svn commit: r354382 - vendor-sys/illumos/dist/common/zfs vendor-sys/illumos/dist/uts/common/fs/zfs vendor-sys/illumos/dist/uts/common/fs/zfs/sys vendor-sys/illumos/dist/uts/common/sys/fs vendor/ill...
Andriy Gapon
avg at FreeBSD.org
Wed Nov 6 08:58:07 UTC 2019
Author: avg
Date: Wed Nov 6 08:58:03 2019
New Revision: 354382
URL: https://svnweb.freebsd.org/changeset/base/354382
Log:
10601 10757 Pool allocation classes
illumos/illumos-gate at 663207adb1669640c01c5ec6949ce78fd806efae
https://github.com/illumos/illumos-gate/commit/663207adb1669640c01c5ec6949ce78fd806efae
10601 Pool allocation classes
https://www.illumos.org/issues/10601
illumos port of ZoL Pool allocation classes. Includes at least these two
commits:
441709695 Pool allocation classes misplacing small file blocks
cc99f275a Pool allocation classes
10757 Add -gLp to zpool subcommands for alt vdev names
https://www.illumos.org/issues/10757
Port from ZoL of
d2f3e292d Add -gLp to zpool subcommands for alt vdev names
Note that a subsequent ZoL commit changed -p to -P
a77f29f93 Change full path subcommand flag from -p to -P
Portions contributed by: Jerry Jelinek <jerry.jelinek at joyent.com>
Portions contributed by: Håkan Johansson <f96hajo at chalmers.se>
Portions contributed by: Richard Yao <ryao at gentoo.org>
Portions contributed by: Chunwei Chen <david.chen at nutanix.com>
Portions contributed by: loli10K <ezomori.nozomu at gmail.com>
Author: Don Brady <don.brady at delphix.com>
Modified:
vendor-sys/illumos/dist/common/zfs/zfeature_common.c
vendor-sys/illumos/dist/common/zfs/zfeature_common.h
vendor-sys/illumos/dist/common/zfs/zfs_prop.c
vendor-sys/illumos/dist/uts/common/fs/zfs/dmu.c
vendor-sys/illumos/dist/uts/common/fs/zfs/dmu_objset.c
vendor-sys/illumos/dist/uts/common/fs/zfs/metaslab.c
vendor-sys/illumos/dist/uts/common/fs/zfs/spa.c
vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu_objset.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/metaslab.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa_impl.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev_impl.h
vendor-sys/illumos/dist/uts/common/fs/zfs/sys/zio.h
vendor-sys/illumos/dist/uts/common/fs/zfs/vdev.c
vendor-sys/illumos/dist/uts/common/fs/zfs/vdev_label.c
vendor-sys/illumos/dist/uts/common/fs/zfs/vdev_removal.c
vendor-sys/illumos/dist/uts/common/fs/zfs/zfs_ioctl.c
vendor-sys/illumos/dist/uts/common/fs/zfs/zio.c
vendor-sys/illumos/dist/uts/common/sys/fs/zfs.h
Changes in other areas also in this revision:
Modified:
vendor/illumos/dist/cmd/zdb/zdb.c
vendor/illumos/dist/cmd/zpool/zpool_main.c
vendor/illumos/dist/cmd/zpool/zpool_vdev.c
vendor/illumos/dist/cmd/ztest/ztest.c
vendor/illumos/dist/lib/libzfs/common/libzfs.h
vendor/illumos/dist/lib/libzfs/common/libzfs_dataset.c
vendor/illumos/dist/lib/libzfs/common/libzfs_pool.c
vendor/illumos/dist/lib/libzpool/common/util.c
vendor/illumos/dist/man/man1m/zfs.1m
vendor/illumos/dist/man/man1m/zpool.1m
vendor/illumos/dist/man/man5/zpool-features.5
Modified: vendor-sys/illumos/dist/common/zfs/zfeature_common.c
==============================================================================
--- vendor-sys/illumos/dist/common/zfs/zfeature_common.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/common/zfs/zfeature_common.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -20,11 +20,12 @@
*/
/*
- * Copyright (c) 2011, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2013, Joyent, Inc. All rights reserved.
* Copyright (c) 2014, Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifdef _KERNEL
@@ -298,4 +299,11 @@ zpool_feature_init(void)
"Reduce memory used by removed devices when their blocks are "
"freed or remapped.",
ZFEATURE_FLAG_READONLY_COMPAT, obsolete_counts_deps);
+
+ {
+ zfeature_register(SPA_FEATURE_ALLOCATION_CLASSES,
+ "org.zfsonlinux:allocation_classes", "allocation_classes",
+ "Support for separate allocation classes.",
+ ZFEATURE_FLAG_READONLY_COMPAT, NULL);
+ }
}
Modified: vendor-sys/illumos/dist/common/zfs/zfeature_common.h
==============================================================================
--- vendor-sys/illumos/dist/common/zfs/zfeature_common.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/common/zfs/zfeature_common.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -24,6 +24,7 @@
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2013, Joyent, Inc. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _ZFEATURE_COMMON_H
@@ -61,6 +62,7 @@ typedef enum spa_feature {
SPA_FEATURE_OBSOLETE_COUNTS,
SPA_FEATURE_POOL_CHECKPOINT,
SPA_FEATURE_SPACEMAP_V2,
+ SPA_FEATURE_ALLOCATION_CLASSES,
SPA_FEATURES
} spa_feature_t;
Modified: vendor-sys/illumos/dist/common/zfs/zfs_prop.c
==============================================================================
--- vendor-sys/illumos/dist/common/zfs/zfs_prop.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/common/zfs/zfs_prop.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -435,6 +435,9 @@ zfs_prop_init(void)
zprop_register_number(ZFS_PROP_RECORDSIZE, "recordsize",
SPA_OLD_MAXBLOCKSIZE, PROP_INHERIT,
ZFS_TYPE_FILESYSTEM, "512 to 1M, power of 2", "RECSIZE");
+ zprop_register_number(ZFS_PROP_SPECIAL_SMALL_BLOCKS,
+ "special_small_blocks", 0, PROP_INHERIT, ZFS_TYPE_FILESYSTEM,
+ "zero or 512 to 128K, power of 2", "SPECIAL_SMALL_BLOCKS");
/* hidden properties */
zprop_register_hidden(ZFS_PROP_REMAPTXG, "remaptxg", PROP_TYPE_NUMBER,
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/dmu.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/dmu.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/dmu.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -2201,6 +2201,8 @@ dmu_write_policy(objset_t *os, dnode_t *dn, int level,
zp->zp_dedup = dedup;
zp->zp_dedup_verify = dedup && dedup_verify;
zp->zp_nopwrite = nopwrite;
+ zp->zp_zpl_smallblk = DMU_OT_IS_FILE(zp->zp_type) ?
