs some work 6794830 reboot command hangs on a failed zfs pool
author George Wilson <George.Wilson@Sun.COM>
date Mon, 30 Mar 2009 20:01:04 -0700
parents 0de2d511cbe7
children 2e98898d0a13
files usr/src/cmd/ztest/ztest.c usr/src/lib/libzfs/common/libzfs_util.c usr/src/uts/common/fs/zfs/spa.c usr/src/uts/common/fs/zfs/spa_misc.c usr/src/uts/common/fs/zfs/sys/spa_impl.h usr/src/uts/common/fs/zfs/sys/zio.h usr/src/uts/common/fs/zfs/zfs_ioctl.c usr/src/uts/common/fs/zfs/zfs_vfsops.c usr/src/uts/common/fs/zfs/zio.c
diffstat 9 files changed, 220 insertions(+), 108 deletions(-) [+]
line wrap: on
line diff
--- a/usr/src/cmd/ztest/ztest.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/cmd/ztest/ztest.c	Mon Mar 30 20:01:04 2009 -0700
@@ -3082,7 +3082,7 @@
 		spa_vdev_state_enter(spa);
 		vdev_clear(spa, NULL);
 		(void) spa_vdev_state_exit(spa, NULL, 0);
-		zio_resume(spa);
+		(void) zio_resume(spa);
 	}
 }
 
--- a/usr/src/lib/libzfs/common/libzfs_util.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/lib/libzfs/common/libzfs_util.c	Mon Mar 30 20:01:04 2009 -0700
@@ -364,6 +364,11 @@
 	case ENOTSUP:
 		zfs_verror(hdl, EZFS_BADVERSION, fmt, ap);
 		break;
+	case EAGAIN:
+		zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
+		    "pool I/O is currently suspended"));
+		zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
+		break;
 	default:
 		zfs_error_aux(hdl, strerror(errno));
 		zfs_verror(hdl, EZFS_UNKNOWN, fmt, ap);
@@ -437,6 +442,11 @@
 	case EDQUOT:
 		zfs_verror(hdl, EZFS_NOSPC, fmt, ap);
 		return (-1);
+	case EAGAIN:
+		zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
+		    "pool I/O is currently suspended"));
+		zfs_verror(hdl, EZFS_POOLUNAVAIL, fmt, ap);
+		break;
 
 	default:
 		zfs_error_aux(hdl, strerror(error));
--- a/usr/src/uts/common/fs/zfs/spa.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/spa.c	Mon Mar 30 20:01:04 2009 -0700
@@ -683,10 +683,10 @@
 	/*
 	 * Wait for any outstanding async I/O to complete.
 	 */
-	mutex_enter(&spa->spa_async_root_lock);
-	while (spa->spa_async_root_count != 0)
-		cv_wait(&spa->spa_async_root_cv, &spa->spa_async_root_lock);
-	mutex_exit(&spa->spa_async_root_lock);
+	if (spa->spa_async_zio_root != NULL) {
+		(void) zio_wait(spa->spa_async_zio_root);
+		spa->spa_async_zio_root = NULL;
+	}
 
 	/*
 	 * Close the dsl pool.
@@ -1107,6 +1107,14 @@
 	spa->spa_load_guid = pool_guid;
 
 	/*
+	 * Create "The Godfather" zio to hold all async IOs
+	 */
+	if (spa->spa_async_zio_root == NULL)
+		spa->spa_async_zio_root = zio_root(spa, NULL, NULL,
+		    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
+		    ZIO_FLAG_GODFATHER);
+
+	/*
 	 * Parse the configuration into a vdev tree.  We explicitly set the
 	 * value that will be returned by spa_version() since parsing the
 	 * configuration requires knowing the version number.
@@ -2010,6 +2018,14 @@
 	spa->spa_ubsync = spa->spa_uberblock;
 
 	/*
+	 * Create "The Godfather" zio to hold all async IOs
+	 */
+	if (spa->spa_async_zio_root == NULL)
+		spa->spa_async_zio_root = zio_root(spa, NULL, NULL,
+		    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
+		    ZIO_FLAG_GODFATHER);
+
+	/*
 	 * Create the root vdev.
 	 */
 	spa_config_enter(spa, SCL_ALL, FTAG, RW_WRITER);
--- a/usr/src/uts/common/fs/zfs/spa_misc.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/spa_misc.c	Mon Mar 30 20:01:04 2009 -0700
@@ -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.
  */
 
