1701 ZFS to support UNMAP/TRIM for SSD

Portions contributed by: Jerry Jelinek <jerry.jelinek@joyent.com>
Portions contributed by: George Wilson <george.wilson@delphix.com>
Portions contributed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Portions contributed by: Tim Chase <tim@chase2k.com>
Portions contributed by: Chunwei Chen <tuxoko@gmail.com>
Reviewed by: Matt Ahrens <mahrens@delphix.com>
Reviewed by: loli10K <ezomori.nozomu@gmail.com>
Reviewed by: Tim Chase <tim@chase2k.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed by: Tom Caputi <tcaputi@datto.com>
Reviewed by: Jorgen Lundman <lundman@lundman.net>
Reviewed by: Serapheim Dimitropoulos <serapheim@delphix.com>
Reviewed by: C Fraire <cfraire@me.com>
Reviewed by: Kody Kantor <kody.kantor@joyent.com>
Approved by: Dan McDonald <danmcd@joyent.com>

4 files changed, 199 insertions, 43 deletions

diff --git a/usr/src/lib/libzfs/common/libzfs.h b/usr/src/lib/libzfs/common/libzfs.h
index b23995b1f9..36b70c11fb 100644
--- a/usr/src/lib/libzfs/common/libzfs.h
+++ b/usr/src/lib/libzfs/common/libzfs.h
@@ -131,6 +131,7 @@ typedef enum zfs_error {
 	EZFS_POOLREADONLY,	/* pool is in read-only mode */
 	EZFS_SCRUB_PAUSED,	/* scrub currently paused */
 	EZFS_ACTIVE_POOL,	/* pool is imported on a different system */
+	EZFS_CRYPTOFAILED,	/* failed to setup encryption */
 	EZFS_NO_PENDING,	/* cannot cancel, no operation is pending */
 	EZFS_CHECKPOINT_EXISTS,	/* checkpoint exists */
 	EZFS_DISCARDING_CHECKPOINT,	/* currently discarding a checkpoint */
@@ -141,8 +142,10 @@ typedef enum zfs_error {
 	EZFS_INITIALIZING,	/* currently initializing */
 	EZFS_NO_INITIALIZE,	/* no active initialize */
 	EZFS_WRONG_PARENT,	/* invalid parent dataset (e.g ZVOL) */
+	EZFS_TRIMMING,		/* currently trimming */
+	EZFS_NO_TRIM,		/* no active trim */
+	EZFS_TRIM_NOTSUP,	/* device does not support trim */
 	EZFS_NO_RESILVER_DEFER,	/* pool doesn't support resilver_defer */
-	EZFS_CRYPTOFAILED,	/* failed to setup encryption */
 	EZFS_UNKNOWN
 } zfs_error_t;
 
@@ -264,12 +267,26 @@ typedef struct splitflags {
 	int name_flags;
 } splitflags_t;
 
+typedef struct trimflags {
+	/* requested vdevs are for the entire pool */
+	boolean_t fullpool;
+
+	/* request a secure trim, requires support from device */
+	boolean_t secure;
+
+	/* trim at the requested rate in bytes/second */
+	uint64_t rate;
+} trimflags_t;
+
 /*
  * 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_trim(zpool_handle_t *, pool_trim_func_t, nvlist_t *,
+    trimflags_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/usr/src/lib/libzfs/common/libzfs_pool.c b/usr/src/lib/libzfs/common/libzfs_pool.c
index e4b3e6b72f..2632eab321 100644
--- a/usr/src/lib/libzfs/common/libzfs_pool.c
+++ b/usr/src/lib/libzfs/common/libzfs_pool.c
@@ -2029,6 +2029,57 @@ zpool_import_props(libzfs_handle_t *hdl, nvlist_t *config, const char *newname,
 }
 