+ os->os_zpl_special_smallblock : 0;
}
int
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/dmu_objset.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/dmu_objset.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/dmu_objset.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -304,6 +304,20 @@ dnodesize_changed_cb(void *arg, uint64_t newval)
}
static void
+smallblk_changed_cb(void *arg, uint64_t newval)
+{
+ objset_t *os = arg;
+
+ /*
+ * Inheritance and range checking should have been done by now.
+ */
+ ASSERT(newval <= SPA_OLD_MAXBLOCKSIZE);
+ ASSERT(ISP2(newval));
+
+ os->os_zpl_special_smallblock = newval;
+}
+
+static void
logbias_changed_cb(void *arg, uint64_t newval)
{
objset_t *os = arg;
@@ -515,6 +529,12 @@ dmu_objset_open_impl(spa_t *spa, dsl_dataset_t *ds, bl
err = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_DNODESIZE),
dnodesize_changed_cb, os);
+ }
+ if (err == 0) {
+ err = dsl_prop_register(ds,
+ zfs_prop_to_name(
+ ZFS_PROP_SPECIAL_SMALL_BLOCKS),
+ smallblk_changed_cb, os);
}
}
if (needlock)
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/metaslab.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/metaslab.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/metaslab.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -23,6 +23,7 @@
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
+ * Copyright (c) 2017, Intel Corporation.
*/
#include <sys/zfs_context.h>
@@ -288,7 +289,7 @@ metaslab_class_validate(metaslab_class_t *mc)
return (0);
}
-void
+static void
metaslab_class_space_update(metaslab_class_t *mc, int64_t alloc_delta,
int64_t defer_delta, int64_t space_delta, int64_t dspace_delta)
{
@@ -325,7 +326,8 @@ metaslab_class_get_dspace(metaslab_class_t *mc)
void
metaslab_class_histogram_verify(metaslab_class_t *mc)
{
- vdev_t *rvd = mc->mc_spa->spa_root_vdev;
+ spa_t *spa = mc->mc_spa;
+ vdev_t *rvd = spa->spa_root_vdev;
uint64_t *mc_hist;
int i;
@@ -831,7 +833,8 @@ metaslab_group_histogram_verify(metaslab_group_t *mg)
for (int m = 0; m < vd->vdev_ms_count; m++) {
metaslab_t *msp = vd->vdev_ms[m];
- if (msp->ms_sm == NULL)
+ /* skip if not active or not a member */
+ if (msp->ms_sm == NULL || msp->ms_group != mg)
continue;
for (i = 0; i < SPACE_MAP_HISTOGRAM_SIZE; i++)
@@ -964,12 +967,14 @@ metaslab_group_fragmentation(metaslab_group_t *mg)
if (msp->ms_fragmentation == ZFS_FRAG_INVALID)
continue;
+ if (msp->ms_group != mg)
+ continue;
valid_ms++;
fragmentation += msp->ms_fragmentation;
}
- if (valid_ms <= vd->vdev_ms_count / 2)
+ if (valid_ms <= mg->mg_vd->vdev_ms_count / 2)
return (ZFS_FRAG_INVALID);
fragmentation /= valid_ms;
@@ -1000,7 +1005,10 @@ metaslab_group_allocatable(metaslab_group_t *mg, metas
* groups to select from. Otherwise, we always consider it eligible
* for allocations.
*/
- if (mc != spa_normal_class(spa) || mc->mc_groups <= 1)
+ if ((mc != spa_normal_class(spa) &&
+ mc != spa_special_class(spa) &&
+ mc != spa_dedup_class(spa)) ||
+ mc->mc_groups <= 1)
return (B_TRUE);
/*
@@ -1534,12 +1542,26 @@ metaslab_unload(metaslab_t *msp)
msp->ms_max_size = 0;
}
+static void
+metaslab_space_update(vdev_t *vd, metaslab_class_t *mc, int64_t alloc_delta,
+ int64_t defer_delta, int64_t space_delta)
+{
+ vdev_space_update(vd, alloc_delta, defer_delta, space_delta);
+
+ ASSERT3P(vd->vdev_spa->spa_root_vdev, ==, vd->vdev_parent);
+ ASSERT(vd->vdev_ms_count != 0);
+
+ metaslab_class_space_update(mc, alloc_delta, defer_delta, space_delta,
+ vdev_deflated_space(vd, space_delta));
+}
+
int
metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object, uint64_t txg,
metaslab_t **msp)
{
vdev_t *vd = mg->mg_vd;
- objset_t *mos = vd->vdev_spa->spa_meta_objset;
+ spa_t *spa = vd->vdev_spa;
+ objset_t *mos = spa->spa_meta_objset;
metaslab_t *ms;
int error;
@@ -1596,8 +1618,8 @@ metaslab_init(metaslab_group_t *mg, uint64_t id, uint6
/*
* If metaslab_debug_load is set and we're initializing a metaslab
- * that has an allocated space map object then load the its space
- * map so that can verify frees.
