Pool allocation classes

Allocation Classes add the ability to have allocation classes in a
pool that are dedicated to serving specific block categories, such
as DDT data, metadata, and small file blocks. A pool can opt-in to
this feature by adding a 'special' or 'dedup' top-level VDEV.

Reviewed by: Pavel Zakharov <[email protected]>
Reviewed-by: Richard Laager <[email protected]>
Reviewed-by: Alek Pinchuk <[email protected]>
Reviewed-by: Håkan Johansson <[email protected]>
Reviewed-by: Andreas Dilger <[email protected]>
Reviewed-by: DHE <[email protected]>
Reviewed-by: Richard Elling <[email protected]>
Reviewed-by: Gregor Kopka <[email protected]>
Reviewed-by: Kash Pande <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed-by: Matthew Ahrens <[email protected]>
Signed-off-by: Don Brady <[email protected]>
Closes #5182 

3 files changed, 83 insertions, 7 deletions

diff --git a/lib/libzfs/libzfs_dataset.c b/lib/libzfs/libzfs_dataset.c
index 85b4c5531..76a316a9c 100644
--- a/lib/libzfs/libzfs_dataset.c
+++ b/lib/libzfs/libzfs_dataset.c
@@ -1195,6 +1195,36 @@ zfs_valid_proplist(libzfs_handle_t *hdl, zfs_type_t type, nvlist_t *nvl,
 			}
 			break;
 		}
+
+		case ZFS_PROP_SPECIAL_SMALL_BLOCKS:
+			if (zpool_hdl != NULL) {
+				char state[64] = "";
+
+				/*
+				 * Issue a warning but do not fail so that
+				 * tests for setable properties succeed.
+				 */
+				if (zpool_prop_get_feature(zpool_hdl,
+				    "feature@allocation_classes", state,
+				    sizeof (state)) != 0 ||
+				    strcmp(state, ZFS_FEATURE_ACTIVE) != 0) {
+					(void) fprintf(stderr, gettext(
+					    "%s: property requires a special "
+					    "device in the pool\n"), propname);
+				}
+			}
+			if (intval != 0 &&
+			    (intval < SPA_MINBLOCKSIZE ||
+			    intval > SPA_OLD_MAXBLOCKSIZE || !ISP2(intval))) {
+				zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
+				    "invalid '%s=%d' property: must be zero or "
+				    "a power of 2 from 512B to 128K"), propname,
+				    intval);
+				(void) zfs_error(hdl, EZFS_BADPROP, errbuf);
+				goto error;
+			}
+			break;
+
 		case ZFS_PROP_MLSLABEL:
 		{
 #ifdef HAVE_MLSLABEL
diff --git a/lib/libzfs/libzfs_pool.c b/lib/libzfs/libzfs_pool.c
index 0785c3170..fca1a4178 100644
--- a/lib/libzfs/libzfs_pool.c
+++ b/lib/libzfs/libzfs_pool.c
@@ -26,6 +26,7 @@
  * Copyright 2016 Igor Kozhukhov <[email protected]>
  * Copyright (c) 2018 Datto Inc.
  * Copyright (c) 2017 Open-E, Inc. All Rights Reserved.
+ * Copyright (c) 2017, Intel Corporation.
  */
 
 #include <ctype.h>
@@ -720,6 +721,7 @@ zpool_valid_proplist(libzfs_handle_t *hdl, const char *poolname,
 				goto error;
 			}
 			break;
+
 		default:
 			zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
 			    "property '%s'(%d) not defined"), propname, prop);
@@ -1173,6 +1175,30 @@ zpool_get_state(zpool_handle_t *zhp)
 }
 