@@ -425,7 +425,6 @@
 	spa = kmem_zalloc(sizeof (spa_t), KM_SLEEP);
 
 	mutex_init(&spa->spa_async_lock, NULL, MUTEX_DEFAULT, NULL);
-	mutex_init(&spa->spa_async_root_lock, NULL, MUTEX_DEFAULT, NULL);
 	mutex_init(&spa->spa_scrub_lock, NULL, MUTEX_DEFAULT, NULL);
 	mutex_init(&spa->spa_errlog_lock, NULL, MUTEX_DEFAULT, NULL);
 	mutex_init(&spa->spa_errlist_lock, NULL, MUTEX_DEFAULT, NULL);
@@ -434,7 +433,6 @@
 	mutex_init(&spa->spa_props_lock, NULL, MUTEX_DEFAULT, NULL);
 
 	cv_init(&spa->spa_async_cv, NULL, CV_DEFAULT, NULL);
-	cv_init(&spa->spa_async_root_cv, NULL, CV_DEFAULT, NULL);
 	cv_init(&spa->spa_scrub_io_cv, NULL, CV_DEFAULT, NULL);
 	cv_init(&spa->spa_suspend_cv, NULL, CV_DEFAULT, NULL);
 
@@ -508,12 +506,10 @@
 	spa_config_lock_destroy(spa);
 
 	cv_destroy(&spa->spa_async_cv);
-	cv_destroy(&spa->spa_async_root_cv);
 	cv_destroy(&spa->spa_scrub_io_cv);
 	cv_destroy(&spa->spa_suspend_cv);
 
 	mutex_destroy(&spa->spa_async_lock);
-	mutex_destroy(&spa->spa_async_root_lock);
 	mutex_destroy(&spa->spa_scrub_lock);
 	mutex_destroy(&spa->spa_errlog_lock);
 	mutex_destroy(&spa->spa_errlist_lock);
--- a/usr/src/uts/common/fs/zfs/sys/spa_impl.h	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/sys/spa_impl.h	Mon Mar 30 20:01:04 2009 -0700
@@ -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.
  */
 
@@ -141,9 +141,6 @@
 	int		spa_async_suspended;	/* async tasks suspended */
 	kcondvar_t	spa_async_cv;		/* wait for thread_exit() */
 	uint16_t	spa_async_tasks;	/* async task mask */
-	kmutex_t	spa_async_root_lock;	/* protects async root count */
-	uint64_t	spa_async_root_count;	/* number of async root zios */
-	kcondvar_t	spa_async_root_cv;	/* notify when count == 0 */
 	char		*spa_root;		/* alternate root directory */
 	uint64_t	spa_ena;		/* spa-wide ereport ENA */
 	boolean_t	spa_last_open_failed;	/* true if last open faled */
@@ -163,6 +160,7 @@
 	uint64_t	spa_failmode;		/* failure mode for the pool */
 	uint64_t	spa_delegation;		/* delegation on/off */
 	list_t		spa_config_list;	/* previous cache file(s) */
+	zio_t		*spa_async_zio_root;	/* root of all async I/O */
 	zio_t		*spa_suspend_zio_root;	/* root of all suspended I/O */
 	kmutex_t	spa_suspend_lock;	/* protects suspend_zio_root */
 	kcondvar_t	spa_suspend_cv;		/* notification of resume */
--- a/usr/src/uts/common/fs/zfs/sys/zio.h	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/sys/zio.h	Mon Mar 30 20:01:04 2009 -0700
@@ -140,6 +140,7 @@
 #define	ZIO_FLAG_PROBE			0x10000
 #define	ZIO_FLAG_GANG_CHILD		0x20000
 #define	ZIO_FLAG_RAW			0x40000
+#define	ZIO_FLAG_GODFATHER		0x80000
 
 #define	ZIO_FLAG_GANG_INHERIT		\
 	(ZIO_FLAG_CANFAIL |		\
@@ -281,7 +282,6 @@
 	int		io_cmd;
 	uint8_t		io_priority;
 	uint8_t		io_reexecute;
-	uint8_t		io_async_root;
 	uint8_t		io_state[ZIO_WAIT_TYPES];
 	uint64_t	io_txg;
 	spa_t		*io_spa;
@@ -415,7 +415,7 @@
 extern uint8_t zio_compress_select(uint8_t child, uint8_t parent);
 
 extern void zio_suspend(spa_t *spa, zio_t *zio);
-extern void zio_resume(spa_t *spa);
+extern int zio_resume(spa_t *spa);
 extern void zio_resume_wait(spa_t *spa);
 