+ * that has an allocated space map object then load the space map
+ * so that we can verify frees.
*/
if (metaslab_debug_load && ms->ms_sm != NULL) {
mutex_enter(&ms->ms_lock);
@@ -1619,16 +1641,19 @@ void
metaslab_fini(metaslab_t *msp)
{
metaslab_group_t *mg = msp->ms_group;
+ vdev_t *vd = mg->mg_vd;
metaslab_group_remove(mg, msp);
mutex_enter(&msp->ms_lock);
VERIFY(msp->ms_group == NULL);
- vdev_space_update(mg->mg_vd, -space_map_allocated(msp->ms_sm),
- 0, -msp->ms_size);
+ metaslab_space_update(vd, mg->mg_class,
+ -space_map_allocated(msp->ms_sm), 0, -msp->ms_size);
+
space_map_close(msp->ms_sm);
metaslab_unload(msp);
+
range_tree_destroy(msp->ms_allocatable);
range_tree_destroy(msp->ms_freeing);
range_tree_destroy(msp->ms_freed);
@@ -2643,7 +2668,7 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg)
ASSERT3P(msp->ms_checkpointing, ==, NULL);
msp->ms_checkpointing = range_tree_create(NULL, NULL);
- vdev_space_update(vd, 0, 0, msp->ms_size);
+ metaslab_space_update(vd, mg->mg_class, 0, 0, msp->ms_size);
}
ASSERT0(range_tree_space(msp->ms_freeing));
ASSERT0(range_tree_space(msp->ms_checkpointing));
@@ -2665,7 +2690,8 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg)
defer_delta -= range_tree_space(*defer_tree);
}
- vdev_space_update(vd, alloc_delta + defer_delta, defer_delta, 0);
+ metaslab_space_update(vd, mg->mg_class, alloc_delta + defer_delta,
+ defer_delta, 0);
/*
* If there's a metaslab_load() in progress, wait for it to complete
@@ -2764,21 +2790,25 @@ metaslab_sync_reassess(metaslab_group_t *mg)
spa_config_exit(spa, SCL_ALLOC, FTAG);
}
-static uint64_t
-metaslab_distance(metaslab_t *msp, dva_t *dva)
+/*
+ * When writing a ditto block (i.e. more than one DVA for a given BP) on
+ * the same vdev as an existing DVA of this BP, then try to allocate it
+ * on a different metaslab than existing DVAs (i.e. a unique metaslab).
+ */
+static boolean_t
+metaslab_is_unique(metaslab_t *msp, dva_t *dva)
{
- uint64_t ms_shift = msp->ms_group->mg_vd->vdev_ms_shift;
- uint64_t offset = DVA_GET_OFFSET(dva) >> ms_shift;
- uint64_t start = msp->ms_id;
+ uint64_t dva_ms_id;
+ if (DVA_GET_ASIZE(dva) == 0)
+ return (B_TRUE);
+
if (msp->ms_group->mg_vd->vdev_id != DVA_GET_VDEV(dva))
- return (1ULL << 63);
+ return (B_TRUE);
- if (offset < start)
- return ((start - offset) << ms_shift);
- if (offset > start)
- return ((offset - start) << ms_shift);
- return (0);
+ dva_ms_id = DVA_GET_OFFSET(dva) >> msp->ms_group->mg_vd->vdev_ms_shift;
+
+ return (msp->ms_id != dva_ms_id);
}
/*
@@ -3012,7 +3042,7 @@ metaslab_block_alloc(metaslab_t *msp, uint64_t size, u
*/
static metaslab_t *
find_valid_metaslab(metaslab_group_t *mg, uint64_t activation_weight,
- dva_t *dva, int d, uint64_t min_distance, uint64_t asize, int allocator,
+ dva_t *dva, int d, boolean_t want_unique, uint64_t asize, int allocator,
zio_alloc_list_t *zal, metaslab_t *search, boolean_t *was_active)
{
avl_index_t idx;
@@ -3047,13 +3077,10 @@ find_valid_metaslab(metaslab_group_t *mg, uint64_t act
if (activation_weight == METASLAB_WEIGHT_PRIMARY || *was_active)
break;
- uint64_t target_distance = min_distance
- + (space_map_allocated(msp->ms_sm) != 0 ? 0 :
- min_distance >> 1);
-
for (i = 0; i < d; i++) {
- if (metaslab_distance(msp, &dva[i]) < target_distance)
- break;
+ if (want_unique &&
+ !metaslab_is_unique(msp, &dva[i]))
+ break; /* try another metaslab */
}
if (i == d)
break;
@@ -3071,8 +3098,8 @@ find_valid_metaslab(metaslab_group_t *mg, uint64_t act
/* ARGSUSED */
static uint64_t
metaslab_group_alloc_normal(metaslab_group_t *mg, zio_alloc_list_t *zal,
- uint64_t asize, uint64_t txg, uint64_t min_distance, dva_t *dva, int d,
- int allocator)
+ uint64_t asize, uint64_t txg, boolean_t want_unique, dva_t *dva,
+ int d, int allocator)
{
metaslab_t *msp = NULL;
uint64_t offset = -1ULL;
@@ -3126,7 +3153,7 @@ metaslab_group_alloc_normal(metaslab_group_t *mg, zio_
was_active = B_TRUE;
} else {
msp = find_valid_metaslab(mg, activation_weight, dva, d,
- min_distance, asize, allocator, zal, search,
+ want_unique, asize, allocator, zal, search,
&was_active);
}
@@ -3264,6 +3291,7 @@ next:
* metaslab.