 /*
+ * Translate vdev names to guids.  If a vdev_path is determined to be
+ * unsuitable then a vd_errlist is allocated and the vdev path and errno
+ * are added to it.
+ */
+static int
+zpool_translate_vdev_guids(zpool_handle_t *zhp, nvlist_t *vds,
+    nvlist_t *vdev_guids, nvlist_t *guids_to_paths, nvlist_t **vd_errlist)
+{
+	nvlist_t *errlist = NULL;
+	int error = 0;
+
+	for (nvpair_t *elem = nvlist_next_nvpair(vds, NULL); elem != NULL;
+	    elem = nvlist_next_nvpair(vds, elem)) {
+		boolean_t spare, cache;
+
+		char *vd_path = nvpair_name(elem);
+		nvlist_t *tgt = zpool_find_vdev(zhp, vd_path, &spare, &cache,
+		    NULL);
+
+		if ((tgt == NULL) || cache || spare) {
+			if (errlist == NULL) {
+				errlist = fnvlist_alloc();
+				error = EINVAL;
+			}
+
+			uint64_t err = (tgt == NULL) ? EZFS_NODEVICE :
+			    (spare ? EZFS_ISSPARE : EZFS_ISL2CACHE);
+			fnvlist_add_int64(errlist, vd_path, err);
+			continue;
+		}
+
+		uint64_t guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID);
+		fnvlist_add_uint64(vdev_guids, vd_path, guid);
+
+		char msg[MAXNAMELEN];
+		(void) snprintf(msg, sizeof (msg), "%llu", (u_longlong_t)guid);
+		fnvlist_add_string(guids_to_paths, msg, vd_path);
+	}
+
+	if (error != 0) {
+		verify(errlist != NULL);
+		if (vd_errlist != NULL)
+			*vd_errlist = errlist;
+		else
+			fnvlist_free(errlist);
+	}
+
+	return (error);
+}
+
+/*
  * Scan the pool.
  */
 int
@@ -2129,72 +2180,151 @@ 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;
+	int err;
 
-	/* 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;
+	nvlist_t *vd_errlist = NULL;
+	nvlist_t *errlist;
 	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);
+	err = zpool_translate_vdev_guids(zhp, vds, vdev_guids,
+	    guids_to_paths, &vd_errlist);
 
-		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);
+	if (err == 0) {
+		err = lzc_initialize(zhp->zpool_name, cmd_type,
+		    vdev_guids, &errlist);
+		if (err == 0) {
 			fnvlist_free(vdev_guids);
 			fnvlist_free(guids_to_paths);
-			return (zfs_error(hdl, err, msg));
+			return (0);
 		}
 
-		uint64_t guid = fnvlist_lookup_uint64(tgt, ZPOOL_CONFIG_GUID);
-		fnvlist_add_uint64(vdev_guids, vd_path, guid);
+		if (errlist != NULL) {
+			vd_errlist = fnvlist_lookup_nvlist(errlist,
+			    ZPOOL_INITIALIZE_VDEVS);
+		}
 
-		(void) snprintf(msg, sizeof (msg), "%llu", guid);
-		fnvlist_add_string(guids_to_paths, msg, vd_path);
+		(void) snprintf(msg, sizeof (msg),
+		    dgettext(TEXT_DOMAIN, "operation failed"));
+	} else {
+		verify(vd_errlist != NULL);
+	}
+
+	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;
+
+		if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem),
+		    &path) != 0)
+			path = nvpair_name(elem);
+
+		(void) zfs_error_fmt(zhp->zpool_hdl, vd_error,
+		    "cannot initialize '%s'", path);
 	}
 
-	int err = lzc_initialize(zhp->zpool_name, cmd_type, vdev_guids,
-	    &errlist);
 	fnvlist_free(vdev_guids);
+	fnvlist_free(guids_to_paths);
 
-	if (err == 0) {
-		fnvlist_free(guids_to_paths);
-		return (0);
+	if (vd_errlist != NULL) {
+		fnvlist_free(vd_errlist);
+		return (-1);
 	}
 
+	return (zpool_standard_error(zhp->zpool_hdl, err, msg));
+}
+
+static int
+xlate_trim_err(int err)
+{
+	switch (err) {
+	case ENODEV:
+		return (EZFS_NODEVICE);
+	case EINVAL:
+	case EROFS:
+		return (EZFS_BADDEV);
+	case EBUSY:
+		return (EZFS_TRIMMING);
+	case ESRCH:
+		return (EZFS_NO_TRIM);
+	case EOPNOTSUPP:
+		return (EZFS_TRIM_NOTSUP);
+	}
+	return (err);
+}
+
+/*
+ * Begin, suspend, or cancel the TRIM (discarding of all free blocks) for
+ * the given vdevs in the given pool.
+ */
+int
+zpool_trim(zpool_handle_t *zhp, pool_trim_func_t cmd_type, nvlist_t *vds,
+    trimflags_t *trim_flags)
+{
+	char msg[1024];
+	int err;
+
+	nvlist_t *vdev_guids = fnvlist_alloc();
+	nvlist_t *guids_to_paths = fnvlist_alloc();
 	nvlist_t *vd_errlist = NULL;
-	if (errlist != NULL) {
-		vd_errlist = fnvlist_lookup_nvlist(errlist,
-		    ZPOOL_INITIALIZE_VDEVS);
+	nvlist_t *errlist;
+	nvpair_t *elem;
+
+	err = zpool_translate_vdev_guids(zhp, vds, vdev_guids,
+	    guids_to_paths, &vd_errlist);
+	if (err == 0) {
+		err = lzc_trim(zhp->zpool_name, cmd_type, trim_flags->rate,
+		    trim_flags->secure, vdev_guids, &errlist);
+		if (err == 0) {
+			fnvlist_free(vdev_guids);
+			fnvlist_free(guids_to_paths);
+			return (0);
+		}
+
+		if (errlist != NULL) {
+			vd_errlist = fnvlist_lookup_nvlist(errlist,
+			    ZPOOL_TRIM_VDEVS);
+		}
+
+		(void) snprintf(msg, sizeof (msg),
+		    dgettext(TEXT_DOMAIN, "operation failed"));
+	} else {
+		verify(vd_errlist != NULL);
 	}
 