 /*
--- a/usr/src/uts/common/fs/zfs/zfs_ioctl.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/zfs_ioctl.c	Mon Mar 30 20:01:04 2009 -0700
@@ -79,15 +79,18 @@
 typedef int zfs_ioc_func_t(zfs_cmd_t *);
 typedef int zfs_secpolicy_func_t(zfs_cmd_t *, cred_t *);
 
+typedef enum {
+	NO_NAME,
+	POOL_NAME,
+	DATASET_NAME
+} zfs_ioc_namecheck_t;
+
 typedef struct zfs_ioc_vec {
 	zfs_ioc_func_t		*zvec_func;
 	zfs_secpolicy_func_t	*zvec_secpolicy;
-	enum {
-		NO_NAME,
-		POOL_NAME,
-		DATASET_NAME
-	} zvec_namecheck;
+	zfs_ioc_namecheck_t	zvec_namecheck;
 	boolean_t		zvec_his_log;
+	boolean_t		zvec_pool_check;
 } zfs_ioc_vec_t;
 
 static void clear_props(char *dataset, nvlist_t *props, nvlist_t *newprops);
@@ -2787,11 +2790,12 @@
 	/*
 	 * Resume any suspended I/Os.
 	 */
-	zio_resume(spa);
+	if (zio_resume(spa) != 0)
+		error = EIO;
 
 	spa_close(spa, FTAG);
 