*/
ASSERT(!metaslab_should_allocate(msp, asize));
+
mutex_exit(&msp->ms_lock);
}
mutex_exit(&msp->ms_lock);
@@ -3273,14 +3301,14 @@ next:
static uint64_t
metaslab_group_alloc(metaslab_group_t *mg, zio_alloc_list_t *zal,
- uint64_t asize, uint64_t txg, uint64_t min_distance, dva_t *dva, int d,
- int allocator)
+ uint64_t asize, uint64_t txg, boolean_t want_unique, dva_t *dva,
+ int d, int allocator)
{
uint64_t offset;
ASSERT(mg->mg_initialized);
- offset = metaslab_group_alloc_normal(mg, zal, asize, txg,
- min_distance, dva, d, allocator);
+ offset = metaslab_group_alloc_normal(mg, zal, asize, txg, want_unique,
+ dva, d, allocator);
mutex_enter(&mg->mg_lock);
if (offset == -1ULL) {
@@ -3308,14 +3336,6 @@ metaslab_group_alloc(metaslab_group_t *mg, zio_alloc_l
}
/*
- * If we have to write a ditto block (i.e. more than one DVA for a given BP)
- * on the same vdev as an existing DVA of this BP, then try to allocate it
- * at least (vdev_asize / (2 ^ ditto_same_vdev_distance_shift)) away from the
- * existing DVAs.
- */
-int ditto_same_vdev_distance_shift = 3;
-
-/*
* Allocate a block for the specified i/o.
*/
int
@@ -3331,6 +3351,7 @@ metaslab_alloc_dva(spa_t *spa, metaslab_class_t *mc, u
/*
* For testing, make some blocks above a certain size be gang blocks.
+ * This will also test spilling from special to normal.
*/
if (psize >= metaslab_force_ganging && (ddi_get_lbolt() & 3) == 0) {
metaslab_trace_add(zal, NULL, NULL, psize, d, TRACE_FORCE_GANG,
@@ -3382,6 +3403,7 @@ metaslab_alloc_dva(spa_t *spa, metaslab_class_t *mc, u
vd = vdev_lookup_top(spa, DVA_GET_VDEV(&dva[d - 1]));
mg = vd->vdev_mg->mg_next;
} else {
+ ASSERT(mc->mc_rotor != NULL);
mg = mc->mc_rotor;
}
@@ -3446,25 +3468,17 @@ top:
ASSERT(mg->mg_class == mc);
- /*
- * If we don't need to try hard, then require that the
- * block be 1/8th of the device away from any other DVAs
- * in this BP. If we are trying hard, allow any offset
- * to be used (distance=0).
- */
- uint64_t distance = 0;
- if (!try_hard) {
- distance = vd->vdev_asize >>
- ditto_same_vdev_distance_shift;
- if (distance <= (1ULL << vd->vdev_ms_shift))
- distance = 0;
- }
-
uint64_t asize = vdev_psize_to_asize(vd, psize);
ASSERT(P2PHASE(asize, 1ULL << vd->vdev_ashift) == 0);
+ /*
+ * If we don't need to try hard, then require that the
+ * block be on an different metaslab from any other DVAs
+ * in this BP (unique=true). If we are trying hard, then
+ * allow any metaslab to be used (unique=false).
+ */
uint64_t offset = metaslab_group_alloc(mg, zal, asize, txg,
- distance, dva, d, allocator);
+ !try_hard, dva, d, allocator);
if (offset != -1ULL) {
/*
@@ -3843,7 +3857,8 @@ metaslab_class_throttle_reserve(metaslab_class_t *mc,
if (reserved_slots < max)
available_slots = max - reserved_slots;
- if (slots <= available_slots || GANG_ALLOCATION(flags)) {
+ if (slots <= available_slots || GANG_ALLOCATION(flags) ||
+ flags & METASLAB_MUST_RESERVE) {
/*
* We reserve the slots individually so that we can unreserve
* them individually when an I/O completes.
@@ -4126,9 +4141,11 @@ metaslab_claim(spa_t *spa, const blkptr_t *bp, uint64_
spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER);
- for (int d = 0; d < ndvas; d++)
- if ((error = metaslab_claim_dva(spa, &dva[d], txg)) != 0)
+ for (int d = 0; d < ndvas; d++) {
+ error = metaslab_claim_dva(spa, &dva[d], txg);
+ if (error != 0)
break;
+ }
spa_config_exit(spa, SCL_ALLOC, FTAG);
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/spa.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/spa.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/spa.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -28,6 +28,7 @@
* Copyright (c) 2014 Integros [integros.com]
* Copyright 2016 Toomas Soome <tsoome at me.com>
* Copyright 2018 Joyent, Inc.
+ * Copyright (c) 2017, Intel Corporation.
* Copyright (c) 2017 Datto Inc.
* Copyright 2018 OmniOS Community Edition (OmniOSce) Association.