-	(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_trim_err(fnvpair_value_int64(elem));
+		char *path;
+		/*
+		 * If only the pool was specified, and it was not a secure
+		 * trim then suppress warnings for individual vdevs which
+		 * do not support trimming.
+		 */
+		if (vd_error == EZFS_TRIM_NOTSUP &&
+		    trim_flags->fullpool &&
+		    !trim_flags->secure) {
+			continue;
+		}
 
-	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);
+		if (nvlist_lookup_string(guids_to_paths, nvpair_name(elem),
+		    &path) != 0)
+			path = nvpair_name(elem);
+
+		(void) zfs_error_fmt(zhp->zpool_hdl, vd_error,
+		    "cannot trim '%s'", path);
 	}
 
+	fnvlist_free(vdev_guids);
 	fnvlist_free(guids_to_paths);
-	if (vd_errlist != NULL)
+
+	if (vd_errlist != NULL) {
+		fnvlist_free(vd_errlist);
 		return (-1);
+	}
 
-	return (zpool_standard_error(hdl, err, msg));
+	return (zpool_standard_error(zhp->zpool_hdl, err, msg));
 }
 
 /*
diff --git a/usr/src/lib/libzfs/common/libzfs_util.c b/usr/src/lib/libzfs/common/libzfs_util.c
index 696a9e10d0..75eb9722b3 100644
--- a/usr/src/lib/libzfs/common/libzfs_util.c
+++ b/usr/src/lib/libzfs/common/libzfs_util.c
@@ -253,6 +253,8 @@ libzfs_error_description(libzfs_handle_t *hdl)
 	case EZFS_ACTIVE_POOL:
 		return (dgettext(TEXT_DOMAIN, "pool is imported on a "
 		    "different host"));
+	case EZFS_CRYPTOFAILED:
+		return (dgettext(TEXT_DOMAIN, "encryption failure"));
 	case EZFS_TOOMANY:
 		return (dgettext(TEXT_DOMAIN, "argument list too long"));
 	case EZFS_INITIALIZING:
@@ -262,11 +264,16 @@ libzfs_error_description(libzfs_handle_t *hdl)
 		    "initialization"));
 	case EZFS_WRONG_PARENT:
 		return (dgettext(TEXT_DOMAIN, "invalid parent dataset"));
+	case EZFS_TRIMMING:
+		return (dgettext(TEXT_DOMAIN, "currently trimming"));
+	case EZFS_NO_TRIM:
+		return (dgettext(TEXT_DOMAIN, "there is no active trim"));
+	case EZFS_TRIM_NOTSUP:
+		return (dgettext(TEXT_DOMAIN, "trim operations are not "
+		    "supported by this device"));
 	case EZFS_NO_RESILVER_DEFER:
 		return (dgettext(TEXT_DOMAIN, "this action requires the "
 		    "resilver_defer feature"));
-	case EZFS_CRYPTOFAILED:
-		return (dgettext(TEXT_DOMAIN, "encryption failure"));
 	case EZFS_UNKNOWN:
 		return (dgettext(TEXT_DOMAIN, "unknown error"));
 	default:
diff --git a/usr/src/lib/libzfs/common/mapfile-vers b/usr/src/lib/libzfs/common/mapfile-vers
index f4412bbd9a..95fb983e8c 100644
--- a/usr/src/lib/libzfs/common/mapfile-vers
+++ b/usr/src/lib/libzfs/common/mapfile-vers
@@ -23,6 +23,7 @@
 # Copyright (c) 2006, 2010, Oracle and/or its affiliates. All rights reserved.
 # Copyright (c) 2011, 2017 by Delphix. All rights reserved.
 # Copyright 2016 Nexenta Systems, Inc.
+# Copyright 2019 Joyent, Inc.
 #
 
 #
@@ -251,6 +252,7 @@ SYMBOL_VERSION SUNWprivate_1.1 {
 	zpool_skip_pool;
 	zpool_state_to_name;
 	zpool_sync_one;
+	zpool_trim;
 	zpool_tryimport;
 	zpool_unmount_datasets;
 	zpool_upgrade;