 /*
+ * Check if vdev list contains a special vdev
+ */
+static boolean_t
+zpool_has_special_vdev(nvlist_t *nvroot)
+{
+	nvlist_t **child;
+	uint_t children;
+
+	if (nvlist_lookup_nvlist_array(nvroot, ZPOOL_CONFIG_CHILDREN, &child,
+	    &children) == 0) {
+		for (uint_t c = 0; c < children; c++) {
+			char *bias;
+
+			if (nvlist_lookup_string(child[c],
+			    ZPOOL_CONFIG_ALLOCATION_BIAS, &bias) == 0 &&
+			    strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0) {
+				return (B_TRUE);
+			}
+		}
+	}
+	return (B_FALSE);
+}
+
+/*
  * Create the named pool, using the provided vdev list.  It is assumed
  * that the consumer has already validated the contents of the nvlist, so we
  * don't have to worry about error semantics.
@@ -1220,6 +1246,17 @@ zpool_create(libzfs_handle_t *hdl, const char *pool, nvlist_t *nvroot,
 		    fsprops, zoned, NULL, NULL, B_TRUE, msg)) == NULL) {
 			goto create_failed;
 		}
+
+		if (nvlist_exists(zc_fsprops,
+		    zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS)) &&
+		    !zpool_has_special_vdev(nvroot)) {
+			zfs_error_aux(hdl, dgettext(TEXT_DOMAIN,
+			    "%s property requires a special vdev"),
+			    zfs_prop_to_name(ZFS_PROP_SPECIAL_SMALL_BLOCKS));
+			(void) zfs_error(hdl, EZFS_BADPROP, msg);
+			goto create_failed;
+		}
+
 		if (!zc_props &&
 		    (nvlist_alloc(&zc_props, NV_UNIQUE_NAME, 0) != 0)) {
 			goto create_failed;
diff --git a/lib/libzpool/util.c b/lib/libzpool/util.c
index bb1710e79..8a6f4c325 100644
--- a/lib/libzpool/util.c
+++ b/lib/libzpool/util.c
@@ -22,6 +22,7 @@
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  * Copyright (c) 2016 by Delphix. All rights reserved.
  * Copyright 2017 Jason King
+ * Copyright (c) 2017, Intel Corporation.
  */
 
 #include <assert.h>
@@ -150,7 +151,6 @@ show_vdev_stats(const char *desc, const char *ctype, nvlist_t *nv, int indent)
 	uint_t c, children;
 	char used[6], avail[6];
 	char rops[6], wops[6], rbytes[6], wbytes[6], rerr[6], werr[6], cerr[6];
-	char *prefix = "";
 
 	v0 = umem_zalloc(sizeof (*v0), UMEM_NOFAIL);
 
@@ -162,15 +162,24 @@ show_vdev_stats(const char *desc, const char *ctype, nvlist_t *nv, int indent)
 	}
 
 	if (desc != NULL) {
-		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log);
-
-		if (is_log)
-			prefix = "log ";
+		char *suffix = "", *bias = NULL;
+		char bias_suffix[32];
 
+		(void) nvlist_lookup_uint64(nv, ZPOOL_CONFIG_IS_LOG, &is_log);
+		(void) nvlist_lookup_string(nv, ZPOOL_CONFIG_ALLOCATION_BIAS,
+		    &bias);
 		if (nvlist_lookup_uint64_array(nv, ZPOOL_CONFIG_VDEV_STATS,
 		    (uint64_t **)&vs, &c) != 0)
 			vs = v0;
 
+		if (bias != NULL) {
+			(void) snprintf(bias_suffix, sizeof (bias_suffix),
+			    " (%s)", bias);
+			suffix = bias_suffix;
+		} else if (is_log) {
+			suffix = " (log)";
+		}
+
 		sec = MAX(1, vs->vs_timestamp / NANOSEC);
 
 		nicenum(vs->vs_alloc, used, sizeof (used));
@@ -187,9 +196,9 @@ show_vdev_stats(const char *desc, const char *ctype, nvlist_t *nv, int indent)
 
 		(void) printf("%*s%s%*s%*s%*s %5s %5s %5s %5s %5s %5s %5s\n",
 		    indent, "",
-		    prefix,
-		    (int)(indent+strlen(prefix)-25-(vs->vs_space ? 0 : 12)),
 		    desc,
+		    (int)(indent+strlen(desc)-25-(vs->vs_space ? 0 : 12)),
+		    suffix,
 		    vs->vs_space ? 6 : 0, vs->vs_space ? used : "",
 		    vs->vs_space ? 6 : 0, vs->vs_space ? avail : "",
 		    rops, wops, rbytes, wbytes, rerr, werr, cerr);