*/
@@ -274,8 +275,14 @@ spa_prop_get_config(spa_t *spa, nvlist_t **nvp)
ASSERT(MUTEX_HELD(&spa->spa_props_lock));
if (rvd != NULL) {
- alloc = metaslab_class_get_alloc(spa_normal_class(spa));
- size = metaslab_class_get_space(spa_normal_class(spa));
+ alloc = metaslab_class_get_alloc(mc);
+ alloc += metaslab_class_get_alloc(spa_special_class(spa));
+ alloc += metaslab_class_get_alloc(spa_dedup_class(spa));
+
+ size = metaslab_class_get_space(mc);
+ size += metaslab_class_get_space(spa_special_class(spa));
+ size += metaslab_class_get_space(spa_dedup_class(spa));
+
spa_prop_add_list(*nvp, ZPOOL_PROP_NAME, spa_name(spa), 0, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_SIZE, NULL, size, src);
spa_prop_add_list(*nvp, ZPOOL_PROP_ALLOCATED, NULL, alloc, src);
@@ -1137,6 +1144,8 @@ spa_activate(spa_t *spa, int mode)
spa->spa_normal_class = metaslab_class_create(spa, zfs_metaslab_ops);
spa->spa_log_class = metaslab_class_create(spa, zfs_metaslab_ops);
+ spa->spa_special_class = metaslab_class_create(spa, zfs_metaslab_ops);
+ spa->spa_dedup_class = metaslab_class_create(spa, zfs_metaslab_ops);
/* Try to create a covering process */
mutex_enter(&spa->spa_proc_lock);
@@ -1232,6 +1241,12 @@ spa_deactivate(spa_t *spa)
metaslab_class_destroy(spa->spa_log_class);
spa->spa_log_class = NULL;
+ metaslab_class_destroy(spa->spa_special_class);
+ spa->spa_special_class = NULL;
+
+ metaslab_class_destroy(spa->spa_dedup_class);
+ spa->spa_dedup_class = NULL;
+
/*
* If this was part of an import or the open otherwise failed, we may
* still have errors left in the queues. Empty them just in case.
@@ -4834,7 +4849,8 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_
char *poolname;
nvlist_t *nvl;
- if (nvlist_lookup_string(props,
+ if (props == NULL ||
+ nvlist_lookup_string(props,
zpool_prop_to_name(ZPOOL_PROP_TNAME), &poolname) != 0)
poolname = (char *)pool;
@@ -4922,9 +4938,15 @@ spa_create(const char *pool, nvlist_t *nvroot, nvlist_
(error = vdev_create(rvd, txg, B_FALSE)) == 0 &&
(error = spa_validate_aux(spa, nvroot, txg,
VDEV_ALLOC_ADD)) == 0) {
- for (int c = 0; c < rvd->vdev_children; c++) {
- vdev_metaslab_set_size(rvd->vdev_child[c]);
- vdev_expand(rvd->vdev_child[c], txg);
+ /*
+ * instantiate the metaslab groups (this will dirty the vdevs)
+ * we can no longer error exit past this point
+ */
+ for (int c = 0; error == 0 && c < rvd->vdev_children; c++) {
+ vdev_t *vd = rvd->vdev_child[c];
+
+ vdev_metaslab_set_size(vd);
+ vdev_expand(vd, txg);
}
}
@@ -7082,8 +7104,14 @@ spa_async_thread(void *arg)
mutex_enter(&spa_namespace_lock);
old_space = metaslab_class_get_space(spa_normal_class(spa));
+ old_space += metaslab_class_get_space(spa_special_class(spa));
+ old_space += metaslab_class_get_space(spa_dedup_class(spa));
+
spa_config_update(spa, SPA_CONFIG_UPDATE_POOL);
+
new_space = metaslab_class_get_space(spa_normal_class(spa));
+ new_space += metaslab_class_get_space(spa_special_class(spa));
+ new_space += metaslab_class_get_space(spa_dedup_class(spa));
mutex_exit(&spa_namespace_lock);
/*
@@ -7780,6 +7808,9 @@ spa_sync(spa_t *spa, uint64_t txg)
dsl_pool_t *dp = spa->spa_dsl_pool;
objset_t *mos = spa->spa_meta_objset;
bplist_t *free_bpl = &spa->spa_free_bplist[txg & TXG_MASK];
+ metaslab_class_t *normal = spa_normal_class(spa);
+ metaslab_class_t *special = spa_special_class(spa);
+ metaslab_class_t *dedup = spa_dedup_class(spa);
vdev_t *rvd = spa->spa_root_vdev;
vdev_t *vd;
dmu_tx_t *tx;
@@ -7872,11 +7903,15 @@ spa_sync(spa_t *spa, uint64_t txg)
for (int c = 0; c < rvd->vdev_children; c++) {
vdev_t *tvd = rvd->vdev_child[c];
metaslab_group_t *mg = tvd->vdev_mg;
+ metaslab_class_t *mc;
- if (mg == NULL || mg->mg_class != spa_normal_class(spa) ||
- !metaslab_group_initialized(mg))
+ if (mg == NULL || !metaslab_group_initialized(mg))
continue;
+ mc = mg->mg_class;
+ if (mc != normal && mc != special && mc != dedup)
+ continue;
+
/*
* It is safe to do a lock-free check here because only async
* allocations look at mg_max_alloc_queue_depth, and async
@@ -7893,12 +7928,18 @@ spa_sync(spa_t *spa, uint64_t txg)
}
slots_per_allocator += zfs_vdev_def_queue_depth;
}
- metaslab_class_t *mc = spa_normal_class(spa);
+
for (int i = 0; i < spa->spa_alloc_count; i++) {
- ASSERT0(zfs_refcount_count(&mc->mc_alloc_slots[i]));
- mc->mc_alloc_max_slots[i] = slots_per_allocator;
+ ASSERT0(zfs_refcount_count(&normal->mc_alloc_slots[i]));
+ ASSERT0(zfs_refcount_count(&special->mc_alloc_slots[i]));
+ ASSERT0(zfs_refcount_count(&dedup->mc_alloc_slots[i]));
+ normal->mc_alloc_max_slots[i] = slots_per_allocator;
+ special->mc_alloc_max_slots[i] = slots_per_allocator;
+ dedup->mc_alloc_max_slots[i] = slots_per_allocator;
}
- mc->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
+ normal->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
+ special->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
+ dedup->mc_alloc_throttle_enabled = zio_dva_throttle_enabled;
for (int c = 0; c < rvd->vdev_children; c++) {
vdev_t *vd = rvd->vdev_child[c];
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/spa_misc.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -26,6 +26,7 @@
* Copyright 2013 Saso Kiselkov. All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
* Copyright (c) 2017 Datto Inc.