-	return (0);
+	return (error);
 }
 
 /*
@@ -3110,59 +3114,123 @@
  * do the logging of those commands.
  */
 static zfs_ioc_vec_t zfs_ioc_vec[] = {
-	{ zfs_ioc_pool_create, zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_pool_destroy,	zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_pool_import, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_pool_export, zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_pool_configs,	zfs_secpolicy_none, NO_NAME, B_FALSE },
-	{ zfs_ioc_pool_stats, zfs_secpolicy_read, POOL_NAME, B_FALSE },
-	{ zfs_ioc_pool_tryimport, zfs_secpolicy_config, NO_NAME, B_FALSE },
-	{ zfs_ioc_pool_scrub, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE },
-	{ zfs_ioc_pool_upgrade,	zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_pool_get_history, zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_vdev_add, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_vdev_remove, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_vdev_set_state, zfs_secpolicy_config,	POOL_NAME, B_TRUE },
-	{ zfs_ioc_vdev_attach, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_vdev_detach, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_vdev_setpath,	zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_objset_stats,	zfs_secpolicy_read, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_objset_zplprops, zfs_secpolicy_read, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_dataset_list_next, zfs_secpolicy_read,
-	    DATASET_NAME, B_FALSE },
-	{ zfs_ioc_snapshot_list_next, zfs_secpolicy_read,
-	    DATASET_NAME, B_FALSE },
-	{ zfs_ioc_set_prop, zfs_secpolicy_none, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_create_minor,	zfs_secpolicy_minor, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_remove_minor,	zfs_secpolicy_minor, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_create, zfs_secpolicy_create, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_destroy, zfs_secpolicy_destroy, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_rename, zfs_secpolicy_rename,	DATASET_NAME, B_TRUE },
-	{ zfs_ioc_recv, zfs_secpolicy_receive, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_send, zfs_secpolicy_send, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_inject_fault,	zfs_secpolicy_inject, NO_NAME, B_FALSE },
-	{ zfs_ioc_clear_fault, zfs_secpolicy_inject, NO_NAME, B_FALSE },
-	{ zfs_ioc_inject_list_next, zfs_secpolicy_inject, NO_NAME, B_FALSE },
-	{ zfs_ioc_error_log, zfs_secpolicy_inject, POOL_NAME, B_FALSE },
-	{ zfs_ioc_clear, zfs_secpolicy_config, POOL_NAME, B_TRUE },
-	{ zfs_ioc_promote, zfs_secpolicy_promote, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_destroy_snaps, zfs_secpolicy_destroy,	DATASET_NAME, B_TRUE },
-	{ zfs_ioc_snapshot, zfs_secpolicy_snapshot, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_dsobj_to_dsname, zfs_secpolicy_config, POOL_NAME, B_FALSE },
-	{ zfs_ioc_obj_to_path, zfs_secpolicy_config, NO_NAME, B_FALSE },
-	{ zfs_ioc_pool_set_props, zfs_secpolicy_config,	POOL_NAME, B_TRUE },
-	{ zfs_ioc_pool_get_props, zfs_secpolicy_read, POOL_NAME, B_FALSE },
-	{ zfs_ioc_set_fsacl, zfs_secpolicy_fsacl, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_get_fsacl, zfs_secpolicy_read, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_iscsi_perm_check, zfs_secpolicy_iscsi,
-	    DATASET_NAME, B_FALSE },
-	{ zfs_ioc_share, zfs_secpolicy_share, DATASET_NAME, B_FALSE },
-	{ zfs_ioc_inherit_prop, zfs_secpolicy_inherit, DATASET_NAME, B_TRUE },
-	{ zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, DATASET_NAME, B_FALSE }
+	{ zfs_ioc_pool_create, zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_destroy,	zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_import, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_FALSE },
+	{ zfs_ioc_pool_export, zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_configs,	zfs_secpolicy_none, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_stats, zfs_secpolicy_read, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_tryimport, zfs_secpolicy_config, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_scrub, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_pool_freeze, zfs_secpolicy_config, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_upgrade,	zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_pool_get_history, zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_vdev_add, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_vdev_remove, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_vdev_set_state, zfs_secpolicy_config,	POOL_NAME, B_TRUE,
+	    B_FALSE },
+	{ zfs_ioc_vdev_attach, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_vdev_detach, zfs_secpolicy_config, POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_vdev_setpath,	zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_TRUE },
+	{ zfs_ioc_objset_stats,	zfs_secpolicy_read, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_objset_zplprops, zfs_secpolicy_read, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_dataset_list_next, zfs_secpolicy_read, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_snapshot_list_next, zfs_secpolicy_read, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_set_prop, zfs_secpolicy_none, DATASET_NAME, B_TRUE, B_TRUE },
+	{ zfs_ioc_create_minor,	zfs_secpolicy_minor, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_remove_minor,	zfs_secpolicy_minor, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_create, zfs_secpolicy_create, DATASET_NAME, B_TRUE, B_TRUE },
+	{ zfs_ioc_destroy, zfs_secpolicy_destroy, DATASET_NAME, B_TRUE,
+	    B_TRUE},
+	{ zfs_ioc_rollback, zfs_secpolicy_rollback, DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_rename, zfs_secpolicy_rename,	DATASET_NAME, B_TRUE, B_TRUE },
+	{ zfs_ioc_recv, zfs_secpolicy_receive, DATASET_NAME, B_TRUE, B_TRUE },
+	{ zfs_ioc_send, zfs_secpolicy_send, DATASET_NAME, B_TRUE, B_FALSE },
+	{ zfs_ioc_inject_fault,	zfs_secpolicy_inject, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_clear_fault, zfs_secpolicy_inject, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_inject_list_next, zfs_secpolicy_inject, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_error_log, zfs_secpolicy_inject, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_clear, zfs_secpolicy_config, POOL_NAME, B_TRUE, B_FALSE },
+	{ zfs_ioc_promote, zfs_secpolicy_promote, DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_destroy_snaps, zfs_secpolicy_destroy,	DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_snapshot, zfs_secpolicy_snapshot, DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_dsobj_to_dsname, zfs_secpolicy_config, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_obj_to_path, zfs_secpolicy_config, NO_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_pool_set_props, zfs_secpolicy_config,	POOL_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_pool_get_props, zfs_secpolicy_read, POOL_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_set_fsacl, zfs_secpolicy_fsacl, DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_get_fsacl, zfs_secpolicy_read, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_iscsi_perm_check, zfs_secpolicy_iscsi, DATASET_NAME, B_FALSE,
+	    B_FALSE },
+	{ zfs_ioc_share, zfs_secpolicy_share, DATASET_NAME, B_FALSE, B_FALSE },
+	{ zfs_ioc_inherit_prop, zfs_secpolicy_inherit, DATASET_NAME, B_TRUE,
+	    B_TRUE },
+	{ zfs_ioc_smb_acl, zfs_secpolicy_smb_acl, DATASET_NAME, B_FALSE,
+	    B_FALSE }
 };
 