+ * Copyright (c) 2017, Intel Corporation.
*/
#include <sys/zfs_context.h>
@@ -388,6 +389,19 @@ spa_load_note(spa_t *spa, const char *fmt, ...)
}
/*
+ * By default dedup and user data indirects land in the special class
+ */
+int zfs_ddt_data_is_special = B_TRUE;
+int zfs_user_indirect_is_special = B_TRUE;
+
+/*
+ * The percentage of special class final space reserved for metadata only.
+ * Once we allocate 100 - zfs_special_class_metadata_reserve_pct we only
+ * let metadata into the class.
+ */
+int zfs_special_class_metadata_reserve_pct = 25;
+
+/*
* ==========================================================================
* SPA config locking
* ==========================================================================
@@ -1173,6 +1187,8 @@ spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t
*/
ASSERT(metaslab_class_validate(spa_normal_class(spa)) == 0);
ASSERT(metaslab_class_validate(spa_log_class(spa)) == 0);
+ ASSERT(metaslab_class_validate(spa_special_class(spa)) == 0);
+ ASSERT(metaslab_class_validate(spa_dedup_class(spa)) == 0);
spa_config_exit(spa, SCL_ALL, spa);
@@ -1516,6 +1532,16 @@ zfs_strtonum(const char *str, char **nptr)
return (val);
}
+void
+spa_activate_allocation_classes(spa_t *spa, dmu_tx_t *tx)
+{
+ /*
+ * We bump the feature refcount for each special vdev added to the pool
+ */
+ ASSERT(spa_feature_is_enabled(spa, SPA_FEATURE_ALLOCATION_CLASSES));
+ spa_feature_incr(spa, SPA_FEATURE_ALLOCATION_CLASSES, tx);
+}
+
/*
* ==========================================================================
* Accessor functions
@@ -1763,6 +1789,79 @@ metaslab_class_t *
spa_log_class(spa_t *spa)
{
return (spa->spa_log_class);
+}
+
+metaslab_class_t *
+spa_special_class(spa_t *spa)
+{
+ return (spa->spa_special_class);
+}
+
+metaslab_class_t *
+spa_dedup_class(spa_t *spa)
+{
+ return (spa->spa_dedup_class);
+}
+
+/*
+ * Locate an appropriate allocation class
+ */
+metaslab_class_t *
+spa_preferred_class(spa_t *spa, uint64_t size, dmu_object_type_t objtype,
+ uint_t level, uint_t special_smallblk)
+{
+ if (DMU_OT_IS_ZIL(objtype)) {
+ if (spa->spa_log_class->mc_groups != 0)
+ return (spa_log_class(spa));
+ else
+ return (spa_normal_class(spa));
+ }
+
+ boolean_t has_special_class = spa->spa_special_class->mc_groups != 0;
+
+ if (DMU_OT_IS_DDT(objtype)) {
+ if (spa->spa_dedup_class->mc_groups != 0)
+ return (spa_dedup_class(spa));
+ else if (has_special_class && zfs_ddt_data_is_special)
+ return (spa_special_class(spa));
+ else
+ return (spa_normal_class(spa));
+ }
+
+ /* Indirect blocks for user data can land in special if allowed */
+ if (level > 0 && (DMU_OT_IS_FILE(objtype) || objtype == DMU_OT_ZVOL)) {
+ if (has_special_class && zfs_user_indirect_is_special)
+ return (spa_special_class(spa));
+ else
+ return (spa_normal_class(spa));
+ }
+
+ if (DMU_OT_IS_METADATA(objtype) || level > 0) {
+ if (has_special_class)
+ return (spa_special_class(spa));
+ else
+ return (spa_normal_class(spa));
+ }
+
+ /*
+ * Allow small file blocks in special class in some cases (like
+ * for the dRAID vdev feature). But always leave a reserve of
+ * zfs_special_class_metadata_reserve_pct exclusively for metadata.
+ */
+ if (DMU_OT_IS_FILE(objtype) &&
+ has_special_class && size <= special_smallblk) {
+ metaslab_class_t *special = spa_special_class(spa);
+ uint64_t alloc = metaslab_class_get_alloc(special);
+ uint64_t space = metaslab_class_get_space(special);
+ uint64_t limit =
+ (space * (100 - zfs_special_class_metadata_reserve_pct))
+ / 100;
+
+ if (alloc < limit)
+ return (special);
+ }
+
+ return (spa_normal_class(spa));
}
void
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -21,13 +21,14 @@
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2011, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2011, 2018 by Delphix. All rights reserved.
* Copyright 2011 Nexenta Systems, Inc. All rights reserved.
* Copyright (c) 2012, Joyent, Inc. All rights reserved.
* Copyright 2013 DEY Storage Systems, Inc.
* Copyright 2014 HybridCluster. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright 2013 Saso Kiselkov. All rights reserved.
+ * Copyright (c) 2017, Intel Corporation.
* Copyright (c) 2014 Integros [integros.com]
*/
@@ -126,6 +127,16 @@ typedef enum dmu_object_byteswap {
((ot) & DMU_OT_METADATA) : \
dmu_ot[(ot)].ot_metadata)
+#define DMU_OT_IS_DDT(ot) \
+ ((ot) == DMU_OT_DDT_ZAP)
+
+#define DMU_OT_IS_ZIL(ot) \
+ ((ot) == DMU_OT_INTENT_LOG)
+
+/* Note: ztest uses DMU_OT_UINT64_OTHER as a proxy for file blocks */
+#define DMU_OT_IS_FILE(ot) \
+ ((ot) == DMU_OT_PLAIN_FILE_CONTENTS || (ot) == DMU_OT_UINT64_OTHER)
+
#define DMU_OT_IS_METADATA_CACHED(ot) (((ot) & DMU_OT_NEWTYPE) ? \
B_TRUE : dmu_ot[(ot)].ot_dbuf_metadata_cache)
@@ -216,6 +227,7 @@ typedef enum dmu_object_type {
*
* The DMU_OTN_* types do not have entries in the dmu_ot table,
* use the DMU_OT_IS_METDATA() and DMU_OT_BYTESWAP() macros instead
+ * use the DMU_OT_IS_METADATA() and DMU_OT_BYTESWAP() macros instead
* of indexing into dmu_ot directly (this works for both DMU_OT_* types
* and DMU_OTN_* types).
*/
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu_objset.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu_objset.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/dmu_objset.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -20,7 +20,7 @@
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright (c) 2014 Integros [integros.com]
@@ -113,6 +113,11 @@ struct objset {
uint64_t os_normalization;
uint64_t os_utf8only;
uint64_t os_casesensitivity;
+ /*
+ * The largest zpl file block allowed in special class.
+ * cached here instead of zfsvfs for easier access.
+ */
+ int os_zpl_special_smallblock;
/*
* Pointer is constant; the blkptr it points to is protected by
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/metaslab.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/metaslab.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/metaslab.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -21,6 +21,7 @@
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _SYS_METASLAB_H
@@ -56,12 +57,17 @@ void metaslab_sync_done(metaslab_t *, uint64_t);
void metaslab_sync_reassess(metaslab_group_t *);
uint64_t metaslab_block_maxsize(metaslab_t *);
+/*
+ * metaslab alloc flags
+ */
#define METASLAB_HINTBP_FAVOR 0x0
#define METASLAB_HINTBP_AVOID 0x1
#define METASLAB_GANG_HEADER 0x2
#define METASLAB_GANG_CHILD 0x4
#define METASLAB_ASYNC_ALLOC 0x8
#define METASLAB_DONT_THROTTLE 0x10
+#define METASLAB_MUST_RESERVE 0x20
+#define METASLAB_FASTWRITE 0x40
int metaslab_alloc(spa_t *, metaslab_class_t *, uint64_t,
blkptr_t *, int, uint64_t, blkptr_t *, int, zio_alloc_list_t *, zio_t *,
@@ -92,8 +98,6 @@ boolean_t metaslab_class_throttle_reserve(metaslab_cla
zio_t *, int);
void metaslab_class_throttle_unreserve(metaslab_class_t *, int, int, zio_t *);
-void metaslab_class_space_update(metaslab_class_t *, int64_t, int64_t,
- int64_t, int64_t);
uint64_t metaslab_class_get_alloc(metaslab_class_t *);
uint64_t metaslab_class_get_space(metaslab_class_t *);
uint64_t metaslab_class_get_dspace(metaslab_class_t *);
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -27,6 +27,7 @@
* Copyright (c) 2014 Integros [integros.com]
* Copyright 2017 Joyent, Inc.
* Copyright (c) 2017 Datto Inc.
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _SYS_SPA_H
@@ -801,6 +802,11 @@ extern uint64_t spa_version(spa_t *spa);
extern boolean_t spa_deflate(spa_t *spa);
extern metaslab_class_t *spa_normal_class(spa_t *spa);
extern metaslab_class_t *spa_log_class(spa_t *spa);
+extern metaslab_class_t *spa_special_class(spa_t *spa);
+extern metaslab_class_t *spa_dedup_class(spa_t *spa);
+extern metaslab_class_t *spa_preferred_class(spa_t *spa, uint64_t size,
+ dmu_object_type_t objtype, uint_t level, uint_t special_smallblk);
+
extern void spa_evicting_os_register(spa_t *, objset_t *os);
extern void spa_evicting_os_deregister(spa_t *, objset_t *os);
extern void spa_evicting_os_wait(spa_t *spa);
@@ -862,6 +868,7 @@ extern boolean_t spa_trust_config(spa_t *spa);
extern uint64_t spa_missing_tvds_allowed(spa_t *spa);
extern void spa_set_missing_tvds(spa_t *spa, uint64_t missing);
extern boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa);
+extern void spa_activate_allocation_classes(spa_t *, dmu_tx_t *);
extern int spa_mode(spa_t *spa);
extern uint64_t zfs_strtonum(const char *str, char **nptr);
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa_impl.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa_impl.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/spa_impl.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -25,6 +25,7 @@
* Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
* Copyright 2013 Saso Kiselkov. All rights reserved.