+int
+pool_status_check(const char *name, zfs_ioc_namecheck_t type)
+{
+	spa_t *spa;
+	char pool[ZFS_MAXNAMELEN];
+	int error;
+
+	ASSERT(type == POOL_NAME || type == DATASET_NAME);
+
+	(void) strlcpy(pool, name, ZFS_MAXNAMELEN);
+	if (type == DATASET_NAME) {
+		char *p;
+
+		if ((p = strpbrk(pool, "/@")) != NULL)
+			*p = '\0';
+	}
+
+	error = spa_open(pool, &spa, FTAG);
+	if (error == 0) {
+		if (spa_suspended(spa))
+			error = EAGAIN;
+		spa_close(spa, FTAG);
+	}
+	return (error);
+}
+
 static int
 zfsdev_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp)
 {
@@ -3196,11 +3264,17 @@
 		case POOL_NAME:
 			if (pool_namecheck(zc->zc_name, NULL, NULL) != 0)
 				error = EINVAL;
+			if (zfs_ioc_vec[vec].zvec_pool_check)
+				error = pool_status_check(zc->zc_name,
+				    zfs_ioc_vec[vec].zvec_namecheck);
 			break;
 
 		case DATASET_NAME:
 			if (dataset_namecheck(zc->zc_name, NULL, NULL) != 0)
 				error = EINVAL;
+			if (zfs_ioc_vec[vec].zvec_pool_check)
+				error = pool_status_check(zc->zc_name,
+				    zfs_ioc_vec[vec].zvec_namecheck);
 			break;
 
 		case NO_NAME:
--- a/usr/src/uts/common/fs/zfs/zfs_vfsops.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/zfs_vfsops.c	Mon Mar 30 20:01:04 2009 -0700
@@ -67,6 +67,8 @@
 static minor_t zfs_minor;
 static kmutex_t	zfs_dev_mtx;
 
+extern int sys_shutdown;
+
 static int zfs_mount(vfs_t *vfsp, vnode_t *mvp, struct mounta *uap, cred_t *cr);
 static int zfs_umount(vfs_t *vfsp, int fflag, cred_t *cr);
 static int zfs_mountroot(vfs_t *vfsp, enum whymountroot);
@@ -145,12 +147,24 @@
 		 * Sync a specific filesystem.
 		 */
 		zfsvfs_t *zfsvfs = vfsp->vfs_data;
+		dsl_pool_t *dp;
 
 		ZFS_ENTER(zfsvfs);
+		dp = dmu_objset_pool(zfsvfs->z_os);
+
+		/*
+		 * If the system is shutting down, then skip any
+		 * filesystems which may exist on a suspended pool.
+		 */
+		if (sys_shutdown && spa_suspended(dp->dp_spa)) {
+			ZFS_EXIT(zfsvfs);
+			return (0);
+		}
+
 		if (zfsvfs->z_log != NULL)
 			zil_commit(zfsvfs->z_log, UINT64_MAX, 0);
 		else
-			txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), 0);
+			txg_wait_synced(dp, 0);
 		ZFS_EXIT(zfsvfs);
 	} else {
 		/*
--- a/usr/src/uts/common/fs/zfs/zio.c	Mon Mar 30 18:41:59 2009 -0700
+++ b/usr/src/uts/common/fs/zfs/zio.c	Mon Mar 30 20:01:04 2009 -0700
@@ -529,21 +529,11 @@
 static void
 zio_destroy(zio_t *zio)
 {
-	spa_t *spa = zio->io_spa;
-	uint8_t async_root = zio->io_async_root;
-
 	list_destroy(&zio->io_parent_list);
 	list_destroy(&zio->io_child_list);
 	mutex_destroy(&zio->io_lock);
 	cv_destroy(&zio->io_cv);
 	kmem_cache_free(zio_cache, zio);
-
-	if (async_root) {
-		mutex_enter(&spa->spa_async_root_lock);
-		if (--spa->spa_async_root_count == 0)
-			cv_broadcast(&spa->spa_async_root_cv);
-		mutex_exit(&spa->spa_async_root_lock);
-	}
 }
 
 zio_t *
@@ -1092,14 +1082,12 @@
 	    zio_unique_parent(zio) == NULL) {
 		/*
 		 * This is a logical async I/O with no parent to wait for it.
-		 * Track how many outstanding I/Os of this type exist so
-		 * that spa_unload() knows when they are all done.
+		 * We add it to the spa_async_root_zio "Godfather" I/O which
+		 * will ensure they complete prior to unloading the pool.
 		 */
 		spa_t *spa = zio->io_spa;
-		zio->io_async_root = B_TRUE;
-		mutex_enter(&spa->spa_async_root_lock);
-		spa->spa_async_root_count++;
-		mutex_exit(&spa->spa_async_root_lock);
+
+		zio_add_child(spa->spa_async_zio_root, zio);
 	}
 