* Copyright (c) 2017 Datto Inc.
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _SYS_SPA_IMPL_H
@@ -218,6 +219,8 @@ struct spa {
boolean_t spa_is_initializing; /* true while opening pool */
metaslab_class_t *spa_normal_class; /* normal data class */
metaslab_class_t *spa_log_class; /* intent log data class */
+ metaslab_class_t *spa_special_class; /* special allocation class */
+ metaslab_class_t *spa_dedup_class; /* dedup allocation class */
uint64_t spa_first_txg; /* first txg after spa_open() */
uint64_t spa_final_txg; /* txg of export/destroy */
uint64_t spa_freeze_txg; /* freeze pool at this txg */
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -22,6 +22,7 @@
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _SYS_VDEV_H
@@ -106,6 +107,8 @@ extern boolean_t vdev_children_are_offline(vdev_t *vd)
extern void vdev_space_update(vdev_t *vd,
int64_t alloc_delta, int64_t defer_delta, int64_t space_delta);
+
+extern int64_t vdev_deflated_space(vdev_t *vd, int64_t space);
extern uint64_t vdev_psize_to_asize(vdev_t *vd, uint64_t psize);
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev_impl.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev_impl.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/vdev_impl.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -21,6 +21,7 @@
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
+ * Copyright (c) 2017, Intel Corporation.
*/
#ifndef _SYS_VDEV_IMPL_H
@@ -146,6 +147,14 @@ struct vdev_queue {
kmutex_t vq_lock;
};
+typedef enum vdev_alloc_bias {
+ VDEV_BIAS_NONE,
+ VDEV_BIAS_LOG, /* dedicated to ZIL data (SLOG) */
+ VDEV_BIAS_SPECIAL, /* dedicated to ddt, metadata, and small blks */
+ VDEV_BIAS_DEDUP /* dedicated to dedup metadata */
+} vdev_alloc_bias_t;
+
+
/*
* On-disk indirect vdev state.
*
@@ -239,6 +248,7 @@ struct vdev {
boolean_t vdev_ishole; /* is a hole in the namespace */
kmutex_t vdev_queue_lock; /* protects vdev_queue_depth */
uint64_t vdev_top_zap;
+ vdev_alloc_bias_t vdev_alloc_bias; /* metaslab allocation bias */
/* pool checkpoint related */
space_map_t *vdev_checkpoint_sm; /* contains reserved blocks */
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/sys/zio.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/sys/zio.h Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/sys/zio.h Wed Nov 6 08:58:03 2019 (r354382)
@@ -313,6 +313,7 @@ typedef struct zio_prop {
boolean_t zp_dedup;
boolean_t zp_dedup_verify;
boolean_t zp_nopwrite;
+ uint32_t zp_zpl_smallblk;
} zio_prop_t;
typedef struct zio_cksum_report zio_cksum_report_t;
@@ -433,6 +434,7 @@ struct zio {
vdev_t *io_vd;
void *io_vsd;
const zio_vsd_ops_t *io_vsd_ops;
+ metaslab_class_t *io_metaslab_class; /* dva throttle class */
uint64_t io_offset;
hrtime_t io_timestamp;
Modified: vendor-sys/illumos/dist/uts/common/fs/zfs/vdev.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/fs/zfs/vdev.c Wed Nov 6 08:55:23 2019 (r354381)
+++ vendor-sys/illumos/dist/uts/common/fs/zfs/vdev.c Wed Nov 6 08:58:03 2019 (r354382)
@@ -26,6 +26,7 @@
* Copyright (c) 2014 Integros [integros.com]
* Copyright 2016 Toomas Soome <tsoome at me.com>
* Copyright 2017 Joyent, Inc.
+ * Copyright (c) 2017, Intel Corporation.
*/
#include <sys/zfs_context.h>
@@ -192,6 +193,25 @@ vdev_getops(const char *type)
return (ops);
}
+/*
+ * Derive the enumerated alloction bias from string input.
+ * String origin is either the per-vdev zap or zpool(1M).
+ */
+static vdev_alloc_bias_t
+vdev_derive_alloc_bias(const char *bias)
+{
+ vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE;
+
+ if (strcmp(bias, VDEV_ALLOC_BIAS_LOG) == 0)
+ alloc_bias = VDEV_BIAS_LOG;
+ else if (strcmp(bias, VDEV_ALLOC_BIAS_SPECIAL) == 0)
+ alloc_bias = VDEV_BIAS_SPECIAL;
+ else if (strcmp(bias, VDEV_ALLOC_BIAS_DEDUP) == 0)
+ alloc_bias = VDEV_BIAS_DEDUP;
+
+ return (alloc_bias);
+}
+
/* ARGSUSED */
void
vdev_default_xlate(vdev_t *vd, const range_seg_t *in, range_seg_t *res)
@@ -515,6 +535,8 @@ vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vde
uint64_t guid = 0, islog, nparity;
vdev_t *vd;
vdev_indirect_config_t *vic;
+ vdev_alloc_bias_t alloc_bias = VDEV_BIAS_NONE;
+ boolean_t top_level = (parent && !parent->vdev_parent);
ASSERT(spa_config_held(spa, SCL_ALL, RW_WRITER) == SCL_ALL);
@@ -601,11 +623,32 @@ vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *nv, vde
}
ASSERT(nparity != -1ULL);
*** DIFF OUTPUT TRUNCATED AT 1000 LINES ***
More information about the svn-src-all
mailing list