 	zio_execute(zio);
@@ -1161,8 +1149,11 @@
 
 	/*
 	 * Now that all children have been reexecuted, execute the parent.
+	 * We don't reexecute "The Godfather" I/O here as it's the
+	 * responsibility of the caller to wait on him.
 	 */
-	zio_execute(pio);
+	if (!(pio->io_flags & ZIO_FLAG_GODFATHER))
+		zio_execute(pio);
 }
 
 void
@@ -1178,11 +1169,14 @@
 	mutex_enter(&spa->spa_suspend_lock);
 
 	if (spa->spa_suspend_zio_root == NULL)
-		spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL, 0);
+		spa->spa_suspend_zio_root = zio_root(spa, NULL, NULL,
+		    ZIO_FLAG_CANFAIL | ZIO_FLAG_SPECULATIVE |
+		    ZIO_FLAG_GODFATHER);
 
 	spa->spa_suspended = B_TRUE;
 
 	if (zio != NULL) {
+		ASSERT(!(zio->io_flags & ZIO_FLAG_GODFATHER));
 		ASSERT(zio != spa->spa_suspend_zio_root);
 		ASSERT(zio->io_child_type == ZIO_CHILD_LOGICAL);
 		ASSERT(zio_unique_parent(zio) == NULL);
@@ -1193,10 +1187,10 @@
 	mutex_exit(&spa->spa_suspend_lock);
 }
 
-void
+int
 zio_resume(spa_t *spa)
 {
-	zio_t *pio, *cio, *cio_next;
+	zio_t *pio;
 
 	/*
 	 * Reexecute all previously suspended i/o.
@@ -1209,18 +1203,10 @@
 	mutex_exit(&spa->spa_suspend_lock);
 
 	if (pio == NULL)
-		return;
+		return (0);
 
-	for (cio = zio_walk_children(pio); cio != NULL; cio = cio_next) {
-		zio_link_t *zl = pio->io_walk_link;
-		cio_next = zio_walk_children(pio);
-		zio_remove_child(pio, cio, zl);
-		zio_reexecute(cio);
-	}
-
-	ASSERT(pio->io_children[ZIO_CHILD_LOGICAL][ZIO_WAIT_DONE] == 0);
-
-	(void) zio_wait(pio);
+	zio_reexecute(pio);
+	return (zio_wait(pio));
 }
 
 void
@@ -2219,7 +2205,7 @@
 	 */
 	zio_inherit_child_errors(zio, ZIO_CHILD_LOGICAL);
 
-	if (zio->io_reexecute) {
+	if (zio->io_reexecute && !(zio->io_flags & ZIO_FLAG_GODFATHER)) {
 		/*
 		 * This is a logical I/O that wants to reexecute.
 		 *
@@ -2244,6 +2230,24 @@
 		zio->io_state[ZIO_WAIT_DONE] = 1;
 		mutex_exit(&zio->io_lock);
 
+		/*
+		 * "The Godfather" I/O monitors its children but is
+		 * not a true parent to them. It will track them through
+		 * the pipeline but severs its ties whenever they get into
+		 * trouble (e.g. suspended). This allows "The Godfather"
+		 * I/O to return status without blocking.
+		 */
+		for (pio = zio_walk_parents(zio); pio != NULL; pio = pio_next) {
+			zio_link_t *zl = zio->io_walk_link;
+			pio_next = zio_walk_parents(zio);
+
+			if ((pio->io_flags & ZIO_FLAG_GODFATHER) &&
+			    (zio->io_reexecute & ZIO_REEXECUTE_SUSPEND)) {
+				zio_remove_child(pio, zio, zl);
+				zio_notify_parent(pio, zio, ZIO_WAIT_DONE);
+			}
+		}
+
 		if ((pio = zio_unique_parent(zio)) != NULL) {
 			/*
 			 * We're not a root i/o, so there's nothing to do
@@ -2271,7 +2275,7 @@
 	}
 
 	ASSERT(zio_walk_children(zio) == NULL);
-	ASSERT(zio->io_reexecute == 0);
+	ASSERT(zio->io_reexecute == 0 || (zio->io_flags & ZIO_FLAG_GODFATHER));
 	ASSERT(zio->io_error == 0 || (zio->io_flags & ZIO_FLAG_CANFAIL));
 
 	/*